1 /*
2 * Copyright (c) 2001, 2016, 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/g1CollectorPolicy.hpp"
32 #include "gc/g1/g1CollectorState.hpp"
33 #include "gc/g1/g1ConcurrentMark.inline.hpp"
34 #include "gc/g1/g1HeapVerifier.hpp"
35 #include "gc/g1/g1OopClosures.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/g1/suspendibleThreadSet.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/taskqueue.inline.hpp"
49 #include "gc/shared/vmGCOperations.hpp"
50 #include "logging/log.hpp"
51 #include "memory/allocation.hpp"
52 #include "memory/resourceArea.hpp"
53 #include "oops/oop.inline.hpp"
54 #include "runtime/atomic.inline.hpp"
55 #include "runtime/handles.inline.hpp"
56 #include "runtime/java.hpp"
57 #include "runtime/prefetch.inline.hpp"
58 #include "services/memTracker.hpp"
59
60 // Concurrent marking bit map wrapper
61
62 G1CMBitMapRO::G1CMBitMapRO(int shifter) :
63 _bm(),
64 _shifter(shifter) {
65 _bmStartWord = 0;
66 _bmWordSize = 0;
67 }
68
69 HeapWord* G1CMBitMapRO::getNextMarkedWordAddress(const HeapWord* addr,
70 const HeapWord* limit) const {
71 // First we must round addr *up* to a possible object boundary.
72 addr = (HeapWord*)align_size_up((intptr_t)addr,
73 HeapWordSize << _shifter);
74 size_t addrOffset = heapWordToOffset(addr);
75 assert(limit != NULL, "limit must not be NULL");
76 size_t limitOffset = heapWordToOffset(limit);
77 size_t nextOffset = _bm.get_next_one_offset(addrOffset, limitOffset);
78 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
79 assert(nextAddr >= addr, "get_next_one postcondition");
80 assert(nextAddr == limit || isMarked(nextAddr),
81 "get_next_one postcondition");
82 return nextAddr;
83 }
84
85 #ifndef PRODUCT
86 bool G1CMBitMapRO::covers(MemRegion heap_rs) const {
87 // assert(_bm.map() == _virtual_space.low(), "map inconsistency");
88 assert(((size_t)_bm.size() * ((size_t)1 << _shifter)) == _bmWordSize,
89 "size inconsistency");
90 return _bmStartWord == (HeapWord*)(heap_rs.start()) &&
91 _bmWordSize == heap_rs.word_size();
92 }
93 #endif
94
95 void G1CMBitMapRO::print_on_error(outputStream* st, const char* prefix) const {
96 _bm.print_on_error(st, prefix);
97 }
98
99 size_t G1CMBitMap::compute_size(size_t heap_size) {
100 return ReservedSpace::allocation_align_size_up(heap_size / mark_distance());
101 }
102
103 size_t G1CMBitMap::mark_distance() {
104 return MinObjAlignmentInBytes * BitsPerByte;
105 }
106
107 void G1CMBitMap::initialize(MemRegion heap, G1RegionToSpaceMapper* storage) {
108 _bmStartWord = heap.start();
109 _bmWordSize = heap.word_size();
110
111 _bm.set_map((BitMap::bm_word_t*) storage->reserved().start());
112 _bm.set_size(_bmWordSize >> _shifter);
113
114 storage->set_mapping_changed_listener(&_listener);
115 }
116
117 void G1CMBitMapMappingChangedListener::on_commit(uint start_region, size_t num_regions, bool zero_filled) {
118 if (zero_filled) {
119 return;
120 }
121 // We need to clear the bitmap on commit, removing any existing information.
122 MemRegion mr(G1CollectedHeap::heap()->bottom_addr_for_region(start_region), num_regions * HeapRegion::GrainWords);
123 _bm->clearRange(mr);
124 }
125
126 // Closure used for clearing the given mark bitmap.
127 class ClearBitmapHRClosure : public HeapRegionClosure {
128 private:
129 G1ConcurrentMark* _cm;
130 G1CMBitMap* _bitmap;
131 bool _may_yield; // The closure may yield during iteration. If yielded, abort the iteration.
132 public:
133 ClearBitmapHRClosure(G1ConcurrentMark* cm, G1CMBitMap* bitmap, bool may_yield) : HeapRegionClosure(), _cm(cm), _bitmap(bitmap), _may_yield(may_yield) {
134 assert(!may_yield || cm != NULL, "CM must be non-NULL if this closure is expected to yield.");
135 }
136
137 virtual bool doHeapRegion(HeapRegion* r) {
138 size_t const chunk_size_in_words = M / HeapWordSize;
139
140 HeapWord* cur = r->bottom();
141 HeapWord* const end = r->end();
142
143 while (cur < end) {
144 MemRegion mr(cur, MIN2(cur + chunk_size_in_words, end));
145 _bitmap->clearRange(mr);
146
147 cur += chunk_size_in_words;
148
149 // Abort iteration if after yielding the marking has been aborted.
150 if (_may_yield && _cm->do_yield_check() && _cm->has_aborted()) {
151 return true;
152 }
153 // Repeat the asserts from before the start of the closure. We will do them
154 // as asserts here to minimize their overhead on the product. However, we
155 // will have them as guarantees at the beginning / end of the bitmap
156 // clearing to get some checking in the product.
157 assert(!_may_yield || _cm->cmThread()->during_cycle(), "invariant");
158 assert(!_may_yield || !G1CollectedHeap::heap()->collector_state()->mark_in_progress(), "invariant");
159 }
160
161 return false;
162 }
163 };
164
165 class ParClearNextMarkBitmapTask : public AbstractGangTask {
166 ClearBitmapHRClosure* _cl;
167 HeapRegionClaimer _hrclaimer;
168 bool _suspendible; // If the task is suspendible, workers must join the STS.
169
170 public:
171 ParClearNextMarkBitmapTask(ClearBitmapHRClosure *cl, uint n_workers, bool suspendible) :
172 _cl(cl), _suspendible(suspendible), AbstractGangTask("Parallel Clear Bitmap Task"), _hrclaimer(n_workers) {}
173
174 void work(uint worker_id) {
175 SuspendibleThreadSetJoiner sts_join(_suspendible);
176 G1CollectedHeap::heap()->heap_region_par_iterate(_cl, worker_id, &_hrclaimer, true);
177 }
178 };
179
180 void G1CMBitMap::clearAll() {
181 G1CollectedHeap* g1h = G1CollectedHeap::heap();
182 ClearBitmapHRClosure cl(NULL, this, false /* may_yield */);
183 uint n_workers = g1h->workers()->active_workers();
184 ParClearNextMarkBitmapTask task(&cl, n_workers, false);
185 g1h->workers()->run_task(&task);
186 guarantee(cl.complete(), "Must have completed iteration.");
187 return;
188 }
189
190 void G1CMBitMap::clearRange(MemRegion mr) {
191 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
192 assert(!mr.is_empty(), "unexpected empty region");
193 // convert address range into offset range
194 _bm.at_put_range(heapWordToOffset(mr.start()),
195 heapWordToOffset(mr.end()), false);
196 }
197
198 G1CMMarkStack::G1CMMarkStack(G1ConcurrentMark* cm) :
199 _base(NULL), _cm(cm)
200 {}
201
202 bool G1CMMarkStack::allocate(size_t capacity) {
203 // allocate a stack of the requisite depth
204 ReservedSpace rs(ReservedSpace::allocation_align_size_up(capacity * sizeof(oop)));
205 if (!rs.is_reserved()) {
206 warning("ConcurrentMark MarkStack allocation failure");
207 return false;
208 }
209 MemTracker::record_virtual_memory_type((address)rs.base(), mtGC);
210 if (!_virtual_space.initialize(rs, rs.size())) {
211 warning("ConcurrentMark MarkStack backing store failure");
212 // Release the virtual memory reserved for the marking stack
213 rs.release();
214 return false;
215 }
216 assert(_virtual_space.committed_size() == rs.size(),
217 "Didn't reserve backing store for all of G1ConcurrentMark stack?");
218 _base = (oop*) _virtual_space.low();
219 setEmpty();
220 _capacity = (jint) capacity;
221 _saved_index = -1;
222 _should_expand = false;
223 return true;
224 }
225
226 void G1CMMarkStack::expand() {
227 // Called, during remark, if we've overflown the marking stack during marking.
228 assert(isEmpty(), "stack should been emptied while handling overflow");
229 assert(_capacity <= (jint) MarkStackSizeMax, "stack bigger than permitted");
230 // Clear expansion flag
231 _should_expand = false;
232 if (_capacity == (jint) MarkStackSizeMax) {
233 log_trace(gc)("(benign) Can't expand marking stack capacity, at max size limit");
234 return;
235 }
236 // Double capacity if possible
237 jint new_capacity = MIN2(_capacity*2, (jint) MarkStackSizeMax);
238 // Do not give up existing stack until we have managed to
239 // get the double capacity that we desired.
240 ReservedSpace rs(ReservedSpace::allocation_align_size_up(new_capacity *
241 sizeof(oop)));
242 if (rs.is_reserved()) {
243 // Release the backing store associated with old stack
244 _virtual_space.release();
245 // Reinitialize virtual space for new stack
246 if (!_virtual_space.initialize(rs, rs.size())) {
247 fatal("Not enough swap for expanded marking stack capacity");
248 }
249 _base = (oop*)(_virtual_space.low());
250 _index = 0;
251 _capacity = new_capacity;
252 } else {
253 // Failed to double capacity, continue;
254 log_trace(gc)("(benign) Failed to expand marking stack capacity from " SIZE_FORMAT "K to " SIZE_FORMAT "K",
255 _capacity / K, new_capacity / K);
256 }
257 }
258
259 void G1CMMarkStack::set_should_expand() {
260 // If we're resetting the marking state because of an
261 // marking stack overflow, record that we should, if
262 // possible, expand the stack.
263 _should_expand = _cm->has_overflown();
264 }
265
266 G1CMMarkStack::~G1CMMarkStack() {
267 if (_base != NULL) {
268 _base = NULL;
269 _virtual_space.release();
270 }
271 }
272
273 void G1CMMarkStack::par_push_arr(oop* ptr_arr, int n) {
274 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
275 jint start = _index;
276 jint next_index = start + n;
277 if (next_index > _capacity) {
278 _overflow = true;
279 return;
280 }
281 // Otherwise.
282 _index = next_index;
283 for (int i = 0; i < n; i++) {
284 int ind = start + i;
285 assert(ind < _capacity, "By overflow test above.");
286 _base[ind] = ptr_arr[i];
287 }
288 }
289
290 bool G1CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) {
291 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
292 jint index = _index;
293 if (index == 0) {
294 *n = 0;
295 return false;
296 } else {
297 int k = MIN2(max, index);
298 jint new_ind = index - k;
299 for (int j = 0; j < k; j++) {
300 ptr_arr[j] = _base[new_ind + j];
301 }
302 _index = new_ind;
303 *n = k;
304 return true;
305 }
306 }
307
308 void G1CMMarkStack::note_start_of_gc() {
309 assert(_saved_index == -1,
310 "note_start_of_gc()/end_of_gc() bracketed incorrectly");
311 _saved_index = _index;
312 }
313
314 void G1CMMarkStack::note_end_of_gc() {
315 // This is intentionally a guarantee, instead of an assert. If we
316 // accidentally add something to the mark stack during GC, it
317 // will be a correctness issue so it's better if we crash. we'll
318 // only check this once per GC anyway, so it won't be a performance
319 // issue in any way.
320 guarantee(_saved_index == _index,
321 "saved index: %d index: %d", _saved_index, _index);
322 _saved_index = -1;
323 }
324
325 G1CMRootRegions::G1CMRootRegions() :
326 _young_list(NULL), _cm(NULL), _scan_in_progress(false),
327 _should_abort(false), _next_survivor(NULL) { }
328
329 void G1CMRootRegions::init(G1CollectedHeap* g1h, G1ConcurrentMark* cm) {
330 _young_list = g1h->young_list();
331 _cm = cm;
332 }
333
334 void G1CMRootRegions::prepare_for_scan() {
335 assert(!scan_in_progress(), "pre-condition");
336
337 // Currently, only survivors can be root regions.
338 assert(_next_survivor == NULL, "pre-condition");
339 _next_survivor = _young_list->first_survivor_region();
340 _scan_in_progress = (_next_survivor != NULL);
341 _should_abort = false;
342 }
343
344 HeapRegion* G1CMRootRegions::claim_next() {
345 if (_should_abort) {
346 // If someone has set the should_abort flag, we return NULL to
347 // force the caller to bail out of their loop.
348 return NULL;
349 }
350
351 // Currently, only survivors can be root regions.
352 HeapRegion* res = _next_survivor;
353 if (res != NULL) {
354 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
355 // Read it again in case it changed while we were waiting for the lock.
356 res = _next_survivor;
357 if (res != NULL) {
358 if (res == _young_list->last_survivor_region()) {
359 // We just claimed the last survivor so store NULL to indicate
360 // that we're done.
361 _next_survivor = NULL;
362 } else {
363 _next_survivor = res->get_next_young_region();
364 }
365 } else {
366 // Someone else claimed the last survivor while we were trying
367 // to take the lock so nothing else to do.
368 }
369 }
370 assert(res == NULL || res->is_survivor(), "post-condition");
371
372 return res;
373 }
374
375 void G1CMRootRegions::notify_scan_done() {
376 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
377 _scan_in_progress = false;
378 RootRegionScan_lock->notify_all();
379 }
380
381 void G1CMRootRegions::cancel_scan() {
382 notify_scan_done();
383 }
384
385 void G1CMRootRegions::scan_finished() {
386 assert(scan_in_progress(), "pre-condition");
387
388 // Currently, only survivors can be root regions.
389 if (!_should_abort) {
390 assert(_next_survivor == NULL, "we should have claimed all survivors");
391 }
392 _next_survivor = NULL;
393
394 notify_scan_done();
395 }
396
397 bool G1CMRootRegions::wait_until_scan_finished() {
398 if (!scan_in_progress()) return false;
399
400 {
401 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
402 while (scan_in_progress()) {
403 RootRegionScan_lock->wait(Mutex::_no_safepoint_check_flag);
404 }
405 }
406 return true;
407 }
408
409 uint G1ConcurrentMark::scale_parallel_threads(uint n_par_threads) {
410 return MAX2((n_par_threads + 2) / 4, 1U);
411 }
412
413 G1ConcurrentMark::G1ConcurrentMark(G1CollectedHeap* g1h, G1RegionToSpaceMapper* prev_bitmap_storage, G1RegionToSpaceMapper* next_bitmap_storage) :
414 _g1h(g1h),
415 _markBitMap1(),
416 _markBitMap2(),
417 _parallel_marking_threads(0),
418 _max_parallel_marking_threads(0),
419 _sleep_factor(0.0),
420 _marking_task_overhead(1.0),
421 _cleanup_list("Cleanup List"),
422 _region_bm((BitMap::idx_t)(g1h->max_regions()), false /* in_resource_area*/),
423 _card_bm((g1h->reserved_region().byte_size() + CardTableModRefBS::card_size - 1) >>
424 CardTableModRefBS::card_shift,
425 false /* in_resource_area*/),
426
427 _prevMarkBitMap(&_markBitMap1),
428 _nextMarkBitMap(&_markBitMap2),
429
430 _markStack(this),
431 // _finger set in set_non_marking_state
432
433 _max_worker_id(ParallelGCThreads),
434 // _active_tasks set in set_non_marking_state
435 // _tasks set inside the constructor
436 _task_queues(new G1CMTaskQueueSet((int) _max_worker_id)),
437 _terminator(ParallelTaskTerminator((int) _max_worker_id, _task_queues)),
438
439 _has_overflown(false),
440 _concurrent(false),
441 _has_aborted(false),
442 _restart_for_overflow(false),
443 _concurrent_marking_in_progress(false),
444 _concurrent_phase_status(ConcPhaseNotStarted),
445
446 // _verbose_level set below
447
448 _init_times(),
449 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(),
450 _cleanup_times(),
451 _total_counting_time(0.0),
452 _total_rs_scrub_time(0.0),
453
454 _parallel_workers(NULL),
455
456 _count_card_bitmaps(NULL),
457 _count_marked_bytes(NULL),
458 _completed_initialization(false) {
459
460 _markBitMap1.initialize(g1h->reserved_region(), prev_bitmap_storage);
461 _markBitMap2.initialize(g1h->reserved_region(), next_bitmap_storage);
462
463 // Create & start a ConcurrentMark thread.
464 _cmThread = new ConcurrentMarkThread(this);
465 assert(cmThread() != NULL, "CM Thread should have been created");
466 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm");
467 if (_cmThread->osthread() == NULL) {
468 vm_shutdown_during_initialization("Could not create ConcurrentMarkThread");
469 }
470
471 assert(CGC_lock != NULL, "Where's the CGC_lock?");
472 assert(_markBitMap1.covers(g1h->reserved_region()), "_markBitMap1 inconsistency");
473 assert(_markBitMap2.covers(g1h->reserved_region()), "_markBitMap2 inconsistency");
474
475 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
476 satb_qs.set_buffer_size(G1SATBBufferSize);
477
478 _root_regions.init(_g1h, this);
479
480 if (ConcGCThreads > ParallelGCThreads) {
481 warning("Can't have more ConcGCThreads (%u) "
482 "than ParallelGCThreads (%u).",
483 ConcGCThreads, ParallelGCThreads);
484 return;
485 }
486 if (!FLAG_IS_DEFAULT(ConcGCThreads) && ConcGCThreads > 0) {
487 // Note: ConcGCThreads has precedence over G1MarkingOverheadPercent
488 // if both are set
489 _sleep_factor = 0.0;
490 _marking_task_overhead = 1.0;
491 } else if (G1MarkingOverheadPercent > 0) {
492 // We will calculate the number of parallel marking threads based
493 // on a target overhead with respect to the soft real-time goal
494 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
495 double overall_cm_overhead =
496 (double) MaxGCPauseMillis * marking_overhead /
497 (double) GCPauseIntervalMillis;
498 double cpu_ratio = 1.0 / (double) os::processor_count();
499 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio);
500 double marking_task_overhead =
501 overall_cm_overhead / marking_thread_num *
502 (double) os::processor_count();
503 double sleep_factor =
504 (1.0 - marking_task_overhead) / marking_task_overhead;
505
506 FLAG_SET_ERGO(uint, ConcGCThreads, (uint) marking_thread_num);
507 _sleep_factor = sleep_factor;
508 _marking_task_overhead = marking_task_overhead;
509 } else {
510 // Calculate the number of parallel marking threads by scaling
511 // the number of parallel GC threads.
512 uint marking_thread_num = scale_parallel_threads(ParallelGCThreads);
513 FLAG_SET_ERGO(uint, ConcGCThreads, marking_thread_num);
514 _sleep_factor = 0.0;
515 _marking_task_overhead = 1.0;
516 }
517
518 assert(ConcGCThreads > 0, "Should have been set");
519 _parallel_marking_threads = ConcGCThreads;
520 _max_parallel_marking_threads = _parallel_marking_threads;
521
522 _parallel_workers = new WorkGang("G1 Marker",
523 _max_parallel_marking_threads, false, true);
524 if (_parallel_workers == NULL) {
525 vm_exit_during_initialization("Failed necessary allocation.");
526 } else {
527 _parallel_workers->initialize_workers();
528 }
529
530 if (FLAG_IS_DEFAULT(MarkStackSize)) {
531 size_t mark_stack_size =
532 MIN2(MarkStackSizeMax,
533 MAX2(MarkStackSize, (size_t) (parallel_marking_threads() * TASKQUEUE_SIZE)));
534 // Verify that the calculated value for MarkStackSize is in range.
535 // It would be nice to use the private utility routine from Arguments.
536 if (!(mark_stack_size >= 1 && mark_stack_size <= MarkStackSizeMax)) {
537 warning("Invalid value calculated for MarkStackSize (" SIZE_FORMAT "): "
538 "must be between 1 and " SIZE_FORMAT,
539 mark_stack_size, MarkStackSizeMax);
540 return;
541 }
542 FLAG_SET_ERGO(size_t, MarkStackSize, mark_stack_size);
543 } else {
544 // Verify MarkStackSize is in range.
545 if (FLAG_IS_CMDLINE(MarkStackSize)) {
546 if (FLAG_IS_DEFAULT(MarkStackSizeMax)) {
547 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
548 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT "): "
549 "must be between 1 and " SIZE_FORMAT,
550 MarkStackSize, MarkStackSizeMax);
551 return;
552 }
553 } else if (FLAG_IS_CMDLINE(MarkStackSizeMax)) {
554 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
555 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT ")"
556 " or for MarkStackSizeMax (" SIZE_FORMAT ")",
557 MarkStackSize, MarkStackSizeMax);
558 return;
559 }
560 }
561 }
562 }
563
564 if (!_markStack.allocate(MarkStackSize)) {
565 warning("Failed to allocate CM marking stack");
566 return;
567 }
568
569 _tasks = NEW_C_HEAP_ARRAY(G1CMTask*, _max_worker_id, mtGC);
570 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_worker_id, mtGC);
571
572 _count_card_bitmaps = NEW_C_HEAP_ARRAY(BitMap, _max_worker_id, mtGC);
573 _count_marked_bytes = NEW_C_HEAP_ARRAY(size_t*, _max_worker_id, mtGC);
574
575 BitMap::idx_t card_bm_size = _card_bm.size();
576
577 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
578 _active_tasks = _max_worker_id;
579
580 uint max_regions = _g1h->max_regions();
581 for (uint i = 0; i < _max_worker_id; ++i) {
582 G1CMTaskQueue* task_queue = new G1CMTaskQueue();
583 task_queue->initialize();
584 _task_queues->register_queue(i, task_queue);
585
586 _count_card_bitmaps[i] = BitMap(card_bm_size, false);
587 _count_marked_bytes[i] = NEW_C_HEAP_ARRAY(size_t, max_regions, mtGC);
588
589 _tasks[i] = new G1CMTask(i, this,
590 _count_marked_bytes[i],
591 &_count_card_bitmaps[i],
592 task_queue, _task_queues);
593
594 _accum_task_vtime[i] = 0.0;
595 }
596
597 // Calculate the card number for the bottom of the heap. Used
598 // in biasing indexes into the accounting card bitmaps.
599 _heap_bottom_card_num =
600 intptr_t(uintptr_t(_g1h->reserved_region().start()) >>
601 CardTableModRefBS::card_shift);
602
603 // Clear all the liveness counting data
604 clear_all_count_data();
605
606 // so that the call below can read a sensible value
607 _heap_start = g1h->reserved_region().start();
608 set_non_marking_state();
609 _completed_initialization = true;
610 }
611
612 void G1ConcurrentMark::reset() {
613 // Starting values for these two. This should be called in a STW
614 // phase.
615 MemRegion reserved = _g1h->g1_reserved();
616 _heap_start = reserved.start();
617 _heap_end = reserved.end();
618
619 // Separated the asserts so that we know which one fires.
620 assert(_heap_start != NULL, "heap bounds should look ok");
621 assert(_heap_end != NULL, "heap bounds should look ok");
622 assert(_heap_start < _heap_end, "heap bounds should look ok");
623
624 // Reset all the marking data structures and any necessary flags
625 reset_marking_state();
626
627 // We do reset all of them, since different phases will use
628 // different number of active threads. So, it's easiest to have all
629 // of them ready.
630 for (uint i = 0; i < _max_worker_id; ++i) {
631 _tasks[i]->reset(_nextMarkBitMap);
632 }
633
634 // we need this to make sure that the flag is on during the evac
635 // pause with initial mark piggy-backed
636 set_concurrent_marking_in_progress();
637 }
638
639
640 void G1ConcurrentMark::reset_marking_state(bool clear_overflow) {
641 _markStack.set_should_expand();
642 _markStack.setEmpty(); // Also clears the _markStack overflow flag
643 if (clear_overflow) {
644 clear_has_overflown();
645 } else {
646 assert(has_overflown(), "pre-condition");
647 }
648 _finger = _heap_start;
649
650 for (uint i = 0; i < _max_worker_id; ++i) {
651 G1CMTaskQueue* queue = _task_queues->queue(i);
652 queue->set_empty();
653 }
654 }
655
656 void G1ConcurrentMark::set_concurrency(uint active_tasks) {
657 assert(active_tasks <= _max_worker_id, "we should not have more");
658
659 _active_tasks = active_tasks;
660 // Need to update the three data structures below according to the
661 // number of active threads for this phase.
662 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
663 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
664 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
665 }
666
667 void G1ConcurrentMark::set_concurrency_and_phase(uint active_tasks, bool concurrent) {
668 set_concurrency(active_tasks);
669
670 _concurrent = concurrent;
671 // We propagate this to all tasks, not just the active ones.
672 for (uint i = 0; i < _max_worker_id; ++i)
673 _tasks[i]->set_concurrent(concurrent);
674
675 if (concurrent) {
676 set_concurrent_marking_in_progress();
677 } else {
678 // We currently assume that the concurrent flag has been set to
679 // false before we start remark. At this point we should also be
680 // in a STW phase.
681 assert(!concurrent_marking_in_progress(), "invariant");
682 assert(out_of_regions(),
683 "only way to get here: _finger: " PTR_FORMAT ", _heap_end: " PTR_FORMAT,
684 p2i(_finger), p2i(_heap_end));
685 }
686 }
687
688 void G1ConcurrentMark::set_non_marking_state() {
689 // We set the global marking state to some default values when we're
690 // not doing marking.
691 reset_marking_state();
692 _active_tasks = 0;
693 clear_concurrent_marking_in_progress();
694 }
695
696 G1ConcurrentMark::~G1ConcurrentMark() {
697 // The G1ConcurrentMark instance is never freed.
698 ShouldNotReachHere();
699 }
700
701 void G1ConcurrentMark::clearNextBitmap() {
702 G1CollectedHeap* g1h = G1CollectedHeap::heap();
703
704 // Make sure that the concurrent mark thread looks to still be in
705 // the current cycle.
706 guarantee(cmThread()->during_cycle(), "invariant");
707
708 // We are finishing up the current cycle by clearing the next
709 // marking bitmap and getting it ready for the next cycle. During
710 // this time no other cycle can start. So, let's make sure that this
711 // is the case.
712 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
713
714 ClearBitmapHRClosure cl(this, _nextMarkBitMap, true /* may_yield */);
715 ParClearNextMarkBitmapTask task(&cl, parallel_marking_threads(), true);
716 _parallel_workers->run_task(&task);
717
718 // Clear the liveness counting data. If the marking has been aborted, the abort()
719 // call already did that.
720 if (cl.complete()) {
721 clear_all_count_data();
722 }
723
724 // Repeat the asserts from above.
725 guarantee(cmThread()->during_cycle(), "invariant");
726 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
727 }
728
729 class CheckBitmapClearHRClosure : public HeapRegionClosure {
730 G1CMBitMap* _bitmap;
731 bool _error;
732 public:
733 CheckBitmapClearHRClosure(G1CMBitMap* bitmap) : _bitmap(bitmap) {
734 }
735
736 virtual bool doHeapRegion(HeapRegion* r) {
737 // This closure can be called concurrently to the mutator, so we must make sure
738 // that the result of the getNextMarkedWordAddress() call is compared to the
739 // value passed to it as limit to detect any found bits.
740 // end never changes in G1.
741 HeapWord* end = r->end();
742 return _bitmap->getNextMarkedWordAddress(r->bottom(), end) != end;
743 }
744 };
745
746 bool G1ConcurrentMark::nextMarkBitmapIsClear() {
747 CheckBitmapClearHRClosure cl(_nextMarkBitMap);
748 _g1h->heap_region_iterate(&cl);
749 return cl.complete();
750 }
751
752 class NoteStartOfMarkHRClosure: public HeapRegionClosure {
753 public:
754 bool doHeapRegion(HeapRegion* r) {
755 r->note_start_of_marking();
756 return false;
757 }
758 };
759
760 void G1ConcurrentMark::checkpointRootsInitialPre() {
761 G1CollectedHeap* g1h = G1CollectedHeap::heap();
762 G1CollectorPolicy* g1p = g1h->g1_policy();
763
764 _has_aborted = false;
765
766 // Initialize marking structures. This has to be done in a STW phase.
767 reset();
768
769 // For each region note start of marking.
770 NoteStartOfMarkHRClosure startcl;
771 g1h->heap_region_iterate(&startcl);
772 }
773
774
775 void G1ConcurrentMark::checkpointRootsInitialPost() {
776 G1CollectedHeap* g1h = G1CollectedHeap::heap();
777
778 // Start Concurrent Marking weak-reference discovery.
779 ReferenceProcessor* rp = g1h->ref_processor_cm();
780 // enable ("weak") refs discovery
781 rp->enable_discovery();
782 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
783
784 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
785 // This is the start of the marking cycle, we're expected all
786 // threads to have SATB queues with active set to false.
787 satb_mq_set.set_active_all_threads(true, /* new active value */
788 false /* expected_active */);
789
790 _root_regions.prepare_for_scan();
791
792 // update_g1_committed() will be called at the end of an evac pause
793 // when marking is on. So, it's also called at the end of the
794 // initial-mark pause to update the heap end, if the heap expands
795 // during it. No need to call it here.
796 }
797
798 /*
799 * Notice that in the next two methods, we actually leave the STS
800 * during the barrier sync and join it immediately afterwards. If we
801 * do not do this, the following deadlock can occur: one thread could
802 * be in the barrier sync code, waiting for the other thread to also
803 * sync up, whereas another one could be trying to yield, while also
804 * waiting for the other threads to sync up too.
805 *
806 * Note, however, that this code is also used during remark and in
807 * this case we should not attempt to leave / enter the STS, otherwise
808 * we'll either hit an assert (debug / fastdebug) or deadlock
809 * (product). So we should only leave / enter the STS if we are
810 * operating concurrently.
811 *
812 * Because the thread that does the sync barrier has left the STS, it
813 * is possible to be suspended for a Full GC or an evacuation pause
814 * could occur. This is actually safe, since the entering the sync
815 * barrier is one of the last things do_marking_step() does, and it
816 * doesn't manipulate any data structures afterwards.
817 */
818
819 void G1ConcurrentMark::enter_first_sync_barrier(uint worker_id) {
820 bool barrier_aborted;
821 {
822 SuspendibleThreadSetLeaver sts_leave(concurrent());
823 barrier_aborted = !_first_overflow_barrier_sync.enter();
824 }
825
826 // at this point everyone should have synced up and not be doing any
827 // more work
828
829 if (barrier_aborted) {
830 // If the barrier aborted we ignore the overflow condition and
831 // just abort the whole marking phase as quickly as possible.
832 return;
833 }
834
835 // If we're executing the concurrent phase of marking, reset the marking
836 // state; otherwise the marking state is reset after reference processing,
837 // during the remark pause.
838 // If we reset here as a result of an overflow during the remark we will
839 // see assertion failures from any subsequent set_concurrency_and_phase()
840 // calls.
841 if (concurrent()) {
842 // let the task associated with with worker 0 do this
843 if (worker_id == 0) {
844 // task 0 is responsible for clearing the global data structures
845 // We should be here because of an overflow. During STW we should
846 // not clear the overflow flag since we rely on it being true when
847 // we exit this method to abort the pause and restart concurrent
848 // marking.
849 reset_marking_state(true /* clear_overflow */);
850
851 log_info(gc)("Concurrent Mark reset for overflow");
852 }
853 }
854
855 // after this, each task should reset its own data structures then
856 // then go into the second barrier
857 }
858
859 void G1ConcurrentMark::enter_second_sync_barrier(uint worker_id) {
860 SuspendibleThreadSetLeaver sts_leave(concurrent());
861 _second_overflow_barrier_sync.enter();
862
863 // at this point everything should be re-initialized and ready to go
864 }
865
866 class G1CMConcurrentMarkingTask: public AbstractGangTask {
867 private:
868 G1ConcurrentMark* _cm;
869 ConcurrentMarkThread* _cmt;
870
871 public:
872 void work(uint worker_id) {
873 assert(Thread::current()->is_ConcurrentGC_thread(),
874 "this should only be done by a conc GC thread");
875 ResourceMark rm;
876
877 double start_vtime = os::elapsedVTime();
878
879 {
880 SuspendibleThreadSetJoiner sts_join;
881
882 assert(worker_id < _cm->active_tasks(), "invariant");
883 G1CMTask* the_task = _cm->task(worker_id);
884 the_task->record_start_time();
885 if (!_cm->has_aborted()) {
886 do {
887 double start_vtime_sec = os::elapsedVTime();
888 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
889
890 the_task->do_marking_step(mark_step_duration_ms,
891 true /* do_termination */,
892 false /* is_serial*/);
893
894 double end_vtime_sec = os::elapsedVTime();
895 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec;
896 _cm->clear_has_overflown();
897
898 _cm->do_yield_check(worker_id);
899
900 jlong sleep_time_ms;
901 if (!_cm->has_aborted() && the_task->has_aborted()) {
902 sleep_time_ms =
903 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0);
904 {
905 SuspendibleThreadSetLeaver sts_leave;
906 os::sleep(Thread::current(), sleep_time_ms, false);
907 }
908 }
909 } while (!_cm->has_aborted() && the_task->has_aborted());
910 }
911 the_task->record_end_time();
912 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant");
913 }
914
915 double end_vtime = os::elapsedVTime();
916 _cm->update_accum_task_vtime(worker_id, end_vtime - start_vtime);
917 }
918
919 G1CMConcurrentMarkingTask(G1ConcurrentMark* cm,
920 ConcurrentMarkThread* cmt) :
921 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { }
922
923 ~G1CMConcurrentMarkingTask() { }
924 };
925
926 // Calculates the number of active workers for a concurrent
927 // phase.
928 uint G1ConcurrentMark::calc_parallel_marking_threads() {
929 uint n_conc_workers = 0;
930 if (!UseDynamicNumberOfGCThreads ||
931 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
932 !ForceDynamicNumberOfGCThreads)) {
933 n_conc_workers = max_parallel_marking_threads();
934 } else {
935 n_conc_workers =
936 AdaptiveSizePolicy::calc_default_active_workers(
937 max_parallel_marking_threads(),
938 1, /* Minimum workers */
939 parallel_marking_threads(),
940 Threads::number_of_non_daemon_threads());
941 // Don't scale down "n_conc_workers" by scale_parallel_threads() because
942 // that scaling has already gone into "_max_parallel_marking_threads".
943 }
944 assert(n_conc_workers > 0, "Always need at least 1");
945 return n_conc_workers;
946 }
947
948 void G1ConcurrentMark::scanRootRegion(HeapRegion* hr, uint worker_id) {
949 // Currently, only survivors can be root regions.
950 assert(hr->next_top_at_mark_start() == hr->bottom(), "invariant");
951 G1RootRegionScanClosure cl(_g1h, this, worker_id);
952
953 const uintx interval = PrefetchScanIntervalInBytes;
954 HeapWord* curr = hr->bottom();
955 const HeapWord* end = hr->top();
956 while (curr < end) {
957 Prefetch::read(curr, interval);
958 oop obj = oop(curr);
959 int size = obj->oop_iterate_size(&cl);
960 assert(size == obj->size(), "sanity");
961 curr += size;
962 }
963 }
964
965 class G1CMRootRegionScanTask : public AbstractGangTask {
966 private:
967 G1ConcurrentMark* _cm;
968
969 public:
970 G1CMRootRegionScanTask(G1ConcurrentMark* cm) :
971 AbstractGangTask("Root Region Scan"), _cm(cm) { }
972
973 void work(uint worker_id) {
974 assert(Thread::current()->is_ConcurrentGC_thread(),
975 "this should only be done by a conc GC thread");
976
977 G1CMRootRegions* root_regions = _cm->root_regions();
978 HeapRegion* hr = root_regions->claim_next();
979 while (hr != NULL) {
980 _cm->scanRootRegion(hr, worker_id);
981 hr = root_regions->claim_next();
982 }
983 }
984 };
985
986 void G1ConcurrentMark::scanRootRegions() {
987 // scan_in_progress() will have been set to true only if there was
988 // at least one root region to scan. So, if it's false, we
989 // should not attempt to do any further work.
990 if (root_regions()->scan_in_progress()) {
991 assert(!has_aborted(), "Aborting before root region scanning is finished not supported.");
992 GCTraceConcTime(Info, gc) tt("Concurrent Root Region Scan");
993
994 _parallel_marking_threads = calc_parallel_marking_threads();
995 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
996 "Maximum number of marking threads exceeded");
997 uint active_workers = MAX2(1U, parallel_marking_threads());
998
999 G1CMRootRegionScanTask task(this);
1000 _parallel_workers->set_active_workers(active_workers);
1001 _parallel_workers->run_task(&task);
1002
1003 // It's possible that has_aborted() is true here without actually
1004 // aborting the survivor scan earlier. This is OK as it's
1005 // mainly used for sanity checking.
1006 root_regions()->scan_finished();
1007 }
1008 }
1009
1010 void G1ConcurrentMark::register_concurrent_phase_start(const char* title) {
1011 uint old_val = 0;
1012 do {
1013 old_val = Atomic::cmpxchg(ConcPhaseStarted, &_concurrent_phase_status, ConcPhaseNotStarted);
1014 } while (old_val != ConcPhaseNotStarted);
1015 _g1h->gc_timer_cm()->register_gc_concurrent_start(title);
1016 }
1017
1018 void G1ConcurrentMark::register_concurrent_phase_end_common(bool end_timer) {
1019 if (_concurrent_phase_status == ConcPhaseNotStarted) {
1020 return;
1021 }
1022
1023 uint old_val = Atomic::cmpxchg(ConcPhaseStopping, &_concurrent_phase_status, ConcPhaseStarted);
1024 if (old_val == ConcPhaseStarted) {
1025 _g1h->gc_timer_cm()->register_gc_concurrent_end();
1026 // If 'end_timer' is true, we came here to end timer which needs concurrent phase ended.
1027 // We need to end it before changing the status to 'ConcPhaseNotStarted' to prevent
1028 // starting a new concurrent phase by 'ConcurrentMarkThread'.
1029 if (end_timer) {
1030 _g1h->gc_timer_cm()->register_gc_end();
1031 }
1032 old_val = Atomic::cmpxchg(ConcPhaseNotStarted, &_concurrent_phase_status, ConcPhaseStopping);
1033 assert(old_val == ConcPhaseStopping, "Should not have changed since we entered this scope.");
1034 } else {
1035 do {
1036 // Let other thread finish changing '_concurrent_phase_status' to 'ConcPhaseNotStarted'.
1037 os::naked_short_sleep(1);
1038 } while (_concurrent_phase_status != ConcPhaseNotStarted);
1039 }
1040 }
1041
1042 void G1ConcurrentMark::register_concurrent_phase_end() {
1043 register_concurrent_phase_end_common(false);
1044 }
1045
1046 void G1ConcurrentMark::register_concurrent_gc_end_and_stop_timer() {
1047 register_concurrent_phase_end_common(true);
1048 }
1049
1050 void G1ConcurrentMark::markFromRoots() {
1051 // we might be tempted to assert that:
1052 // assert(asynch == !SafepointSynchronize::is_at_safepoint(),
1053 // "inconsistent argument?");
1054 // However that wouldn't be right, because it's possible that
1055 // a safepoint is indeed in progress as a younger generation
1056 // stop-the-world GC happens even as we mark in this generation.
1057
1058 _restart_for_overflow = false;
1059
1060 // _g1h has _n_par_threads
1061 _parallel_marking_threads = calc_parallel_marking_threads();
1062 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1063 "Maximum number of marking threads exceeded");
1064
1065 uint active_workers = MAX2(1U, parallel_marking_threads());
1066 assert(active_workers > 0, "Should have been set");
1067
1068 // Parallel task terminator is set in "set_concurrency_and_phase()"
1069 set_concurrency_and_phase(active_workers, true /* concurrent */);
1070
1071 G1CMConcurrentMarkingTask markingTask(this, cmThread());
1072 _parallel_workers->set_active_workers(active_workers);
1073 _parallel_workers->run_task(&markingTask);
1074 print_stats();
1075 }
1076
1077 void G1ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) {
1078 // world is stopped at this checkpoint
1079 assert(SafepointSynchronize::is_at_safepoint(),
1080 "world should be stopped");
1081
1082 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1083
1084 // If a full collection has happened, we shouldn't do this.
1085 if (has_aborted()) {
1086 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1087 return;
1088 }
1089
1090 SvcGCMarker sgcm(SvcGCMarker::OTHER);
1091
1092 if (VerifyDuringGC) {
1093 HandleMark hm; // handle scope
1094 g1h->prepare_for_verify();
1095 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (before)");
1096 }
1097 g1h->verifier()->check_bitmaps("Remark Start");
1098
1099 G1CollectorPolicy* g1p = g1h->g1_policy();
1100 g1p->record_concurrent_mark_remark_start();
1101
1102 double start = os::elapsedTime();
1103
1104 checkpointRootsFinalWork();
1105
1106 double mark_work_end = os::elapsedTime();
1107
1108 weakRefsWork(clear_all_soft_refs);
1109
1110 if (has_overflown()) {
1111 // Oops. We overflowed. Restart concurrent marking.
1112 _restart_for_overflow = true;
1113 log_develop_trace(gc)("Remark led to restart for overflow.");
1114
1115 // Verify the heap w.r.t. the previous marking bitmap.
1116 if (VerifyDuringGC) {
1117 HandleMark hm; // handle scope
1118 g1h->prepare_for_verify();
1119 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (overflow)");
1120 }
1121
1122 // Clear the marking state because we will be restarting
1123 // marking due to overflowing the global mark stack.
1124 reset_marking_state();
1125 } else {
1126 {
1127 GCTraceTime(Debug, gc) trace("Aggregate Data", g1h->gc_timer_cm());
1128
1129 // Aggregate the per-task counting data that we have accumulated
1130 // while marking.
1131 aggregate_count_data();
1132 }
1133
1134 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1135 // We're done with marking.
1136 // This is the end of the marking cycle, we're expected all
1137 // threads to have SATB queues with active set to true.
1138 satb_mq_set.set_active_all_threads(false, /* new active value */
1139 true /* expected_active */);
1140
1141 if (VerifyDuringGC) {
1142 HandleMark hm; // handle scope
1143 g1h->prepare_for_verify();
1144 Universe::verify(VerifyOption_G1UseNextMarking, "During GC (after)");
1145 }
1146 g1h->verifier()->check_bitmaps("Remark End");
1147 assert(!restart_for_overflow(), "sanity");
1148 // Completely reset the marking state since marking completed
1149 set_non_marking_state();
1150 }
1151
1152 // Expand the marking stack, if we have to and if we can.
1153 if (_markStack.should_expand()) {
1154 _markStack.expand();
1155 }
1156
1157 // Statistics
1158 double now = os::elapsedTime();
1159 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1160 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1161 _remark_times.add((now - start) * 1000.0);
1162
1163 g1p->record_concurrent_mark_remark_end();
1164
1165 G1CMIsAliveClosure is_alive(g1h);
1166 g1h->gc_tracer_cm()->report_object_count_after_gc(&is_alive);
1167 }
1168
1169 // Base class of the closures that finalize and verify the
1170 // liveness counting data.
1171 class G1CMCountDataClosureBase: public HeapRegionClosure {
1172 protected:
1173 G1CollectedHeap* _g1h;
1174 G1ConcurrentMark* _cm;
1175 CardTableModRefBS* _ct_bs;
1176
1177 BitMap* _region_bm;
1178 BitMap* _card_bm;
1179
1180 // Takes a region that's not empty (i.e., it has at least one
1181 // live object in it and sets its corresponding bit on the region
1182 // bitmap to 1.
1183 void set_bit_for_region(HeapRegion* hr) {
1184 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1185 _region_bm->par_at_put(index, true);
1186 }
1187
1188 public:
1189 G1CMCountDataClosureBase(G1CollectedHeap* g1h,
1190 BitMap* region_bm, BitMap* card_bm):
1191 _g1h(g1h), _cm(g1h->concurrent_mark()),
1192 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
1193 _region_bm(region_bm), _card_bm(card_bm) { }
1194 };
1195
1196 // Closure that calculates the # live objects per region. Used
1197 // for verification purposes during the cleanup pause.
1198 class CalcLiveObjectsClosure: public G1CMCountDataClosureBase {
1199 G1CMBitMapRO* _bm;
1200 size_t _region_marked_bytes;
1201
1202 public:
1203 CalcLiveObjectsClosure(G1CMBitMapRO *bm, G1CollectedHeap* g1h,
1204 BitMap* region_bm, BitMap* card_bm) :
1205 G1CMCountDataClosureBase(g1h, region_bm, card_bm),
1206 _bm(bm), _region_marked_bytes(0) { }
1207
1208 bool doHeapRegion(HeapRegion* hr) {
1209 HeapWord* ntams = hr->next_top_at_mark_start();
1210 HeapWord* start = hr->bottom();
1211
1212 assert(start <= hr->end() && start <= ntams && ntams <= hr->end(),
1213 "Preconditions not met - "
1214 "start: " PTR_FORMAT ", ntams: " PTR_FORMAT ", end: " PTR_FORMAT,
1215 p2i(start), p2i(ntams), p2i(hr->end()));
1216
1217 // Find the first marked object at or after "start".
1218 start = _bm->getNextMarkedWordAddress(start, ntams);
1219
1220 size_t marked_bytes = 0;
1221
1222 while (start < ntams) {
1223 oop obj = oop(start);
1224 int obj_sz = obj->size();
1225 HeapWord* obj_end = start + obj_sz;
1226
1227 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
1228 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(obj_end);
1229
1230 // Note: if we're looking at the last region in heap - obj_end
1231 // could be actually just beyond the end of the heap; end_idx
1232 // will then correspond to a (non-existent) card that is also
1233 // just beyond the heap.
1234 if (_g1h->is_in_g1_reserved(obj_end) && !_ct_bs->is_card_aligned(obj_end)) {
1235 // end of object is not card aligned - increment to cover
1236 // all the cards spanned by the object
1237 end_idx += 1;
1238 }
1239
1240 // Set the bits in the card BM for the cards spanned by this object.
1241 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1242
1243 // Add the size of this object to the number of marked bytes.
1244 marked_bytes += (size_t)obj_sz * HeapWordSize;
1245
1246 // This will happen if we are handling a humongous object that spans
1247 // several heap regions.
1248 if (obj_end > hr->end()) {
1249 break;
1250 }
1251 // Find the next marked object after this one.
1252 start = _bm->getNextMarkedWordAddress(obj_end, ntams);
1253 }
1254
1255 // Mark the allocated-since-marking portion...
1256 HeapWord* top = hr->top();
1257 if (ntams < top) {
1258 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1259 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1260
1261 // Note: if we're looking at the last region in heap - top
1262 // could be actually just beyond the end of the heap; end_idx
1263 // will then correspond to a (non-existent) card that is also
1264 // just beyond the heap.
1265 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1266 // end of object is not card aligned - increment to cover
1267 // all the cards spanned by the object
1268 end_idx += 1;
1269 }
1270 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1271
1272 // This definitely means the region has live objects.
1273 set_bit_for_region(hr);
1274 }
1275
1276 // Update the live region bitmap.
1277 if (marked_bytes > 0) {
1278 set_bit_for_region(hr);
1279 }
1280
1281 // Set the marked bytes for the current region so that
1282 // it can be queried by a calling verification routine
1283 _region_marked_bytes = marked_bytes;
1284
1285 return false;
1286 }
1287
1288 size_t region_marked_bytes() const { return _region_marked_bytes; }
1289 };
1290
1291 // Heap region closure used for verifying the counting data
1292 // that was accumulated concurrently and aggregated during
1293 // the remark pause. This closure is applied to the heap
1294 // regions during the STW cleanup pause.
1295
1296 class VerifyLiveObjectDataHRClosure: public HeapRegionClosure {
1297 G1CollectedHeap* _g1h;
1298 G1ConcurrentMark* _cm;
1299 CalcLiveObjectsClosure _calc_cl;
1300 BitMap* _region_bm; // Region BM to be verified
1301 BitMap* _card_bm; // Card BM to be verified
1302
1303 BitMap* _exp_region_bm; // Expected Region BM values
1304 BitMap* _exp_card_bm; // Expected card BM values
1305
1306 int _failures;
1307
1308 public:
1309 VerifyLiveObjectDataHRClosure(G1CollectedHeap* g1h,
1310 BitMap* region_bm,
1311 BitMap* card_bm,
1312 BitMap* exp_region_bm,
1313 BitMap* exp_card_bm) :
1314 _g1h(g1h), _cm(g1h->concurrent_mark()),
1315 _calc_cl(_cm->nextMarkBitMap(), g1h, exp_region_bm, exp_card_bm),
1316 _region_bm(region_bm), _card_bm(card_bm),
1317 _exp_region_bm(exp_region_bm), _exp_card_bm(exp_card_bm),
1318 _failures(0) { }
1319
1320 int failures() const { return _failures; }
1321
1322 bool doHeapRegion(HeapRegion* hr) {
1323 int failures = 0;
1324
1325 // Call the CalcLiveObjectsClosure to walk the marking bitmap for
1326 // this region and set the corresponding bits in the expected region
1327 // and card bitmaps.
1328 bool res = _calc_cl.doHeapRegion(hr);
1329 assert(res == false, "should be continuing");
1330
1331 // Verify the marked bytes for this region.
1332 size_t exp_marked_bytes = _calc_cl.region_marked_bytes();
1333 size_t act_marked_bytes = hr->next_marked_bytes();
1334
1335 if (exp_marked_bytes > act_marked_bytes) {
1336 if (hr->is_starts_humongous()) {
1337 // For start_humongous regions, the size of the whole object will be
1338 // in exp_marked_bytes.
1339 HeapRegion* region = hr;
1340 int num_regions;
1341 for (num_regions = 0; region != NULL; num_regions++) {
1342 region = _g1h->next_region_in_humongous(region);
1343 }
1344 if ((num_regions-1) * HeapRegion::GrainBytes >= exp_marked_bytes) {
1345 failures += 1;
1346 } else if (num_regions * HeapRegion::GrainBytes < exp_marked_bytes) {
1347 failures += 1;
1348 }
1349 } else {
1350 // We're not OK if expected marked bytes > actual marked bytes. It means
1351 // we have missed accounting some objects during the actual marking.
1352 failures += 1;
1353 }
1354 }
1355
1356 // Verify the bit, for this region, in the actual and expected
1357 // (which was just calculated) region bit maps.
1358 // We're not OK if the bit in the calculated expected region
1359 // bitmap is set and the bit in the actual region bitmap is not.
1360 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1361
1362 bool expected = _exp_region_bm->at(index);
1363 bool actual = _region_bm->at(index);
1364 if (expected && !actual) {
1365 failures += 1;
1366 }
1367
1368 // Verify that the card bit maps for the cards spanned by the current
1369 // region match. We have an error if we have a set bit in the expected
1370 // bit map and the corresponding bit in the actual bitmap is not set.
1371
1372 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(hr->bottom());
1373 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(hr->top());
1374
1375 for (BitMap::idx_t i = start_idx; i < end_idx; i+=1) {
1376 expected = _exp_card_bm->at(i);
1377 actual = _card_bm->at(i);
1378
1379 if (expected && !actual) {
1380 failures += 1;
1381 }
1382 }
1383
1384 _failures += failures;
1385
1386 // We could stop iteration over the heap when we
1387 // find the first violating region by returning true.
1388 return false;
1389 }
1390 };
1391
1392 class G1ParVerifyFinalCountTask: public AbstractGangTask {
1393 protected:
1394 G1CollectedHeap* _g1h;
1395 G1ConcurrentMark* _cm;
1396 BitMap* _actual_region_bm;
1397 BitMap* _actual_card_bm;
1398
1399 uint _n_workers;
1400
1401 BitMap* _expected_region_bm;
1402 BitMap* _expected_card_bm;
1403
1404 int _failures;
1405
1406 HeapRegionClaimer _hrclaimer;
1407
1408 public:
1409 G1ParVerifyFinalCountTask(G1CollectedHeap* g1h,
1410 BitMap* region_bm, BitMap* card_bm,
1411 BitMap* expected_region_bm, BitMap* expected_card_bm)
1412 : AbstractGangTask("G1 verify final counting"),
1413 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1414 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1415 _expected_region_bm(expected_region_bm), _expected_card_bm(expected_card_bm),
1416 _failures(0),
1417 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1418 assert(VerifyDuringGC, "don't call this otherwise");
1419 assert(_expected_card_bm->size() == _actual_card_bm->size(), "sanity");
1420 assert(_expected_region_bm->size() == _actual_region_bm->size(), "sanity");
1421 }
1422
1423 void work(uint worker_id) {
1424 assert(worker_id < _n_workers, "invariant");
1425
1426 VerifyLiveObjectDataHRClosure verify_cl(_g1h,
1427 _actual_region_bm, _actual_card_bm,
1428 _expected_region_bm,
1429 _expected_card_bm);
1430
1431 _g1h->heap_region_par_iterate(&verify_cl, worker_id, &_hrclaimer);
1432
1433 Atomic::add(verify_cl.failures(), &_failures);
1434 }
1435
1436 int failures() const { return _failures; }
1437 };
1438
1439 // Closure that finalizes the liveness counting data.
1440 // Used during the cleanup pause.
1441 // Sets the bits corresponding to the interval [NTAMS, top]
1442 // (which contains the implicitly live objects) in the
1443 // card liveness bitmap. Also sets the bit for each region,
1444 // containing live data, in the region liveness bitmap.
1445
1446 class FinalCountDataUpdateClosure: public G1CMCountDataClosureBase {
1447 public:
1448 FinalCountDataUpdateClosure(G1CollectedHeap* g1h,
1449 BitMap* region_bm,
1450 BitMap* card_bm) :
1451 G1CMCountDataClosureBase(g1h, region_bm, card_bm) { }
1452
1453 bool doHeapRegion(HeapRegion* hr) {
1454 HeapWord* ntams = hr->next_top_at_mark_start();
1455 HeapWord* top = hr->top();
1456
1457 assert(hr->bottom() <= ntams && ntams <= hr->end(), "Preconditions.");
1458
1459 // Mark the allocated-since-marking portion...
1460 if (ntams < top) {
1461 // This definitely means the region has live objects.
1462 set_bit_for_region(hr);
1463
1464 // Now set the bits in the card bitmap for [ntams, top)
1465 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1466 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1467
1468 // Note: if we're looking at the last region in heap - top
1469 // could be actually just beyond the end of the heap; end_idx
1470 // will then correspond to a (non-existent) card that is also
1471 // just beyond the heap.
1472 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1473 // end of object is not card aligned - increment to cover
1474 // all the cards spanned by the object
1475 end_idx += 1;
1476 }
1477
1478 assert(end_idx <= _card_bm->size(),
1479 "oob: end_idx= " SIZE_FORMAT ", bitmap size= " SIZE_FORMAT,
1480 end_idx, _card_bm->size());
1481 assert(start_idx < _card_bm->size(),
1482 "oob: start_idx= " SIZE_FORMAT ", bitmap size= " SIZE_FORMAT,
1483 start_idx, _card_bm->size());
1484
1485 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1486 }
1487
1488 // Set the bit for the region if it contains live data
1489 if (hr->next_marked_bytes() > 0) {
1490 set_bit_for_region(hr);
1491 }
1492
1493 return false;
1494 }
1495 };
1496
1497 class G1ParFinalCountTask: public AbstractGangTask {
1498 protected:
1499 G1CollectedHeap* _g1h;
1500 G1ConcurrentMark* _cm;
1501 BitMap* _actual_region_bm;
1502 BitMap* _actual_card_bm;
1503
1504 uint _n_workers;
1505 HeapRegionClaimer _hrclaimer;
1506
1507 public:
1508 G1ParFinalCountTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm)
1509 : AbstractGangTask("G1 final counting"),
1510 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1511 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1512 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1513 }
1514
1515 void work(uint worker_id) {
1516 assert(worker_id < _n_workers, "invariant");
1517
1518 FinalCountDataUpdateClosure final_update_cl(_g1h,
1519 _actual_region_bm,
1520 _actual_card_bm);
1521
1522 _g1h->heap_region_par_iterate(&final_update_cl, worker_id, &_hrclaimer);
1523 }
1524 };
1525
1526 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure {
1527 G1CollectedHeap* _g1;
1528 size_t _freed_bytes;
1529 FreeRegionList* _local_cleanup_list;
1530 uint _old_regions_removed;
1531 uint _humongous_regions_removed;
1532 HRRSCleanupTask* _hrrs_cleanup_task;
1533
1534 public:
1535 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1,
1536 FreeRegionList* local_cleanup_list,
1537 HRRSCleanupTask* hrrs_cleanup_task) :
1538 _g1(g1),
1539 _freed_bytes(0),
1540 _local_cleanup_list(local_cleanup_list),
1541 _old_regions_removed(0),
1542 _humongous_regions_removed(0),
1543 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1544
1545 size_t freed_bytes() { return _freed_bytes; }
1546 const uint old_regions_removed() { return _old_regions_removed; }
1547 const uint humongous_regions_removed() { return _humongous_regions_removed; }
1548
1549 bool doHeapRegion(HeapRegion *hr) {
1550 if (hr->is_archive()) {
1551 return false;
1552 }
1553 // We use a claim value of zero here because all regions
1554 // were claimed with value 1 in the FinalCount task.
1555 _g1->reset_gc_time_stamps(hr);
1556 hr->note_end_of_marking();
1557
1558 if (hr->used() > 0 && hr->max_live_bytes() == 0 && !hr->is_young()) {
1559 _freed_bytes += hr->used();
1560 hr->set_containing_set(NULL);
1561 if (hr->is_humongous()) {
1562 _humongous_regions_removed++;
1563 _g1->free_humongous_region(hr, _local_cleanup_list, true);
1564 } else {
1565 _old_regions_removed++;
1566 _g1->free_region(hr, _local_cleanup_list, true);
1567 }
1568 } else {
1569 hr->rem_set()->do_cleanup_work(_hrrs_cleanup_task);
1570 }
1571
1572 return false;
1573 }
1574 };
1575
1576 class G1ParNoteEndTask: public AbstractGangTask {
1577 friend class G1NoteEndOfConcMarkClosure;
1578
1579 protected:
1580 G1CollectedHeap* _g1h;
1581 FreeRegionList* _cleanup_list;
1582 HeapRegionClaimer _hrclaimer;
1583
1584 public:
1585 G1ParNoteEndTask(G1CollectedHeap* g1h, FreeRegionList* cleanup_list, uint n_workers) :
1586 AbstractGangTask("G1 note end"), _g1h(g1h), _cleanup_list(cleanup_list), _hrclaimer(n_workers) {
1587 }
1588
1589 void work(uint worker_id) {
1590 FreeRegionList local_cleanup_list("Local Cleanup List");
1591 HRRSCleanupTask hrrs_cleanup_task;
1592 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, &local_cleanup_list,
1593 &hrrs_cleanup_task);
1594 _g1h->heap_region_par_iterate(&g1_note_end, worker_id, &_hrclaimer);
1595 assert(g1_note_end.complete(), "Shouldn't have yielded!");
1596
1597 // Now update the lists
1598 _g1h->remove_from_old_sets(g1_note_end.old_regions_removed(), g1_note_end.humongous_regions_removed());
1599 {
1600 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1601 _g1h->decrement_summary_bytes(g1_note_end.freed_bytes());
1602
1603 // If we iterate over the global cleanup list at the end of
1604 // cleanup to do this printing we will not guarantee to only
1605 // generate output for the newly-reclaimed regions (the list
1606 // might not be empty at the beginning of cleanup; we might
1607 // still be working on its previous contents). So we do the
1608 // printing here, before we append the new regions to the global
1609 // cleanup list.
1610
1611 G1HRPrinter* hr_printer = _g1h->hr_printer();
1612 if (hr_printer->is_active()) {
1613 FreeRegionListIterator iter(&local_cleanup_list);
1614 while (iter.more_available()) {
1615 HeapRegion* hr = iter.get_next();
1616 hr_printer->cleanup(hr);
1617 }
1618 }
1619
1620 _cleanup_list->add_ordered(&local_cleanup_list);
1621 assert(local_cleanup_list.is_empty(), "post-condition");
1622
1623 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1624 }
1625 }
1626 };
1627
1628 void G1ConcurrentMark::cleanup() {
1629 // world is stopped at this checkpoint
1630 assert(SafepointSynchronize::is_at_safepoint(),
1631 "world should be stopped");
1632 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1633
1634 // If a full collection has happened, we shouldn't do this.
1635 if (has_aborted()) {
1636 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1637 return;
1638 }
1639
1640 g1h->verifier()->verify_region_sets_optional();
1641
1642 if (VerifyDuringGC) {
1643 HandleMark hm; // handle scope
1644 g1h->prepare_for_verify();
1645 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (before)");
1646 }
1647 g1h->verifier()->check_bitmaps("Cleanup Start");
1648
1649 G1CollectorPolicy* g1p = g1h->g1_policy();
1650 g1p->record_concurrent_mark_cleanup_start();
1651
1652 double start = os::elapsedTime();
1653
1654 HeapRegionRemSet::reset_for_cleanup_tasks();
1655
1656 // Do counting once more with the world stopped for good measure.
1657 G1ParFinalCountTask g1_par_count_task(g1h, &_region_bm, &_card_bm);
1658
1659 g1h->workers()->run_task(&g1_par_count_task);
1660
1661 if (VerifyDuringGC) {
1662 // Verify that the counting data accumulated during marking matches
1663 // that calculated by walking the marking bitmap.
1664
1665 // Bitmaps to hold expected values
1666 BitMap expected_region_bm(_region_bm.size(), true);
1667 BitMap expected_card_bm(_card_bm.size(), true);
1668
1669 G1ParVerifyFinalCountTask g1_par_verify_task(g1h,
1670 &_region_bm,
1671 &_card_bm,
1672 &expected_region_bm,
1673 &expected_card_bm);
1674
1675 g1h->workers()->run_task(&g1_par_verify_task);
1676
1677 guarantee(g1_par_verify_task.failures() == 0, "Unexpected accounting failures");
1678 }
1679
1680 size_t start_used_bytes = g1h->used();
1681 g1h->collector_state()->set_mark_in_progress(false);
1682
1683 double count_end = os::elapsedTime();
1684 double this_final_counting_time = (count_end - start);
1685 _total_counting_time += this_final_counting_time;
1686
1687 if (log_is_enabled(Trace, gc, liveness)) {
1688 G1PrintRegionLivenessInfoClosure cl("Post-Marking");
1689 _g1h->heap_region_iterate(&cl);
1690 }
1691
1692 // Install newly created mark bitMap as "prev".
1693 swapMarkBitMaps();
1694
1695 g1h->reset_gc_time_stamp();
1696
1697 uint n_workers = _g1h->workers()->active_workers();
1698
1699 // Note end of marking in all heap regions.
1700 G1ParNoteEndTask g1_par_note_end_task(g1h, &_cleanup_list, n_workers);
1701 g1h->workers()->run_task(&g1_par_note_end_task);
1702 g1h->check_gc_time_stamps();
1703
1704 if (!cleanup_list_is_empty()) {
1705 // The cleanup list is not empty, so we'll have to process it
1706 // concurrently. Notify anyone else that might be wanting free
1707 // regions that there will be more free regions coming soon.
1708 g1h->set_free_regions_coming();
1709 }
1710
1711 // call below, since it affects the metric by which we sort the heap
1712 // regions.
1713 if (G1ScrubRemSets) {
1714 double rs_scrub_start = os::elapsedTime();
1715 g1h->scrub_rem_set(&_region_bm, &_card_bm);
1716 _total_rs_scrub_time += (os::elapsedTime() - rs_scrub_start);
1717 }
1718
1719 // this will also free any regions totally full of garbage objects,
1720 // and sort the regions.
1721 g1h->g1_policy()->record_concurrent_mark_cleanup_end();
1722
1723 // Statistics.
1724 double end = os::elapsedTime();
1725 _cleanup_times.add((end - start) * 1000.0);
1726
1727 // Clean up will have freed any regions completely full of garbage.
1728 // Update the soft reference policy with the new heap occupancy.
1729 Universe::update_heap_info_at_gc();
1730
1731 if (VerifyDuringGC) {
1732 HandleMark hm; // handle scope
1733 g1h->prepare_for_verify();
1734 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (after)");
1735 }
1736
1737 g1h->verifier()->check_bitmaps("Cleanup End");
1738
1739 g1h->verifier()->verify_region_sets_optional();
1740
1741 // We need to make this be a "collection" so any collection pause that
1742 // races with it goes around and waits for completeCleanup to finish.
1743 g1h->increment_total_collections();
1744
1745 // Clean out dead classes and update Metaspace sizes.
1746 if (ClassUnloadingWithConcurrentMark) {
1747 ClassLoaderDataGraph::purge();
1748 }
1749 MetaspaceGC::compute_new_size();
1750
1751 // We reclaimed old regions so we should calculate the sizes to make
1752 // sure we update the old gen/space data.
1753 g1h->g1mm()->update_sizes();
1754 g1h->allocation_context_stats().update_after_mark();
1755
1756 g1h->trace_heap_after_concurrent_cycle();
1757 }
1758
1759 void G1ConcurrentMark::completeCleanup() {
1760 if (has_aborted()) return;
1761
1762 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1763
1764 _cleanup_list.verify_optional();
1765 FreeRegionList tmp_free_list("Tmp Free List");
1766
1767 log_develop_trace(gc, freelist)("G1ConcRegionFreeing [complete cleanup] : "
1768 "cleanup list has %u entries",
1769 _cleanup_list.length());
1770
1771 // No one else should be accessing the _cleanup_list at this point,
1772 // so it is not necessary to take any locks
1773 while (!_cleanup_list.is_empty()) {
1774 HeapRegion* hr = _cleanup_list.remove_region(true /* from_head */);
1775 assert(hr != NULL, "Got NULL from a non-empty list");
1776 hr->par_clear();
1777 tmp_free_list.add_ordered(hr);
1778
1779 // Instead of adding one region at a time to the secondary_free_list,
1780 // we accumulate them in the local list and move them a few at a
1781 // time. This also cuts down on the number of notify_all() calls
1782 // we do during this process. We'll also append the local list when
1783 // _cleanup_list is empty (which means we just removed the last
1784 // region from the _cleanup_list).
1785 if ((tmp_free_list.length() % G1SecondaryFreeListAppendLength == 0) ||
1786 _cleanup_list.is_empty()) {
1787 log_develop_trace(gc, freelist)("G1ConcRegionFreeing [complete cleanup] : "
1788 "appending %u entries to the secondary_free_list, "
1789 "cleanup list still has %u entries",
1790 tmp_free_list.length(),
1791 _cleanup_list.length());
1792
1793 {
1794 MutexLockerEx x(SecondaryFreeList_lock, Mutex::_no_safepoint_check_flag);
1795 g1h->secondary_free_list_add(&tmp_free_list);
1796 SecondaryFreeList_lock->notify_all();
1797 }
1798 #ifndef PRODUCT
1799 if (G1StressConcRegionFreeing) {
1800 for (uintx i = 0; i < G1StressConcRegionFreeingDelayMillis; ++i) {
1801 os::sleep(Thread::current(), (jlong) 1, false);
1802 }
1803 }
1804 #endif
1805 }
1806 }
1807 assert(tmp_free_list.is_empty(), "post-condition");
1808 }
1809
1810 // Supporting Object and Oop closures for reference discovery
1811 // and processing in during marking
1812
1813 bool G1CMIsAliveClosure::do_object_b(oop obj) {
1814 HeapWord* addr = (HeapWord*)obj;
1815 return addr != NULL &&
1816 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj));
1817 }
1818
1819 // 'Keep Alive' oop closure used by both serial parallel reference processing.
1820 // Uses the G1CMTask associated with a worker thread (for serial reference
1821 // processing the G1CMTask for worker 0 is used) to preserve (mark) and
1822 // trace referent objects.
1823 //
1824 // Using the G1CMTask and embedded local queues avoids having the worker
1825 // threads operating on the global mark stack. This reduces the risk
1826 // of overflowing the stack - which we would rather avoid at this late
1827 // state. Also using the tasks' local queues removes the potential
1828 // of the workers interfering with each other that could occur if
1829 // operating on the global stack.
1830
1831 class G1CMKeepAliveAndDrainClosure: public OopClosure {
1832 G1ConcurrentMark* _cm;
1833 G1CMTask* _task;
1834 int _ref_counter_limit;
1835 int _ref_counter;
1836 bool _is_serial;
1837 public:
1838 G1CMKeepAliveAndDrainClosure(G1ConcurrentMark* cm, G1CMTask* task, bool is_serial) :
1839 _cm(cm), _task(task), _is_serial(is_serial),
1840 _ref_counter_limit(G1RefProcDrainInterval) {
1841 assert(_ref_counter_limit > 0, "sanity");
1842 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1843 _ref_counter = _ref_counter_limit;
1844 }
1845
1846 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
1847 virtual void do_oop( oop* p) { do_oop_work(p); }
1848
1849 template <class T> void do_oop_work(T* p) {
1850 if (!_cm->has_overflown()) {
1851 oop obj = oopDesc::load_decode_heap_oop(p);
1852 _task->deal_with_reference(obj);
1853 _ref_counter--;
1854
1855 if (_ref_counter == 0) {
1856 // We have dealt with _ref_counter_limit references, pushing them
1857 // and objects reachable from them on to the local stack (and
1858 // possibly the global stack). Call G1CMTask::do_marking_step() to
1859 // process these entries.
1860 //
1861 // We call G1CMTask::do_marking_step() in a loop, which we'll exit if
1862 // there's nothing more to do (i.e. we're done with the entries that
1863 // were pushed as a result of the G1CMTask::deal_with_reference() calls
1864 // above) or we overflow.
1865 //
1866 // Note: G1CMTask::do_marking_step() can set the G1CMTask::has_aborted()
1867 // flag while there may still be some work to do. (See the comment at
1868 // the beginning of G1CMTask::do_marking_step() for those conditions -
1869 // one of which is reaching the specified time target.) It is only
1870 // when G1CMTask::do_marking_step() returns without setting the
1871 // has_aborted() flag that the marking step has completed.
1872 do {
1873 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1874 _task->do_marking_step(mark_step_duration_ms,
1875 false /* do_termination */,
1876 _is_serial);
1877 } while (_task->has_aborted() && !_cm->has_overflown());
1878 _ref_counter = _ref_counter_limit;
1879 }
1880 }
1881 }
1882 };
1883
1884 // 'Drain' oop closure used by both serial and parallel reference processing.
1885 // Uses the G1CMTask associated with a given worker thread (for serial
1886 // reference processing the G1CMtask for worker 0 is used). Calls the
1887 // do_marking_step routine, with an unbelievably large timeout value,
1888 // to drain the marking data structures of the remaining entries
1889 // added by the 'keep alive' oop closure above.
1890
1891 class G1CMDrainMarkingStackClosure: public VoidClosure {
1892 G1ConcurrentMark* _cm;
1893 G1CMTask* _task;
1894 bool _is_serial;
1895 public:
1896 G1CMDrainMarkingStackClosure(G1ConcurrentMark* cm, G1CMTask* task, bool is_serial) :
1897 _cm(cm), _task(task), _is_serial(is_serial) {
1898 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1899 }
1900
1901 void do_void() {
1902 do {
1903 // We call G1CMTask::do_marking_step() to completely drain the local
1904 // and global marking stacks of entries pushed by the 'keep alive'
1905 // oop closure (an instance of G1CMKeepAliveAndDrainClosure above).
1906 //
1907 // G1CMTask::do_marking_step() is called in a loop, which we'll exit
1908 // if there's nothing more to do (i.e. we've completely drained the
1909 // entries that were pushed as a a result of applying the 'keep alive'
1910 // closure to the entries on the discovered ref lists) or we overflow
1911 // the global marking stack.
1912 //
1913 // Note: G1CMTask::do_marking_step() can set the G1CMTask::has_aborted()
1914 // flag while there may still be some work to do. (See the comment at
1915 // the beginning of G1CMTask::do_marking_step() for those conditions -
1916 // one of which is reaching the specified time target.) It is only
1917 // when G1CMTask::do_marking_step() returns without setting the
1918 // has_aborted() flag that the marking step has completed.
1919
1920 _task->do_marking_step(1000000000.0 /* something very large */,
1921 true /* do_termination */,
1922 _is_serial);
1923 } while (_task->has_aborted() && !_cm->has_overflown());
1924 }
1925 };
1926
1927 // Implementation of AbstractRefProcTaskExecutor for parallel
1928 // reference processing at the end of G1 concurrent marking
1929
1930 class G1CMRefProcTaskExecutor: public AbstractRefProcTaskExecutor {
1931 private:
1932 G1CollectedHeap* _g1h;
1933 G1ConcurrentMark* _cm;
1934 WorkGang* _workers;
1935 uint _active_workers;
1936
1937 public:
1938 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
1939 G1ConcurrentMark* cm,
1940 WorkGang* workers,
1941 uint n_workers) :
1942 _g1h(g1h), _cm(cm),
1943 _workers(workers), _active_workers(n_workers) { }
1944
1945 // Executes the given task using concurrent marking worker threads.
1946 virtual void execute(ProcessTask& task);
1947 virtual void execute(EnqueueTask& task);
1948 };
1949
1950 class G1CMRefProcTaskProxy: public AbstractGangTask {
1951 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
1952 ProcessTask& _proc_task;
1953 G1CollectedHeap* _g1h;
1954 G1ConcurrentMark* _cm;
1955
1956 public:
1957 G1CMRefProcTaskProxy(ProcessTask& proc_task,
1958 G1CollectedHeap* g1h,
1959 G1ConcurrentMark* cm) :
1960 AbstractGangTask("Process reference objects in parallel"),
1961 _proc_task(proc_task), _g1h(g1h), _cm(cm) {
1962 ReferenceProcessor* rp = _g1h->ref_processor_cm();
1963 assert(rp->processing_is_mt(), "shouldn't be here otherwise");
1964 }
1965
1966 virtual void work(uint worker_id) {
1967 ResourceMark rm;
1968 HandleMark hm;
1969 G1CMTask* task = _cm->task(worker_id);
1970 G1CMIsAliveClosure g1_is_alive(_g1h);
1971 G1CMKeepAliveAndDrainClosure g1_par_keep_alive(_cm, task, false /* is_serial */);
1972 G1CMDrainMarkingStackClosure g1_par_drain(_cm, task, false /* is_serial */);
1973
1974 _proc_task.work(worker_id, g1_is_alive, g1_par_keep_alive, g1_par_drain);
1975 }
1976 };
1977
1978 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
1979 assert(_workers != NULL, "Need parallel worker threads.");
1980 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
1981
1982 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
1983
1984 // We need to reset the concurrency level before each
1985 // proxy task execution, so that the termination protocol
1986 // and overflow handling in G1CMTask::do_marking_step() knows
1987 // how many workers to wait for.
1988 _cm->set_concurrency(_active_workers);
1989 _workers->run_task(&proc_task_proxy);
1990 }
1991
1992 class G1CMRefEnqueueTaskProxy: public AbstractGangTask {
1993 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
1994 EnqueueTask& _enq_task;
1995
1996 public:
1997 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
1998 AbstractGangTask("Enqueue reference objects in parallel"),
1999 _enq_task(enq_task) { }
2000
2001 virtual void work(uint worker_id) {
2002 _enq_task.work(worker_id);
2003 }
2004 };
2005
2006 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
2007 assert(_workers != NULL, "Need parallel worker threads.");
2008 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2009
2010 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
2011
2012 // Not strictly necessary but...
2013 //
2014 // We need to reset the concurrency level before each
2015 // proxy task execution, so that the termination protocol
2016 // and overflow handling in G1CMTask::do_marking_step() knows
2017 // how many workers to wait for.
2018 _cm->set_concurrency(_active_workers);
2019 _workers->run_task(&enq_task_proxy);
2020 }
2021
2022 void G1ConcurrentMark::weakRefsWorkParallelPart(BoolObjectClosure* is_alive, bool purged_classes) {
2023 G1CollectedHeap::heap()->parallel_cleaning(is_alive, true, true, purged_classes);
2024 }
2025
2026 void G1ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) {
2027 if (has_overflown()) {
2028 // Skip processing the discovered references if we have
2029 // overflown the global marking stack. Reference objects
2030 // only get discovered once so it is OK to not
2031 // de-populate the discovered reference lists. We could have,
2032 // but the only benefit would be that, when marking restarts,
2033 // less reference objects are discovered.
2034 return;
2035 }
2036
2037 ResourceMark rm;
2038 HandleMark hm;
2039
2040 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2041
2042 // Is alive closure.
2043 G1CMIsAliveClosure g1_is_alive(g1h);
2044
2045 // Inner scope to exclude the cleaning of the string and symbol
2046 // tables from the displayed time.
2047 {
2048 GCTraceTime(Debug, gc) trace("Reference Processing", g1h->gc_timer_cm());
2049
2050 ReferenceProcessor* rp = g1h->ref_processor_cm();
2051
2052 // See the comment in G1CollectedHeap::ref_processing_init()
2053 // about how reference processing currently works in G1.
2054
2055 // Set the soft reference policy
2056 rp->setup_policy(clear_all_soft_refs);
2057 assert(_markStack.isEmpty(), "mark stack should be empty");
2058
2059 // Instances of the 'Keep Alive' and 'Complete GC' closures used
2060 // in serial reference processing. Note these closures are also
2061 // used for serially processing (by the the current thread) the
2062 // JNI references during parallel reference processing.
2063 //
2064 // These closures do not need to synchronize with the worker
2065 // threads involved in parallel reference processing as these
2066 // instances are executed serially by the current thread (e.g.
2067 // reference processing is not multi-threaded and is thus
2068 // performed by the current thread instead of a gang worker).
2069 //
2070 // The gang tasks involved in parallel reference processing create
2071 // their own instances of these closures, which do their own
2072 // synchronization among themselves.
2073 G1CMKeepAliveAndDrainClosure g1_keep_alive(this, task(0), true /* is_serial */);
2074 G1CMDrainMarkingStackClosure g1_drain_mark_stack(this, task(0), true /* is_serial */);
2075
2076 // We need at least one active thread. If reference processing
2077 // is not multi-threaded we use the current (VMThread) thread,
2078 // otherwise we use the work gang from the G1CollectedHeap and
2079 // we utilize all the worker threads we can.
2080 bool processing_is_mt = rp->processing_is_mt();
2081 uint active_workers = (processing_is_mt ? g1h->workers()->active_workers() : 1U);
2082 active_workers = MAX2(MIN2(active_workers, _max_worker_id), 1U);
2083
2084 // Parallel processing task executor.
2085 G1CMRefProcTaskExecutor par_task_executor(g1h, this,
2086 g1h->workers(), active_workers);
2087 AbstractRefProcTaskExecutor* executor = (processing_is_mt ? &par_task_executor : NULL);
2088
2089 // Set the concurrency level. The phase was already set prior to
2090 // executing the remark task.
2091 set_concurrency(active_workers);
2092
2093 // Set the degree of MT processing here. If the discovery was done MT,
2094 // the number of threads involved during discovery could differ from
2095 // the number of active workers. This is OK as long as the discovered
2096 // Reference lists are balanced (see balance_all_queues() and balance_queues()).
2097 rp->set_active_mt_degree(active_workers);
2098
2099 // Process the weak references.
2100 const ReferenceProcessorStats& stats =
2101 rp->process_discovered_references(&g1_is_alive,
2102 &g1_keep_alive,
2103 &g1_drain_mark_stack,
2104 executor,
2105 g1h->gc_timer_cm());
2106 g1h->gc_tracer_cm()->report_gc_reference_stats(stats);
2107
2108 // The do_oop work routines of the keep_alive and drain_marking_stack
2109 // oop closures will set the has_overflown flag if we overflow the
2110 // global marking stack.
2111
2112 assert(_markStack.overflow() || _markStack.isEmpty(),
2113 "mark stack should be empty (unless it overflowed)");
2114
2115 if (_markStack.overflow()) {
2116 // This should have been done already when we tried to push an
2117 // entry on to the global mark stack. But let's do it again.
2118 set_has_overflown();
2119 }
2120
2121 assert(rp->num_q() == active_workers, "why not");
2122
2123 rp->enqueue_discovered_references(executor);
2124
2125 rp->verify_no_references_recorded();
2126 assert(!rp->discovery_enabled(), "Post condition");
2127 }
2128
2129 if (has_overflown()) {
2130 // We can not trust g1_is_alive if the marking stack overflowed
2131 return;
2132 }
2133
2134 assert(_markStack.isEmpty(), "Marking should have completed");
2135
2136 // Unload Klasses, String, Symbols, Code Cache, etc.
2137 {
2138 GCTraceTime(Debug, gc) trace("Unloading", g1h->gc_timer_cm());
2139
2140 if (ClassUnloadingWithConcurrentMark) {
2141 bool purged_classes;
2142
2143 {
2144 GCTraceTime(Trace, gc) trace("System Dictionary Unloading", g1h->gc_timer_cm());
2145 purged_classes = SystemDictionary::do_unloading(&g1_is_alive, false /* Defer klass cleaning */);
2146 }
2147
2148 {
2149 GCTraceTime(Trace, gc) trace("Parallel Unloading", g1h->gc_timer_cm());
2150 weakRefsWorkParallelPart(&g1_is_alive, purged_classes);
2151 }
2152 }
2153
2154 if (G1StringDedup::is_enabled()) {
2155 GCTraceTime(Trace, gc) trace("String Deduplication Unlink", g1h->gc_timer_cm());
2156 G1StringDedup::unlink(&g1_is_alive);
2157 }
2158 }
2159 }
2160
2161 void G1ConcurrentMark::swapMarkBitMaps() {
2162 G1CMBitMapRO* temp = _prevMarkBitMap;
2163 _prevMarkBitMap = (G1CMBitMapRO*)_nextMarkBitMap;
2164 _nextMarkBitMap = (G1CMBitMap*) temp;
2165 }
2166
2167 // Closure for marking entries in SATB buffers.
2168 class G1CMSATBBufferClosure : public SATBBufferClosure {
2169 private:
2170 G1CMTask* _task;
2171 G1CollectedHeap* _g1h;
2172
2173 // This is very similar to G1CMTask::deal_with_reference, but with
2174 // more relaxed requirements for the argument, so this must be more
2175 // circumspect about treating the argument as an object.
2176 void do_entry(void* entry) const {
2177 _task->increment_refs_reached();
2178 HeapRegion* hr = _g1h->heap_region_containing(entry);
2179 if (entry < hr->next_top_at_mark_start()) {
2180 // Until we get here, we don't know whether entry refers to a valid
2181 // object; it could instead have been a stale reference.
2182 oop obj = static_cast<oop>(entry);
2183 assert(obj->is_oop(true /* ignore mark word */),
2184 "Invalid oop in SATB buffer: " PTR_FORMAT, p2i(obj));
2185 _task->make_reference_grey(obj, hr);
2186 }
2187 }
2188
2189 public:
2190 G1CMSATBBufferClosure(G1CMTask* task, G1CollectedHeap* g1h)
2191 : _task(task), _g1h(g1h) { }
2192
2193 virtual void do_buffer(void** buffer, size_t size) {
2194 for (size_t i = 0; i < size; ++i) {
2195 do_entry(buffer[i]);
2196 }
2197 }
2198 };
2199
2200 class G1RemarkThreadsClosure : public ThreadClosure {
2201 G1CMSATBBufferClosure _cm_satb_cl;
2202 G1CMOopClosure _cm_cl;
2203 MarkingCodeBlobClosure _code_cl;
2204 int _thread_parity;
2205
2206 public:
2207 G1RemarkThreadsClosure(G1CollectedHeap* g1h, G1CMTask* task) :
2208 _cm_satb_cl(task, g1h),
2209 _cm_cl(g1h, g1h->concurrent_mark(), task),
2210 _code_cl(&_cm_cl, !CodeBlobToOopClosure::FixRelocations),
2211 _thread_parity(Threads::thread_claim_parity()) {}
2212
2213 void do_thread(Thread* thread) {
2214 if (thread->is_Java_thread()) {
2215 if (thread->claim_oops_do(true, _thread_parity)) {
2216 JavaThread* jt = (JavaThread*)thread;
2217
2218 // In theory it should not be neccessary to explicitly walk the nmethods to find roots for concurrent marking
2219 // however the liveness of oops reachable from nmethods have very complex lifecycles:
2220 // * Alive if on the stack of an executing method
2221 // * Weakly reachable otherwise
2222 // Some objects reachable from nmethods, such as the class loader (or klass_holder) of the receiver should be
2223 // live by the SATB invariant but other oops recorded in nmethods may behave differently.
2224 jt->nmethods_do(&_code_cl);
2225
2226 jt->satb_mark_queue().apply_closure_and_empty(&_cm_satb_cl);
2227 }
2228 } else if (thread->is_VM_thread()) {
2229 if (thread->claim_oops_do(true, _thread_parity)) {
2230 JavaThread::satb_mark_queue_set().shared_satb_queue()->apply_closure_and_empty(&_cm_satb_cl);
2231 }
2232 }
2233 }
2234 };
2235
2236 class G1CMRemarkTask: public AbstractGangTask {
2237 private:
2238 G1ConcurrentMark* _cm;
2239 public:
2240 void work(uint worker_id) {
2241 // Since all available tasks are actually started, we should
2242 // only proceed if we're supposed to be active.
2243 if (worker_id < _cm->active_tasks()) {
2244 G1CMTask* task = _cm->task(worker_id);
2245 task->record_start_time();
2246 {
2247 ResourceMark rm;
2248 HandleMark hm;
2249
2250 G1RemarkThreadsClosure threads_f(G1CollectedHeap::heap(), task);
2251 Threads::threads_do(&threads_f);
2252 }
2253
2254 do {
2255 task->do_marking_step(1000000000.0 /* something very large */,
2256 true /* do_termination */,
2257 false /* is_serial */);
2258 } while (task->has_aborted() && !_cm->has_overflown());
2259 // If we overflow, then we do not want to restart. We instead
2260 // want to abort remark and do concurrent marking again.
2261 task->record_end_time();
2262 }
2263 }
2264
2265 G1CMRemarkTask(G1ConcurrentMark* cm, uint active_workers) :
2266 AbstractGangTask("Par Remark"), _cm(cm) {
2267 _cm->terminator()->reset_for_reuse(active_workers);
2268 }
2269 };
2270
2271 void G1ConcurrentMark::checkpointRootsFinalWork() {
2272 ResourceMark rm;
2273 HandleMark hm;
2274 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2275
2276 GCTraceTime(Debug, gc) trace("Finalize Marking", g1h->gc_timer_cm());
2277
2278 g1h->ensure_parsability(false);
2279
2280 // this is remark, so we'll use up all active threads
2281 uint active_workers = g1h->workers()->active_workers();
2282 set_concurrency_and_phase(active_workers, false /* concurrent */);
2283 // Leave _parallel_marking_threads at it's
2284 // value originally calculated in the G1ConcurrentMark
2285 // constructor and pass values of the active workers
2286 // through the gang in the task.
2287
2288 {
2289 StrongRootsScope srs(active_workers);
2290
2291 G1CMRemarkTask remarkTask(this, active_workers);
2292 // We will start all available threads, even if we decide that the
2293 // active_workers will be fewer. The extra ones will just bail out
2294 // immediately.
2295 g1h->workers()->run_task(&remarkTask);
2296 }
2297
2298 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2299 guarantee(has_overflown() ||
2300 satb_mq_set.completed_buffers_num() == 0,
2301 "Invariant: has_overflown = %s, num buffers = " SIZE_FORMAT,
2302 BOOL_TO_STR(has_overflown()),
2303 satb_mq_set.completed_buffers_num());
2304
2305 print_stats();
2306 }
2307
2308 void G1ConcurrentMark::clearRangePrevBitmap(MemRegion mr) {
2309 // Note we are overriding the read-only view of the prev map here, via
2310 // the cast.
2311 ((G1CMBitMap*)_prevMarkBitMap)->clearRange(mr);
2312 }
2313
2314 HeapRegion*
2315 G1ConcurrentMark::claim_region(uint worker_id) {
2316 // "checkpoint" the finger
2317 HeapWord* finger = _finger;
2318
2319 // _heap_end will not change underneath our feet; it only changes at
2320 // yield points.
2321 while (finger < _heap_end) {
2322 assert(_g1h->is_in_g1_reserved(finger), "invariant");
2323
2324 HeapRegion* curr_region = _g1h->heap_region_containing(finger);
2325
2326 // Above heap_region_containing may return NULL as we always scan claim
2327 // until the end of the heap. In this case, just jump to the next region.
2328 HeapWord* end = curr_region != NULL ? curr_region->end() : finger + HeapRegion::GrainWords;
2329
2330 // Is the gap between reading the finger and doing the CAS too long?
2331 HeapWord* res = (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger);
2332 if (res == finger && curr_region != NULL) {
2333 // we succeeded
2334 HeapWord* bottom = curr_region->bottom();
2335 HeapWord* limit = curr_region->next_top_at_mark_start();
2336
2337 // notice that _finger == end cannot be guaranteed here since,
2338 // someone else might have moved the finger even further
2339 assert(_finger >= end, "the finger should have moved forward");
2340
2341 if (limit > bottom) {
2342 return curr_region;
2343 } else {
2344 assert(limit == bottom,
2345 "the region limit should be at bottom");
2346 // we return NULL and the caller should try calling
2347 // claim_region() again.
2348 return NULL;
2349 }
2350 } else {
2351 assert(_finger > finger, "the finger should have moved forward");
2352 // read it again
2353 finger = _finger;
2354 }
2355 }
2356
2357 return NULL;
2358 }
2359
2360 #ifndef PRODUCT
2361 class VerifyNoCSetOops VALUE_OBJ_CLASS_SPEC {
2362 private:
2363 G1CollectedHeap* _g1h;
2364 const char* _phase;
2365 int _info;
2366
2367 public:
2368 VerifyNoCSetOops(const char* phase, int info = -1) :
2369 _g1h(G1CollectedHeap::heap()),
2370 _phase(phase),
2371 _info(info)
2372 { }
2373
2374 void operator()(oop obj) const {
2375 guarantee(obj->is_oop(),
2376 "Non-oop " PTR_FORMAT ", phase: %s, info: %d",
2377 p2i(obj), _phase, _info);
2378 guarantee(!_g1h->obj_in_cs(obj),
2379 "obj: " PTR_FORMAT " in CSet, phase: %s, info: %d",
2380 p2i(obj), _phase, _info);
2381 }
2382 };
2383
2384 void G1ConcurrentMark::verify_no_cset_oops() {
2385 assert(SafepointSynchronize::is_at_safepoint(), "should be at a safepoint");
2386 if (!G1CollectedHeap::heap()->collector_state()->mark_in_progress()) {
2387 return;
2388 }
2389
2390 // Verify entries on the global mark stack
2391 _markStack.iterate(VerifyNoCSetOops("Stack"));
2392
2393 // Verify entries on the task queues
2394 for (uint i = 0; i < _max_worker_id; ++i) {
2395 G1CMTaskQueue* queue = _task_queues->queue(i);
2396 queue->iterate(VerifyNoCSetOops("Queue", i));
2397 }
2398
2399 // Verify the global finger
2400 HeapWord* global_finger = finger();
2401 if (global_finger != NULL && global_finger < _heap_end) {
2402 // Since we always iterate over all regions, we might get a NULL HeapRegion
2403 // here.
2404 HeapRegion* global_hr = _g1h->heap_region_containing(global_finger);
2405 guarantee(global_hr == NULL || global_finger == global_hr->bottom(),
2406 "global finger: " PTR_FORMAT " region: " HR_FORMAT,
2407 p2i(global_finger), HR_FORMAT_PARAMS(global_hr));
2408 }
2409
2410 // Verify the task fingers
2411 assert(parallel_marking_threads() <= _max_worker_id, "sanity");
2412 for (uint i = 0; i < parallel_marking_threads(); ++i) {
2413 G1CMTask* task = _tasks[i];
2414 HeapWord* task_finger = task->finger();
2415 if (task_finger != NULL && task_finger < _heap_end) {
2416 // See above note on the global finger verification.
2417 HeapRegion* task_hr = _g1h->heap_region_containing(task_finger);
2418 guarantee(task_hr == NULL || task_finger == task_hr->bottom() ||
2419 !task_hr->in_collection_set(),
2420 "task finger: " PTR_FORMAT " region: " HR_FORMAT,
2421 p2i(task_finger), HR_FORMAT_PARAMS(task_hr));
2422 }
2423 }
2424 }
2425 #endif // PRODUCT
2426
2427 // Aggregate the counting data that was constructed concurrently
2428 // with marking.
2429 class AggregateCountDataHRClosure: public HeapRegionClosure {
2430 G1CollectedHeap* _g1h;
2431 G1ConcurrentMark* _cm;
2432 CardTableModRefBS* _ct_bs;
2433 BitMap* _cm_card_bm;
2434 uint _max_worker_id;
2435
2436 public:
2437 AggregateCountDataHRClosure(G1CollectedHeap* g1h,
2438 BitMap* cm_card_bm,
2439 uint max_worker_id) :
2440 _g1h(g1h), _cm(g1h->concurrent_mark()),
2441 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
2442 _cm_card_bm(cm_card_bm), _max_worker_id(max_worker_id) { }
2443
2444 bool doHeapRegion(HeapRegion* hr) {
2445 HeapWord* start = hr->bottom();
2446 HeapWord* limit = hr->next_top_at_mark_start();
2447 HeapWord* end = hr->end();
2448
2449 assert(start <= limit && limit <= hr->top() && hr->top() <= hr->end(),
2450 "Preconditions not met - "
2451 "start: " PTR_FORMAT ", limit: " PTR_FORMAT ", "
2452 "top: " PTR_FORMAT ", end: " PTR_FORMAT,
2453 p2i(start), p2i(limit), p2i(hr->top()), p2i(hr->end()));
2454
2455 assert(hr->next_marked_bytes() == 0, "Precondition");
2456
2457 if (start == limit) {
2458 // NTAMS of this region has not been set so nothing to do.
2459 return false;
2460 }
2461
2462 // 'start' should be in the heap.
2463 assert(_g1h->is_in_g1_reserved(start) && _ct_bs->is_card_aligned(start), "sanity");
2464 // 'end' *may* be just beyond the end of the heap (if hr is the last region)
2465 assert(!_g1h->is_in_g1_reserved(end) || _ct_bs->is_card_aligned(end), "sanity");
2466
2467 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
2468 BitMap::idx_t limit_idx = _cm->card_bitmap_index_for(limit);
2469 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(end);
2470
2471 // If ntams is not card aligned then we bump card bitmap index
2472 // for limit so that we get the all the cards spanned by
2473 // the object ending at ntams.
2474 // Note: if this is the last region in the heap then ntams
2475 // could be actually just beyond the end of the the heap;
2476 // limit_idx will then correspond to a (non-existent) card
2477 // that is also outside the heap.
2478 if (_g1h->is_in_g1_reserved(limit) && !_ct_bs->is_card_aligned(limit)) {
2479 limit_idx += 1;
2480 }
2481
2482 assert(limit_idx <= end_idx, "or else use atomics");
2483
2484 // Aggregate the "stripe" in the count data associated with hr.
2485 uint hrm_index = hr->hrm_index();
2486 size_t marked_bytes = 0;
2487
2488 for (uint i = 0; i < _max_worker_id; i += 1) {
2489 size_t* marked_bytes_array = _cm->count_marked_bytes_array_for(i);
2490 BitMap* task_card_bm = _cm->count_card_bitmap_for(i);
2491
2492 // Fetch the marked_bytes in this region for task i and
2493 // add it to the running total for this region.
2494 marked_bytes += marked_bytes_array[hrm_index];
2495
2496 // Now union the bitmaps[0,max_worker_id)[start_idx..limit_idx)
2497 // into the global card bitmap.
2498 BitMap::idx_t scan_idx = task_card_bm->get_next_one_offset(start_idx, limit_idx);
2499
2500 while (scan_idx < limit_idx) {
2501 assert(task_card_bm->at(scan_idx) == true, "should be");
2502 _cm_card_bm->set_bit(scan_idx);
2503 assert(_cm_card_bm->at(scan_idx) == true, "should be");
2504
2505 // BitMap::get_next_one_offset() can handle the case when
2506 // its left_offset parameter is greater than its right_offset
2507 // parameter. It does, however, have an early exit if
2508 // left_offset == right_offset. So let's limit the value
2509 // passed in for left offset here.
2510 BitMap::idx_t next_idx = MIN2(scan_idx + 1, limit_idx);
2511 scan_idx = task_card_bm->get_next_one_offset(next_idx, limit_idx);
2512 }
2513 }
2514
2515 // Update the marked bytes for this region.
2516 hr->add_to_marked_bytes(marked_bytes);
2517
2518 // Next heap region
2519 return false;
2520 }
2521 };
2522
2523 class G1AggregateCountDataTask: public AbstractGangTask {
2524 protected:
2525 G1CollectedHeap* _g1h;
2526 G1ConcurrentMark* _cm;
2527 BitMap* _cm_card_bm;
2528 uint _max_worker_id;
2529 uint _active_workers;
2530 HeapRegionClaimer _hrclaimer;
2531
2532 public:
2533 G1AggregateCountDataTask(G1CollectedHeap* g1h,
2534 G1ConcurrentMark* cm,
2535 BitMap* cm_card_bm,
2536 uint max_worker_id,
2537 uint n_workers) :
2538 AbstractGangTask("Count Aggregation"),
2539 _g1h(g1h), _cm(cm), _cm_card_bm(cm_card_bm),
2540 _max_worker_id(max_worker_id),
2541 _active_workers(n_workers),
2542 _hrclaimer(_active_workers) {
2543 }
2544
2545 void work(uint worker_id) {
2546 AggregateCountDataHRClosure cl(_g1h, _cm_card_bm, _max_worker_id);
2547
2548 _g1h->heap_region_par_iterate(&cl, worker_id, &_hrclaimer);
2549 }
2550 };
2551
2552
2553 void G1ConcurrentMark::aggregate_count_data() {
2554 uint n_workers = _g1h->workers()->active_workers();
2555
2556 G1AggregateCountDataTask g1_par_agg_task(_g1h, this, &_card_bm,
2557 _max_worker_id, n_workers);
2558
2559 _g1h->workers()->run_task(&g1_par_agg_task);
2560 }
2561
2562 // Clear the per-worker arrays used to store the per-region counting data
2563 void G1ConcurrentMark::clear_all_count_data() {
2564 // Clear the global card bitmap - it will be filled during
2565 // liveness count aggregation (during remark) and the
2566 // final counting task.
2567 _card_bm.clear();
2568
2569 // Clear the global region bitmap - it will be filled as part
2570 // of the final counting task.
2571 _region_bm.clear();
2572
2573 uint max_regions = _g1h->max_regions();
2574 assert(_max_worker_id > 0, "uninitialized");
2575
2576 for (uint i = 0; i < _max_worker_id; i += 1) {
2577 BitMap* task_card_bm = count_card_bitmap_for(i);
2578 size_t* marked_bytes_array = count_marked_bytes_array_for(i);
2579
2580 assert(task_card_bm->size() == _card_bm.size(), "size mismatch");
2581 assert(marked_bytes_array != NULL, "uninitialized");
2582
2583 memset(marked_bytes_array, 0, (size_t) max_regions * sizeof(size_t));
2584 task_card_bm->clear();
2585 }
2586 }
2587
2588 void G1ConcurrentMark::print_stats() {
2589 if (!log_is_enabled(Debug, gc, stats)) {
2590 return;
2591 }
2592 log_debug(gc, stats)("---------------------------------------------------------------------");
2593 for (size_t i = 0; i < _active_tasks; ++i) {
2594 _tasks[i]->print_stats();
2595 log_debug(gc, stats)("---------------------------------------------------------------------");
2596 }
2597 }
2598
2599 // abandon current marking iteration due to a Full GC
2600 void G1ConcurrentMark::abort() {
2601 if (!cmThread()->during_cycle() || _has_aborted) {
2602 // We haven't started a concurrent cycle or we have already aborted it. No need to do anything.
2603 return;
2604 }
2605
2606 // Clear all marks in the next bitmap for the next marking cycle. This will allow us to skip the next
2607 // concurrent bitmap clearing.
2608 _nextMarkBitMap->clearAll();
2609
2610 // Note we cannot clear the previous marking bitmap here
2611 // since VerifyDuringGC verifies the objects marked during
2612 // a full GC against the previous bitmap.
2613
2614 // Clear the liveness counting data
2615 clear_all_count_data();
2616 // Empty mark stack
2617 reset_marking_state();
2618 for (uint i = 0; i < _max_worker_id; ++i) {
2619 _tasks[i]->clear_region_fields();
2620 }
2621 _first_overflow_barrier_sync.abort();
2622 _second_overflow_barrier_sync.abort();
2623 _has_aborted = true;
2624
2625 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2626 satb_mq_set.abandon_partial_marking();
2627 // This can be called either during or outside marking, we'll read
2628 // the expected_active value from the SATB queue set.
2629 satb_mq_set.set_active_all_threads(
2630 false, /* new active value */
2631 satb_mq_set.is_active() /* expected_active */);
2632
2633 _g1h->trace_heap_after_concurrent_cycle();
2634
2635 _g1h->register_concurrent_cycle_end();
2636 }
2637
2638 static void print_ms_time_info(const char* prefix, const char* name,
2639 NumberSeq& ns) {
2640 log_trace(gc, marking)("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
2641 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
2642 if (ns.num() > 0) {
2643 log_trace(gc, marking)("%s [std. dev = %8.2f ms, max = %8.2f ms]",
2644 prefix, ns.sd(), ns.maximum());
2645 }
2646 }
2647
2648 void G1ConcurrentMark::print_summary_info() {
2649 LogHandle(gc, marking) log;
2650 if (!log.is_trace()) {
2651 return;
2652 }
2653
2654 log.trace(" Concurrent marking:");
2655 print_ms_time_info(" ", "init marks", _init_times);
2656 print_ms_time_info(" ", "remarks", _remark_times);
2657 {
2658 print_ms_time_info(" ", "final marks", _remark_mark_times);
2659 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
2660
2661 }
2662 print_ms_time_info(" ", "cleanups", _cleanup_times);
2663 log.trace(" Final counting total time = %8.2f s (avg = %8.2f ms).",
2664 _total_counting_time, (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 / (double)_cleanup_times.num() : 0.0));
2665 if (G1ScrubRemSets) {
2666 log.trace(" RS scrub total time = %8.2f s (avg = %8.2f ms).",
2667 _total_rs_scrub_time, (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 / (double)_cleanup_times.num() : 0.0));
2668 }
2669 log.trace(" Total stop_world time = %8.2f s.",
2670 (_init_times.sum() + _remark_times.sum() + _cleanup_times.sum())/1000.0);
2671 log.trace(" Total concurrent time = %8.2f s (%8.2f s marking).",
2672 cmThread()->vtime_accum(), cmThread()->vtime_mark_accum());
2673 }
2674
2675 void G1ConcurrentMark::print_worker_threads_on(outputStream* st) const {
2676 _parallel_workers->print_worker_threads_on(st);
2677 }
2678
2679 void G1ConcurrentMark::print_on_error(outputStream* st) const {
2680 st->print_cr("Marking Bits (Prev, Next): (CMBitMap*) " PTR_FORMAT ", (CMBitMap*) " PTR_FORMAT,
2681 p2i(_prevMarkBitMap), p2i(_nextMarkBitMap));
2682 _prevMarkBitMap->print_on_error(st, " Prev Bits: ");
2683 _nextMarkBitMap->print_on_error(st, " Next Bits: ");
2684 }
2685
2686 // We take a break if someone is trying to stop the world.
2687 bool G1ConcurrentMark::do_yield_check(uint worker_id) {
2688 if (SuspendibleThreadSet::should_yield()) {
2689 SuspendibleThreadSet::yield();
2690 return true;
2691 } else {
2692 return false;
2693 }
2694 }
2695
2696 // Closure for iteration over bitmaps
2697 class G1CMBitMapClosure : public BitMapClosure {
2698 private:
2699 // the bitmap that is being iterated over
2700 G1CMBitMap* _nextMarkBitMap;
2701 G1ConcurrentMark* _cm;
2702 G1CMTask* _task;
2703
2704 public:
2705 G1CMBitMapClosure(G1CMTask *task, G1ConcurrentMark* cm, G1CMBitMap* nextMarkBitMap) :
2706 _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { }
2707
2708 bool do_bit(size_t offset) {
2709 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset);
2710 assert(_nextMarkBitMap->isMarked(addr), "invariant");
2711 assert( addr < _cm->finger(), "invariant");
2712 assert(addr >= _task->finger(), "invariant");
2713
2714 // We move that task's local finger along.
2715 _task->move_finger_to(addr);
2716
2717 _task->scan_object(oop(addr));
2718 // we only partially drain the local queue and global stack
2719 _task->drain_local_queue(true);
2720 _task->drain_global_stack(true);
2721
2722 // if the has_aborted flag has been raised, we need to bail out of
2723 // the iteration
2724 return !_task->has_aborted();
2725 }
2726 };
2727
2728 static ReferenceProcessor* get_cm_oop_closure_ref_processor(G1CollectedHeap* g1h) {
2729 ReferenceProcessor* result = g1h->ref_processor_cm();
2730 assert(result != NULL, "CM reference processor should not be NULL");
2731 return result;
2732 }
2733
2734 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
2735 G1ConcurrentMark* cm,
2736 G1CMTask* task)
2737 : MetadataAwareOopClosure(get_cm_oop_closure_ref_processor(g1h)),
2738 _g1h(g1h), _cm(cm), _task(task)
2739 { }
2740
2741 void G1CMTask::setup_for_region(HeapRegion* hr) {
2742 assert(hr != NULL,
2743 "claim_region() should have filtered out NULL regions");
2744 _curr_region = hr;
2745 _finger = hr->bottom();
2746 update_region_limit();
2747 }
2748
2749 void G1CMTask::update_region_limit() {
2750 HeapRegion* hr = _curr_region;
2751 HeapWord* bottom = hr->bottom();
2752 HeapWord* limit = hr->next_top_at_mark_start();
2753
2754 if (limit == bottom) {
2755 // The region was collected underneath our feet.
2756 // We set the finger to bottom to ensure that the bitmap
2757 // iteration that will follow this will not do anything.
2758 // (this is not a condition that holds when we set the region up,
2759 // as the region is not supposed to be empty in the first place)
2760 _finger = bottom;
2761 } else if (limit >= _region_limit) {
2762 assert(limit >= _finger, "peace of mind");
2763 } else {
2764 assert(limit < _region_limit, "only way to get here");
2765 // This can happen under some pretty unusual circumstances. An
2766 // evacuation pause empties the region underneath our feet (NTAMS
2767 // at bottom). We then do some allocation in the region (NTAMS
2768 // stays at bottom), followed by the region being used as a GC
2769 // alloc region (NTAMS will move to top() and the objects
2770 // originally below it will be grayed). All objects now marked in
2771 // the region are explicitly grayed, if below the global finger,
2772 // and we do not need in fact to scan anything else. So, we simply
2773 // set _finger to be limit to ensure that the bitmap iteration
2774 // doesn't do anything.
2775 _finger = limit;
2776 }
2777
2778 _region_limit = limit;
2779 }
2780
2781 void G1CMTask::giveup_current_region() {
2782 assert(_curr_region != NULL, "invariant");
2783 clear_region_fields();
2784 }
2785
2786 void G1CMTask::clear_region_fields() {
2787 // Values for these three fields that indicate that we're not
2788 // holding on to a region.
2789 _curr_region = NULL;
2790 _finger = NULL;
2791 _region_limit = NULL;
2792 }
2793
2794 void G1CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
2795 if (cm_oop_closure == NULL) {
2796 assert(_cm_oop_closure != NULL, "invariant");
2797 } else {
2798 assert(_cm_oop_closure == NULL, "invariant");
2799 }
2800 _cm_oop_closure = cm_oop_closure;
2801 }
2802
2803 void G1CMTask::reset(G1CMBitMap* nextMarkBitMap) {
2804 guarantee(nextMarkBitMap != NULL, "invariant");
2805 _nextMarkBitMap = nextMarkBitMap;
2806 clear_region_fields();
2807
2808 _calls = 0;
2809 _elapsed_time_ms = 0.0;
2810 _termination_time_ms = 0.0;
2811 _termination_start_time_ms = 0.0;
2812 }
2813
2814 bool G1CMTask::should_exit_termination() {
2815 regular_clock_call();
2816 // This is called when we are in the termination protocol. We should
2817 // quit if, for some reason, this task wants to abort or the global
2818 // stack is not empty (this means that we can get work from it).
2819 return !_cm->mark_stack_empty() || has_aborted();
2820 }
2821
2822 void G1CMTask::reached_limit() {
2823 assert(_words_scanned >= _words_scanned_limit ||
2824 _refs_reached >= _refs_reached_limit ,
2825 "shouldn't have been called otherwise");
2826 regular_clock_call();
2827 }
2828
2829 void G1CMTask::regular_clock_call() {
2830 if (has_aborted()) return;
2831
2832 // First, we need to recalculate the words scanned and refs reached
2833 // limits for the next clock call.
2834 recalculate_limits();
2835
2836 // During the regular clock call we do the following
2837
2838 // (1) If an overflow has been flagged, then we abort.
2839 if (_cm->has_overflown()) {
2840 set_has_aborted();
2841 return;
2842 }
2843
2844 // If we are not concurrent (i.e. we're doing remark) we don't need
2845 // to check anything else. The other steps are only needed during
2846 // the concurrent marking phase.
2847 if (!concurrent()) return;
2848
2849 // (2) If marking has been aborted for Full GC, then we also abort.
2850 if (_cm->has_aborted()) {
2851 set_has_aborted();
2852 return;
2853 }
2854
2855 double curr_time_ms = os::elapsedVTime() * 1000.0;
2856
2857 // (4) We check whether we should yield. If we have to, then we abort.
2858 if (SuspendibleThreadSet::should_yield()) {
2859 // We should yield. To do this we abort the task. The caller is
2860 // responsible for yielding.
2861 set_has_aborted();
2862 return;
2863 }
2864
2865 // (5) We check whether we've reached our time quota. If we have,
2866 // then we abort.
2867 double elapsed_time_ms = curr_time_ms - _start_time_ms;
2868 if (elapsed_time_ms > _time_target_ms) {
2869 set_has_aborted();
2870 _has_timed_out = true;
2871 return;
2872 }
2873
2874 // (6) Finally, we check whether there are enough completed STAB
2875 // buffers available for processing. If there are, we abort.
2876 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2877 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
2878 // we do need to process SATB buffers, we'll abort and restart
2879 // the marking task to do so
2880 set_has_aborted();
2881 return;
2882 }
2883 }
2884
2885 void G1CMTask::recalculate_limits() {
2886 _real_words_scanned_limit = _words_scanned + words_scanned_period;
2887 _words_scanned_limit = _real_words_scanned_limit;
2888
2889 _real_refs_reached_limit = _refs_reached + refs_reached_period;
2890 _refs_reached_limit = _real_refs_reached_limit;
2891 }
2892
2893 void G1CMTask::decrease_limits() {
2894 // This is called when we believe that we're going to do an infrequent
2895 // operation which will increase the per byte scanned cost (i.e. move
2896 // entries to/from the global stack). It basically tries to decrease the
2897 // scanning limit so that the clock is called earlier.
2898
2899 _words_scanned_limit = _real_words_scanned_limit -
2900 3 * words_scanned_period / 4;
2901 _refs_reached_limit = _real_refs_reached_limit -
2902 3 * refs_reached_period / 4;
2903 }
2904
2905 void G1CMTask::move_entries_to_global_stack() {
2906 // local array where we'll store the entries that will be popped
2907 // from the local queue
2908 oop buffer[global_stack_transfer_size];
2909
2910 int n = 0;
2911 oop obj;
2912 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) {
2913 buffer[n] = obj;
2914 ++n;
2915 }
2916
2917 if (n > 0) {
2918 // we popped at least one entry from the local queue
2919
2920 if (!_cm->mark_stack_push(buffer, n)) {
2921 set_has_aborted();
2922 }
2923 }
2924
2925 // this operation was quite expensive, so decrease the limits
2926 decrease_limits();
2927 }
2928
2929 void G1CMTask::get_entries_from_global_stack() {
2930 // local array where we'll store the entries that will be popped
2931 // from the global stack.
2932 oop buffer[global_stack_transfer_size];
2933 int n;
2934 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n);
2935 assert(n <= global_stack_transfer_size,
2936 "we should not pop more than the given limit");
2937 if (n > 0) {
2938 // yes, we did actually pop at least one entry
2939 for (int i = 0; i < n; ++i) {
2940 bool success = _task_queue->push(buffer[i]);
2941 // We only call this when the local queue is empty or under a
2942 // given target limit. So, we do not expect this push to fail.
2943 assert(success, "invariant");
2944 }
2945 }
2946
2947 // this operation was quite expensive, so decrease the limits
2948 decrease_limits();
2949 }
2950
2951 void G1CMTask::drain_local_queue(bool partially) {
2952 if (has_aborted()) return;
2953
2954 // Decide what the target size is, depending whether we're going to
2955 // drain it partially (so that other tasks can steal if they run out
2956 // of things to do) or totally (at the very end).
2957 size_t target_size;
2958 if (partially) {
2959 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
2960 } else {
2961 target_size = 0;
2962 }
2963
2964 if (_task_queue->size() > target_size) {
2965 oop obj;
2966 bool ret = _task_queue->pop_local(obj);
2967 while (ret) {
2968 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" );
2969 assert(!_g1h->is_on_master_free_list(
2970 _g1h->heap_region_containing((HeapWord*) obj)), "invariant");
2971
2972 scan_object(obj);
2973
2974 if (_task_queue->size() <= target_size || has_aborted()) {
2975 ret = false;
2976 } else {
2977 ret = _task_queue->pop_local(obj);
2978 }
2979 }
2980 }
2981 }
2982
2983 void G1CMTask::drain_global_stack(bool partially) {
2984 if (has_aborted()) return;
2985
2986 // We have a policy to drain the local queue before we attempt to
2987 // drain the global stack.
2988 assert(partially || _task_queue->size() == 0, "invariant");
2989
2990 // Decide what the target size is, depending whether we're going to
2991 // drain it partially (so that other tasks can steal if they run out
2992 // of things to do) or totally (at the very end). Notice that,
2993 // because we move entries from the global stack in chunks or
2994 // because another task might be doing the same, we might in fact
2995 // drop below the target. But, this is not a problem.
2996 size_t target_size;
2997 if (partially) {
2998 target_size = _cm->partial_mark_stack_size_target();
2999 } else {
3000 target_size = 0;
3001 }
3002
3003 if (_cm->mark_stack_size() > target_size) {
3004 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
3005 get_entries_from_global_stack();
3006 drain_local_queue(partially);
3007 }
3008 }
3009 }
3010
3011 // SATB Queue has several assumptions on whether to call the par or
3012 // non-par versions of the methods. this is why some of the code is
3013 // replicated. We should really get rid of the single-threaded version
3014 // of the code to simplify things.
3015 void G1CMTask::drain_satb_buffers() {
3016 if (has_aborted()) return;
3017
3018 // We set this so that the regular clock knows that we're in the
3019 // middle of draining buffers and doesn't set the abort flag when it
3020 // notices that SATB buffers are available for draining. It'd be
3021 // very counter productive if it did that. :-)
3022 _draining_satb_buffers = true;
3023
3024 G1CMSATBBufferClosure satb_cl(this, _g1h);
3025 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3026
3027 // This keeps claiming and applying the closure to completed buffers
3028 // until we run out of buffers or we need to abort.
3029 while (!has_aborted() &&
3030 satb_mq_set.apply_closure_to_completed_buffer(&satb_cl)) {
3031 regular_clock_call();
3032 }
3033
3034 _draining_satb_buffers = false;
3035
3036 assert(has_aborted() ||
3037 concurrent() ||
3038 satb_mq_set.completed_buffers_num() == 0, "invariant");
3039
3040 // again, this was a potentially expensive operation, decrease the
3041 // limits to get the regular clock call early
3042 decrease_limits();
3043 }
3044
3045 void G1CMTask::print_stats() {
3046 log_debug(gc, stats)("Marking Stats, task = %u, calls = %d",
3047 _worker_id, _calls);
3048 log_debug(gc, stats)(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
3049 _elapsed_time_ms, _termination_time_ms);
3050 log_debug(gc, stats)(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3051 _step_times_ms.num(), _step_times_ms.avg(),
3052 _step_times_ms.sd());
3053 log_debug(gc, stats)(" max = %1.2lfms, total = %1.2lfms",
3054 _step_times_ms.maximum(), _step_times_ms.sum());
3055 }
3056
3057 bool G1ConcurrentMark::try_stealing(uint worker_id, int* hash_seed, oop& obj) {
3058 return _task_queues->steal(worker_id, hash_seed, obj);
3059 }
3060
3061 /*****************************************************************************
3062
3063 The do_marking_step(time_target_ms, ...) method is the building
3064 block of the parallel marking framework. It can be called in parallel
3065 with other invocations of do_marking_step() on different tasks
3066 (but only one per task, obviously) and concurrently with the
3067 mutator threads, or during remark, hence it eliminates the need
3068 for two versions of the code. When called during remark, it will
3069 pick up from where the task left off during the concurrent marking
3070 phase. Interestingly, tasks are also claimable during evacuation
3071 pauses too, since do_marking_step() ensures that it aborts before
3072 it needs to yield.
3073
3074 The data structures that it uses to do marking work are the
3075 following:
3076
3077 (1) Marking Bitmap. If there are gray objects that appear only
3078 on the bitmap (this happens either when dealing with an overflow
3079 or when the initial marking phase has simply marked the roots
3080 and didn't push them on the stack), then tasks claim heap
3081 regions whose bitmap they then scan to find gray objects. A
3082 global finger indicates where the end of the last claimed region
3083 is. A local finger indicates how far into the region a task has
3084 scanned. The two fingers are used to determine how to gray an
3085 object (i.e. whether simply marking it is OK, as it will be
3086 visited by a task in the future, or whether it needs to be also
3087 pushed on a stack).
3088
3089 (2) Local Queue. The local queue of the task which is accessed
3090 reasonably efficiently by the task. Other tasks can steal from
3091 it when they run out of work. Throughout the marking phase, a
3092 task attempts to keep its local queue short but not totally
3093 empty, so that entries are available for stealing by other
3094 tasks. Only when there is no more work, a task will totally
3095 drain its local queue.
3096
3097 (3) Global Mark Stack. This handles local queue overflow. During
3098 marking only sets of entries are moved between it and the local
3099 queues, as access to it requires a mutex and more fine-grain
3100 interaction with it which might cause contention. If it
3101 overflows, then the marking phase should restart and iterate
3102 over the bitmap to identify gray objects. Throughout the marking
3103 phase, tasks attempt to keep the global mark stack at a small
3104 length but not totally empty, so that entries are available for
3105 popping by other tasks. Only when there is no more work, tasks
3106 will totally drain the global mark stack.
3107
3108 (4) SATB Buffer Queue. This is where completed SATB buffers are
3109 made available. Buffers are regularly removed from this queue
3110 and scanned for roots, so that the queue doesn't get too
3111 long. During remark, all completed buffers are processed, as
3112 well as the filled in parts of any uncompleted buffers.
3113
3114 The do_marking_step() method tries to abort when the time target
3115 has been reached. There are a few other cases when the
3116 do_marking_step() method also aborts:
3117
3118 (1) When the marking phase has been aborted (after a Full GC).
3119
3120 (2) When a global overflow (on the global stack) has been
3121 triggered. Before the task aborts, it will actually sync up with
3122 the other tasks to ensure that all the marking data structures
3123 (local queues, stacks, fingers etc.) are re-initialized so that
3124 when do_marking_step() completes, the marking phase can
3125 immediately restart.
3126
3127 (3) When enough completed SATB buffers are available. The
3128 do_marking_step() method only tries to drain SATB buffers right
3129 at the beginning. So, if enough buffers are available, the
3130 marking step aborts and the SATB buffers are processed at
3131 the beginning of the next invocation.
3132
3133 (4) To yield. when we have to yield then we abort and yield
3134 right at the end of do_marking_step(). This saves us from a lot
3135 of hassle as, by yielding we might allow a Full GC. If this
3136 happens then objects will be compacted underneath our feet, the
3137 heap might shrink, etc. We save checking for this by just
3138 aborting and doing the yield right at the end.
3139
3140 From the above it follows that the do_marking_step() method should
3141 be called in a loop (or, otherwise, regularly) until it completes.
3142
3143 If a marking step completes without its has_aborted() flag being
3144 true, it means it has completed the current marking phase (and
3145 also all other marking tasks have done so and have all synced up).
3146
3147 A method called regular_clock_call() is invoked "regularly" (in
3148 sub ms intervals) throughout marking. It is this clock method that
3149 checks all the abort conditions which were mentioned above and
3150 decides when the task should abort. A work-based scheme is used to
3151 trigger this clock method: when the number of object words the
3152 marking phase has scanned or the number of references the marking
3153 phase has visited reach a given limit. Additional invocations to
3154 the method clock have been planted in a few other strategic places
3155 too. The initial reason for the clock method was to avoid calling
3156 vtime too regularly, as it is quite expensive. So, once it was in
3157 place, it was natural to piggy-back all the other conditions on it
3158 too and not constantly check them throughout the code.
3159
3160 If do_termination is true then do_marking_step will enter its
3161 termination protocol.
3162
3163 The value of is_serial must be true when do_marking_step is being
3164 called serially (i.e. by the VMThread) and do_marking_step should
3165 skip any synchronization in the termination and overflow code.
3166 Examples include the serial remark code and the serial reference
3167 processing closures.
3168
3169 The value of is_serial must be false when do_marking_step is
3170 being called by any of the worker threads in a work gang.
3171 Examples include the concurrent marking code (CMMarkingTask),
3172 the MT remark code, and the MT reference processing closures.
3173
3174 *****************************************************************************/
3175
3176 void G1CMTask::do_marking_step(double time_target_ms,
3177 bool do_termination,
3178 bool is_serial) {
3179 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
3180 assert(concurrent() == _cm->concurrent(), "they should be the same");
3181
3182 G1CollectorPolicy* g1_policy = _g1h->g1_policy();
3183 assert(_task_queues != NULL, "invariant");
3184 assert(_task_queue != NULL, "invariant");
3185 assert(_task_queues->queue(_worker_id) == _task_queue, "invariant");
3186
3187 assert(!_claimed,
3188 "only one thread should claim this task at any one time");
3189
3190 // OK, this doesn't safeguard again all possible scenarios, as it is
3191 // possible for two threads to set the _claimed flag at the same
3192 // time. But it is only for debugging purposes anyway and it will
3193 // catch most problems.
3194 _claimed = true;
3195
3196 _start_time_ms = os::elapsedVTime() * 1000.0;
3197
3198 // If do_stealing is true then do_marking_step will attempt to
3199 // steal work from the other G1CMTasks. It only makes sense to
3200 // enable stealing when the termination protocol is enabled
3201 // and do_marking_step() is not being called serially.
3202 bool do_stealing = do_termination && !is_serial;
3203
3204 double diff_prediction_ms = _g1h->g1_policy()->predictor().get_new_prediction(&_marking_step_diffs_ms);
3205 _time_target_ms = time_target_ms - diff_prediction_ms;
3206
3207 // set up the variables that are used in the work-based scheme to
3208 // call the regular clock method
3209 _words_scanned = 0;
3210 _refs_reached = 0;
3211 recalculate_limits();
3212
3213 // clear all flags
3214 clear_has_aborted();
3215 _has_timed_out = false;
3216 _draining_satb_buffers = false;
3217
3218 ++_calls;
3219
3220 // Set up the bitmap and oop closures. Anything that uses them is
3221 // eventually called from this method, so it is OK to allocate these
3222 // statically.
3223 G1CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap);
3224 G1CMOopClosure cm_oop_closure(_g1h, _cm, this);
3225 set_cm_oop_closure(&cm_oop_closure);
3226
3227 if (_cm->has_overflown()) {
3228 // This can happen if the mark stack overflows during a GC pause
3229 // and this task, after a yield point, restarts. We have to abort
3230 // as we need to get into the overflow protocol which happens
3231 // right at the end of this task.
3232 set_has_aborted();
3233 }
3234
3235 // First drain any available SATB buffers. After this, we will not
3236 // look at SATB buffers before the next invocation of this method.
3237 // If enough completed SATB buffers are queued up, the regular clock
3238 // will abort this task so that it restarts.
3239 drain_satb_buffers();
3240 // ...then partially drain the local queue and the global stack
3241 drain_local_queue(true);
3242 drain_global_stack(true);
3243
3244 do {
3245 if (!has_aborted() && _curr_region != NULL) {
3246 // This means that we're already holding on to a region.
3247 assert(_finger != NULL, "if region is not NULL, then the finger "
3248 "should not be NULL either");
3249
3250 // We might have restarted this task after an evacuation pause
3251 // which might have evacuated the region we're holding on to
3252 // underneath our feet. Let's read its limit again to make sure
3253 // that we do not iterate over a region of the heap that
3254 // contains garbage (update_region_limit() will also move
3255 // _finger to the start of the region if it is found empty).
3256 update_region_limit();
3257 // We will start from _finger not from the start of the region,
3258 // as we might be restarting this task after aborting half-way
3259 // through scanning this region. In this case, _finger points to
3260 // the address where we last found a marked object. If this is a
3261 // fresh region, _finger points to start().
3262 MemRegion mr = MemRegion(_finger, _region_limit);
3263
3264 assert(!_curr_region->is_humongous() || mr.start() == _curr_region->bottom(),
3265 "humongous regions should go around loop once only");
3266
3267 // Some special cases:
3268 // If the memory region is empty, we can just give up the region.
3269 // If the current region is humongous then we only need to check
3270 // the bitmap for the bit associated with the start of the object,
3271 // scan the object if it's live, and give up the region.
3272 // Otherwise, let's iterate over the bitmap of the part of the region
3273 // that is left.
3274 // If the iteration is successful, give up the region.
3275 if (mr.is_empty()) {
3276 giveup_current_region();
3277 regular_clock_call();
3278 } else if (_curr_region->is_humongous() && mr.start() == _curr_region->bottom()) {
3279 if (_nextMarkBitMap->isMarked(mr.start())) {
3280 // The object is marked - apply the closure
3281 BitMap::idx_t offset = _nextMarkBitMap->heapWordToOffset(mr.start());
3282 bitmap_closure.do_bit(offset);
3283 }
3284 // Even if this task aborted while scanning the humongous object
3285 // we can (and should) give up the current region.
3286 giveup_current_region();
3287 regular_clock_call();
3288 } else if (_nextMarkBitMap->iterate(&bitmap_closure, mr)) {
3289 giveup_current_region();
3290 regular_clock_call();
3291 } else {
3292 assert(has_aborted(), "currently the only way to do so");
3293 // The only way to abort the bitmap iteration is to return
3294 // false from the do_bit() method. However, inside the
3295 // do_bit() method we move the _finger to point to the
3296 // object currently being looked at. So, if we bail out, we
3297 // have definitely set _finger to something non-null.
3298 assert(_finger != NULL, "invariant");
3299
3300 // Region iteration was actually aborted. So now _finger
3301 // points to the address of the object we last scanned. If we
3302 // leave it there, when we restart this task, we will rescan
3303 // the object. It is easy to avoid this. We move the finger by
3304 // enough to point to the next possible object header (the
3305 // bitmap knows by how much we need to move it as it knows its
3306 // granularity).
3307 assert(_finger < _region_limit, "invariant");
3308 HeapWord* new_finger = _nextMarkBitMap->nextObject(_finger);
3309 // Check if bitmap iteration was aborted while scanning the last object
3310 if (new_finger >= _region_limit) {
3311 giveup_current_region();
3312 } else {
3313 move_finger_to(new_finger);
3314 }
3315 }
3316 }
3317 // At this point we have either completed iterating over the
3318 // region we were holding on to, or we have aborted.
3319
3320 // We then partially drain the local queue and the global stack.
3321 // (Do we really need this?)
3322 drain_local_queue(true);
3323 drain_global_stack(true);
3324
3325 // Read the note on the claim_region() method on why it might
3326 // return NULL with potentially more regions available for
3327 // claiming and why we have to check out_of_regions() to determine
3328 // whether we're done or not.
3329 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
3330 // We are going to try to claim a new region. We should have
3331 // given up on the previous one.
3332 // Separated the asserts so that we know which one fires.
3333 assert(_curr_region == NULL, "invariant");
3334 assert(_finger == NULL, "invariant");
3335 assert(_region_limit == NULL, "invariant");
3336 HeapRegion* claimed_region = _cm->claim_region(_worker_id);
3337 if (claimed_region != NULL) {
3338 // Yes, we managed to claim one
3339 setup_for_region(claimed_region);
3340 assert(_curr_region == claimed_region, "invariant");
3341 }
3342 // It is important to call the regular clock here. It might take
3343 // a while to claim a region if, for example, we hit a large
3344 // block of empty regions. So we need to call the regular clock
3345 // method once round the loop to make sure it's called
3346 // frequently enough.
3347 regular_clock_call();
3348 }
3349
3350 if (!has_aborted() && _curr_region == NULL) {
3351 assert(_cm->out_of_regions(),
3352 "at this point we should be out of regions");
3353 }
3354 } while ( _curr_region != NULL && !has_aborted());
3355
3356 if (!has_aborted()) {
3357 // We cannot check whether the global stack is empty, since other
3358 // tasks might be pushing objects to it concurrently.
3359 assert(_cm->out_of_regions(),
3360 "at this point we should be out of regions");
3361 // Try to reduce the number of available SATB buffers so that
3362 // remark has less work to do.
3363 drain_satb_buffers();
3364 }
3365
3366 // Since we've done everything else, we can now totally drain the
3367 // local queue and global stack.
3368 drain_local_queue(false);
3369 drain_global_stack(false);
3370
3371 // Attempt at work stealing from other task's queues.
3372 if (do_stealing && !has_aborted()) {
3373 // We have not aborted. This means that we have finished all that
3374 // we could. Let's try to do some stealing...
3375
3376 // We cannot check whether the global stack is empty, since other
3377 // tasks might be pushing objects to it concurrently.
3378 assert(_cm->out_of_regions() && _task_queue->size() == 0,
3379 "only way to reach here");
3380 while (!has_aborted()) {
3381 oop obj;
3382 if (_cm->try_stealing(_worker_id, &_hash_seed, obj)) {
3383 assert(_nextMarkBitMap->isMarked((HeapWord*) obj),
3384 "any stolen object should be marked");
3385 scan_object(obj);
3386
3387 // And since we're towards the end, let's totally drain the
3388 // local queue and global stack.
3389 drain_local_queue(false);
3390 drain_global_stack(false);
3391 } else {
3392 break;
3393 }
3394 }
3395 }
3396
3397 // We still haven't aborted. Now, let's try to get into the
3398 // termination protocol.
3399 if (do_termination && !has_aborted()) {
3400 // We cannot check whether the global stack is empty, since other
3401 // tasks might be concurrently pushing objects on it.
3402 // Separated the asserts so that we know which one fires.
3403 assert(_cm->out_of_regions(), "only way to reach here");
3404 assert(_task_queue->size() == 0, "only way to reach here");
3405 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
3406
3407 // The G1CMTask class also extends the TerminatorTerminator class,
3408 // hence its should_exit_termination() method will also decide
3409 // whether to exit the termination protocol or not.
3410 bool finished = (is_serial ||
3411 _cm->terminator()->offer_termination(this));
3412 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
3413 _termination_time_ms +=
3414 termination_end_time_ms - _termination_start_time_ms;
3415
3416 if (finished) {
3417 // We're all done.
3418
3419 if (_worker_id == 0) {
3420 // let's allow task 0 to do this
3421 if (concurrent()) {
3422 assert(_cm->concurrent_marking_in_progress(), "invariant");
3423 // we need to set this to false before the next
3424 // safepoint. This way we ensure that the marking phase
3425 // doesn't observe any more heap expansions.
3426 _cm->clear_concurrent_marking_in_progress();
3427 }
3428 }
3429
3430 // We can now guarantee that the global stack is empty, since
3431 // all other tasks have finished. We separated the guarantees so
3432 // that, if a condition is false, we can immediately find out
3433 // which one.
3434 guarantee(_cm->out_of_regions(), "only way to reach here");
3435 guarantee(_cm->mark_stack_empty(), "only way to reach here");
3436 guarantee(_task_queue->size() == 0, "only way to reach here");
3437 guarantee(!_cm->has_overflown(), "only way to reach here");
3438 guarantee(!_cm->mark_stack_overflow(), "only way to reach here");
3439 } else {
3440 // Apparently there's more work to do. Let's abort this task. It
3441 // will restart it and we can hopefully find more things to do.
3442 set_has_aborted();
3443 }
3444 }
3445
3446 // Mainly for debugging purposes to make sure that a pointer to the
3447 // closure which was statically allocated in this frame doesn't
3448 // escape it by accident.
3449 set_cm_oop_closure(NULL);
3450 double end_time_ms = os::elapsedVTime() * 1000.0;
3451 double elapsed_time_ms = end_time_ms - _start_time_ms;
3452 // Update the step history.
3453 _step_times_ms.add(elapsed_time_ms);
3454
3455 if (has_aborted()) {
3456 // The task was aborted for some reason.
3457 if (_has_timed_out) {
3458 double diff_ms = elapsed_time_ms - _time_target_ms;
3459 // Keep statistics of how well we did with respect to hitting
3460 // our target only if we actually timed out (if we aborted for
3461 // other reasons, then the results might get skewed).
3462 _marking_step_diffs_ms.add(diff_ms);
3463 }
3464
3465 if (_cm->has_overflown()) {
3466 // This is the interesting one. We aborted because a global
3467 // overflow was raised. This means we have to restart the
3468 // marking phase and start iterating over regions. However, in
3469 // order to do this we have to make sure that all tasks stop
3470 // what they are doing and re-initialize in a safe manner. We
3471 // will achieve this with the use of two barrier sync points.
3472
3473 if (!is_serial) {
3474 // We only need to enter the sync barrier if being called
3475 // from a parallel context
3476 _cm->enter_first_sync_barrier(_worker_id);
3477
3478 // When we exit this sync barrier we know that all tasks have
3479 // stopped doing marking work. So, it's now safe to
3480 // re-initialize our data structures. At the end of this method,
3481 // task 0 will clear the global data structures.
3482 }
3483
3484 // We clear the local state of this task...
3485 clear_region_fields();
3486
3487 if (!is_serial) {
3488 // ...and enter the second barrier.
3489 _cm->enter_second_sync_barrier(_worker_id);
3490 }
3491 // At this point, if we're during the concurrent phase of
3492 // marking, everything has been re-initialized and we're
3493 // ready to restart.
3494 }
3495 }
3496
3497 _claimed = false;
3498 }
3499
3500 G1CMTask::G1CMTask(uint worker_id,
3501 G1ConcurrentMark* cm,
3502 size_t* marked_bytes,
3503 BitMap* card_bm,
3504 G1CMTaskQueue* task_queue,
3505 G1CMTaskQueueSet* task_queues)
3506 : _g1h(G1CollectedHeap::heap()),
3507 _worker_id(worker_id), _cm(cm),
3508 _claimed(false),
3509 _nextMarkBitMap(NULL), _hash_seed(17),
3510 _task_queue(task_queue),
3511 _task_queues(task_queues),
3512 _cm_oop_closure(NULL),
3513 _marked_bytes_array(marked_bytes),
3514 _card_bm(card_bm) {
3515 guarantee(task_queue != NULL, "invariant");
3516 guarantee(task_queues != NULL, "invariant");
3517
3518 _marking_step_diffs_ms.add(0.5);
3519 }
3520
3521 // These are formatting macros that are used below to ensure
3522 // consistent formatting. The *_H_* versions are used to format the
3523 // header for a particular value and they should be kept consistent
3524 // with the corresponding macro. Also note that most of the macros add
3525 // the necessary white space (as a prefix) which makes them a bit
3526 // easier to compose.
3527
3528 // All the output lines are prefixed with this string to be able to
3529 // identify them easily in a large log file.
3530 #define G1PPRL_LINE_PREFIX "###"
3531
3532 #define G1PPRL_ADDR_BASE_FORMAT " " PTR_FORMAT "-" PTR_FORMAT
3533 #ifdef _LP64
3534 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
3535 #else // _LP64
3536 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
3537 #endif // _LP64
3538
3539 // For per-region info
3540 #define G1PPRL_TYPE_FORMAT " %-4s"
3541 #define G1PPRL_TYPE_H_FORMAT " %4s"
3542 #define G1PPRL_BYTE_FORMAT " " SIZE_FORMAT_W(9)
3543 #define G1PPRL_BYTE_H_FORMAT " %9s"
3544 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
3545 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
3546
3547 // For summary info
3548 #define G1PPRL_SUM_ADDR_FORMAT(tag) " " tag ":" G1PPRL_ADDR_BASE_FORMAT
3549 #define G1PPRL_SUM_BYTE_FORMAT(tag) " " tag ": " SIZE_FORMAT
3550 #define G1PPRL_SUM_MB_FORMAT(tag) " " tag ": %1.2f MB"
3551 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag) " / %1.2f %%"
3552
3553 G1PrintRegionLivenessInfoClosure::
3554 G1PrintRegionLivenessInfoClosure(const char* phase_name)
3555 : _total_used_bytes(0), _total_capacity_bytes(0),
3556 _total_prev_live_bytes(0), _total_next_live_bytes(0),
3557 _hum_used_bytes(0), _hum_capacity_bytes(0),
3558 _hum_prev_live_bytes(0), _hum_next_live_bytes(0),
3559 _total_remset_bytes(0), _total_strong_code_roots_bytes(0) {
3560 G1CollectedHeap* g1h = G1CollectedHeap::heap();
3561 MemRegion g1_reserved = g1h->g1_reserved();
3562 double now = os::elapsedTime();
3563
3564 // Print the header of the output.
3565 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
3566 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" HEAP"
3567 G1PPRL_SUM_ADDR_FORMAT("reserved")
3568 G1PPRL_SUM_BYTE_FORMAT("region-size"),
3569 p2i(g1_reserved.start()), p2i(g1_reserved.end()),
3570 HeapRegion::GrainBytes);
3571 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
3572 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3573 G1PPRL_TYPE_H_FORMAT
3574 G1PPRL_ADDR_BASE_H_FORMAT
3575 G1PPRL_BYTE_H_FORMAT
3576 G1PPRL_BYTE_H_FORMAT
3577 G1PPRL_BYTE_H_FORMAT
3578 G1PPRL_DOUBLE_H_FORMAT
3579 G1PPRL_BYTE_H_FORMAT
3580 G1PPRL_BYTE_H_FORMAT,
3581 "type", "address-range",
3582 "used", "prev-live", "next-live", "gc-eff",
3583 "remset", "code-roots");
3584 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3585 G1PPRL_TYPE_H_FORMAT
3586 G1PPRL_ADDR_BASE_H_FORMAT
3587 G1PPRL_BYTE_H_FORMAT
3588 G1PPRL_BYTE_H_FORMAT
3589 G1PPRL_BYTE_H_FORMAT
3590 G1PPRL_DOUBLE_H_FORMAT
3591 G1PPRL_BYTE_H_FORMAT
3592 G1PPRL_BYTE_H_FORMAT,
3593 "", "",
3594 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)",
3595 "(bytes)", "(bytes)");
3596 }
3597
3598 // It takes as a parameter a reference to one of the _hum_* fields, it
3599 // deduces the corresponding value for a region in a humongous region
3600 // series (either the region size, or what's left if the _hum_* field
3601 // is < the region size), and updates the _hum_* field accordingly.
3602 size_t G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* hum_bytes) {
3603 size_t bytes = 0;
3604 // The > 0 check is to deal with the prev and next live bytes which
3605 // could be 0.
3606 if (*hum_bytes > 0) {
3607 bytes = MIN2(HeapRegion::GrainBytes, *hum_bytes);
3608 *hum_bytes -= bytes;
3609 }
3610 return bytes;
3611 }
3612
3613 // It deduces the values for a region in a humongous region series
3614 // from the _hum_* fields and updates those accordingly. It assumes
3615 // that that _hum_* fields have already been set up from the "starts
3616 // humongous" region and we visit the regions in address order.
3617 void G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* used_bytes,
3618 size_t* capacity_bytes,
3619 size_t* prev_live_bytes,
3620 size_t* next_live_bytes) {
3621 assert(_hum_used_bytes > 0 && _hum_capacity_bytes > 0, "pre-condition");
3622 *used_bytes = get_hum_bytes(&_hum_used_bytes);
3623 *capacity_bytes = get_hum_bytes(&_hum_capacity_bytes);
3624 *prev_live_bytes = get_hum_bytes(&_hum_prev_live_bytes);
3625 *next_live_bytes = get_hum_bytes(&_hum_next_live_bytes);
3626 }
3627
3628 bool G1PrintRegionLivenessInfoClosure::doHeapRegion(HeapRegion* r) {
3629 const char* type = r->get_type_str();
3630 HeapWord* bottom = r->bottom();
3631 HeapWord* end = r->end();
3632 size_t capacity_bytes = r->capacity();
3633 size_t used_bytes = r->used();
3634 size_t prev_live_bytes = r->live_bytes();
3635 size_t next_live_bytes = r->next_live_bytes();
3636 double gc_eff = r->gc_efficiency();
3637 size_t remset_bytes = r->rem_set()->mem_size();
3638 size_t strong_code_roots_bytes = r->rem_set()->strong_code_roots_mem_size();
3639
3640 if (r->is_starts_humongous()) {
3641 assert(_hum_used_bytes == 0 && _hum_capacity_bytes == 0 &&
3642 _hum_prev_live_bytes == 0 && _hum_next_live_bytes == 0,
3643 "they should have been zeroed after the last time we used them");
3644 // Set up the _hum_* fields.
3645 _hum_capacity_bytes = capacity_bytes;
3646 _hum_used_bytes = used_bytes;
3647 _hum_prev_live_bytes = prev_live_bytes;
3648 _hum_next_live_bytes = next_live_bytes;
3649 get_hum_bytes(&used_bytes, &capacity_bytes,
3650 &prev_live_bytes, &next_live_bytes);
3651 end = bottom + HeapRegion::GrainWords;
3652 } else if (r->is_continues_humongous()) {
3653 get_hum_bytes(&used_bytes, &capacity_bytes,
3654 &prev_live_bytes, &next_live_bytes);
3655 assert(end == bottom + HeapRegion::GrainWords, "invariant");
3656 }
3657
3658 _total_used_bytes += used_bytes;
3659 _total_capacity_bytes += capacity_bytes;
3660 _total_prev_live_bytes += prev_live_bytes;
3661 _total_next_live_bytes += next_live_bytes;
3662 _total_remset_bytes += remset_bytes;
3663 _total_strong_code_roots_bytes += strong_code_roots_bytes;
3664
3665 // Print a line for this particular region.
3666 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3667 G1PPRL_TYPE_FORMAT
3668 G1PPRL_ADDR_BASE_FORMAT
3669 G1PPRL_BYTE_FORMAT
3670 G1PPRL_BYTE_FORMAT
3671 G1PPRL_BYTE_FORMAT
3672 G1PPRL_DOUBLE_FORMAT
3673 G1PPRL_BYTE_FORMAT
3674 G1PPRL_BYTE_FORMAT,
3675 type, p2i(bottom), p2i(end),
3676 used_bytes, prev_live_bytes, next_live_bytes, gc_eff,
3677 remset_bytes, strong_code_roots_bytes);
3678
3679 return false;
3680 }
3681
3682 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
3683 // add static memory usages to remembered set sizes
3684 _total_remset_bytes += HeapRegionRemSet::fl_mem_size() + HeapRegionRemSet::static_mem_size();
3685 // Print the footer of the output.
3686 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
3687 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3688 " SUMMARY"
3689 G1PPRL_SUM_MB_FORMAT("capacity")
3690 G1PPRL_SUM_MB_PERC_FORMAT("used")
3691 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
3692 G1PPRL_SUM_MB_PERC_FORMAT("next-live")
3693 G1PPRL_SUM_MB_FORMAT("remset")
3694 G1PPRL_SUM_MB_FORMAT("code-roots"),
3695 bytes_to_mb(_total_capacity_bytes),
3696 bytes_to_mb(_total_used_bytes),
3697 perc(_total_used_bytes, _total_capacity_bytes),
3698 bytes_to_mb(_total_prev_live_bytes),
3699 perc(_total_prev_live_bytes, _total_capacity_bytes),
3700 bytes_to_mb(_total_next_live_bytes),
3701 perc(_total_next_live_bytes, _total_capacity_bytes),
3702 bytes_to_mb(_total_remset_bytes),
3703 bytes_to_mb(_total_strong_code_roots_bytes));
3704 }