1 /*
2 * Copyright (c) 2001, 2018, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25 #include "precompiled.hpp"
26 #include "classfile/metadataOnStackMark.hpp"
27 #include "classfile/symbolTable.hpp"
28 #include "code/codeCache.hpp"
29 #include "gc/g1/concurrentMarkThread.inline.hpp"
30 #include "gc/g1/g1CollectedHeap.inline.hpp"
31 #include "gc/g1/g1CollectorState.hpp"
32 #include "gc/g1/g1ConcurrentMark.inline.hpp"
33 #include "gc/g1/g1OopClosures.inline.hpp"
34 #include "gc/g1/g1Policy.hpp"
35 #include "gc/g1/g1RegionMarkStatsCache.inline.hpp"
36 #include "gc/g1/g1StringDedup.hpp"
37 #include "gc/g1/heapRegion.inline.hpp"
38 #include "gc/g1/heapRegionRemSet.hpp"
39 #include "gc/g1/heapRegionSet.inline.hpp"
40 #include "gc/shared/adaptiveSizePolicy.hpp"
41 #include "gc/shared/gcId.hpp"
42 #include "gc/shared/gcTimer.hpp"
43 #include "gc/shared/gcTrace.hpp"
44 #include "gc/shared/gcTraceTime.inline.hpp"
45 #include "gc/shared/genOopClosures.inline.hpp"
46 #include "gc/shared/referencePolicy.hpp"
47 #include "gc/shared/strongRootsScope.hpp"
48 #include "gc/shared/suspendibleThreadSet.hpp"
49 #include "gc/shared/taskqueue.inline.hpp"
50 #include "gc/shared/vmGCOperations.hpp"
51 #include "gc/shared/weakProcessor.hpp"
52 #include "include/jvm.h"
53 #include "logging/log.hpp"
54 #include "memory/allocation.hpp"
55 #include "memory/resourceArea.hpp"
56 #include "oops/access.inline.hpp"
57 #include "oops/oop.inline.hpp"
58 #include "runtime/atomic.hpp"
59 #include "runtime/handles.inline.hpp"
60 #include "runtime/java.hpp"
61 #include "runtime/prefetch.inline.hpp"
62 #include "services/memTracker.hpp"
63 #include "utilities/align.hpp"
64 #include "utilities/growableArray.hpp"
65
66 bool G1CMBitMapClosure::do_addr(HeapWord* const addr) {
67 assert(addr < _cm->finger(), "invariant");
68 assert(addr >= _task->finger(), "invariant");
69
70 // We move that task's local finger along.
71 _task->move_finger_to(addr);
72
73 _task->scan_task_entry(G1TaskQueueEntry::from_oop(oop(addr)));
74 // we only partially drain the local queue and global stack
75 _task->drain_local_queue(true);
76 _task->drain_global_stack(true);
77
78 // if the has_aborted flag has been raised, we need to bail out of
79 // the iteration
80 return !_task->has_aborted();
81 }
82
83 G1CMMarkStack::G1CMMarkStack() :
84 _max_chunk_capacity(0),
85 _base(NULL),
86 _chunk_capacity(0) {
87 set_empty();
88 }
89
90 bool G1CMMarkStack::resize(size_t new_capacity) {
91 assert(is_empty(), "Only resize when stack is empty.");
92 assert(new_capacity <= _max_chunk_capacity,
93 "Trying to resize stack to " SIZE_FORMAT " chunks when the maximum is " SIZE_FORMAT, new_capacity, _max_chunk_capacity);
94
95 TaskQueueEntryChunk* new_base = MmapArrayAllocator<TaskQueueEntryChunk>::allocate_or_null(new_capacity, mtGC);
96
97 if (new_base == NULL) {
98 log_warning(gc)("Failed to reserve memory for new overflow mark stack with " SIZE_FORMAT " chunks and size " SIZE_FORMAT "B.", new_capacity, new_capacity * sizeof(TaskQueueEntryChunk));
99 return false;
100 }
101 // Release old mapping.
102 if (_base != NULL) {
103 MmapArrayAllocator<TaskQueueEntryChunk>::free(_base, _chunk_capacity);
104 }
105
106 _base = new_base;
107 _chunk_capacity = new_capacity;
108 set_empty();
109
110 return true;
111 }
112
113 size_t G1CMMarkStack::capacity_alignment() {
114 return (size_t)lcm(os::vm_allocation_granularity(), sizeof(TaskQueueEntryChunk)) / sizeof(G1TaskQueueEntry);
115 }
116
117 bool G1CMMarkStack::initialize(size_t initial_capacity, size_t max_capacity) {
118 guarantee(_max_chunk_capacity == 0, "G1CMMarkStack already initialized.");
119
120 size_t const TaskEntryChunkSizeInVoidStar = sizeof(TaskQueueEntryChunk) / sizeof(G1TaskQueueEntry);
121
122 _max_chunk_capacity = align_up(max_capacity, capacity_alignment()) / TaskEntryChunkSizeInVoidStar;
123 size_t initial_chunk_capacity = align_up(initial_capacity, capacity_alignment()) / TaskEntryChunkSizeInVoidStar;
124
125 guarantee(initial_chunk_capacity <= _max_chunk_capacity,
126 "Maximum chunk capacity " SIZE_FORMAT " smaller than initial capacity " SIZE_FORMAT,
127 _max_chunk_capacity,
128 initial_chunk_capacity);
129
130 log_debug(gc)("Initialize mark stack with " SIZE_FORMAT " chunks, maximum " SIZE_FORMAT,
131 initial_chunk_capacity, _max_chunk_capacity);
132
133 return resize(initial_chunk_capacity);
134 }
135
136 void G1CMMarkStack::expand() {
137 if (_chunk_capacity == _max_chunk_capacity) {
138 log_debug(gc)("Can not expand overflow mark stack further, already at maximum capacity of " SIZE_FORMAT " chunks.", _chunk_capacity);
139 return;
140 }
141 size_t old_capacity = _chunk_capacity;
142 // Double capacity if possible
143 size_t new_capacity = MIN2(old_capacity * 2, _max_chunk_capacity);
144
145 if (resize(new_capacity)) {
146 log_debug(gc)("Expanded mark stack capacity from " SIZE_FORMAT " to " SIZE_FORMAT " chunks",
147 old_capacity, new_capacity);
148 } else {
149 log_warning(gc)("Failed to expand mark stack capacity from " SIZE_FORMAT " to " SIZE_FORMAT " chunks",
150 old_capacity, new_capacity);
151 }
152 }
153
154 G1CMMarkStack::~G1CMMarkStack() {
155 if (_base != NULL) {
156 MmapArrayAllocator<TaskQueueEntryChunk>::free(_base, _chunk_capacity);
157 }
158 }
159
160 void G1CMMarkStack::add_chunk_to_list(TaskQueueEntryChunk* volatile* list, TaskQueueEntryChunk* elem) {
161 elem->next = *list;
162 *list = elem;
163 }
164
165 void G1CMMarkStack::add_chunk_to_chunk_list(TaskQueueEntryChunk* elem) {
166 MutexLockerEx x(MarkStackChunkList_lock, Mutex::_no_safepoint_check_flag);
167 add_chunk_to_list(&_chunk_list, elem);
168 _chunks_in_chunk_list++;
169 }
170
171 void G1CMMarkStack::add_chunk_to_free_list(TaskQueueEntryChunk* elem) {
172 MutexLockerEx x(MarkStackFreeList_lock, Mutex::_no_safepoint_check_flag);
173 add_chunk_to_list(&_free_list, elem);
174 }
175
176 G1CMMarkStack::TaskQueueEntryChunk* G1CMMarkStack::remove_chunk_from_list(TaskQueueEntryChunk* volatile* list) {
177 TaskQueueEntryChunk* result = *list;
178 if (result != NULL) {
179 *list = (*list)->next;
180 }
181 return result;
182 }
183
184 G1CMMarkStack::TaskQueueEntryChunk* G1CMMarkStack::remove_chunk_from_chunk_list() {
185 MutexLockerEx x(MarkStackChunkList_lock, Mutex::_no_safepoint_check_flag);
186 TaskQueueEntryChunk* result = remove_chunk_from_list(&_chunk_list);
187 if (result != NULL) {
188 _chunks_in_chunk_list--;
189 }
190 return result;
191 }
192
193 G1CMMarkStack::TaskQueueEntryChunk* G1CMMarkStack::remove_chunk_from_free_list() {
194 MutexLockerEx x(MarkStackFreeList_lock, Mutex::_no_safepoint_check_flag);
195 return remove_chunk_from_list(&_free_list);
196 }
197
198 G1CMMarkStack::TaskQueueEntryChunk* G1CMMarkStack::allocate_new_chunk() {
199 // This dirty read of _hwm is okay because we only ever increase the _hwm in parallel code.
200 // Further this limits _hwm to a value of _chunk_capacity + #threads, avoiding
201 // wraparound of _hwm.
202 if (_hwm >= _chunk_capacity) {
203 return NULL;
204 }
205
206 size_t cur_idx = Atomic::add(1u, &_hwm) - 1;
207 if (cur_idx >= _chunk_capacity) {
208 return NULL;
209 }
210
211 TaskQueueEntryChunk* result = ::new (&_base[cur_idx]) TaskQueueEntryChunk;
212 result->next = NULL;
213 return result;
214 }
215
216 bool G1CMMarkStack::par_push_chunk(G1TaskQueueEntry* ptr_arr) {
217 // Get a new chunk.
218 TaskQueueEntryChunk* new_chunk = remove_chunk_from_free_list();
219
220 if (new_chunk == NULL) {
221 // Did not get a chunk from the free list. Allocate from backing memory.
222 new_chunk = allocate_new_chunk();
223
224 if (new_chunk == NULL) {
225 return false;
226 }
227 }
228
229 Copy::conjoint_memory_atomic(ptr_arr, new_chunk->data, EntriesPerChunk * sizeof(G1TaskQueueEntry));
230
231 add_chunk_to_chunk_list(new_chunk);
232
233 return true;
234 }
235
236 bool G1CMMarkStack::par_pop_chunk(G1TaskQueueEntry* ptr_arr) {
237 TaskQueueEntryChunk* cur = remove_chunk_from_chunk_list();
238
239 if (cur == NULL) {
240 return false;
241 }
242
243 Copy::conjoint_memory_atomic(cur->data, ptr_arr, EntriesPerChunk * sizeof(G1TaskQueueEntry));
244
245 add_chunk_to_free_list(cur);
246 return true;
247 }
248
249 void G1CMMarkStack::set_empty() {
250 _chunks_in_chunk_list = 0;
251 _hwm = 0;
252 _chunk_list = NULL;
253 _free_list = NULL;
254 }
255
256 G1CMRootRegions::G1CMRootRegions() :
257 _survivors(NULL), _cm(NULL), _scan_in_progress(false),
258 _should_abort(false), _claimed_survivor_index(0) { }
259
260 void G1CMRootRegions::init(const G1SurvivorRegions* survivors, G1ConcurrentMark* cm) {
261 _survivors = survivors;
262 _cm = cm;
263 }
264
265 void G1CMRootRegions::prepare_for_scan() {
266 assert(!scan_in_progress(), "pre-condition");
267
268 // Currently, only survivors can be root regions.
269 _claimed_survivor_index = 0;
270 _scan_in_progress = _survivors->regions()->is_nonempty();
271 _should_abort = false;
272 }
273
274 HeapRegion* G1CMRootRegions::claim_next() {
275 if (_should_abort) {
276 // If someone has set the should_abort flag, we return NULL to
277 // force the caller to bail out of their loop.
278 return NULL;
279 }
280
281 // Currently, only survivors can be root regions.
282 const GrowableArray<HeapRegion*>* survivor_regions = _survivors->regions();
283
284 int claimed_index = Atomic::add(1, &_claimed_survivor_index) - 1;
285 if (claimed_index < survivor_regions->length()) {
286 return survivor_regions->at(claimed_index);
287 }
288 return NULL;
289 }
290
291 uint G1CMRootRegions::num_root_regions() const {
292 return (uint)_survivors->regions()->length();
293 }
294
295 void G1CMRootRegions::notify_scan_done() {
296 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
297 _scan_in_progress = false;
298 RootRegionScan_lock->notify_all();
299 }
300
301 void G1CMRootRegions::cancel_scan() {
302 notify_scan_done();
303 }
304
305 void G1CMRootRegions::scan_finished() {
306 assert(scan_in_progress(), "pre-condition");
307
308 // Currently, only survivors can be root regions.
309 if (!_should_abort) {
310 assert(_claimed_survivor_index >= 0, "otherwise comparison is invalid: %d", _claimed_survivor_index);
311 assert((uint)_claimed_survivor_index >= _survivors->length(),
312 "we should have claimed all survivors, claimed index = %u, length = %u",
313 (uint)_claimed_survivor_index, _survivors->length());
314 }
315
316 notify_scan_done();
317 }
318
319 bool G1CMRootRegions::wait_until_scan_finished() {
320 if (!scan_in_progress()) {
321 return false;
322 }
323
324 {
325 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
326 while (scan_in_progress()) {
327 RootRegionScan_lock->wait(Mutex::_no_safepoint_check_flag);
328 }
329 }
330 return true;
331 }
332
333 // Returns the maximum number of workers to be used in a concurrent
334 // phase based on the number of GC workers being used in a STW
335 // phase.
336 static uint scale_concurrent_worker_threads(uint num_gc_workers) {
337 return MAX2((num_gc_workers + 2) / 4, 1U);
338 }
339
340 G1ConcurrentMark::G1ConcurrentMark(G1CollectedHeap* g1h,
341 G1RegionToSpaceMapper* prev_bitmap_storage,
342 G1RegionToSpaceMapper* next_bitmap_storage) :
343 // _cm_thread set inside the constructor
344 _g1h(g1h),
345 _completed_initialization(false),
346
347 _mark_bitmap_1(),
348 _mark_bitmap_2(),
349 _prev_mark_bitmap(&_mark_bitmap_1),
350 _next_mark_bitmap(&_mark_bitmap_2),
351
352 _heap(_g1h->reserved_region()),
353
354 _root_regions(),
355
356 _global_mark_stack(),
357
358 // _finger set in set_non_marking_state
359
360 _worker_id_offset(DirtyCardQueueSet::num_par_ids() + G1ConcRefinementThreads),
361 _max_num_tasks(ParallelGCThreads),
362 // _num_active_tasks set in set_non_marking_state()
363 // _tasks set inside the constructor
364
365 _task_queues(new G1CMTaskQueueSet((int) _max_num_tasks)),
366 _terminator(ParallelTaskTerminator((int) _max_num_tasks, _task_queues)),
367
368 _first_overflow_barrier_sync(),
369 _second_overflow_barrier_sync(),
370
371 _has_overflown(false),
372 _concurrent(false),
373 _has_aborted(false),
374 _restart_for_overflow(false),
375 _gc_timer_cm(new (ResourceObj::C_HEAP, mtGC) ConcurrentGCTimer()),
376 _gc_tracer_cm(new (ResourceObj::C_HEAP, mtGC) G1OldTracer()),
377
378 // _verbose_level set below
379
380 _init_times(),
381 _remark_times(),
382 _remark_mark_times(),
383 _remark_weak_ref_times(),
384 _cleanup_times(),
385 _total_cleanup_time(0.0),
386
387 _accum_task_vtime(NULL),
388
389 _concurrent_workers(NULL),
390 _num_concurrent_workers(0),
391 _max_concurrent_workers(0),
392
393 _region_mark_stats(NEW_C_HEAP_ARRAY(G1RegionMarkStats, _g1h->max_regions(), mtGC)),
394 _top_at_rebuild_starts(NEW_C_HEAP_ARRAY(HeapWord*, _g1h->max_regions(), mtGC))
395 {
396 _mark_bitmap_1.initialize(g1h->reserved_region(), prev_bitmap_storage);
397 _mark_bitmap_2.initialize(g1h->reserved_region(), next_bitmap_storage);
398
399 // Create & start ConcurrentMark thread.
400 _cm_thread = new ConcurrentMarkThread(this);
401 if (_cm_thread->osthread() == NULL) {
402 vm_shutdown_during_initialization("Could not create ConcurrentMarkThread");
403 }
404
405 assert(CGC_lock != NULL, "CGC_lock must be initialized");
406
407 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
408 satb_qs.set_buffer_size(G1SATBBufferSize);
409
410 _root_regions.init(_g1h->survivor(), this);
411
412 if (FLAG_IS_DEFAULT(ConcGCThreads) || ConcGCThreads == 0) {
413 // Calculate the number of concurrent worker threads by scaling
414 // the number of parallel GC threads.
415 uint marking_thread_num = scale_concurrent_worker_threads(ParallelGCThreads);
416 FLAG_SET_ERGO(uint, ConcGCThreads, marking_thread_num);
417 }
418
419 assert(ConcGCThreads > 0, "ConcGCThreads have been set.");
420 if (ConcGCThreads > ParallelGCThreads) {
421 log_warning(gc)("More ConcGCThreads (%u) than ParallelGCThreads (%u).",
422 ConcGCThreads, ParallelGCThreads);
423 return;
424 }
425
426 log_debug(gc)("ConcGCThreads: %u offset %u", ConcGCThreads, _worker_id_offset);
427 log_debug(gc)("ParallelGCThreads: %u", ParallelGCThreads);
428
429 _num_concurrent_workers = ConcGCThreads;
430 _max_concurrent_workers = _num_concurrent_workers;
431
432 _concurrent_workers = new WorkGang("G1 Conc", _max_concurrent_workers, false, true);
433 _concurrent_workers->initialize_workers();
434
435 if (FLAG_IS_DEFAULT(MarkStackSize)) {
436 size_t mark_stack_size =
437 MIN2(MarkStackSizeMax,
438 MAX2(MarkStackSize, (size_t) (_max_concurrent_workers * TASKQUEUE_SIZE)));
439 // Verify that the calculated value for MarkStackSize is in range.
440 // It would be nice to use the private utility routine from Arguments.
441 if (!(mark_stack_size >= 1 && mark_stack_size <= MarkStackSizeMax)) {
442 log_warning(gc)("Invalid value calculated for MarkStackSize (" SIZE_FORMAT "): "
443 "must be between 1 and " SIZE_FORMAT,
444 mark_stack_size, MarkStackSizeMax);
445 return;
446 }
447 FLAG_SET_ERGO(size_t, MarkStackSize, mark_stack_size);
448 } else {
449 // Verify MarkStackSize is in range.
450 if (FLAG_IS_CMDLINE(MarkStackSize)) {
451 if (FLAG_IS_DEFAULT(MarkStackSizeMax)) {
452 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
453 log_warning(gc)("Invalid value specified for MarkStackSize (" SIZE_FORMAT "): "
454 "must be between 1 and " SIZE_FORMAT,
455 MarkStackSize, MarkStackSizeMax);
456 return;
457 }
458 } else if (FLAG_IS_CMDLINE(MarkStackSizeMax)) {
459 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
460 log_warning(gc)("Invalid value specified for MarkStackSize (" SIZE_FORMAT ")"
461 " or for MarkStackSizeMax (" SIZE_FORMAT ")",
462 MarkStackSize, MarkStackSizeMax);
463 return;
464 }
465 }
466 }
467 }
468
469 if (!_global_mark_stack.initialize(MarkStackSize, MarkStackSizeMax)) {
470 vm_exit_during_initialization("Failed to allocate initial concurrent mark overflow mark stack.");
471 }
472
473 _tasks = NEW_C_HEAP_ARRAY(G1CMTask*, _max_num_tasks, mtGC);
474 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_num_tasks, mtGC);
475
476 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
477 _num_active_tasks = _max_num_tasks;
478
479 for (uint i = 0; i < _max_num_tasks; ++i) {
480 G1CMTaskQueue* task_queue = new G1CMTaskQueue();
481 task_queue->initialize();
482 _task_queues->register_queue(i, task_queue);
483
484 _tasks[i] = new G1CMTask(i, this, task_queue, _region_mark_stats, _g1h->max_regions());
485
486 _accum_task_vtime[i] = 0.0;
487 }
488
489 reset_at_marking_complete();
490 _completed_initialization = true;
491 }
492
493 void G1ConcurrentMark::reset() {
494 _has_aborted = false;
495
496 reset_marking_for_restart();
497
498 // Reset all tasks, since different phases will use different number of active
499 // threads. So, it's easiest to have all of them ready.
500 for (uint i = 0; i < _max_num_tasks; ++i) {
501 _tasks[i]->reset(_next_mark_bitmap);
502 }
503
504 uint max_regions = _g1h->max_regions();
505 for (uint i = 0; i < max_regions; i++) {
506 _top_at_rebuild_starts[i] = NULL;
507 _region_mark_stats[i].clear();
508 }
509 }
510
511 void G1ConcurrentMark::clear_statistics_in_region(uint region_idx) {
512 for (uint j = 0; j < _max_num_tasks; ++j) {
513 _tasks[j]->clear_mark_stats_cache(region_idx);
514 }
515 _top_at_rebuild_starts[region_idx] = NULL;
516 _region_mark_stats[region_idx].clear();
517 }
518
519 void G1ConcurrentMark::clear_statistics(HeapRegion* r) {
520 uint const region_idx = r->hrm_index();
521 if (r->is_humongous()) {
522 assert(r->is_starts_humongous(), "Got humongous continues region here");
523 uint const size_in_regions = (uint)_g1h->humongous_obj_size_in_regions(oop(r->humongous_start_region()->bottom())->size());
524 for (uint j = region_idx; j < (region_idx + size_in_regions); j++) {
525 clear_statistics_in_region(j);
526 }
527 } else {
528 clear_statistics_in_region(region_idx);
529 }
530 }
531
532 static void clear_mark_if_set(G1CMBitMap* bitmap, HeapWord* addr) {
533 if (bitmap->is_marked(addr)) {
534 bitmap->clear(addr);
535 }
536 }
537
538 void G1ConcurrentMark::humongous_object_eagerly_reclaimed(HeapRegion* r) {
539 assert_at_safepoint_on_vm_thread();
540
541 // Need to clear all mark bits of the humongous object.
542 clear_mark_if_set(_prev_mark_bitmap, r->bottom());
543 clear_mark_if_set(_next_mark_bitmap, r->bottom());
544
545 if (!_g1h->collector_state()->mark_or_rebuild_in_progress()) {
546 return;
547 }
548
549 // Clear any statistics about the region gathered so far.
550 clear_statistics(r);
551 }
552
553 void G1ConcurrentMark::reset_marking_for_restart() {
554 _global_mark_stack.set_empty();
555
556 // Expand the marking stack, if we have to and if we can.
557 if (has_overflown()) {
558 _global_mark_stack.expand();
559
560 uint max_regions = _g1h->max_regions();
561 for (uint i = 0; i < max_regions; i++) {
562 _region_mark_stats[i].clear_during_overflow();
563 }
564 }
565
566 clear_has_overflown();
567 _finger = _heap.start();
568
569 for (uint i = 0; i < _max_num_tasks; ++i) {
570 G1CMTaskQueue* queue = _task_queues->queue(i);
571 queue->set_empty();
572 }
573 }
574
575 void G1ConcurrentMark::set_concurrency(uint active_tasks) {
576 assert(active_tasks <= _max_num_tasks, "we should not have more");
577
578 _num_active_tasks = active_tasks;
579 // Need to update the three data structures below according to the
580 // number of active threads for this phase.
581 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
582 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
583 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
584 }
585
586 void G1ConcurrentMark::set_concurrency_and_phase(uint active_tasks, bool concurrent) {
587 set_concurrency(active_tasks);
588
589 _concurrent = concurrent;
590
591 if (!concurrent) {
592 // At this point we should be in a STW phase, and completed marking.
593 assert_at_safepoint_on_vm_thread();
594 assert(out_of_regions(),
595 "only way to get here: _finger: " PTR_FORMAT ", _heap_end: " PTR_FORMAT,
596 p2i(_finger), p2i(_heap.end()));
597 }
598 }
599
600 void G1ConcurrentMark::reset_at_marking_complete() {
601 // We set the global marking state to some default values when we're
602 // not doing marking.
603 reset_marking_for_restart();
604 _num_active_tasks = 0;
605 }
606
607 G1ConcurrentMark::~G1ConcurrentMark() {
608 FREE_C_HEAP_ARRAY(HeapWord*, _top_at_rebuild_starts);
609 FREE_C_HEAP_ARRAY(G1RegionMarkStats, _region_mark_stats);
610 // The G1ConcurrentMark instance is never freed.
611 ShouldNotReachHere();
612 }
613
614 class G1ClearBitMapTask : public AbstractGangTask {
615 public:
616 static size_t chunk_size() { return M; }
617
618 private:
619 // Heap region closure used for clearing the given mark bitmap.
620 class G1ClearBitmapHRClosure : public HeapRegionClosure {
621 private:
622 G1CMBitMap* _bitmap;
623 G1ConcurrentMark* _cm;
624 public:
625 G1ClearBitmapHRClosure(G1CMBitMap* bitmap, G1ConcurrentMark* cm) : HeapRegionClosure(), _cm(cm), _bitmap(bitmap) {
626 }
627
628 virtual bool do_heap_region(HeapRegion* r) {
629 size_t const chunk_size_in_words = G1ClearBitMapTask::chunk_size() / HeapWordSize;
630
631 HeapWord* cur = r->bottom();
632 HeapWord* const end = r->end();
633
634 while (cur < end) {
635 MemRegion mr(cur, MIN2(cur + chunk_size_in_words, end));
636 _bitmap->clear_range(mr);
637
638 cur += chunk_size_in_words;
639
640 // Abort iteration if after yielding the marking has been aborted.
641 if (_cm != NULL && _cm->do_yield_check() && _cm->has_aborted()) {
642 return true;
643 }
644 // Repeat the asserts from before the start of the closure. We will do them
645 // as asserts here to minimize their overhead on the product. However, we
646 // will have them as guarantees at the beginning / end of the bitmap
647 // clearing to get some checking in the product.
648 assert(_cm == NULL || _cm->cm_thread()->during_cycle(), "invariant");
649 assert(_cm == NULL || !G1CollectedHeap::heap()->collector_state()->mark_or_rebuild_in_progress(), "invariant");
650 }
651 assert(cur == end, "Must have completed iteration over the bitmap for region %u.", r->hrm_index());
652
653 return false;
654 }
655 };
656
657 G1ClearBitmapHRClosure _cl;
658 HeapRegionClaimer _hr_claimer;
659 bool _suspendible; // If the task is suspendible, workers must join the STS.
660
661 public:
662 G1ClearBitMapTask(G1CMBitMap* bitmap, G1ConcurrentMark* cm, uint n_workers, bool suspendible) :
663 AbstractGangTask("G1 Clear Bitmap"),
664 _cl(bitmap, suspendible ? cm : NULL),
665 _hr_claimer(n_workers),
666 _suspendible(suspendible)
667 { }
668
669 void work(uint worker_id) {
670 SuspendibleThreadSetJoiner sts_join(_suspendible);
671 G1CollectedHeap::heap()->heap_region_par_iterate_from_worker_offset(&_cl, &_hr_claimer, worker_id);
672 }
673
674 bool is_complete() {
675 return _cl.is_complete();
676 }
677 };
678
679 void G1ConcurrentMark::clear_bitmap(G1CMBitMap* bitmap, WorkGang* workers, bool may_yield) {
680 assert(may_yield || SafepointSynchronize::is_at_safepoint(), "Non-yielding bitmap clear only allowed at safepoint.");
681
682 size_t const num_bytes_to_clear = (HeapRegion::GrainBytes * _g1h->num_regions()) / G1CMBitMap::heap_map_factor();
683 size_t const num_chunks = align_up(num_bytes_to_clear, G1ClearBitMapTask::chunk_size()) / G1ClearBitMapTask::chunk_size();
684
685 uint const num_workers = (uint)MIN2(num_chunks, (size_t)workers->active_workers());
686
687 G1ClearBitMapTask cl(bitmap, this, num_workers, may_yield);
688
689 log_debug(gc, ergo)("Running %s with %u workers for " SIZE_FORMAT " work units.", cl.name(), num_workers, num_chunks);
690 workers->run_task(&cl, num_workers);
691 guarantee(!may_yield || cl.is_complete(), "Must have completed iteration when not yielding.");
692 }
693
694 void G1ConcurrentMark::cleanup_for_next_mark() {
695 // Make sure that the concurrent mark thread looks to still be in
696 // the current cycle.
697 guarantee(cm_thread()->during_cycle(), "invariant");
698
699 // We are finishing up the current cycle by clearing the next
700 // marking bitmap and getting it ready for the next cycle. During
701 // this time no other cycle can start. So, let's make sure that this
702 // is the case.
703 guarantee(!_g1h->collector_state()->mark_or_rebuild_in_progress(), "invariant");
704
705 clear_bitmap(_next_mark_bitmap, _concurrent_workers, true);
706
707 // Repeat the asserts from above.
708 guarantee(cm_thread()->during_cycle(), "invariant");
709 guarantee(!_g1h->collector_state()->mark_or_rebuild_in_progress(), "invariant");
710 }
711
712 void G1ConcurrentMark::clear_prev_bitmap(WorkGang* workers) {
713 assert_at_safepoint_on_vm_thread();
714 clear_bitmap(_prev_mark_bitmap, workers, false);
715 }
716
717 class CheckBitmapClearHRClosure : public HeapRegionClosure {
718 G1CMBitMap* _bitmap;
719 public:
720 CheckBitmapClearHRClosure(G1CMBitMap* bitmap) : _bitmap(bitmap) {
721 }
722
723 virtual bool do_heap_region(HeapRegion* r) {
724 // This closure can be called concurrently to the mutator, so we must make sure
725 // that the result of the getNextMarkedWordAddress() call is compared to the
726 // value passed to it as limit to detect any found bits.
727 // end never changes in G1.
728 HeapWord* end = r->end();
729 return _bitmap->get_next_marked_addr(r->bottom(), end) != end;
730 }
731 };
732
733 bool G1ConcurrentMark::next_mark_bitmap_is_clear() {
734 CheckBitmapClearHRClosure cl(_next_mark_bitmap);
735 _g1h->heap_region_iterate(&cl);
736 return cl.is_complete();
737 }
738
739 class NoteStartOfMarkHRClosure : public HeapRegionClosure {
740 public:
741 bool do_heap_region(HeapRegion* r) {
742 r->note_start_of_marking();
743 return false;
744 }
745 };
746
747 void G1ConcurrentMark::pre_initial_mark() {
748 // Initialize marking structures. This has to be done in a STW phase.
749 reset();
750
751 // For each region note start of marking.
752 NoteStartOfMarkHRClosure startcl;
753 _g1h->heap_region_iterate(&startcl);
754 }
755
756
757 void G1ConcurrentMark::post_initial_mark() {
758 // Start Concurrent Marking weak-reference discovery.
759 ReferenceProcessor* rp = _g1h->ref_processor_cm();
760 // enable ("weak") refs discovery
761 rp->enable_discovery();
762 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
763
764 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
765 // This is the start of the marking cycle, we're expected all
766 // threads to have SATB queues with active set to false.
767 satb_mq_set.set_active_all_threads(true, /* new active value */
768 false /* expected_active */);
769
770 _root_regions.prepare_for_scan();
771
772 // update_g1_committed() will be called at the end of an evac pause
773 // when marking is on. So, it's also called at the end of the
774 // initial-mark pause to update the heap end, if the heap expands
775 // during it. No need to call it here.
776 }
777
778 /*
779 * Notice that in the next two methods, we actually leave the STS
780 * during the barrier sync and join it immediately afterwards. If we
781 * do not do this, the following deadlock can occur: one thread could
782 * be in the barrier sync code, waiting for the other thread to also
783 * sync up, whereas another one could be trying to yield, while also
784 * waiting for the other threads to sync up too.
785 *
786 * Note, however, that this code is also used during remark and in
787 * this case we should not attempt to leave / enter the STS, otherwise
788 * we'll either hit an assert (debug / fastdebug) or deadlock
789 * (product). So we should only leave / enter the STS if we are
790 * operating concurrently.
791 *
792 * Because the thread that does the sync barrier has left the STS, it
793 * is possible to be suspended for a Full GC or an evacuation pause
794 * could occur. This is actually safe, since the entering the sync
795 * barrier is one of the last things do_marking_step() does, and it
796 * doesn't manipulate any data structures afterwards.
797 */
798
799 void G1ConcurrentMark::enter_first_sync_barrier(uint worker_id) {
800 bool barrier_aborted;
801 {
802 SuspendibleThreadSetLeaver sts_leave(concurrent());
803 barrier_aborted = !_first_overflow_barrier_sync.enter();
804 }
805
806 // at this point everyone should have synced up and not be doing any
807 // more work
808
809 if (barrier_aborted) {
810 // If the barrier aborted we ignore the overflow condition and
811 // just abort the whole marking phase as quickly as possible.
812 return;
813 }
814 }
815
816 void G1ConcurrentMark::enter_second_sync_barrier(uint worker_id) {
817 SuspendibleThreadSetLeaver sts_leave(concurrent());
818 _second_overflow_barrier_sync.enter();
819
820 // at this point everything should be re-initialized and ready to go
821 }
822
823 class G1CMConcurrentMarkingTask : public AbstractGangTask {
824 G1ConcurrentMark* _cm;
825 public:
826 void work(uint worker_id) {
827 assert(Thread::current()->is_ConcurrentGC_thread(), "Not a concurrent GC thread");
828 ResourceMark rm;
829
830 double start_vtime = os::elapsedVTime();
831
832 {
833 SuspendibleThreadSetJoiner sts_join;
834
835 assert(worker_id < _cm->active_tasks(), "invariant");
836
837 G1CMTask* task = _cm->task(worker_id);
838 task->record_start_time();
839 if (!_cm->has_aborted()) {
840 do {
841 task->do_marking_step(G1ConcMarkStepDurationMillis,
842 true /* do_termination */,
843 false /* is_serial*/);
844
845 _cm->do_yield_check();
846 } while (!_cm->has_aborted() && task->has_aborted());
847 }
848 task->record_end_time();
849 guarantee(!task->has_aborted() || _cm->has_aborted(), "invariant");
850 }
851
852 double end_vtime = os::elapsedVTime();
853 _cm->update_accum_task_vtime(worker_id, end_vtime - start_vtime);
854 }
855
856 G1CMConcurrentMarkingTask(G1ConcurrentMark* cm) :
857 AbstractGangTask("Concurrent Mark"), _cm(cm) { }
858
859 ~G1CMConcurrentMarkingTask() { }
860 };
861
862 uint G1ConcurrentMark::calc_active_marking_workers() {
863 uint result = 0;
864 if (!UseDynamicNumberOfGCThreads ||
865 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
866 !ForceDynamicNumberOfGCThreads)) {
867 result = _max_concurrent_workers;
868 } else {
869 result =
870 AdaptiveSizePolicy::calc_default_active_workers(_max_concurrent_workers,
871 1, /* Minimum workers */
872 _num_concurrent_workers,
873 Threads::number_of_non_daemon_threads());
874 // Don't scale the result down by scale_concurrent_workers() because
875 // that scaling has already gone into "_max_concurrent_workers".
876 }
877 assert(result > 0 && result <= _max_concurrent_workers,
878 "Calculated number of marking workers must be larger than zero and at most the maximum %u, but is %u",
879 _max_concurrent_workers, result);
880 return result;
881 }
882
883 void G1ConcurrentMark::scan_root_region(HeapRegion* hr, uint worker_id) {
884 // Currently, only survivors can be root regions.
885 assert(hr->next_top_at_mark_start() == hr->bottom(), "invariant");
886 G1RootRegionScanClosure cl(_g1h, this, worker_id);
887
888 const uintx interval = PrefetchScanIntervalInBytes;
889 HeapWord* curr = hr->bottom();
890 const HeapWord* end = hr->top();
891 while (curr < end) {
892 Prefetch::read(curr, interval);
893 oop obj = oop(curr);
894 int size = obj->oop_iterate_size(&cl);
895 assert(size == obj->size(), "sanity");
896 curr += size;
897 }
898 }
899
900 class G1CMRootRegionScanTask : public AbstractGangTask {
901 G1ConcurrentMark* _cm;
902 public:
903 G1CMRootRegionScanTask(G1ConcurrentMark* cm) :
904 AbstractGangTask("G1 Root Region Scan"), _cm(cm) { }
905
906 void work(uint worker_id) {
907 assert(Thread::current()->is_ConcurrentGC_thread(),
908 "this should only be done by a conc GC thread");
909
910 G1CMRootRegions* root_regions = _cm->root_regions();
911 HeapRegion* hr = root_regions->claim_next();
912 while (hr != NULL) {
913 _cm->scan_root_region(hr, worker_id);
914 hr = root_regions->claim_next();
915 }
916 }
917 };
918
919 void G1ConcurrentMark::scan_root_regions() {
920 // scan_in_progress() will have been set to true only if there was
921 // at least one root region to scan. So, if it's false, we
922 // should not attempt to do any further work.
923 if (root_regions()->scan_in_progress()) {
924 assert(!has_aborted(), "Aborting before root region scanning is finished not supported.");
925
926 _num_concurrent_workers = MIN2(calc_active_marking_workers(),
927 // We distribute work on a per-region basis, so starting
928 // more threads than that is useless.
929 root_regions()->num_root_regions());
930 assert(_num_concurrent_workers <= _max_concurrent_workers,
931 "Maximum number of marking threads exceeded");
932
933 G1CMRootRegionScanTask task(this);
934 log_debug(gc, ergo)("Running %s using %u workers for %u work units.",
935 task.name(), _num_concurrent_workers, root_regions()->num_root_regions());
936 _concurrent_workers->run_task(&task, _num_concurrent_workers);
937
938 // It's possible that has_aborted() is true here without actually
939 // aborting the survivor scan earlier. This is OK as it's
940 // mainly used for sanity checking.
941 root_regions()->scan_finished();
942 }
943 }
944
945 void G1ConcurrentMark::concurrent_cycle_start() {
946 _gc_timer_cm->register_gc_start();
947
948 _gc_tracer_cm->report_gc_start(GCCause::_no_gc /* first parameter is not used */, _gc_timer_cm->gc_start());
949
950 _g1h->trace_heap_before_gc(_gc_tracer_cm);
951 }
952
953 void G1ConcurrentMark::concurrent_cycle_end() {
954 _g1h->collector_state()->set_clearing_next_bitmap(false);
955
956 _g1h->trace_heap_after_gc(_gc_tracer_cm);
957
958 if (has_aborted()) {
959 log_info(gc, marking)("Concurrent Mark Abort");
960 _gc_tracer_cm->report_concurrent_mode_failure();
961 }
962
963 _gc_timer_cm->register_gc_end();
964
965 _gc_tracer_cm->report_gc_end(_gc_timer_cm->gc_end(), _gc_timer_cm->time_partitions());
966 }
967
968 void G1ConcurrentMark::mark_from_roots() {
969 _restart_for_overflow = false;
970
971 _num_concurrent_workers = calc_active_marking_workers();
972
973 uint active_workers = MAX2(1U, _num_concurrent_workers);
974
975 // Setting active workers is not guaranteed since fewer
976 // worker threads may currently exist and more may not be
977 // available.
978 active_workers = _concurrent_workers->update_active_workers(active_workers);
979 log_info(gc, task)("Using %u workers of %u for marking", active_workers, _concurrent_workers->total_workers());
980
981 // Parallel task terminator is set in "set_concurrency_and_phase()"
982 set_concurrency_and_phase(active_workers, true /* concurrent */);
983
984 G1CMConcurrentMarkingTask marking_task(this);
985 _concurrent_workers->run_task(&marking_task);
986 print_stats();
987 }
988
989 void G1ConcurrentMark::verify_during_pause(G1HeapVerifier::G1VerifyType type, VerifyOption vo, const char* caller) {
990 G1HeapVerifier* verifier = _g1h->verifier();
991
992 verifier->verify_region_sets_optional();
993
994 if (VerifyDuringGC) {
995 GCTraceTime(Debug, gc, phases) trace(caller, _gc_timer_cm);
996
997 size_t const BufLen = 512;
998 char buffer[BufLen];
999
1000 jio_snprintf(buffer, BufLen, "During GC (%s)", caller);
1001 verifier->verify(type, vo, buffer);
1002 }
1003
1004 verifier->check_bitmaps(caller);
1005 }
1006
1007 class G1UpdateRemSetTrackingBeforeRebuild : public HeapRegionClosure {
1008 G1CollectedHeap* _g1h;
1009 G1ConcurrentMark* _cm;
1010
1011 uint _num_regions_selected_for_rebuild; // The number of regions actually selected for rebuild.
1012
1013 void update_remset_before_rebuild(HeapRegion * hr) {
1014 G1RemSetTrackingPolicy* tracking_policy = _g1h->g1_policy()->remset_tracker();
1015
1016 size_t live_bytes = _cm->liveness(hr->hrm_index()) * HeapWordSize;
1017 bool selected_for_rebuild = tracking_policy->update_before_rebuild(hr, live_bytes);
1018 if (selected_for_rebuild) {
1019 _num_regions_selected_for_rebuild++;
1020 }
1021 _cm->update_top_at_rebuild_start(hr);
1022 }
1023
1024 public:
1025 G1UpdateRemSetTrackingBeforeRebuild(G1CollectedHeap* g1h, G1ConcurrentMark* cm) :
1026 _g1h(g1h), _cm(cm), _num_regions_selected_for_rebuild(0) { }
1027
1028 virtual bool do_heap_region(HeapRegion* r) {
1029 update_remset_before_rebuild(r);
1030 return false;
1031 }
1032
1033 uint num_selected_for_rebuild() const { return _num_regions_selected_for_rebuild; }
1034 };
1035
1036 class G1UpdateRemSetTrackingAfterRebuild : public HeapRegionClosure {
1037 G1CollectedHeap* _g1h;
1038 public:
1039 G1UpdateRemSetTrackingAfterRebuild(G1CollectedHeap* g1h) : _g1h(g1h) { }
1040
1041 virtual bool do_heap_region(HeapRegion* r) {
1042 _g1h->g1_policy()->remset_tracker()->update_after_rebuild(r);
1043 return false;
1044 }
1045 };
1046
1047 void G1ConcurrentMark::remark() {
1048 assert_at_safepoint_on_vm_thread();
1049
1050 // If a full collection has happened, we should not continue. However we might
1051 // have ended up here as the Remark VM operation has been scheduled already.
1052 if (has_aborted()) {
1053 return;
1054 }
1055
1056 G1Policy* g1p = _g1h->g1_policy();
1057 g1p->record_concurrent_mark_remark_start();
1058
1059 double start = os::elapsedTime();
1060
1061 verify_during_pause(G1HeapVerifier::G1VerifyRemark, VerifyOption_G1UsePrevMarking, "Remark before");
1062
1063 {
1064 GCTraceTime(Debug, gc, phases) trace("Finalize Marking", _gc_timer_cm);
1065 finalize_marking();
1066 }
1067
1068 double mark_work_end = os::elapsedTime();
1069
1070 bool const mark_finished = !has_overflown();
1071 if (mark_finished) {
1072 weak_refs_work(false /* clear_all_soft_refs */);
1073
1074 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1075 // We're done with marking.
1076 // This is the end of the marking cycle, we're expected all
1077 // threads to have SATB queues with active set to true.
1078 satb_mq_set.set_active_all_threads(false, /* new active value */
1079 true /* expected_active */);
1080
1081 {
1082 GCTraceTime(Debug, gc, phases)("Flush Task Caches");
1083 flush_all_task_caches();
1084 }
1085
1086 {
1087 GCTraceTime(Debug, gc, phases)("Update Remembered Set Tracking Before Rebuild");
1088 G1UpdateRemSetTrackingBeforeRebuild cl(_g1h, this);
1089 _g1h->heap_region_iterate(&cl);
1090 log_debug(gc, remset, tracking)("Remembered Set Tracking update regions total %u, selected %u",
1091 _g1h->num_regions(), cl.num_selected_for_rebuild());
1092 }
1093
1094 verify_during_pause(G1HeapVerifier::G1VerifyRemark, VerifyOption_G1UseNextMarking, "Remark after");
1095
1096 assert(!restart_for_overflow(), "sanity");
1097 // Completely reset the marking state since marking completed
1098 reset_at_marking_complete();
1099 } else {
1100 // We overflowed. Restart concurrent marking.
1101 _restart_for_overflow = true;
1102
1103 verify_during_pause(G1HeapVerifier::G1VerifyRemark, VerifyOption_G1UsePrevMarking, "Remark overflow");
1104
1105 // Clear the marking state because we will be restarting
1106 // marking due to overflowing the global mark stack.
1107 reset_marking_for_restart();
1108 }
1109
1110 {
1111 GCTraceTime(Debug, gc, phases)("Report Object Count");
1112 report_object_count();
1113 }
1114
1115 // Statistics
1116 double now = os::elapsedTime();
1117 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1118 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1119 _remark_times.add((now - start) * 1000.0);
1120
1121 g1p->record_concurrent_mark_remark_end();
1122 }
1123
1124 class G1CleanupTask : public AbstractGangTask {
1125 // Per-region work during the Cleanup pause.
1126 class G1CleanupRegionsClosure : public HeapRegionClosure {
1127 G1CollectedHeap* _g1h;
1128 size_t _freed_bytes;
1129 FreeRegionList* _local_cleanup_list;
1130 uint _old_regions_removed;
1131 uint _humongous_regions_removed;
1132 HRRSCleanupTask* _hrrs_cleanup_task;
1133
1134 public:
1135 G1CleanupRegionsClosure(G1CollectedHeap* g1,
1136 FreeRegionList* local_cleanup_list,
1137 HRRSCleanupTask* hrrs_cleanup_task) :
1138 _g1h(g1),
1139 _freed_bytes(0),
1140 _local_cleanup_list(local_cleanup_list),
1141 _old_regions_removed(0),
1142 _humongous_regions_removed(0),
1143 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1144
1145 size_t freed_bytes() { return _freed_bytes; }
1146 const uint old_regions_removed() { return _old_regions_removed; }
1147 const uint humongous_regions_removed() { return _humongous_regions_removed; }
1148
1149 bool do_heap_region(HeapRegion *hr) {
1150 hr->note_end_of_marking();
1151
1152 if (hr->used() > 0 && hr->max_live_bytes() == 0 && !hr->is_young() && !hr->is_archive()) {
1153 _freed_bytes += hr->used();
1154 hr->set_containing_set(NULL);
1155 if (hr->is_humongous()) {
1156 _humongous_regions_removed++;
1157 _g1h->free_humongous_region(hr, _local_cleanup_list);
1158 } else {
1159 _old_regions_removed++;
1160 _g1h->free_region(hr, _local_cleanup_list, false /* skip_remset */, false /* skip_hcc */, true /* locked */);
1161 }
1162 hr->clear_cardtable();
1163 _g1h->concurrent_mark()->clear_statistics_in_region(hr->hrm_index());
1164 log_trace(gc)("Reclaimed empty region %u (%s) bot " PTR_FORMAT, hr->hrm_index(), hr->get_short_type_str(), p2i(hr->bottom()));
1165 } else {
1166 hr->rem_set()->do_cleanup_work(_hrrs_cleanup_task);
1167 }
1168
1169 return false;
1170 }
1171 };
1172
1173 G1CollectedHeap* _g1h;
1174 FreeRegionList* _cleanup_list;
1175 HeapRegionClaimer _hrclaimer;
1176
1177 public:
1178 G1CleanupTask(G1CollectedHeap* g1h, FreeRegionList* cleanup_list, uint n_workers) :
1179 AbstractGangTask("G1 Cleanup"),
1180 _g1h(g1h),
1181 _cleanup_list(cleanup_list),
1182 _hrclaimer(n_workers) {
1183
1184 HeapRegionRemSet::reset_for_cleanup_tasks();
1185 }
1186
1187 void work(uint worker_id) {
1188 FreeRegionList local_cleanup_list("Local Cleanup List");
1189 HRRSCleanupTask hrrs_cleanup_task;
1190 G1CleanupRegionsClosure cl(_g1h,
1191 &local_cleanup_list,
1192 &hrrs_cleanup_task);
1193 _g1h->heap_region_par_iterate_from_worker_offset(&cl, &_hrclaimer, worker_id);
1194 assert(cl.is_complete(), "Shouldn't have aborted!");
1195
1196 // Now update the old/humongous region sets
1197 _g1h->remove_from_old_sets(cl.old_regions_removed(), cl.humongous_regions_removed());
1198 {
1199 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1200 _g1h->decrement_summary_bytes(cl.freed_bytes());
1201
1202 _cleanup_list->add_ordered(&local_cleanup_list);
1203 assert(local_cleanup_list.is_empty(), "post-condition");
1204
1205 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1206 }
1207 }
1208 };
1209
1210 void G1ConcurrentMark::reclaim_empty_regions() {
1211 WorkGang* workers = _g1h->workers();
1212 FreeRegionList empty_regions_list("Empty Regions After Mark List");
1213
1214 G1CleanupTask cl(_g1h, &empty_regions_list, workers->active_workers());
1215 workers->run_task(&cl);
1216
1217 if (!empty_regions_list.is_empty()) {
1218 log_debug(gc)("Reclaimed %u empty regions", empty_regions_list.length());
1219 // Now print the empty regions list.
1220 G1HRPrinter* hrp = _g1h->hr_printer();
1221 if (hrp->is_active()) {
1222 FreeRegionListIterator iter(&empty_regions_list);
1223 while (iter.more_available()) {
1224 HeapRegion* hr = iter.get_next();
1225 hrp->cleanup(hr);
1226 }
1227 }
1228 // And actually make them available.
1229 _g1h->prepend_to_freelist(&empty_regions_list);
1230 }
1231 }
1232
1233 void G1ConcurrentMark::cleanup() {
1234 assert_at_safepoint_on_vm_thread();
1235
1236 // If a full collection has happened, we shouldn't do this.
1237 if (has_aborted()) {
1238 return;
1239 }
1240
1241 G1Policy* g1p = _g1h->g1_policy();
1242 g1p->record_concurrent_mark_cleanup_start();
1243
1244 double start = os::elapsedTime();
1245
1246 verify_during_pause(G1HeapVerifier::G1VerifyCleanup, VerifyOption_G1UseNextMarking, "Cleanup before");
1247
1248 {
1249 GCTraceTime(Debug, gc, phases)("Update Remembered Set Tracking After Rebuild");
1250 G1UpdateRemSetTrackingAfterRebuild cl(_g1h);
1251 _g1h->heap_region_iterate(&cl);
1252 }
1253
1254 if (log_is_enabled(Trace, gc, liveness)) {
1255 G1PrintRegionLivenessInfoClosure cl("Post-Cleanup");
1256 _g1h->heap_region_iterate(&cl);
1257 }
1258
1259 // Install newly created mark bitmap as "prev".
1260 swap_mark_bitmaps();
1261 {
1262 GCTraceTime(Debug, gc, phases)("Reclaim Empty Regions");
1263 reclaim_empty_regions();
1264 }
1265
1266 // Cleanup will have freed any regions completely full of garbage.
1267 // Update the soft reference policy with the new heap occupancy.
1268 Universe::update_heap_info_at_gc();
1269
1270 // Clean out dead classes and update Metaspace sizes.
1271 if (ClassUnloadingWithConcurrentMark) {
1272 GCTraceTime(Debug, gc, phases)("Purge Metaspace");
1273 ClassLoaderDataGraph::purge();
1274 }
1275 MetaspaceGC::compute_new_size();
1276
1277 // We reclaimed old regions so we should calculate the sizes to make
1278 // sure we update the old gen/space data.
1279 _g1h->g1mm()->update_sizes();
1280
1281 verify_during_pause(G1HeapVerifier::G1VerifyCleanup, VerifyOption_G1UsePrevMarking, "Cleanup after");
1282
1283 // We need to make this be a "collection" so any collection pause that
1284 // races with it goes around and waits for Cleanup to finish.
1285 _g1h->increment_total_collections();
1286
1287 // Local statistics
1288 double recent_cleanup_time = (os::elapsedTime() - start);
1289 _total_cleanup_time += recent_cleanup_time;
1290 _cleanup_times.add(recent_cleanup_time);
1291
1292 {
1293 GCTraceTime(Debug, gc, phases)("Finalize Concurrent Mark Cleanup");
1294 _g1h->g1_policy()->record_concurrent_mark_cleanup_end();
1295 }
1296 }
1297
1298 // Supporting Object and Oop closures for reference discovery
1299 // and processing in during marking
1300
1301 bool G1CMIsAliveClosure::do_object_b(oop obj) {
1302 HeapWord* addr = (HeapWord*)obj;
1303 return addr != NULL &&
1304 (!_g1h->is_in_g1_reserved(addr) || !_g1h->is_obj_ill(obj));
1305 }
1306
1307 // 'Keep Alive' oop closure used by both serial parallel reference processing.
1308 // Uses the G1CMTask associated with a worker thread (for serial reference
1309 // processing the G1CMTask for worker 0 is used) to preserve (mark) and
1310 // trace referent objects.
1311 //
1312 // Using the G1CMTask and embedded local queues avoids having the worker
1313 // threads operating on the global mark stack. This reduces the risk
1314 // of overflowing the stack - which we would rather avoid at this late
1315 // state. Also using the tasks' local queues removes the potential
1316 // of the workers interfering with each other that could occur if
1317 // operating on the global stack.
1318
1319 class G1CMKeepAliveAndDrainClosure : public OopClosure {
1320 G1ConcurrentMark* _cm;
1321 G1CMTask* _task;
1322 int _ref_counter_limit;
1323 int _ref_counter;
1324 bool _is_serial;
1325 public:
1326 G1CMKeepAliveAndDrainClosure(G1ConcurrentMark* cm, G1CMTask* task, bool is_serial) :
1327 _cm(cm), _task(task), _is_serial(is_serial),
1328 _ref_counter_limit(G1RefProcDrainInterval) {
1329 assert(_ref_counter_limit > 0, "sanity");
1330 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1331 _ref_counter = _ref_counter_limit;
1332 }
1333
1334 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
1335 virtual void do_oop( oop* p) { do_oop_work(p); }
1336
1337 template <class T> void do_oop_work(T* p) {
1338 if (!_cm->has_overflown()) {
1339 _task->deal_with_reference(p);
1340 _ref_counter--;
1341
1342 if (_ref_counter == 0) {
1343 // We have dealt with _ref_counter_limit references, pushing them
1344 // and objects reachable from them on to the local stack (and
1345 // possibly the global stack). Call G1CMTask::do_marking_step() to
1346 // process these entries.
1347 //
1348 // We call G1CMTask::do_marking_step() in a loop, which we'll exit if
1349 // there's nothing more to do (i.e. we're done with the entries that
1350 // were pushed as a result of the G1CMTask::deal_with_reference() calls
1351 // above) or we overflow.
1352 //
1353 // Note: G1CMTask::do_marking_step() can set the G1CMTask::has_aborted()
1354 // flag while there may still be some work to do. (See the comment at
1355 // the beginning of G1CMTask::do_marking_step() for those conditions -
1356 // one of which is reaching the specified time target.) It is only
1357 // when G1CMTask::do_marking_step() returns without setting the
1358 // has_aborted() flag that the marking step has completed.
1359 do {
1360 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1361 _task->do_marking_step(mark_step_duration_ms,
1362 false /* do_termination */,
1363 _is_serial);
1364 } while (_task->has_aborted() && !_cm->has_overflown());
1365 _ref_counter = _ref_counter_limit;
1366 }
1367 }
1368 }
1369 };
1370
1371 // 'Drain' oop closure used by both serial and parallel reference processing.
1372 // Uses the G1CMTask associated with a given worker thread (for serial
1373 // reference processing the G1CMtask for worker 0 is used). Calls the
1374 // do_marking_step routine, with an unbelievably large timeout value,
1375 // to drain the marking data structures of the remaining entries
1376 // added by the 'keep alive' oop closure above.
1377
1378 class G1CMDrainMarkingStackClosure : public VoidClosure {
1379 G1ConcurrentMark* _cm;
1380 G1CMTask* _task;
1381 bool _is_serial;
1382 public:
1383 G1CMDrainMarkingStackClosure(G1ConcurrentMark* cm, G1CMTask* task, bool is_serial) :
1384 _cm(cm), _task(task), _is_serial(is_serial) {
1385 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1386 }
1387
1388 void do_void() {
1389 do {
1390 // We call G1CMTask::do_marking_step() to completely drain the local
1391 // and global marking stacks of entries pushed by the 'keep alive'
1392 // oop closure (an instance of G1CMKeepAliveAndDrainClosure above).
1393 //
1394 // G1CMTask::do_marking_step() is called in a loop, which we'll exit
1395 // if there's nothing more to do (i.e. we've completely drained the
1396 // entries that were pushed as a a result of applying the 'keep alive'
1397 // closure to the entries on the discovered ref lists) or we overflow
1398 // the global marking stack.
1399 //
1400 // Note: G1CMTask::do_marking_step() can set the G1CMTask::has_aborted()
1401 // flag while there may still be some work to do. (See the comment at
1402 // the beginning of G1CMTask::do_marking_step() for those conditions -
1403 // one of which is reaching the specified time target.) It is only
1404 // when G1CMTask::do_marking_step() returns without setting the
1405 // has_aborted() flag that the marking step has completed.
1406
1407 _task->do_marking_step(1000000000.0 /* something very large */,
1408 true /* do_termination */,
1409 _is_serial);
1410 } while (_task->has_aborted() && !_cm->has_overflown());
1411 }
1412 };
1413
1414 // Implementation of AbstractRefProcTaskExecutor for parallel
1415 // reference processing at the end of G1 concurrent marking
1416
1417 class G1CMRefProcTaskExecutor : public AbstractRefProcTaskExecutor {
1418 private:
1419 G1CollectedHeap* _g1h;
1420 G1ConcurrentMark* _cm;
1421 WorkGang* _workers;
1422 uint _active_workers;
1423
1424 public:
1425 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
1426 G1ConcurrentMark* cm,
1427 WorkGang* workers,
1428 uint n_workers) :
1429 _g1h(g1h), _cm(cm),
1430 _workers(workers), _active_workers(n_workers) { }
1431
1432 // Executes the given task using concurrent marking worker threads.
1433 virtual void execute(ProcessTask& task);
1434 virtual void execute(EnqueueTask& task);
1435 };
1436
1437 class G1CMRefProcTaskProxy : public AbstractGangTask {
1438 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
1439 ProcessTask& _proc_task;
1440 G1CollectedHeap* _g1h;
1441 G1ConcurrentMark* _cm;
1442
1443 public:
1444 G1CMRefProcTaskProxy(ProcessTask& proc_task,
1445 G1CollectedHeap* g1h,
1446 G1ConcurrentMark* cm) :
1447 AbstractGangTask("Process reference objects in parallel"),
1448 _proc_task(proc_task), _g1h(g1h), _cm(cm) {
1449 ReferenceProcessor* rp = _g1h->ref_processor_cm();
1450 assert(rp->processing_is_mt(), "shouldn't be here otherwise");
1451 }
1452
1453 virtual void work(uint worker_id) {
1454 ResourceMark rm;
1455 HandleMark hm;
1456 G1CMTask* task = _cm->task(worker_id);
1457 G1CMIsAliveClosure g1_is_alive(_g1h);
1458 G1CMKeepAliveAndDrainClosure g1_par_keep_alive(_cm, task, false /* is_serial */);
1459 G1CMDrainMarkingStackClosure g1_par_drain(_cm, task, false /* is_serial */);
1460
1461 _proc_task.work(worker_id, g1_is_alive, g1_par_keep_alive, g1_par_drain);
1462 }
1463 };
1464
1465 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
1466 assert(_workers != NULL, "Need parallel worker threads.");
1467 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
1468
1469 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
1470
1471 // We need to reset the concurrency level before each
1472 // proxy task execution, so that the termination protocol
1473 // and overflow handling in G1CMTask::do_marking_step() knows
1474 // how many workers to wait for.
1475 _cm->set_concurrency(_active_workers);
1476 _workers->run_task(&proc_task_proxy);
1477 }
1478
1479 class G1CMRefEnqueueTaskProxy : public AbstractGangTask {
1480 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
1481 EnqueueTask& _enq_task;
1482
1483 public:
1484 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
1485 AbstractGangTask("Enqueue reference objects in parallel"),
1486 _enq_task(enq_task) { }
1487
1488 virtual void work(uint worker_id) {
1489 _enq_task.work(worker_id);
1490 }
1491 };
1492
1493 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
1494 assert(_workers != NULL, "Need parallel worker threads.");
1495 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
1496
1497 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
1498
1499 // Not strictly necessary but...
1500 //
1501 // We need to reset the concurrency level before each
1502 // proxy task execution, so that the termination protocol
1503 // and overflow handling in G1CMTask::do_marking_step() knows
1504 // how many workers to wait for.
1505 _cm->set_concurrency(_active_workers);
1506 _workers->run_task(&enq_task_proxy);
1507 }
1508
1509 void G1ConcurrentMark::weak_refs_work(bool clear_all_soft_refs) {
1510 ResourceMark rm;
1511 HandleMark hm;
1512
1513 // Is alive closure.
1514 G1CMIsAliveClosure g1_is_alive(_g1h);
1515
1516 // Inner scope to exclude the cleaning of the string and symbol
1517 // tables from the displayed time.
1518 {
1519 GCTraceTime(Debug, gc, phases) trace("Reference Processing", _gc_timer_cm);
1520
1521 ReferenceProcessor* rp = _g1h->ref_processor_cm();
1522
1523 // See the comment in G1CollectedHeap::ref_processing_init()
1524 // about how reference processing currently works in G1.
1525
1526 // Set the soft reference policy
1527 rp->setup_policy(clear_all_soft_refs);
1528 assert(_global_mark_stack.is_empty(), "mark stack should be empty");
1529
1530 // Instances of the 'Keep Alive' and 'Complete GC' closures used
1531 // in serial reference processing. Note these closures are also
1532 // used for serially processing (by the the current thread) the
1533 // JNI references during parallel reference processing.
1534 //
1535 // These closures do not need to synchronize with the worker
1536 // threads involved in parallel reference processing as these
1537 // instances are executed serially by the current thread (e.g.
1538 // reference processing is not multi-threaded and is thus
1539 // performed by the current thread instead of a gang worker).
1540 //
1541 // The gang tasks involved in parallel reference processing create
1542 // their own instances of these closures, which do their own
1543 // synchronization among themselves.
1544 G1CMKeepAliveAndDrainClosure g1_keep_alive(this, task(0), true /* is_serial */);
1545 G1CMDrainMarkingStackClosure g1_drain_mark_stack(this, task(0), true /* is_serial */);
1546
1547 // We need at least one active thread. If reference processing
1548 // is not multi-threaded we use the current (VMThread) thread,
1549 // otherwise we use the work gang from the G1CollectedHeap and
1550 // we utilize all the worker threads we can.
1551 bool processing_is_mt = rp->processing_is_mt();
1552 uint active_workers = (processing_is_mt ? _g1h->workers()->active_workers() : 1U);
1553 active_workers = MAX2(MIN2(active_workers, _max_num_tasks), 1U);
1554
1555 // Parallel processing task executor.
1556 G1CMRefProcTaskExecutor par_task_executor(_g1h, this,
1557 _g1h->workers(), active_workers);
1558 AbstractRefProcTaskExecutor* executor = (processing_is_mt ? &par_task_executor : NULL);
1559
1560 // Set the concurrency level. The phase was already set prior to
1561 // executing the remark task.
1562 set_concurrency(active_workers);
1563
1564 // Set the degree of MT processing here. If the discovery was done MT,
1565 // the number of threads involved during discovery could differ from
1566 // the number of active workers. This is OK as long as the discovered
1567 // Reference lists are balanced (see balance_all_queues() and balance_queues()).
1568 rp->set_active_mt_degree(active_workers);
1569
1570 ReferenceProcessorPhaseTimes pt(_gc_timer_cm, rp->num_q());
1571
1572 // Process the weak references.
1573 const ReferenceProcessorStats& stats =
1574 rp->process_discovered_references(&g1_is_alive,
1575 &g1_keep_alive,
1576 &g1_drain_mark_stack,
1577 executor,
1578 &pt);
1579 _gc_tracer_cm->report_gc_reference_stats(stats);
1580 pt.print_all_references();
1581
1582 // The do_oop work routines of the keep_alive and drain_marking_stack
1583 // oop closures will set the has_overflown flag if we overflow the
1584 // global marking stack.
1585
1586 assert(has_overflown() || _global_mark_stack.is_empty(),
1587 "Mark stack should be empty (unless it has overflown)");
1588
1589 assert(rp->num_q() == active_workers, "why not");
1590
1591 rp->enqueue_discovered_references(executor, &pt);
1592
1593 rp->verify_no_references_recorded();
1594
1595 pt.print_enqueue_phase();
1596
1597 assert(!rp->discovery_enabled(), "Post condition");
1598 }
1599
1600 assert(has_overflown() || _global_mark_stack.is_empty(),
1601 "Mark stack should be empty (unless it has overflown)");
1602
1603 {
1604 GCTraceTime(Debug, gc, phases) debug("Weak Processing", _gc_timer_cm);
1605 WeakProcessor::weak_oops_do(&g1_is_alive, &do_nothing_cl);
1606 }
1607
1608 if (has_overflown()) {
1609 // We can not trust g1_is_alive if the marking stack overflowed
1610 return;
1611 }
1612
1613 assert(_global_mark_stack.is_empty(), "Marking should have completed");
1614
1615 // Unload Klasses, String, Symbols, Code Cache, etc.
1616 if (ClassUnloadingWithConcurrentMark) {
1617 GCTraceTime(Debug, gc, phases) debug("Class Unloading", _gc_timer_cm);
1618 bool purged_classes = SystemDictionary::do_unloading(&g1_is_alive, _gc_timer_cm, false /* Defer cleaning */);
1619 _g1h->complete_cleaning(&g1_is_alive, purged_classes);
1620 } else {
1621 GCTraceTime(Debug, gc, phases) debug("Cleanup", _gc_timer_cm);
1622 // No need to clean string table and symbol table as they are treated as strong roots when
1623 // class unloading is disabled.
1624 _g1h->partial_cleaning(&g1_is_alive, false, false, G1StringDedup::is_enabled());
1625 }
1626 }
1627
1628 void G1ConcurrentMark::report_object_count() {
1629 G1CMIsAliveClosure is_alive(_g1h);
1630 _gc_tracer_cm->report_object_count_after_gc(&is_alive);
1631 }
1632
1633 void G1ConcurrentMark::swap_mark_bitmaps() {
1634 G1CMBitMap* temp = _prev_mark_bitmap;
1635 _prev_mark_bitmap = _next_mark_bitmap;
1636 _next_mark_bitmap = temp;
1637 _g1h->collector_state()->set_clearing_next_bitmap(true);
1638 }
1639
1640 // Closure for marking entries in SATB buffers.
1641 class G1CMSATBBufferClosure : public SATBBufferClosure {
1642 private:
1643 G1CMTask* _task;
1644 G1CollectedHeap* _g1h;
1645
1646 // This is very similar to G1CMTask::deal_with_reference, but with
1647 // more relaxed requirements for the argument, so this must be more
1648 // circumspect about treating the argument as an object.
1649 void do_entry(void* entry) const {
1650 _task->increment_refs_reached();
1651 oop const obj = static_cast<oop>(entry);
1652 _task->make_reference_grey(obj);
1653 }
1654
1655 public:
1656 G1CMSATBBufferClosure(G1CMTask* task, G1CollectedHeap* g1h)
1657 : _task(task), _g1h(g1h) { }
1658
1659 virtual void do_buffer(void** buffer, size_t size) {
1660 for (size_t i = 0; i < size; ++i) {
1661 do_entry(buffer[i]);
1662 }
1663 }
1664 };
1665
1666 class G1RemarkThreadsClosure : public ThreadClosure {
1667 G1CMSATBBufferClosure _cm_satb_cl;
1668 G1CMOopClosure _cm_cl;
1669 MarkingCodeBlobClosure _code_cl;
1670 int _thread_parity;
1671
1672 public:
1673 G1RemarkThreadsClosure(G1CollectedHeap* g1h, G1CMTask* task) :
1674 _cm_satb_cl(task, g1h),
1675 _cm_cl(g1h, task),
1676 _code_cl(&_cm_cl, !CodeBlobToOopClosure::FixRelocations),
1677 _thread_parity(Threads::thread_claim_parity()) {}
1678
1679 void do_thread(Thread* thread) {
1680 if (thread->is_Java_thread()) {
1681 if (thread->claim_oops_do(true, _thread_parity)) {
1682 JavaThread* jt = (JavaThread*)thread;
1683
1684 // In theory it should not be neccessary to explicitly walk the nmethods to find roots for concurrent marking
1685 // however the liveness of oops reachable from nmethods have very complex lifecycles:
1686 // * Alive if on the stack of an executing method
1687 // * Weakly reachable otherwise
1688 // Some objects reachable from nmethods, such as the class loader (or klass_holder) of the receiver should be
1689 // live by the SATB invariant but other oops recorded in nmethods may behave differently.
1690 jt->nmethods_do(&_code_cl);
1691
1692 jt->satb_mark_queue().apply_closure_and_empty(&_cm_satb_cl);
1693 }
1694 } else if (thread->is_VM_thread()) {
1695 if (thread->claim_oops_do(true, _thread_parity)) {
1696 JavaThread::satb_mark_queue_set().shared_satb_queue()->apply_closure_and_empty(&_cm_satb_cl);
1697 }
1698 }
1699 }
1700 };
1701
1702 class G1CMRemarkTask : public AbstractGangTask {
1703 G1ConcurrentMark* _cm;
1704 public:
1705 void work(uint worker_id) {
1706 G1CMTask* task = _cm->task(worker_id);
1707 task->record_start_time();
1708 {
1709 ResourceMark rm;
1710 HandleMark hm;
1711
1712 G1RemarkThreadsClosure threads_f(G1CollectedHeap::heap(), task);
1713 Threads::threads_do(&threads_f);
1714 }
1715
1716 do {
1717 task->do_marking_step(1000000000.0 /* something very large */,
1718 true /* do_termination */,
1719 false /* is_serial */);
1720 } while (task->has_aborted() && !_cm->has_overflown());
1721 // If we overflow, then we do not want to restart. We instead
1722 // want to abort remark and do concurrent marking again.
1723 task->record_end_time();
1724 }
1725
1726 G1CMRemarkTask(G1ConcurrentMark* cm, uint active_workers) :
1727 AbstractGangTask("Par Remark"), _cm(cm) {
1728 _cm->terminator()->reset_for_reuse(active_workers);
1729 }
1730 };
1731
1732 void G1ConcurrentMark::finalize_marking() {
1733 ResourceMark rm;
1734 HandleMark hm;
1735
1736 _g1h->ensure_parsability(false);
1737
1738 // this is remark, so we'll use up all active threads
1739 uint active_workers = _g1h->workers()->active_workers();
1740 set_concurrency_and_phase(active_workers, false /* concurrent */);
1741 // Leave _parallel_marking_threads at it's
1742 // value originally calculated in the G1ConcurrentMark
1743 // constructor and pass values of the active workers
1744 // through the gang in the task.
1745
1746 {
1747 StrongRootsScope srs(active_workers);
1748
1749 G1CMRemarkTask remarkTask(this, active_workers);
1750 // We will start all available threads, even if we decide that the
1751 // active_workers will be fewer. The extra ones will just bail out
1752 // immediately.
1753 _g1h->workers()->run_task(&remarkTask);
1754 }
1755
1756 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1757 guarantee(has_overflown() ||
1758 satb_mq_set.completed_buffers_num() == 0,
1759 "Invariant: has_overflown = %s, num buffers = " SIZE_FORMAT,
1760 BOOL_TO_STR(has_overflown()),
1761 satb_mq_set.completed_buffers_num());
1762
1763 print_stats();
1764 }
1765
1766 void G1ConcurrentMark::flush_all_task_caches() {
1767 size_t hits = 0;
1768 size_t misses = 0;
1769 for (uint i = 0; i < _max_num_tasks; i++) {
1770 Pair<size_t, size_t> stats = _tasks[i]->flush_mark_stats_cache();
1771 hits += stats.first;
1772 misses += stats.second;
1773 }
1774 size_t sum = hits + misses;
1775 log_debug(gc, stats)("Mark stats cache hits " SIZE_FORMAT " misses " SIZE_FORMAT " ratio %1.3lf",
1776 hits, misses, percent_of(hits, sum));
1777 }
1778
1779 void G1ConcurrentMark::clear_range_in_prev_bitmap(MemRegion mr) {
1780 _prev_mark_bitmap->clear_range(mr);
1781 }
1782
1783 HeapRegion*
1784 G1ConcurrentMark::claim_region(uint worker_id) {
1785 // "checkpoint" the finger
1786 HeapWord* finger = _finger;
1787
1788 while (finger < _heap.end()) {
1789 assert(_g1h->is_in_g1_reserved(finger), "invariant");
1790
1791 HeapRegion* curr_region = _g1h->heap_region_containing(finger);
1792 // Make sure that the reads below do not float before loading curr_region.
1793 OrderAccess::loadload();
1794 // Above heap_region_containing may return NULL as we always scan claim
1795 // until the end of the heap. In this case, just jump to the next region.
1796 HeapWord* end = curr_region != NULL ? curr_region->end() : finger + HeapRegion::GrainWords;
1797
1798 // Is the gap between reading the finger and doing the CAS too long?
1799 HeapWord* res = Atomic::cmpxchg(end, &_finger, finger);
1800 if (res == finger && curr_region != NULL) {
1801 // we succeeded
1802 HeapWord* bottom = curr_region->bottom();
1803 HeapWord* limit = curr_region->next_top_at_mark_start();
1804
1805 // notice that _finger == end cannot be guaranteed here since,
1806 // someone else might have moved the finger even further
1807 assert(_finger >= end, "the finger should have moved forward");
1808
1809 if (limit > bottom) {
1810 return curr_region;
1811 } else {
1812 assert(limit == bottom,
1813 "the region limit should be at bottom");
1814 // we return NULL and the caller should try calling
1815 // claim_region() again.
1816 return NULL;
1817 }
1818 } else {
1819 assert(_finger > finger, "the finger should have moved forward");
1820 // read it again
1821 finger = _finger;
1822 }
1823 }
1824
1825 return NULL;
1826 }
1827
1828 #ifndef PRODUCT
1829 class VerifyNoCSetOops {
1830 G1CollectedHeap* _g1h;
1831 const char* _phase;
1832 int _info;
1833
1834 public:
1835 VerifyNoCSetOops(const char* phase, int info = -1) :
1836 _g1h(G1CollectedHeap::heap()),
1837 _phase(phase),
1838 _info(info)
1839 { }
1840
1841 void operator()(G1TaskQueueEntry task_entry) const {
1842 if (task_entry.is_array_slice()) {
1843 guarantee(_g1h->is_in_reserved(task_entry.slice()), "Slice " PTR_FORMAT " must be in heap.", p2i(task_entry.slice()));
1844 return;
1845 }
1846 guarantee(oopDesc::is_oop(task_entry.obj()),
1847 "Non-oop " PTR_FORMAT ", phase: %s, info: %d",
1848 p2i(task_entry.obj()), _phase, _info);
1849 guarantee(!_g1h->is_in_cset(task_entry.obj()),
1850 "obj: " PTR_FORMAT " in CSet, phase: %s, info: %d",
1851 p2i(task_entry.obj()), _phase, _info);
1852 }
1853 };
1854
1855 void G1ConcurrentMark::verify_no_cset_oops() {
1856 assert(SafepointSynchronize::is_at_safepoint(), "should be at a safepoint");
1857 if (!_g1h->collector_state()->mark_or_rebuild_in_progress()) {
1858 return;
1859 }
1860
1861 // Verify entries on the global mark stack
1862 _global_mark_stack.iterate(VerifyNoCSetOops("Stack"));
1863
1864 // Verify entries on the task queues
1865 for (uint i = 0; i < _max_num_tasks; ++i) {
1866 G1CMTaskQueue* queue = _task_queues->queue(i);
1867 queue->iterate(VerifyNoCSetOops("Queue", i));
1868 }
1869
1870 // Verify the global finger
1871 HeapWord* global_finger = finger();
1872 if (global_finger != NULL && global_finger < _heap.end()) {
1873 // Since we always iterate over all regions, we might get a NULL HeapRegion
1874 // here.
1875 HeapRegion* global_hr = _g1h->heap_region_containing(global_finger);
1876 guarantee(global_hr == NULL || global_finger == global_hr->bottom(),
1877 "global finger: " PTR_FORMAT " region: " HR_FORMAT,
1878 p2i(global_finger), HR_FORMAT_PARAMS(global_hr));
1879 }
1880
1881 // Verify the task fingers
1882 assert(_num_concurrent_workers <= _max_num_tasks, "sanity");
1883 for (uint i = 0; i < _num_concurrent_workers; ++i) {
1884 G1CMTask* task = _tasks[i];
1885 HeapWord* task_finger = task->finger();
1886 if (task_finger != NULL && task_finger < _heap.end()) {
1887 // See above note on the global finger verification.
1888 HeapRegion* task_hr = _g1h->heap_region_containing(task_finger);
1889 guarantee(task_hr == NULL || task_finger == task_hr->bottom() ||
1890 !task_hr->in_collection_set(),
1891 "task finger: " PTR_FORMAT " region: " HR_FORMAT,
1892 p2i(task_finger), HR_FORMAT_PARAMS(task_hr));
1893 }
1894 }
1895 }
1896 #endif // PRODUCT
1897
1898 void G1ConcurrentMark::rebuild_rem_set_concurrently() {
1899 _g1h->g1_rem_set()->rebuild_rem_set(this, _concurrent_workers, _worker_id_offset);
1900 }
1901
1902 void G1ConcurrentMark::print_stats() {
1903 if (!log_is_enabled(Debug, gc, stats)) {
1904 return;
1905 }
1906 log_debug(gc, stats)("---------------------------------------------------------------------");
1907 for (size_t i = 0; i < _num_active_tasks; ++i) {
1908 _tasks[i]->print_stats();
1909 log_debug(gc, stats)("---------------------------------------------------------------------");
1910 }
1911 }
1912
1913 void G1ConcurrentMark::concurrent_cycle_abort() {
1914 if (!cm_thread()->during_cycle() || _has_aborted) {
1915 // We haven't started a concurrent cycle or we have already aborted it. No need to do anything.
1916 return;
1917 }
1918
1919 // Clear all marks in the next bitmap for the next marking cycle. This will allow us to skip the next
1920 // concurrent bitmap clearing.
1921 {
1922 GCTraceTime(Debug, gc)("Clear Next Bitmap");
1923 clear_bitmap(_next_mark_bitmap, _g1h->workers(), false);
1924 }
1925 // Note we cannot clear the previous marking bitmap here
1926 // since VerifyDuringGC verifies the objects marked during
1927 // a full GC against the previous bitmap.
1928
1929 // Empty mark stack
1930 reset_marking_for_restart();
1931 for (uint i = 0; i < _max_num_tasks; ++i) {
1932 _tasks[i]->clear_region_fields();
1933 }
1934 _first_overflow_barrier_sync.abort();
1935 _second_overflow_barrier_sync.abort();
1936 _has_aborted = true;
1937
1938 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1939 satb_mq_set.abandon_partial_marking();
1940 // This can be called either during or outside marking, we'll read
1941 // the expected_active value from the SATB queue set.
1942 satb_mq_set.set_active_all_threads(
1943 false, /* new active value */
1944 satb_mq_set.is_active() /* expected_active */);
1945 }
1946
1947 static void print_ms_time_info(const char* prefix, const char* name,
1948 NumberSeq& ns) {
1949 log_trace(gc, marking)("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
1950 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
1951 if (ns.num() > 0) {
1952 log_trace(gc, marking)("%s [std. dev = %8.2f ms, max = %8.2f ms]",
1953 prefix, ns.sd(), ns.maximum());
1954 }
1955 }
1956
1957 void G1ConcurrentMark::print_summary_info() {
1958 Log(gc, marking) log;
1959 if (!log.is_trace()) {
1960 return;
1961 }
1962
1963 log.trace(" Concurrent marking:");
1964 print_ms_time_info(" ", "init marks", _init_times);
1965 print_ms_time_info(" ", "remarks", _remark_times);
1966 {
1967 print_ms_time_info(" ", "final marks", _remark_mark_times);
1968 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
1969
1970 }
1971 print_ms_time_info(" ", "cleanups", _cleanup_times);
1972 log.trace(" Finalize live data total time = %8.2f s (avg = %8.2f ms).",
1973 _total_cleanup_time, (_cleanup_times.num() > 0 ? _total_cleanup_time * 1000.0 / (double)_cleanup_times.num() : 0.0));
1974 log.trace(" Total stop_world time = %8.2f s.",
1975 (_init_times.sum() + _remark_times.sum() + _cleanup_times.sum())/1000.0);
1976 log.trace(" Total concurrent time = %8.2f s (%8.2f s marking).",
1977 cm_thread()->vtime_accum(), cm_thread()->vtime_mark_accum());
1978 }
1979
1980 void G1ConcurrentMark::print_worker_threads_on(outputStream* st) const {
1981 _concurrent_workers->print_worker_threads_on(st);
1982 }
1983
1984 void G1ConcurrentMark::threads_do(ThreadClosure* tc) const {
1985 _concurrent_workers->threads_do(tc);
1986 }
1987
1988 void G1ConcurrentMark::print_on_error(outputStream* st) const {
1989 st->print_cr("Marking Bits (Prev, Next): (CMBitMap*) " PTR_FORMAT ", (CMBitMap*) " PTR_FORMAT,
1990 p2i(_prev_mark_bitmap), p2i(_next_mark_bitmap));
1991 _prev_mark_bitmap->print_on_error(st, " Prev Bits: ");
1992 _next_mark_bitmap->print_on_error(st, " Next Bits: ");
1993 }
1994
1995 static ReferenceProcessor* get_cm_oop_closure_ref_processor(G1CollectedHeap* g1h) {
1996 ReferenceProcessor* result = g1h->ref_processor_cm();
1997 assert(result != NULL, "CM reference processor should not be NULL");
1998 return result;
1999 }
2000
2001 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
2002 G1CMTask* task)
2003 : MetadataAwareOopClosure(get_cm_oop_closure_ref_processor(g1h)),
2004 _g1h(g1h), _task(task)
2005 { }
2006
2007 void G1CMTask::setup_for_region(HeapRegion* hr) {
2008 assert(hr != NULL,
2009 "claim_region() should have filtered out NULL regions");
2010 _curr_region = hr;
2011 _finger = hr->bottom();
2012 update_region_limit();
2013 }
2014
2015 void G1CMTask::update_region_limit() {
2016 HeapRegion* hr = _curr_region;
2017 HeapWord* bottom = hr->bottom();
2018 HeapWord* limit = hr->next_top_at_mark_start();
2019
2020 if (limit == bottom) {
2021 // The region was collected underneath our feet.
2022 // We set the finger to bottom to ensure that the bitmap
2023 // iteration that will follow this will not do anything.
2024 // (this is not a condition that holds when we set the region up,
2025 // as the region is not supposed to be empty in the first place)
2026 _finger = bottom;
2027 } else if (limit >= _region_limit) {
2028 assert(limit >= _finger, "peace of mind");
2029 } else {
2030 assert(limit < _region_limit, "only way to get here");
2031 // This can happen under some pretty unusual circumstances. An
2032 // evacuation pause empties the region underneath our feet (NTAMS
2033 // at bottom). We then do some allocation in the region (NTAMS
2034 // stays at bottom), followed by the region being used as a GC
2035 // alloc region (NTAMS will move to top() and the objects
2036 // originally below it will be grayed). All objects now marked in
2037 // the region are explicitly grayed, if below the global finger,
2038 // and we do not need in fact to scan anything else. So, we simply
2039 // set _finger to be limit to ensure that the bitmap iteration
2040 // doesn't do anything.
2041 _finger = limit;
2042 }
2043
2044 _region_limit = limit;
2045 }
2046
2047 void G1CMTask::giveup_current_region() {
2048 assert(_curr_region != NULL, "invariant");
2049 clear_region_fields();
2050 }
2051
2052 void G1CMTask::clear_region_fields() {
2053 // Values for these three fields that indicate that we're not
2054 // holding on to a region.
2055 _curr_region = NULL;
2056 _finger = NULL;
2057 _region_limit = NULL;
2058 }
2059
2060 void G1CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
2061 if (cm_oop_closure == NULL) {
2062 assert(_cm_oop_closure != NULL, "invariant");
2063 } else {
2064 assert(_cm_oop_closure == NULL, "invariant");
2065 }
2066 _cm_oop_closure = cm_oop_closure;
2067 }
2068
2069 void G1CMTask::reset(G1CMBitMap* next_mark_bitmap) {
2070 guarantee(next_mark_bitmap != NULL, "invariant");
2071 _next_mark_bitmap = next_mark_bitmap;
2072 clear_region_fields();
2073
2074 _calls = 0;
2075 _elapsed_time_ms = 0.0;
2076 _termination_time_ms = 0.0;
2077 _termination_start_time_ms = 0.0;
2078
2079 _mark_stats_cache.reset();
2080 }
2081
2082 bool G1CMTask::should_exit_termination() {
2083 regular_clock_call();
2084 // This is called when we are in the termination protocol. We should
2085 // quit if, for some reason, this task wants to abort or the global
2086 // stack is not empty (this means that we can get work from it).
2087 return !_cm->mark_stack_empty() || has_aborted();
2088 }
2089
2090 void G1CMTask::reached_limit() {
2091 assert(_words_scanned >= _words_scanned_limit ||
2092 _refs_reached >= _refs_reached_limit ,
2093 "shouldn't have been called otherwise");
2094 regular_clock_call();
2095 }
2096
2097 void G1CMTask::regular_clock_call() {
2098 if (has_aborted()) {
2099 return;
2100 }
2101
2102 // First, we need to recalculate the words scanned and refs reached
2103 // limits for the next clock call.
2104 recalculate_limits();
2105
2106 // During the regular clock call we do the following
2107
2108 // (1) If an overflow has been flagged, then we abort.
2109 if (_cm->has_overflown()) {
2110 set_has_aborted();
2111 return;
2112 }
2113
2114 // If we are not concurrent (i.e. we're doing remark) we don't need
2115 // to check anything else. The other steps are only needed during
2116 // the concurrent marking phase.
2117 if (!_cm->concurrent()) {
2118 return;
2119 }
2120
2121 // (2) If marking has been aborted for Full GC, then we also abort.
2122 if (_cm->has_aborted()) {
2123 set_has_aborted();
2124 return;
2125 }
2126
2127 double curr_time_ms = os::elapsedVTime() * 1000.0;
2128
2129 // (4) We check whether we should yield. If we have to, then we abort.
2130 if (SuspendibleThreadSet::should_yield()) {
2131 // We should yield. To do this we abort the task. The caller is
2132 // responsible for yielding.
2133 set_has_aborted();
2134 return;
2135 }
2136
2137 // (5) We check whether we've reached our time quota. If we have,
2138 // then we abort.
2139 double elapsed_time_ms = curr_time_ms - _start_time_ms;
2140 if (elapsed_time_ms > _time_target_ms) {
2141 set_has_aborted();
2142 _has_timed_out = true;
2143 return;
2144 }
2145
2146 // (6) Finally, we check whether there are enough completed STAB
2147 // buffers available for processing. If there are, we abort.
2148 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2149 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
2150 // we do need to process SATB buffers, we'll abort and restart
2151 // the marking task to do so
2152 set_has_aborted();
2153 return;
2154 }
2155 }
2156
2157 void G1CMTask::recalculate_limits() {
2158 _real_words_scanned_limit = _words_scanned + words_scanned_period;
2159 _words_scanned_limit = _real_words_scanned_limit;
2160
2161 _real_refs_reached_limit = _refs_reached + refs_reached_period;
2162 _refs_reached_limit = _real_refs_reached_limit;
2163 }
2164
2165 void G1CMTask::decrease_limits() {
2166 // This is called when we believe that we're going to do an infrequent
2167 // operation which will increase the per byte scanned cost (i.e. move
2168 // entries to/from the global stack). It basically tries to decrease the
2169 // scanning limit so that the clock is called earlier.
2170
2171 _words_scanned_limit = _real_words_scanned_limit - 3 * words_scanned_period / 4;
2172 _refs_reached_limit = _real_refs_reached_limit - 3 * refs_reached_period / 4;
2173 }
2174
2175 void G1CMTask::move_entries_to_global_stack() {
2176 // Local array where we'll store the entries that will be popped
2177 // from the local queue.
2178 G1TaskQueueEntry buffer[G1CMMarkStack::EntriesPerChunk];
2179
2180 size_t n = 0;
2181 G1TaskQueueEntry task_entry;
2182 while (n < G1CMMarkStack::EntriesPerChunk && _task_queue->pop_local(task_entry)) {
2183 buffer[n] = task_entry;
2184 ++n;
2185 }
2186 if (n < G1CMMarkStack::EntriesPerChunk) {
2187 buffer[n] = G1TaskQueueEntry();
2188 }
2189
2190 if (n > 0) {
2191 if (!_cm->mark_stack_push(buffer)) {
2192 set_has_aborted();
2193 }
2194 }
2195
2196 // This operation was quite expensive, so decrease the limits.
2197 decrease_limits();
2198 }
2199
2200 bool G1CMTask::get_entries_from_global_stack() {
2201 // Local array where we'll store the entries that will be popped
2202 // from the global stack.
2203 G1TaskQueueEntry buffer[G1CMMarkStack::EntriesPerChunk];
2204
2205 if (!_cm->mark_stack_pop(buffer)) {
2206 return false;
2207 }
2208
2209 // We did actually pop at least one entry.
2210 for (size_t i = 0; i < G1CMMarkStack::EntriesPerChunk; ++i) {
2211 G1TaskQueueEntry task_entry = buffer[i];
2212 if (task_entry.is_null()) {
2213 break;
2214 }
2215 assert(task_entry.is_array_slice() || oopDesc::is_oop(task_entry.obj()), "Element " PTR_FORMAT " must be an array slice or oop", p2i(task_entry.obj()));
2216 bool success = _task_queue->push(task_entry);
2217 // We only call this when the local queue is empty or under a
2218 // given target limit. So, we do not expect this push to fail.
2219 assert(success, "invariant");
2220 }
2221
2222 // This operation was quite expensive, so decrease the limits
2223 decrease_limits();
2224 return true;
2225 }
2226
2227 void G1CMTask::drain_local_queue(bool partially) {
2228 if (has_aborted()) {
2229 return;
2230 }
2231
2232 // Decide what the target size is, depending whether we're going to
2233 // drain it partially (so that other tasks can steal if they run out
2234 // of things to do) or totally (at the very end).
2235 size_t target_size;
2236 if (partially) {
2237 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
2238 } else {
2239 target_size = 0;
2240 }
2241
2242 if (_task_queue->size() > target_size) {
2243 G1TaskQueueEntry entry;
2244 bool ret = _task_queue->pop_local(entry);
2245 while (ret) {
2246 scan_task_entry(entry);
2247 if (_task_queue->size() <= target_size || has_aborted()) {
2248 ret = false;
2249 } else {
2250 ret = _task_queue->pop_local(entry);
2251 }
2252 }
2253 }
2254 }
2255
2256 void G1CMTask::drain_global_stack(bool partially) {
2257 if (has_aborted()) {
2258 return;
2259 }
2260
2261 // We have a policy to drain the local queue before we attempt to
2262 // drain the global stack.
2263 assert(partially || _task_queue->size() == 0, "invariant");
2264
2265 // Decide what the target size is, depending whether we're going to
2266 // drain it partially (so that other tasks can steal if they run out
2267 // of things to do) or totally (at the very end).
2268 // Notice that when draining the global mark stack partially, due to the racyness
2269 // of the mark stack size update we might in fact drop below the target. But,
2270 // this is not a problem.
2271 // In case of total draining, we simply process until the global mark stack is
2272 // totally empty, disregarding the size counter.
2273 if (partially) {
2274 size_t const target_size = _cm->partial_mark_stack_size_target();
2275 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
2276 if (get_entries_from_global_stack()) {
2277 drain_local_queue(partially);
2278 }
2279 }
2280 } else {
2281 while (!has_aborted() && get_entries_from_global_stack()) {
2282 drain_local_queue(partially);
2283 }
2284 }
2285 }
2286
2287 // SATB Queue has several assumptions on whether to call the par or
2288 // non-par versions of the methods. this is why some of the code is
2289 // replicated. We should really get rid of the single-threaded version
2290 // of the code to simplify things.
2291 void G1CMTask::drain_satb_buffers() {
2292 if (has_aborted()) {
2293 return;
2294 }
2295
2296 // We set this so that the regular clock knows that we're in the
2297 // middle of draining buffers and doesn't set the abort flag when it
2298 // notices that SATB buffers are available for draining. It'd be
2299 // very counter productive if it did that. :-)
2300 _draining_satb_buffers = true;
2301
2302 G1CMSATBBufferClosure satb_cl(this, _g1h);
2303 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2304
2305 // This keeps claiming and applying the closure to completed buffers
2306 // until we run out of buffers or we need to abort.
2307 while (!has_aborted() &&
2308 satb_mq_set.apply_closure_to_completed_buffer(&satb_cl)) {
2309 regular_clock_call();
2310 }
2311
2312 _draining_satb_buffers = false;
2313
2314 assert(has_aborted() ||
2315 _cm->concurrent() ||
2316 satb_mq_set.completed_buffers_num() == 0, "invariant");
2317
2318 // again, this was a potentially expensive operation, decrease the
2319 // limits to get the regular clock call early
2320 decrease_limits();
2321 }
2322
2323 void G1CMTask::clear_mark_stats_cache(uint region_idx) {
2324 _mark_stats_cache.reset(region_idx);
2325 }
2326
2327 Pair<size_t, size_t> G1CMTask::flush_mark_stats_cache() {
2328 return _mark_stats_cache.evict_all();
2329 }
2330
2331 void G1CMTask::print_stats() {
2332 log_debug(gc, stats)("Marking Stats, task = %u, calls = %u", _worker_id, _calls);
2333 log_debug(gc, stats)(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
2334 _elapsed_time_ms, _termination_time_ms);
2335 log_debug(gc, stats)(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms max = %1.2lfms, total = %1.2lfms",
2336 _step_times_ms.num(),
2337 _step_times_ms.avg(),
2338 _step_times_ms.sd(),
2339 _step_times_ms.maximum(),
2340 _step_times_ms.sum());
2341 size_t const hits = _mark_stats_cache.hits();
2342 size_t const misses = _mark_stats_cache.misses();
2343 log_debug(gc, stats)(" Mark Stats Cache: hits " SIZE_FORMAT " misses " SIZE_FORMAT " ratio %.3f",
2344 hits, misses, percent_of(hits, hits + misses));
2345 }
2346
2347 bool G1ConcurrentMark::try_stealing(uint worker_id, int* hash_seed, G1TaskQueueEntry& task_entry) {
2348 return _task_queues->steal(worker_id, hash_seed, task_entry);
2349 }
2350
2351 /*****************************************************************************
2352
2353 The do_marking_step(time_target_ms, ...) method is the building
2354 block of the parallel marking framework. It can be called in parallel
2355 with other invocations of do_marking_step() on different tasks
2356 (but only one per task, obviously) and concurrently with the
2357 mutator threads, or during remark, hence it eliminates the need
2358 for two versions of the code. When called during remark, it will
2359 pick up from where the task left off during the concurrent marking
2360 phase. Interestingly, tasks are also claimable during evacuation
2361 pauses too, since do_marking_step() ensures that it aborts before
2362 it needs to yield.
2363
2364 The data structures that it uses to do marking work are the
2365 following:
2366
2367 (1) Marking Bitmap. If there are gray objects that appear only
2368 on the bitmap (this happens either when dealing with an overflow
2369 or when the initial marking phase has simply marked the roots
2370 and didn't push them on the stack), then tasks claim heap
2371 regions whose bitmap they then scan to find gray objects. A
2372 global finger indicates where the end of the last claimed region
2373 is. A local finger indicates how far into the region a task has
2374 scanned. The two fingers are used to determine how to gray an
2375 object (i.e. whether simply marking it is OK, as it will be
2376 visited by a task in the future, or whether it needs to be also
2377 pushed on a stack).
2378
2379 (2) Local Queue. The local queue of the task which is accessed
2380 reasonably efficiently by the task. Other tasks can steal from
2381 it when they run out of work. Throughout the marking phase, a
2382 task attempts to keep its local queue short but not totally
2383 empty, so that entries are available for stealing by other
2384 tasks. Only when there is no more work, a task will totally
2385 drain its local queue.
2386
2387 (3) Global Mark Stack. This handles local queue overflow. During
2388 marking only sets of entries are moved between it and the local
2389 queues, as access to it requires a mutex and more fine-grain
2390 interaction with it which might cause contention. If it
2391 overflows, then the marking phase should restart and iterate
2392 over the bitmap to identify gray objects. Throughout the marking
2393 phase, tasks attempt to keep the global mark stack at a small
2394 length but not totally empty, so that entries are available for
2395 popping by other tasks. Only when there is no more work, tasks
2396 will totally drain the global mark stack.
2397
2398 (4) SATB Buffer Queue. This is where completed SATB buffers are
2399 made available. Buffers are regularly removed from this queue
2400 and scanned for roots, so that the queue doesn't get too
2401 long. During remark, all completed buffers are processed, as
2402 well as the filled in parts of any uncompleted buffers.
2403
2404 The do_marking_step() method tries to abort when the time target
2405 has been reached. There are a few other cases when the
2406 do_marking_step() method also aborts:
2407
2408 (1) When the marking phase has been aborted (after a Full GC).
2409
2410 (2) When a global overflow (on the global stack) has been
2411 triggered. Before the task aborts, it will actually sync up with
2412 the other tasks to ensure that all the marking data structures
2413 (local queues, stacks, fingers etc.) are re-initialized so that
2414 when do_marking_step() completes, the marking phase can
2415 immediately restart.
2416
2417 (3) When enough completed SATB buffers are available. The
2418 do_marking_step() method only tries to drain SATB buffers right
2419 at the beginning. So, if enough buffers are available, the
2420 marking step aborts and the SATB buffers are processed at
2421 the beginning of the next invocation.
2422
2423 (4) To yield. when we have to yield then we abort and yield
2424 right at the end of do_marking_step(). This saves us from a lot
2425 of hassle as, by yielding we might allow a Full GC. If this
2426 happens then objects will be compacted underneath our feet, the
2427 heap might shrink, etc. We save checking for this by just
2428 aborting and doing the yield right at the end.
2429
2430 From the above it follows that the do_marking_step() method should
2431 be called in a loop (or, otherwise, regularly) until it completes.
2432
2433 If a marking step completes without its has_aborted() flag being
2434 true, it means it has completed the current marking phase (and
2435 also all other marking tasks have done so and have all synced up).
2436
2437 A method called regular_clock_call() is invoked "regularly" (in
2438 sub ms intervals) throughout marking. It is this clock method that
2439 checks all the abort conditions which were mentioned above and
2440 decides when the task should abort. A work-based scheme is used to
2441 trigger this clock method: when the number of object words the
2442 marking phase has scanned or the number of references the marking
2443 phase has visited reach a given limit. Additional invocations to
2444 the method clock have been planted in a few other strategic places
2445 too. The initial reason for the clock method was to avoid calling
2446 vtime too regularly, as it is quite expensive. So, once it was in
2447 place, it was natural to piggy-back all the other conditions on it
2448 too and not constantly check them throughout the code.
2449
2450 If do_termination is true then do_marking_step will enter its
2451 termination protocol.
2452
2453 The value of is_serial must be true when do_marking_step is being
2454 called serially (i.e. by the VMThread) and do_marking_step should
2455 skip any synchronization in the termination and overflow code.
2456 Examples include the serial remark code and the serial reference
2457 processing closures.
2458
2459 The value of is_serial must be false when do_marking_step is
2460 being called by any of the worker threads in a work gang.
2461 Examples include the concurrent marking code (CMMarkingTask),
2462 the MT remark code, and the MT reference processing closures.
2463
2464 *****************************************************************************/
2465
2466 void G1CMTask::do_marking_step(double time_target_ms,
2467 bool do_termination,
2468 bool is_serial) {
2469 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
2470
2471 _start_time_ms = os::elapsedVTime() * 1000.0;
2472
2473 // If do_stealing is true then do_marking_step will attempt to
2474 // steal work from the other G1CMTasks. It only makes sense to
2475 // enable stealing when the termination protocol is enabled
2476 // and do_marking_step() is not being called serially.
2477 bool do_stealing = do_termination && !is_serial;
2478
2479 double diff_prediction_ms = _g1h->g1_policy()->predictor().get_new_prediction(&_marking_step_diffs_ms);
2480 _time_target_ms = time_target_ms - diff_prediction_ms;
2481
2482 // set up the variables that are used in the work-based scheme to
2483 // call the regular clock method
2484 _words_scanned = 0;
2485 _refs_reached = 0;
2486 recalculate_limits();
2487
2488 // clear all flags
2489 clear_has_aborted();
2490 _has_timed_out = false;
2491 _draining_satb_buffers = false;
2492
2493 ++_calls;
2494
2495 // Set up the bitmap and oop closures. Anything that uses them is
2496 // eventually called from this method, so it is OK to allocate these
2497 // statically.
2498 G1CMBitMapClosure bitmap_closure(this, _cm);
2499 G1CMOopClosure cm_oop_closure(_g1h, this);
2500 set_cm_oop_closure(&cm_oop_closure);
2501
2502 if (_cm->has_overflown()) {
2503 // This can happen if the mark stack overflows during a GC pause
2504 // and this task, after a yield point, restarts. We have to abort
2505 // as we need to get into the overflow protocol which happens
2506 // right at the end of this task.
2507 set_has_aborted();
2508 }
2509
2510 // First drain any available SATB buffers. After this, we will not
2511 // look at SATB buffers before the next invocation of this method.
2512 // If enough completed SATB buffers are queued up, the regular clock
2513 // will abort this task so that it restarts.
2514 drain_satb_buffers();
2515 // ...then partially drain the local queue and the global stack
2516 drain_local_queue(true);
2517 drain_global_stack(true);
2518
2519 do {
2520 if (!has_aborted() && _curr_region != NULL) {
2521 // This means that we're already holding on to a region.
2522 assert(_finger != NULL, "if region is not NULL, then the finger "
2523 "should not be NULL either");
2524
2525 // We might have restarted this task after an evacuation pause
2526 // which might have evacuated the region we're holding on to
2527 // underneath our feet. Let's read its limit again to make sure
2528 // that we do not iterate over a region of the heap that
2529 // contains garbage (update_region_limit() will also move
2530 // _finger to the start of the region if it is found empty).
2531 update_region_limit();
2532 // We will start from _finger not from the start of the region,
2533 // as we might be restarting this task after aborting half-way
2534 // through scanning this region. In this case, _finger points to
2535 // the address where we last found a marked object. If this is a
2536 // fresh region, _finger points to start().
2537 MemRegion mr = MemRegion(_finger, _region_limit);
2538
2539 assert(!_curr_region->is_humongous() || mr.start() == _curr_region->bottom(),
2540 "humongous regions should go around loop once only");
2541
2542 // Some special cases:
2543 // If the memory region is empty, we can just give up the region.
2544 // If the current region is humongous then we only need to check
2545 // the bitmap for the bit associated with the start of the object,
2546 // scan the object if it's live, and give up the region.
2547 // Otherwise, let's iterate over the bitmap of the part of the region
2548 // that is left.
2549 // If the iteration is successful, give up the region.
2550 if (mr.is_empty()) {
2551 giveup_current_region();
2552 regular_clock_call();
2553 } else if (_curr_region->is_humongous() && mr.start() == _curr_region->bottom()) {
2554 if (_next_mark_bitmap->is_marked(mr.start())) {
2555 // The object is marked - apply the closure
2556 bitmap_closure.do_addr(mr.start());
2557 }
2558 // Even if this task aborted while scanning the humongous object
2559 // we can (and should) give up the current region.
2560 giveup_current_region();
2561 regular_clock_call();
2562 } else if (_next_mark_bitmap->iterate(&bitmap_closure, mr)) {
2563 giveup_current_region();
2564 regular_clock_call();
2565 } else {
2566 assert(has_aborted(), "currently the only way to do so");
2567 // The only way to abort the bitmap iteration is to return
2568 // false from the do_bit() method. However, inside the
2569 // do_bit() method we move the _finger to point to the
2570 // object currently being looked at. So, if we bail out, we
2571 // have definitely set _finger to something non-null.
2572 assert(_finger != NULL, "invariant");
2573
2574 // Region iteration was actually aborted. So now _finger
2575 // points to the address of the object we last scanned. If we
2576 // leave it there, when we restart this task, we will rescan
2577 // the object. It is easy to avoid this. We move the finger by
2578 // enough to point to the next possible object header.
2579 assert(_finger < _region_limit, "invariant");
2580 HeapWord* const new_finger = _finger + ((oop)_finger)->size();
2581 // Check if bitmap iteration was aborted while scanning the last object
2582 if (new_finger >= _region_limit) {
2583 giveup_current_region();
2584 } else {
2585 move_finger_to(new_finger);
2586 }
2587 }
2588 }
2589 // At this point we have either completed iterating over the
2590 // region we were holding on to, or we have aborted.
2591
2592 // We then partially drain the local queue and the global stack.
2593 // (Do we really need this?)
2594 drain_local_queue(true);
2595 drain_global_stack(true);
2596
2597 // Read the note on the claim_region() method on why it might
2598 // return NULL with potentially more regions available for
2599 // claiming and why we have to check out_of_regions() to determine
2600 // whether we're done or not.
2601 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
2602 // We are going to try to claim a new region. We should have
2603 // given up on the previous one.
2604 // Separated the asserts so that we know which one fires.
2605 assert(_curr_region == NULL, "invariant");
2606 assert(_finger == NULL, "invariant");
2607 assert(_region_limit == NULL, "invariant");
2608 HeapRegion* claimed_region = _cm->claim_region(_worker_id);
2609 if (claimed_region != NULL) {
2610 // Yes, we managed to claim one
2611 setup_for_region(claimed_region);
2612 assert(_curr_region == claimed_region, "invariant");
2613 }
2614 // It is important to call the regular clock here. It might take
2615 // a while to claim a region if, for example, we hit a large
2616 // block of empty regions. So we need to call the regular clock
2617 // method once round the loop to make sure it's called
2618 // frequently enough.
2619 regular_clock_call();
2620 }
2621
2622 if (!has_aborted() && _curr_region == NULL) {
2623 assert(_cm->out_of_regions(),
2624 "at this point we should be out of regions");
2625 }
2626 } while ( _curr_region != NULL && !has_aborted());
2627
2628 if (!has_aborted()) {
2629 // We cannot check whether the global stack is empty, since other
2630 // tasks might be pushing objects to it concurrently.
2631 assert(_cm->out_of_regions(),
2632 "at this point we should be out of regions");
2633 // Try to reduce the number of available SATB buffers so that
2634 // remark has less work to do.
2635 drain_satb_buffers();
2636 }
2637
2638 // Since we've done everything else, we can now totally drain the
2639 // local queue and global stack.
2640 drain_local_queue(false);
2641 drain_global_stack(false);
2642
2643 // Attempt at work stealing from other task's queues.
2644 if (do_stealing && !has_aborted()) {
2645 // We have not aborted. This means that we have finished all that
2646 // we could. Let's try to do some stealing...
2647
2648 // We cannot check whether the global stack is empty, since other
2649 // tasks might be pushing objects to it concurrently.
2650 assert(_cm->out_of_regions() && _task_queue->size() == 0,
2651 "only way to reach here");
2652 while (!has_aborted()) {
2653 G1TaskQueueEntry entry;
2654 if (_cm->try_stealing(_worker_id, &_hash_seed, entry)) {
2655 scan_task_entry(entry);
2656
2657 // And since we're towards the end, let's totally drain the
2658 // local queue and global stack.
2659 drain_local_queue(false);
2660 drain_global_stack(false);
2661 } else {
2662 break;
2663 }
2664 }
2665 }
2666
2667 // We still haven't aborted. Now, let's try to get into the
2668 // termination protocol.
2669 if (do_termination && !has_aborted()) {
2670 // We cannot check whether the global stack is empty, since other
2671 // tasks might be concurrently pushing objects on it.
2672 // Separated the asserts so that we know which one fires.
2673 assert(_cm->out_of_regions(), "only way to reach here");
2674 assert(_task_queue->size() == 0, "only way to reach here");
2675 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
2676
2677 // The G1CMTask class also extends the TerminatorTerminator class,
2678 // hence its should_exit_termination() method will also decide
2679 // whether to exit the termination protocol or not.
2680 bool finished = (is_serial ||
2681 _cm->terminator()->offer_termination(this));
2682 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
2683 _termination_time_ms +=
2684 termination_end_time_ms - _termination_start_time_ms;
2685
2686 if (finished) {
2687 // We're all done.
2688
2689 // We can now guarantee that the global stack is empty, since
2690 // all other tasks have finished. We separated the guarantees so
2691 // that, if a condition is false, we can immediately find out
2692 // which one.
2693 guarantee(_cm->out_of_regions(), "only way to reach here");
2694 guarantee(_cm->mark_stack_empty(), "only way to reach here");
2695 guarantee(_task_queue->size() == 0, "only way to reach here");
2696 guarantee(!_cm->has_overflown(), "only way to reach here");
2697 } else {
2698 // Apparently there's more work to do. Let's abort this task. It
2699 // will restart it and we can hopefully find more things to do.
2700 set_has_aborted();
2701 }
2702 }
2703
2704 // Mainly for debugging purposes to make sure that a pointer to the
2705 // closure which was statically allocated in this frame doesn't
2706 // escape it by accident.
2707 set_cm_oop_closure(NULL);
2708 double end_time_ms = os::elapsedVTime() * 1000.0;
2709 double elapsed_time_ms = end_time_ms - _start_time_ms;
2710 // Update the step history.
2711 _step_times_ms.add(elapsed_time_ms);
2712
2713 if (has_aborted()) {
2714 // The task was aborted for some reason.
2715 if (_has_timed_out) {
2716 double diff_ms = elapsed_time_ms - _time_target_ms;
2717 // Keep statistics of how well we did with respect to hitting
2718 // our target only if we actually timed out (if we aborted for
2719 // other reasons, then the results might get skewed).
2720 _marking_step_diffs_ms.add(diff_ms);
2721 }
2722
2723 if (_cm->has_overflown()) {
2724 // This is the interesting one. We aborted because a global
2725 // overflow was raised. This means we have to restart the
2726 // marking phase and start iterating over regions. However, in
2727 // order to do this we have to make sure that all tasks stop
2728 // what they are doing and re-initialize in a safe manner. We
2729 // will achieve this with the use of two barrier sync points.
2730
2731 if (!is_serial) {
2732 // We only need to enter the sync barrier if being called
2733 // from a parallel context
2734 _cm->enter_first_sync_barrier(_worker_id);
2735
2736 // When we exit this sync barrier we know that all tasks have
2737 // stopped doing marking work. So, it's now safe to
2738 // re-initialize our data structures.
2739 }
2740
2741 clear_region_fields();
2742 flush_mark_stats_cache();
2743
2744 if (!is_serial) {
2745 // If we're executing the concurrent phase of marking, reset the marking
2746 // state; otherwise the marking state is reset after reference processing,
2747 // during the remark pause.
2748 // If we reset here as a result of an overflow during the remark we will
2749 // see assertion failures from any subsequent set_concurrency_and_phase()
2750 // calls.
2751 if (_cm->concurrent() && _worker_id == 0) {
2752 // Worker 0 is responsible for clearing the global data structures because
2753 // of an overflow. During STW we should not clear the overflow flag (in
2754 // G1ConcurrentMark::reset_marking_state()) since we rely on it being true when we exit
2755 // method to abort the pause and restart concurrent marking.
2756 _cm->reset_marking_for_restart();
2757
2758 log_info(gc, marking)("Concurrent Mark reset for overflow");
2759 }
2760
2761 // ...and enter the second barrier.
2762 _cm->enter_second_sync_barrier(_worker_id);
2763 }
2764 // At this point, if we're during the concurrent phase of
2765 // marking, everything has been re-initialized and we're
2766 // ready to restart.
2767 }
2768 }
2769 }
2770
2771 G1CMTask::G1CMTask(uint worker_id,
2772 G1ConcurrentMark* cm,
2773 G1CMTaskQueue* task_queue,
2774 G1RegionMarkStats* mark_stats,
2775 uint max_regions) :
2776 _objArray_processor(this),
2777 _worker_id(worker_id),
2778 _g1h(G1CollectedHeap::heap()),
2779 _cm(cm),
2780 _next_mark_bitmap(NULL),
2781 _task_queue(task_queue),
2782 _mark_stats_cache(mark_stats, max_regions, RegionMarkStatsCacheSize),
2783 _calls(0),
2784 _time_target_ms(0.0),
2785 _start_time_ms(0.0),
2786 _cm_oop_closure(NULL),
2787 _curr_region(NULL),
2788 _finger(NULL),
2789 _region_limit(NULL),
2790 _words_scanned(0),
2791 _words_scanned_limit(0),
2792 _real_words_scanned_limit(0),
2793 _refs_reached(0),
2794 _refs_reached_limit(0),
2795 _real_refs_reached_limit(0),
2796 _hash_seed(17),
2797 _has_aborted(false),
2798 _has_timed_out(false),
2799 _draining_satb_buffers(false),
2800 _step_times_ms(),
2801 _elapsed_time_ms(0.0),
2802 _termination_time_ms(0.0),
2803 _termination_start_time_ms(0.0),
2804 _marking_step_diffs_ms()
2805 {
2806 guarantee(task_queue != NULL, "invariant");
2807
2808 _marking_step_diffs_ms.add(0.5);
2809 }
2810
2811 // These are formatting macros that are used below to ensure
2812 // consistent formatting. The *_H_* versions are used to format the
2813 // header for a particular value and they should be kept consistent
2814 // with the corresponding macro. Also note that most of the macros add
2815 // the necessary white space (as a prefix) which makes them a bit
2816 // easier to compose.
2817
2818 // All the output lines are prefixed with this string to be able to
2819 // identify them easily in a large log file.
2820 #define G1PPRL_LINE_PREFIX "###"
2821
2822 #define G1PPRL_ADDR_BASE_FORMAT " " PTR_FORMAT "-" PTR_FORMAT
2823 #ifdef _LP64
2824 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
2825 #else // _LP64
2826 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
2827 #endif // _LP64
2828
2829 // For per-region info
2830 #define G1PPRL_TYPE_FORMAT " %-4s"
2831 #define G1PPRL_TYPE_H_FORMAT " %4s"
2832 #define G1PPRL_STATE_FORMAT " %-5s"
2833 #define G1PPRL_STATE_H_FORMAT " %5s"
2834 #define G1PPRL_BYTE_FORMAT " " SIZE_FORMAT_W(9)
2835 #define G1PPRL_BYTE_H_FORMAT " %9s"
2836 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
2837 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
2838
2839 // For summary info
2840 #define G1PPRL_SUM_ADDR_FORMAT(tag) " " tag ":" G1PPRL_ADDR_BASE_FORMAT
2841 #define G1PPRL_SUM_BYTE_FORMAT(tag) " " tag ": " SIZE_FORMAT
2842 #define G1PPRL_SUM_MB_FORMAT(tag) " " tag ": %1.2f MB"
2843 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag) " / %1.2f %%"
2844
2845 G1PrintRegionLivenessInfoClosure::G1PrintRegionLivenessInfoClosure(const char* phase_name) :
2846 _total_used_bytes(0), _total_capacity_bytes(0),
2847 _total_prev_live_bytes(0), _total_next_live_bytes(0),
2848 _total_remset_bytes(0), _total_strong_code_roots_bytes(0)
2849 {
2850 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2851 MemRegion g1_reserved = g1h->g1_reserved();
2852 double now = os::elapsedTime();
2853
2854 // Print the header of the output.
2855 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
2856 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" HEAP"
2857 G1PPRL_SUM_ADDR_FORMAT("reserved")
2858 G1PPRL_SUM_BYTE_FORMAT("region-size"),
2859 p2i(g1_reserved.start()), p2i(g1_reserved.end()),
2860 HeapRegion::GrainBytes);
2861 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
2862 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
2863 G1PPRL_TYPE_H_FORMAT
2864 G1PPRL_ADDR_BASE_H_FORMAT
2865 G1PPRL_BYTE_H_FORMAT
2866 G1PPRL_BYTE_H_FORMAT
2867 G1PPRL_BYTE_H_FORMAT
2868 G1PPRL_DOUBLE_H_FORMAT
2869 G1PPRL_BYTE_H_FORMAT
2870 G1PPRL_STATE_H_FORMAT
2871 G1PPRL_BYTE_H_FORMAT,
2872 "type", "address-range",
2873 "used", "prev-live", "next-live", "gc-eff",
2874 "remset", "state", "code-roots");
2875 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
2876 G1PPRL_TYPE_H_FORMAT
2877 G1PPRL_ADDR_BASE_H_FORMAT
2878 G1PPRL_BYTE_H_FORMAT
2879 G1PPRL_BYTE_H_FORMAT
2880 G1PPRL_BYTE_H_FORMAT
2881 G1PPRL_DOUBLE_H_FORMAT
2882 G1PPRL_BYTE_H_FORMAT
2883 G1PPRL_STATE_H_FORMAT
2884 G1PPRL_BYTE_H_FORMAT,
2885 "", "",
2886 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)",
2887 "(bytes)", "", "(bytes)");
2888 }
2889
2890 bool G1PrintRegionLivenessInfoClosure::do_heap_region(HeapRegion* r) {
2891 const char* type = r->get_type_str();
2892 HeapWord* bottom = r->bottom();
2893 HeapWord* end = r->end();
2894 size_t capacity_bytes = r->capacity();
2895 size_t used_bytes = r->used();
2896 size_t prev_live_bytes = r->live_bytes();
2897 size_t next_live_bytes = r->next_live_bytes();
2898 double gc_eff = r->gc_efficiency();
2899 size_t remset_bytes = r->rem_set()->mem_size();
2900 size_t strong_code_roots_bytes = r->rem_set()->strong_code_roots_mem_size();
2901 const char* remset_type = r->rem_set()->get_short_state_str();
2902
2903 _total_used_bytes += used_bytes;
2904 _total_capacity_bytes += capacity_bytes;
2905 _total_prev_live_bytes += prev_live_bytes;
2906 _total_next_live_bytes += next_live_bytes;
2907 _total_remset_bytes += remset_bytes;
2908 _total_strong_code_roots_bytes += strong_code_roots_bytes;
2909
2910 // Print a line for this particular region.
2911 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
2912 G1PPRL_TYPE_FORMAT
2913 G1PPRL_ADDR_BASE_FORMAT
2914 G1PPRL_BYTE_FORMAT
2915 G1PPRL_BYTE_FORMAT
2916 G1PPRL_BYTE_FORMAT
2917 G1PPRL_DOUBLE_FORMAT
2918 G1PPRL_BYTE_FORMAT
2919 G1PPRL_STATE_FORMAT
2920 G1PPRL_BYTE_FORMAT,
2921 type, p2i(bottom), p2i(end),
2922 used_bytes, prev_live_bytes, next_live_bytes, gc_eff,
2923 remset_bytes, remset_type, strong_code_roots_bytes);
2924
2925 return false;
2926 }
2927
2928 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
2929 // add static memory usages to remembered set sizes
2930 _total_remset_bytes += HeapRegionRemSet::fl_mem_size() + HeapRegionRemSet::static_mem_size();
2931 // Print the footer of the output.
2932 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
2933 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
2934 " SUMMARY"
2935 G1PPRL_SUM_MB_FORMAT("capacity")
2936 G1PPRL_SUM_MB_PERC_FORMAT("used")
2937 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
2938 G1PPRL_SUM_MB_PERC_FORMAT("next-live")
2939 G1PPRL_SUM_MB_FORMAT("remset")
2940 G1PPRL_SUM_MB_FORMAT("code-roots"),
2941 bytes_to_mb(_total_capacity_bytes),
2942 bytes_to_mb(_total_used_bytes),
2943 percent_of(_total_used_bytes, _total_capacity_bytes),
2944 bytes_to_mb(_total_prev_live_bytes),
2945 percent_of(_total_prev_live_bytes, _total_capacity_bytes),
2946 bytes_to_mb(_total_next_live_bytes),
2947 percent_of(_total_next_live_bytes, _total_capacity_bytes),
2948 bytes_to_mb(_total_remset_bytes),
2949 bytes_to_mb(_total_strong_code_roots_bytes));
2950 }