src/hotspot/share/gc/shenandoah/shenandoahHeap.cpp

/*
 * Copyright (c) 2023, 2025, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2013, 2022, Red Hat, Inc. All rights reserved.
 * Copyright Amazon.com Inc. 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 "cds/archiveHeapWriter.hpp"
#include "classfile/systemDictionary.hpp"
#include "gc/shared/classUnloadingContext.hpp"
#include "gc/shared/fullGCForwarding.hpp"
#include "gc/shared/gcArguments.hpp"
#include "gc/shared/gcTimer.hpp"
#include "gc/shared/gcTraceTime.inline.hpp"
#include "gc/shared/locationPrinter.inline.hpp"
#include "gc/shared/memAllocator.hpp"
#include "gc/shared/plab.hpp"
#include "gc/shared/tlab_globals.hpp"
#include "gc/shenandoah/heuristics/shenandoahOldHeuristics.hpp"
#include "gc/shenandoah/heuristics/shenandoahYoungHeuristics.hpp"
#include "gc/shenandoah/mode/shenandoahGenerationalMode.hpp"
#include "gc/shenandoah/mode/shenandoahPassiveMode.hpp"
#include "gc/shenandoah/mode/shenandoahSATBMode.hpp"
#include "gc/shenandoah/shenandoahAllocRequest.hpp"
#include "gc/shenandoah/shenandoahBarrierSet.hpp"
#include "gc/shenandoah/shenandoahClosures.inline.hpp"
#include "gc/shenandoah/shenandoahCodeRoots.hpp"
#include "gc/shenandoah/shenandoahCollectionSet.hpp"
#include "gc/shenandoah/shenandoahCollectorPolicy.hpp"
#include "gc/shenandoah/shenandoahConcurrentMark.hpp"
#include "gc/shenandoah/shenandoahControlThread.hpp"
#include "gc/shenandoah/shenandoahFreeSet.hpp"
#include "gc/shenandoah/shenandoahGenerationalEvacuationTask.hpp"
#include "gc/shenandoah/shenandoahGenerationalHeap.hpp"
#include "gc/shenandoah/shenandoahGlobalGeneration.hpp"
#include "gc/shenandoah/shenandoahHeap.inline.hpp"
#include "gc/shenandoah/shenandoahHeapRegion.inline.hpp"
#include "gc/shenandoah/shenandoahHeapRegionClosures.hpp"
#include "gc/shenandoah/shenandoahHeapRegionSet.hpp"
#include "gc/shenandoah/shenandoahInitLogger.hpp"
#include "gc/shenandoah/shenandoahMarkingContext.inline.hpp"
#include "gc/shenandoah/shenandoahMemoryPool.hpp"
#include "gc/shenandoah/shenandoahMonitoringSupport.hpp"
#include "gc/shenandoah/shenandoahOldGeneration.hpp"
#include "gc/shenandoah/shenandoahPacer.inline.hpp"
#include "gc/shenandoah/shenandoahPadding.hpp"
#include "gc/shenandoah/shenandoahParallelCleaning.inline.hpp"
#include "gc/shenandoah/shenandoahPhaseTimings.hpp"
#include "gc/shenandoah/shenandoahReferenceProcessor.hpp"
#include "gc/shenandoah/shenandoahRootProcessor.inline.hpp"
#include "gc/shenandoah/shenandoahScanRemembered.inline.hpp"
#include "gc/shenandoah/shenandoahSTWMark.hpp"
#include "gc/shenandoah/shenandoahUncommitThread.hpp"
#include "gc/shenandoah/shenandoahUtils.hpp"
#include "gc/shenandoah/shenandoahVerifier.hpp"
#include "gc/shenandoah/shenandoahVMOperations.hpp"
#include "gc/shenandoah/shenandoahWorkerPolicy.hpp"
#include "gc/shenandoah/shenandoahWorkGroup.hpp"
#include "gc/shenandoah/shenandoahYoungGeneration.hpp"
#include "memory/allocation.hpp"
#include "memory/classLoaderMetaspace.hpp"
#include "memory/memoryReserver.hpp"
#include "memory/metaspaceUtils.hpp"
#include "memory/universe.hpp"
#include "nmt/mallocTracker.hpp"
#include "nmt/memTracker.hpp"
#include "oops/compressedOops.inline.hpp"
#include "prims/jvmtiTagMap.hpp"
#include "runtime/atomic.hpp"
#include "runtime/globals.hpp"
#include "runtime/interfaceSupport.inline.hpp"
#include "runtime/java.hpp"
#include "runtime/orderAccess.hpp"
#include "runtime/safepointMechanism.hpp"
#include "runtime/stackWatermarkSet.hpp"
#include "runtime/threads.hpp"
#include "runtime/vmThread.hpp"
#include "utilities/events.hpp"
#include "utilities/globalDefinitions.hpp"
#include "utilities/powerOfTwo.hpp"
#if INCLUDE_JVMCI
#include "jvmci/jvmci.hpp"
#endif
#if INCLUDE_JFR
#include "gc/shenandoah/shenandoahJfrSupport.hpp"
#endif

class ShenandoahPretouchHeapTask : public WorkerTask {
private:
  ShenandoahRegionIterator _regions;
  const size_t _page_size;
public:
  ShenandoahPretouchHeapTask(size_t page_size) :
    WorkerTask("Shenandoah Pretouch Heap"),
    _page_size(page_size) {}

  virtual void work(uint worker_id) {
    ShenandoahHeapRegion* r = _regions.next();
    while (r != nullptr) {
      if (r->is_committed()) {
        os::pretouch_memory(r->bottom(), r->end(), _page_size);
      }
      r = _regions.next();
    }
  }
};

class ShenandoahPretouchBitmapTask : public WorkerTask {
private:
  ShenandoahRegionIterator _regions;
  char* _bitmap_base;
  const size_t _bitmap_size;
  const size_t _page_size;
public:
  ShenandoahPretouchBitmapTask(char* bitmap_base, size_t bitmap_size, size_t page_size) :
    WorkerTask("Shenandoah Pretouch Bitmap"),
    _bitmap_base(bitmap_base),
    _bitmap_size(bitmap_size),
    _page_size(page_size) {}

  virtual void work(uint worker_id) {
    ShenandoahHeapRegion* r = _regions.next();
    while (r != nullptr) {
      size_t start = r->index()       * ShenandoahHeapRegion::region_size_bytes() / MarkBitMap::heap_map_factor();
      size_t end   = (r->index() + 1) * ShenandoahHeapRegion::region_size_bytes() / MarkBitMap::heap_map_factor();
      assert (end <= _bitmap_size, "end is sane: %zu < %zu", end, _bitmap_size);

      if (r->is_committed()) {
        os::pretouch_memory(_bitmap_base + start, _bitmap_base + end, _page_size);
      }

      r = _regions.next();
    }
  }
};

static ReservedSpace reserve(size_t size, size_t preferred_page_size) {
  // When a page size is given we don't want to mix large
  // and normal pages. If the size is not a multiple of the
  // page size it will be aligned up to achieve this.
  size_t alignment = os::vm_allocation_granularity();
  if (preferred_page_size != os::vm_page_size()) {
    alignment = MAX2(preferred_page_size, alignment);
    size = align_up(size, alignment);
  }

  const ReservedSpace reserved = MemoryReserver::reserve(size, alignment, preferred_page_size, mtGC);
  if (!reserved.is_reserved()) {
    vm_exit_during_initialization("Could not reserve space");
  }
  return reserved;
}

jint ShenandoahHeap::initialize() {
  //
  // Figure out heap sizing
  //

  size_t init_byte_size = InitialHeapSize;
  size_t min_byte_size  = MinHeapSize;
  size_t max_byte_size  = MaxHeapSize;
  size_t heap_alignment = HeapAlignment;

  size_t reg_size_bytes = ShenandoahHeapRegion::region_size_bytes();

  Universe::check_alignment(max_byte_size,  reg_size_bytes, "Shenandoah heap");
  Universe::check_alignment(init_byte_size, reg_size_bytes, "Shenandoah heap");

  _num_regions = ShenandoahHeapRegion::region_count();
  assert(_num_regions == (max_byte_size / reg_size_bytes),
         "Regions should cover entire heap exactly: %zu != %zu/%zu",
         _num_regions, max_byte_size, reg_size_bytes);

  size_t num_committed_regions = init_byte_size / reg_size_bytes;
  num_committed_regions = MIN2(num_committed_regions, _num_regions);
  assert(num_committed_regions <= _num_regions, "sanity");
  _initial_size = num_committed_regions * reg_size_bytes;

  size_t num_min_regions = min_byte_size / reg_size_bytes;
  num_min_regions = MIN2(num_min_regions, _num_regions);
  assert(num_min_regions <= _num_regions, "sanity");
  _minimum_size = num_min_regions * reg_size_bytes;

  // Default to max heap size.
  _soft_max_size = _num_regions * reg_size_bytes;

  _committed = _initial_size;

  size_t heap_page_size   = UseLargePages ? os::large_page_size() : os::vm_page_size();
  size_t bitmap_page_size = UseLargePages ? os::large_page_size() : os::vm_page_size();
  size_t region_page_size = UseLargePages ? os::large_page_size() : os::vm_page_size();

  //
  // Reserve and commit memory for heap
  //

  ReservedHeapSpace heap_rs = Universe::reserve_heap(max_byte_size, heap_alignment);
  initialize_reserved_region(heap_rs);
  _heap_region = MemRegion((HeapWord*)heap_rs.base(), heap_rs.size() / HeapWordSize);
  _heap_region_special = heap_rs.special();

  assert((((size_t) base()) & ShenandoahHeapRegion::region_size_bytes_mask()) == 0,
         "Misaligned heap: " PTR_FORMAT, p2i(base()));
  os::trace_page_sizes_for_requested_size("Heap",
                                          max_byte_size, heap_alignment,
                                          heap_rs.base(),
                                          heap_rs.size(), heap_rs.page_size());

#if SHENANDOAH_OPTIMIZED_MARKTASK
  // The optimized ShenandoahMarkTask takes some bits away from the full object bits.
  // Fail if we ever attempt to address more than we can.
  if ((uintptr_t)heap_rs.end() >= ShenandoahMarkTask::max_addressable()) {
    FormatBuffer<512> buf("Shenandoah reserved [" PTR_FORMAT ", " PTR_FORMAT") for the heap, \n"
                          "but max object address is " PTR_FORMAT ". Try to reduce heap size, or try other \n"
                          "VM options that allocate heap at lower addresses (HeapBaseMinAddress, AllocateHeapAt, etc).",
                p2i(heap_rs.base()), p2i(heap_rs.end()), ShenandoahMarkTask::max_addressable());
    vm_exit_during_initialization("Fatal Error", buf);
  }
#endif

  ReservedSpace sh_rs = heap_rs.first_part(max_byte_size);
  if (!_heap_region_special) {
    os::commit_memory_or_exit(sh_rs.base(), _initial_size, heap_alignment, false,
                              "Cannot commit heap memory");
  }

  BarrierSet::set_barrier_set(new ShenandoahBarrierSet(this, _heap_region));

  // Now we know the number of regions and heap sizes, initialize the heuristics.
  initialize_heuristics();

  assert(_heap_region.byte_size() == heap_rs.size(), "Need to know reserved size for card table");

  //
  // Worker threads must be initialized after the barrier is configured
  //
  _workers = new ShenandoahWorkerThreads("Shenandoah GC Threads", _max_workers);
  if (_workers == nullptr) {
    vm_exit_during_initialization("Failed necessary allocation.");
  } else {
    _workers->initialize_workers();
  }

  if (ParallelGCThreads > 1) {
    _safepoint_workers = new ShenandoahWorkerThreads("Safepoint Cleanup Thread", ParallelGCThreads);
    _safepoint_workers->initialize_workers();
  }

  //
  // Reserve and commit memory for bitmap(s)
  //

  size_t bitmap_size_orig = ShenandoahMarkBitMap::compute_size(heap_rs.size());
  _bitmap_size = align_up(bitmap_size_orig, bitmap_page_size);

  size_t bitmap_bytes_per_region = reg_size_bytes / ShenandoahMarkBitMap::heap_map_factor();

  guarantee(bitmap_bytes_per_region != 0,
            "Bitmap bytes per region should not be zero");
  guarantee(is_power_of_2(bitmap_bytes_per_region),
            "Bitmap bytes per region should be power of two: %zu", bitmap_bytes_per_region);

  if (bitmap_page_size > bitmap_bytes_per_region) {
    _bitmap_regions_per_slice = bitmap_page_size / bitmap_bytes_per_region;
    _bitmap_bytes_per_slice = bitmap_page_size;
  } else {
    _bitmap_regions_per_slice = 1;
    _bitmap_bytes_per_slice = bitmap_bytes_per_region;
  }

  guarantee(_bitmap_regions_per_slice >= 1,
            "Should have at least one region per slice: %zu",
            _bitmap_regions_per_slice);

  guarantee(((_bitmap_bytes_per_slice) % bitmap_page_size) == 0,
            "Bitmap slices should be page-granular: bps = %zu, page size = %zu",
            _bitmap_bytes_per_slice, bitmap_page_size);

  ReservedSpace bitmap = reserve(_bitmap_size, bitmap_page_size);
  os::trace_page_sizes_for_requested_size("Mark Bitmap",
                                          bitmap_size_orig, bitmap_page_size,
                                          bitmap.base(),
                                          bitmap.size(), bitmap.page_size());
  MemTracker::record_virtual_memory_tag(bitmap, mtGC);
  _bitmap_region = MemRegion((HeapWord*) bitmap.base(), bitmap.size() / HeapWordSize);
  _bitmap_region_special = bitmap.special();

  size_t bitmap_init_commit = _bitmap_bytes_per_slice *
    align_up(num_committed_regions, _bitmap_regions_per_slice) / _bitmap_regions_per_slice;
  bitmap_init_commit = MIN2(_bitmap_size, bitmap_init_commit);
  if (!_bitmap_region_special) {
    os::commit_memory_or_exit((char *) _bitmap_region.start(), bitmap_init_commit, bitmap_page_size, false,
                              "Cannot commit bitmap memory");
  }

  _marking_context = new ShenandoahMarkingContext(_heap_region, _bitmap_region, _num_regions);

  if (ShenandoahVerify) {
    ReservedSpace verify_bitmap = reserve(_bitmap_size, bitmap_page_size);
    os::trace_page_sizes_for_requested_size("Verify Bitmap",
                                            bitmap_size_orig, bitmap_page_size,
                                            verify_bitmap.base(),
                                            verify_bitmap.size(), verify_bitmap.page_size());
    if (!verify_bitmap.special()) {
      os::commit_memory_or_exit(verify_bitmap.base(), verify_bitmap.size(), bitmap_page_size, false,
                                "Cannot commit verification bitmap memory");
    }
    MemTracker::record_virtual_memory_tag(verify_bitmap, mtGC);
    MemRegion verify_bitmap_region = MemRegion((HeapWord *) verify_bitmap.base(), verify_bitmap.size() / HeapWordSize);
    _verification_bit_map.initialize(_heap_region, verify_bitmap_region);
    _verifier = new ShenandoahVerifier(this, &_verification_bit_map);
  }

  // Reserve aux bitmap for use in object_iterate(). We don't commit it here.
  size_t aux_bitmap_page_size = bitmap_page_size;

  ReservedSpace aux_bitmap = reserve(_bitmap_size, aux_bitmap_page_size);
  os::trace_page_sizes_for_requested_size("Aux Bitmap",
                                          bitmap_size_orig, aux_bitmap_page_size,
                                          aux_bitmap.base(),
                                          aux_bitmap.size(), aux_bitmap.page_size());
  MemTracker::record_virtual_memory_tag(aux_bitmap, mtGC);
  _aux_bitmap_region = MemRegion((HeapWord*) aux_bitmap.base(), aux_bitmap.size() / HeapWordSize);
  _aux_bitmap_region_special = aux_bitmap.special();
  _aux_bit_map.initialize(_heap_region, _aux_bitmap_region);

  //
  // Create regions and region sets
  //
  size_t region_align = align_up(sizeof(ShenandoahHeapRegion), SHENANDOAH_CACHE_LINE_SIZE);
  size_t region_storage_size_orig = region_align * _num_regions;
  size_t region_storage_size = align_up(region_storage_size_orig,
                                        MAX2(region_page_size, os::vm_allocation_granularity()));

  ReservedSpace region_storage = reserve(region_storage_size, region_page_size);
  os::trace_page_sizes_for_requested_size("Region Storage",
                                          region_storage_size_orig, region_page_size,
                                          region_storage.base(),
                                          region_storage.size(), region_storage.page_size());
  MemTracker::record_virtual_memory_tag(region_storage, mtGC);
  if (!region_storage.special()) {
    os::commit_memory_or_exit(region_storage.base(), region_storage_size, region_page_size, false,
                              "Cannot commit region memory");
  }

  // Try to fit the collection set bitmap at lower addresses. This optimizes code generation for cset checks.
  // Go up until a sensible limit (subject to encoding constraints) and try to reserve the space there.
  // If not successful, bite a bullet and allocate at whatever address.
  {
    const size_t cset_align = MAX2<size_t>(os::vm_page_size(), os::vm_allocation_granularity());
    const size_t cset_size = align_up(((size_t) sh_rs.base() + sh_rs.size()) >> ShenandoahHeapRegion::region_size_bytes_shift(), cset_align);
    const size_t cset_page_size = os::vm_page_size();

    uintptr_t min = round_up_power_of_2(cset_align);
    uintptr_t max = (1u << 30u);
    ReservedSpace cset_rs;

    for (uintptr_t addr = min; addr <= max; addr <<= 1u) {
      char* req_addr = (char*)addr;
      assert(is_aligned(req_addr, cset_align), "Should be aligned");
      cset_rs = MemoryReserver::reserve(req_addr, cset_size, cset_align, cset_page_size, mtGC);
      if (cset_rs.is_reserved()) {
        assert(cset_rs.base() == req_addr, "Allocated where requested: " PTR_FORMAT ", " PTR_FORMAT, p2i(cset_rs.base()), addr);
        _collection_set = new ShenandoahCollectionSet(this, cset_rs, sh_rs.base());
        break;
      }
    }

    if (_collection_set == nullptr) {
      cset_rs = MemoryReserver::reserve(cset_size, cset_align, os::vm_page_size(), mtGC);
      if (!cset_rs.is_reserved()) {
        vm_exit_during_initialization("Cannot reserve memory for collection set");
      }

      _collection_set = new ShenandoahCollectionSet(this, cset_rs, sh_rs.base());
    }
    os::trace_page_sizes_for_requested_size("Collection Set",
                                            cset_size, cset_page_size,
                                            cset_rs.base(),
                                            cset_rs.size(), cset_rs.page_size());
  }

  _regions = NEW_C_HEAP_ARRAY(ShenandoahHeapRegion*, _num_regions, mtGC);
  _affiliations = NEW_C_HEAP_ARRAY(uint8_t, _num_regions, mtGC);

  {
    ShenandoahHeapLocker locker(lock());
    _free_set = new ShenandoahFreeSet(this, _num_regions);
    for (size_t i = 0; i < _num_regions; i++) {
      HeapWord* start = (HeapWord*)sh_rs.base() + ShenandoahHeapRegion::region_size_words() * i;
      bool is_committed = i < num_committed_regions;
      void* loc = region_storage.base() + i * region_align;

      ShenandoahHeapRegion* r = new (loc) ShenandoahHeapRegion(start, i, is_committed);
      assert(is_aligned(r, SHENANDOAH_CACHE_LINE_SIZE), "Sanity");

      _marking_context->initialize_top_at_mark_start(r);
      _regions[i] = r;
      assert(!collection_set()->is_in(i), "New region should not be in collection set");

      _affiliations[i] = ShenandoahAffiliation::FREE;
    }

    size_t young_cset_regions, old_cset_regions;

    // We are initializing free set.  We ignore cset region tallies.
    size_t first_old, last_old, num_old;
    _free_set->prepare_to_rebuild(young_cset_regions, old_cset_regions, first_old, last_old, num_old);
    _free_set->finish_rebuild(young_cset_regions, old_cset_regions, num_old);
  }

  if (AlwaysPreTouch) {
    // For NUMA, it is important to pre-touch the storage under bitmaps with worker threads,
    // before initialize() below zeroes it with initializing thread. For any given region,
    // we touch the region and the corresponding bitmaps from the same thread.
    ShenandoahPushWorkerScope scope(workers(), _max_workers, false);

    _pretouch_heap_page_size = heap_page_size;
    _pretouch_bitmap_page_size = bitmap_page_size;

    // OS memory managers may want to coalesce back-to-back pages. Make their jobs
    // simpler by pre-touching continuous spaces (heap and bitmap) separately.

    ShenandoahPretouchBitmapTask bcl(bitmap.base(), _bitmap_size, _pretouch_bitmap_page_size);
    _workers->run_task(&bcl);

    ShenandoahPretouchHeapTask hcl(_pretouch_heap_page_size);
    _workers->run_task(&hcl);
  }

  //
  // Initialize the rest of GC subsystems
  //

  _liveness_cache = NEW_C_HEAP_ARRAY(ShenandoahLiveData*, _max_workers, mtGC);
  for (uint worker = 0; worker < _max_workers; worker++) {
    _liveness_cache[worker] = NEW_C_HEAP_ARRAY(ShenandoahLiveData, _num_regions, mtGC);
    Copy::fill_to_bytes(_liveness_cache[worker], _num_regions * sizeof(ShenandoahLiveData));
  }

  // There should probably be Shenandoah-specific options for these,
  // just as there are G1-specific options.
  {
    ShenandoahSATBMarkQueueSet& satbqs = ShenandoahBarrierSet::satb_mark_queue_set();
    satbqs.set_process_completed_buffers_threshold(20); // G1SATBProcessCompletedThreshold
    satbqs.set_buffer_enqueue_threshold_percentage(60); // G1SATBBufferEnqueueingThresholdPercent
  }

  _monitoring_support = new ShenandoahMonitoringSupport(this);
  _phase_timings = new ShenandoahPhaseTimings(max_workers());
  ShenandoahCodeRoots::initialize();

  if (ShenandoahPacing) {
    _pacer = new ShenandoahPacer(this);
    _pacer->setup_for_idle();
  }

  initialize_controller();

  if (ShenandoahUncommit) {
    _uncommit_thread = new ShenandoahUncommitThread(this);
  }

  print_init_logger();

  FullGCForwarding::initialize(_heap_region);

  return JNI_OK;
}

void ShenandoahHeap::initialize_controller() {
  _control_thread = new ShenandoahControlThread();
}

void ShenandoahHeap::print_init_logger() const {
  ShenandoahInitLogger::print();
}

void ShenandoahHeap::initialize_mode() {
  if (ShenandoahGCMode != nullptr) {
    if (strcmp(ShenandoahGCMode, "satb") == 0) {
      _gc_mode = new ShenandoahSATBMode();
    } else if (strcmp(ShenandoahGCMode, "passive") == 0) {
      _gc_mode = new ShenandoahPassiveMode();
    } else if (strcmp(ShenandoahGCMode, "generational") == 0) {
      _gc_mode = new ShenandoahGenerationalMode();
    } else {
      vm_exit_during_initialization("Unknown -XX:ShenandoahGCMode option");
    }
  } else {
    vm_exit_during_initialization("Unknown -XX:ShenandoahGCMode option (null)");
  }
  _gc_mode->initialize_flags();
  if (_gc_mode->is_diagnostic() && !UnlockDiagnosticVMOptions) {
    vm_exit_during_initialization(
            err_msg("GC mode \"%s\" is diagnostic, and must be enabled via -XX:+UnlockDiagnosticVMOptions.",
                    _gc_mode->name()));
  }
  if (_gc_mode->is_experimental() && !UnlockExperimentalVMOptions) {
    vm_exit_during_initialization(
            err_msg("GC mode \"%s\" is experimental, and must be enabled via -XX:+UnlockExperimentalVMOptions.",
                    _gc_mode->name()));
  }
}

void ShenandoahHeap::initialize_heuristics() {
  _global_generation = new ShenandoahGlobalGeneration(mode()->is_generational(), max_workers(), max_capacity(), max_capacity());
  _global_generation->initialize_heuristics(mode());
}

#ifdef _MSC_VER
#pragma warning( push )
#pragma warning( disable:4355 ) // 'this' : used in base member initializer list
#endif

ShenandoahHeap::ShenandoahHeap(ShenandoahCollectorPolicy* policy) :
  CollectedHeap(),
  _gc_generation(nullptr),
  _active_generation(nullptr),
  _initial_size(0),
  _committed(0),
  _max_workers(MAX3(ConcGCThreads, ParallelGCThreads, 1U)),
  _workers(nullptr),
  _safepoint_workers(nullptr),
  _heap_region_special(false),
  _num_regions(0),
  _regions(nullptr),
  _affiliations(nullptr),
  _gc_state_changed(false),
  _gc_no_progress_count(0),
  _cancel_requested_time(0),
  _update_refs_iterator(this),
  _global_generation(nullptr),
  _control_thread(nullptr),
  _uncommit_thread(nullptr),
  _young_generation(nullptr),
  _old_generation(nullptr),
  _shenandoah_policy(policy),
  _gc_mode(nullptr),
  _free_set(nullptr),
  _pacer(nullptr),
  _verifier(nullptr),
  _phase_timings(nullptr),
  _monitoring_support(nullptr),
  _memory_pool(nullptr),
  _stw_memory_manager("Shenandoah Pauses"),
  _cycle_memory_manager("Shenandoah Cycles"),
  _gc_timer(new ConcurrentGCTimer()),
  _log_min_obj_alignment_in_bytes(LogMinObjAlignmentInBytes),
  _marking_context(nullptr),
  _bitmap_size(0),
  _bitmap_regions_per_slice(0),
  _bitmap_bytes_per_slice(0),
  _bitmap_region_special(false),
  _aux_bitmap_region_special(false),
  _liveness_cache(nullptr),
  _collection_set(nullptr)
{
  // Initialize GC mode early, many subsequent initialization procedures depend on it
  initialize_mode();
  _cancelled_gc.set(GCCause::_no_gc);
}

#ifdef _MSC_VER
#pragma warning( pop )
#endif

void ShenandoahHeap::print_heap_on(outputStream* st) const {
  st->print_cr("Shenandoah Heap");
  st->print_cr(" %zu%s max, %zu%s soft max, %zu%s committed, %zu%s used",
               byte_size_in_proper_unit(max_capacity()), proper_unit_for_byte_size(max_capacity()),
               byte_size_in_proper_unit(soft_max_capacity()), proper_unit_for_byte_size(soft_max_capacity()),
               byte_size_in_proper_unit(committed()),    proper_unit_for_byte_size(committed()),
               byte_size_in_proper_unit(used()),         proper_unit_for_byte_size(used()));
  st->print_cr(" %zu x %zu %s regions",
               num_regions(),
               byte_size_in_proper_unit(ShenandoahHeapRegion::region_size_bytes()),
               proper_unit_for_byte_size(ShenandoahHeapRegion::region_size_bytes()));

  st->print("Status: ");
  if (has_forwarded_objects())                 st->print("has forwarded objects, ");
  if (!mode()->is_generational()) {
    if (is_concurrent_mark_in_progress())      st->print("marking,");
  } else {
    if (is_concurrent_old_mark_in_progress())    st->print("old marking, ");
    if (is_concurrent_young_mark_in_progress())  st->print("young marking, ");
  }
  if (is_evacuation_in_progress())             st->print("evacuating, ");
  if (is_update_refs_in_progress())            st->print("updating refs, ");
  if (is_degenerated_gc_in_progress())         st->print("degenerated gc, ");
  if (is_full_gc_in_progress())                st->print("full gc, ");
  if (is_full_gc_move_in_progress())           st->print("full gc move, ");
  if (is_concurrent_weak_root_in_progress())   st->print("concurrent weak roots, ");
  if (is_concurrent_strong_root_in_progress() &&
      !is_concurrent_weak_root_in_progress())  st->print("concurrent strong roots, ");

  if (cancelled_gc()) {
    st->print("cancelled");
  } else {
    st->print("not cancelled");
  }
  st->cr();

  st->print_cr("Reserved region:");
  st->print_cr(" - [" PTR_FORMAT ", " PTR_FORMAT ") ",
               p2i(reserved_region().start()),
               p2i(reserved_region().end()));

  ShenandoahCollectionSet* cset = collection_set();
  st->print_cr("Collection set:");
  if (cset != nullptr) {
    st->print_cr(" - map (vanilla): " PTR_FORMAT, p2i(cset->map_address()));
    st->print_cr(" - map (biased):  " PTR_FORMAT, p2i(cset->biased_map_address()));
  } else {
    st->print_cr(" (null)");
  }

  st->cr();

  if (Verbose) {
    st->cr();
    print_heap_regions_on(st);
  }
}

void ShenandoahHeap::print_gc_on(outputStream* st) const {
  print_heap_regions_on(st);
}

class ShenandoahInitWorkerGCLABClosure : public ThreadClosure {
public:
  void do_thread(Thread* thread) {
    assert(thread != nullptr, "Sanity");
    ShenandoahThreadLocalData::initialize_gclab(thread);
  }
};

void ShenandoahHeap::post_initialize() {
  CollectedHeap::post_initialize();

  // Schedule periodic task to report on gc thread CPU utilization
  _mmu_tracker.initialize();

  MutexLocker ml(Threads_lock);

  ShenandoahInitWorkerGCLABClosure init_gclabs;
  _workers->threads_do(&init_gclabs);

  // gclab can not be initialized early during VM startup, as it can not determinate its max_size.
  // Now, we will let WorkerThreads to initialize gclab when new worker is created.
  _workers->set_initialize_gclab();

  // Note that the safepoint workers may require gclabs if the threads are used to create a heap dump
  // during a concurrent evacuation phase.
  if (_safepoint_workers != nullptr) {
    _safepoint_workers->threads_do(&init_gclabs);
    _safepoint_workers->set_initialize_gclab();
  }

  JFR_ONLY(ShenandoahJFRSupport::register_jfr_type_serializers();)
}

ShenandoahHeuristics* ShenandoahHeap::heuristics() {
  return _global_generation->heuristics();
}

size_t ShenandoahHeap::used() const {
  return global_generation()->used();
}

size_t ShenandoahHeap::committed() const {
  return Atomic::load(&_committed);
}

void ShenandoahHeap::increase_committed(size_t bytes) {
  shenandoah_assert_heaplocked_or_safepoint();
  _committed += bytes;
}

void ShenandoahHeap::decrease_committed(size_t bytes) {
  shenandoah_assert_heaplocked_or_safepoint();
  _committed -= bytes;
}

// For tracking usage based on allocations, it should be the case that:
// * The sum of regions::used == heap::used
// * The sum of a generation's regions::used == generation::used
// * The sum of a generation's humongous regions::free == generation::humongous_waste
// These invariants are checked by the verifier on GC safepoints.
//
// Additional notes:
// * When a mutator's allocation request causes a region to be retired, the
//   free memory left in that region is considered waste. It does not contribute
//   to the usage, but it _does_ contribute to allocation rate.
// * The bottom of a PLAB must be aligned on card size. In some cases this will
//   require padding in front of the PLAB (a filler object). Because this padding
//   is included in the region's used memory we include the padding in the usage
//   accounting as waste.
// * Mutator allocations are used to compute an allocation rate. They are also
//   sent to the Pacer for those purposes.
// * There are three sources of waste:
//  1. The padding used to align a PLAB on card size
//  2. Region's free is less than minimum TLAB size and is retired
//  3. The unused portion of memory in the last region of a humongous object
void ShenandoahHeap::increase_used(const ShenandoahAllocRequest& req) {
  size_t actual_bytes = req.actual_size() * HeapWordSize;
  size_t wasted_bytes = req.waste() * HeapWordSize;
  ShenandoahGeneration* generation = generation_for(req.affiliation());

  if (req.is_gc_alloc()) {
    assert(wasted_bytes == 0 || req.type() == ShenandoahAllocRequest::_alloc_plab, "Only PLABs have waste");
    increase_used(generation, actual_bytes + wasted_bytes);
  } else {
    assert(req.is_mutator_alloc(), "Expected mutator alloc here");
    // padding and actual size both count towards allocation counter
    generation->increase_allocated(actual_bytes + wasted_bytes);

    // only actual size counts toward usage for mutator allocations
    increase_used(generation, actual_bytes);

    // notify pacer of both actual size and waste
    notify_mutator_alloc_words(req.actual_size(), req.waste());

    if (wasted_bytes > 0 && ShenandoahHeapRegion::requires_humongous(req.actual_size())) {
      increase_humongous_waste(generation,wasted_bytes);
    }
  }
}

void ShenandoahHeap::increase_humongous_waste(ShenandoahGeneration* generation, size_t bytes) {
  generation->increase_humongous_waste(bytes);
  if (!generation->is_global()) {
    global_generation()->increase_humongous_waste(bytes);
  }
}

void ShenandoahHeap::decrease_humongous_waste(ShenandoahGeneration* generation, size_t bytes) {
  generation->decrease_humongous_waste(bytes);
  if (!generation->is_global()) {
    global_generation()->decrease_humongous_waste(bytes);
  }
}

void ShenandoahHeap::increase_used(ShenandoahGeneration* generation, size_t bytes) {
  generation->increase_used(bytes);
  if (!generation->is_global()) {
    global_generation()->increase_used(bytes);
  }
}

void ShenandoahHeap::decrease_used(ShenandoahGeneration* generation, size_t bytes) {
  generation->decrease_used(bytes);
  if (!generation->is_global()) {
    global_generation()->decrease_used(bytes);
  }
}

void ShenandoahHeap::notify_mutator_alloc_words(size_t words, size_t waste) {
  if (ShenandoahPacing) {
    control_thread()->pacing_notify_alloc(words);
    if (waste > 0) {
      pacer()->claim_for_alloc<true>(waste);
    }
  }
}

size_t ShenandoahHeap::capacity() const {
  return committed();
}

size_t ShenandoahHeap::max_capacity() const {
  return _num_regions * ShenandoahHeapRegion::region_size_bytes();
}

size_t ShenandoahHeap::soft_max_capacity() const {
  size_t v = Atomic::load(&_soft_max_size);
  assert(min_capacity() <= v && v <= max_capacity(),
         "Should be in bounds: %zu <= %zu <= %zu",
         min_capacity(), v, max_capacity());
  return v;
}

void ShenandoahHeap::set_soft_max_capacity(size_t v) {
  assert(min_capacity() <= v && v <= max_capacity(),
         "Should be in bounds: %zu <= %zu <= %zu",
         min_capacity(), v, max_capacity());
  Atomic::store(&_soft_max_size, v);
}

size_t ShenandoahHeap::min_capacity() const {
  return _minimum_size;
}

size_t ShenandoahHeap::initial_capacity() const {
  return _initial_size;
}

bool ShenandoahHeap::is_in(const void* p) const {
  if (!is_in_reserved(p)) {
    return false;
  }

  if (is_full_gc_move_in_progress()) {
    // Full GC move is running, we do not have a consistent region
    // information yet. But we know the pointer is in heap.
    return true;
  }

  // Now check if we point to a live section in active region.
  const ShenandoahHeapRegion* r = heap_region_containing(p);
  if (p >= r->top()) {
    return false;
  }

  if (r->is_active()) {
    return true;
  }

  // The region is trash, but won't be recycled until after concurrent weak
  // roots. We also don't allow mutators to allocate from trash regions
  // during weak roots. Concurrent class unloading may access unmarked oops
  // in trash regions.
  return r->is_trash() && is_concurrent_weak_root_in_progress();
}

void ShenandoahHeap::notify_soft_max_changed() {
  if (_uncommit_thread != nullptr) {
    _uncommit_thread->notify_soft_max_changed();
  }
}

void ShenandoahHeap::notify_explicit_gc_requested() {
  if (_uncommit_thread != nullptr) {
    _uncommit_thread->notify_explicit_gc_requested();
  }
}

bool ShenandoahHeap::check_soft_max_changed() {
  size_t new_soft_max = Atomic::load(&SoftMaxHeapSize);
  size_t old_soft_max = soft_max_capacity();
  if (new_soft_max != old_soft_max) {
    new_soft_max = MAX2(min_capacity(), new_soft_max);
    new_soft_max = MIN2(max_capacity(), new_soft_max);
    if (new_soft_max != old_soft_max) {
      log_info(gc)("Soft Max Heap Size: %zu%s -> %zu%s",
                   byte_size_in_proper_unit(old_soft_max), proper_unit_for_byte_size(old_soft_max),
                   byte_size_in_proper_unit(new_soft_max), proper_unit_for_byte_size(new_soft_max)
      );
      set_soft_max_capacity(new_soft_max);
      return true;
    }
  }
  return false;
}

void ShenandoahHeap::notify_heap_changed() {
  // Update monitoring counters when we took a new region. This amortizes the
  // update costs on slow path.
  monitoring_support()->notify_heap_changed();
  _heap_changed.try_set();
}

void ShenandoahHeap::set_forced_counters_update(bool value) {
  monitoring_support()->set_forced_counters_update(value);
}

void ShenandoahHeap::handle_force_counters_update() {
  monitoring_support()->handle_force_counters_update();
}

HeapWord* ShenandoahHeap::allocate_from_gclab_slow(Thread* thread, size_t size) {
  // New object should fit the GCLAB size
  size_t min_size = MAX2(size, PLAB::min_size());

  // Figure out size of new GCLAB, looking back at heuristics. Expand aggressively.
  size_t new_size = ShenandoahThreadLocalData::gclab_size(thread) * 2;

  new_size = MIN2(new_size, PLAB::max_size());
  new_size = MAX2(new_size, PLAB::min_size());

  // Record new heuristic value even if we take any shortcut. This captures
  // the case when moderately-sized objects always take a shortcut. At some point,
  // heuristics should catch up with them.
  log_debug(gc, free)("Set new GCLAB size: %zu", new_size);
  ShenandoahThreadLocalData::set_gclab_size(thread, new_size);

  if (new_size < size) {
    // New size still does not fit the object. Fall back to shared allocation.
    // This avoids retiring perfectly good GCLABs, when we encounter a large object.
    log_debug(gc, free)("New gclab size (%zu) is too small for %zu", new_size, size);
    return nullptr;
  }

  // Retire current GCLAB, and allocate a new one.
  PLAB* gclab = ShenandoahThreadLocalData::gclab(thread);
  gclab->retire();

  size_t actual_size = 0;
  HeapWord* gclab_buf = allocate_new_gclab(min_size, new_size, &actual_size);
  if (gclab_buf == nullptr) {
    return nullptr;
  }

  assert (size <= actual_size, "allocation should fit");

  // ...and clear or zap just allocated TLAB, if needed.
  if (ZeroTLAB) {
    Copy::zero_to_words(gclab_buf, actual_size);
  } else if (ZapTLAB) {
    // Skip mangling the space corresponding to the object header to
    // ensure that the returned space is not considered parsable by
    // any concurrent GC thread.
    size_t hdr_size = oopDesc::header_size();
    Copy::fill_to_words(gclab_buf + hdr_size, actual_size - hdr_size, badHeapWordVal);
  }
  gclab->set_buf(gclab_buf, actual_size);
  return gclab->allocate(size);
}

// Called from stubs in JIT code or interpreter
HeapWord* ShenandoahHeap::allocate_new_tlab(size_t min_size,
                                            size_t requested_size,
                                            size_t* actual_size) {
  ShenandoahAllocRequest req = ShenandoahAllocRequest::for_tlab(min_size, requested_size);
  HeapWord* res = allocate_memory(req);
  if (res != nullptr) {
    *actual_size = req.actual_size();
  } else {
    *actual_size = 0;
  }
  return res;
}

HeapWord* ShenandoahHeap::allocate_new_gclab(size_t min_size,
                                             size_t word_size,
                                             size_t* actual_size) {
  ShenandoahAllocRequest req = ShenandoahAllocRequest::for_gclab(min_size, word_size);
  HeapWord* res = allocate_memory(req);
  if (res != nullptr) {
    *actual_size = req.actual_size();
  } else {
    *actual_size = 0;
  }
  return res;
}

HeapWord* ShenandoahHeap::allocate_memory(ShenandoahAllocRequest& req) {
  intptr_t pacer_epoch = 0;
  bool in_new_region = false;
  HeapWord* result = nullptr;

  if (req.is_mutator_alloc()) {
    if (ShenandoahPacing) {
      pacer()->pace_for_alloc(req.size());
      pacer_epoch = pacer()->epoch();
    }

    if (!ShenandoahAllocFailureALot || !should_inject_alloc_failure()) {
      result = allocate_memory_under_lock(req, in_new_region);
    }

    // Check that gc overhead is not exceeded.
    //
    // Shenandoah will grind along for quite a while allocating one
    // object at a time using shared (non-tlab) allocations. This check
    // is testing that the GC overhead limit has not been exceeded.
    // This will notify the collector to start a cycle, but will raise
    // an OOME to the mutator if the last Full GCs have not made progress.
    // gc_no_progress_count is incremented following each degen or full GC that fails to achieve is_good_progress().
    if (result == nullptr && !req.is_lab_alloc() && get_gc_no_progress_count() > ShenandoahNoProgressThreshold) {
      control_thread()->handle_alloc_failure(req, false);
      req.set_actual_size(0);
      return nullptr;
    }

    if (result == nullptr) {
      // Block until control thread reacted, then retry allocation.
      //
      // It might happen that one of the threads requesting allocation would unblock
      // way later after GC happened, only to fail the second allocation, because
      // other threads have already depleted the free storage. In this case, a better
      // strategy is to try again, until at least one full GC has completed.
      //
      // Stop retrying and return nullptr to cause OOMError exception if our allocation failed even after:
      //   a) We experienced a GC that had good progress, or
      //   b) We experienced at least one Full GC (whether or not it had good progress)

      const size_t original_count = shenandoah_policy()->full_gc_count();
      while (result == nullptr && should_retry_allocation(original_count)) {
        control_thread()->handle_alloc_failure(req, true);
        result = allocate_memory_under_lock(req, in_new_region);
      }
      if (result != nullptr) {
        // If our allocation request has been satisfied after it initially failed, we count this as good gc progress
        notify_gc_progress();
      }
      if (log_develop_is_enabled(Debug, gc, alloc)) {
        ResourceMark rm;
        log_debug(gc, alloc)("Thread: %s, Result: " PTR_FORMAT ", Request: %s, Size: %zu"
                             ", Original: %zu, Latest: %zu",
                             Thread::current()->name(), p2i(result), req.type_string(), req.size(),
                             original_count, get_gc_no_progress_count());
      }
    }
  } else {
    assert(req.is_gc_alloc(), "Can only accept GC allocs here");
    result = allocate_memory_under_lock(req, in_new_region);
    // Do not call handle_alloc_failure() here, because we cannot block.
    // The allocation failure would be handled by the LRB slowpath with handle_alloc_failure_evac().
  }

  if (in_new_region) {
    notify_heap_changed();
  }

  if (result == nullptr) {
    req.set_actual_size(0);
  }

  // This is called regardless of the outcome of the allocation to account
  // for any waste created by retiring regions with this request.
  increase_used(req);

  if (result != nullptr) {
    size_t requested = req.size();
    size_t actual = req.actual_size();

    assert (req.is_lab_alloc() || (requested == actual),
            "Only LAB allocations are elastic: %s, requested = %zu, actual = %zu",
            ShenandoahAllocRequest::alloc_type_to_string(req.type()), requested, actual);

    if (req.is_mutator_alloc()) {
      // If we requested more than we were granted, give the rest back to pacer.
      // This only matters if we are in the same pacing epoch: do not try to unpace
      // over the budget for the other phase.
      if (ShenandoahPacing && (pacer_epoch > 0) && (requested > actual)) {
        pacer()->unpace_for_alloc(pacer_epoch, requested - actual);
      }
    }
  }

  return result;
}

inline bool ShenandoahHeap::should_retry_allocation(size_t original_full_gc_count) const {
  return shenandoah_policy()->full_gc_count() == original_full_gc_count
      && !shenandoah_policy()->is_at_shutdown();
}

HeapWord* ShenandoahHeap::allocate_memory_under_lock(ShenandoahAllocRequest& req, bool& in_new_region) {
  // If we are dealing with mutator allocation, then we may need to block for safepoint.
  // We cannot block for safepoint for GC allocations, because there is a high chance
  // we are already running at safepoint or from stack watermark machinery, and we cannot
  // block again.
  ShenandoahHeapLocker locker(lock(), req.is_mutator_alloc());

  // Make sure the old generation has room for either evacuations or promotions before trying to allocate.
  if (req.is_old() && !old_generation()->can_allocate(req)) {
    return nullptr;
  }

  // If TLAB request size is greater than available, allocate() will attempt to downsize request to fit within available
  // memory.
  HeapWord* result = _free_set->allocate(req, in_new_region);

  // Record the plab configuration for this result and register the object.
  if (result != nullptr && req.is_old()) {
    old_generation()->configure_plab_for_current_thread(req);
    if (req.type() == ShenandoahAllocRequest::_alloc_shared_gc) {
      // Register the newly allocated object while we're holding the global lock since there's no synchronization
      // built in to the implementation of register_object().  There are potential races when multiple independent
      // threads are allocating objects, some of which might span the same card region.  For example, consider
      // a card table's memory region within which three objects are being allocated by three different threads:
      //
      // objects being "concurrently" allocated:
      //    [-----a------][-----b-----][--------------c------------------]
      //            [---- card table memory range --------------]
      //
      // Before any objects are allocated, this card's memory range holds no objects.  Note that allocation of object a
      // wants to set the starts-object, first-start, and last-start attributes of the preceding card region.
      // Allocation of object b wants to set the starts-object, first-start, and last-start attributes of this card region.
      // Allocation of object c also wants to set the starts-object, first-start, and last-start attributes of this
      // card region.
      //
      // The thread allocating b and the thread allocating c can "race" in various ways, resulting in confusion, such as
      // last-start representing object b while first-start represents object c.  This is why we need to require all
      // register_object() invocations to be "mutually exclusive" with respect to each card's memory range.
      old_generation()->card_scan()->register_object(result);
    }
  }

  return result;
}

HeapWord* ShenandoahHeap::mem_allocate(size_t size,
                                        bool*  gc_overhead_limit_was_exceeded) {
  ShenandoahAllocRequest req = ShenandoahAllocRequest::for_shared(size);
  return allocate_memory(req);
}

MetaWord* ShenandoahHeap::satisfy_failed_metadata_allocation(ClassLoaderData* loader_data,
                                                             size_t size,
                                                             Metaspace::MetadataType mdtype) {
  MetaWord* result;

  // Inform metaspace OOM to GC heuristics if class unloading is possible.
  ShenandoahHeuristics* h = global_generation()->heuristics();
  if (h->can_unload_classes()) {
    h->record_metaspace_oom();
  }

  // Expand and retry allocation
  result = loader_data->metaspace_non_null()->expand_and_allocate(size, mdtype);
  if (result != nullptr) {
    return result;
  }

  // Start full GC
  collect(GCCause::_metadata_GC_clear_soft_refs);

  // Retry allocation
  result = loader_data->metaspace_non_null()->allocate(size, mdtype);
  if (result != nullptr) {
    return result;
  }

  // Expand and retry allocation
  result = loader_data->metaspace_non_null()->expand_and_allocate(size, mdtype);
  if (result != nullptr) {
    return result;
  }

  // Out of memory
  return nullptr;
}

class ShenandoahConcurrentEvacuateRegionObjectClosure : public ObjectClosure {
private:
  ShenandoahHeap* const _heap;
  Thread* const _thread;
public:
  ShenandoahConcurrentEvacuateRegionObjectClosure(ShenandoahHeap* heap) :
    _heap(heap), _thread(Thread::current()) {}

  void do_object(oop p) {
    shenandoah_assert_marked(nullptr, p);
    if (!p->is_forwarded()) {
      _heap->evacuate_object(p, _thread);
    }
  }
};

class ShenandoahEvacuationTask : public WorkerTask {
private:
  ShenandoahHeap* const _sh;
  ShenandoahCollectionSet* const _cs;
  bool _concurrent;
public:
  ShenandoahEvacuationTask(ShenandoahHeap* sh,
                           ShenandoahCollectionSet* cs,
                           bool concurrent) :
    WorkerTask("Shenandoah Evacuation"),
    _sh(sh),
    _cs(cs),
    _concurrent(concurrent)
  {}

  void work(uint worker_id) {
    if (_concurrent) {
      ShenandoahConcurrentWorkerSession worker_session(worker_id);
      ShenandoahSuspendibleThreadSetJoiner stsj;
      ShenandoahEvacOOMScope oom_evac_scope;
      do_work();
    } else {
      ShenandoahParallelWorkerSession worker_session(worker_id);
      ShenandoahEvacOOMScope oom_evac_scope;
      do_work();
    }
  }

private:
  void do_work() {
    ShenandoahConcurrentEvacuateRegionObjectClosure cl(_sh);
    ShenandoahHeapRegion* r;
    while ((r =_cs->claim_next()) != nullptr) {
      assert(r->has_live(), "Region %zu should have been reclaimed early", r->index());
      _sh->marked_object_iterate(r, &cl);

      if (ShenandoahPacing) {
        _sh->pacer()->report_evac(r->used() >> LogHeapWordSize);
      }

      if (_sh->check_cancelled_gc_and_yield(_concurrent)) {
        break;
      }
    }
  }
};

class ShenandoahRetireGCLABClosure : public ThreadClosure {
private:
  bool const _resize;
public:
  explicit ShenandoahRetireGCLABClosure(bool resize) : _resize(resize) {}
  void do_thread(Thread* thread) override {
    PLAB* gclab = ShenandoahThreadLocalData::gclab(thread);
    assert(gclab != nullptr, "GCLAB should be initialized for %s", thread->name());
    gclab->retire();
    if (_resize && ShenandoahThreadLocalData::gclab_size(thread) > 0) {
      ShenandoahThreadLocalData::set_gclab_size(thread, 0);
    }

    if (ShenandoahHeap::heap()->mode()->is_generational()) {
      PLAB* plab = ShenandoahThreadLocalData::plab(thread);
      assert(plab != nullptr, "PLAB should be initialized for %s", thread->name());

      // There are two reasons to retire all plabs between old-gen evacuation passes.
      //  1. We need to make the plab memory parsable by remembered-set scanning.
      //  2. We need to establish a trustworthy UpdateWaterMark value within each old-gen heap region
      ShenandoahGenerationalHeap::heap()->retire_plab(plab, thread);
      if (_resize && ShenandoahThreadLocalData::plab_size(thread) > 0) {
        ShenandoahThreadLocalData::set_plab_size(thread, 0);
      }
    }
  }
};

class ShenandoahGCStatePropagator : public HandshakeClosure {
public:
  explicit ShenandoahGCStatePropagator(char gc_state) :
    HandshakeClosure("Shenandoah GC State Change"),
    _gc_state(gc_state) {}

  void do_thread(Thread* thread) override {
    ShenandoahThreadLocalData::set_gc_state(thread, _gc_state);
  }
private:
  char _gc_state;
};

class ShenandoahPrepareForUpdateRefs : public HandshakeClosure {
public:
  explicit ShenandoahPrepareForUpdateRefs(char gc_state) :
    HandshakeClosure("Shenandoah Prepare for Update Refs"),
    _retire(ResizeTLAB), _propagator(gc_state) {}

  void do_thread(Thread* thread) override {
    _propagator.do_thread(thread);
    if (ShenandoahThreadLocalData::gclab(thread) != nullptr) {
      _retire.do_thread(thread);
    }
  }
private:
  ShenandoahRetireGCLABClosure _retire;
  ShenandoahGCStatePropagator _propagator;
};

void ShenandoahHeap::evacuate_collection_set(bool concurrent) {
  ShenandoahEvacuationTask task(this, _collection_set, concurrent);
  workers()->run_task(&task);
}

void ShenandoahHeap::concurrent_prepare_for_update_refs() {
  {
    // Java threads take this lock while they are being attached and added to the list of thread.
    // If another thread holds this lock before we update the gc state, it will receive a stale
    // gc state, but they will have been added to the list of java threads and so will be corrected
    // by the following handshake.
    MutexLocker lock(Threads_lock);

    // A cancellation at this point means the degenerated cycle must resume from update-refs.
    set_gc_state_concurrent(EVACUATION, false);
    set_gc_state_concurrent(WEAK_ROOTS, false);
    set_gc_state_concurrent(UPDATE_REFS, true);
  }

  // This will propagate the gc state and retire gclabs and plabs for threads that require it.
  ShenandoahPrepareForUpdateRefs prepare_for_update_refs(_gc_state.raw_value());

  // The handshake won't touch worker threads (or control thread, or VM thread), so do those separately.
  Threads::non_java_threads_do(&prepare_for_update_refs);

  // Now retire gclabs and plabs and propagate gc_state for mutator threads
  Handshake::execute(&prepare_for_update_refs);

  _update_refs_iterator.reset();
}

class ShenandoahCompositeHandshakeClosure : public HandshakeClosure {
  HandshakeClosure* _handshake_1;
  HandshakeClosure* _handshake_2;
  public:
    ShenandoahCompositeHandshakeClosure(HandshakeClosure* handshake_1, HandshakeClosure* handshake_2) :
      HandshakeClosure(handshake_2->name()),
      _handshake_1(handshake_1), _handshake_2(handshake_2) {}

  void do_thread(Thread* thread) override {
      _handshake_1->do_thread(thread);
      _handshake_2->do_thread(thread);
    }
};

void ShenandoahHeap::concurrent_final_roots(HandshakeClosure* handshake_closure) {
  {
    assert(!is_evacuation_in_progress(), "Should not evacuate for abbreviated or old cycles");
    MutexLocker lock(Threads_lock);
    set_gc_state_concurrent(WEAK_ROOTS, false);
  }

  ShenandoahGCStatePropagator propagator(_gc_state.raw_value());
  Threads::non_java_threads_do(&propagator);
  if (handshake_closure == nullptr) {
    Handshake::execute(&propagator);
  } else {
    ShenandoahCompositeHandshakeClosure composite(&propagator, handshake_closure);
    Handshake::execute(&composite);
  }
}

oop ShenandoahHeap::evacuate_object(oop p, Thread* thread) {
  assert(thread == Thread::current(), "Expected thread parameter to be current thread.");
  if (ShenandoahThreadLocalData::is_oom_during_evac(thread)) {
    // This thread went through the OOM during evac protocol. It is safe to return
    // the forward pointer. It must not attempt to evacuate any other objects.
    return ShenandoahBarrierSet::resolve_forwarded(p);
  }

  assert(ShenandoahThreadLocalData::is_evac_allowed(thread), "must be enclosed in oom-evac scope");

  ShenandoahHeapRegion* r = heap_region_containing(p);
  assert(!r->is_humongous(), "never evacuate humongous objects");

  ShenandoahAffiliation target_gen = r->affiliation();
  return try_evacuate_object(p, thread, r, target_gen);
}

oop ShenandoahHeap::try_evacuate_object(oop p, Thread* thread, ShenandoahHeapRegion* from_region,
                                               ShenandoahAffiliation target_gen) {
  assert(target_gen == YOUNG_GENERATION, "Only expect evacuations to young in this mode");
  assert(from_region->is_young(), "Only expect evacuations from young in this mode");
  bool alloc_from_lab = true;
  HeapWord* copy = nullptr;
  size_t size = ShenandoahForwarding::size(p);

#ifdef ASSERT
  if (ShenandoahOOMDuringEvacALot &&
      (os::random() & 1) == 0) { // Simulate OOM every ~2nd slow-path call
    copy = nullptr;
  } else {
#endif
    if (UseTLAB) {
      copy = allocate_from_gclab(thread, size);
    }
    if (copy == nullptr) {
      // If we failed to allocate in LAB, we'll try a shared allocation.
      ShenandoahAllocRequest req = ShenandoahAllocRequest::for_shared_gc(size, target_gen);
      copy = allocate_memory(req);
      alloc_from_lab = false;
    }
#ifdef ASSERT
  }
#endif

  if (copy == nullptr) {
    control_thread()->handle_alloc_failure_evac(size);

    _oom_evac_handler.handle_out_of_memory_during_evacuation();

    return ShenandoahBarrierSet::resolve_forwarded(p);
  }

  // Copy the object:
  Copy::aligned_disjoint_words(cast_from_oop<HeapWord*>(p), copy, size);

  // Try to install the new forwarding pointer.
  oop copy_val = cast_to_oop(copy);
  oop result = ShenandoahForwarding::try_update_forwardee(p, copy_val);
  if (result == copy_val) {
    // Successfully evacuated. Our copy is now the public one!
    ContinuationGCSupport::relativize_stack_chunk(copy_val);
    shenandoah_assert_correct(nullptr, copy_val);
    return copy_val;
  }  else {
    // Failed to evacuate. We need to deal with the object that is left behind. Since this
    // new allocation is certainly after TAMS, it will be considered live in the next cycle.
    // But if it happens to contain references to evacuated regions, those references would
    // not get updated for this stale copy during this cycle, and we will crash while scanning
    // it the next cycle.
    if (alloc_from_lab) {
      // For LAB allocations, it is enough to rollback the allocation ptr. Either the next
      // object will overwrite this stale copy, or the filler object on LAB retirement will
      // do this.
      ShenandoahThreadLocalData::gclab(thread)->undo_allocation(copy, size);
    } else {
      // For non-LAB allocations, we have no way to retract the allocation, and
      // have to explicitly overwrite the copy with the filler object. With that overwrite,
      // we have to keep the fwdptr initialized and pointing to our (stale) copy.
      assert(size >= ShenandoahHeap::min_fill_size(), "previously allocated object known to be larger than min_size");
      fill_with_object(copy, size);
      shenandoah_assert_correct(nullptr, copy_val);
      // For non-LAB allocations, the object has already been registered
    }
    shenandoah_assert_correct(nullptr, result);
    return result;
  }
}

void ShenandoahHeap::trash_cset_regions() {
  ShenandoahHeapLocker locker(lock());

  ShenandoahCollectionSet* set = collection_set();
  ShenandoahHeapRegion* r;
  set->clear_current_index();
  while ((r = set->next()) != nullptr) {
    r->make_trash();
  }
  collection_set()->clear();
}

void ShenandoahHeap::print_heap_regions_on(outputStream* st) const {
  st->print_cr("Heap Regions:");
  st->print_cr("Region state: EU=empty-uncommitted, EC=empty-committed, R=regular, H=humongous start, HP=pinned humongous start");
  st->print_cr("              HC=humongous continuation, CS=collection set, TR=trash, P=pinned, CSP=pinned collection set");
  st->print_cr("BTE=bottom/top/end, TAMS=top-at-mark-start");
  st->print_cr("UWM=update watermark, U=used");
  st->print_cr("T=TLAB allocs, G=GCLAB allocs");
  st->print_cr("S=shared allocs, L=live data");
  st->print_cr("CP=critical pins");

  for (size_t i = 0; i < num_regions(); i++) {
    get_region(i)->print_on(st);
  }
}

size_t ShenandoahHeap::trash_humongous_region_at(ShenandoahHeapRegion* start) {
  assert(start->is_humongous_start(), "reclaim regions starting with the first one");

  oop humongous_obj = cast_to_oop(start->bottom());
  size_t size = humongous_obj->size();
  size_t required_regions = ShenandoahHeapRegion::required_regions(size * HeapWordSize);
  size_t index = start->index() + required_regions - 1;

  assert(!start->has_live(), "liveness must be zero");

  for(size_t i = 0; i < required_regions; i++) {
    // Reclaim from tail. Otherwise, assertion fails when printing region to trace log,
    // as it expects that every region belongs to a humongous region starting with a humongous start region.
    ShenandoahHeapRegion* region = get_region(index --);

    assert(region->is_humongous(), "expect correct humongous start or continuation");
    assert(!region->is_cset(), "Humongous region should not be in collection set");

    region->make_trash_immediate();
  }
  return required_regions;
}

class ShenandoahCheckCleanGCLABClosure : public ThreadClosure {
public:
  ShenandoahCheckCleanGCLABClosure() {}
  void do_thread(Thread* thread) {
    PLAB* gclab = ShenandoahThreadLocalData::gclab(thread);
    assert(gclab != nullptr, "GCLAB should be initialized for %s", thread->name());
    assert(gclab->words_remaining() == 0, "GCLAB should not need retirement");

    if (ShenandoahHeap::heap()->mode()->is_generational()) {
      PLAB* plab = ShenandoahThreadLocalData::plab(thread);
      assert(plab != nullptr, "PLAB should be initialized for %s", thread->name());
      assert(plab->words_remaining() == 0, "PLAB should not need retirement");
    }
  }
};

void ShenandoahHeap::labs_make_parsable() {
  assert(UseTLAB, "Only call with UseTLAB");

  ShenandoahRetireGCLABClosure cl(false);

  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *t = jtiwh.next(); ) {
    ThreadLocalAllocBuffer& tlab = t->tlab();
    tlab.make_parsable();
    cl.do_thread(t);
  }

  workers()->threads_do(&cl);

  if (safepoint_workers() != nullptr) {
    safepoint_workers()->threads_do(&cl);
  }
}

void ShenandoahHeap::tlabs_retire(bool resize) {
  assert(UseTLAB, "Only call with UseTLAB");
  assert(!resize || ResizeTLAB, "Only call for resize when ResizeTLAB is enabled");

  ThreadLocalAllocStats stats;

  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *t = jtiwh.next(); ) {
    ThreadLocalAllocBuffer& tlab = t->tlab();
    tlab.retire(&stats);
    if (resize) {
      tlab.resize();
    }
  }

  stats.publish();

#ifdef ASSERT
  ShenandoahCheckCleanGCLABClosure cl;
  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *t = jtiwh.next(); ) {
    cl.do_thread(t);
  }
  workers()->threads_do(&cl);
#endif
}

void ShenandoahHeap::gclabs_retire(bool resize) {
  assert(UseTLAB, "Only call with UseTLAB");
  assert(!resize || ResizeTLAB, "Only call for resize when ResizeTLAB is enabled");

  ShenandoahRetireGCLABClosure cl(resize);
  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *t = jtiwh.next(); ) {
    cl.do_thread(t);
  }

  workers()->threads_do(&cl);

  if (safepoint_workers() != nullptr) {
    safepoint_workers()->threads_do(&cl);
  }
}

// Returns size in bytes
size_t ShenandoahHeap::unsafe_max_tlab_alloc(Thread *thread) const {
  // Return the max allowed size, and let the allocation path
  // figure out the safe size for current allocation.
  return ShenandoahHeapRegion::max_tlab_size_bytes();
}

size_t ShenandoahHeap::max_tlab_size() const {
  // Returns size in words
  return ShenandoahHeapRegion::max_tlab_size_words();
}

void ShenandoahHeap::collect_as_vm_thread(GCCause::Cause cause) {
  // These requests are ignored because we can't easily have Shenandoah jump into
  // a synchronous (degenerated or full) cycle while it is in the middle of a concurrent
  // cycle. We _could_ cancel the concurrent cycle and then try to run a cycle directly
  // on the VM thread, but this would confuse the control thread mightily and doesn't
  // seem worth the trouble. Instead, we will have the caller thread run (and wait for) a
  // concurrent cycle in the prologue of the heap inspect/dump operation. This is how
  // other concurrent collectors in the JVM handle this scenario as well.
  assert(Thread::current()->is_VM_thread(), "Should be the VM thread");
  guarantee(cause == GCCause::_heap_dump || cause == GCCause::_heap_inspection, "Invalid cause");
}

void ShenandoahHeap::collect(GCCause::Cause cause) {
  control_thread()->request_gc(cause);
}

void ShenandoahHeap::do_full_collection(bool clear_all_soft_refs) {
  //assert(false, "Shouldn't need to do full collections");
}

HeapWord* ShenandoahHeap::block_start(const void* addr) const {
  ShenandoahHeapRegion* r = heap_region_containing(addr);
  if (r != nullptr) {
    return r->block_start(addr);
  }
  return nullptr;
}

bool ShenandoahHeap::block_is_obj(const HeapWord* addr) const {
  ShenandoahHeapRegion* r = heap_region_containing(addr);
  return r->block_is_obj(addr);
}

bool ShenandoahHeap::print_location(outputStream* st, void* addr) const {
  return BlockLocationPrinter<ShenandoahHeap>::print_location(st, addr);
}

void ShenandoahHeap::prepare_for_verify() {
  if (SafepointSynchronize::is_at_safepoint() && UseTLAB) {
    labs_make_parsable();
  }
}

void ShenandoahHeap::gc_threads_do(ThreadClosure* tcl) const {
  if (_shenandoah_policy->is_at_shutdown()) {
    return;
  }

  if (_control_thread != nullptr) {
    tcl->do_thread(_control_thread);
  }

  if (_uncommit_thread != nullptr) {
    tcl->do_thread(_uncommit_thread);
  }

  workers()->threads_do(tcl);
  if (_safepoint_workers != nullptr) {
    _safepoint_workers->threads_do(tcl);
  }
}

void ShenandoahHeap::print_tracing_info() const {
  LogTarget(Info, gc, stats) lt;
  if (lt.is_enabled()) {
    ResourceMark rm;
    LogStream ls(lt);

    phase_timings()->print_global_on(&ls);

    ls.cr();
    ls.cr();

    shenandoah_policy()->print_gc_stats(&ls);

    ls.cr();
    ls.cr();
  }
}

void ShenandoahHeap::set_gc_generation(ShenandoahGeneration* generation) {
  shenandoah_assert_control_or_vm_thread_at_safepoint();
  _gc_generation = generation;
}

// Active generation may only be set by the VM thread at a safepoint.
void ShenandoahHeap::set_active_generation() {
  assert(Thread::current()->is_VM_thread(), "Only the VM Thread");
  assert(SafepointSynchronize::is_at_safepoint(), "Only at a safepoint!");
  assert(_gc_generation != nullptr, "Will set _active_generation to nullptr");
  _active_generation = _gc_generation;
}

void ShenandoahHeap::on_cycle_start(GCCause::Cause cause, ShenandoahGeneration* generation) {
  shenandoah_policy()->record_collection_cause(cause);

  const GCCause::Cause current = gc_cause();
  assert(current == GCCause::_no_gc, "Over-writing cause: %s, with: %s",
         GCCause::to_string(current), GCCause::to_string(cause));
  assert(_gc_generation == nullptr, "Over-writing _gc_generation");

  set_gc_cause(cause);
  set_gc_generation(generation);

  generation->heuristics()->record_cycle_start();
}

void ShenandoahHeap::on_cycle_end(ShenandoahGeneration* generation) {
  assert(gc_cause() != GCCause::_no_gc, "cause wasn't set");
  assert(_gc_generation != nullptr, "_gc_generation wasn't set");

  generation->heuristics()->record_cycle_end();
  if (mode()->is_generational() && generation->is_global()) {
    // If we just completed a GLOBAL GC, claim credit for completion of young-gen and old-gen GC as well
    young_generation()->heuristics()->record_cycle_end();
    old_generation()->heuristics()->record_cycle_end();
  }

  set_gc_generation(nullptr);
  set_gc_cause(GCCause::_no_gc);
}

void ShenandoahHeap::verify(VerifyOption vo) {
  if (ShenandoahSafepoint::is_at_shenandoah_safepoint()) {
    if (ShenandoahVerify) {
      verifier()->verify_generic(vo);
    } else {
      // TODO: Consider allocating verification bitmaps on demand,
      // and turn this on unconditionally.
    }
  }
}
size_t ShenandoahHeap::tlab_capacity(Thread *thr) const {
  return _free_set->capacity();
}

class ObjectIterateScanRootClosure : public BasicOopIterateClosure {
private:
  MarkBitMap* _bitmap;
  ShenandoahScanObjectStack* _oop_stack;
  ShenandoahHeap* const _heap;
  ShenandoahMarkingContext* const _marking_context;

  template <class T>
  void do_oop_work(T* p) {
    T o = RawAccess<>::oop_load(p);
    if (!CompressedOops::is_null(o)) {
      oop obj = CompressedOops::decode_not_null(o);
      if (_heap->is_concurrent_weak_root_in_progress() && !_marking_context->is_marked(obj)) {
        // There may be dead oops in weak roots in concurrent root phase, do not touch them.
        return;
      }
      obj = ShenandoahBarrierSet::barrier_set()->load_reference_barrier(obj);

      assert(oopDesc::is_oop(obj), "must be a valid oop");
      if (!_bitmap->is_marked(obj)) {
        _bitmap->mark(obj);
        _oop_stack->push(obj);
      }
    }
  }
public:
  ObjectIterateScanRootClosure(MarkBitMap* bitmap, ShenandoahScanObjectStack* oop_stack) :
    _bitmap(bitmap), _oop_stack(oop_stack), _heap(ShenandoahHeap::heap()),
    _marking_context(_heap->marking_context()) {}
  void do_oop(oop* p)       { do_oop_work(p); }
  void do_oop(narrowOop* p) { do_oop_work(p); }
};

/*
 * This is public API, used in preparation of object_iterate().
 * Since we don't do linear scan of heap in object_iterate() (see comment below), we don't
 * need to make the heap parsable. For Shenandoah-internal linear heap scans that we can
 * control, we call SH::tlabs_retire, SH::gclabs_retire.
 */
void ShenandoahHeap::ensure_parsability(bool retire_tlabs) {
  // No-op.
}

/*
 * Iterates objects in the heap. This is public API, used for, e.g., heap dumping.
 *
 * We cannot safely iterate objects by doing a linear scan at random points in time. Linear
 * scanning needs to deal with dead objects, which may have dead Klass* pointers (e.g.
 * calling oopDesc::size() would crash) or dangling reference fields (crashes) etc. Linear
 * scanning therefore depends on having a valid marking bitmap to support it. However, we only
 * have a valid marking bitmap after successful marking. In particular, we *don't* have a valid
 * marking bitmap during marking, after aborted marking or during/after cleanup (when we just
 * wiped the bitmap in preparation for next marking).
 *
 * For all those reasons, we implement object iteration as a single marking traversal, reporting
 * objects as we mark+traverse through the heap, starting from GC roots. JVMTI IterateThroughHeap
 * is allowed to report dead objects, but is not required to do so.
 */
void ShenandoahHeap::object_iterate(ObjectClosure* cl) {
  // Reset bitmap
  if (!prepare_aux_bitmap_for_iteration())
    return;

  ShenandoahScanObjectStack oop_stack;
  ObjectIterateScanRootClosure oops(&_aux_bit_map, &oop_stack);
  // Seed the stack with root scan
  scan_roots_for_iteration(&oop_stack, &oops);

  // Work through the oop stack to traverse heap
  while (! oop_stack.is_empty()) {
    oop obj = oop_stack.pop();
    assert(oopDesc::is_oop(obj), "must be a valid oop");
    cl->do_object(obj);
    obj->oop_iterate(&oops);
  }

  assert(oop_stack.is_empty(), "should be empty");
  // Reclaim bitmap
  reclaim_aux_bitmap_for_iteration();
}

bool ShenandoahHeap::prepare_aux_bitmap_for_iteration() {
  assert(SafepointSynchronize::is_at_safepoint(), "safe iteration is only available during safepoints");

  if (!_aux_bitmap_region_special && !os::commit_memory((char*)_aux_bitmap_region.start(), _aux_bitmap_region.byte_size(), false)) {
    log_warning(gc)("Could not commit native memory for auxiliary marking bitmap for heap iteration");
    return false;
  }
  // Reset bitmap
  _aux_bit_map.clear();
  return true;
}

void ShenandoahHeap::scan_roots_for_iteration(ShenandoahScanObjectStack* oop_stack, ObjectIterateScanRootClosure* oops) {
  // Process GC roots according to current GC cycle
  // This populates the work stack with initial objects
  // It is important to relinquish the associated locks before diving
  // into heap dumper
  uint n_workers = safepoint_workers() != nullptr ? safepoint_workers()->active_workers() : 1;
  ShenandoahHeapIterationRootScanner rp(n_workers);
  rp.roots_do(oops);
}

void ShenandoahHeap::reclaim_aux_bitmap_for_iteration() {
  if (!_aux_bitmap_region_special && !os::uncommit_memory((char*)_aux_bitmap_region.start(), _aux_bitmap_region.byte_size())) {
    log_warning(gc)("Could not uncommit native memory for auxiliary marking bitmap for heap iteration");
  }
}

// Closure for parallelly iterate objects
class ShenandoahObjectIterateParScanClosure : public BasicOopIterateClosure {
private:
  MarkBitMap* _bitmap;
  ShenandoahObjToScanQueue* _queue;
  ShenandoahHeap* const _heap;
  ShenandoahMarkingContext* const _marking_context;

  template <class T>
  void do_oop_work(T* p) {
    T o = RawAccess<>::oop_load(p);
    if (!CompressedOops::is_null(o)) {
      oop obj = CompressedOops::decode_not_null(o);
      if (_heap->is_concurrent_weak_root_in_progress() && !_marking_context->is_marked(obj)) {
        // There may be dead oops in weak roots in concurrent root phase, do not touch them.
        return;
      }
      obj = ShenandoahBarrierSet::barrier_set()->load_reference_barrier(obj);

      assert(oopDesc::is_oop(obj), "Must be a valid oop");
      if (_bitmap->par_mark(obj)) {
        _queue->push(ShenandoahMarkTask(obj));
      }
    }
  }
public:
  ShenandoahObjectIterateParScanClosure(MarkBitMap* bitmap, ShenandoahObjToScanQueue* q) :
    _bitmap(bitmap), _queue(q), _heap(ShenandoahHeap::heap()),
    _marking_context(_heap->marking_context()) {}
  void do_oop(oop* p)       { do_oop_work(p); }
  void do_oop(narrowOop* p) { do_oop_work(p); }
};

// Object iterator for parallel heap iteraion.
// The root scanning phase happenes in construction as a preparation of
// parallel marking queues.
// Every worker processes it's own marking queue. work-stealing is used
// to balance workload.
class ShenandoahParallelObjectIterator : public ParallelObjectIteratorImpl {
private:
  uint                         _num_workers;
  bool                         _init_ready;
  MarkBitMap*                  _aux_bit_map;
  ShenandoahHeap*              _heap;
  ShenandoahScanObjectStack    _roots_stack; // global roots stack
  ShenandoahObjToScanQueueSet* _task_queues;
public:
  ShenandoahParallelObjectIterator(uint num_workers, MarkBitMap* bitmap) :
        _num_workers(num_workers),
        _init_ready(false),
        _aux_bit_map(bitmap),
        _heap(ShenandoahHeap::heap()) {
    // Initialize bitmap
    _init_ready = _heap->prepare_aux_bitmap_for_iteration();
    if (!_init_ready) {
      return;
    }

    ObjectIterateScanRootClosure oops(_aux_bit_map, &_roots_stack);
    _heap->scan_roots_for_iteration(&_roots_stack, &oops);

    _init_ready = prepare_worker_queues();
  }

  ~ShenandoahParallelObjectIterator() {
    // Reclaim bitmap
    _heap->reclaim_aux_bitmap_for_iteration();
    // Reclaim queue for workers
    if (_task_queues!= nullptr) {
      for (uint i = 0; i < _num_workers; ++i) {
        ShenandoahObjToScanQueue* q = _task_queues->queue(i);
        if (q != nullptr) {
          delete q;
          _task_queues->register_queue(i, nullptr);
        }
      }
      delete _task_queues;
      _task_queues = nullptr;
    }
  }

  virtual void object_iterate(ObjectClosure* cl, uint worker_id) {
    if (_init_ready) {
      object_iterate_parallel(cl, worker_id, _task_queues);
    }
  }

private:
  // Divide global root_stack into worker queues
  bool prepare_worker_queues() {
    _task_queues = new ShenandoahObjToScanQueueSet((int) _num_workers);
    // Initialize queues for every workers
    for (uint i = 0; i < _num_workers; ++i) {
      ShenandoahObjToScanQueue* task_queue = new ShenandoahObjToScanQueue();
      _task_queues->register_queue(i, task_queue);
    }
    // Divide roots among the workers. Assume that object referencing distribution
    // is related with root kind, use round-robin to make every worker have same chance
    // to process every kind of roots
    size_t roots_num = _roots_stack.size();
    if (roots_num == 0) {
      // No work to do
      return false;
    }

    for (uint j = 0; j < roots_num; j++) {
      uint stack_id = j % _num_workers;
      oop obj = _roots_stack.pop();
      _task_queues->queue(stack_id)->push(ShenandoahMarkTask(obj));
    }
    return true;
  }

  void object_iterate_parallel(ObjectClosure* cl,
                               uint worker_id,
                               ShenandoahObjToScanQueueSet* queue_set) {
    assert(SafepointSynchronize::is_at_safepoint(), "safe iteration is only available during safepoints");
    assert(queue_set != nullptr, "task queue must not be null");

    ShenandoahObjToScanQueue* q = queue_set->queue(worker_id);
    assert(q != nullptr, "object iterate queue must not be null");

    ShenandoahMarkTask t;
    ShenandoahObjectIterateParScanClosure oops(_aux_bit_map, q);

    // Work through the queue to traverse heap.
    // Steal when there is no task in queue.
    while (q->pop(t) || queue_set->steal(worker_id, t)) {
      oop obj = t.obj();
      assert(oopDesc::is_oop(obj), "must be a valid oop");
      cl->do_object(obj);
      obj->oop_iterate(&oops);
    }
    assert(q->is_empty(), "should be empty");
  }
};

ParallelObjectIteratorImpl* ShenandoahHeap::parallel_object_iterator(uint workers) {
  return new ShenandoahParallelObjectIterator(workers, &_aux_bit_map);
}

// Keep alive an object that was loaded with AS_NO_KEEPALIVE.
void ShenandoahHeap::keep_alive(oop obj) {
  if (is_concurrent_mark_in_progress() && (obj != nullptr)) {
    ShenandoahBarrierSet::barrier_set()->enqueue(obj);
  }
}

void ShenandoahHeap::heap_region_iterate(ShenandoahHeapRegionClosure* blk) const {
  for (size_t i = 0; i < num_regions(); i++) {
    ShenandoahHeapRegion* current = get_region(i);
    blk->heap_region_do(current);
  }
}

class ShenandoahParallelHeapRegionTask : public WorkerTask {
private:
  ShenandoahHeap* const _heap;
  ShenandoahHeapRegionClosure* const _blk;
  size_t const _stride;

  shenandoah_padding(0);
  volatile size_t _index;
  shenandoah_padding(1);

public:
  ShenandoahParallelHeapRegionTask(ShenandoahHeapRegionClosure* blk, size_t stride) :
          WorkerTask("Shenandoah Parallel Region Operation"),
          _heap(ShenandoahHeap::heap()), _blk(blk), _stride(stride), _index(0) {}

  void work(uint worker_id) {
    ShenandoahParallelWorkerSession worker_session(worker_id);
    size_t stride = _stride;

    size_t max = _heap->num_regions();
    while (Atomic::load(&_index) < max) {
      size_t cur = Atomic::fetch_then_add(&_index, stride, memory_order_relaxed);
      size_t start = cur;
      size_t end = MIN2(cur + stride, max);
      if (start >= max) break;

      for (size_t i = cur; i < end; i++) {
        ShenandoahHeapRegion* current = _heap->get_region(i);
        _blk->heap_region_do(current);
      }
    }
  }
};

void ShenandoahHeap::parallel_heap_region_iterate(ShenandoahHeapRegionClosure* blk) const {
  assert(blk->is_thread_safe(), "Only thread-safe closures here");
  const uint active_workers = workers()->active_workers();
  const size_t n_regions = num_regions();
  size_t stride = ShenandoahParallelRegionStride;
  if (stride == 0 && active_workers > 1) {
    // Automatically derive the stride to balance the work between threads
    // evenly. Do not try to split work if below the reasonable threshold.
    constexpr size_t threshold = 4096;
    stride = n_regions <= threshold ?
            threshold :
            (n_regions + active_workers - 1) / active_workers;
  }

  if (n_regions > stride && active_workers > 1) {
    ShenandoahParallelHeapRegionTask task(blk, stride);
    workers()->run_task(&task);
  } else {
    heap_region_iterate(blk);
  }
}

class ShenandoahRendezvousClosure : public HandshakeClosure {
public:
  inline ShenandoahRendezvousClosure(const char* name) : HandshakeClosure(name) {}
  inline void do_thread(Thread* thread) {}
};

void ShenandoahHeap::rendezvous_threads(const char* name) {
  ShenandoahRendezvousClosure cl(name);
  Handshake::execute(&cl);
}

void ShenandoahHeap::recycle_trash() {
  free_set()->recycle_trash();
}

void ShenandoahHeap::do_class_unloading() {
  _unloader.unload();
  if (mode()->is_generational()) {
    old_generation()->set_parsable(false);
  }
}

void ShenandoahHeap::stw_weak_refs(bool full_gc) {
  // Weak refs processing
  ShenandoahPhaseTimings::Phase phase = full_gc ? ShenandoahPhaseTimings::full_gc_weakrefs
                                                : ShenandoahPhaseTimings::degen_gc_weakrefs;
  ShenandoahTimingsTracker t(phase);
  ShenandoahGCWorkerPhase worker_phase(phase);
  shenandoah_assert_generations_reconciled();
  gc_generation()->ref_processor()->process_references(phase, workers(), false /* concurrent */);
}

void ShenandoahHeap::prepare_update_heap_references() {
  assert(ShenandoahSafepoint::is_at_shenandoah_safepoint(), "must be at safepoint");

  // Evacuation is over, no GCLABs are needed anymore. GCLABs are under URWM, so we need to
  // make them parsable for update code to work correctly. Plus, we can compute new sizes
  // for future GCLABs here.
  if (UseTLAB) {
    ShenandoahGCPhase phase(ShenandoahPhaseTimings::degen_gc_init_update_refs_manage_gclabs);
    gclabs_retire(ResizeTLAB);
  }

  _update_refs_iterator.reset();
}

void ShenandoahHeap::propagate_gc_state_to_all_threads() {
  assert(ShenandoahSafepoint::is_at_shenandoah_safepoint(), "Must be at Shenandoah safepoint");
  if (_gc_state_changed) {
    ShenandoahGCStatePropagator propagator(_gc_state.raw_value());
    Threads::threads_do(&propagator);
    _gc_state_changed = false;
  }
}

void ShenandoahHeap::set_gc_state_at_safepoint(uint mask, bool value) {
  assert(ShenandoahSafepoint::is_at_shenandoah_safepoint(), "Must be at Shenandoah safepoint");
  _gc_state.set_cond(mask, value);
  _gc_state_changed = true;
}

void ShenandoahHeap::set_gc_state_concurrent(uint mask, bool value) {
  // Holding the thread lock here assures that any thread created after we change the gc
  // state will have the correct state. It also prevents attaching threads from seeing
  // an inconsistent state. See ShenandoahBarrierSet::on_thread_attach for reference. Established
  // threads will use their thread local copy of the gc state (changed by a handshake, or on a
  // safepoint).
  assert(Threads_lock->is_locked(), "Must hold thread lock for concurrent gc state change");
  _gc_state.set_cond(mask, value);
}

void ShenandoahHeap::set_concurrent_young_mark_in_progress(bool in_progress) {
  uint mask;
  assert(!has_forwarded_objects(), "Young marking is not concurrent with evacuation");
  if (!in_progress && is_concurrent_old_mark_in_progress()) {
    assert(mode()->is_generational(), "Only generational GC has old marking");
    assert(_gc_state.is_set(MARKING), "concurrent_old_marking_in_progress implies MARKING");
    // If old-marking is in progress when we turn off YOUNG_MARKING, leave MARKING (and OLD_MARKING) on
    mask = YOUNG_MARKING;
  } else {
    mask = MARKING | YOUNG_MARKING;
  }
  set_gc_state_at_safepoint(mask, in_progress);
  manage_satb_barrier(in_progress);
}

void ShenandoahHeap::set_concurrent_old_mark_in_progress(bool in_progress) {
#ifdef ASSERT
  // has_forwarded_objects() iff UPDATE_REFS or EVACUATION
  bool has_forwarded = has_forwarded_objects();
  bool updating_or_evacuating = _gc_state.is_set(UPDATE_REFS | EVACUATION);
  bool evacuating = _gc_state.is_set(EVACUATION);
  assert ((has_forwarded == updating_or_evacuating) || (evacuating && !has_forwarded && collection_set()->is_empty()),
          "Updating or evacuating iff has forwarded objects, or if evacuation phase is promoting in place without forwarding");
#endif
  if (!in_progress && is_concurrent_young_mark_in_progress()) {
    // If young-marking is in progress when we turn off OLD_MARKING, leave MARKING (and YOUNG_MARKING) on
    assert(_gc_state.is_set(MARKING), "concurrent_young_marking_in_progress implies MARKING");
    set_gc_state_at_safepoint(OLD_MARKING, in_progress);
  } else {
    set_gc_state_at_safepoint(MARKING | OLD_MARKING, in_progress);
  }
  manage_satb_barrier(in_progress);
}

bool ShenandoahHeap::is_prepare_for_old_mark_in_progress() const {
  return old_generation()->is_preparing_for_mark();
}

void ShenandoahHeap::manage_satb_barrier(bool active) {
  if (is_concurrent_mark_in_progress()) {
    // Ignore request to deactivate barrier while concurrent mark is in progress.
    // Do not attempt to re-activate the barrier if it is already active.
    if (active && !ShenandoahBarrierSet::satb_mark_queue_set().is_active()) {
      ShenandoahBarrierSet::satb_mark_queue_set().set_active_all_threads(active, !active);
    }
  } else {
    // No concurrent marking is in progress so honor request to deactivate,
    // but only if the barrier is already active.
    if (!active && ShenandoahBarrierSet::satb_mark_queue_set().is_active()) {
      ShenandoahBarrierSet::satb_mark_queue_set().set_active_all_threads(active, !active);
    }
  }
}

void ShenandoahHeap::set_evacuation_in_progress(bool in_progress) {
  assert(ShenandoahSafepoint::is_at_shenandoah_safepoint(), "Only call this at safepoint");
  set_gc_state_at_safepoint(EVACUATION, in_progress);
}

void ShenandoahHeap::set_concurrent_strong_root_in_progress(bool in_progress) {
  if (in_progress) {
    _concurrent_strong_root_in_progress.set();
  } else {
    _concurrent_strong_root_in_progress.unset();
  }
}

void ShenandoahHeap::set_concurrent_weak_root_in_progress(bool cond) {
  set_gc_state_at_safepoint(WEAK_ROOTS, cond);
}

GCTracer* ShenandoahHeap::tracer() {
  return shenandoah_policy()->tracer();
}

size_t ShenandoahHeap::tlab_used(Thread* thread) const {
  return _free_set->used();
}

bool ShenandoahHeap::try_cancel_gc(GCCause::Cause cause) {
  const GCCause::Cause prev = _cancelled_gc.xchg(cause);
  return prev == GCCause::_no_gc || prev == GCCause::_shenandoah_concurrent_gc;
}

void ShenandoahHeap::cancel_concurrent_mark() {
  if (mode()->is_generational()) {
    young_generation()->cancel_marking();
    old_generation()->cancel_marking();
  }

  global_generation()->cancel_marking();

  ShenandoahBarrierSet::satb_mark_queue_set().abandon_partial_marking();
}

bool ShenandoahHeap::cancel_gc(GCCause::Cause cause) {
  if (try_cancel_gc(cause)) {
    FormatBuffer<> msg("Cancelling GC: %s", GCCause::to_string(cause));
    log_info(gc,thread)("%s", msg.buffer());
    Events::log(Thread::current(), "%s", msg.buffer());
    _cancel_requested_time = os::elapsedTime();
    return true;
  }
  return false;
}

uint ShenandoahHeap::max_workers() {
  return _max_workers;
}

void ShenandoahHeap::stop() {
  // The shutdown sequence should be able to terminate when GC is running.

  // Step 0. Notify policy to disable event recording and prevent visiting gc threads during shutdown
  _shenandoah_policy->record_shutdown();

  // Step 1. Stop reporting on gc thread cpu utilization
  mmu_tracker()->stop();

  // Step 2. Wait until GC worker exits normally (this will cancel any ongoing GC).
  control_thread()->stop();

  // Stop 4. Shutdown uncommit thread.
  if (_uncommit_thread != nullptr) {
    _uncommit_thread->stop();
  }
}

void ShenandoahHeap::stw_unload_classes(bool full_gc) {
  if (!unload_classes()) return;
  ClassUnloadingContext ctx(_workers->active_workers(),
                            true /* unregister_nmethods_during_purge */,
                            false /* lock_nmethod_free_separately */);

  // Unload classes and purge SystemDictionary.
  {
    ShenandoahPhaseTimings::Phase phase = full_gc ?
                                          ShenandoahPhaseTimings::full_gc_purge_class_unload :
                                          ShenandoahPhaseTimings::degen_gc_purge_class_unload;
    ShenandoahIsAliveSelector is_alive;
    {
      CodeCache::UnlinkingScope scope(is_alive.is_alive_closure());
      ShenandoahGCPhase gc_phase(phase);
      ShenandoahGCWorkerPhase worker_phase(phase);
      bool unloading_occurred = SystemDictionary::do_unloading(gc_timer());

      // Clean JVMCI metadata handles.
      JVMCI_ONLY(JVMCI::do_unloading(unloading_occurred));

      uint num_workers = _workers->active_workers();
      ShenandoahClassUnloadingTask unlink_task(phase, num_workers, unloading_occurred);
      _workers->run_task(&unlink_task);
    }
    // Release unloaded nmethods's memory.
    ClassUnloadingContext::context()->purge_and_free_nmethods();
  }

  {
    ShenandoahGCPhase phase(full_gc ?
                            ShenandoahPhaseTimings::full_gc_purge_cldg :
                            ShenandoahPhaseTimings::degen_gc_purge_cldg);
    ClassLoaderDataGraph::purge(true /* at_safepoint */);
  }
  // Resize and verify metaspace
  MetaspaceGC::compute_new_size();
  DEBUG_ONLY(MetaspaceUtils::verify();)
}

// Weak roots are either pre-evacuated (final mark) or updated (final update refs),
// so they should not have forwarded oops.
// However, we do need to "null" dead oops in the roots, if can not be done
// in concurrent cycles.
void ShenandoahHeap::stw_process_weak_roots(bool full_gc) {
  uint num_workers = _workers->active_workers();
  ShenandoahPhaseTimings::Phase timing_phase = full_gc ?
                                               ShenandoahPhaseTimings::full_gc_purge_weak_par :
                                               ShenandoahPhaseTimings::degen_gc_purge_weak_par;
  ShenandoahGCPhase phase(timing_phase);
  ShenandoahGCWorkerPhase worker_phase(timing_phase);
  // Cleanup weak roots
  if (has_forwarded_objects()) {
    ShenandoahForwardedIsAliveClosure is_alive;
    ShenandoahNonConcUpdateRefsClosure keep_alive;
    ShenandoahParallelWeakRootsCleaningTask<ShenandoahForwardedIsAliveClosure, ShenandoahNonConcUpdateRefsClosure>
      cleaning_task(timing_phase, &is_alive, &keep_alive, num_workers);
    _workers->run_task(&cleaning_task);
  } else {
    ShenandoahIsAliveClosure is_alive;
#ifdef ASSERT
    ShenandoahAssertNotForwardedClosure verify_cl;
    ShenandoahParallelWeakRootsCleaningTask<ShenandoahIsAliveClosure, ShenandoahAssertNotForwardedClosure>
      cleaning_task(timing_phase, &is_alive, &verify_cl, num_workers);
#else
    ShenandoahParallelWeakRootsCleaningTask<ShenandoahIsAliveClosure, DoNothingClosure>
      cleaning_task(timing_phase, &is_alive, &do_nothing_cl, num_workers);
#endif
    _workers->run_task(&cleaning_task);
  }
}

void ShenandoahHeap::parallel_cleaning(bool full_gc) {
  assert(SafepointSynchronize::is_at_safepoint(), "Must be at a safepoint");
  assert(is_stw_gc_in_progress(), "Only for Degenerated and Full GC");
  ShenandoahGCPhase phase(full_gc ?
                          ShenandoahPhaseTimings::full_gc_purge :
                          ShenandoahPhaseTimings::degen_gc_purge);
  stw_weak_refs(full_gc);
  stw_process_weak_roots(full_gc);
  stw_unload_classes(full_gc);
}

void ShenandoahHeap::set_has_forwarded_objects(bool cond) {
  set_gc_state_at_safepoint(HAS_FORWARDED, cond);
}

void ShenandoahHeap::set_unload_classes(bool uc) {
  _unload_classes.set_cond(uc);
}

bool ShenandoahHeap::unload_classes() const {
  return _unload_classes.is_set();
}

address ShenandoahHeap::in_cset_fast_test_addr() {
  ShenandoahHeap* heap = ShenandoahHeap::heap();
  assert(heap->collection_set() != nullptr, "Sanity");
  return (address) heap->collection_set()->biased_map_address();
}

void ShenandoahHeap::reset_bytes_allocated_since_gc_start() {
  if (mode()->is_generational()) {
    young_generation()->reset_bytes_allocated_since_gc_start();
    old_generation()->reset_bytes_allocated_since_gc_start();
  }

  global_generation()->reset_bytes_allocated_since_gc_start();
}

void ShenandoahHeap::set_degenerated_gc_in_progress(bool in_progress) {
  _degenerated_gc_in_progress.set_cond(in_progress);
}

void ShenandoahHeap::set_full_gc_in_progress(bool in_progress) {
  _full_gc_in_progress.set_cond(in_progress);
}

void ShenandoahHeap::set_full_gc_move_in_progress(bool in_progress) {
  assert (is_full_gc_in_progress(), "should be");
  _full_gc_move_in_progress.set_cond(in_progress);
}

void ShenandoahHeap::set_update_refs_in_progress(bool in_progress) {
  set_gc_state_at_safepoint(UPDATE_REFS, in_progress);
}

void ShenandoahHeap::register_nmethod(nmethod* nm) {
  ShenandoahCodeRoots::register_nmethod(nm);
}

void ShenandoahHeap::unregister_nmethod(nmethod* nm) {
  ShenandoahCodeRoots::unregister_nmethod(nm);
}

void ShenandoahHeap::pin_object(JavaThread* thr, oop o) {
  heap_region_containing(o)->record_pin();
}

void ShenandoahHeap::unpin_object(JavaThread* thr, oop o) {
  ShenandoahHeapRegion* r = heap_region_containing(o);
  assert(r != nullptr, "Sanity");
  assert(r->pin_count() > 0, "Region %zu should have non-zero pins", r->index());
  r->record_unpin();
}

void ShenandoahHeap::sync_pinned_region_status() {
  ShenandoahHeapLocker locker(lock());

  for (size_t i = 0; i < num_regions(); i++) {
    ShenandoahHeapRegion *r = get_region(i);
    if (r->is_active()) {
      if (r->is_pinned()) {
        if (r->pin_count() == 0) {
          r->make_unpinned();
        }
      } else {
        if (r->pin_count() > 0) {
          r->make_pinned();
        }
      }
    }
  }

  assert_pinned_region_status();
}

#ifdef ASSERT
void ShenandoahHeap::assert_pinned_region_status() {
  for (size_t i = 0; i < num_regions(); i++) {
    ShenandoahHeapRegion* r = get_region(i);
    shenandoah_assert_generations_reconciled();
    if (gc_generation()->contains(r)) {
      assert((r->is_pinned() && r->pin_count() > 0) || (!r->is_pinned() && r->pin_count() == 0),
             "Region %zu pinning status is inconsistent", i);
    }
  }
}
#endif

ConcurrentGCTimer* ShenandoahHeap::gc_timer() const {
  return _gc_timer;
}

void ShenandoahHeap::prepare_concurrent_roots() {
  assert(SafepointSynchronize::is_at_safepoint(), "Must be at a safepoint");
  assert(!is_stw_gc_in_progress(), "Only concurrent GC");
  set_concurrent_strong_root_in_progress(!collection_set()->is_empty());
  set_concurrent_weak_root_in_progress(true);
  if (unload_classes()) {
    _unloader.prepare();
  }
}

void ShenandoahHeap::finish_concurrent_roots() {
  assert(SafepointSynchronize::is_at_safepoint(), "Must be at a safepoint");
  assert(!is_stw_gc_in_progress(), "Only concurrent GC");
  if (unload_classes()) {
    _unloader.finish();
  }
}

#ifdef ASSERT
void ShenandoahHeap::assert_gc_workers(uint nworkers) {
  assert(nworkers > 0 && nworkers <= max_workers(), "Sanity");

  if (ShenandoahSafepoint::is_at_shenandoah_safepoint()) {
    // Use ParallelGCThreads inside safepoints
    assert(nworkers == ParallelGCThreads, "Use ParallelGCThreads (%u) within safepoint, not %u",
           ParallelGCThreads, nworkers);
  } else {
    // Use ConcGCThreads outside safepoints
    assert(nworkers == ConcGCThreads, "Use ConcGCThreads (%u) outside safepoints, %u",
           ConcGCThreads, nworkers);
  }
}
#endif

ShenandoahVerifier* ShenandoahHeap::verifier() {
  guarantee(ShenandoahVerify, "Should be enabled");
  assert (_verifier != nullptr, "sanity");
  return _verifier;
}

template<bool CONCURRENT>
class ShenandoahUpdateHeapRefsTask : public WorkerTask {
private:
  ShenandoahHeap* _heap;
  ShenandoahRegionIterator* _regions;
public:
  explicit ShenandoahUpdateHeapRefsTask(ShenandoahRegionIterator* regions) :
    WorkerTask("Shenandoah Update References"),
    _heap(ShenandoahHeap::heap()),
    _regions(regions) {
  }

  void work(uint worker_id) {
    if (CONCURRENT) {
      ShenandoahConcurrentWorkerSession worker_session(worker_id);
      ShenandoahSuspendibleThreadSetJoiner stsj;
      do_work<ShenandoahConcUpdateRefsClosure>(worker_id);
    } else {
      ShenandoahParallelWorkerSession worker_session(worker_id);
      do_work<ShenandoahNonConcUpdateRefsClosure>(worker_id);
    }
  }

private:
  template<class T>
  void do_work(uint worker_id) {
    if (CONCURRENT && (worker_id == 0)) {
      // We ask the first worker to replenish the Mutator free set by moving regions previously reserved to hold the
      // results of evacuation.  These reserves are no longer necessary because evacuation has completed.
      size_t cset_regions = _heap->collection_set()->count();

      // Now that evacuation is done, we can reassign any regions that had been reserved to hold the results of evacuation
      // to the mutator free set.  At the end of GC, we will have cset_regions newly evacuated fully empty regions from
      // which we will be able to replenish the Collector free set and the OldCollector free set in preparation for the
      // next GC cycle.
      _heap->free_set()->move_regions_from_collector_to_mutator(cset_regions);
    }
    // If !CONCURRENT, there's no value in expanding Mutator free set
    T cl;
    ShenandoahHeapRegion* r = _regions->next();
    while (r != nullptr) {
      HeapWord* update_watermark = r->get_update_watermark();
      assert (update_watermark >= r->bottom(), "sanity");
      if (r->is_active() && !r->is_cset()) {
        _heap->marked_object_oop_iterate(r, &cl, update_watermark);
        if (ShenandoahPacing) {
          _heap->pacer()->report_update_refs(pointer_delta(update_watermark, r->bottom()));
        }
      }
      if (_heap->check_cancelled_gc_and_yield(CONCURRENT)) {
        return;
      }
      r = _regions->next();
    }
  }
};

void ShenandoahHeap::update_heap_references(bool concurrent) {
  assert(!is_full_gc_in_progress(), "Only for concurrent and degenerated GC");

  if (concurrent) {
    ShenandoahUpdateHeapRefsTask<true> task(&_update_refs_iterator);
    workers()->run_task(&task);
  } else {
    ShenandoahUpdateHeapRefsTask<false> task(&_update_refs_iterator);
    workers()->run_task(&task);
  }
}

void ShenandoahHeap::update_heap_region_states(bool concurrent) {
  assert(SafepointSynchronize::is_at_safepoint(), "Must be at a safepoint");
  assert(!is_full_gc_in_progress(), "Only for concurrent and degenerated GC");

  {
    ShenandoahGCPhase phase(concurrent ?
                            ShenandoahPhaseTimings::final_update_refs_update_region_states :
                            ShenandoahPhaseTimings::degen_gc_final_update_refs_update_region_states);

    final_update_refs_update_region_states();

    assert_pinned_region_status();
  }

  {
    ShenandoahGCPhase phase(concurrent ?
                            ShenandoahPhaseTimings::final_update_refs_trash_cset :
                            ShenandoahPhaseTimings::degen_gc_final_update_refs_trash_cset);
    trash_cset_regions();
  }
}

void ShenandoahHeap::final_update_refs_update_region_states() {
  ShenandoahSynchronizePinnedRegionStates cl;
  parallel_heap_region_iterate(&cl);
}

void ShenandoahHeap::rebuild_free_set(bool concurrent) {
  ShenandoahGCPhase phase(concurrent ?
                          ShenandoahPhaseTimings::final_update_refs_rebuild_freeset :
                          ShenandoahPhaseTimings::degen_gc_final_update_refs_rebuild_freeset);
  ShenandoahHeapLocker locker(lock());
  size_t young_cset_regions, old_cset_regions;
  size_t first_old_region, last_old_region, old_region_count;
  _free_set->prepare_to_rebuild(young_cset_regions, old_cset_regions, first_old_region, last_old_region, old_region_count);
  // If there are no old regions, first_old_region will be greater than last_old_region
  assert((first_old_region > last_old_region) ||
         ((last_old_region + 1 - first_old_region >= old_region_count) &&
          get_region(first_old_region)->is_old() && get_region(last_old_region)->is_old()),
         "sanity: old_region_count: %zu, first_old_region: %zu, last_old_region: %zu",
         old_region_count, first_old_region, last_old_region);

  if (mode()->is_generational()) {
#ifdef ASSERT
    if (ShenandoahVerify) {
      verifier()->verify_before_rebuilding_free_set();
    }
#endif

    // The computation of bytes_of_allocation_runway_before_gc_trigger is quite conservative so consider all of this
    // available for transfer to old. Note that transfer of humongous regions does not impact available.
    ShenandoahGenerationalHeap* gen_heap = ShenandoahGenerationalHeap::heap();
    size_t allocation_runway = gen_heap->young_generation()->heuristics()->bytes_of_allocation_runway_before_gc_trigger(young_cset_regions);
    gen_heap->compute_old_generation_balance(allocation_runway, old_cset_regions);

    // Total old_available may have been expanded to hold anticipated promotions.  We trigger if the fragmented available
    // memory represents more than 16 regions worth of data.  Note that fragmentation may increase when we promote regular
    // regions in place when many of these regular regions have an abundant amount of available memory within them.  Fragmentation
    // will decrease as promote-by-copy consumes the available memory within these partially consumed regions.
    //
    // We consider old-gen to have excessive fragmentation if more than 12.5% of old-gen is free memory that resides
    // within partially consumed regions of memory.
  }
  // Rebuild free set based on adjusted generation sizes.
  _free_set->finish_rebuild(young_cset_regions, old_cset_regions, old_region_count);

  if (mode()->is_generational()) {
    ShenandoahGenerationalHeap* gen_heap = ShenandoahGenerationalHeap::heap();
    ShenandoahOldGeneration* old_gen = gen_heap->old_generation();
    old_gen->heuristics()->evaluate_triggers(first_old_region, last_old_region, old_region_count, num_regions());
  }
}

bool ShenandoahHeap::is_bitmap_slice_committed(ShenandoahHeapRegion* r, bool skip_self) {
  size_t slice = r->index() / _bitmap_regions_per_slice;

  size_t regions_from = _bitmap_regions_per_slice * slice;
  size_t regions_to   = MIN2(num_regions(), _bitmap_regions_per_slice * (slice + 1));
  for (size_t g = regions_from; g < regions_to; g++) {
    assert (g / _bitmap_regions_per_slice == slice, "same slice");
    if (skip_self && g == r->index()) continue;
    if (get_region(g)->is_committed()) {
      return true;
    }
  }
  return false;
}

bool ShenandoahHeap::commit_bitmap_slice(ShenandoahHeapRegion* r) {
  shenandoah_assert_heaplocked();

  // Bitmaps in special regions do not need commits
  if (_bitmap_region_special) {
    return true;
  }

  if (is_bitmap_slice_committed(r, true)) {
    // Some other region from the group is already committed, meaning the bitmap
    // slice is already committed, we exit right away.
    return true;
  }

  // Commit the bitmap slice:
  size_t slice = r->index() / _bitmap_regions_per_slice;
  size_t off = _bitmap_bytes_per_slice * slice;
  size_t len = _bitmap_bytes_per_slice;
  char* start = (char*) _bitmap_region.start() + off;

  if (!os::commit_memory(start, len, false)) {
    return false;
  }

  if (AlwaysPreTouch) {
    os::pretouch_memory(start, start + len, _pretouch_bitmap_page_size);
  }

  return true;
}

bool ShenandoahHeap::uncommit_bitmap_slice(ShenandoahHeapRegion *r) {
  shenandoah_assert_heaplocked();

  // Bitmaps in special regions do not need uncommits
  if (_bitmap_region_special) {
    return true;
  }

  if (is_bitmap_slice_committed(r, true)) {
    // Some other region from the group is still committed, meaning the bitmap
    // slice should stay committed, exit right away.
    return true;
  }

  // Uncommit the bitmap slice:
  size_t slice = r->index() / _bitmap_regions_per_slice;
  size_t off = _bitmap_bytes_per_slice * slice;
  size_t len = _bitmap_bytes_per_slice;
  if (!os::uncommit_memory((char*)_bitmap_region.start() + off, len)) {
    return false;
  }
  return true;
}

void ShenandoahHeap::forbid_uncommit() {
  if (_uncommit_thread != nullptr) {
    _uncommit_thread->forbid_uncommit();
  }
}

void ShenandoahHeap::allow_uncommit() {
  if (_uncommit_thread != nullptr) {
    _uncommit_thread->allow_uncommit();
  }
}

#ifdef ASSERT
bool ShenandoahHeap::is_uncommit_in_progress() {
  if (_uncommit_thread != nullptr) {
    return _uncommit_thread->is_uncommit_in_progress();
  }
  return false;
}
#endif

void ShenandoahHeap::safepoint_synchronize_begin() {
  StackWatermarkSet::safepoint_synchronize_begin();
  SuspendibleThreadSet::synchronize();
}

void ShenandoahHeap::safepoint_synchronize_end() {
  SuspendibleThreadSet::desynchronize();
}

void ShenandoahHeap::try_inject_alloc_failure() {
  if (ShenandoahAllocFailureALot && !cancelled_gc() && ((os::random() % 1000) > 950)) {
    _inject_alloc_failure.set();
    os::naked_short_sleep(1);
    if (cancelled_gc()) {
      log_info(gc)("Allocation failure was successfully injected");
    }
  }
}

bool ShenandoahHeap::should_inject_alloc_failure() {
  return _inject_alloc_failure.is_set() && _inject_alloc_failure.try_unset();
}

void ShenandoahHeap::initialize_serviceability() {
  _memory_pool = new ShenandoahMemoryPool(this);
  _cycle_memory_manager.add_pool(_memory_pool);
  _stw_memory_manager.add_pool(_memory_pool);
}

GrowableArray<GCMemoryManager*> ShenandoahHeap::memory_managers() {
  GrowableArray<GCMemoryManager*> memory_managers(2);
  memory_managers.append(&_cycle_memory_manager);
  memory_managers.append(&_stw_memory_manager);
  return memory_managers;
}

GrowableArray<MemoryPool*> ShenandoahHeap::memory_pools() {
  GrowableArray<MemoryPool*> memory_pools(1);
  memory_pools.append(_memory_pool);
  return memory_pools;
}

MemoryUsage ShenandoahHeap::memory_usage() {
  return MemoryUsage(_initial_size, used(), committed(), max_capacity());
}

ShenandoahRegionIterator::ShenandoahRegionIterator() :
  _heap(ShenandoahHeap::heap()),
  _index(0) {}

ShenandoahRegionIterator::ShenandoahRegionIterator(ShenandoahHeap* heap) :
  _heap(heap),
  _index(0) {}

void ShenandoahRegionIterator::reset() {
  _index = 0;
}

bool ShenandoahRegionIterator::has_next() const {
  return _index < _heap->num_regions();
}

char ShenandoahHeap::gc_state() const {
  return _gc_state.raw_value();
}

bool ShenandoahHeap::is_gc_state(GCState state) const {
  // If the global gc state has been changed, but hasn't yet been propagated to all threads, then
  // the global gc state is the correct value. Once the gc state has been synchronized with all threads,
  // _gc_state_changed will be toggled to false and we need to use the thread local state.
  return _gc_state_changed ? _gc_state.is_set(state) : ShenandoahThreadLocalData::is_gc_state(state);
}


ShenandoahLiveData* ShenandoahHeap::get_liveness_cache(uint worker_id) {
#ifdef ASSERT
  assert(_liveness_cache != nullptr, "sanity");
  assert(worker_id < _max_workers, "sanity");
  for (uint i = 0; i < num_regions(); i++) {
    assert(_liveness_cache[worker_id][i] == 0, "liveness cache should be empty");
  }
#endif
  return _liveness_cache[worker_id];
}

void ShenandoahHeap::flush_liveness_cache(uint worker_id) {
  assert(worker_id < _max_workers, "sanity");
  assert(_liveness_cache != nullptr, "sanity");
  ShenandoahLiveData* ld = _liveness_cache[worker_id];
  for (uint i = 0; i < num_regions(); i++) {
    ShenandoahLiveData live = ld[i];
    if (live > 0) {
      ShenandoahHeapRegion* r = get_region(i);
      r->increase_live_data_gc_words(live);
      ld[i] = 0;
    }
  }
}

bool ShenandoahHeap::requires_barriers(stackChunkOop obj) const {
  if (is_idle()) return false;

  // Objects allocated after marking start are implicitly alive, don't need any barriers during
  // marking phase.
  if (is_concurrent_mark_in_progress() &&
     !marking_context()->allocated_after_mark_start(obj)) {
    return true;
  }

  // Can not guarantee obj is deeply good.
  if (has_forwarded_objects()) {
    return true;
  }

  return false;
}

HeapWord* ShenandoahHeap::allocate_loaded_archive_space(size_t size) {
#if INCLUDE_CDS_JAVA_HEAP
  // CDS wants a continuous memory range to load a bunch of objects.
  // This effectively bypasses normal allocation paths, and requires
  // a bit of massaging to unbreak GC invariants.

  ShenandoahAllocRequest req = ShenandoahAllocRequest::for_shared(size);

  // Easy case: a single regular region, no further adjustments needed.
  if (!ShenandoahHeapRegion::requires_humongous(size)) {
    return allocate_memory(req);
  }

  // Hard case: the requested size would cause a humongous allocation.
  // We need to make sure it looks like regular allocation to the rest of GC.

  // CDS code would guarantee no objects straddle multiple regions, as long as
  // regions are as large as MIN_GC_REGION_ALIGNMENT. It is impractical at this
  // point to deal with case when Shenandoah runs with smaller regions.
  // TODO: This check can be dropped once MIN_GC_REGION_ALIGNMENT agrees more with Shenandoah.
  if (ShenandoahHeapRegion::region_size_bytes() < ArchiveHeapWriter::MIN_GC_REGION_ALIGNMENT) {
    return nullptr;
  }

  HeapWord* mem = allocate_memory(req);
  size_t start_idx = heap_region_index_containing(mem);
  size_t num_regions = ShenandoahHeapRegion::required_regions(size * HeapWordSize);

  // Flip humongous -> regular.
  {
    ShenandoahHeapLocker locker(lock(), false);
    for (size_t c = start_idx; c < start_idx + num_regions; c++) {
      get_region(c)->make_regular_bypass();
    }
  }

  return mem;
#else
  assert(false, "Archive heap loader should not be available, should not be here");
  return nullptr;
#endif // INCLUDE_CDS_JAVA_HEAP
}

void ShenandoahHeap::complete_loaded_archive_space(MemRegion archive_space) {
  // Nothing to do here, except checking that heap looks fine.
#ifdef ASSERT
  HeapWord* start = archive_space.start();
  HeapWord* end = archive_space.end();

  // No unclaimed space between the objects.
  // Objects are properly allocated in correct regions.
  HeapWord* cur = start;
  while (cur < end) {
    oop oop = cast_to_oop(cur);
    shenandoah_assert_in_correct_region(nullptr, oop);
    cur += oop->size();
  }

  // No unclaimed tail at the end of archive space.
  assert(cur == end,
         "Archive space should be fully used: " PTR_FORMAT " " PTR_FORMAT,
         p2i(cur), p2i(end));

  // Region bounds are good.
  ShenandoahHeapRegion* begin_reg = heap_region_containing(start);
  ShenandoahHeapRegion* end_reg = heap_region_containing(end);
  assert(begin_reg->is_regular(), "Must be");
  assert(end_reg->is_regular(), "Must be");
  assert(begin_reg->bottom() == start,
         "Must agree: archive-space-start: " PTR_FORMAT ", begin-region-bottom: " PTR_FORMAT,
         p2i(start), p2i(begin_reg->bottom()));
  assert(end_reg->top() == end,
         "Must agree: archive-space-end: " PTR_FORMAT ", end-region-top: " PTR_FORMAT,
         p2i(end), p2i(end_reg->top()));
#endif
}

ShenandoahGeneration* ShenandoahHeap::generation_for(ShenandoahAffiliation affiliation) const {
  if (!mode()->is_generational()) {
    return global_generation();
  } else if (affiliation == YOUNG_GENERATION) {
    return young_generation();
  } else if (affiliation == OLD_GENERATION) {
    return old_generation();
  }

  ShouldNotReachHere();
  return nullptr;
}

void ShenandoahHeap::log_heap_status(const char* msg) const {
  if (mode()->is_generational()) {
    young_generation()->log_status(msg);
    old_generation()->log_status(msg);
  } else {
    global_generation()->log_status(msg);
  }
}