/* * Copyright (c) 2001, 2010, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ # include "incls/_precompiled.incl" # include "incls/_referenceProcessor.cpp.incl" ReferencePolicy* ReferenceProcessor::_always_clear_soft_ref_policy = NULL; ReferencePolicy* ReferenceProcessor::_default_soft_ref_policy = NULL; oop ReferenceProcessor::_sentinelRef = NULL; const int subclasses_of_ref = REF_PHANTOM - REF_OTHER; // List of discovered references. class DiscoveredList { public: DiscoveredList() : _len(0), _compressed_head(0), _oop_head(NULL) { } oop head() const { return UseCompressedOops ? oopDesc::decode_heap_oop_not_null(_compressed_head) : _oop_head; } HeapWord* adr_head() { return UseCompressedOops ? (HeapWord*)&_compressed_head : (HeapWord*)&_oop_head; } void set_head(oop o) { if (UseCompressedOops) { // Must compress the head ptr. _compressed_head = oopDesc::encode_heap_oop_not_null(o); } else { _oop_head = o; } } bool empty() const { return head() == ReferenceProcessor::sentinel_ref(); } size_t length() { return _len; } void set_length(size_t len) { _len = len; } void inc_length(size_t inc) { _len += inc; assert(_len > 0, "Error"); } void dec_length(size_t dec) { _len -= dec; } private: // Set value depending on UseCompressedOops. This could be a template class // but then we have to fix all the instantiations and declarations that use this class. oop _oop_head; narrowOop _compressed_head; size_t _len; }; void referenceProcessor_init() { ReferenceProcessor::init_statics(); } void ReferenceProcessor::init_statics() { assert(_sentinelRef == NULL, "should be initialized precisely once"); EXCEPTION_MARK; _sentinelRef = instanceKlass::cast( SystemDictionary::Reference_klass())-> allocate_permanent_instance(THREAD); // Initialize the master soft ref clock. java_lang_ref_SoftReference::set_clock(os::javaTimeMillis()); if (HAS_PENDING_EXCEPTION) { Handle ex(THREAD, PENDING_EXCEPTION); vm_exit_during_initialization(ex); } assert(_sentinelRef != NULL && _sentinelRef->is_oop(), "Just constructed it!"); _always_clear_soft_ref_policy = new AlwaysClearPolicy(); _default_soft_ref_policy = new COMPILER2_PRESENT(LRUMaxHeapPolicy()) NOT_COMPILER2(LRUCurrentHeapPolicy()); if (_always_clear_soft_ref_policy == NULL || _default_soft_ref_policy == NULL) { vm_exit_during_initialization("Could not allocate reference policy object"); } guarantee(RefDiscoveryPolicy == ReferenceBasedDiscovery || RefDiscoveryPolicy == ReferentBasedDiscovery, "Unrecongnized RefDiscoveryPolicy"); } ReferenceProcessor* ReferenceProcessor::create_ref_processor(MemRegion span, bool atomic_discovery, bool mt_discovery, BoolObjectClosure* is_alive_non_header, int parallel_gc_threads, bool mt_processing, bool dl_needs_barrier) { int mt_degree = 1; if (parallel_gc_threads > 1) { mt_degree = parallel_gc_threads; } ReferenceProcessor* rp = new ReferenceProcessor(span, atomic_discovery, mt_discovery, mt_degree, mt_processing && (parallel_gc_threads > 0), dl_needs_barrier); if (rp == NULL) { vm_exit_during_initialization("Could not allocate ReferenceProcessor object"); } rp->set_is_alive_non_header(is_alive_non_header); rp->setup_policy(false /* default soft ref policy */); return rp; } ReferenceProcessor::ReferenceProcessor(MemRegion span, bool atomic_discovery, bool mt_discovery, int mt_degree, bool mt_processing, bool discovered_list_needs_barrier) : _discovering_refs(false), _enqueuing_is_done(false), _is_alive_non_header(NULL), _discovered_list_needs_barrier(discovered_list_needs_barrier), _bs(NULL), _processing_is_mt(mt_processing), _next_id(0) { _span = span; _discovery_is_atomic = atomic_discovery; _discovery_is_mt = mt_discovery; _num_q = mt_degree; _max_num_q = mt_degree; _discoveredSoftRefs = NEW_C_HEAP_ARRAY(DiscoveredList, _max_num_q * subclasses_of_ref); if (_discoveredSoftRefs == NULL) { vm_exit_during_initialization("Could not allocated RefProc Array"); } _discoveredWeakRefs = &_discoveredSoftRefs[_max_num_q]; _discoveredFinalRefs = &_discoveredWeakRefs[_max_num_q]; _discoveredPhantomRefs = &_discoveredFinalRefs[_max_num_q]; assert(sentinel_ref() != NULL, "_sentinelRef is NULL"); // Initialized all entries to _sentinelRef for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { _discoveredSoftRefs[i].set_head(sentinel_ref()); _discoveredSoftRefs[i].set_length(0); } // If we do barreirs, cache a copy of the barrier set. if (discovered_list_needs_barrier) { _bs = Universe::heap()->barrier_set(); } } #ifndef PRODUCT void ReferenceProcessor::verify_no_references_recorded() { guarantee(!_discovering_refs, "Discovering refs?"); for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { guarantee(_discoveredSoftRefs[i].empty(), "Found non-empty discovered list"); } } #endif void ReferenceProcessor::weak_oops_do(OopClosure* f) { // Should this instead be // for (int i = 0; i < subclasses_of_ref; i++_ { // for (int j = 0; j < _num_q; j++) { // int index = i * _max_num_q + j; for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { if (UseCompressedOops) { f->do_oop((narrowOop*)_discoveredSoftRefs[i].adr_head()); } else { f->do_oop((oop*)_discoveredSoftRefs[i].adr_head()); } } } void ReferenceProcessor::oops_do(OopClosure* f) { f->do_oop(adr_sentinel_ref()); } void ReferenceProcessor::update_soft_ref_master_clock() { // Update (advance) the soft ref master clock field. This must be done // after processing the soft ref list. jlong now = os::javaTimeMillis(); jlong clock = java_lang_ref_SoftReference::clock(); NOT_PRODUCT( if (now < clock) { warning("time warp: %d to %d", clock, now); } ) // In product mode, protect ourselves from system time being adjusted // externally and going backward; see note in the implementation of // GenCollectedHeap::time_since_last_gc() for the right way to fix // this uniformly throughout the VM; see bug-id 4741166. XXX if (now > clock) { java_lang_ref_SoftReference::set_clock(now); } // Else leave clock stalled at its old value until time progresses // past clock value. } void ReferenceProcessor::process_discovered_references( BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc, AbstractRefProcTaskExecutor* task_executor) { NOT_PRODUCT(verify_ok_to_handle_reflists()); assert(!enqueuing_is_done(), "If here enqueuing should not be complete"); // Stop treating discovered references specially. disable_discovery(); bool trace_time = PrintGCDetails && PrintReferenceGC; // Soft references { TraceTime tt("SoftReference", trace_time, false, gclog_or_tty); process_discovered_reflist(_discoveredSoftRefs, _current_soft_ref_policy, true, is_alive, keep_alive, complete_gc, task_executor); } update_soft_ref_master_clock(); // Weak references { TraceTime tt("WeakReference", trace_time, false, gclog_or_tty); process_discovered_reflist(_discoveredWeakRefs, NULL, true, is_alive, keep_alive, complete_gc, task_executor); } // Final references { TraceTime tt("FinalReference", trace_time, false, gclog_or_tty); process_discovered_reflist(_discoveredFinalRefs, NULL, false, is_alive, keep_alive, complete_gc, task_executor); } // Phantom references { TraceTime tt("PhantomReference", trace_time, false, gclog_or_tty); process_discovered_reflist(_discoveredPhantomRefs, NULL, false, is_alive, keep_alive, complete_gc, task_executor); } // Weak global JNI references. It would make more sense (semantically) to // traverse these simultaneously with the regular weak references above, but // that is not how the JDK1.2 specification is. See #4126360. Native code can // thus use JNI weak references to circumvent the phantom references and // resurrect a "post-mortem" object. { TraceTime tt("JNI Weak Reference", trace_time, false, gclog_or_tty); if (task_executor != NULL) { task_executor->set_single_threaded_mode(); } process_phaseJNI(is_alive, keep_alive, complete_gc); } } #ifndef PRODUCT // Calculate the number of jni handles. uint ReferenceProcessor::count_jni_refs() { class AlwaysAliveClosure: public BoolObjectClosure { public: virtual bool do_object_b(oop obj) { return true; } virtual void do_object(oop obj) { assert(false, "Don't call"); } }; class CountHandleClosure: public OopClosure { private: int _count; public: CountHandleClosure(): _count(0) {} void do_oop(oop* unused) { _count++; } void do_oop(narrowOop* unused) { ShouldNotReachHere(); } int count() { return _count; } }; CountHandleClosure global_handle_count; AlwaysAliveClosure always_alive; JNIHandles::weak_oops_do(&always_alive, &global_handle_count); return global_handle_count.count(); } #endif void ReferenceProcessor::process_phaseJNI(BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc) { #ifndef PRODUCT if (PrintGCDetails && PrintReferenceGC) { unsigned int count = count_jni_refs(); gclog_or_tty->print(", %u refs", count); } #endif JNIHandles::weak_oops_do(is_alive, keep_alive); // Finally remember to keep sentinel around keep_alive->do_oop(adr_sentinel_ref()); complete_gc->do_void(); } template bool enqueue_discovered_ref_helper(ReferenceProcessor* ref, AbstractRefProcTaskExecutor* task_executor) { // Remember old value of pending references list T* pending_list_addr = (T*)java_lang_ref_Reference::pending_list_addr(); T old_pending_list_value = *pending_list_addr; // Enqueue references that are not made active again, and // clear the decks for the next collection (cycle). ref->enqueue_discovered_reflists((HeapWord*)pending_list_addr, task_executor); // Do the oop-check on pending_list_addr missed in // enqueue_discovered_reflist. We should probably // do a raw oop_check so that future such idempotent // oop_stores relying on the oop-check side-effect // may be elided automatically and safely without // affecting correctness. oop_store(pending_list_addr, oopDesc::load_decode_heap_oop(pending_list_addr)); // Stop treating discovered references specially. ref->disable_discovery(); // Return true if new pending references were added return old_pending_list_value != *pending_list_addr; } bool ReferenceProcessor::enqueue_discovered_references(AbstractRefProcTaskExecutor* task_executor) { NOT_PRODUCT(verify_ok_to_handle_reflists()); if (UseCompressedOops) { return enqueue_discovered_ref_helper(this, task_executor); } else { return enqueue_discovered_ref_helper(this, task_executor); } } void ReferenceProcessor::enqueue_discovered_reflist(DiscoveredList& refs_list, HeapWord* pending_list_addr) { // Given a list of refs linked through the "discovered" field // (java.lang.ref.Reference.discovered) chain them through the // "next" field (java.lang.ref.Reference.next) and prepend // to the pending list. if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr("ReferenceProcessor::enqueue_discovered_reflist list " INTPTR_FORMAT, (address)refs_list.head()); } oop obj = refs_list.head(); // Walk down the list, copying the discovered field into // the next field and clearing it (except for the last // non-sentinel object which is treated specially to avoid // confusion with an active reference). while (obj != sentinel_ref()) { assert(obj->is_instanceRef(), "should be reference object"); oop next = java_lang_ref_Reference::discovered(obj); if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr(" obj " INTPTR_FORMAT "/next " INTPTR_FORMAT, obj, next); } assert(java_lang_ref_Reference::next(obj) == NULL, "The reference should not be enqueued"); if (next == sentinel_ref()) { // obj is last // Swap refs_list into pendling_list_addr and // set obj's next to what we read from pending_list_addr. oop old = oopDesc::atomic_exchange_oop(refs_list.head(), pending_list_addr); // Need oop_check on pending_list_addr above; // see special oop-check code at the end of // enqueue_discovered_reflists() further below. if (old == NULL) { // obj should be made to point to itself, since // pending list was empty. java_lang_ref_Reference::set_next(obj, obj); } else { java_lang_ref_Reference::set_next(obj, old); } } else { java_lang_ref_Reference::set_next(obj, next); } java_lang_ref_Reference::set_discovered(obj, (oop) NULL); obj = next; } } // Parallel enqueue task class RefProcEnqueueTask: public AbstractRefProcTaskExecutor::EnqueueTask { public: RefProcEnqueueTask(ReferenceProcessor& ref_processor, DiscoveredList discovered_refs[], HeapWord* pending_list_addr, oop sentinel_ref, int n_queues) : EnqueueTask(ref_processor, discovered_refs, pending_list_addr, sentinel_ref, n_queues) { } virtual void work(unsigned int work_id) { assert(work_id < (unsigned int)_ref_processor.num_q(), "Index out-of-bounds"); // Simplest first cut: static partitioning. int index = work_id; // The increment on "index" must correspond to the maximum number of queues // (n_queues) with which that ReferenceProcessor was created. That // is because of the "clever" way the discovered references lists were // allocated and are indexed into. That number is ParallelGCThreads // currently. Assert that. assert(_n_queues == (int) ParallelGCThreads, "Different number not expected"); for (int j = 0; j < subclasses_of_ref; j++, index += _n_queues) { _ref_processor.enqueue_discovered_reflist( _refs_lists[index], _pending_list_addr); _refs_lists[index].set_head(_sentinel_ref); _refs_lists[index].set_length(0); } } }; // Enqueue references that are not made active again void ReferenceProcessor::enqueue_discovered_reflists(HeapWord* pending_list_addr, AbstractRefProcTaskExecutor* task_executor) { if (_processing_is_mt && task_executor != NULL) { // Parallel code RefProcEnqueueTask tsk(*this, _discoveredSoftRefs, pending_list_addr, sentinel_ref(), _max_num_q); task_executor->execute(tsk); } else { // Serial code: call the parent class's implementation for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { enqueue_discovered_reflist(_discoveredSoftRefs[i], pending_list_addr); _discoveredSoftRefs[i].set_head(sentinel_ref()); _discoveredSoftRefs[i].set_length(0); } } } // Iterator for the list of discovered references. class DiscoveredListIterator { public: inline DiscoveredListIterator(DiscoveredList& refs_list, OopClosure* keep_alive, BoolObjectClosure* is_alive); // End Of List. inline bool has_next() const { return _next != ReferenceProcessor::sentinel_ref(); } // Get oop to the Reference object. inline oop obj() const { return _ref; } // Get oop to the referent object. inline oop referent() const { return _referent; } // Returns true if referent is alive. inline bool is_referent_alive() const; // Loads data for the current reference. // The "allow_null_referent" argument tells us to allow for the possibility // of a NULL referent in the discovered Reference object. This typically // happens in the case of concurrent collectors that may have done the // discovery concurrently, or interleaved, with mutator execution. inline void load_ptrs(DEBUG_ONLY(bool allow_null_referent)); // Move to the next discovered reference. inline void next(); // Remove the current reference from the list inline void remove(); // Make the Reference object active again. inline void make_active() { java_lang_ref_Reference::set_next(_ref, NULL); } // Make the referent alive. inline void make_referent_alive() { if (UseCompressedOops) { _keep_alive->do_oop((narrowOop*)_referent_addr); } else { _keep_alive->do_oop((oop*)_referent_addr); } } // Update the discovered field. inline void update_discovered() { // First _prev_next ref actually points into DiscoveredList (gross). if (UseCompressedOops) { _keep_alive->do_oop((narrowOop*)_prev_next); } else { _keep_alive->do_oop((oop*)_prev_next); } } // NULL out referent pointer. inline void clear_referent() { oop_store_raw(_referent_addr, NULL); } // Statistics NOT_PRODUCT( inline size_t processed() const { return _processed; } inline size_t removed() const { return _removed; } ) inline void move_to_next(); private: DiscoveredList& _refs_list; HeapWord* _prev_next; oop _ref; HeapWord* _discovered_addr; oop _next; HeapWord* _referent_addr; oop _referent; OopClosure* _keep_alive; BoolObjectClosure* _is_alive; DEBUG_ONLY( oop _first_seen; // cyclic linked list check ) NOT_PRODUCT( size_t _processed; size_t _removed; ) }; inline DiscoveredListIterator::DiscoveredListIterator(DiscoveredList& refs_list, OopClosure* keep_alive, BoolObjectClosure* is_alive) : _refs_list(refs_list), _prev_next(refs_list.adr_head()), _ref(refs_list.head()), #ifdef ASSERT _first_seen(refs_list.head()), #endif #ifndef PRODUCT _processed(0), _removed(0), #endif _next(refs_list.head()), _keep_alive(keep_alive), _is_alive(is_alive) { } inline bool DiscoveredListIterator::is_referent_alive() const { return _is_alive->do_object_b(_referent); } inline void DiscoveredListIterator::load_ptrs(DEBUG_ONLY(bool allow_null_referent)) { _discovered_addr = java_lang_ref_Reference::discovered_addr(_ref); oop discovered = java_lang_ref_Reference::discovered(_ref); assert(_discovered_addr && discovered->is_oop_or_null(), "discovered field is bad"); _next = discovered; _referent_addr = java_lang_ref_Reference::referent_addr(_ref); _referent = java_lang_ref_Reference::referent(_ref); assert(Universe::heap()->is_in_reserved_or_null(_referent), "Wrong oop found in java.lang.Reference object"); assert(allow_null_referent ? _referent->is_oop_or_null() : _referent->is_oop(), "bad referent"); } inline void DiscoveredListIterator::next() { _prev_next = _discovered_addr; move_to_next(); } inline void DiscoveredListIterator::remove() { assert(_ref->is_oop(), "Dropping a bad reference"); oop_store_raw(_discovered_addr, NULL); // First _prev_next ref actually points into DiscoveredList (gross). if (UseCompressedOops) { // Remove Reference object from list. oopDesc::encode_store_heap_oop_not_null((narrowOop*)_prev_next, _next); } else { // Remove Reference object from list. oopDesc::store_heap_oop((oop*)_prev_next, _next); } NOT_PRODUCT(_removed++); _refs_list.dec_length(1); } inline void DiscoveredListIterator::move_to_next() { _ref = _next; assert(_ref != _first_seen, "cyclic ref_list found"); NOT_PRODUCT(_processed++); } // NOTE: process_phase*() are largely similar, and at a high level // merely iterate over the extant list applying a predicate to // each of its elements and possibly removing that element from the // list and applying some further closures to that element. // We should consider the possibility of replacing these // process_phase*() methods by abstracting them into // a single general iterator invocation that receives appropriate // closures that accomplish this work. // (SoftReferences only) Traverse the list and remove any SoftReferences whose // referents are not alive, but that should be kept alive for policy reasons. // Keep alive the transitive closure of all such referents. void ReferenceProcessor::process_phase1(DiscoveredList& refs_list, ReferencePolicy* policy, BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc) { assert(policy != NULL, "Must have a non-NULL policy"); DiscoveredListIterator iter(refs_list, keep_alive, is_alive); // Decide which softly reachable refs should be kept alive. while (iter.has_next()) { iter.load_ptrs(DEBUG_ONLY(!discovery_is_atomic() /* allow_null_referent */)); bool referent_is_dead = (iter.referent() != NULL) && !iter.is_referent_alive(); if (referent_is_dead && !policy->should_clear_reference(iter.obj())) { if (TraceReferenceGC) { gclog_or_tty->print_cr("Dropping reference (" INTPTR_FORMAT ": %s" ") by policy", iter.obj(), iter.obj()->blueprint()->internal_name()); } // Remove Reference object from list iter.remove(); // Make the Reference object active again iter.make_active(); // keep the referent around iter.make_referent_alive(); iter.move_to_next(); } else { iter.next(); } } // Close the reachable set complete_gc->do_void(); NOT_PRODUCT( if (PrintGCDetails && TraceReferenceGC) { gclog_or_tty->print_cr(" Dropped %d dead Refs out of %d " "discovered Refs by policy list " INTPTR_FORMAT, iter.removed(), iter.processed(), (address)refs_list.head()); } ) } // Traverse the list and remove any Refs that are not active, or // whose referents are either alive or NULL. void ReferenceProcessor::pp2_work(DiscoveredList& refs_list, BoolObjectClosure* is_alive, OopClosure* keep_alive) { assert(discovery_is_atomic(), "Error"); DiscoveredListIterator iter(refs_list, keep_alive, is_alive); while (iter.has_next()) { iter.load_ptrs(DEBUG_ONLY(false /* allow_null_referent */)); DEBUG_ONLY(oop next = java_lang_ref_Reference::next(iter.obj());) assert(next == NULL, "Should not discover inactive Reference"); if (iter.is_referent_alive()) { if (TraceReferenceGC) { gclog_or_tty->print_cr("Dropping strongly reachable reference (" INTPTR_FORMAT ": %s)", iter.obj(), iter.obj()->blueprint()->internal_name()); } // The referent is reachable after all. // Remove Reference object from list. iter.remove(); // Update the referent pointer as necessary: Note that this // should not entail any recursive marking because the // referent must already have been traversed. iter.make_referent_alive(); iter.move_to_next(); } else { iter.next(); } } NOT_PRODUCT( if (PrintGCDetails && TraceReferenceGC) { gclog_or_tty->print_cr(" Dropped %d active Refs out of %d " "Refs in discovered list " INTPTR_FORMAT, iter.removed(), iter.processed(), (address)refs_list.head()); } ) } void ReferenceProcessor::pp2_work_concurrent_discovery(DiscoveredList& refs_list, BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc) { assert(!discovery_is_atomic(), "Error"); DiscoveredListIterator iter(refs_list, keep_alive, is_alive); while (iter.has_next()) { iter.load_ptrs(DEBUG_ONLY(true /* allow_null_referent */)); HeapWord* next_addr = java_lang_ref_Reference::next_addr(iter.obj()); oop next = java_lang_ref_Reference::next(iter.obj()); if ((iter.referent() == NULL || iter.is_referent_alive() || next != NULL)) { assert(next->is_oop_or_null(), "bad next field"); // Remove Reference object from list iter.remove(); // Trace the cohorts iter.make_referent_alive(); if (UseCompressedOops) { keep_alive->do_oop((narrowOop*)next_addr); } else { keep_alive->do_oop((oop*)next_addr); } iter.move_to_next(); } else { iter.next(); } } // Now close the newly reachable set complete_gc->do_void(); NOT_PRODUCT( if (PrintGCDetails && TraceReferenceGC) { gclog_or_tty->print_cr(" Dropped %d active Refs out of %d " "Refs in discovered list " INTPTR_FORMAT, iter.removed(), iter.processed(), (address)refs_list.head()); } ) } // Traverse the list and process the referents, by either // clearing them or keeping them (and their reachable // closure) alive. void ReferenceProcessor::process_phase3(DiscoveredList& refs_list, bool clear_referent, BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc) { ResourceMark rm; DiscoveredListIterator iter(refs_list, keep_alive, is_alive); while (iter.has_next()) { iter.update_discovered(); iter.load_ptrs(DEBUG_ONLY(false /* allow_null_referent */)); if (clear_referent) { // NULL out referent pointer iter.clear_referent(); } else { // keep the referent around iter.make_referent_alive(); } if (TraceReferenceGC) { gclog_or_tty->print_cr("Adding %sreference (" INTPTR_FORMAT ": %s) as pending", clear_referent ? "cleared " : "", iter.obj(), iter.obj()->blueprint()->internal_name()); } assert(iter.obj()->is_oop(UseConcMarkSweepGC), "Adding a bad reference"); iter.next(); } // Remember to keep sentinel pointer around iter.update_discovered(); // Close the reachable set complete_gc->do_void(); } void ReferenceProcessor::abandon_partial_discovered_list(DiscoveredList& refs_list) { oop obj = refs_list.head(); while (obj != sentinel_ref()) { oop discovered = java_lang_ref_Reference::discovered(obj); java_lang_ref_Reference::set_discovered_raw(obj, NULL); obj = discovered; } refs_list.set_head(sentinel_ref()); refs_list.set_length(0); } void ReferenceProcessor::abandon_partial_discovery() { // loop over the lists for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { if (TraceReferenceGC && PrintGCDetails && ((i % _max_num_q) == 0)) { gclog_or_tty->print_cr( "\nAbandoning %s discovered list", list_name(i)); } abandon_partial_discovered_list(_discoveredSoftRefs[i]); } } class RefProcPhase1Task: public AbstractRefProcTaskExecutor::ProcessTask { public: RefProcPhase1Task(ReferenceProcessor& ref_processor, DiscoveredList refs_lists[], ReferencePolicy* policy, bool marks_oops_alive) : ProcessTask(ref_processor, refs_lists, marks_oops_alive), _policy(policy) { } virtual void work(unsigned int i, BoolObjectClosure& is_alive, OopClosure& keep_alive, VoidClosure& complete_gc) { Thread* thr = Thread::current(); int refs_list_index = ((WorkerThread*)thr)->id(); _ref_processor.process_phase1(_refs_lists[refs_list_index], _policy, &is_alive, &keep_alive, &complete_gc); } private: ReferencePolicy* _policy; }; class RefProcPhase2Task: public AbstractRefProcTaskExecutor::ProcessTask { public: RefProcPhase2Task(ReferenceProcessor& ref_processor, DiscoveredList refs_lists[], bool marks_oops_alive) : ProcessTask(ref_processor, refs_lists, marks_oops_alive) { } virtual void work(unsigned int i, BoolObjectClosure& is_alive, OopClosure& keep_alive, VoidClosure& complete_gc) { _ref_processor.process_phase2(_refs_lists[i], &is_alive, &keep_alive, &complete_gc); } }; class RefProcPhase3Task: public AbstractRefProcTaskExecutor::ProcessTask { public: RefProcPhase3Task(ReferenceProcessor& ref_processor, DiscoveredList refs_lists[], bool clear_referent, bool marks_oops_alive) : ProcessTask(ref_processor, refs_lists, marks_oops_alive), _clear_referent(clear_referent) { } virtual void work(unsigned int i, BoolObjectClosure& is_alive, OopClosure& keep_alive, VoidClosure& complete_gc) { // Don't use "refs_list_index" calculated in this way because // balance_queues() has moved the Ref's into the first n queues. // Thread* thr = Thread::current(); // int refs_list_index = ((WorkerThread*)thr)->id(); // _ref_processor.process_phase3(_refs_lists[refs_list_index], _clear_referent, _ref_processor.process_phase3(_refs_lists[i], _clear_referent, &is_alive, &keep_alive, &complete_gc); } private: bool _clear_referent; }; // Balances reference queues. // Move entries from all queues[0, 1, ..., _max_num_q-1] to // queues[0, 1, ..., _num_q-1] because only the first _num_q // corresponding to the active workers will be processed. void ReferenceProcessor::balance_queues(DiscoveredList ref_lists[]) { // calculate total length size_t total_refs = 0; if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr("\nBalance ref_lists "); } for (int i = 0; i < _max_num_q; ++i) { total_refs += ref_lists[i].length(); if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print("%d ", ref_lists[i].length()); } } if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr(" = %d", total_refs); } size_t avg_refs = total_refs / _num_q + 1; int to_idx = 0; for (int from_idx = 0; from_idx < _max_num_q; from_idx++) { bool move_all = false; if (from_idx >= _num_q) { move_all = ref_lists[from_idx].length() > 0; } while ((ref_lists[from_idx].length() > avg_refs) || move_all) { assert(to_idx < _num_q, "Sanity Check!"); if (ref_lists[to_idx].length() < avg_refs) { // move superfluous refs size_t refs_to_move; // Move all the Ref's if the from queue will not be processed. if (move_all) { refs_to_move = MIN2(ref_lists[from_idx].length(), avg_refs - ref_lists[to_idx].length()); } else { refs_to_move = MIN2(ref_lists[from_idx].length() - avg_refs, avg_refs - ref_lists[to_idx].length()); } oop move_head = ref_lists[from_idx].head(); oop move_tail = move_head; oop new_head = move_head; // find an element to split the list on for (size_t j = 0; j < refs_to_move; ++j) { move_tail = new_head; new_head = java_lang_ref_Reference::discovered(new_head); } java_lang_ref_Reference::set_discovered(move_tail, ref_lists[to_idx].head()); ref_lists[to_idx].set_head(move_head); ref_lists[to_idx].inc_length(refs_to_move); ref_lists[from_idx].set_head(new_head); ref_lists[from_idx].dec_length(refs_to_move); if (ref_lists[from_idx].length() == 0) { break; } } else { to_idx = (to_idx + 1) % _num_q; } } } #ifdef ASSERT size_t balanced_total_refs = 0; for (int i = 0; i < _max_num_q; ++i) { balanced_total_refs += ref_lists[i].length(); if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print("%d ", ref_lists[i].length()); } } if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr(" = %d", balanced_total_refs); gclog_or_tty->flush(); } assert(total_refs == balanced_total_refs, "Balancing was incomplete"); #endif } void ReferenceProcessor::balance_all_queues() { balance_queues(_discoveredSoftRefs); balance_queues(_discoveredWeakRefs); balance_queues(_discoveredFinalRefs); balance_queues(_discoveredPhantomRefs); } void ReferenceProcessor::process_discovered_reflist( DiscoveredList refs_lists[], ReferencePolicy* policy, bool clear_referent, BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc, AbstractRefProcTaskExecutor* task_executor) { bool mt_processing = task_executor != NULL && _processing_is_mt; // If discovery used MT and a dynamic number of GC threads, then // the queues must be balanced for correctness if fewer than the // maximum number of queues were used. The number of queue used // during discovery may be different than the number to be used // for processing so don't depend of _num_q < _max_num_q as part // of the test. bool must_balance = _discovery_is_mt; if ((mt_processing && ParallelRefProcBalancingEnabled) || must_balance) { balance_queues(refs_lists); } if (PrintReferenceGC && PrintGCDetails) { size_t total = 0; for (int i = 0; i < _num_q; ++i) { total += refs_lists[i].length(); } gclog_or_tty->print(", %u refs", total); } // Phase 1 (soft refs only): // . Traverse the list and remove any SoftReferences whose // referents are not alive, but that should be kept alive for // policy reasons. Keep alive the transitive closure of all // such referents. if (policy != NULL) { if (mt_processing) { RefProcPhase1Task phase1(*this, refs_lists, policy, true /*marks_oops_alive*/); task_executor->execute(phase1); } else { for (int i = 0; i < _num_q; i++) { process_phase1(refs_lists[i], policy, is_alive, keep_alive, complete_gc); } } } else { // policy == NULL assert(refs_lists != _discoveredSoftRefs, "Policy must be specified for soft references."); } // Phase 2: // . Traverse the list and remove any refs whose referents are alive. if (mt_processing) { RefProcPhase2Task phase2(*this, refs_lists, !discovery_is_atomic() /*marks_oops_alive*/); task_executor->execute(phase2); } else { for (int i = 0; i < _num_q; i++) { process_phase2(refs_lists[i], is_alive, keep_alive, complete_gc); } } // Phase 3: // . Traverse the list and process referents as appropriate. if (mt_processing) { RefProcPhase3Task phase3(*this, refs_lists, clear_referent, true /*marks_oops_alive*/); task_executor->execute(phase3); } else { for (int i = 0; i < _num_q; i++) { process_phase3(refs_lists[i], clear_referent, is_alive, keep_alive, complete_gc); } } } void ReferenceProcessor::clean_up_discovered_references() { // loop over the lists // Should this instead be // for (int i = 0; i < subclasses_of_ref; i++_ { // for (int j = 0; j < _num_q; j++) { // int index = i * _max_num_q + j; for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { if (TraceReferenceGC && PrintGCDetails && ((i % _num_q) == 0)) { gclog_or_tty->print_cr( "\nScrubbing %s discovered list of Null referents", list_name(i)); } clean_up_discovered_reflist(_discoveredSoftRefs[i]); } } void ReferenceProcessor::clean_up_discovered_reflist(DiscoveredList& refs_list) { assert(!discovery_is_atomic(), "Else why call this method?"); DiscoveredListIterator iter(refs_list, NULL, NULL); while (iter.has_next()) { iter.load_ptrs(DEBUG_ONLY(true /* allow_null_referent */)); oop next = java_lang_ref_Reference::next(iter.obj()); assert(next->is_oop_or_null(), "bad next field"); // If referent has been cleared or Reference is not active, // drop it. if (iter.referent() == NULL || next != NULL) { debug_only( if (PrintGCDetails && TraceReferenceGC) { gclog_or_tty->print_cr("clean_up_discovered_list: Dropping Reference: " INTPTR_FORMAT " with next field: " INTPTR_FORMAT " and referent: " INTPTR_FORMAT, iter.obj(), next, iter.referent()); } ) // Remove Reference object from list iter.remove(); iter.move_to_next(); } else { iter.next(); } } NOT_PRODUCT( if (PrintGCDetails && TraceReferenceGC) { gclog_or_tty->print( " Removed %d Refs with NULL referents out of %d discovered Refs", iter.removed(), iter.processed()); } ) } inline DiscoveredList* ReferenceProcessor::get_discovered_list(ReferenceType rt) { int id = 0; // Determine the queue index to use for this object. if (_discovery_is_mt) { // During a multi-threaded discovery phase, // each thread saves to its "own" list. Thread* thr = Thread::current(); assert(thr->is_GC_task_thread(), "Dubious cast from Thread* to WorkerThread*?"); id = ((WorkerThread*)thr)->id(); } else { // single-threaded discovery, we save in round-robin // fashion to each of the lists. if (_processing_is_mt) { id = next_id(); } } assert(0 <= id && id < _max_num_q, "Id is out-of-bounds (call Freud?)"); // Get the discovered queue to which we will add DiscoveredList* list = NULL; switch (rt) { case REF_OTHER: // Unknown reference type, no special treatment break; case REF_SOFT: list = &_discoveredSoftRefs[id]; break; case REF_WEAK: list = &_discoveredWeakRefs[id]; break; case REF_FINAL: list = &_discoveredFinalRefs[id]; break; case REF_PHANTOM: list = &_discoveredPhantomRefs[id]; break; case REF_NONE: // we should not reach here if we are an instanceRefKlass default: ShouldNotReachHere(); } if (TraceReferenceGC && PrintGCDetails) { gclog_or_tty->print_cr("Thread %d gets list " INTPTR_FORMAT, id, list); } return list; } inline void ReferenceProcessor::add_to_discovered_list_mt(DiscoveredList& refs_list, oop obj, HeapWord* discovered_addr) { assert(_discovery_is_mt, "!_discovery_is_mt should have been handled by caller"); // First we must make sure this object is only enqueued once. CAS in a non null // discovered_addr. oop current_head = refs_list.head(); // Note: In the case of G1, this specific pre-barrier is strictly // not necessary because the only case we are interested in // here is when *discovered_addr is NULL (see the CAS further below), // so this will expand to nothing. As a result, we have manually // elided this out for G1, but left in the test for some future // collector that might have need for a pre-barrier here. if (_discovered_list_needs_barrier && !UseG1GC) { if (UseCompressedOops) { _bs->write_ref_field_pre((narrowOop*)discovered_addr, current_head); } else { _bs->write_ref_field_pre((oop*)discovered_addr, current_head); } guarantee(false, "Need to check non-G1 collector"); } oop retest = oopDesc::atomic_compare_exchange_oop(current_head, discovered_addr, NULL); if (retest == NULL) { // This thread just won the right to enqueue the object. // We have separate lists for enqueueing so no synchronization // is necessary. refs_list.set_head(obj); refs_list.inc_length(1); if (_discovered_list_needs_barrier) { _bs->write_ref_field((void*)discovered_addr, current_head); } } else { // If retest was non NULL, another thread beat us to it: // The reference has already been discovered... if (TraceReferenceGC) { gclog_or_tty->print_cr("Already enqueued reference (" INTPTR_FORMAT ": %s)", obj, obj->blueprint()->internal_name()); } } } // We mention two of several possible choices here: // #0: if the reference object is not in the "originating generation" // (or part of the heap being collected, indicated by our "span" // we don't treat it specially (i.e. we scan it as we would // a normal oop, treating its references as strong references). // This means that references can't be enqueued unless their // referent is also in the same span. This is the simplest, // most "local" and most conservative approach, albeit one // that may cause weak references to be enqueued least promptly. // We call this choice the "ReferenceBasedDiscovery" policy. // #1: the reference object may be in any generation (span), but if // the referent is in the generation (span) being currently collected // then we can discover the reference object, provided // the object has not already been discovered by // a different concurrently running collector (as may be the // case, for instance, if the reference object is in CMS and // the referent in DefNewGeneration), and provided the processing // of this reference object by the current collector will // appear atomic to every other collector in the system. // (Thus, for instance, a concurrent collector may not // discover references in other generations even if the // referent is in its own generation). This policy may, // in certain cases, enqueue references somewhat sooner than // might Policy #0 above, but at marginally increased cost // and complexity in processing these references. // We call this choice the "RefeferentBasedDiscovery" policy. bool ReferenceProcessor::discover_reference(oop obj, ReferenceType rt) { // We enqueue references only if we are discovering refs // (rather than processing discovered refs). if (!_discovering_refs || !RegisterReferences) { return false; } // We only enqueue active references. oop next = java_lang_ref_Reference::next(obj); if (next != NULL) { return false; } HeapWord* obj_addr = (HeapWord*)obj; if (RefDiscoveryPolicy == ReferenceBasedDiscovery && !_span.contains(obj_addr)) { // Reference is not in the originating generation; // don't treat it specially (i.e. we want to scan it as a normal // object with strong references). return false; } // We only enqueue references whose referents are not (yet) strongly // reachable. if (is_alive_non_header() != NULL) { oop referent = java_lang_ref_Reference::referent(obj); // In the case of non-concurrent discovery, the last // disjunct below should hold. It may not hold in the // case of concurrent discovery because mutators may // concurrently clear() a Reference. assert(UseConcMarkSweepGC || UseG1GC || referent != NULL, "Refs with null referents already filtered"); if (is_alive_non_header()->do_object_b(referent)) { return false; // referent is reachable } } if (rt == REF_SOFT) { // For soft refs we can decide now if these are not // current candidates for clearing, in which case we // can mark through them now, rather than delaying that // to the reference-processing phase. Since all current // time-stamp policies advance the soft-ref clock only // at a major collection cycle, this is always currently // accurate. if (!_current_soft_ref_policy->should_clear_reference(obj)) { return false; } } HeapWord* const discovered_addr = java_lang_ref_Reference::discovered_addr(obj); const oop discovered = java_lang_ref_Reference::discovered(obj); assert(discovered->is_oop_or_null(), "bad discovered field"); if (discovered != NULL) { // The reference has already been discovered... if (TraceReferenceGC) { gclog_or_tty->print_cr("Already enqueued reference (" INTPTR_FORMAT ": %s)", obj, obj->blueprint()->internal_name()); } if (RefDiscoveryPolicy == ReferentBasedDiscovery) { // assumes that an object is not processed twice; // if it's been already discovered it must be on another // generation's discovered list; so we won't discover it. return false; } else { assert(RefDiscoveryPolicy == ReferenceBasedDiscovery, "Unrecognized policy"); // Check assumption that an object is not potentially // discovered twice except by concurrent collectors that potentially // trace the same Reference object twice. assert(UseConcMarkSweepGC, "Only possible with an incremental-update concurrent collector"); return true; } } if (RefDiscoveryPolicy == ReferentBasedDiscovery) { oop referent = java_lang_ref_Reference::referent(obj); assert(referent->is_oop(), "bad referent"); // enqueue if and only if either: // reference is in our span or // we are an atomic collector and referent is in our span if (_span.contains(obj_addr) || (discovery_is_atomic() && _span.contains(referent))) { // should_enqueue = true; } else { return false; } } else { assert(RefDiscoveryPolicy == ReferenceBasedDiscovery && _span.contains(obj_addr), "code inconsistency"); } // Get the right type of discovered queue head. DiscoveredList* list = get_discovered_list(rt); if (list == NULL) { return false; // nothing special needs to be done } if (_discovery_is_mt) { add_to_discovered_list_mt(*list, obj, discovered_addr); } else { // If "_discovered_list_needs_barrier", we do write barriers when // updating the discovered reference list. Otherwise, we do a raw store // here: the field will be visited later when processing the discovered // references. oop current_head = list->head(); // As in the case further above, since we are over-writing a NULL // pre-value, we can safely elide the pre-barrier here for the case of G1. assert(discovered == NULL, "control point invariant"); if (_discovered_list_needs_barrier && !UseG1GC) { // safe to elide for G1 if (UseCompressedOops) { _bs->write_ref_field_pre((narrowOop*)discovered_addr, current_head); } else { _bs->write_ref_field_pre((oop*)discovered_addr, current_head); } guarantee(false, "Need to check non-G1 collector"); } oop_store_raw(discovered_addr, current_head); if (_discovered_list_needs_barrier) { _bs->write_ref_field((void*)discovered_addr, current_head); } list->set_head(obj); list->inc_length(1); } // In the MT discovery case, it is currently possible to see // the following message multiple times if several threads // discover a reference about the same time. Only one will // however have actually added it to the disocvered queue. // One could let add_to_discovered_list_mt() return an // indication for success in queueing (by 1 thread) or // failure (by all other threads), but I decided the extra // code was not worth the effort for something that is // only used for debugging support. if (TraceReferenceGC) { oop referent = java_lang_ref_Reference::referent(obj); if (PrintGCDetails) { gclog_or_tty->print_cr("Enqueued reference (" INTPTR_FORMAT ": %s)", obj, obj->blueprint()->internal_name()); } assert(referent->is_oop(), "Enqueued a bad referent"); } assert(obj->is_oop(), "Enqueued a bad reference"); return true; } // Preclean the discovered references by removing those // whose referents are alive, and by marking from those that // are not active. These lists can be handled here // in any order and, indeed, concurrently. void ReferenceProcessor::preclean_discovered_references( BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc, YieldClosure* yield, bool should_unload_classes) { NOT_PRODUCT(verify_ok_to_handle_reflists()); #ifdef ASSERT bool must_remember_klasses = ClassUnloading && !UseConcMarkSweepGC || CMSClassUnloadingEnabled && UseConcMarkSweepGC || ExplicitGCInvokesConcurrentAndUnloadsClasses && UseConcMarkSweepGC && should_unload_classes; RememberKlassesChecker mx(must_remember_klasses); #endif // Soft references { TraceTime tt("Preclean SoftReferences", PrintGCDetails && PrintReferenceGC, false, gclog_or_tty); for (int i = 0; i < _max_num_q; i++) { if (yield->should_return()) { return; } preclean_discovered_reflist(_discoveredSoftRefs[i], is_alive, keep_alive, complete_gc, yield); } } // Weak references { TraceTime tt("Preclean WeakReferences", PrintGCDetails && PrintReferenceGC, false, gclog_or_tty); for (int i = 0; i < _num_q; i++) { if (yield->should_return()) { return; } preclean_discovered_reflist(_discoveredWeakRefs[i], is_alive, keep_alive, complete_gc, yield); } } // Final references { TraceTime tt("Preclean FinalReferences", PrintGCDetails && PrintReferenceGC, false, gclog_or_tty); for (int i = 0; i < _num_q; i++) { if (yield->should_return()) { return; } preclean_discovered_reflist(_discoveredFinalRefs[i], is_alive, keep_alive, complete_gc, yield); } } // Phantom references { TraceTime tt("Preclean PhantomReferences", PrintGCDetails && PrintReferenceGC, false, gclog_or_tty); for (int i = 0; i < _num_q; i++) { if (yield->should_return()) { return; } preclean_discovered_reflist(_discoveredPhantomRefs[i], is_alive, keep_alive, complete_gc, yield); } } } // Walk the given discovered ref list, and remove all reference objects // whose referents are still alive, whose referents are NULL or which // are not active (have a non-NULL next field). NOTE: When we are // thus precleaning the ref lists (which happens single-threaded today), // we do not disable refs discovery to honour the correct semantics of // java.lang.Reference. As a result, we need to be careful below // that ref removal steps interleave safely with ref discovery steps // (in this thread). void ReferenceProcessor::preclean_discovered_reflist(DiscoveredList& refs_list, BoolObjectClosure* is_alive, OopClosure* keep_alive, VoidClosure* complete_gc, YieldClosure* yield) { DiscoveredListIterator iter(refs_list, keep_alive, is_alive); while (iter.has_next()) { iter.load_ptrs(DEBUG_ONLY(true /* allow_null_referent */)); oop obj = iter.obj(); oop next = java_lang_ref_Reference::next(obj); if (iter.referent() == NULL || iter.is_referent_alive() || next != NULL) { // The referent has been cleared, or is alive, or the Reference is not // active; we need to trace and mark its cohort. if (TraceReferenceGC) { gclog_or_tty->print_cr("Precleaning Reference (" INTPTR_FORMAT ": %s)", iter.obj(), iter.obj()->blueprint()->internal_name()); } // Remove Reference object from list iter.remove(); // Keep alive its cohort. iter.make_referent_alive(); if (UseCompressedOops) { narrowOop* next_addr = (narrowOop*)java_lang_ref_Reference::next_addr(obj); keep_alive->do_oop(next_addr); } else { oop* next_addr = (oop*)java_lang_ref_Reference::next_addr(obj); keep_alive->do_oop(next_addr); } iter.move_to_next(); } else { iter.next(); } } // Close the reachable set complete_gc->do_void(); NOT_PRODUCT( if (PrintGCDetails && PrintReferenceGC) { gclog_or_tty->print_cr(" Dropped %d Refs out of %d " "Refs in discovered list " INTPTR_FORMAT, iter.removed(), iter.processed(), (address)refs_list.head()); } ) } const char* ReferenceProcessor::list_name(int i) { assert(i >= 0 && i <= _max_num_q * subclasses_of_ref, "Out of bounds index"); int j = i / _max_num_q; switch (j) { case 0: return "SoftRef"; case 1: return "WeakRef"; case 2: return "FinalRef"; case 3: return "PhantomRef"; } ShouldNotReachHere(); return NULL; } #ifndef PRODUCT void ReferenceProcessor::verify_ok_to_handle_reflists() { // empty for now } #endif void ReferenceProcessor::verify() { guarantee(sentinel_ref() != NULL && sentinel_ref()->is_oop(), "Lost _sentinelRef"); } #ifndef PRODUCT void ReferenceProcessor::clear_discovered_references() { guarantee(!_discovering_refs, "Discovering refs?"); for (int i = 0; i < _max_num_q * subclasses_of_ref; i++) { oop obj = _discoveredSoftRefs[i].head(); while (obj != sentinel_ref()) { oop next = java_lang_ref_Reference::discovered(obj); java_lang_ref_Reference::set_discovered(obj, (oop) NULL); obj = next; } _discoveredSoftRefs[i].set_head(sentinel_ref()); _discoveredSoftRefs[i].set_length(0); } } #endif // PRODUCT