< prev index next >

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

Print this page
@@ -1,7 +1,8 @@
  /*
   * Copyright (c) 2014, 2021, 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.

@@ -31,28 +32,32 @@
  #include "gc/shared/tlab_globals.hpp"
  #include "gc/shared/workerThread.hpp"
  #include "gc/shenandoah/heuristics/shenandoahHeuristics.hpp"
  #include "gc/shenandoah/shenandoahConcurrentGC.hpp"
  #include "gc/shenandoah/shenandoahCollectionSet.hpp"
+ #include "gc/shenandoah/shenandoahCollectorPolicy.hpp"
  #include "gc/shenandoah/shenandoahFreeSet.hpp"
  #include "gc/shenandoah/shenandoahFullGC.hpp"
+ #include "gc/shenandoah/shenandoahGlobalGeneration.hpp"
  #include "gc/shenandoah/shenandoahPhaseTimings.hpp"
  #include "gc/shenandoah/shenandoahMark.inline.hpp"
  #include "gc/shenandoah/shenandoahMonitoringSupport.hpp"
  #include "gc/shenandoah/shenandoahHeapRegionSet.hpp"
  #include "gc/shenandoah/shenandoahHeap.inline.hpp"
  #include "gc/shenandoah/shenandoahHeapRegion.inline.hpp"
  #include "gc/shenandoah/shenandoahMarkingContext.inline.hpp"
  #include "gc/shenandoah/shenandoahMetrics.hpp"
+ #include "gc/shenandoah/shenandoahOldGeneration.hpp"
  #include "gc/shenandoah/shenandoahOopClosures.inline.hpp"
  #include "gc/shenandoah/shenandoahReferenceProcessor.hpp"
  #include "gc/shenandoah/shenandoahRootProcessor.inline.hpp"
  #include "gc/shenandoah/shenandoahSTWMark.hpp"
  #include "gc/shenandoah/shenandoahUtils.hpp"
  #include "gc/shenandoah/shenandoahVerifier.hpp"
  #include "gc/shenandoah/shenandoahVMOperations.hpp"
  #include "gc/shenandoah/shenandoahWorkerPolicy.hpp"
+ #include "gc/shenandoah/shenandoahYoungGeneration.hpp"
  #include "memory/metaspaceUtils.hpp"
  #include "memory/universe.hpp"
  #include "oops/compressedOops.inline.hpp"
  #include "oops/oop.inline.hpp"
  #include "runtime/javaThread.hpp"

@@ -60,10 +65,73 @@
  #include "runtime/vmThread.hpp"
  #include "utilities/copy.hpp"
  #include "utilities/events.hpp"
  #include "utilities/growableArray.hpp"
  
+ // After Full GC is done, reconstruct the remembered set by iterating over OLD regions,
+ // registering all objects between bottom() and top(), and setting remembered set cards to
+ // DIRTY if they hold interesting pointers.
+ class ShenandoahReconstructRememberedSetTask : public WorkerTask {
+ private:
+   ShenandoahRegionIterator _regions;
+ 
+ public:
+   ShenandoahReconstructRememberedSetTask() :
+     WorkerTask("Shenandoah Reset Bitmap") { }
+ 
+   void work(uint worker_id) {
+     ShenandoahParallelWorkerSession worker_session(worker_id);
+     ShenandoahHeapRegion* r = _regions.next();
+     ShenandoahHeap* heap = ShenandoahHeap::heap();
+     RememberedScanner* scanner = heap->card_scan();
+     ShenandoahSetRememberedCardsToDirtyClosure dirty_cards_for_interesting_pointers;
+ 
+     while (r != nullptr) {
+       if (r->is_old() && r->is_active()) {
+         HeapWord* obj_addr = r->bottom();
+         if (r->is_humongous_start()) {
+           // First, clear the remembered set
+           oop obj = cast_to_oop(obj_addr);
+           size_t size = obj->size();
+ 
+           // First, clear the remembered set for all spanned humongous regions
+           size_t num_regions = ShenandoahHeapRegion::required_regions(size * HeapWordSize);
+           size_t region_span = num_regions * ShenandoahHeapRegion::region_size_words();
+           scanner->reset_remset(r->bottom(), region_span);
+           size_t region_index = r->index();
+           ShenandoahHeapRegion* humongous_region = heap->get_region(region_index);
+           while (num_regions-- != 0) {
+             scanner->reset_object_range(humongous_region->bottom(), humongous_region->end());
+             region_index++;
+             humongous_region = heap->get_region(region_index);
+           }
+ 
+           // Then register the humongous object and DIRTY relevant remembered set cards
+           scanner->register_object_without_lock(obj_addr);
+           obj->oop_iterate(&dirty_cards_for_interesting_pointers);
+         } else if (!r->is_humongous()) {
+           // First, clear the remembered set
+           scanner->reset_remset(r->bottom(), ShenandoahHeapRegion::region_size_words());
+           scanner->reset_object_range(r->bottom(), r->end());
+ 
+           // Then iterate over all objects, registering object and DIRTYing relevant remembered set cards
+           HeapWord* t = r->top();
+           while (obj_addr < t) {
+             oop obj = cast_to_oop(obj_addr);
+             size_t size = obj->size();
+             scanner->register_object_without_lock(obj_addr);
+             obj_addr += obj->oop_iterate_size(&dirty_cards_for_interesting_pointers);
+           }
+         } // else, ignore humongous continuation region
+       }
+       // else, this region is FREE or YOUNG or inactive and we can ignore it.
+       // TODO: Assert this.
+       r = _regions.next();
+     }
+   }
+ };
+ 
  ShenandoahFullGC::ShenandoahFullGC() :
    _gc_timer(ShenandoahHeap::heap()->gc_timer()),
    _preserved_marks(new PreservedMarksSet(true)) {}
  
  ShenandoahFullGC::~ShenandoahFullGC() {

@@ -97,29 +165,72 @@
  
    op_full(cause);
  }
  
  void ShenandoahFullGC::op_full(GCCause::Cause cause) {
+   ShenandoahHeap* const heap = ShenandoahHeap::heap();
    ShenandoahMetricsSnapshot metrics;
    metrics.snap_before();
  
    // Perform full GC
    do_it(cause);
  
    metrics.snap_after();
- 
+   if (heap->mode()->is_generational()) {
+     // Full GC should reset time since last gc for young and old heuristics
+     heap->young_generation()->heuristics()->record_cycle_end();
+     heap->old_generation()->heuristics()->record_cycle_end();
+ 
+     heap->mmu_tracker()->record_full(GCId::current());
+     heap->log_heap_status("At end of Full GC");
+ 
+     assert(heap->old_generation()->state() == ShenandoahOldGeneration::WAITING_FOR_BOOTSTRAP,
+            "After full GC, old generation should be waiting for bootstrap.");
+ 
+     // Since we allow temporary violation of these constraints during Full GC, we want to enforce that the assertions are
+     // made valid by the time Full GC completes.
+     assert(heap->old_generation()->used_regions_size() <= heap->old_generation()->max_capacity(),
+            "Old generation affiliated regions must be less than capacity");
+     assert(heap->young_generation()->used_regions_size() <= heap->young_generation()->max_capacity(),
+            "Young generation affiliated regions must be less than capacity");
+ 
+     assert((heap->young_generation()->used() + heap->young_generation()->get_humongous_waste())
+            <= heap->young_generation()->used_regions_size(), "Young consumed can be no larger than span of affiliated regions");
+     assert((heap->old_generation()->used() + heap->old_generation()->get_humongous_waste())
+            <= heap->old_generation()->used_regions_size(), "Old consumed can be no larger than span of affiliated regions");
+ 
+     // Establish baseline for next old-has-grown trigger.
+     heap->old_generation()->set_live_bytes_after_last_mark(heap->old_generation()->used() +
+                                                            heap->old_generation()->get_humongous_waste());
+   }
    if (metrics.is_good_progress()) {
      ShenandoahHeap::heap()->notify_gc_progress();
    } else {
      // Nothing to do. Tell the allocation path that we have failed to make
      // progress, and it can finally fail.
      ShenandoahHeap::heap()->notify_gc_no_progress();
    }
+ 
+   // Regardless if progress was made, we record that we completed a "successful" full GC.
+   heap->global_generation()->heuristics()->record_success_full();
+   heap->shenandoah_policy()->record_success_full();
  }
  
  void ShenandoahFullGC::do_it(GCCause::Cause gc_cause) {
    ShenandoahHeap* heap = ShenandoahHeap::heap();
+   // Since we may arrive here from degenerated GC failure of either young or old, establish generation as GLOBAL.
+   heap->set_gc_generation(heap->global_generation());
+ 
+   if (heap->mode()->is_generational()) {
+     // No need for old_gen->increase_used() as this was done when plabs were allocated.
+     heap->set_young_evac_reserve(0);
+     heap->set_old_evac_reserve(0);
+     heap->set_promoted_reserve(0);
+ 
+     // Full GC supersedes any marking or coalescing in old generation.
+     heap->cancel_old_gc();
+   }
  
    if (ShenandoahVerify) {
      heap->verifier()->verify_before_fullgc();
    }
  

@@ -159,41 +270,50 @@
      if (heap->is_update_refs_in_progress()) {
        heap->set_update_refs_in_progress(false);
      }
      assert(!heap->is_update_refs_in_progress(), "sanity");
  
-     // b. Cancel concurrent mark, if in progress
+     // b. Cancel all concurrent marks, if in progress
      if (heap->is_concurrent_mark_in_progress()) {
-       ShenandoahConcurrentGC::cancel();
-       heap->set_concurrent_mark_in_progress(false);
+       heap->cancel_concurrent_mark();
      }
      assert(!heap->is_concurrent_mark_in_progress(), "sanity");
  
      // c. Update roots if this full GC is due to evac-oom, which may carry from-space pointers in roots.
      if (has_forwarded_objects) {
        update_roots(true /*full_gc*/);
      }
  
      // d. Reset the bitmaps for new marking
-     heap->reset_mark_bitmap();
+     heap->global_generation()->reset_mark_bitmap();
      assert(heap->marking_context()->is_bitmap_clear(), "sanity");
-     assert(!heap->marking_context()->is_complete(), "sanity");
+     assert(!heap->global_generation()->is_mark_complete(), "sanity");
  
      // e. Abandon reference discovery and clear all discovered references.
-     ShenandoahReferenceProcessor* rp = heap->ref_processor();
+     ShenandoahReferenceProcessor* rp = heap->global_generation()->ref_processor();
      rp->abandon_partial_discovery();
  
      // f. Sync pinned region status from the CP marks
      heap->sync_pinned_region_status();
  
+     if (heap->mode()->is_generational()) {
+       for (size_t i = 0; i < heap->num_regions(); i++) {
+         ShenandoahHeapRegion* r = heap->get_region(i);
+         if (r->get_top_before_promote() != nullptr) {
+           r->restore_top_before_promote();
+         }
+       }
+     }
+ 
      // The rest of prologue:
      _preserved_marks->init(heap->workers()->active_workers());
  
      assert(heap->has_forwarded_objects() == has_forwarded_objects, "This should not change");
    }
  
    if (UseTLAB) {
+     // TODO: Do we need to explicitly retire PLABs?
      heap->gclabs_retire(ResizeTLAB);
      heap->tlabs_retire(ResizeTLAB);
    }
  
    OrderAccess::fence();

@@ -226,16 +346,25 @@
      OrderAccess::fence();
  
      phase3_update_references();
  
      phase4_compact_objects(worker_slices);
+ 
+     phase5_epilog();
    }
  
    {
      // Epilogue
+     // TODO: Merge with phase5_epilog?
      _preserved_marks->restore(heap->workers());
      _preserved_marks->reclaim();
+ 
+     if (heap->mode()->is_generational()) {
+       ShenandoahGCPhase phase(ShenandoahPhaseTimings::full_gc_reconstruct_remembered_set);
+       ShenandoahReconstructRememberedSetTask task;
+       heap->workers()->run_task(&task);
+     }
    }
  
    // Resize metaspace
    MetaspaceGC::compute_new_size();
  

@@ -250,10 +379,11 @@
  
    if (ShenandoahVerify) {
      heap->verifier()->verify_after_fullgc();
    }
  
+   // Humongous regions are promoted on demand and are accounted for by normal Full GC mechanisms.
    if (VerifyAfterGC) {
      Universe::verify();
    }
  
    {

@@ -268,35 +398,283 @@
  
  public:
    ShenandoahPrepareForMarkClosure() : _ctx(ShenandoahHeap::heap()->marking_context()) {}
  
    void heap_region_do(ShenandoahHeapRegion *r) {
-     _ctx->capture_top_at_mark_start(r);
-     r->clear_live_data();
+     if (r->affiliation() != FREE) {
+       _ctx->capture_top_at_mark_start(r);
+       r->clear_live_data();
+     }
    }
+ 
+   bool is_thread_safe() { return true; }
  };
  
  void ShenandoahFullGC::phase1_mark_heap() {
    GCTraceTime(Info, gc, phases) time("Phase 1: Mark live objects", _gc_timer);
    ShenandoahGCPhase mark_phase(ShenandoahPhaseTimings::full_gc_mark);
  
    ShenandoahHeap* heap = ShenandoahHeap::heap();
  
    ShenandoahPrepareForMarkClosure cl;
-   heap->heap_region_iterate(&cl);
+   heap->parallel_heap_region_iterate(&cl);
  
-   heap->set_unload_classes(heap->heuristics()->can_unload_classes());
+   heap->set_unload_classes(heap->global_generation()->heuristics()->can_unload_classes());
  
-   ShenandoahReferenceProcessor* rp = heap->ref_processor();
+   ShenandoahReferenceProcessor* rp = heap->global_generation()->ref_processor();
    // enable ("weak") refs discovery
    rp->set_soft_reference_policy(true); // forcefully purge all soft references
  
-   ShenandoahSTWMark mark(true /*full_gc*/);
+   ShenandoahSTWMark mark(heap->global_generation(), true /*full_gc*/);
    mark.mark();
    heap->parallel_cleaning(true /* full_gc */);
+ 
+   size_t live_bytes_in_old = 0;
+   for (size_t i = 0; i < heap->num_regions(); i++) {
+     ShenandoahHeapRegion* r = heap->get_region(i);
+     if (r->is_old()) {
+       live_bytes_in_old += r->get_live_data_bytes();
+     }
+   }
+   log_info(gc)("Live bytes in old after STW mark: " PROPERFMT, PROPERFMTARGS(live_bytes_in_old));
  }
  
+ class ShenandoahPrepareForCompactionTask : public WorkerTask {
+ private:
+   PreservedMarksSet*        const _preserved_marks;
+   ShenandoahHeap*           const _heap;
+   ShenandoahHeapRegionSet** const _worker_slices;
+   size_t                    const _num_workers;
+ 
+ public:
+   ShenandoahPrepareForCompactionTask(PreservedMarksSet *preserved_marks,
+                                      ShenandoahHeapRegionSet **worker_slices,
+                                      size_t num_workers);
+ 
+   static bool is_candidate_region(ShenandoahHeapRegion* r) {
+     // Empty region: get it into the slice to defragment the slice itself.
+     // We could have skipped this without violating correctness, but we really
+     // want to compact all live regions to the start of the heap, which sometimes
+     // means moving them into the fully empty regions.
+     if (r->is_empty()) return true;
+ 
+     // Can move the region, and this is not the humongous region. Humongous
+     // moves are special cased here, because their moves are handled separately.
+     return r->is_stw_move_allowed() && !r->is_humongous();
+   }
+ 
+   void work(uint worker_id);
+ };
+ 
+ class ShenandoahPrepareForGenerationalCompactionObjectClosure : public ObjectClosure {
+ private:
+   PreservedMarks*          const _preserved_marks;
+   ShenandoahHeap*          const _heap;
+   uint                           _tenuring_threshold;
+ 
+   // _empty_regions is a thread-local list of heap regions that have been completely emptied by this worker thread's
+   // compaction efforts.  The worker thread that drives these efforts adds compacted regions to this list if the
+   // region has not been compacted onto itself.
+   GrowableArray<ShenandoahHeapRegion*>& _empty_regions;
+   int _empty_regions_pos;
+   ShenandoahHeapRegion*          _old_to_region;
+   ShenandoahHeapRegion*          _young_to_region;
+   ShenandoahHeapRegion*          _from_region;
+   ShenandoahAffiliation          _from_affiliation;
+   HeapWord*                      _old_compact_point;
+   HeapWord*                      _young_compact_point;
+   uint                           _worker_id;
+ 
+ public:
+   ShenandoahPrepareForGenerationalCompactionObjectClosure(PreservedMarks* preserved_marks,
+                                                           GrowableArray<ShenandoahHeapRegion*>& empty_regions,
+                                                           ShenandoahHeapRegion* old_to_region,
+                                                           ShenandoahHeapRegion* young_to_region, uint worker_id) :
+       _preserved_marks(preserved_marks),
+       _heap(ShenandoahHeap::heap()),
+       _tenuring_threshold(0),
+       _empty_regions(empty_regions),
+       _empty_regions_pos(0),
+       _old_to_region(old_to_region),
+       _young_to_region(young_to_region),
+       _from_region(nullptr),
+       _old_compact_point((old_to_region != nullptr)? old_to_region->bottom(): nullptr),
+       _young_compact_point((young_to_region != nullptr)? young_to_region->bottom(): nullptr),
+       _worker_id(worker_id) {
+     if (_heap->mode()->is_generational()) {
+       _tenuring_threshold = _heap->age_census()->tenuring_threshold();
+     }
+   }
+ 
+   void set_from_region(ShenandoahHeapRegion* from_region) {
+     _from_region = from_region;
+     _from_affiliation = from_region->affiliation();
+     if (_from_region->has_live()) {
+       if (_from_affiliation == ShenandoahAffiliation::OLD_GENERATION) {
+         if (_old_to_region == nullptr) {
+           _old_to_region = from_region;
+           _old_compact_point = from_region->bottom();
+         }
+       } else {
+         assert(_from_affiliation == ShenandoahAffiliation::YOUNG_GENERATION, "from_region must be OLD or YOUNG");
+         if (_young_to_region == nullptr) {
+           _young_to_region = from_region;
+           _young_compact_point = from_region->bottom();
+         }
+       }
+     } // else, we won't iterate over this _from_region so we don't need to set up to region to hold copies
+   }
+ 
+   void finish() {
+     finish_old_region();
+     finish_young_region();
+   }
+ 
+   void finish_old_region() {
+     if (_old_to_region != nullptr) {
+       log_debug(gc)("Planned compaction into Old Region " SIZE_FORMAT ", used: " SIZE_FORMAT " tabulated by worker %u",
+                     _old_to_region->index(), _old_compact_point - _old_to_region->bottom(), _worker_id);
+       _old_to_region->set_new_top(_old_compact_point);
+       _old_to_region = nullptr;
+     }
+   }
+ 
+   void finish_young_region() {
+     if (_young_to_region != nullptr) {
+       log_debug(gc)("Worker %u planned compaction into Young Region " SIZE_FORMAT ", used: " SIZE_FORMAT,
+                     _worker_id, _young_to_region->index(), _young_compact_point - _young_to_region->bottom());
+       _young_to_region->set_new_top(_young_compact_point);
+       _young_to_region = nullptr;
+     }
+   }
+ 
+   bool is_compact_same_region() {
+     return (_from_region == _old_to_region) || (_from_region == _young_to_region);
+   }
+ 
+   int empty_regions_pos() {
+     return _empty_regions_pos;
+   }
+ 
+   void do_object(oop p) {
+     assert(_from_region != nullptr, "must set before work");
+     assert((_from_region->bottom() <= cast_from_oop<HeapWord*>(p)) && (cast_from_oop<HeapWord*>(p) < _from_region->top()),
+            "Object must reside in _from_region");
+     assert(_heap->complete_marking_context()->is_marked(p), "must be marked");
+     assert(!_heap->complete_marking_context()->allocated_after_mark_start(p), "must be truly marked");
+ 
+     size_t obj_size = p->size();
+     uint from_region_age = _from_region->age();
+     uint object_age = p->age();
+ 
+     bool promote_object = false;
+     if ((_from_affiliation == ShenandoahAffiliation::YOUNG_GENERATION) &&
+         (from_region_age + object_age >= _tenuring_threshold)) {
+       if ((_old_to_region != nullptr) && (_old_compact_point + obj_size > _old_to_region->end())) {
+         finish_old_region();
+         _old_to_region = nullptr;
+       }
+       if (_old_to_region == nullptr) {
+         if (_empty_regions_pos < _empty_regions.length()) {
+           ShenandoahHeapRegion* new_to_region = _empty_regions.at(_empty_regions_pos);
+           _empty_regions_pos++;
+           new_to_region->set_affiliation(OLD_GENERATION);
+           _old_to_region = new_to_region;
+           _old_compact_point = _old_to_region->bottom();
+           promote_object = true;
+         }
+         // Else this worker thread does not yet have any empty regions into which this aged object can be promoted so
+         // we leave promote_object as false, deferring the promotion.
+       } else {
+         promote_object = true;
+       }
+     }
+ 
+     if (promote_object || (_from_affiliation == ShenandoahAffiliation::OLD_GENERATION)) {
+       assert(_old_to_region != nullptr, "_old_to_region should not be nullptr when evacuating to OLD region");
+       if (_old_compact_point + obj_size > _old_to_region->end()) {
+         ShenandoahHeapRegion* new_to_region;
+ 
+         log_debug(gc)("Worker %u finishing old region " SIZE_FORMAT ", compact_point: " PTR_FORMAT ", obj_size: " SIZE_FORMAT
+                       ", &compact_point[obj_size]: " PTR_FORMAT ", region end: " PTR_FORMAT,  _worker_id, _old_to_region->index(),
+                       p2i(_old_compact_point), obj_size, p2i(_old_compact_point + obj_size), p2i(_old_to_region->end()));
+ 
+         // Object does not fit.  Get a new _old_to_region.
+         finish_old_region();
+         if (_empty_regions_pos < _empty_regions.length()) {
+           new_to_region = _empty_regions.at(_empty_regions_pos);
+           _empty_regions_pos++;
+           new_to_region->set_affiliation(OLD_GENERATION);
+         } else {
+           // If we've exhausted the previously selected _old_to_region, we know that the _old_to_region is distinct
+           // from _from_region.  That's because there is always room for _from_region to be compacted into itself.
+           // Since we're out of empty regions, let's use _from_region to hold the results of its own compaction.
+           new_to_region = _from_region;
+         }
+ 
+         assert(new_to_region != _old_to_region, "must not reuse same OLD to-region");
+         assert(new_to_region != nullptr, "must not be nullptr");
+         _old_to_region = new_to_region;
+         _old_compact_point = _old_to_region->bottom();
+       }
+ 
+       // Object fits into current region, record new location:
+       assert(_old_compact_point + obj_size <= _old_to_region->end(), "must fit");
+       shenandoah_assert_not_forwarded(nullptr, p);
+       _preserved_marks->push_if_necessary(p, p->mark());
+       p->forward_to(cast_to_oop(_old_compact_point));
+       _old_compact_point += obj_size;
+     } else {
+       assert(_from_affiliation == ShenandoahAffiliation::YOUNG_GENERATION,
+              "_from_region must be OLD_GENERATION or YOUNG_GENERATION");
+       assert(_young_to_region != nullptr, "_young_to_region should not be nullptr when compacting YOUNG _from_region");
+ 
+       // After full gc compaction, all regions have age 0.  Embed the region's age into the object's age in order to preserve
+       // tenuring progress.
+       if (_heap->is_aging_cycle()) {
+         _heap->increase_object_age(p, from_region_age + 1);
+       } else {
+         _heap->increase_object_age(p, from_region_age);
+       }
+ 
+       if (_young_compact_point + obj_size > _young_to_region->end()) {
+         ShenandoahHeapRegion* new_to_region;
+ 
+         log_debug(gc)("Worker %u finishing young region " SIZE_FORMAT ", compact_point: " PTR_FORMAT ", obj_size: " SIZE_FORMAT
+                       ", &compact_point[obj_size]: " PTR_FORMAT ", region end: " PTR_FORMAT,  _worker_id, _young_to_region->index(),
+                       p2i(_young_compact_point), obj_size, p2i(_young_compact_point + obj_size), p2i(_young_to_region->end()));
+ 
+         // Object does not fit.  Get a new _young_to_region.
+         finish_young_region();
+         if (_empty_regions_pos < _empty_regions.length()) {
+           new_to_region = _empty_regions.at(_empty_regions_pos);
+           _empty_regions_pos++;
+           new_to_region->set_affiliation(YOUNG_GENERATION);
+         } else {
+           // If we've exhausted the previously selected _young_to_region, we know that the _young_to_region is distinct
+           // from _from_region.  That's because there is always room for _from_region to be compacted into itself.
+           // Since we're out of empty regions, let's use _from_region to hold the results of its own compaction.
+           new_to_region = _from_region;
+         }
+ 
+         assert(new_to_region != _young_to_region, "must not reuse same OLD to-region");
+         assert(new_to_region != nullptr, "must not be nullptr");
+         _young_to_region = new_to_region;
+         _young_compact_point = _young_to_region->bottom();
+       }
+ 
+       // Object fits into current region, record new location:
+       assert(_young_compact_point + obj_size <= _young_to_region->end(), "must fit");
+       shenandoah_assert_not_forwarded(nullptr, p);
+       _preserved_marks->push_if_necessary(p, p->mark());
+       p->forward_to(cast_to_oop(_young_compact_point));
+       _young_compact_point += obj_size;
+     }
+   }
+ };
+ 
+ 
  class ShenandoahPrepareForCompactionObjectClosure : public ObjectClosure {
  private:
    PreservedMarks*          const _preserved_marks;
    ShenandoahHeap*          const _heap;
    GrowableArray<ShenandoahHeapRegion*>& _empty_regions;

@@ -321,10 +699,11 @@
      _from_region = from_region;
    }
  
    void finish_region() {
      assert(_to_region != nullptr, "should not happen");
+     assert(!_heap->mode()->is_generational(), "Generational GC should use different Closure");
      _to_region->set_new_top(_compact_point);
    }
  
    bool is_compact_same_region() {
      return _from_region == _to_region;

@@ -366,56 +745,68 @@
      p->forward_to(cast_to_oop(_compact_point));
      _compact_point += obj_size;
    }
  };
  
- class ShenandoahPrepareForCompactionTask : public WorkerTask {
- private:
-   PreservedMarksSet*        const _preserved_marks;
-   ShenandoahHeap*           const _heap;
-   ShenandoahHeapRegionSet** const _worker_slices;
  
- public:
-   ShenandoahPrepareForCompactionTask(PreservedMarksSet *preserved_marks, ShenandoahHeapRegionSet **worker_slices) :
+ ShenandoahPrepareForCompactionTask::ShenandoahPrepareForCompactionTask(PreservedMarksSet *preserved_marks,
+                                                                        ShenandoahHeapRegionSet **worker_slices,
+                                                                        size_t num_workers) :
      WorkerTask("Shenandoah Prepare For Compaction"),
-     _preserved_marks(preserved_marks),
-     _heap(ShenandoahHeap::heap()), _worker_slices(worker_slices) {
+     _preserved_marks(preserved_marks), _heap(ShenandoahHeap::heap()),
+     _worker_slices(worker_slices), _num_workers(num_workers) { }
+ 
+ 
+ void ShenandoahPrepareForCompactionTask::work(uint worker_id) {
+   ShenandoahParallelWorkerSession worker_session(worker_id);
+   ShenandoahHeapRegionSet* slice = _worker_slices[worker_id];
+   ShenandoahHeapRegionSetIterator it(slice);
+   ShenandoahHeapRegion* from_region = it.next();
+   // No work?
+   if (from_region == nullptr) {
+     return;
    }
  
-   static bool is_candidate_region(ShenandoahHeapRegion* r) {
-     // Empty region: get it into the slice to defragment the slice itself.
-     // We could have skipped this without violating correctness, but we really
-     // want to compact all live regions to the start of the heap, which sometimes
-     // means moving them into the fully empty regions.
-     if (r->is_empty()) return true;
+   // Sliding compaction. Walk all regions in the slice, and compact them.
+   // Remember empty regions and reuse them as needed.
+   ResourceMark rm;
  
-     // Can move the region, and this is not the humongous region. Humongous
-     // moves are special cased here, because their moves are handled separately.
-     return r->is_stw_move_allowed() && !r->is_humongous();
-   }
+   GrowableArray<ShenandoahHeapRegion*> empty_regions((int)_heap->num_regions());
  
-   void work(uint worker_id) {
-     ShenandoahParallelWorkerSession worker_session(worker_id);
-     ShenandoahHeapRegionSet* slice = _worker_slices[worker_id];
-     ShenandoahHeapRegionSetIterator it(slice);
-     ShenandoahHeapRegion* from_region = it.next();
-     // No work?
-     if (from_region == nullptr) {
-        return;
+   if (_heap->mode()->is_generational()) {
+     ShenandoahHeapRegion* old_to_region = (from_region->is_old())? from_region: nullptr;
+     ShenandoahHeapRegion* young_to_region = (from_region->is_young())? from_region: nullptr;
+     ShenandoahPrepareForGenerationalCompactionObjectClosure cl(_preserved_marks->get(worker_id),
+                                                                empty_regions,
+                                                                old_to_region, young_to_region,
+                                                                worker_id);
+     while (from_region != nullptr) {
+       assert(is_candidate_region(from_region), "Sanity");
+       log_debug(gc)("Worker %u compacting %s Region " SIZE_FORMAT " which had used " SIZE_FORMAT " and %s live",
+                     worker_id, from_region->affiliation_name(),
+                     from_region->index(), from_region->used(), from_region->has_live()? "has": "does not have");
+       cl.set_from_region(from_region);
+       if (from_region->has_live()) {
+         _heap->marked_object_iterate(from_region, &cl);
+       }
+       // Compacted the region to somewhere else? From-region is empty then.
+       if (!cl.is_compact_same_region()) {
+         empty_regions.append(from_region);
+       }
+       from_region = it.next();
      }
+     cl.finish();
  
-     // Sliding compaction. Walk all regions in the slice, and compact them.
-     // Remember empty regions and reuse them as needed.
-     ResourceMark rm;
- 
-     GrowableArray<ShenandoahHeapRegion*> empty_regions((int)_heap->num_regions());
- 
+     // Mark all remaining regions as empty
+     for (int pos = cl.empty_regions_pos(); pos < empty_regions.length(); ++pos) {
+       ShenandoahHeapRegion* r = empty_regions.at(pos);
+       r->set_new_top(r->bottom());
+     }
+   } else {
      ShenandoahPrepareForCompactionObjectClosure cl(_preserved_marks->get(worker_id), empty_regions, from_region);
- 
      while (from_region != nullptr) {
        assert(is_candidate_region(from_region), "Sanity");
- 
        cl.set_from_region(from_region);
        if (from_region->has_live()) {
          _heap->marked_object_iterate(from_region, &cl);
        }
  

@@ -431,11 +822,11 @@
      for (int pos = cl.empty_regions_pos(); pos < empty_regions.length(); ++pos) {
        ShenandoahHeapRegion* r = empty_regions.at(pos);
        r->set_new_top(r->bottom());
      }
    }
- };
+ }
  
  void ShenandoahFullGC::calculate_target_humongous_objects() {
    ShenandoahHeap* heap = ShenandoahHeap::heap();
  
    // Compute the new addresses for humongous objects. We need to do this after addresses

@@ -450,10 +841,11 @@
    // detected, then sliding restarts towards that non-movable region.
  
    size_t to_begin = heap->num_regions();
    size_t to_end = heap->num_regions();
  
+   log_debug(gc)("Full GC calculating target humongous objects from end " SIZE_FORMAT, to_end);
    for (size_t c = heap->num_regions(); c > 0; c--) {
      ShenandoahHeapRegion *r = heap->get_region(c - 1);
      if (r->is_humongous_continuation() || (r->new_top() == r->bottom())) {
        // To-region candidate: record this, and continue scan
        to_begin = r->index();

@@ -492,10 +884,11 @@
    void heap_region_do(ShenandoahHeapRegion* r) {
      if (r->is_trash()) {
        r->recycle();
      }
      if (r->is_cset()) {
+       // Leave affiliation unchanged
        r->make_regular_bypass();
      }
      if (r->is_empty_uncommitted()) {
        r->make_committed_bypass();
      }

@@ -516,26 +909,35 @@
    ShenandoahTrashImmediateGarbageClosure() :
      _heap(ShenandoahHeap::heap()),
      _ctx(ShenandoahHeap::heap()->complete_marking_context()) {}
  
    void heap_region_do(ShenandoahHeapRegion* r) {
+     if (!r->is_affiliated()) {
+       // Ignore free regions
+       // TODO: change iterators so they do not process FREE regions.
+       return;
+     }
+ 
      if (r->is_humongous_start()) {
        oop humongous_obj = cast_to_oop(r->bottom());
        if (!_ctx->is_marked(humongous_obj)) {
          assert(!r->has_live(),
-                "Region " SIZE_FORMAT " is not marked, should not have live", r->index());
+                "Humongous Start %s Region " SIZE_FORMAT " is not marked, should not have live",
+                r->affiliation_name(),  r->index());
+         log_debug(gc)("Trashing immediate humongous region " SIZE_FORMAT " because not marked", r->index());
          _heap->trash_humongous_region_at(r);
        } else {
          assert(r->has_live(),
-                "Region " SIZE_FORMAT " should have live", r->index());
+                "Humongous Start %s Region " SIZE_FORMAT " should have live", r->affiliation_name(),  r->index());
        }
      } else if (r->is_humongous_continuation()) {
        // If we hit continuation, the non-live humongous starts should have been trashed already
        assert(r->humongous_start_region()->has_live(),
-              "Region " SIZE_FORMAT " should have live", r->index());
+              "Humongous Continuation %s Region " SIZE_FORMAT " should have live", r->affiliation_name(),  r->index());
      } else if (r->is_regular()) {
        if (!r->has_live()) {
+         log_debug(gc)("Trashing immediate regular region " SIZE_FORMAT " because has no live", r->index());
          r->make_trash_immediate();
        }
      }
    }
  };

@@ -680,10 +1082,15 @@
      assert(is_distributed || !is_candidate, "All candidates are distributed: " SIZE_FORMAT, rid);
    }
  #endif
  }
  
+ // TODO:
+ //  Consider compacting old-gen objects toward the high end of memory and young-gen objects towards the low-end
+ //  of memory.  As currently implemented, all regions are compacted toward the low-end of memory.  This creates more
+ //  fragmentation of the heap, because old-gen regions get scattered among low-address regions such that it becomes
+ //  more difficult to find contiguous regions for humongous objects.
  void ShenandoahFullGC::phase2_calculate_target_addresses(ShenandoahHeapRegionSet** worker_slices) {
    GCTraceTime(Info, gc, phases) time("Phase 2: Compute new object addresses", _gc_timer);
    ShenandoahGCPhase calculate_address_phase(ShenandoahPhaseTimings::full_gc_calculate_addresses);
  
    ShenandoahHeap* heap = ShenandoahHeap::heap();

@@ -707,11 +1114,14 @@
    {
      ShenandoahGCPhase phase(ShenandoahPhaseTimings::full_gc_calculate_addresses_regular);
  
      distribute_slices(worker_slices);
  
-     ShenandoahPrepareForCompactionTask task(_preserved_marks, worker_slices);
+     size_t num_workers = heap->max_workers();
+ 
+     ResourceMark rm;
+     ShenandoahPrepareForCompactionTask task(_preserved_marks, worker_slices, num_workers);
      heap->workers()->run_task(&task);
    }
  
    // Compute the new addresses for humongous objects
    {

@@ -781,10 +1191,17 @@
      ShenandoahHeapRegion* r = _regions.next();
      while (r != nullptr) {
        if (!r->is_humongous_continuation() && r->has_live()) {
          _heap->marked_object_iterate(r, &obj_cl);
        }
+       if (r->is_pinned() && r->is_old() && r->is_active() && !r->is_humongous()) {
+         // Pinned regions are not compacted so they may still hold unmarked objects with
+         // reference to reclaimed memory. Remembered set scanning will crash if it attempts
+         // to iterate the oops in these objects.
+         r->begin_preemptible_coalesce_and_fill();
+         r->oop_fill_and_coalesce_without_cancel();
+       }
        r = _regions.next();
      }
    }
  };
  

@@ -881,17 +1298,44 @@
        r = slice.next();
      }
    }
  };
  
+ static void account_for_region(ShenandoahHeapRegion* r, size_t &region_count, size_t &region_usage, size_t &humongous_waste) {
+   region_count++;
+   region_usage += r->used();
+   if (r->is_humongous_start()) {
+     // For each humongous object, we take this path once regardless of how many regions it spans.
+     HeapWord* obj_addr = r->bottom();
+     oop obj = cast_to_oop(obj_addr);
+     size_t word_size = obj->size();
+     size_t region_size_words = ShenandoahHeapRegion::region_size_words();
+     size_t overreach = word_size % region_size_words;
+     if (overreach != 0) {
+       humongous_waste += (region_size_words - overreach) * HeapWordSize;
+     }
+     // else, this humongous object aligns exactly on region size, so no waste.
+   }
+ }
+ 
  class ShenandoahPostCompactClosure : public ShenandoahHeapRegionClosure {
  private:
    ShenandoahHeap* const _heap;
-   size_t _live;
+   bool _is_generational;
+   size_t _young_regions, _young_usage, _young_humongous_waste;
+   size_t _old_regions, _old_usage, _old_humongous_waste;
  
  public:
-   ShenandoahPostCompactClosure() : _heap(ShenandoahHeap::heap()), _live(0) {
+   ShenandoahPostCompactClosure() : _heap(ShenandoahHeap::heap()),
+                                    _is_generational(_heap->mode()->is_generational()),
+                                    _young_regions(0),
+                                    _young_usage(0),
+                                    _young_humongous_waste(0),
+                                    _old_regions(0),
+                                    _old_usage(0),
+                                    _old_humongous_waste(0)
+   {
      _heap->free_set()->clear();
    }
  
    void heap_region_do(ShenandoahHeapRegion* r) {
      assert (!r->is_cset(), "cset regions should have been demoted already");

@@ -907,10 +1351,14 @@
  
      size_t live = r->used();
  
      // Make empty regions that have been allocated into regular
      if (r->is_empty() && live > 0) {
+       if (!_is_generational) {
+         r->make_young_maybe();
+       }
+       // else, generational mode compaction has already established affiliation.
        r->make_regular_bypass();
        if (ZapUnusedHeapArea) {
          SpaceMangler::mangle_region(MemRegion(r->top(), r->end()));
        }
      }

@@ -922,28 +1370,45 @@
  
      // Recycle all trash regions
      if (r->is_trash()) {
        live = 0;
        r->recycle();
+     } else {
+       if (r->is_old()) {
+         account_for_region(r, _old_regions, _old_usage, _old_humongous_waste);
+       } else if (r->is_young()) {
+         account_for_region(r, _young_regions, _young_usage, _young_humongous_waste);
+       }
      }
- 
      r->set_live_data(live);
      r->reset_alloc_metadata();
-     _live += live;
    }
  
-   size_t get_live() {
-     return _live;
+   void update_generation_usage() {
+     if (_is_generational) {
+       _heap->old_generation()->establish_usage(_old_regions, _old_usage, _old_humongous_waste);
+       _heap->young_generation()->establish_usage(_young_regions, _young_usage, _young_humongous_waste);
+     } else {
+       assert(_old_regions == 0, "Old regions only expected in generational mode");
+       assert(_old_usage == 0, "Old usage only expected in generational mode");
+       assert(_old_humongous_waste == 0, "Old humongous waste only expected in generational mode");
+     }
+ 
+     // In generational mode, global usage should be the sum of young and old. This is also true
+     // for non-generational modes except that there are no old regions.
+     _heap->global_generation()->establish_usage(_old_regions + _young_regions,
+                                                 _old_usage + _young_usage,
+                                                 _old_humongous_waste + _young_humongous_waste);
    }
  };
  
  void ShenandoahFullGC::compact_humongous_objects() {
    // Compact humongous regions, based on their fwdptr objects.
    //
    // This code is serial, because doing the in-slice parallel sliding is tricky. In most cases,
    // humongous regions are already compacted, and do not require further moves, which alleviates
-   // sliding costs. We may consider doing this in parallel in future.
+   // sliding costs. We may consider doing this in parallel in the future.
  
    ShenandoahHeap* heap = ShenandoahHeap::heap();
  
    for (size_t c = heap->num_regions(); c > 0; c--) {
      ShenandoahHeapRegion* r = heap->get_region(c - 1);

@@ -961,29 +1426,36 @@
        size_t new_start = heap->heap_region_index_containing(old_obj->forwardee());
        size_t new_end   = new_start + num_regions - 1;
        assert(old_start != new_start, "must be real move");
        assert(r->is_stw_move_allowed(), "Region " SIZE_FORMAT " should be movable", r->index());
  
-       Copy::aligned_conjoint_words(r->bottom(), heap->get_region(new_start)->bottom(), words_size);
-       ContinuationGCSupport::relativize_stack_chunk(cast_to_oop<HeapWord*>(r->bottom()));
+       ContinuationGCSupport::relativize_stack_chunk(cast_to_oop<HeapWord*>(heap->get_region(old_start)->bottom()));
+       log_debug(gc)("Full GC compaction moves humongous object from region " SIZE_FORMAT " to region " SIZE_FORMAT,
+                     old_start, new_start);
+ 
+       Copy::aligned_conjoint_words(heap->get_region(old_start)->bottom(),
+                                    heap->get_region(new_start)->bottom(),
+                                    words_size);
  
        oop new_obj = cast_to_oop(heap->get_region(new_start)->bottom());
        new_obj->init_mark();
  
        {
+         ShenandoahAffiliation original_affiliation = r->affiliation();
          for (size_t c = old_start; c <= old_end; c++) {
            ShenandoahHeapRegion* r = heap->get_region(c);
+           // Leave humongous region affiliation unchanged.
            r->make_regular_bypass();
            r->set_top(r->bottom());
          }
  
          for (size_t c = new_start; c <= new_end; c++) {
            ShenandoahHeapRegion* r = heap->get_region(c);
            if (c == new_start) {
-             r->make_humongous_start_bypass();
+             r->make_humongous_start_bypass(original_affiliation);
            } else {
-             r->make_humongous_cont_bypass();
+             r->make_humongous_cont_bypass(original_affiliation);
            }
  
            // Trailing region may be non-full, record the remainder there
            size_t remainder = words_size & ShenandoahHeapRegion::region_size_words_mask();
            if ((c == new_end) && (remainder != 0)) {

@@ -1045,10 +1517,15 @@
    // Compact humongous objects after regular object moves
    {
      ShenandoahGCPhase phase(ShenandoahPhaseTimings::full_gc_copy_objects_humong);
      compact_humongous_objects();
    }
+ }
+ 
+ void ShenandoahFullGC::phase5_epilog() {
+   GCTraceTime(Info, gc, phases) time("Phase 5: Full GC epilog", _gc_timer);
+   ShenandoahHeap* heap = ShenandoahHeap::heap();
  
    // Reset complete bitmap. We're about to reset the complete-top-at-mark-start pointer
    // and must ensure the bitmap is in sync.
    {
      ShenandoahGCPhase phase(ShenandoahPhaseTimings::full_gc_copy_objects_reset_complete);

@@ -1057,16 +1534,89 @@
    }
  
    // Bring regions in proper states after the collection, and set heap properties.
    {
      ShenandoahGCPhase phase(ShenandoahPhaseTimings::full_gc_copy_objects_rebuild);
- 
      ShenandoahPostCompactClosure post_compact;
      heap->heap_region_iterate(&post_compact);
-     heap->set_used(post_compact.get_live());
+     post_compact.update_generation_usage();
+     if (heap->mode()->is_generational()) {
+       size_t old_usage = heap->old_generation()->used_regions_size();
+       size_t old_capacity = heap->old_generation()->max_capacity();
+ 
+       assert(old_usage % ShenandoahHeapRegion::region_size_bytes() == 0, "Old usage must aligh with region size");
+       assert(old_capacity % ShenandoahHeapRegion::region_size_bytes() == 0, "Old capacity must aligh with region size");
+ 
+       if (old_capacity > old_usage) {
+         size_t excess_old_regions = (old_capacity - old_usage) / ShenandoahHeapRegion::region_size_bytes();
+         heap->generation_sizer()->transfer_to_young(excess_old_regions);
+       } else if (old_capacity < old_usage) {
+         size_t old_regions_deficit = (old_usage - old_capacity) / ShenandoahHeapRegion::region_size_bytes();
+         heap->generation_sizer()->force_transfer_to_old(old_regions_deficit);
+       }
  
+       log_info(gc)("FullGC done: young usage: " SIZE_FORMAT "%s, old usage: " SIZE_FORMAT "%s",
+                    byte_size_in_proper_unit(heap->young_generation()->used()), proper_unit_for_byte_size(heap->young_generation()->used()),
+                    byte_size_in_proper_unit(heap->old_generation()->used()),   proper_unit_for_byte_size(heap->old_generation()->used()));
+     }
      heap->collection_set()->clear();
-     heap->free_set()->rebuild();
-   }
+     size_t young_cset_regions, old_cset_regions;
+     size_t first_old, last_old, num_old;
+     heap->free_set()->prepare_to_rebuild(young_cset_regions, old_cset_regions, first_old, last_old, num_old);
+ 
+     // We also do not expand old generation size following Full GC because we have scrambled age populations and
+     // no longer have objects separated by age into distinct regions.
+ 
+     // TODO: Do we need to fix FullGC so that it maintains aged segregation of objects into distinct regions?
+     //       A partial solution would be to remember how many objects are of tenure age following Full GC, but
+     //       this is probably suboptimal, because most of these objects will not reside in a region that will be
+     //       selected for the next evacuation phase.
+ 
+     // In case this Full GC resulted from degeneration, clear the tally on anticipated promotion.
+     heap->clear_promotion_potential();
  
-   heap->clear_cancelled_gc();
+     if (heap->mode()->is_generational()) {
+       // Invoke this in case we are able to transfer memory from OLD to YOUNG.
+       heap->adjust_generation_sizes_for_next_cycle(0, 0, 0);
+     }
+     heap->free_set()->rebuild(young_cset_regions, old_cset_regions);
+ 
+     // We defer generation resizing actions until after cset regions have been recycled.  We do this even following an
+     // abbreviated cycle.
+     if (heap->mode()->is_generational()) {
+       bool success;
+       size_t region_xfer;
+       const char* region_destination;
+       ShenandoahYoungGeneration* young_gen = heap->young_generation();
+       ShenandoahGeneration* old_gen = heap->old_generation();
+ 
+       size_t old_region_surplus = heap->get_old_region_surplus();
+       size_t old_region_deficit = heap->get_old_region_deficit();
+       if (old_region_surplus) {
+         success = heap->generation_sizer()->transfer_to_young(old_region_surplus);
+         region_destination = "young";
+         region_xfer = old_region_surplus;
+       } else if (old_region_deficit) {
+         success = heap->generation_sizer()->transfer_to_old(old_region_deficit);
+         region_destination = "old";
+         region_xfer = old_region_deficit;
+         if (!success) {
+           ((ShenandoahOldHeuristics *) old_gen->heuristics())->trigger_cannot_expand();
+         }
+       } else {
+         region_destination = "none";
+         region_xfer = 0;
+         success = true;
+       }
+       heap->set_old_region_surplus(0);
+       heap->set_old_region_deficit(0);
+       size_t young_available = young_gen->available();
+       size_t old_available = old_gen->available();
+       log_info(gc, ergo)("After cleanup, %s " SIZE_FORMAT " regions to %s to prepare for next gc, old available: "
+                          SIZE_FORMAT "%s, young_available: " SIZE_FORMAT "%s",
+                          success? "successfully transferred": "failed to transfer", region_xfer, region_destination,
+                          byte_size_in_proper_unit(old_available), proper_unit_for_byte_size(old_available),
+                          byte_size_in_proper_unit(young_available), proper_unit_for_byte_size(young_available));
+     }
+     heap->clear_cancelled_gc(true /* clear oom handler */);
+   }
  }
< prev index next >