// Copyright 2012 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #ifndef V8_HEAP_MARK_COMPACT_H_ #define V8_HEAP_MARK_COMPACT_H_ #include <atomic> #include <vector> #include "include/v8-internal.h" #include "src/heap/base/worklist.h" #include "src/heap/concurrent-marking.h" #include "src/heap/marking-visitor.h" #include "src/heap/marking-worklist.h" #include "src/heap/marking.h" #include "src/heap/memory-measurement.h" #include "src/heap/parallel-work-item.h" #include "src/heap/spaces.h" #include "src/heap/sweeper.h" namespace v8 { namespace internal { // Forward declarations. class EvacuationJobTraits; class HeapObjectVisitor; class ItemParallelJob; class LargeObjectSpace; class LargePage; class MigrationObserver; class ReadOnlySpace; class RecordMigratedSlotVisitor; class UpdatingItem; class MarkBitCellIterator { public: MarkBitCellIterator(const MemoryChunk* chunk, Bitmap* bitmap) : chunk_(chunk) { last_cell_index_ = Bitmap::IndexToCell(chunk_->AddressToMarkbitIndex(chunk_->area_end())); cell_base_ = chunk_->address(); cell_index_ = Bitmap::IndexToCell(chunk_->AddressToMarkbitIndex(cell_base_)); cells_ = bitmap->cells(); } inline bool Done() { return cell_index_ >= last_cell_index_; } inline bool HasNext() { return cell_index_ < last_cell_index_ - 1; } inline MarkBit::CellType* CurrentCell() { DCHECK_EQ(cell_index_, Bitmap::IndexToCell(Bitmap::CellAlignIndex( chunk_->AddressToMarkbitIndex(cell_base_)))); return &cells_[cell_index_]; } inline Address CurrentCellBase() { DCHECK_EQ(cell_index_, Bitmap::IndexToCell(Bitmap::CellAlignIndex( chunk_->AddressToMarkbitIndex(cell_base_)))); return cell_base_; } V8_WARN_UNUSED_RESULT inline bool Advance() { cell_base_ += Bitmap::kBitsPerCell * kTaggedSize; return ++cell_index_ != last_cell_index_; } inline bool Advance(unsigned int new_cell_index) { if (new_cell_index != cell_index_) { DCHECK_GT(new_cell_index, cell_index_); DCHECK_LE(new_cell_index, last_cell_index_); unsigned int diff = new_cell_index - cell_index_; cell_index_ = new_cell_index; cell_base_ += diff * (Bitmap::kBitsPerCell * kTaggedSize); return true; } return false; } // Return the next mark bit cell. If there is no next it returns 0; inline MarkBit::CellType PeekNext() { if (HasNext()) { return cells_[cell_index_ + 1]; } return 0; } private: const MemoryChunk* chunk_; MarkBit::CellType* cells_; unsigned int last_cell_index_; unsigned int cell_index_; Address cell_base_; }; enum LiveObjectIterationMode { kBlackObjects, kGreyObjects, kAllLiveObjects }; template <LiveObjectIterationMode mode> class LiveObjectRange { public: class iterator { public: using value_type = std::pair<HeapObject, int /* size */>; using pointer = const value_type*; using reference = const value_type&; using iterator_category = std::forward_iterator_tag; inline iterator(const MemoryChunk* chunk, Bitmap* bitmap, Address start); inline iterator& operator++(); inline iterator operator++(int); bool operator==(iterator other) const { return current_object_ == other.current_object_; } bool operator!=(iterator other) const { return !(*this == other); } value_type operator*() { return std::make_pair(current_object_, current_size_); } private: inline void AdvanceToNextValidObject(); const MemoryChunk* const chunk_; Map const one_word_filler_map_; Map const two_word_filler_map_; Map const free_space_map_; MarkBitCellIterator it_; Address cell_base_; MarkBit::CellType current_cell_; HeapObject current_object_; int current_size_; }; LiveObjectRange(const MemoryChunk* chunk, Bitmap* bitmap) : chunk_(chunk), bitmap_(bitmap), start_(chunk_->area_start()), end_(chunk->area_end()) { DCHECK(!chunk->IsLargePage()); } inline iterator begin(); inline iterator end(); private: const MemoryChunk* const chunk_; Bitmap* bitmap_; Address start_; Address end_; }; class LiveObjectVisitor : AllStatic { public: enum IterationMode { kKeepMarking, kClearMarkbits, }; // Visits black objects on a MemoryChunk until the Visitor returns |false| for // an object. If IterationMode::kClearMarkbits is passed the markbits and // slots for visited objects are cleared for each successfully visited object. template <class Visitor, typename MarkingState> static bool VisitBlackObjects(MemoryChunk* chunk, MarkingState* state, Visitor* visitor, IterationMode iteration_mode, HeapObject* failed_object); // Visits black objects on a MemoryChunk. The visitor is not allowed to fail // visitation for an object. template <class Visitor, typename MarkingState> static void VisitBlackObjectsNoFail(MemoryChunk* chunk, MarkingState* state, Visitor* visitor, IterationMode iteration_mode); template <typename MarkingState> static void RecomputeLiveBytes(MemoryChunk* chunk, MarkingState* state); }; enum class AlwaysPromoteYoung { kYes, kNo }; enum PageEvacuationMode { NEW_TO_NEW, NEW_TO_OLD }; enum class RememberedSetUpdatingMode { ALL, OLD_TO_NEW_ONLY }; // This is used by marking visitors. class MarkingState final : public MarkingStateBase<MarkingState, AccessMode::ATOMIC> { public: explicit MarkingState(PtrComprCageBase cage_base) : MarkingStateBase(cage_base) {} ConcurrentBitmap<AccessMode::ATOMIC>* bitmap( const BasicMemoryChunk* chunk) const { return chunk->marking_bitmap<AccessMode::ATOMIC>(); } // Concurrent marking uses local live bytes so we may do these accesses // non-atomically. void IncrementLiveBytes(MemoryChunk* chunk, intptr_t by) { chunk->live_byte_count_.fetch_add(by, std::memory_order_relaxed); } intptr_t live_bytes(const MemoryChunk* chunk) const { return chunk->live_byte_count_.load(std::memory_order_relaxed); } void SetLiveBytes(MemoryChunk* chunk, intptr_t value) { chunk->live_byte_count_.store(value, std::memory_order_relaxed); } }; // This is used by Scavenger and Evacuator in TransferColor. // Live byte increments have to be atomic. class AtomicMarkingState final : public MarkingStateBase<AtomicMarkingState, AccessMode::ATOMIC> { public: explicit AtomicMarkingState(PtrComprCageBase cage_base) : MarkingStateBase(cage_base) {} ConcurrentBitmap<AccessMode::ATOMIC>* bitmap( const BasicMemoryChunk* chunk) const { return chunk->marking_bitmap<AccessMode::ATOMIC>(); } void IncrementLiveBytes(MemoryChunk* chunk, intptr_t by) { chunk->live_byte_count_.fetch_add(by); } }; class NonAtomicMarkingState final : public MarkingStateBase<NonAtomicMarkingState, AccessMode::NON_ATOMIC> { public: explicit NonAtomicMarkingState(PtrComprCageBase cage_base) : MarkingStateBase(cage_base) {} ConcurrentBitmap<AccessMode::NON_ATOMIC>* bitmap( const BasicMemoryChunk* chunk) const { return chunk->marking_bitmap<AccessMode::NON_ATOMIC>(); } void IncrementLiveBytes(MemoryChunk* chunk, intptr_t by) { chunk->live_byte_count_.fetch_add(by, std::memory_order_relaxed); } intptr_t live_bytes(const MemoryChunk* chunk) const { return chunk->live_byte_count_.load(std::memory_order_relaxed); } void SetLiveBytes(MemoryChunk* chunk, intptr_t value) { chunk->live_byte_count_.store(value, std::memory_order_relaxed); } }; // This visitor is used for marking on the main thread. It is cheaper than // the concurrent marking visitor because it does not snapshot JSObjects. template <typename MarkingState> class MainMarkingVisitor final : public MarkingVisitorBase<MainMarkingVisitor<MarkingState>, MarkingState> { public: MainMarkingVisitor(MarkingState* marking_state, MarkingWorklists::Local* local_marking_worklists, WeakObjects::Local* local_weak_objects, Heap* heap, unsigned mark_compact_epoch, base::EnumSet<CodeFlushMode> code_flush_mode, bool embedder_tracing_enabled, bool should_keep_ages_unchanged) : MarkingVisitorBase<MainMarkingVisitor<MarkingState>, MarkingState>( local_marking_worklists, local_weak_objects, heap, mark_compact_epoch, code_flush_mode, embedder_tracing_enabled, should_keep_ages_unchanged), marking_state_(marking_state) {} // HeapVisitor override. bool ShouldVisit(HeapObject object) { return marking_state_->GreyToBlack(object); } private: // Functions required by MarkingVisitorBase. template <typename T, typename TBodyDescriptor = typename T::BodyDescriptor> int VisitJSObjectSubclass(Map map, T object); template <typename T> int VisitLeftTrimmableArray(Map map, T object); template <typename TSlot> void RecordSlot(HeapObject object, TSlot slot, HeapObject target); void RecordRelocSlot(Code host, RelocInfo* rinfo, HeapObject target); MarkingState* marking_state() { return marking_state_; } TraceRetainingPathMode retaining_path_mode() { return (V8_UNLIKELY(FLAG_track_retaining_path)) ? TraceRetainingPathMode::kEnabled : TraceRetainingPathMode::kDisabled; } MarkingState* const marking_state_; friend class MarkingVisitorBase<MainMarkingVisitor<MarkingState>, MarkingState>; }; class YoungGenerationMainMarkingVisitor final : public YoungGenerationMarkingVisitorBase< YoungGenerationMainMarkingVisitor, MarkingState> { public: YoungGenerationMainMarkingVisitor(Isolate* isolate, MarkingState* marking_state, MarkingWorklists::Local* worklists_local); // HeapVisitor override. bool ShouldVisit(HeapObject object); private: MarkingState* marking_state() { return marking_state_; } MarkingState* const marking_state_; friend class YoungGenerationMarkingVisitorBase< YoungGenerationMainMarkingVisitor, MarkingState>; }; class CollectorBase { public: GarbageCollector garbage_collector() { return garbage_collector_; } virtual void SetUp() {} virtual void TearDown() {} virtual void CollectGarbage() = 0; virtual void Prepare() = 0; virtual void StartMarking() = 0; MarkingState* marking_state() { return &marking_state_; } NonAtomicMarkingState* non_atomic_marking_state() { return &non_atomic_marking_state_; } inline Heap* heap() const { return heap_; } inline Isolate* isolate(); MarkingWorklists* marking_worklists() { return &marking_worklists_; } MarkingWorklists::Local* local_marking_worklists() { return local_marking_worklists_.get(); } // Drains the main thread marking worklist until the specified number of // bytes are processed. If the number of bytes is zero, then the worklist // is drained until it is empty. virtual std::pair<size_t, size_t> ProcessMarkingWorklist( size_t bytes_to_process) = 0; // Used by incremental marking for object that change their layout. virtual void VisitObject(HeapObject obj) = 0; virtual bool sweeping_in_progress() const = 0; virtual void Finish() = 0; bool IsMajorMC(); private: std::vector<Page*> new_space_evacuation_pages_; std::vector<LargePage*> promoted_large_pages_; protected: Heap* heap_; GarbageCollector garbage_collector_; MarkingWorklists marking_worklists_; std::unique_ptr<MarkingWorklists::Local> local_marking_worklists_; MarkingState marking_state_; NonAtomicMarkingState non_atomic_marking_state_; explicit CollectorBase(Heap* heap, GarbageCollector collector); virtual ~CollectorBase() = default; }; // Collector for young and old generation. class MarkCompactCollector final : public CollectorBase { public: using MarkingVisitor = MainMarkingVisitor<MarkingState>; class CustomRootBodyMarkingVisitor; class SharedHeapObjectVisitor; class RootMarkingVisitor; enum class StartCompactionMode { kIncremental, kAtomic, }; enum class MarkingWorklistProcessingMode { kDefault, kTrackNewlyDiscoveredObjects }; static MarkCompactCollector* From(CollectorBase* collector) { return static_cast<MarkCompactCollector*>(collector); } std::pair<size_t, size_t> ProcessMarkingWorklist( size_t bytes_to_process) final; std::pair<size_t, size_t> ProcessMarkingWorklist( size_t bytes_to_process, MarkingWorklistProcessingMode mode); void SetUp() final; void TearDown() final; // Performs a global garbage collection. void CollectGarbage() final; void CollectEvacuationCandidates(PagedSpace* space); void AddEvacuationCandidate(Page* p); // Prepares for GC by resetting relocation info in old and map spaces and // choosing spaces to compact. void Prepare() final; // Stop concurrent marking (either by preempting it right away or waiting for // it to complete as requested by |stop_request|). void FinishConcurrentMarking(); // Returns whether compaction is running. bool StartCompaction(StartCompactionMode mode); void AbortCompaction(); void StartMarking() final; static inline bool IsOnEvacuationCandidate(Object obj) { return Page::FromAddress(obj.ptr())->IsEvacuationCandidate(); } static bool IsOnEvacuationCandidate(MaybeObject obj); struct RecordRelocSlotInfo { MemoryChunk* memory_chunk; SlotType slot_type; uint32_t offset; }; static V8_EXPORT_PRIVATE bool IsMapOrForwarded(Map map); static bool ShouldRecordRelocSlot(Code host, RelocInfo* rinfo, HeapObject target); static RecordRelocSlotInfo ProcessRelocInfo(Code host, RelocInfo* rinfo, HeapObject target); static void RecordRelocSlot(Code host, RelocInfo* rinfo, HeapObject target); V8_INLINE static void RecordSlot(HeapObject object, ObjectSlot slot, HeapObject target); V8_INLINE static void RecordSlot(HeapObject object, HeapObjectSlot slot, HeapObject target); V8_INLINE static void RecordSlot(MemoryChunk* source_page, HeapObjectSlot slot, HeapObject target); void RecordLiveSlotsOnPage(Page* page); bool is_compacting() const { return compacting_; } bool is_shared_heap() const { return is_shared_heap_; } void FinishSweepingIfOutOfWork(); enum class SweepingForcedFinalizationMode { kUnifiedHeap, kV8Only }; // Ensures that sweeping is finished. // // Note: Can only be called safely from main thread. V8_EXPORT_PRIVATE void EnsureSweepingCompleted( SweepingForcedFinalizationMode mode); void EnsurePageIsSwept(Page* page); void DrainSweepingWorklistForSpace(AllocationSpace space); // Checks if sweeping is in progress right now on any space. bool sweeping_in_progress() const final { return sweeper_->sweeping_in_progress(); } void set_evacuation(bool evacuation) { evacuation_ = evacuation; } bool evacuation() const { return evacuation_; } inline void AddTransitionArray(TransitionArray array); Sweeper* sweeper() { return sweeper_; } #ifdef DEBUG // Checks whether performing mark-compact collection. bool in_use() { return state_ > PREPARE_GC; } bool are_map_pointers_encoded() { return state_ == UPDATE_POINTERS; } #endif void VerifyMarking(); #ifdef VERIFY_HEAP void VerifyMarkbitsAreClean(); void VerifyMarkbitsAreDirty(ReadOnlySpace* space); void VerifyMarkbitsAreClean(PagedSpaceBase* space); void VerifyMarkbitsAreClean(NewSpace* space); void VerifyMarkbitsAreClean(LargeObjectSpace* space); #endif unsigned epoch() const { return epoch_; } base::EnumSet<CodeFlushMode> code_flush_mode() const { return code_flush_mode_; } WeakObjects* weak_objects() { return &weak_objects_; } WeakObjects::Local* local_weak_objects() { return local_weak_objects_.get(); } void VisitObject(HeapObject obj) final; void AddNewlyDiscovered(HeapObject object) { if (ephemeron_marking_.newly_discovered_overflowed) return; if (ephemeron_marking_.newly_discovered.size() < ephemeron_marking_.newly_discovered_limit) { ephemeron_marking_.newly_discovered.push_back(object); } else { ephemeron_marking_.newly_discovered_overflowed = true; } } void ResetNewlyDiscovered() { ephemeron_marking_.newly_discovered_overflowed = false; ephemeron_marking_.newly_discovered.clear(); } explicit MarkCompactCollector(Heap* heap); ~MarkCompactCollector() final; // Used by wrapper tracing. V8_INLINE void MarkExternallyReferencedObject(HeapObject obj); std::unique_ptr<UpdatingItem> CreateRememberedSetUpdatingItem( MemoryChunk* chunk, RememberedSetUpdatingMode updating_mode); #ifdef V8_ENABLE_INNER_POINTER_RESOLUTION_MB // Finds an object header based on a `maybe_inner_ptr`. It returns // `kNullAddress` if the parameter does not point to (the interior of) a valid // heap object, or if it points to (the interior of) some object that is // already marked as live (black or grey). Address FindBasePtrForMarking(Address maybe_inner_ptr); #endif // V8_ENABLE_INNER_POINTER_RESOLUTION_MB private: void ComputeEvacuationHeuristics(size_t area_size, int* target_fragmentation_percent, size_t* max_evacuated_bytes); void RecordObjectStats(); // Finishes GC, performs heap verification if enabled. void Finish() final; // Free unmarked ArrayBufferExtensions. void SweepArrayBufferExtensions(); // Free unmarked entries in the ExternalPointerTable. void SweepExternalPointerTable(); void MarkLiveObjects(); // Marks the object grey and adds it to the marking work list. // This is for non-incremental marking only. V8_INLINE void MarkObject(HeapObject host, HeapObject obj); // Marks the object grey and adds it to the marking work list. // This is for non-incremental marking only. V8_INLINE void MarkRootObject(Root root, HeapObject obj); // Mark the heap roots and all objects reachable from them. void MarkRoots(RootVisitor* root_visitor, ObjectVisitor* custom_root_body_visitor); // Mark the stack roots and all objects reachable from them. void MarkRootsFromStack(RootVisitor* root_visitor); // Mark all objects that are directly referenced from one of the clients // heaps. void MarkObjectsFromClientHeaps(); // Updates pointers to shared objects from client heaps. void UpdatePointersInClientHeaps(); void UpdatePointersInClientHeap(Isolate* client); // Marks object reachable from harmony weak maps and wrapper tracing. void MarkTransitiveClosure(); void VerifyEphemeronMarking(); // If the call-site of the top optimized code was not prepared for // deoptimization, then treat embedded pointers in the code as strong as // otherwise they can die and try to deoptimize the underlying code. void ProcessTopOptimizedFrame(ObjectVisitor* visitor, Isolate* isolate); // Implements ephemeron semantics: Marks value if key is already reachable. // Returns true if value was actually marked. bool ProcessEphemeron(HeapObject key, HeapObject value); // Marks the transitive closure by draining the marking worklist iteratively, // applying ephemerons semantics and invoking embedder tracing until a // fixpoint is reached. Returns false if too many iterations have been tried // and the linear approach should be used. bool MarkTransitiveClosureUntilFixpoint(); // Marks the transitive closure applying ephemeron semantics and invoking // embedder tracing with a linear algorithm for ephemerons. Only used if // fixpoint iteration doesn't finish within a few iterations. void MarkTransitiveClosureLinear(); // Drains ephemeron and marking worklists. Single iteration of the // fixpoint iteration. bool ProcessEphemerons(); // Perform Wrapper Tracing if in use. void PerformWrapperTracing(); // Callback function for telling whether the object *p is an unmarked // heap object. static bool IsUnmarkedHeapObject(Heap* heap, FullObjectSlot p); // Retain dying maps for `FLAG_retain_maps_for_n_gc` garbage collections to // increase chances of reusing of map transition tree in future. void RetainMaps(); // Clear non-live references in weak cells, transition and descriptor arrays, // and deoptimize dependent code of non-live maps. void ClearNonLiveReferences(); void MarkDependentCodeForDeoptimization(); // Checks if the given weak cell is a simple transition from the parent map // of the given dead target. If so it clears the transition and trims // the descriptor array of the parent if needed. void ClearPotentialSimpleMapTransition(Map dead_target); void ClearPotentialSimpleMapTransition(Map map, Map dead_target); // Flushes a weakly held bytecode array from a shared function info. void FlushBytecodeFromSFI(SharedFunctionInfo shared_info); // Clears bytecode arrays / baseline code that have not been executed for // multiple collections. void ProcessOldCodeCandidates(); void ProcessFlushedBaselineCandidates(); // Resets any JSFunctions which have had their bytecode flushed. void ClearFlushedJsFunctions(); // Compact every array in the global list of transition arrays and // trim the corresponding descriptor array if a transition target is non-live. void ClearFullMapTransitions(); void TrimDescriptorArray(Map map, DescriptorArray descriptors); void TrimEnumCache(Map map, DescriptorArray descriptors); bool CompactTransitionArray(Map map, TransitionArray transitions, DescriptorArray descriptors); bool TransitionArrayNeedsCompaction(TransitionArray transitions, int num_transitions); // After all reachable objects have been marked those weak map entries // with an unreachable key are removed from all encountered weak maps. // The linked list of all encountered weak maps is destroyed. void ClearWeakCollections(); // Goes through the list of encountered weak references and clears those with // dead values. If the value is a dead map and the parent map transitions to // the dead map via weak cell, then this function also clears the map // transition. void ClearWeakReferences(); // Goes through the list of encountered JSWeakRefs and WeakCells and clears // those with dead values. void ClearJSWeakRefs(); // Starts sweeping of spaces by contributing on the main thread and setting // up other pages for sweeping. Does not start sweeper tasks. void Sweep(); void StartSweepSpace(PagedSpace* space); void SweepLargeSpace(LargeObjectSpace* space); void EvacuatePrologue(); void EvacuateEpilogue(); void Evacuate(); void EvacuatePagesInParallel(); void UpdatePointersAfterEvacuation(); void ReleaseEvacuationCandidates(); // Returns number of aborted pages. size_t PostProcessEvacuationCandidates(); void ReportAbortedEvacuationCandidateDueToOOM(Address failed_start, Page* page); void ReportAbortedEvacuationCandidateDueToFlags(Address failed_start, Page* page); static const int kEphemeronChunkSize = 8 * KB; int NumberOfParallelEphemeronVisitingTasks(size_t elements); void RightTrimDescriptorArray(DescriptorArray array, int descriptors_to_trim); base::Mutex mutex_; base::Semaphore page_parallel_job_semaphore_{0}; #ifdef DEBUG enum CollectorState{IDLE, PREPARE_GC, MARK_LIVE_OBJECTS, SWEEP_SPACES, ENCODE_FORWARDING_ADDRESSES, UPDATE_POINTERS, RELOCATE_OBJECTS}; // The current stage of the collector. CollectorState state_; #endif const bool is_shared_heap_; bool evacuation_ = false; // True if we are collecting slots to perform evacuation from evacuation // candidates. bool compacting_ = false; bool black_allocation_ = false; bool have_code_to_deoptimize_ = false; bool parallel_marking_ = false; WeakObjects weak_objects_; EphemeronMarking ephemeron_marking_; std::unique_ptr<MarkingVisitor> marking_visitor_; std::unique_ptr<WeakObjects::Local> local_weak_objects_; NativeContextInferrer native_context_inferrer_; NativeContextStats native_context_stats_; // Candidates for pages that should be evacuated. std::vector<Page*> evacuation_candidates_; // Pages that are actually processed during evacuation. std::vector<Page*> old_space_evacuation_pages_; std::vector<Page*> new_space_evacuation_pages_; std::vector<std::pair<Address, Page*>> aborted_evacuation_candidates_due_to_oom_; std::vector<std::pair<Address, Page*>> aborted_evacuation_candidates_due_to_flags_; std::vector<LargePage*> promoted_large_pages_; Sweeper* sweeper_; // Counts the number of major mark-compact collections. The counter is // incremented right after marking. This is used for: // - marking descriptor arrays. See NumberOfMarkedDescriptors. Only the lower // two bits are used, so it is okay if this counter overflows and wraps // around. unsigned epoch_ = 0; // Bytecode flushing is disabled when the code coverage mode is changed. Since // that can happen while a GC is happening and we need the // code_flush_mode_ to remain the same through out a GC, we record this at // the start of each GC. base::EnumSet<CodeFlushMode> code_flush_mode_; friend class FullEvacuator; friend class RecordMigratedSlotVisitor; }; class V8_NODISCARD EvacuationScope { public: explicit EvacuationScope(MarkCompactCollector* collector) : collector_(collector) { collector_->set_evacuation(true); } ~EvacuationScope() { collector_->set_evacuation(false); } private: MarkCompactCollector* collector_; }; // Collector for young-generation only. class MinorMarkCompactCollector final : public CollectorBase { public: static constexpr size_t kMaxParallelTasks = 8; static MinorMarkCompactCollector* From(CollectorBase* collector) { return static_cast<MinorMarkCompactCollector*>(collector); } explicit MinorMarkCompactCollector(Heap* heap); ~MinorMarkCompactCollector() final; std::pair<size_t, size_t> ProcessMarkingWorklist( size_t bytes_to_process) final; void SetUp() final; void TearDown() final; void CollectGarbage() final; void Prepare() final; void StartMarking() final; void MakeIterable(Page* page, FreeSpaceTreatmentMode free_space_mode); void CleanupPromotedPages(); std::unique_ptr<UpdatingItem> CreateRememberedSetUpdatingItem( MemoryChunk* chunk, RememberedSetUpdatingMode updating_mode); void Finish() final; bool sweeping_in_progress() const final { // TODO(v8:13012): Fix this once sweeping is implemented. return false; } void VisitObject(HeapObject obj) final; private: class RootMarkingVisitor; static const int kNumMarkers = 8; static const int kMainMarker = 0; void MarkLiveObjects(); void MarkRootSetInParallel(RootMarkingVisitor* root_visitor); V8_INLINE void MarkRootObject(HeapObject obj); void DrainMarkingWorklist(); void TraceFragmentation(); void ClearNonLiveReferences(); void EvacuatePrologue(); void EvacuateEpilogue(); void Evacuate(); void EvacuatePagesInParallel(); void UpdatePointersAfterEvacuation(); std::unique_ptr<UpdatingItem> CreateToSpaceUpdatingItem(MemoryChunk* chunk, Address start, Address end); int CollectToSpaceUpdatingItems( std::vector<std::unique_ptr<UpdatingItem>>* items); void SweepArrayBufferExtensions(); std::unique_ptr<YoungGenerationMainMarkingVisitor> main_marking_visitor_; base::Semaphore page_parallel_job_semaphore_; std::vector<Page*> new_space_evacuation_pages_; std::vector<Page*> promoted_pages_; std::vector<LargePage*> promoted_large_pages_; friend class YoungGenerationMarkingTask; friend class YoungGenerationMarkingJob; friend class YoungGenerationMainMarkingVisitor; }; } // namespace internal } // namespace v8 #endif // V8_HEAP_MARK_COMPACT_H_