1 /*
2 * Copyright (c) 2001, 2015, 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/concurrentMark.inline.hpp"
30 #include "gc/g1/concurrentMarkThread.inline.hpp"
31 #include "gc/g1/g1CollectedHeap.inline.hpp"
32 #include "gc/g1/g1CollectorPolicy.hpp"
33 #include "gc/g1/g1CollectorState.hpp"
34 #include "gc/g1/g1OopClosures.inline.hpp"
35 #include "gc/g1/g1RemSet.hpp"
36 #include "gc/g1/g1StringDedup.hpp"
37 #include "gc/g1/heapRegion.inline.hpp"
38 #include "gc/g1/heapRegionManager.inline.hpp"
39 #include "gc/g1/heapRegionRemSet.hpp"
40 #include "gc/g1/heapRegionSet.inline.hpp"
41 #include "gc/g1/suspendibleThreadSet.hpp"
42 #include "gc/shared/gcId.hpp"
43 #include "gc/shared/gcTimer.hpp"
44 #include "gc/shared/gcTrace.hpp"
45 #include "gc/shared/gcTraceTime.hpp"
46 #include "gc/shared/genOopClosures.inline.hpp"
47 #include "gc/shared/referencePolicy.hpp"
48 #include "gc/shared/strongRootsScope.hpp"
49 #include "gc/shared/taskqueue.inline.hpp"
50 #include "gc/shared/vmGCOperations.hpp"
51 #include "logging/log.hpp"
52 #include "memory/allocation.hpp"
53 #include "memory/resourceArea.hpp"
54 #include "oops/oop.inline.hpp"
55 #include "runtime/atomic.inline.hpp"
56 #include "runtime/handles.inline.hpp"
57 #include "runtime/java.hpp"
58 #include "runtime/prefetch.inline.hpp"
59 #include "services/memTracker.hpp"
60
61 // Concurrent marking bit map wrapper
62
63 CMBitMapRO::CMBitMapRO(int shifter) :
64 _bm(),
65 _shifter(shifter) {
66 _bmStartWord = 0;
67 _bmWordSize = 0;
68 }
69
70 HeapWord* CMBitMapRO::getNextMarkedWordAddress(const HeapWord* addr,
71 const HeapWord* limit) const {
72 // First we must round addr *up* to a possible object boundary.
73 addr = (HeapWord*)align_size_up((intptr_t)addr,
74 HeapWordSize << _shifter);
75 size_t addrOffset = heapWordToOffset(addr);
76 if (limit == NULL) {
77 limit = _bmStartWord + _bmWordSize;
78 }
79 size_t limitOffset = heapWordToOffset(limit);
80 size_t nextOffset = _bm.get_next_one_offset(addrOffset, limitOffset);
81 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
82 assert(nextAddr >= addr, "get_next_one postcondition");
83 assert(nextAddr == limit || isMarked(nextAddr),
84 "get_next_one postcondition");
85 return nextAddr;
86 }
87
88 HeapWord* CMBitMapRO::getNextUnmarkedWordAddress(const HeapWord* addr,
89 const HeapWord* limit) const {
90 size_t addrOffset = heapWordToOffset(addr);
91 if (limit == NULL) {
92 limit = _bmStartWord + _bmWordSize;
93 }
94 size_t limitOffset = heapWordToOffset(limit);
95 size_t nextOffset = _bm.get_next_zero_offset(addrOffset, limitOffset);
96 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
97 assert(nextAddr >= addr, "get_next_one postcondition");
98 assert(nextAddr == limit || !isMarked(nextAddr),
99 "get_next_one postcondition");
100 return nextAddr;
101 }
102
103 int CMBitMapRO::heapWordDiffToOffsetDiff(size_t diff) const {
104 assert((diff & ((1 << _shifter) - 1)) == 0, "argument check");
105 return (int) (diff >> _shifter);
106 }
107
108 #ifndef PRODUCT
109 bool CMBitMapRO::covers(MemRegion heap_rs) const {
110 // assert(_bm.map() == _virtual_space.low(), "map inconsistency");
111 assert(((size_t)_bm.size() * ((size_t)1 << _shifter)) == _bmWordSize,
112 "size inconsistency");
113 return _bmStartWord == (HeapWord*)(heap_rs.start()) &&
114 _bmWordSize == heap_rs.word_size();
115 }
116 #endif
117
118 void CMBitMapRO::print_on_error(outputStream* st, const char* prefix) const {
119 _bm.print_on_error(st, prefix);
120 }
121
122 size_t CMBitMap::compute_size(size_t heap_size) {
123 return ReservedSpace::allocation_align_size_up(heap_size / mark_distance());
124 }
125
126 size_t CMBitMap::mark_distance() {
127 return MinObjAlignmentInBytes * BitsPerByte;
128 }
129
130 void CMBitMap::initialize(MemRegion heap, G1RegionToSpaceMapper* storage) {
131 _bmStartWord = heap.start();
132 _bmWordSize = heap.word_size();
133
134 _bm.set_map((BitMap::bm_word_t*) storage->reserved().start());
135 _bm.set_size(_bmWordSize >> _shifter);
136
137 storage->set_mapping_changed_listener(&_listener);
138 }
139
140 void CMBitMapMappingChangedListener::on_commit(uint start_region, size_t num_regions, bool zero_filled) {
141 if (zero_filled) {
142 return;
143 }
144 // We need to clear the bitmap on commit, removing any existing information.
145 MemRegion mr(G1CollectedHeap::heap()->bottom_addr_for_region(start_region), num_regions * HeapRegion::GrainWords);
146 _bm->clearRange(mr);
147 }
148
149 // Closure used for clearing the given mark bitmap.
150 class ClearBitmapHRClosure : public HeapRegionClosure {
151 private:
152 ConcurrentMark* _cm;
153 CMBitMap* _bitmap;
154 bool _may_yield; // The closure may yield during iteration. If yielded, abort the iteration.
155 public:
156 ClearBitmapHRClosure(ConcurrentMark* cm, CMBitMap* bitmap, bool may_yield) : HeapRegionClosure(), _cm(cm), _bitmap(bitmap), _may_yield(may_yield) {
157 assert(!may_yield || cm != NULL, "CM must be non-NULL if this closure is expected to yield.");
158 }
159
160 virtual bool doHeapRegion(HeapRegion* r) {
161 size_t const chunk_size_in_words = M / HeapWordSize;
162
163 HeapWord* cur = r->bottom();
164 HeapWord* const end = r->end();
165
166 while (cur < end) {
167 MemRegion mr(cur, MIN2(cur + chunk_size_in_words, end));
168 _bitmap->clearRange(mr);
169
170 cur += chunk_size_in_words;
171
172 // Abort iteration if after yielding the marking has been aborted.
173 if (_may_yield && _cm->do_yield_check() && _cm->has_aborted()) {
174 return true;
175 }
176 // Repeat the asserts from before the start of the closure. We will do them
177 // as asserts here to minimize their overhead on the product. However, we
178 // will have them as guarantees at the beginning / end of the bitmap
179 // clearing to get some checking in the product.
180 assert(!_may_yield || _cm->cmThread()->during_cycle(), "invariant");
181 assert(!_may_yield || !G1CollectedHeap::heap()->collector_state()->mark_in_progress(), "invariant");
182 }
183
184 return false;
185 }
186 };
187
188 class ParClearNextMarkBitmapTask : public AbstractGangTask {
189 ClearBitmapHRClosure* _cl;
190 HeapRegionClaimer _hrclaimer;
191 bool _suspendible; // If the task is suspendible, workers must join the STS.
192
193 public:
194 ParClearNextMarkBitmapTask(ClearBitmapHRClosure *cl, uint n_workers, bool suspendible) :
195 _cl(cl), _suspendible(suspendible), AbstractGangTask("Parallel Clear Bitmap Task"), _hrclaimer(n_workers) {}
196
197 void work(uint worker_id) {
198 SuspendibleThreadSetJoiner sts_join(_suspendible);
199 G1CollectedHeap::heap()->heap_region_par_iterate(_cl, worker_id, &_hrclaimer, true);
200 }
201 };
202
203 void CMBitMap::clearAll() {
204 G1CollectedHeap* g1h = G1CollectedHeap::heap();
205 ClearBitmapHRClosure cl(NULL, this, false /* may_yield */);
206 uint n_workers = g1h->workers()->active_workers();
207 ParClearNextMarkBitmapTask task(&cl, n_workers, false);
208 g1h->workers()->run_task(&task);
209 guarantee(cl.complete(), "Must have completed iteration.");
210 return;
211 }
212
213 void CMBitMap::markRange(MemRegion mr) {
214 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
215 assert(!mr.is_empty(), "unexpected empty region");
216 assert((offsetToHeapWord(heapWordToOffset(mr.end())) ==
217 ((HeapWord *) mr.end())),
218 "markRange memory region end is not card aligned");
219 // convert address range into offset range
220 _bm.at_put_range(heapWordToOffset(mr.start()),
221 heapWordToOffset(mr.end()), true);
222 }
223
224 void CMBitMap::clearRange(MemRegion mr) {
225 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
226 assert(!mr.is_empty(), "unexpected empty region");
227 // convert address range into offset range
228 _bm.at_put_range(heapWordToOffset(mr.start()),
229 heapWordToOffset(mr.end()), false);
230 }
231
232 MemRegion CMBitMap::getAndClearMarkedRegion(HeapWord* addr,
233 HeapWord* end_addr) {
234 HeapWord* start = getNextMarkedWordAddress(addr);
235 start = MIN2(start, end_addr);
236 HeapWord* end = getNextUnmarkedWordAddress(start);
237 end = MIN2(end, end_addr);
238 assert(start <= end, "Consistency check");
239 MemRegion mr(start, end);
240 if (!mr.is_empty()) {
241 clearRange(mr);
242 }
243 return mr;
244 }
245
246 CMMarkStack::CMMarkStack(ConcurrentMark* cm) :
247 _base(NULL), _cm(cm)
248 {}
249
250 bool CMMarkStack::allocate(size_t capacity) {
251 // allocate a stack of the requisite depth
252 ReservedSpace rs(ReservedSpace::allocation_align_size_up(capacity * sizeof(oop)));
253 if (!rs.is_reserved()) {
254 warning("ConcurrentMark MarkStack allocation failure");
255 return false;
256 }
257 MemTracker::record_virtual_memory_type((address)rs.base(), mtGC);
258 if (!_virtual_space.initialize(rs, rs.size())) {
259 warning("ConcurrentMark MarkStack backing store failure");
260 // Release the virtual memory reserved for the marking stack
261 rs.release();
262 return false;
263 }
264 assert(_virtual_space.committed_size() == rs.size(),
265 "Didn't reserve backing store for all of ConcurrentMark stack?");
266 _base = (oop*) _virtual_space.low();
267 setEmpty();
268 _capacity = (jint) capacity;
269 _saved_index = -1;
270 _should_expand = false;
271 return true;
272 }
273
274 void CMMarkStack::expand() {
275 // Called, during remark, if we've overflown the marking stack during marking.
276 assert(isEmpty(), "stack should been emptied while handling overflow");
277 assert(_capacity <= (jint) MarkStackSizeMax, "stack bigger than permitted");
278 // Clear expansion flag
279 _should_expand = false;
280 if (_capacity == (jint) MarkStackSizeMax) {
281 log_trace(gc)("(benign) Can't expand marking stack capacity, at max size limit");
282 return;
283 }
284 // Double capacity if possible
285 jint new_capacity = MIN2(_capacity*2, (jint) MarkStackSizeMax);
286 // Do not give up existing stack until we have managed to
287 // get the double capacity that we desired.
288 ReservedSpace rs(ReservedSpace::allocation_align_size_up(new_capacity *
289 sizeof(oop)));
290 if (rs.is_reserved()) {
291 // Release the backing store associated with old stack
292 _virtual_space.release();
293 // Reinitialize virtual space for new stack
294 if (!_virtual_space.initialize(rs, rs.size())) {
295 fatal("Not enough swap for expanded marking stack capacity");
296 }
297 _base = (oop*)(_virtual_space.low());
298 _index = 0;
299 _capacity = new_capacity;
300 } else {
301 // Failed to double capacity, continue;
302 log_trace(gc)("(benign) Failed to expand marking stack capacity from " SIZE_FORMAT "K to " SIZE_FORMAT "K",
303 _capacity / K, new_capacity / K);
304 }
305 }
306
307 void CMMarkStack::set_should_expand() {
308 // If we're resetting the marking state because of an
309 // marking stack overflow, record that we should, if
310 // possible, expand the stack.
311 _should_expand = _cm->has_overflown();
312 }
313
314 CMMarkStack::~CMMarkStack() {
315 if (_base != NULL) {
316 _base = NULL;
317 _virtual_space.release();
318 }
319 }
320
321 void CMMarkStack::par_push_arr(oop* ptr_arr, int n) {
322 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
323 jint start = _index;
324 jint next_index = start + n;
325 if (next_index > _capacity) {
326 _overflow = true;
327 return;
328 }
329 // Otherwise.
330 _index = next_index;
331 for (int i = 0; i < n; i++) {
332 int ind = start + i;
333 assert(ind < _capacity, "By overflow test above.");
334 _base[ind] = ptr_arr[i];
335 }
336 }
337
338 bool CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) {
339 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
340 jint index = _index;
341 if (index == 0) {
342 *n = 0;
343 return false;
344 } else {
345 int k = MIN2(max, index);
346 jint new_ind = index - k;
347 for (int j = 0; j < k; j++) {
348 ptr_arr[j] = _base[new_ind + j];
349 }
350 _index = new_ind;
351 *n = k;
352 return true;
353 }
354 }
355
356 void CMMarkStack::note_start_of_gc() {
357 assert(_saved_index == -1,
358 "note_start_of_gc()/end_of_gc() bracketed incorrectly");
359 _saved_index = _index;
360 }
361
362 void CMMarkStack::note_end_of_gc() {
363 // This is intentionally a guarantee, instead of an assert. If we
364 // accidentally add something to the mark stack during GC, it
365 // will be a correctness issue so it's better if we crash. we'll
366 // only check this once per GC anyway, so it won't be a performance
367 // issue in any way.
368 guarantee(_saved_index == _index,
369 "saved index: %d index: %d", _saved_index, _index);
370 _saved_index = -1;
371 }
372
373 CMRootRegions::CMRootRegions() :
374 _young_list(NULL), _cm(NULL), _scan_in_progress(false),
375 _should_abort(false), _next_survivor(NULL) { }
376
377 void CMRootRegions::init(G1CollectedHeap* g1h, ConcurrentMark* cm) {
378 _young_list = g1h->young_list();
379 _cm = cm;
380 }
381
382 void CMRootRegions::prepare_for_scan() {
383 assert(!scan_in_progress(), "pre-condition");
384
385 // Currently, only survivors can be root regions.
386 assert(_next_survivor == NULL, "pre-condition");
387 _next_survivor = _young_list->first_survivor_region();
388 _scan_in_progress = (_next_survivor != NULL);
389 _should_abort = false;
390 }
391
392 HeapRegion* CMRootRegions::claim_next() {
393 if (_should_abort) {
394 // If someone has set the should_abort flag, we return NULL to
395 // force the caller to bail out of their loop.
396 return NULL;
397 }
398
399 // Currently, only survivors can be root regions.
400 HeapRegion* res = _next_survivor;
401 if (res != NULL) {
402 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
403 // Read it again in case it changed while we were waiting for the lock.
404 res = _next_survivor;
405 if (res != NULL) {
406 if (res == _young_list->last_survivor_region()) {
407 // We just claimed the last survivor so store NULL to indicate
408 // that we're done.
409 _next_survivor = NULL;
410 } else {
411 _next_survivor = res->get_next_young_region();
412 }
413 } else {
414 // Someone else claimed the last survivor while we were trying
415 // to take the lock so nothing else to do.
416 }
417 }
418 assert(res == NULL || res->is_survivor(), "post-condition");
419
420 return res;
421 }
422
423 void CMRootRegions::scan_finished() {
424 assert(scan_in_progress(), "pre-condition");
425
426 // Currently, only survivors can be root regions.
427 if (!_should_abort) {
428 assert(_next_survivor == NULL, "we should have claimed all survivors");
429 }
430 _next_survivor = NULL;
431
432 {
433 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
434 _scan_in_progress = false;
435 RootRegionScan_lock->notify_all();
436 }
437 }
438
439 bool CMRootRegions::wait_until_scan_finished() {
440 if (!scan_in_progress()) return false;
441
442 {
443 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
444 while (scan_in_progress()) {
445 RootRegionScan_lock->wait(Mutex::_no_safepoint_check_flag);
446 }
447 }
448 return true;
449 }
450
451 uint ConcurrentMark::scale_parallel_threads(uint n_par_threads) {
452 return MAX2((n_par_threads + 2) / 4, 1U);
453 }
454
455 ConcurrentMark::ConcurrentMark(G1CollectedHeap* g1h, G1RegionToSpaceMapper* prev_bitmap_storage, G1RegionToSpaceMapper* next_bitmap_storage) :
456 _g1h(g1h),
457 _markBitMap1(),
458 _markBitMap2(),
459 _parallel_marking_threads(0),
460 _max_parallel_marking_threads(0),
461 _sleep_factor(0.0),
462 _marking_task_overhead(1.0),
463 _cleanup_sleep_factor(0.0),
464 _cleanup_task_overhead(1.0),
465 _cleanup_list("Cleanup List"),
466 _region_bm((BitMap::idx_t)(g1h->max_regions()), false /* in_resource_area*/),
467 _card_bm((g1h->reserved_region().byte_size() + CardTableModRefBS::card_size - 1) >>
468 CardTableModRefBS::card_shift,
469 false /* in_resource_area*/),
470
471 _prevMarkBitMap(&_markBitMap1),
472 _nextMarkBitMap(&_markBitMap2),
473
474 _markStack(this),
475 // _finger set in set_non_marking_state
476
477 _max_worker_id(ParallelGCThreads),
478 // _active_tasks set in set_non_marking_state
479 // _tasks set inside the constructor
480 _task_queues(new CMTaskQueueSet((int) _max_worker_id)),
481 _terminator(ParallelTaskTerminator((int) _max_worker_id, _task_queues)),
482
483 _has_overflown(false),
484 _concurrent(false),
485 _has_aborted(false),
486 _restart_for_overflow(false),
487 _concurrent_marking_in_progress(false),
488
489 // _verbose_level set below
490
491 _init_times(),
492 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(),
493 _cleanup_times(),
494 _total_counting_time(0.0),
495 _total_rs_scrub_time(0.0),
496
497 _parallel_workers(NULL),
498
499 _count_card_bitmaps(NULL),
500 _count_marked_bytes(NULL),
501 _completed_initialization(false) {
502
503 _markBitMap1.initialize(g1h->reserved_region(), prev_bitmap_storage);
504 _markBitMap2.initialize(g1h->reserved_region(), next_bitmap_storage);
505
506 // Create & start a ConcurrentMark thread.
507 _cmThread = new ConcurrentMarkThread(this);
508 assert(cmThread() != NULL, "CM Thread should have been created");
509 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm");
510 if (_cmThread->osthread() == NULL) {
511 vm_shutdown_during_initialization("Could not create ConcurrentMarkThread");
512 }
513
514 assert(CGC_lock != NULL, "Where's the CGC_lock?");
515 assert(_markBitMap1.covers(g1h->reserved_region()), "_markBitMap1 inconsistency");
516 assert(_markBitMap2.covers(g1h->reserved_region()), "_markBitMap2 inconsistency");
517
518 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
519 satb_qs.set_buffer_size(G1SATBBufferSize);
520
521 _root_regions.init(_g1h, this);
522
523 if (ConcGCThreads > ParallelGCThreads) {
524 warning("Can't have more ConcGCThreads (%u) "
525 "than ParallelGCThreads (%u).",
526 ConcGCThreads, ParallelGCThreads);
527 return;
528 }
529 if (!FLAG_IS_DEFAULT(ConcGCThreads) && ConcGCThreads > 0) {
530 // Note: ConcGCThreads has precedence over G1MarkingOverheadPercent
531 // if both are set
532 _sleep_factor = 0.0;
533 _marking_task_overhead = 1.0;
534 } else if (G1MarkingOverheadPercent > 0) {
535 // We will calculate the number of parallel marking threads based
536 // on a target overhead with respect to the soft real-time goal
537 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
538 double overall_cm_overhead =
539 (double) MaxGCPauseMillis * marking_overhead /
540 (double) GCPauseIntervalMillis;
541 double cpu_ratio = 1.0 / (double) os::processor_count();
542 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio);
543 double marking_task_overhead =
544 overall_cm_overhead / marking_thread_num *
545 (double) os::processor_count();
546 double sleep_factor =
547 (1.0 - marking_task_overhead) / marking_task_overhead;
548
549 FLAG_SET_ERGO(uint, ConcGCThreads, (uint) marking_thread_num);
550 _sleep_factor = sleep_factor;
551 _marking_task_overhead = marking_task_overhead;
552 } else {
553 // Calculate the number of parallel marking threads by scaling
554 // the number of parallel GC threads.
555 uint marking_thread_num = scale_parallel_threads(ParallelGCThreads);
556 FLAG_SET_ERGO(uint, ConcGCThreads, marking_thread_num);
557 _sleep_factor = 0.0;
558 _marking_task_overhead = 1.0;
559 }
560
561 assert(ConcGCThreads > 0, "Should have been set");
562 _parallel_marking_threads = ConcGCThreads;
563 _max_parallel_marking_threads = _parallel_marking_threads;
564
565 if (parallel_marking_threads() > 1) {
566 _cleanup_task_overhead = 1.0;
567 } else {
568 _cleanup_task_overhead = marking_task_overhead();
569 }
570 _cleanup_sleep_factor =
571 (1.0 - cleanup_task_overhead()) / cleanup_task_overhead();
572
573 _parallel_workers = new WorkGang("G1 Marker",
574 _max_parallel_marking_threads, false, true);
575 if (_parallel_workers == NULL) {
576 vm_exit_during_initialization("Failed necessary allocation.");
577 } else {
578 _parallel_workers->initialize_workers();
579 }
580
581 if (FLAG_IS_DEFAULT(MarkStackSize)) {
582 size_t mark_stack_size =
583 MIN2(MarkStackSizeMax,
584 MAX2(MarkStackSize, (size_t) (parallel_marking_threads() * TASKQUEUE_SIZE)));
585 // Verify that the calculated value for MarkStackSize is in range.
586 // It would be nice to use the private utility routine from Arguments.
587 if (!(mark_stack_size >= 1 && mark_stack_size <= MarkStackSizeMax)) {
588 warning("Invalid value calculated for MarkStackSize (" SIZE_FORMAT "): "
589 "must be between 1 and " SIZE_FORMAT,
590 mark_stack_size, MarkStackSizeMax);
591 return;
592 }
593 FLAG_SET_ERGO(size_t, MarkStackSize, mark_stack_size);
594 } else {
595 // Verify MarkStackSize is in range.
596 if (FLAG_IS_CMDLINE(MarkStackSize)) {
597 if (FLAG_IS_DEFAULT(MarkStackSizeMax)) {
598 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
599 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT "): "
600 "must be between 1 and " SIZE_FORMAT,
601 MarkStackSize, MarkStackSizeMax);
602 return;
603 }
604 } else if (FLAG_IS_CMDLINE(MarkStackSizeMax)) {
605 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
606 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT ")"
607 " or for MarkStackSizeMax (" SIZE_FORMAT ")",
608 MarkStackSize, MarkStackSizeMax);
609 return;
610 }
611 }
612 }
613 }
614
615 if (!_markStack.allocate(MarkStackSize)) {
616 warning("Failed to allocate CM marking stack");
617 return;
618 }
619
620 _tasks = NEW_C_HEAP_ARRAY(CMTask*, _max_worker_id, mtGC);
621 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_worker_id, mtGC);
622
623 _count_card_bitmaps = NEW_C_HEAP_ARRAY(BitMap, _max_worker_id, mtGC);
624 _count_marked_bytes = NEW_C_HEAP_ARRAY(size_t*, _max_worker_id, mtGC);
625
626 BitMap::idx_t card_bm_size = _card_bm.size();
627
628 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
629 _active_tasks = _max_worker_id;
630
631 uint max_regions = _g1h->max_regions();
632 for (uint i = 0; i < _max_worker_id; ++i) {
633 CMTaskQueue* task_queue = new CMTaskQueue();
634 task_queue->initialize();
635 _task_queues->register_queue(i, task_queue);
636
637 _count_card_bitmaps[i] = BitMap(card_bm_size, false);
638 _count_marked_bytes[i] = NEW_C_HEAP_ARRAY(size_t, max_regions, mtGC);
639
640 _tasks[i] = new CMTask(i, this,
641 _count_marked_bytes[i],
642 &_count_card_bitmaps[i],
643 task_queue, _task_queues);
644
645 _accum_task_vtime[i] = 0.0;
646 }
647
648 // Calculate the card number for the bottom of the heap. Used
649 // in biasing indexes into the accounting card bitmaps.
650 _heap_bottom_card_num =
651 intptr_t(uintptr_t(_g1h->reserved_region().start()) >>
652 CardTableModRefBS::card_shift);
653
654 // Clear all the liveness counting data
655 clear_all_count_data();
656
657 // so that the call below can read a sensible value
658 _heap_start = g1h->reserved_region().start();
659 set_non_marking_state();
660 _completed_initialization = true;
661 }
662
663 void ConcurrentMark::reset() {
664 // Starting values for these two. This should be called in a STW
665 // phase.
666 MemRegion reserved = _g1h->g1_reserved();
667 _heap_start = reserved.start();
668 _heap_end = reserved.end();
669
670 // Separated the asserts so that we know which one fires.
671 assert(_heap_start != NULL, "heap bounds should look ok");
672 assert(_heap_end != NULL, "heap bounds should look ok");
673 assert(_heap_start < _heap_end, "heap bounds should look ok");
674
675 // Reset all the marking data structures and any necessary flags
676 reset_marking_state();
677
678 // We do reset all of them, since different phases will use
679 // different number of active threads. So, it's easiest to have all
680 // of them ready.
681 for (uint i = 0; i < _max_worker_id; ++i) {
682 _tasks[i]->reset(_nextMarkBitMap);
683 }
684
685 // we need this to make sure that the flag is on during the evac
686 // pause with initial mark piggy-backed
687 set_concurrent_marking_in_progress();
688 }
689
690
691 void ConcurrentMark::reset_marking_state(bool clear_overflow) {
692 _markStack.set_should_expand();
693 _markStack.setEmpty(); // Also clears the _markStack overflow flag
694 if (clear_overflow) {
695 clear_has_overflown();
696 } else {
697 assert(has_overflown(), "pre-condition");
698 }
699 _finger = _heap_start;
700
701 for (uint i = 0; i < _max_worker_id; ++i) {
702 CMTaskQueue* queue = _task_queues->queue(i);
703 queue->set_empty();
704 }
705 }
706
707 void ConcurrentMark::set_concurrency(uint active_tasks) {
708 assert(active_tasks <= _max_worker_id, "we should not have more");
709
710 _active_tasks = active_tasks;
711 // Need to update the three data structures below according to the
712 // number of active threads for this phase.
713 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
714 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
715 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
716 }
717
718 void ConcurrentMark::set_concurrency_and_phase(uint active_tasks, bool concurrent) {
719 set_concurrency(active_tasks);
720
721 _concurrent = concurrent;
722 // We propagate this to all tasks, not just the active ones.
723 for (uint i = 0; i < _max_worker_id; ++i)
724 _tasks[i]->set_concurrent(concurrent);
725
726 if (concurrent) {
727 set_concurrent_marking_in_progress();
728 } else {
729 // We currently assume that the concurrent flag has been set to
730 // false before we start remark. At this point we should also be
731 // in a STW phase.
732 assert(!concurrent_marking_in_progress(), "invariant");
733 assert(out_of_regions(),
734 "only way to get here: _finger: " PTR_FORMAT ", _heap_end: " PTR_FORMAT,
735 p2i(_finger), p2i(_heap_end));
736 }
737 }
738
739 void ConcurrentMark::set_non_marking_state() {
740 // We set the global marking state to some default values when we're
741 // not doing marking.
742 reset_marking_state();
743 _active_tasks = 0;
744 clear_concurrent_marking_in_progress();
745 }
746
747 ConcurrentMark::~ConcurrentMark() {
748 // The ConcurrentMark instance is never freed.
749 ShouldNotReachHere();
750 }
751
752 void ConcurrentMark::clearNextBitmap() {
753 G1CollectedHeap* g1h = G1CollectedHeap::heap();
754
755 // Make sure that the concurrent mark thread looks to still be in
756 // the current cycle.
757 guarantee(cmThread()->during_cycle(), "invariant");
758
759 // We are finishing up the current cycle by clearing the next
760 // marking bitmap and getting it ready for the next cycle. During
761 // this time no other cycle can start. So, let's make sure that this
762 // is the case.
763 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
764
765 ClearBitmapHRClosure cl(this, _nextMarkBitMap, true /* may_yield */);
766 ParClearNextMarkBitmapTask task(&cl, parallel_marking_threads(), true);
767 _parallel_workers->run_task(&task);
768
769 // Clear the liveness counting data. If the marking has been aborted, the abort()
770 // call already did that.
771 if (cl.complete()) {
772 clear_all_count_data();
773 }
774
775 // Repeat the asserts from above.
776 guarantee(cmThread()->during_cycle(), "invariant");
777 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
778 }
779
780 class CheckBitmapClearHRClosure : public HeapRegionClosure {
781 CMBitMap* _bitmap;
782 bool _error;
783 public:
784 CheckBitmapClearHRClosure(CMBitMap* bitmap) : _bitmap(bitmap) {
785 }
786
787 virtual bool doHeapRegion(HeapRegion* r) {
788 // This closure can be called concurrently to the mutator, so we must make sure
789 // that the result of the getNextMarkedWordAddress() call is compared to the
790 // value passed to it as limit to detect any found bits.
791 // end never changes in G1.
792 HeapWord* end = r->end();
793 return _bitmap->getNextMarkedWordAddress(r->bottom(), end) != end;
794 }
795 };
796
797 bool ConcurrentMark::nextMarkBitmapIsClear() {
798 CheckBitmapClearHRClosure cl(_nextMarkBitMap);
799 _g1h->heap_region_iterate(&cl);
800 return cl.complete();
801 }
802
803 class NoteStartOfMarkHRClosure: public HeapRegionClosure {
804 public:
805 bool doHeapRegion(HeapRegion* r) {
806 r->note_start_of_marking();
807 return false;
808 }
809 };
810
811 void ConcurrentMark::checkpointRootsInitialPre() {
812 G1CollectedHeap* g1h = G1CollectedHeap::heap();
813 G1CollectorPolicy* g1p = g1h->g1_policy();
814
815 _has_aborted = false;
816
817 // Initialize marking structures. This has to be done in a STW phase.
818 reset();
819
820 // For each region note start of marking.
821 NoteStartOfMarkHRClosure startcl;
822 g1h->heap_region_iterate(&startcl);
823 }
824
825
826 void ConcurrentMark::checkpointRootsInitialPost() {
827 G1CollectedHeap* g1h = G1CollectedHeap::heap();
828
829 // If we force an overflow during remark, the remark operation will
830 // actually abort and we'll restart concurrent marking. If we always
831 // force an overflow during remark we'll never actually complete the
832 // marking phase. So, we initialize this here, at the start of the
833 // cycle, so that at the remaining overflow number will decrease at
834 // every remark and we'll eventually not need to cause one.
835 force_overflow_stw()->init();
836
837 // Start Concurrent Marking weak-reference discovery.
838 ReferenceProcessor* rp = g1h->ref_processor_cm();
839 // enable ("weak") refs discovery
840 rp->enable_discovery();
841 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
842
843 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
844 // This is the start of the marking cycle, we're expected all
845 // threads to have SATB queues with active set to false.
846 satb_mq_set.set_active_all_threads(true, /* new active value */
847 false /* expected_active */);
848
849 _root_regions.prepare_for_scan();
850
851 // update_g1_committed() will be called at the end of an evac pause
852 // when marking is on. So, it's also called at the end of the
853 // initial-mark pause to update the heap end, if the heap expands
854 // during it. No need to call it here.
855 }
856
857 /*
858 * Notice that in the next two methods, we actually leave the STS
859 * during the barrier sync and join it immediately afterwards. If we
860 * do not do this, the following deadlock can occur: one thread could
861 * be in the barrier sync code, waiting for the other thread to also
862 * sync up, whereas another one could be trying to yield, while also
863 * waiting for the other threads to sync up too.
864 *
865 * Note, however, that this code is also used during remark and in
866 * this case we should not attempt to leave / enter the STS, otherwise
867 * we'll either hit an assert (debug / fastdebug) or deadlock
868 * (product). So we should only leave / enter the STS if we are
869 * operating concurrently.
870 *
871 * Because the thread that does the sync barrier has left the STS, it
872 * is possible to be suspended for a Full GC or an evacuation pause
873 * could occur. This is actually safe, since the entering the sync
874 * barrier is one of the last things do_marking_step() does, and it
875 * doesn't manipulate any data structures afterwards.
876 */
877
878 void ConcurrentMark::enter_first_sync_barrier(uint worker_id) {
879 bool barrier_aborted;
880 {
881 SuspendibleThreadSetLeaver sts_leave(concurrent());
882 barrier_aborted = !_first_overflow_barrier_sync.enter();
883 }
884
885 // at this point everyone should have synced up and not be doing any
886 // more work
887
888 if (barrier_aborted) {
889 // If the barrier aborted we ignore the overflow condition and
890 // just abort the whole marking phase as quickly as possible.
891 return;
892 }
893
894 // If we're executing the concurrent phase of marking, reset the marking
895 // state; otherwise the marking state is reset after reference processing,
896 // during the remark pause.
897 // If we reset here as a result of an overflow during the remark we will
898 // see assertion failures from any subsequent set_concurrency_and_phase()
899 // calls.
900 if (concurrent()) {
901 // let the task associated with with worker 0 do this
902 if (worker_id == 0) {
903 // task 0 is responsible for clearing the global data structures
904 // We should be here because of an overflow. During STW we should
905 // not clear the overflow flag since we rely on it being true when
906 // we exit this method to abort the pause and restart concurrent
907 // marking.
908 reset_marking_state(true /* clear_overflow */);
909 force_overflow()->update();
910
911 log_info(gc)("Concurrent Mark reset for overflow");
912 }
913 }
914
915 // after this, each task should reset its own data structures then
916 // then go into the second barrier
917 }
918
919 void ConcurrentMark::enter_second_sync_barrier(uint worker_id) {
920 SuspendibleThreadSetLeaver sts_leave(concurrent());
921 _second_overflow_barrier_sync.enter();
922
923 // at this point everything should be re-initialized and ready to go
924 }
925
926 #ifndef PRODUCT
927 void ForceOverflowSettings::init() {
928 _num_remaining = G1ConcMarkForceOverflow;
929 _force = false;
930 update();
931 }
932
933 void ForceOverflowSettings::update() {
934 if (_num_remaining > 0) {
935 _num_remaining -= 1;
936 _force = true;
937 } else {
938 _force = false;
939 }
940 }
941
942 bool ForceOverflowSettings::should_force() {
943 if (_force) {
944 _force = false;
945 return true;
946 } else {
947 return false;
948 }
949 }
950 #endif // !PRODUCT
951
952 class CMConcurrentMarkingTask: public AbstractGangTask {
953 private:
954 ConcurrentMark* _cm;
955 ConcurrentMarkThread* _cmt;
956
957 public:
958 void work(uint worker_id) {
959 assert(Thread::current()->is_ConcurrentGC_thread(),
960 "this should only be done by a conc GC thread");
961 ResourceMark rm;
962
963 double start_vtime = os::elapsedVTime();
964
965 {
966 SuspendibleThreadSetJoiner sts_join;
967
968 assert(worker_id < _cm->active_tasks(), "invariant");
969 CMTask* the_task = _cm->task(worker_id);
970 the_task->record_start_time();
971 if (!_cm->has_aborted()) {
972 do {
973 double start_vtime_sec = os::elapsedVTime();
974 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
975
976 the_task->do_marking_step(mark_step_duration_ms,
977 true /* do_termination */,
978 false /* is_serial*/);
979
980 double end_vtime_sec = os::elapsedVTime();
981 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec;
982 _cm->clear_has_overflown();
983
984 _cm->do_yield_check(worker_id);
985
986 jlong sleep_time_ms;
987 if (!_cm->has_aborted() && the_task->has_aborted()) {
988 sleep_time_ms =
989 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0);
990 {
991 SuspendibleThreadSetLeaver sts_leave;
992 os::sleep(Thread::current(), sleep_time_ms, false);
993 }
994 }
995 } while (!_cm->has_aborted() && the_task->has_aborted());
996 }
997 the_task->record_end_time();
998 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant");
999 }
1000
1001 double end_vtime = os::elapsedVTime();
1002 _cm->update_accum_task_vtime(worker_id, end_vtime - start_vtime);
1003 }
1004
1005 CMConcurrentMarkingTask(ConcurrentMark* cm,
1006 ConcurrentMarkThread* cmt) :
1007 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { }
1008
1009 ~CMConcurrentMarkingTask() { }
1010 };
1011
1012 // Calculates the number of active workers for a concurrent
1013 // phase.
1014 uint ConcurrentMark::calc_parallel_marking_threads() {
1015 uint n_conc_workers = 0;
1016 if (!UseDynamicNumberOfGCThreads ||
1017 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
1018 !ForceDynamicNumberOfGCThreads)) {
1019 n_conc_workers = max_parallel_marking_threads();
1020 } else {
1021 n_conc_workers =
1022 AdaptiveSizePolicy::calc_default_active_workers(
1023 max_parallel_marking_threads(),
1024 1, /* Minimum workers */
1025 parallel_marking_threads(),
1026 Threads::number_of_non_daemon_threads());
1027 // Don't scale down "n_conc_workers" by scale_parallel_threads() because
1028 // that scaling has already gone into "_max_parallel_marking_threads".
1029 }
1030 assert(n_conc_workers > 0, "Always need at least 1");
1031 return n_conc_workers;
1032 }
1033
1034 void ConcurrentMark::scanRootRegion(HeapRegion* hr, uint worker_id) {
1035 // Currently, only survivors can be root regions.
1036 assert(hr->next_top_at_mark_start() == hr->bottom(), "invariant");
1037 G1RootRegionScanClosure cl(_g1h, this, worker_id);
1038
1039 const uintx interval = PrefetchScanIntervalInBytes;
1040 HeapWord* curr = hr->bottom();
1041 const HeapWord* end = hr->top();
1042 while (curr < end) {
1043 Prefetch::read(curr, interval);
1044 oop obj = oop(curr);
1045 int size = obj->oop_iterate_size(&cl);
1046 assert(size == obj->size(), "sanity");
1047 curr += size;
1048 }
1049 }
1050
1051 class CMRootRegionScanTask : public AbstractGangTask {
1052 private:
1053 ConcurrentMark* _cm;
1054
1055 public:
1056 CMRootRegionScanTask(ConcurrentMark* cm) :
1057 AbstractGangTask("Root Region Scan"), _cm(cm) { }
1058
1059 void work(uint worker_id) {
1060 assert(Thread::current()->is_ConcurrentGC_thread(),
1061 "this should only be done by a conc GC thread");
1062
1063 CMRootRegions* root_regions = _cm->root_regions();
1064 HeapRegion* hr = root_regions->claim_next();
1065 while (hr != NULL) {
1066 _cm->scanRootRegion(hr, worker_id);
1067 hr = root_regions->claim_next();
1068 }
1069 }
1070 };
1071
1072 void ConcurrentMark::scanRootRegions() {
1073 jlong scan_start = os::elapsed_counter();
1074
1075 // Start of concurrent marking.
1076 ClassLoaderDataGraph::clear_claimed_marks();
1077
1078 // scan_in_progress() will have been set to true only if there was
1079 // at least one root region to scan. So, if it's false, we
1080 // should not attempt to do any further work.
1081 if (root_regions()->scan_in_progress()) {
1082 log_info(gc)("Concurrent Root Region Scan (%.3fs)", TimeHelper::counter_to_seconds(scan_start));
1083
1084 _parallel_marking_threads = calc_parallel_marking_threads();
1085 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1086 "Maximum number of marking threads exceeded");
1087 uint active_workers = MAX2(1U, parallel_marking_threads());
1088
1089 CMRootRegionScanTask task(this);
1090 _parallel_workers->set_active_workers(active_workers);
1091 _parallel_workers->run_task(&task);
1092
1093 double end_time_ms = os::elapsed_counter();
1094 log_info(gc)("Concurrent Root Region Scan (%.3fs, %.3fs) %.3fms",
1095 TimeHelper::counter_to_seconds(scan_start),
1096 TimeHelper::counter_to_seconds(end_time_ms),
1097 TimeHelper::counter_to_millis(end_time_ms - scan_start));
1098
1099 // It's possible that has_aborted() is true here without actually
1100 // aborting the survivor scan earlier. This is OK as it's
1101 // mainly used for sanity checking.
1102 root_regions()->scan_finished();
1103 }
1104 }
1105
1106 void ConcurrentMark::markFromRoots() {
1107 // we might be tempted to assert that:
1108 // assert(asynch == !SafepointSynchronize::is_at_safepoint(),
1109 // "inconsistent argument?");
1110 // However that wouldn't be right, because it's possible that
1111 // a safepoint is indeed in progress as a younger generation
1112 // stop-the-world GC happens even as we mark in this generation.
1113
1114 _restart_for_overflow = false;
1115 force_overflow_conc()->init();
1116
1117 // _g1h has _n_par_threads
1118 _parallel_marking_threads = calc_parallel_marking_threads();
1119 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1120 "Maximum number of marking threads exceeded");
1121
1122 uint active_workers = MAX2(1U, parallel_marking_threads());
1123 assert(active_workers > 0, "Should have been set");
1124
1125 // Parallel task terminator is set in "set_concurrency_and_phase()"
1126 set_concurrency_and_phase(active_workers, true /* concurrent */);
1127
1128 CMConcurrentMarkingTask markingTask(this, cmThread());
1129 _parallel_workers->set_active_workers(active_workers);
1130 _parallel_workers->run_task(&markingTask);
1131 print_stats();
1132 }
1133
1134 void ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) {
1135 // world is stopped at this checkpoint
1136 assert(SafepointSynchronize::is_at_safepoint(),
1137 "world should be stopped");
1138
1139 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1140
1141 // If a full collection has happened, we shouldn't do this.
1142 if (has_aborted()) {
1143 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1144 return;
1145 }
1146
1147 SvcGCMarker sgcm(SvcGCMarker::OTHER);
1148
1149 if (VerifyDuringGC) {
1150 HandleMark hm; // handle scope
1151 g1h->prepare_for_verify();
1152 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (before)");
1153 }
1154 g1h->check_bitmaps("Remark Start");
1155
1156 G1CollectorPolicy* g1p = g1h->g1_policy();
1157 g1p->record_concurrent_mark_remark_start();
1158
1159 double start = os::elapsedTime();
1160
1161 checkpointRootsFinalWork();
1162
1163 double mark_work_end = os::elapsedTime();
1164
1165 weakRefsWork(clear_all_soft_refs);
1166
1167 if (has_overflown()) {
1168 // Oops. We overflowed. Restart concurrent marking.
1169 _restart_for_overflow = true;
1170 log_develop(gc)("Remark led to restart for overflow.");
1171
1172 // Verify the heap w.r.t. the previous marking bitmap.
1173 if (VerifyDuringGC) {
1174 HandleMark hm; // handle scope
1175 g1h->prepare_for_verify();
1176 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (overflow)");
1177 }
1178
1179 // Clear the marking state because we will be restarting
1180 // marking due to overflowing the global mark stack.
1181 reset_marking_state();
1182 } else {
1183 {
1184 GCTraceTime(Debug, gc) trace("GC aggregate-data", g1h->gc_timer_cm());
1185
1186 // Aggregate the per-task counting data that we have accumulated
1187 // while marking.
1188 aggregate_count_data();
1189 }
1190
1191 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1192 // We're done with marking.
1193 // This is the end of the marking cycle, we're expected all
1194 // threads to have SATB queues with active set to true.
1195 satb_mq_set.set_active_all_threads(false, /* new active value */
1196 true /* expected_active */);
1197
1198 if (VerifyDuringGC) {
1199 HandleMark hm; // handle scope
1200 g1h->prepare_for_verify();
1201 Universe::verify(VerifyOption_G1UseNextMarking, "During GC (after)");
1202 }
1203 g1h->check_bitmaps("Remark End");
1204 assert(!restart_for_overflow(), "sanity");
1205 // Completely reset the marking state since marking completed
1206 set_non_marking_state();
1207 }
1208
1209 // Expand the marking stack, if we have to and if we can.
1210 if (_markStack.should_expand()) {
1211 _markStack.expand();
1212 }
1213
1214 // Statistics
1215 double now = os::elapsedTime();
1216 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1217 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1218 _remark_times.add((now - start) * 1000.0);
1219
1220 g1p->record_concurrent_mark_remark_end();
1221
1222 G1CMIsAliveClosure is_alive(g1h);
1223 g1h->gc_tracer_cm()->report_object_count_after_gc(&is_alive);
1224 }
1225
1226 // Base class of the closures that finalize and verify the
1227 // liveness counting data.
1228 class CMCountDataClosureBase: public HeapRegionClosure {
1229 protected:
1230 G1CollectedHeap* _g1h;
1231 ConcurrentMark* _cm;
1232 CardTableModRefBS* _ct_bs;
1233
1234 BitMap* _region_bm;
1235 BitMap* _card_bm;
1236
1237 // Takes a region that's not empty (i.e., it has at least one
1238 // live object in it and sets its corresponding bit on the region
1239 // bitmap to 1.
1240 void set_bit_for_region(HeapRegion* hr) {
1241 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1242 _region_bm->par_at_put(index, true);
1243 }
1244
1245 public:
1246 CMCountDataClosureBase(G1CollectedHeap* g1h,
1247 BitMap* region_bm, BitMap* card_bm):
1248 _g1h(g1h), _cm(g1h->concurrent_mark()),
1249 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
1250 _region_bm(region_bm), _card_bm(card_bm) { }
1251 };
1252
1253 // Closure that calculates the # live objects per region. Used
1254 // for verification purposes during the cleanup pause.
1255 class CalcLiveObjectsClosure: public CMCountDataClosureBase {
1256 CMBitMapRO* _bm;
1257 size_t _region_marked_bytes;
1258
1259 public:
1260 CalcLiveObjectsClosure(CMBitMapRO *bm, G1CollectedHeap* g1h,
1261 BitMap* region_bm, BitMap* card_bm) :
1262 CMCountDataClosureBase(g1h, region_bm, card_bm),
1263 _bm(bm), _region_marked_bytes(0) { }
1264
1265 bool doHeapRegion(HeapRegion* hr) {
1266 HeapWord* ntams = hr->next_top_at_mark_start();
1267 HeapWord* start = hr->bottom();
1268
1269 assert(start <= hr->end() && start <= ntams && ntams <= hr->end(),
1270 "Preconditions not met - "
1271 "start: " PTR_FORMAT ", ntams: " PTR_FORMAT ", end: " PTR_FORMAT,
1272 p2i(start), p2i(ntams), p2i(hr->end()));
1273
1274 // Find the first marked object at or after "start".
1275 start = _bm->getNextMarkedWordAddress(start, ntams);
1276
1277 size_t marked_bytes = 0;
1278
1279 while (start < ntams) {
1280 oop obj = oop(start);
1281 int obj_sz = obj->size();
1282 HeapWord* obj_end = start + obj_sz;
1283
1284 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
1285 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(obj_end);
1286
1287 // Note: if we're looking at the last region in heap - obj_end
1288 // could be actually just beyond the end of the heap; end_idx
1289 // will then correspond to a (non-existent) card that is also
1290 // just beyond the heap.
1291 if (_g1h->is_in_g1_reserved(obj_end) && !_ct_bs->is_card_aligned(obj_end)) {
1292 // end of object is not card aligned - increment to cover
1293 // all the cards spanned by the object
1294 end_idx += 1;
1295 }
1296
1297 // Set the bits in the card BM for the cards spanned by this object.
1298 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1299
1300 // Add the size of this object to the number of marked bytes.
1301 marked_bytes += (size_t)obj_sz * HeapWordSize;
1302
1303 // This will happen if we are handling a humongous object that spans
1304 // several heap regions.
1305 if (obj_end > hr->end()) {
1306 break;
1307 }
1308 // Find the next marked object after this one.
1309 start = _bm->getNextMarkedWordAddress(obj_end, ntams);
1310 }
1311
1312 // Mark the allocated-since-marking portion...
1313 HeapWord* top = hr->top();
1314 if (ntams < top) {
1315 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1316 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1317
1318 // Note: if we're looking at the last region in heap - top
1319 // could be actually just beyond the end of the heap; end_idx
1320 // will then correspond to a (non-existent) card that is also
1321 // just beyond the heap.
1322 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1323 // end of object is not card aligned - increment to cover
1324 // all the cards spanned by the object
1325 end_idx += 1;
1326 }
1327 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1328
1329 // This definitely means the region has live objects.
1330 set_bit_for_region(hr);
1331 }
1332
1333 // Update the live region bitmap.
1334 if (marked_bytes > 0) {
1335 set_bit_for_region(hr);
1336 }
1337
1338 // Set the marked bytes for the current region so that
1339 // it can be queried by a calling verification routine
1340 _region_marked_bytes = marked_bytes;
1341
1342 return false;
1343 }
1344
1345 size_t region_marked_bytes() const { return _region_marked_bytes; }
1346 };
1347
1348 // Heap region closure used for verifying the counting data
1349 // that was accumulated concurrently and aggregated during
1350 // the remark pause. This closure is applied to the heap
1351 // regions during the STW cleanup pause.
1352
1353 class VerifyLiveObjectDataHRClosure: public HeapRegionClosure {
1354 G1CollectedHeap* _g1h;
1355 ConcurrentMark* _cm;
1356 CalcLiveObjectsClosure _calc_cl;
1357 BitMap* _region_bm; // Region BM to be verified
1358 BitMap* _card_bm; // Card BM to be verified
1359
1360 BitMap* _exp_region_bm; // Expected Region BM values
1361 BitMap* _exp_card_bm; // Expected card BM values
1362
1363 int _failures;
1364
1365 public:
1366 VerifyLiveObjectDataHRClosure(G1CollectedHeap* g1h,
1367 BitMap* region_bm,
1368 BitMap* card_bm,
1369 BitMap* exp_region_bm,
1370 BitMap* exp_card_bm) :
1371 _g1h(g1h), _cm(g1h->concurrent_mark()),
1372 _calc_cl(_cm->nextMarkBitMap(), g1h, exp_region_bm, exp_card_bm),
1373 _region_bm(region_bm), _card_bm(card_bm),
1374 _exp_region_bm(exp_region_bm), _exp_card_bm(exp_card_bm),
1375 _failures(0) { }
1376
1377 int failures() const { return _failures; }
1378
1379 bool doHeapRegion(HeapRegion* hr) {
1380 int failures = 0;
1381
1382 // Call the CalcLiveObjectsClosure to walk the marking bitmap for
1383 // this region and set the corresponding bits in the expected region
1384 // and card bitmaps.
1385 bool res = _calc_cl.doHeapRegion(hr);
1386 assert(res == false, "should be continuing");
1387
1388 // Verify the marked bytes for this region.
1389 size_t exp_marked_bytes = _calc_cl.region_marked_bytes();
1390 size_t act_marked_bytes = hr->next_marked_bytes();
1391
1392 if (exp_marked_bytes > act_marked_bytes) {
1393 if (hr->is_starts_humongous()) {
1394 // For start_humongous regions, the size of the whole object will be
1395 // in exp_marked_bytes.
1396 HeapRegion* region = hr;
1397 int num_regions;
1398 for (num_regions = 0; region != NULL; num_regions++) {
1399 region = _g1h->next_region_in_humongous(region);
1400 }
1401 if ((num_regions-1) * HeapRegion::GrainBytes >= exp_marked_bytes) {
1402 failures += 1;
1403 } else if (num_regions * HeapRegion::GrainBytes < exp_marked_bytes) {
1404 failures += 1;
1405 }
1406 } else {
1407 // We're not OK if expected marked bytes > actual marked bytes. It means
1408 // we have missed accounting some objects during the actual marking.
1409 failures += 1;
1410 }
1411 }
1412
1413 // Verify the bit, for this region, in the actual and expected
1414 // (which was just calculated) region bit maps.
1415 // We're not OK if the bit in the calculated expected region
1416 // bitmap is set and the bit in the actual region bitmap is not.
1417 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1418
1419 bool expected = _exp_region_bm->at(index);
1420 bool actual = _region_bm->at(index);
1421 if (expected && !actual) {
1422 failures += 1;
1423 }
1424
1425 // Verify that the card bit maps for the cards spanned by the current
1426 // region match. We have an error if we have a set bit in the expected
1427 // bit map and the corresponding bit in the actual bitmap is not set.
1428
1429 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(hr->bottom());
1430 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(hr->top());
1431
1432 for (BitMap::idx_t i = start_idx; i < end_idx; i+=1) {
1433 expected = _exp_card_bm->at(i);
1434 actual = _card_bm->at(i);
1435
1436 if (expected && !actual) {
1437 failures += 1;
1438 }
1439 }
1440
1441 _failures += failures;
1442
1443 // We could stop iteration over the heap when we
1444 // find the first violating region by returning true.
1445 return false;
1446 }
1447 };
1448
1449 class G1ParVerifyFinalCountTask: public AbstractGangTask {
1450 protected:
1451 G1CollectedHeap* _g1h;
1452 ConcurrentMark* _cm;
1453 BitMap* _actual_region_bm;
1454 BitMap* _actual_card_bm;
1455
1456 uint _n_workers;
1457
1458 BitMap* _expected_region_bm;
1459 BitMap* _expected_card_bm;
1460
1461 int _failures;
1462
1463 HeapRegionClaimer _hrclaimer;
1464
1465 public:
1466 G1ParVerifyFinalCountTask(G1CollectedHeap* g1h,
1467 BitMap* region_bm, BitMap* card_bm,
1468 BitMap* expected_region_bm, BitMap* expected_card_bm)
1469 : AbstractGangTask("G1 verify final counting"),
1470 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1471 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1472 _expected_region_bm(expected_region_bm), _expected_card_bm(expected_card_bm),
1473 _failures(0),
1474 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1475 assert(VerifyDuringGC, "don't call this otherwise");
1476 assert(_expected_card_bm->size() == _actual_card_bm->size(), "sanity");
1477 assert(_expected_region_bm->size() == _actual_region_bm->size(), "sanity");
1478 }
1479
1480 void work(uint worker_id) {
1481 assert(worker_id < _n_workers, "invariant");
1482
1483 VerifyLiveObjectDataHRClosure verify_cl(_g1h,
1484 _actual_region_bm, _actual_card_bm,
1485 _expected_region_bm,
1486 _expected_card_bm);
1487
1488 _g1h->heap_region_par_iterate(&verify_cl, worker_id, &_hrclaimer);
1489
1490 Atomic::add(verify_cl.failures(), &_failures);
1491 }
1492
1493 int failures() const { return _failures; }
1494 };
1495
1496 // Closure that finalizes the liveness counting data.
1497 // Used during the cleanup pause.
1498 // Sets the bits corresponding to the interval [NTAMS, top]
1499 // (which contains the implicitly live objects) in the
1500 // card liveness bitmap. Also sets the bit for each region,
1501 // containing live data, in the region liveness bitmap.
1502
1503 class FinalCountDataUpdateClosure: public CMCountDataClosureBase {
1504 public:
1505 FinalCountDataUpdateClosure(G1CollectedHeap* g1h,
1506 BitMap* region_bm,
1507 BitMap* card_bm) :
1508 CMCountDataClosureBase(g1h, region_bm, card_bm) { }
1509
1510 bool doHeapRegion(HeapRegion* hr) {
1511 HeapWord* ntams = hr->next_top_at_mark_start();
1512 HeapWord* top = hr->top();
1513
1514 assert(hr->bottom() <= ntams && ntams <= hr->end(), "Preconditions.");
1515
1516 // Mark the allocated-since-marking portion...
1517 if (ntams < top) {
1518 // This definitely means the region has live objects.
1519 set_bit_for_region(hr);
1520
1521 // Now set the bits in the card bitmap for [ntams, top)
1522 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1523 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1524
1525 // Note: if we're looking at the last region in heap - top
1526 // could be actually just beyond the end of the heap; end_idx
1527 // will then correspond to a (non-existent) card that is also
1528 // just beyond the heap.
1529 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1530 // end of object is not card aligned - increment to cover
1531 // all the cards spanned by the object
1532 end_idx += 1;
1533 }
1534
1535 assert(end_idx <= _card_bm->size(),
1536 "oob: end_idx= " SIZE_FORMAT ", bitmap size= " SIZE_FORMAT,
1537 end_idx, _card_bm->size());
1538 assert(start_idx < _card_bm->size(),
1539 "oob: start_idx= " SIZE_FORMAT ", bitmap size= " SIZE_FORMAT,
1540 start_idx, _card_bm->size());
1541
1542 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1543 }
1544
1545 // Set the bit for the region if it contains live data
1546 if (hr->next_marked_bytes() > 0) {
1547 set_bit_for_region(hr);
1548 }
1549
1550 return false;
1551 }
1552 };
1553
1554 class G1ParFinalCountTask: public AbstractGangTask {
1555 protected:
1556 G1CollectedHeap* _g1h;
1557 ConcurrentMark* _cm;
1558 BitMap* _actual_region_bm;
1559 BitMap* _actual_card_bm;
1560
1561 uint _n_workers;
1562 HeapRegionClaimer _hrclaimer;
1563
1564 public:
1565 G1ParFinalCountTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm)
1566 : AbstractGangTask("G1 final counting"),
1567 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1568 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1569 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1570 }
1571
1572 void work(uint worker_id) {
1573 assert(worker_id < _n_workers, "invariant");
1574
1575 FinalCountDataUpdateClosure final_update_cl(_g1h,
1576 _actual_region_bm,
1577 _actual_card_bm);
1578
1579 _g1h->heap_region_par_iterate(&final_update_cl, worker_id, &_hrclaimer);
1580 }
1581 };
1582
1583 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure {
1584 G1CollectedHeap* _g1;
1585 size_t _freed_bytes;
1586 FreeRegionList* _local_cleanup_list;
1587 HeapRegionSetCount _old_regions_removed;
1588 HeapRegionSetCount _humongous_regions_removed;
1589 HRRSCleanupTask* _hrrs_cleanup_task;
1590
1591 public:
1592 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1,
1593 FreeRegionList* local_cleanup_list,
1594 HRRSCleanupTask* hrrs_cleanup_task) :
1595 _g1(g1),
1596 _freed_bytes(0),
1597 _local_cleanup_list(local_cleanup_list),
1598 _old_regions_removed(),
1599 _humongous_regions_removed(),
1600 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1601
1602 size_t freed_bytes() { return _freed_bytes; }
1603 const HeapRegionSetCount& old_regions_removed() { return _old_regions_removed; }
1604 const HeapRegionSetCount& humongous_regions_removed() { return _humongous_regions_removed; }
1605
1606 bool doHeapRegion(HeapRegion *hr) {
1607 if (hr->is_archive()) {
1608 return false;
1609 }
1610 // We use a claim value of zero here because all regions
1611 // were claimed with value 1 in the FinalCount task.
1612 _g1->reset_gc_time_stamps(hr);
1613 hr->note_end_of_marking();
1614
1615 if (hr->used() > 0 && hr->max_live_bytes() == 0 && !hr->is_young()) {
1616 _freed_bytes += hr->used();
1617 hr->set_containing_set(NULL);
1618 if (hr->is_humongous()) {
1619 _humongous_regions_removed.increment(1u, hr->capacity());
1620 _g1->free_humongous_region(hr, _local_cleanup_list, true);
1621 } else {
1622 _old_regions_removed.increment(1u, hr->capacity());
1623 _g1->free_region(hr, _local_cleanup_list, true);
1624 }
1625 } else {
1626 hr->rem_set()->do_cleanup_work(_hrrs_cleanup_task);
1627 }
1628
1629 return false;
1630 }
1631 };
1632
1633 class G1ParNoteEndTask: public AbstractGangTask {
1634 friend class G1NoteEndOfConcMarkClosure;
1635
1636 protected:
1637 G1CollectedHeap* _g1h;
1638 FreeRegionList* _cleanup_list;
1639 HeapRegionClaimer _hrclaimer;
1640
1641 public:
1642 G1ParNoteEndTask(G1CollectedHeap* g1h, FreeRegionList* cleanup_list, uint n_workers) :
1643 AbstractGangTask("G1 note end"), _g1h(g1h), _cleanup_list(cleanup_list), _hrclaimer(n_workers) {
1644 }
1645
1646 void work(uint worker_id) {
1647 FreeRegionList local_cleanup_list("Local Cleanup List");
1648 HRRSCleanupTask hrrs_cleanup_task;
1649 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, &local_cleanup_list,
1650 &hrrs_cleanup_task);
1651 _g1h->heap_region_par_iterate(&g1_note_end, worker_id, &_hrclaimer);
1652 assert(g1_note_end.complete(), "Shouldn't have yielded!");
1653
1654 // Now update the lists
1655 _g1h->remove_from_old_sets(g1_note_end.old_regions_removed(), g1_note_end.humongous_regions_removed());
1656 {
1657 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1658 _g1h->decrement_summary_bytes(g1_note_end.freed_bytes());
1659
1660 // If we iterate over the global cleanup list at the end of
1661 // cleanup to do this printing we will not guarantee to only
1662 // generate output for the newly-reclaimed regions (the list
1663 // might not be empty at the beginning of cleanup; we might
1664 // still be working on its previous contents). So we do the
1665 // printing here, before we append the new regions to the global
1666 // cleanup list.
1667
1668 G1HRPrinter* hr_printer = _g1h->hr_printer();
1669 if (hr_printer->is_active()) {
1670 FreeRegionListIterator iter(&local_cleanup_list);
1671 while (iter.more_available()) {
1672 HeapRegion* hr = iter.get_next();
1673 hr_printer->cleanup(hr);
1674 }
1675 }
1676
1677 _cleanup_list->add_ordered(&local_cleanup_list);
1678 assert(local_cleanup_list.is_empty(), "post-condition");
1679
1680 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1681 }
1682 }
1683 };
1684
1685 class G1ParScrubRemSetTask: public AbstractGangTask {
1686 protected:
1687 G1RemSet* _g1rs;
1688 BitMap* _region_bm;
1689 BitMap* _card_bm;
1690 HeapRegionClaimer _hrclaimer;
1691
1692 public:
1693 G1ParScrubRemSetTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm, uint n_workers) :
1694 AbstractGangTask("G1 ScrubRS"), _g1rs(g1h->g1_rem_set()), _region_bm(region_bm), _card_bm(card_bm), _hrclaimer(n_workers) {
1695 }
1696
1697 void work(uint worker_id) {
1698 _g1rs->scrub(_region_bm, _card_bm, worker_id, &_hrclaimer);
1699 }
1700
1701 };
1702
1703 void ConcurrentMark::cleanup() {
1704 // world is stopped at this checkpoint
1705 assert(SafepointSynchronize::is_at_safepoint(),
1706 "world should be stopped");
1707 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1708
1709 // If a full collection has happened, we shouldn't do this.
1710 if (has_aborted()) {
1711 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1712 return;
1713 }
1714
1715 g1h->verify_region_sets_optional();
1716
1717 if (VerifyDuringGC) {
1718 HandleMark hm; // handle scope
1719 g1h->prepare_for_verify();
1720 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (before)");
1721 }
1722 g1h->check_bitmaps("Cleanup Start");
1723
1724 G1CollectorPolicy* g1p = g1h->g1_policy();
1725 g1p->record_concurrent_mark_cleanup_start();
1726
1727 double start = os::elapsedTime();
1728
1729 HeapRegionRemSet::reset_for_cleanup_tasks();
1730
1731 // Do counting once more with the world stopped for good measure.
1732 G1ParFinalCountTask g1_par_count_task(g1h, &_region_bm, &_card_bm);
1733
1734 g1h->workers()->run_task(&g1_par_count_task);
1735
1736 if (VerifyDuringGC) {
1737 // Verify that the counting data accumulated during marking matches
1738 // that calculated by walking the marking bitmap.
1739
1740 // Bitmaps to hold expected values
1741 BitMap expected_region_bm(_region_bm.size(), true);
1742 BitMap expected_card_bm(_card_bm.size(), true);
1743
1744 G1ParVerifyFinalCountTask g1_par_verify_task(g1h,
1745 &_region_bm,
1746 &_card_bm,
1747 &expected_region_bm,
1748 &expected_card_bm);
1749
1750 g1h->workers()->run_task(&g1_par_verify_task);
1751
1752 guarantee(g1_par_verify_task.failures() == 0, "Unexpected accounting failures");
1753 }
1754
1755 size_t start_used_bytes = g1h->used();
1756 g1h->collector_state()->set_mark_in_progress(false);
1757
1758 double count_end = os::elapsedTime();
1759 double this_final_counting_time = (count_end - start);
1760 _total_counting_time += this_final_counting_time;
1761
1762 if (Log<LOG_TAGS(gc, liveness)>::is_trace()) {
1763 G1PrintRegionLivenessInfoClosure cl("Post-Marking");
1764 _g1h->heap_region_iterate(&cl);
1765 }
1766
1767 // Install newly created mark bitMap as "prev".
1768 swapMarkBitMaps();
1769
1770 g1h->reset_gc_time_stamp();
1771
1772 uint n_workers = _g1h->workers()->active_workers();
1773
1774 // Note end of marking in all heap regions.
1775 G1ParNoteEndTask g1_par_note_end_task(g1h, &_cleanup_list, n_workers);
1776 g1h->workers()->run_task(&g1_par_note_end_task);
1777 g1h->check_gc_time_stamps();
1778
1779 if (!cleanup_list_is_empty()) {
1780 // The cleanup list is not empty, so we'll have to process it
1781 // concurrently. Notify anyone else that might be wanting free
1782 // regions that there will be more free regions coming soon.
1783 g1h->set_free_regions_coming();
1784 }
1785
1786 // call below, since it affects the metric by which we sort the heap
1787 // regions.
1788 if (G1ScrubRemSets) {
1789 double rs_scrub_start = os::elapsedTime();
1790 G1ParScrubRemSetTask g1_par_scrub_rs_task(g1h, &_region_bm, &_card_bm, n_workers);
1791 g1h->workers()->run_task(&g1_par_scrub_rs_task);
1792
1793 double rs_scrub_end = os::elapsedTime();
1794 double this_rs_scrub_time = (rs_scrub_end - rs_scrub_start);
1795 _total_rs_scrub_time += this_rs_scrub_time;
1796 }
1797
1798 // this will also free any regions totally full of garbage objects,
1799 // and sort the regions.
1800 g1h->g1_policy()->record_concurrent_mark_cleanup_end();
1801
1802 // Statistics.
1803 double end = os::elapsedTime();
1804 _cleanup_times.add((end - start) * 1000.0);
1805
1806 // Clean up will have freed any regions completely full of garbage.
1807 // Update the soft reference policy with the new heap occupancy.
1808 Universe::update_heap_info_at_gc();
1809
1810 if (VerifyDuringGC) {
1811 HandleMark hm; // handle scope
1812 g1h->prepare_for_verify();
1813 Universe::verify(VerifyOption_G1UsePrevMarking, "During GC (after)");
1814 }
1815
1816 g1h->check_bitmaps("Cleanup End");
1817
1818 g1h->verify_region_sets_optional();
1819
1820 // We need to make this be a "collection" so any collection pause that
1821 // races with it goes around and waits for completeCleanup to finish.
1822 g1h->increment_total_collections();
1823
1824 // Clean out dead classes and update Metaspace sizes.
1825 if (ClassUnloadingWithConcurrentMark) {
1826 ClassLoaderDataGraph::purge();
1827 }
1828 MetaspaceGC::compute_new_size();
1829
1830 // We reclaimed old regions so we should calculate the sizes to make
1831 // sure we update the old gen/space data.
1832 g1h->g1mm()->update_sizes();
1833 g1h->allocation_context_stats().update_after_mark();
1834
1835 g1h->trace_heap_after_concurrent_cycle();
1836 }
1837
1838 void ConcurrentMark::completeCleanup() {
1839 if (has_aborted()) return;
1840
1841 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1842
1843 _cleanup_list.verify_optional();
1844 FreeRegionList tmp_free_list("Tmp Free List");
1845
1846 log_develop(gc, freelist)("G1ConcRegionFreeing [complete cleanup] : "
1847 "cleanup list has %u entries",
1848 _cleanup_list.length());
1849
1850 // No one else should be accessing the _cleanup_list at this point,
1851 // so it is not necessary to take any locks
1852 while (!_cleanup_list.is_empty()) {
1853 HeapRegion* hr = _cleanup_list.remove_region(true /* from_head */);
1854 assert(hr != NULL, "Got NULL from a non-empty list");
1855 hr->par_clear();
1856 tmp_free_list.add_ordered(hr);
1857
1858 // Instead of adding one region at a time to the secondary_free_list,
1859 // we accumulate them in the local list and move them a few at a
1860 // time. This also cuts down on the number of notify_all() calls
1861 // we do during this process. We'll also append the local list when
1862 // _cleanup_list is empty (which means we just removed the last
1863 // region from the _cleanup_list).
1864 if ((tmp_free_list.length() % G1SecondaryFreeListAppendLength == 0) ||
1865 _cleanup_list.is_empty()) {
1866 log_develop(gc, freelist)("G1ConcRegionFreeing [complete cleanup] : "
1867 "appending %u entries to the secondary_free_list, "
1868 "cleanup list still has %u entries",
1869 tmp_free_list.length(),
1870 _cleanup_list.length());
1871
1872 {
1873 MutexLockerEx x(SecondaryFreeList_lock, Mutex::_no_safepoint_check_flag);
1874 g1h->secondary_free_list_add(&tmp_free_list);
1875 SecondaryFreeList_lock->notify_all();
1876 }
1877 #ifndef PRODUCT
1878 if (G1StressConcRegionFreeing) {
1879 for (uintx i = 0; i < G1StressConcRegionFreeingDelayMillis; ++i) {
1880 os::sleep(Thread::current(), (jlong) 1, false);
1881 }
1882 }
1883 #endif
1884 }
1885 }
1886 assert(tmp_free_list.is_empty(), "post-condition");
1887 }
1888
1889 // Supporting Object and Oop closures for reference discovery
1890 // and processing in during marking
1891
1892 bool G1CMIsAliveClosure::do_object_b(oop obj) {
1893 HeapWord* addr = (HeapWord*)obj;
1894 return addr != NULL &&
1895 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj));
1896 }
1897
1898 // 'Keep Alive' oop closure used by both serial parallel reference processing.
1899 // Uses the CMTask associated with a worker thread (for serial reference
1900 // processing the CMTask for worker 0 is used) to preserve (mark) and
1901 // trace referent objects.
1902 //
1903 // Using the CMTask and embedded local queues avoids having the worker
1904 // threads operating on the global mark stack. This reduces the risk
1905 // of overflowing the stack - which we would rather avoid at this late
1906 // state. Also using the tasks' local queues removes the potential
1907 // of the workers interfering with each other that could occur if
1908 // operating on the global stack.
1909
1910 class G1CMKeepAliveAndDrainClosure: public OopClosure {
1911 ConcurrentMark* _cm;
1912 CMTask* _task;
1913 int _ref_counter_limit;
1914 int _ref_counter;
1915 bool _is_serial;
1916 public:
1917 G1CMKeepAliveAndDrainClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
1918 _cm(cm), _task(task), _is_serial(is_serial),
1919 _ref_counter_limit(G1RefProcDrainInterval) {
1920 assert(_ref_counter_limit > 0, "sanity");
1921 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1922 _ref_counter = _ref_counter_limit;
1923 }
1924
1925 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
1926 virtual void do_oop( oop* p) { do_oop_work(p); }
1927
1928 template <class T> void do_oop_work(T* p) {
1929 if (!_cm->has_overflown()) {
1930 oop obj = oopDesc::load_decode_heap_oop(p);
1931 _task->deal_with_reference(obj);
1932 _ref_counter--;
1933
1934 if (_ref_counter == 0) {
1935 // We have dealt with _ref_counter_limit references, pushing them
1936 // and objects reachable from them on to the local stack (and
1937 // possibly the global stack). Call CMTask::do_marking_step() to
1938 // process these entries.
1939 //
1940 // We call CMTask::do_marking_step() in a loop, which we'll exit if
1941 // there's nothing more to do (i.e. we're done with the entries that
1942 // were pushed as a result of the CMTask::deal_with_reference() calls
1943 // above) or we overflow.
1944 //
1945 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
1946 // flag while there may still be some work to do. (See the comment at
1947 // the beginning of CMTask::do_marking_step() for those conditions -
1948 // one of which is reaching the specified time target.) It is only
1949 // when CMTask::do_marking_step() returns without setting the
1950 // has_aborted() flag that the marking step has completed.
1951 do {
1952 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1953 _task->do_marking_step(mark_step_duration_ms,
1954 false /* do_termination */,
1955 _is_serial);
1956 } while (_task->has_aborted() && !_cm->has_overflown());
1957 _ref_counter = _ref_counter_limit;
1958 }
1959 }
1960 }
1961 };
1962
1963 // 'Drain' oop closure used by both serial and parallel reference processing.
1964 // Uses the CMTask associated with a given worker thread (for serial
1965 // reference processing the CMtask for worker 0 is used). Calls the
1966 // do_marking_step routine, with an unbelievably large timeout value,
1967 // to drain the marking data structures of the remaining entries
1968 // added by the 'keep alive' oop closure above.
1969
1970 class G1CMDrainMarkingStackClosure: public VoidClosure {
1971 ConcurrentMark* _cm;
1972 CMTask* _task;
1973 bool _is_serial;
1974 public:
1975 G1CMDrainMarkingStackClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
1976 _cm(cm), _task(task), _is_serial(is_serial) {
1977 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
1978 }
1979
1980 void do_void() {
1981 do {
1982 // We call CMTask::do_marking_step() to completely drain the local
1983 // and global marking stacks of entries pushed by the 'keep alive'
1984 // oop closure (an instance of G1CMKeepAliveAndDrainClosure above).
1985 //
1986 // CMTask::do_marking_step() is called in a loop, which we'll exit
1987 // if there's nothing more to do (i.e. we've completely drained the
1988 // entries that were pushed as a a result of applying the 'keep alive'
1989 // closure to the entries on the discovered ref lists) or we overflow
1990 // the global marking stack.
1991 //
1992 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
1993 // flag while there may still be some work to do. (See the comment at
1994 // the beginning of CMTask::do_marking_step() for those conditions -
1995 // one of which is reaching the specified time target.) It is only
1996 // when CMTask::do_marking_step() returns without setting the
1997 // has_aborted() flag that the marking step has completed.
1998
1999 _task->do_marking_step(1000000000.0 /* something very large */,
2000 true /* do_termination */,
2001 _is_serial);
2002 } while (_task->has_aborted() && !_cm->has_overflown());
2003 }
2004 };
2005
2006 // Implementation of AbstractRefProcTaskExecutor for parallel
2007 // reference processing at the end of G1 concurrent marking
2008
2009 class G1CMRefProcTaskExecutor: public AbstractRefProcTaskExecutor {
2010 private:
2011 G1CollectedHeap* _g1h;
2012 ConcurrentMark* _cm;
2013 WorkGang* _workers;
2014 uint _active_workers;
2015
2016 public:
2017 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
2018 ConcurrentMark* cm,
2019 WorkGang* workers,
2020 uint n_workers) :
2021 _g1h(g1h), _cm(cm),
2022 _workers(workers), _active_workers(n_workers) { }
2023
2024 // Executes the given task using concurrent marking worker threads.
2025 virtual void execute(ProcessTask& task);
2026 virtual void execute(EnqueueTask& task);
2027 };
2028
2029 class G1CMRefProcTaskProxy: public AbstractGangTask {
2030 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
2031 ProcessTask& _proc_task;
2032 G1CollectedHeap* _g1h;
2033 ConcurrentMark* _cm;
2034
2035 public:
2036 G1CMRefProcTaskProxy(ProcessTask& proc_task,
2037 G1CollectedHeap* g1h,
2038 ConcurrentMark* cm) :
2039 AbstractGangTask("Process reference objects in parallel"),
2040 _proc_task(proc_task), _g1h(g1h), _cm(cm) {
2041 ReferenceProcessor* rp = _g1h->ref_processor_cm();
2042 assert(rp->processing_is_mt(), "shouldn't be here otherwise");
2043 }
2044
2045 virtual void work(uint worker_id) {
2046 ResourceMark rm;
2047 HandleMark hm;
2048 CMTask* task = _cm->task(worker_id);
2049 G1CMIsAliveClosure g1_is_alive(_g1h);
2050 G1CMKeepAliveAndDrainClosure g1_par_keep_alive(_cm, task, false /* is_serial */);
2051 G1CMDrainMarkingStackClosure g1_par_drain(_cm, task, false /* is_serial */);
2052
2053 _proc_task.work(worker_id, g1_is_alive, g1_par_keep_alive, g1_par_drain);
2054 }
2055 };
2056
2057 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
2058 assert(_workers != NULL, "Need parallel worker threads.");
2059 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2060
2061 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
2062
2063 // We need to reset the concurrency level before each
2064 // proxy task execution, so that the termination protocol
2065 // and overflow handling in CMTask::do_marking_step() knows
2066 // how many workers to wait for.
2067 _cm->set_concurrency(_active_workers);
2068 _workers->run_task(&proc_task_proxy);
2069 }
2070
2071 class G1CMRefEnqueueTaskProxy: public AbstractGangTask {
2072 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
2073 EnqueueTask& _enq_task;
2074
2075 public:
2076 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
2077 AbstractGangTask("Enqueue reference objects in parallel"),
2078 _enq_task(enq_task) { }
2079
2080 virtual void work(uint worker_id) {
2081 _enq_task.work(worker_id);
2082 }
2083 };
2084
2085 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
2086 assert(_workers != NULL, "Need parallel worker threads.");
2087 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2088
2089 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
2090
2091 // Not strictly necessary but...
2092 //
2093 // We need to reset the concurrency level before each
2094 // proxy task execution, so that the termination protocol
2095 // and overflow handling in CMTask::do_marking_step() knows
2096 // how many workers to wait for.
2097 _cm->set_concurrency(_active_workers);
2098 _workers->run_task(&enq_task_proxy);
2099 }
2100
2101 void ConcurrentMark::weakRefsWorkParallelPart(BoolObjectClosure* is_alive, bool purged_classes) {
2102 G1CollectedHeap::heap()->parallel_cleaning(is_alive, true, true, purged_classes);
2103 }
2104
2105 void ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) {
2106 if (has_overflown()) {
2107 // Skip processing the discovered references if we have
2108 // overflown the global marking stack. Reference objects
2109 // only get discovered once so it is OK to not
2110 // de-populate the discovered reference lists. We could have,
2111 // but the only benefit would be that, when marking restarts,
2112 // less reference objects are discovered.
2113 return;
2114 }
2115
2116 ResourceMark rm;
2117 HandleMark hm;
2118
2119 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2120
2121 // Is alive closure.
2122 G1CMIsAliveClosure g1_is_alive(g1h);
2123
2124 // Inner scope to exclude the cleaning of the string and symbol
2125 // tables from the displayed time.
2126 {
2127 GCTraceTime(Debug, gc) trace("GC ref-proc", g1h->gc_timer_cm());
2128
2129 ReferenceProcessor* rp = g1h->ref_processor_cm();
2130
2131 // See the comment in G1CollectedHeap::ref_processing_init()
2132 // about how reference processing currently works in G1.
2133
2134 // Set the soft reference policy
2135 rp->setup_policy(clear_all_soft_refs);
2136 assert(_markStack.isEmpty(), "mark stack should be empty");
2137
2138 // Instances of the 'Keep Alive' and 'Complete GC' closures used
2139 // in serial reference processing. Note these closures are also
2140 // used for serially processing (by the the current thread) the
2141 // JNI references during parallel reference processing.
2142 //
2143 // These closures do not need to synchronize with the worker
2144 // threads involved in parallel reference processing as these
2145 // instances are executed serially by the current thread (e.g.
2146 // reference processing is not multi-threaded and is thus
2147 // performed by the current thread instead of a gang worker).
2148 //
2149 // The gang tasks involved in parallel reference processing create
2150 // their own instances of these closures, which do their own
2151 // synchronization among themselves.
2152 G1CMKeepAliveAndDrainClosure g1_keep_alive(this, task(0), true /* is_serial */);
2153 G1CMDrainMarkingStackClosure g1_drain_mark_stack(this, task(0), true /* is_serial */);
2154
2155 // We need at least one active thread. If reference processing
2156 // is not multi-threaded we use the current (VMThread) thread,
2157 // otherwise we use the work gang from the G1CollectedHeap and
2158 // we utilize all the worker threads we can.
2159 bool processing_is_mt = rp->processing_is_mt();
2160 uint active_workers = (processing_is_mt ? g1h->workers()->active_workers() : 1U);
2161 active_workers = MAX2(MIN2(active_workers, _max_worker_id), 1U);
2162
2163 // Parallel processing task executor.
2164 G1CMRefProcTaskExecutor par_task_executor(g1h, this,
2165 g1h->workers(), active_workers);
2166 AbstractRefProcTaskExecutor* executor = (processing_is_mt ? &par_task_executor : NULL);
2167
2168 // Set the concurrency level. The phase was already set prior to
2169 // executing the remark task.
2170 set_concurrency(active_workers);
2171
2172 // Set the degree of MT processing here. If the discovery was done MT,
2173 // the number of threads involved during discovery could differ from
2174 // the number of active workers. This is OK as long as the discovered
2175 // Reference lists are balanced (see balance_all_queues() and balance_queues()).
2176 rp->set_active_mt_degree(active_workers);
2177
2178 // Process the weak references.
2179 const ReferenceProcessorStats& stats =
2180 rp->process_discovered_references(&g1_is_alive,
2181 &g1_keep_alive,
2182 &g1_drain_mark_stack,
2183 executor,
2184 g1h->gc_timer_cm());
2185 g1h->gc_tracer_cm()->report_gc_reference_stats(stats);
2186
2187 // The do_oop work routines of the keep_alive and drain_marking_stack
2188 // oop closures will set the has_overflown flag if we overflow the
2189 // global marking stack.
2190
2191 assert(_markStack.overflow() || _markStack.isEmpty(),
2192 "mark stack should be empty (unless it overflowed)");
2193
2194 if (_markStack.overflow()) {
2195 // This should have been done already when we tried to push an
2196 // entry on to the global mark stack. But let's do it again.
2197 set_has_overflown();
2198 }
2199
2200 assert(rp->num_q() == active_workers, "why not");
2201
2202 rp->enqueue_discovered_references(executor);
2203
2204 rp->verify_no_references_recorded();
2205 assert(!rp->discovery_enabled(), "Post condition");
2206 }
2207
2208 if (has_overflown()) {
2209 // We can not trust g1_is_alive if the marking stack overflowed
2210 return;
2211 }
2212
2213 assert(_markStack.isEmpty(), "Marking should have completed");
2214
2215 // Unload Klasses, String, Symbols, Code Cache, etc.
2216 {
2217 GCTraceTime(Debug, gc) trace("Unloading", g1h->gc_timer_cm());
2218
2219 if (ClassUnloadingWithConcurrentMark) {
2220 bool purged_classes;
2221
2222 {
2223 GCTraceTime(Trace, gc) trace("System Dictionary Unloading", g1h->gc_timer_cm());
2224 purged_classes = SystemDictionary::do_unloading(&g1_is_alive, false /* Defer klass cleaning */);
2225 }
2226
2227 {
2228 GCTraceTime(Trace, gc) trace("Parallel Unloading", g1h->gc_timer_cm());
2229 weakRefsWorkParallelPart(&g1_is_alive, purged_classes);
2230 }
2231 }
2232
2233 if (G1StringDedup::is_enabled()) {
2234 GCTraceTime(Trace, gc) trace("String Deduplication Unlink", g1h->gc_timer_cm());
2235 G1StringDedup::unlink(&g1_is_alive);
2236 }
2237 }
2238 }
2239
2240 void ConcurrentMark::swapMarkBitMaps() {
2241 CMBitMapRO* temp = _prevMarkBitMap;
2242 _prevMarkBitMap = (CMBitMapRO*)_nextMarkBitMap;
2243 _nextMarkBitMap = (CMBitMap*) temp;
2244 }
2245
2246 // Closure for marking entries in SATB buffers.
2247 class CMSATBBufferClosure : public SATBBufferClosure {
2248 private:
2249 CMTask* _task;
2250 G1CollectedHeap* _g1h;
2251
2252 // This is very similar to CMTask::deal_with_reference, but with
2253 // more relaxed requirements for the argument, so this must be more
2254 // circumspect about treating the argument as an object.
2255 void do_entry(void* entry) const {
2256 _task->increment_refs_reached();
2257 HeapRegion* hr = _g1h->heap_region_containing(entry);
2258 if (entry < hr->next_top_at_mark_start()) {
2259 // Until we get here, we don't know whether entry refers to a valid
2260 // object; it could instead have been a stale reference.
2261 oop obj = static_cast<oop>(entry);
2262 assert(obj->is_oop(true /* ignore mark word */),
2263 "Invalid oop in SATB buffer: " PTR_FORMAT, p2i(obj));
2264 _task->make_reference_grey(obj, hr);
2265 }
2266 }
2267
2268 public:
2269 CMSATBBufferClosure(CMTask* task, G1CollectedHeap* g1h)
2270 : _task(task), _g1h(g1h) { }
2271
2272 virtual void do_buffer(void** buffer, size_t size) {
2273 for (size_t i = 0; i < size; ++i) {
2274 do_entry(buffer[i]);
2275 }
2276 }
2277 };
2278
2279 class G1RemarkThreadsClosure : public ThreadClosure {
2280 CMSATBBufferClosure _cm_satb_cl;
2281 G1CMOopClosure _cm_cl;
2282 MarkingCodeBlobClosure _code_cl;
2283 int _thread_parity;
2284
2285 public:
2286 G1RemarkThreadsClosure(G1CollectedHeap* g1h, CMTask* task) :
2287 _cm_satb_cl(task, g1h),
2288 _cm_cl(g1h, g1h->concurrent_mark(), task),
2289 _code_cl(&_cm_cl, !CodeBlobToOopClosure::FixRelocations),
2290 _thread_parity(Threads::thread_claim_parity()) {}
2291
2292 void do_thread(Thread* thread) {
2293 if (thread->is_Java_thread()) {
2294 if (thread->claim_oops_do(true, _thread_parity)) {
2295 JavaThread* jt = (JavaThread*)thread;
2296
2297 // In theory it should not be neccessary to explicitly walk the nmethods to find roots for concurrent marking
2298 // however the liveness of oops reachable from nmethods have very complex lifecycles:
2299 // * Alive if on the stack of an executing method
2300 // * Weakly reachable otherwise
2301 // Some objects reachable from nmethods, such as the class loader (or klass_holder) of the receiver should be
2302 // live by the SATB invariant but other oops recorded in nmethods may behave differently.
2303 jt->nmethods_do(&_code_cl);
2304
2305 jt->satb_mark_queue().apply_closure_and_empty(&_cm_satb_cl);
2306 }
2307 } else if (thread->is_VM_thread()) {
2308 if (thread->claim_oops_do(true, _thread_parity)) {
2309 JavaThread::satb_mark_queue_set().shared_satb_queue()->apply_closure_and_empty(&_cm_satb_cl);
2310 }
2311 }
2312 }
2313 };
2314
2315 class CMRemarkTask: public AbstractGangTask {
2316 private:
2317 ConcurrentMark* _cm;
2318 public:
2319 void work(uint worker_id) {
2320 // Since all available tasks are actually started, we should
2321 // only proceed if we're supposed to be active.
2322 if (worker_id < _cm->active_tasks()) {
2323 CMTask* task = _cm->task(worker_id);
2324 task->record_start_time();
2325 {
2326 ResourceMark rm;
2327 HandleMark hm;
2328
2329 G1RemarkThreadsClosure threads_f(G1CollectedHeap::heap(), task);
2330 Threads::threads_do(&threads_f);
2331 }
2332
2333 do {
2334 task->do_marking_step(1000000000.0 /* something very large */,
2335 true /* do_termination */,
2336 false /* is_serial */);
2337 } while (task->has_aborted() && !_cm->has_overflown());
2338 // If we overflow, then we do not want to restart. We instead
2339 // want to abort remark and do concurrent marking again.
2340 task->record_end_time();
2341 }
2342 }
2343
2344 CMRemarkTask(ConcurrentMark* cm, uint active_workers) :
2345 AbstractGangTask("Par Remark"), _cm(cm) {
2346 _cm->terminator()->reset_for_reuse(active_workers);
2347 }
2348 };
2349
2350 void ConcurrentMark::checkpointRootsFinalWork() {
2351 ResourceMark rm;
2352 HandleMark hm;
2353 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2354
2355 GCTraceTime(Debug, gc) trace("Finalize Marking", g1h->gc_timer_cm());
2356
2357 g1h->ensure_parsability(false);
2358
2359 // this is remark, so we'll use up all active threads
2360 uint active_workers = g1h->workers()->active_workers();
2361 set_concurrency_and_phase(active_workers, false /* concurrent */);
2362 // Leave _parallel_marking_threads at it's
2363 // value originally calculated in the ConcurrentMark
2364 // constructor and pass values of the active workers
2365 // through the gang in the task.
2366
2367 {
2368 StrongRootsScope srs(active_workers);
2369
2370 CMRemarkTask remarkTask(this, active_workers);
2371 // We will start all available threads, even if we decide that the
2372 // active_workers will be fewer. The extra ones will just bail out
2373 // immediately.
2374 g1h->workers()->run_task(&remarkTask);
2375 }
2376
2377 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2378 guarantee(has_overflown() ||
2379 satb_mq_set.completed_buffers_num() == 0,
2380 "Invariant: has_overflown = %s, num buffers = %d",
2381 BOOL_TO_STR(has_overflown()),
2382 satb_mq_set.completed_buffers_num());
2383
2384 print_stats();
2385 }
2386
2387 void ConcurrentMark::clearRangePrevBitmap(MemRegion mr) {
2388 // Note we are overriding the read-only view of the prev map here, via
2389 // the cast.
2390 ((CMBitMap*)_prevMarkBitMap)->clearRange(mr);
2391 }
2392
2393 void ConcurrentMark::clearRangeNextBitmap(MemRegion mr) {
2394 _nextMarkBitMap->clearRange(mr);
2395 }
2396
2397 HeapRegion*
2398 ConcurrentMark::claim_region(uint worker_id) {
2399 // "checkpoint" the finger
2400 HeapWord* finger = _finger;
2401
2402 // _heap_end will not change underneath our feet; it only changes at
2403 // yield points.
2404 while (finger < _heap_end) {
2405 assert(_g1h->is_in_g1_reserved(finger), "invariant");
2406
2407 HeapRegion* curr_region = _g1h->heap_region_containing(finger);
2408
2409 // Above heap_region_containing may return NULL as we always scan claim
2410 // until the end of the heap. In this case, just jump to the next region.
2411 HeapWord* end = curr_region != NULL ? curr_region->end() : finger + HeapRegion::GrainWords;
2412
2413 // Is the gap between reading the finger and doing the CAS too long?
2414 HeapWord* res = (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger);
2415 if (res == finger && curr_region != NULL) {
2416 // we succeeded
2417 HeapWord* bottom = curr_region->bottom();
2418 HeapWord* limit = curr_region->next_top_at_mark_start();
2419
2420 // notice that _finger == end cannot be guaranteed here since,
2421 // someone else might have moved the finger even further
2422 assert(_finger >= end, "the finger should have moved forward");
2423
2424 if (limit > bottom) {
2425 return curr_region;
2426 } else {
2427 assert(limit == bottom,
2428 "the region limit should be at bottom");
2429 // we return NULL and the caller should try calling
2430 // claim_region() again.
2431 return NULL;
2432 }
2433 } else {
2434 assert(_finger > finger, "the finger should have moved forward");
2435 // read it again
2436 finger = _finger;
2437 }
2438 }
2439
2440 return NULL;
2441 }
2442
2443 #ifndef PRODUCT
2444 class VerifyNoCSetOops VALUE_OBJ_CLASS_SPEC {
2445 private:
2446 G1CollectedHeap* _g1h;
2447 const char* _phase;
2448 int _info;
2449
2450 public:
2451 VerifyNoCSetOops(const char* phase, int info = -1) :
2452 _g1h(G1CollectedHeap::heap()),
2453 _phase(phase),
2454 _info(info)
2455 { }
2456
2457 void operator()(oop obj) const {
2458 guarantee(obj->is_oop(),
2459 "Non-oop " PTR_FORMAT ", phase: %s, info: %d",
2460 p2i(obj), _phase, _info);
2461 guarantee(!_g1h->obj_in_cs(obj),
2462 "obj: " PTR_FORMAT " in CSet, phase: %s, info: %d",
2463 p2i(obj), _phase, _info);
2464 }
2465 };
2466
2467 void ConcurrentMark::verify_no_cset_oops() {
2468 assert(SafepointSynchronize::is_at_safepoint(), "should be at a safepoint");
2469 if (!G1CollectedHeap::heap()->collector_state()->mark_in_progress()) {
2470 return;
2471 }
2472
2473 // Verify entries on the global mark stack
2474 _markStack.iterate(VerifyNoCSetOops("Stack"));
2475
2476 // Verify entries on the task queues
2477 for (uint i = 0; i < _max_worker_id; ++i) {
2478 CMTaskQueue* queue = _task_queues->queue(i);
2479 queue->iterate(VerifyNoCSetOops("Queue", i));
2480 }
2481
2482 // Verify the global finger
2483 HeapWord* global_finger = finger();
2484 if (global_finger != NULL && global_finger < _heap_end) {
2485 // Since we always iterate over all regions, we might get a NULL HeapRegion
2486 // here.
2487 HeapRegion* global_hr = _g1h->heap_region_containing(global_finger);
2488 guarantee(global_hr == NULL || global_finger == global_hr->bottom(),
2489 "global finger: " PTR_FORMAT " region: " HR_FORMAT,
2490 p2i(global_finger), HR_FORMAT_PARAMS(global_hr));
2491 }
2492
2493 // Verify the task fingers
2494 assert(parallel_marking_threads() <= _max_worker_id, "sanity");
2495 for (uint i = 0; i < parallel_marking_threads(); ++i) {
2496 CMTask* task = _tasks[i];
2497 HeapWord* task_finger = task->finger();
2498 if (task_finger != NULL && task_finger < _heap_end) {
2499 // See above note on the global finger verification.
2500 HeapRegion* task_hr = _g1h->heap_region_containing(task_finger);
2501 guarantee(task_hr == NULL || task_finger == task_hr->bottom() ||
2502 !task_hr->in_collection_set(),
2503 "task finger: " PTR_FORMAT " region: " HR_FORMAT,
2504 p2i(task_finger), HR_FORMAT_PARAMS(task_hr));
2505 }
2506 }
2507 }
2508 #endif // PRODUCT
2509
2510 // Aggregate the counting data that was constructed concurrently
2511 // with marking.
2512 class AggregateCountDataHRClosure: public HeapRegionClosure {
2513 G1CollectedHeap* _g1h;
2514 ConcurrentMark* _cm;
2515 CardTableModRefBS* _ct_bs;
2516 BitMap* _cm_card_bm;
2517 uint _max_worker_id;
2518
2519 public:
2520 AggregateCountDataHRClosure(G1CollectedHeap* g1h,
2521 BitMap* cm_card_bm,
2522 uint max_worker_id) :
2523 _g1h(g1h), _cm(g1h->concurrent_mark()),
2524 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
2525 _cm_card_bm(cm_card_bm), _max_worker_id(max_worker_id) { }
2526
2527 bool doHeapRegion(HeapRegion* hr) {
2528 HeapWord* start = hr->bottom();
2529 HeapWord* limit = hr->next_top_at_mark_start();
2530 HeapWord* end = hr->end();
2531
2532 assert(start <= limit && limit <= hr->top() && hr->top() <= hr->end(),
2533 "Preconditions not met - "
2534 "start: " PTR_FORMAT ", limit: " PTR_FORMAT ", "
2535 "top: " PTR_FORMAT ", end: " PTR_FORMAT,
2536 p2i(start), p2i(limit), p2i(hr->top()), p2i(hr->end()));
2537
2538 assert(hr->next_marked_bytes() == 0, "Precondition");
2539
2540 if (start == limit) {
2541 // NTAMS of this region has not been set so nothing to do.
2542 return false;
2543 }
2544
2545 // 'start' should be in the heap.
2546 assert(_g1h->is_in_g1_reserved(start) && _ct_bs->is_card_aligned(start), "sanity");
2547 // 'end' *may* be just beyond the end of the heap (if hr is the last region)
2548 assert(!_g1h->is_in_g1_reserved(end) || _ct_bs->is_card_aligned(end), "sanity");
2549
2550 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
2551 BitMap::idx_t limit_idx = _cm->card_bitmap_index_for(limit);
2552 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(end);
2553
2554 // If ntams is not card aligned then we bump card bitmap index
2555 // for limit so that we get the all the cards spanned by
2556 // the object ending at ntams.
2557 // Note: if this is the last region in the heap then ntams
2558 // could be actually just beyond the end of the the heap;
2559 // limit_idx will then correspond to a (non-existent) card
2560 // that is also outside the heap.
2561 if (_g1h->is_in_g1_reserved(limit) && !_ct_bs->is_card_aligned(limit)) {
2562 limit_idx += 1;
2563 }
2564
2565 assert(limit_idx <= end_idx, "or else use atomics");
2566
2567 // Aggregate the "stripe" in the count data associated with hr.
2568 uint hrm_index = hr->hrm_index();
2569 size_t marked_bytes = 0;
2570
2571 for (uint i = 0; i < _max_worker_id; i += 1) {
2572 size_t* marked_bytes_array = _cm->count_marked_bytes_array_for(i);
2573 BitMap* task_card_bm = _cm->count_card_bitmap_for(i);
2574
2575 // Fetch the marked_bytes in this region for task i and
2576 // add it to the running total for this region.
2577 marked_bytes += marked_bytes_array[hrm_index];
2578
2579 // Now union the bitmaps[0,max_worker_id)[start_idx..limit_idx)
2580 // into the global card bitmap.
2581 BitMap::idx_t scan_idx = task_card_bm->get_next_one_offset(start_idx, limit_idx);
2582
2583 while (scan_idx < limit_idx) {
2584 assert(task_card_bm->at(scan_idx) == true, "should be");
2585 _cm_card_bm->set_bit(scan_idx);
2586 assert(_cm_card_bm->at(scan_idx) == true, "should be");
2587
2588 // BitMap::get_next_one_offset() can handle the case when
2589 // its left_offset parameter is greater than its right_offset
2590 // parameter. It does, however, have an early exit if
2591 // left_offset == right_offset. So let's limit the value
2592 // passed in for left offset here.
2593 BitMap::idx_t next_idx = MIN2(scan_idx + 1, limit_idx);
2594 scan_idx = task_card_bm->get_next_one_offset(next_idx, limit_idx);
2595 }
2596 }
2597
2598 // Update the marked bytes for this region.
2599 hr->add_to_marked_bytes(marked_bytes);
2600
2601 // Next heap region
2602 return false;
2603 }
2604 };
2605
2606 class G1AggregateCountDataTask: public AbstractGangTask {
2607 protected:
2608 G1CollectedHeap* _g1h;
2609 ConcurrentMark* _cm;
2610 BitMap* _cm_card_bm;
2611 uint _max_worker_id;
2612 uint _active_workers;
2613 HeapRegionClaimer _hrclaimer;
2614
2615 public:
2616 G1AggregateCountDataTask(G1CollectedHeap* g1h,
2617 ConcurrentMark* cm,
2618 BitMap* cm_card_bm,
2619 uint max_worker_id,
2620 uint n_workers) :
2621 AbstractGangTask("Count Aggregation"),
2622 _g1h(g1h), _cm(cm), _cm_card_bm(cm_card_bm),
2623 _max_worker_id(max_worker_id),
2624 _active_workers(n_workers),
2625 _hrclaimer(_active_workers) {
2626 }
2627
2628 void work(uint worker_id) {
2629 AggregateCountDataHRClosure cl(_g1h, _cm_card_bm, _max_worker_id);
2630
2631 _g1h->heap_region_par_iterate(&cl, worker_id, &_hrclaimer);
2632 }
2633 };
2634
2635
2636 void ConcurrentMark::aggregate_count_data() {
2637 uint n_workers = _g1h->workers()->active_workers();
2638
2639 G1AggregateCountDataTask g1_par_agg_task(_g1h, this, &_card_bm,
2640 _max_worker_id, n_workers);
2641
2642 _g1h->workers()->run_task(&g1_par_agg_task);
2643 }
2644
2645 // Clear the per-worker arrays used to store the per-region counting data
2646 void ConcurrentMark::clear_all_count_data() {
2647 // Clear the global card bitmap - it will be filled during
2648 // liveness count aggregation (during remark) and the
2649 // final counting task.
2650 _card_bm.clear();
2651
2652 // Clear the global region bitmap - it will be filled as part
2653 // of the final counting task.
2654 _region_bm.clear();
2655
2656 uint max_regions = _g1h->max_regions();
2657 assert(_max_worker_id > 0, "uninitialized");
2658
2659 for (uint i = 0; i < _max_worker_id; i += 1) {
2660 BitMap* task_card_bm = count_card_bitmap_for(i);
2661 size_t* marked_bytes_array = count_marked_bytes_array_for(i);
2662
2663 assert(task_card_bm->size() == _card_bm.size(), "size mismatch");
2664 assert(marked_bytes_array != NULL, "uninitialized");
2665
2666 memset(marked_bytes_array, 0, (size_t) max_regions * sizeof(size_t));
2667 task_card_bm->clear();
2668 }
2669 }
2670
2671 void ConcurrentMark::print_stats() {
2672 LogHandle(gc, stats) log;
2673 if (!log.is_debug()) {
2674 return;
2675 }
2676 log.debug("---------------------------------------------------------------------");
2677 for (size_t i = 0; i < _active_tasks; ++i) {
2678 _tasks[i]->print_stats();
2679 log.debug("---------------------------------------------------------------------");
2680 }
2681 }
2682
2683 // abandon current marking iteration due to a Full GC
2684 void ConcurrentMark::abort() {
2685 if (!cmThread()->during_cycle() || _has_aborted) {
2686 // We haven't started a concurrent cycle or we have already aborted it. No need to do anything.
2687 return;
2688 }
2689
2690 // Clear all marks in the next bitmap for the next marking cycle. This will allow us to skip the next
2691 // concurrent bitmap clearing.
2692 _nextMarkBitMap->clearAll();
2693
2694 // Note we cannot clear the previous marking bitmap here
2695 // since VerifyDuringGC verifies the objects marked during
2696 // a full GC against the previous bitmap.
2697
2698 // Clear the liveness counting data
2699 clear_all_count_data();
2700 // Empty mark stack
2701 reset_marking_state();
2702 for (uint i = 0; i < _max_worker_id; ++i) {
2703 _tasks[i]->clear_region_fields();
2704 }
2705 _first_overflow_barrier_sync.abort();
2706 _second_overflow_barrier_sync.abort();
2707 _has_aborted = true;
2708
2709 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2710 satb_mq_set.abandon_partial_marking();
2711 // This can be called either during or outside marking, we'll read
2712 // the expected_active value from the SATB queue set.
2713 satb_mq_set.set_active_all_threads(
2714 false, /* new active value */
2715 satb_mq_set.is_active() /* expected_active */);
2716
2717 _g1h->trace_heap_after_concurrent_cycle();
2718 _g1h->register_concurrent_cycle_end();
2719 }
2720
2721 static void print_ms_time_info(const char* prefix, const char* name,
2722 NumberSeq& ns) {
2723 log_trace(gc, marking, stats, exit)("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
2724 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
2725 if (ns.num() > 0) {
2726 log_trace(gc, marking, stats, exit)("%s [std. dev = %8.2f ms, max = %8.2f ms]",
2727 prefix, ns.sd(), ns.maximum());
2728 }
2729 }
2730
2731 void ConcurrentMark::print_summary_info() {
2732 LogHandle(gc, marking, stats, exit) log;
2733 if (!log.is_trace()) {
2734 return;
2735 }
2736
2737 log.trace(" Concurrent marking:");
2738 print_ms_time_info(" ", "init marks", _init_times);
2739 print_ms_time_info(" ", "remarks", _remark_times);
2740 {
2741 print_ms_time_info(" ", "final marks", _remark_mark_times);
2742 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
2743
2744 }
2745 print_ms_time_info(" ", "cleanups", _cleanup_times);
2746 log.trace(" Final counting total time = %8.2f s (avg = %8.2f ms).",
2747 _total_counting_time, (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 / (double)_cleanup_times.num() : 0.0));
2748 if (G1ScrubRemSets) {
2749 log.trace(" RS scrub total time = %8.2f s (avg = %8.2f ms).",
2750 _total_rs_scrub_time, (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 / (double)_cleanup_times.num() : 0.0));
2751 }
2752 log.trace(" Total stop_world time = %8.2f s.",
2753 (_init_times.sum() + _remark_times.sum() + _cleanup_times.sum())/1000.0);
2754 log.trace(" Total concurrent time = %8.2f s (%8.2f s marking).",
2755 cmThread()->vtime_accum(), cmThread()->vtime_mark_accum());
2756 }
2757
2758 void ConcurrentMark::print_worker_threads_on(outputStream* st) const {
2759 _parallel_workers->print_worker_threads_on(st);
2760 }
2761
2762 void ConcurrentMark::print_on_error(outputStream* st) const {
2763 st->print_cr("Marking Bits (Prev, Next): (CMBitMap*) " PTR_FORMAT ", (CMBitMap*) " PTR_FORMAT,
2764 p2i(_prevMarkBitMap), p2i(_nextMarkBitMap));
2765 _prevMarkBitMap->print_on_error(st, " Prev Bits: ");
2766 _nextMarkBitMap->print_on_error(st, " Next Bits: ");
2767 }
2768
2769 // We take a break if someone is trying to stop the world.
2770 bool ConcurrentMark::do_yield_check(uint worker_id) {
2771 if (SuspendibleThreadSet::should_yield()) {
2772 if (worker_id == 0) {
2773 _g1h->g1_policy()->record_concurrent_pause();
2774 }
2775 SuspendibleThreadSet::yield();
2776 return true;
2777 } else {
2778 return false;
2779 }
2780 }
2781
2782 // Closure for iteration over bitmaps
2783 class CMBitMapClosure : public BitMapClosure {
2784 private:
2785 // the bitmap that is being iterated over
2786 CMBitMap* _nextMarkBitMap;
2787 ConcurrentMark* _cm;
2788 CMTask* _task;
2789
2790 public:
2791 CMBitMapClosure(CMTask *task, ConcurrentMark* cm, CMBitMap* nextMarkBitMap) :
2792 _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { }
2793
2794 bool do_bit(size_t offset) {
2795 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset);
2796 assert(_nextMarkBitMap->isMarked(addr), "invariant");
2797 assert( addr < _cm->finger(), "invariant");
2798 assert(addr >= _task->finger(), "invariant");
2799
2800 // We move that task's local finger along.
2801 _task->move_finger_to(addr);
2802
2803 _task->scan_object(oop(addr));
2804 // we only partially drain the local queue and global stack
2805 _task->drain_local_queue(true);
2806 _task->drain_global_stack(true);
2807
2808 // if the has_aborted flag has been raised, we need to bail out of
2809 // the iteration
2810 return !_task->has_aborted();
2811 }
2812 };
2813
2814 static ReferenceProcessor* get_cm_oop_closure_ref_processor(G1CollectedHeap* g1h) {
2815 ReferenceProcessor* result = NULL;
2816 if (G1UseConcMarkReferenceProcessing) {
2817 result = g1h->ref_processor_cm();
2818 assert(result != NULL, "should not be NULL");
2819 }
2820 return result;
2821 }
2822
2823 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
2824 ConcurrentMark* cm,
2825 CMTask* task)
2826 : MetadataAwareOopClosure(get_cm_oop_closure_ref_processor(g1h)),
2827 _g1h(g1h), _cm(cm), _task(task)
2828 { }
2829
2830 void CMTask::setup_for_region(HeapRegion* hr) {
2831 assert(hr != NULL,
2832 "claim_region() should have filtered out NULL regions");
2833 _curr_region = hr;
2834 _finger = hr->bottom();
2835 update_region_limit();
2836 }
2837
2838 void CMTask::update_region_limit() {
2839 HeapRegion* hr = _curr_region;
2840 HeapWord* bottom = hr->bottom();
2841 HeapWord* limit = hr->next_top_at_mark_start();
2842
2843 if (limit == bottom) {
2844 // The region was collected underneath our feet.
2845 // We set the finger to bottom to ensure that the bitmap
2846 // iteration that will follow this will not do anything.
2847 // (this is not a condition that holds when we set the region up,
2848 // as the region is not supposed to be empty in the first place)
2849 _finger = bottom;
2850 } else if (limit >= _region_limit) {
2851 assert(limit >= _finger, "peace of mind");
2852 } else {
2853 assert(limit < _region_limit, "only way to get here");
2854 // This can happen under some pretty unusual circumstances. An
2855 // evacuation pause empties the region underneath our feet (NTAMS
2856 // at bottom). We then do some allocation in the region (NTAMS
2857 // stays at bottom), followed by the region being used as a GC
2858 // alloc region (NTAMS will move to top() and the objects
2859 // originally below it will be grayed). All objects now marked in
2860 // the region are explicitly grayed, if below the global finger,
2861 // and we do not need in fact to scan anything else. So, we simply
2862 // set _finger to be limit to ensure that the bitmap iteration
2863 // doesn't do anything.
2864 _finger = limit;
2865 }
2866
2867 _region_limit = limit;
2868 }
2869
2870 void CMTask::giveup_current_region() {
2871 assert(_curr_region != NULL, "invariant");
2872 clear_region_fields();
2873 }
2874
2875 void CMTask::clear_region_fields() {
2876 // Values for these three fields that indicate that we're not
2877 // holding on to a region.
2878 _curr_region = NULL;
2879 _finger = NULL;
2880 _region_limit = NULL;
2881 }
2882
2883 void CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
2884 if (cm_oop_closure == NULL) {
2885 assert(_cm_oop_closure != NULL, "invariant");
2886 } else {
2887 assert(_cm_oop_closure == NULL, "invariant");
2888 }
2889 _cm_oop_closure = cm_oop_closure;
2890 }
2891
2892 void CMTask::reset(CMBitMap* nextMarkBitMap) {
2893 guarantee(nextMarkBitMap != NULL, "invariant");
2894 _nextMarkBitMap = nextMarkBitMap;
2895 clear_region_fields();
2896
2897 _calls = 0;
2898 _elapsed_time_ms = 0.0;
2899 _termination_time_ms = 0.0;
2900 _termination_start_time_ms = 0.0;
2901 }
2902
2903 bool CMTask::should_exit_termination() {
2904 regular_clock_call();
2905 // This is called when we are in the termination protocol. We should
2906 // quit if, for some reason, this task wants to abort or the global
2907 // stack is not empty (this means that we can get work from it).
2908 return !_cm->mark_stack_empty() || has_aborted();
2909 }
2910
2911 void CMTask::reached_limit() {
2912 assert(_words_scanned >= _words_scanned_limit ||
2913 _refs_reached >= _refs_reached_limit ,
2914 "shouldn't have been called otherwise");
2915 regular_clock_call();
2916 }
2917
2918 void CMTask::regular_clock_call() {
2919 if (has_aborted()) return;
2920
2921 // First, we need to recalculate the words scanned and refs reached
2922 // limits for the next clock call.
2923 recalculate_limits();
2924
2925 // During the regular clock call we do the following
2926
2927 // (1) If an overflow has been flagged, then we abort.
2928 if (_cm->has_overflown()) {
2929 set_has_aborted();
2930 return;
2931 }
2932
2933 // If we are not concurrent (i.e. we're doing remark) we don't need
2934 // to check anything else. The other steps are only needed during
2935 // the concurrent marking phase.
2936 if (!concurrent()) return;
2937
2938 // (2) If marking has been aborted for Full GC, then we also abort.
2939 if (_cm->has_aborted()) {
2940 set_has_aborted();
2941 return;
2942 }
2943
2944 double curr_time_ms = os::elapsedVTime() * 1000.0;
2945
2946 // (4) We check whether we should yield. If we have to, then we abort.
2947 if (SuspendibleThreadSet::should_yield()) {
2948 // We should yield. To do this we abort the task. The caller is
2949 // responsible for yielding.
2950 set_has_aborted();
2951 return;
2952 }
2953
2954 // (5) We check whether we've reached our time quota. If we have,
2955 // then we abort.
2956 double elapsed_time_ms = curr_time_ms - _start_time_ms;
2957 if (elapsed_time_ms > _time_target_ms) {
2958 set_has_aborted();
2959 _has_timed_out = true;
2960 return;
2961 }
2962
2963 // (6) Finally, we check whether there are enough completed STAB
2964 // buffers available for processing. If there are, we abort.
2965 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2966 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
2967 // we do need to process SATB buffers, we'll abort and restart
2968 // the marking task to do so
2969 set_has_aborted();
2970 return;
2971 }
2972 }
2973
2974 void CMTask::recalculate_limits() {
2975 _real_words_scanned_limit = _words_scanned + words_scanned_period;
2976 _words_scanned_limit = _real_words_scanned_limit;
2977
2978 _real_refs_reached_limit = _refs_reached + refs_reached_period;
2979 _refs_reached_limit = _real_refs_reached_limit;
2980 }
2981
2982 void CMTask::decrease_limits() {
2983 // This is called when we believe that we're going to do an infrequent
2984 // operation which will increase the per byte scanned cost (i.e. move
2985 // entries to/from the global stack). It basically tries to decrease the
2986 // scanning limit so that the clock is called earlier.
2987
2988 _words_scanned_limit = _real_words_scanned_limit -
2989 3 * words_scanned_period / 4;
2990 _refs_reached_limit = _real_refs_reached_limit -
2991 3 * refs_reached_period / 4;
2992 }
2993
2994 void CMTask::move_entries_to_global_stack() {
2995 // local array where we'll store the entries that will be popped
2996 // from the local queue
2997 oop buffer[global_stack_transfer_size];
2998
2999 int n = 0;
3000 oop obj;
3001 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) {
3002 buffer[n] = obj;
3003 ++n;
3004 }
3005
3006 if (n > 0) {
3007 // we popped at least one entry from the local queue
3008
3009 if (!_cm->mark_stack_push(buffer, n)) {
3010 set_has_aborted();
3011 }
3012 }
3013
3014 // this operation was quite expensive, so decrease the limits
3015 decrease_limits();
3016 }
3017
3018 void CMTask::get_entries_from_global_stack() {
3019 // local array where we'll store the entries that will be popped
3020 // from the global stack.
3021 oop buffer[global_stack_transfer_size];
3022 int n;
3023 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n);
3024 assert(n <= global_stack_transfer_size,
3025 "we should not pop more than the given limit");
3026 if (n > 0) {
3027 // yes, we did actually pop at least one entry
3028 for (int i = 0; i < n; ++i) {
3029 bool success = _task_queue->push(buffer[i]);
3030 // We only call this when the local queue is empty or under a
3031 // given target limit. So, we do not expect this push to fail.
3032 assert(success, "invariant");
3033 }
3034 }
3035
3036 // this operation was quite expensive, so decrease the limits
3037 decrease_limits();
3038 }
3039
3040 void CMTask::drain_local_queue(bool partially) {
3041 if (has_aborted()) return;
3042
3043 // Decide what the target size is, depending whether we're going to
3044 // drain it partially (so that other tasks can steal if they run out
3045 // of things to do) or totally (at the very end).
3046 size_t target_size;
3047 if (partially) {
3048 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
3049 } else {
3050 target_size = 0;
3051 }
3052
3053 if (_task_queue->size() > target_size) {
3054 oop obj;
3055 bool ret = _task_queue->pop_local(obj);
3056 while (ret) {
3057 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" );
3058 assert(!_g1h->is_on_master_free_list(
3059 _g1h->heap_region_containing((HeapWord*) obj)), "invariant");
3060
3061 scan_object(obj);
3062
3063 if (_task_queue->size() <= target_size || has_aborted()) {
3064 ret = false;
3065 } else {
3066 ret = _task_queue->pop_local(obj);
3067 }
3068 }
3069 }
3070 }
3071
3072 void CMTask::drain_global_stack(bool partially) {
3073 if (has_aborted()) return;
3074
3075 // We have a policy to drain the local queue before we attempt to
3076 // drain the global stack.
3077 assert(partially || _task_queue->size() == 0, "invariant");
3078
3079 // Decide what the target size is, depending whether we're going to
3080 // drain it partially (so that other tasks can steal if they run out
3081 // of things to do) or totally (at the very end). Notice that,
3082 // because we move entries from the global stack in chunks or
3083 // because another task might be doing the same, we might in fact
3084 // drop below the target. But, this is not a problem.
3085 size_t target_size;
3086 if (partially) {
3087 target_size = _cm->partial_mark_stack_size_target();
3088 } else {
3089 target_size = 0;
3090 }
3091
3092 if (_cm->mark_stack_size() > target_size) {
3093 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
3094 get_entries_from_global_stack();
3095 drain_local_queue(partially);
3096 }
3097 }
3098 }
3099
3100 // SATB Queue has several assumptions on whether to call the par or
3101 // non-par versions of the methods. this is why some of the code is
3102 // replicated. We should really get rid of the single-threaded version
3103 // of the code to simplify things.
3104 void CMTask::drain_satb_buffers() {
3105 if (has_aborted()) return;
3106
3107 // We set this so that the regular clock knows that we're in the
3108 // middle of draining buffers and doesn't set the abort flag when it
3109 // notices that SATB buffers are available for draining. It'd be
3110 // very counter productive if it did that. :-)
3111 _draining_satb_buffers = true;
3112
3113 CMSATBBufferClosure satb_cl(this, _g1h);
3114 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3115
3116 // This keeps claiming and applying the closure to completed buffers
3117 // until we run out of buffers or we need to abort.
3118 while (!has_aborted() &&
3119 satb_mq_set.apply_closure_to_completed_buffer(&satb_cl)) {
3120 regular_clock_call();
3121 }
3122
3123 _draining_satb_buffers = false;
3124
3125 assert(has_aborted() ||
3126 concurrent() ||
3127 satb_mq_set.completed_buffers_num() == 0, "invariant");
3128
3129 // again, this was a potentially expensive operation, decrease the
3130 // limits to get the regular clock call early
3131 decrease_limits();
3132 }
3133
3134 void CMTask::print_stats() {
3135 log_debug(gc, stats)("Marking Stats, task = %u, calls = %d",
3136 _worker_id, _calls);
3137 log_debug(gc, stats)(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
3138 _elapsed_time_ms, _termination_time_ms);
3139 log_debug(gc, stats)(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3140 _step_times_ms.num(), _step_times_ms.avg(),
3141 _step_times_ms.sd());
3142 log_debug(gc, stats)(" max = %1.2lfms, total = %1.2lfms",
3143 _step_times_ms.maximum(), _step_times_ms.sum());
3144 }
3145
3146 bool ConcurrentMark::try_stealing(uint worker_id, int* hash_seed, oop& obj) {
3147 return _task_queues->steal(worker_id, hash_seed, obj);
3148 }
3149
3150 /*****************************************************************************
3151
3152 The do_marking_step(time_target_ms, ...) method is the building
3153 block of the parallel marking framework. It can be called in parallel
3154 with other invocations of do_marking_step() on different tasks
3155 (but only one per task, obviously) and concurrently with the
3156 mutator threads, or during remark, hence it eliminates the need
3157 for two versions of the code. When called during remark, it will
3158 pick up from where the task left off during the concurrent marking
3159 phase. Interestingly, tasks are also claimable during evacuation
3160 pauses too, since do_marking_step() ensures that it aborts before
3161 it needs to yield.
3162
3163 The data structures that it uses to do marking work are the
3164 following:
3165
3166 (1) Marking Bitmap. If there are gray objects that appear only
3167 on the bitmap (this happens either when dealing with an overflow
3168 or when the initial marking phase has simply marked the roots
3169 and didn't push them on the stack), then tasks claim heap
3170 regions whose bitmap they then scan to find gray objects. A
3171 global finger indicates where the end of the last claimed region
3172 is. A local finger indicates how far into the region a task has
3173 scanned. The two fingers are used to determine how to gray an
3174 object (i.e. whether simply marking it is OK, as it will be
3175 visited by a task in the future, or whether it needs to be also
3176 pushed on a stack).
3177
3178 (2) Local Queue. The local queue of the task which is accessed
3179 reasonably efficiently by the task. Other tasks can steal from
3180 it when they run out of work. Throughout the marking phase, a
3181 task attempts to keep its local queue short but not totally
3182 empty, so that entries are available for stealing by other
3183 tasks. Only when there is no more work, a task will totally
3184 drain its local queue.
3185
3186 (3) Global Mark Stack. This handles local queue overflow. During
3187 marking only sets of entries are moved between it and the local
3188 queues, as access to it requires a mutex and more fine-grain
3189 interaction with it which might cause contention. If it
3190 overflows, then the marking phase should restart and iterate
3191 over the bitmap to identify gray objects. Throughout the marking
3192 phase, tasks attempt to keep the global mark stack at a small
3193 length but not totally empty, so that entries are available for
3194 popping by other tasks. Only when there is no more work, tasks
3195 will totally drain the global mark stack.
3196
3197 (4) SATB Buffer Queue. This is where completed SATB buffers are
3198 made available. Buffers are regularly removed from this queue
3199 and scanned for roots, so that the queue doesn't get too
3200 long. During remark, all completed buffers are processed, as
3201 well as the filled in parts of any uncompleted buffers.
3202
3203 The do_marking_step() method tries to abort when the time target
3204 has been reached. There are a few other cases when the
3205 do_marking_step() method also aborts:
3206
3207 (1) When the marking phase has been aborted (after a Full GC).
3208
3209 (2) When a global overflow (on the global stack) has been
3210 triggered. Before the task aborts, it will actually sync up with
3211 the other tasks to ensure that all the marking data structures
3212 (local queues, stacks, fingers etc.) are re-initialized so that
3213 when do_marking_step() completes, the marking phase can
3214 immediately restart.
3215
3216 (3) When enough completed SATB buffers are available. The
3217 do_marking_step() method only tries to drain SATB buffers right
3218 at the beginning. So, if enough buffers are available, the
3219 marking step aborts and the SATB buffers are processed at
3220 the beginning of the next invocation.
3221
3222 (4) To yield. when we have to yield then we abort and yield
3223 right at the end of do_marking_step(). This saves us from a lot
3224 of hassle as, by yielding we might allow a Full GC. If this
3225 happens then objects will be compacted underneath our feet, the
3226 heap might shrink, etc. We save checking for this by just
3227 aborting and doing the yield right at the end.
3228
3229 From the above it follows that the do_marking_step() method should
3230 be called in a loop (or, otherwise, regularly) until it completes.
3231
3232 If a marking step completes without its has_aborted() flag being
3233 true, it means it has completed the current marking phase (and
3234 also all other marking tasks have done so and have all synced up).
3235
3236 A method called regular_clock_call() is invoked "regularly" (in
3237 sub ms intervals) throughout marking. It is this clock method that
3238 checks all the abort conditions which were mentioned above and
3239 decides when the task should abort. A work-based scheme is used to
3240 trigger this clock method: when the number of object words the
3241 marking phase has scanned or the number of references the marking
3242 phase has visited reach a given limit. Additional invocations to
3243 the method clock have been planted in a few other strategic places
3244 too. The initial reason for the clock method was to avoid calling
3245 vtime too regularly, as it is quite expensive. So, once it was in
3246 place, it was natural to piggy-back all the other conditions on it
3247 too and not constantly check them throughout the code.
3248
3249 If do_termination is true then do_marking_step will enter its
3250 termination protocol.
3251
3252 The value of is_serial must be true when do_marking_step is being
3253 called serially (i.e. by the VMThread) and do_marking_step should
3254 skip any synchronization in the termination and overflow code.
3255 Examples include the serial remark code and the serial reference
3256 processing closures.
3257
3258 The value of is_serial must be false when do_marking_step is
3259 being called by any of the worker threads in a work gang.
3260 Examples include the concurrent marking code (CMMarkingTask),
3261 the MT remark code, and the MT reference processing closures.
3262
3263 *****************************************************************************/
3264
3265 void CMTask::do_marking_step(double time_target_ms,
3266 bool do_termination,
3267 bool is_serial) {
3268 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
3269 assert(concurrent() == _cm->concurrent(), "they should be the same");
3270
3271 G1CollectorPolicy* g1_policy = _g1h->g1_policy();
3272 assert(_task_queues != NULL, "invariant");
3273 assert(_task_queue != NULL, "invariant");
3274 assert(_task_queues->queue(_worker_id) == _task_queue, "invariant");
3275
3276 assert(!_claimed,
3277 "only one thread should claim this task at any one time");
3278
3279 // OK, this doesn't safeguard again all possible scenarios, as it is
3280 // possible for two threads to set the _claimed flag at the same
3281 // time. But it is only for debugging purposes anyway and it will
3282 // catch most problems.
3283 _claimed = true;
3284
3285 _start_time_ms = os::elapsedVTime() * 1000.0;
3286
3287 // If do_stealing is true then do_marking_step will attempt to
3288 // steal work from the other CMTasks. It only makes sense to
3289 // enable stealing when the termination protocol is enabled
3290 // and do_marking_step() is not being called serially.
3291 bool do_stealing = do_termination && !is_serial;
3292
3293 double diff_prediction_ms = _g1h->g1_policy()->predictor().get_new_prediction(&_marking_step_diffs_ms);
3294 _time_target_ms = time_target_ms - diff_prediction_ms;
3295
3296 // set up the variables that are used in the work-based scheme to
3297 // call the regular clock method
3298 _words_scanned = 0;
3299 _refs_reached = 0;
3300 recalculate_limits();
3301
3302 // clear all flags
3303 clear_has_aborted();
3304 _has_timed_out = false;
3305 _draining_satb_buffers = false;
3306
3307 ++_calls;
3308
3309 // Set up the bitmap and oop closures. Anything that uses them is
3310 // eventually called from this method, so it is OK to allocate these
3311 // statically.
3312 CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap);
3313 G1CMOopClosure cm_oop_closure(_g1h, _cm, this);
3314 set_cm_oop_closure(&cm_oop_closure);
3315
3316 if (_cm->has_overflown()) {
3317 // This can happen if the mark stack overflows during a GC pause
3318 // and this task, after a yield point, restarts. We have to abort
3319 // as we need to get into the overflow protocol which happens
3320 // right at the end of this task.
3321 set_has_aborted();
3322 }
3323
3324 // First drain any available SATB buffers. After this, we will not
3325 // look at SATB buffers before the next invocation of this method.
3326 // If enough completed SATB buffers are queued up, the regular clock
3327 // will abort this task so that it restarts.
3328 drain_satb_buffers();
3329 // ...then partially drain the local queue and the global stack
3330 drain_local_queue(true);
3331 drain_global_stack(true);
3332
3333 do {
3334 if (!has_aborted() && _curr_region != NULL) {
3335 // This means that we're already holding on to a region.
3336 assert(_finger != NULL, "if region is not NULL, then the finger "
3337 "should not be NULL either");
3338
3339 // We might have restarted this task after an evacuation pause
3340 // which might have evacuated the region we're holding on to
3341 // underneath our feet. Let's read its limit again to make sure
3342 // that we do not iterate over a region of the heap that
3343 // contains garbage (update_region_limit() will also move
3344 // _finger to the start of the region if it is found empty).
3345 update_region_limit();
3346 // We will start from _finger not from the start of the region,
3347 // as we might be restarting this task after aborting half-way
3348 // through scanning this region. In this case, _finger points to
3349 // the address where we last found a marked object. If this is a
3350 // fresh region, _finger points to start().
3351 MemRegion mr = MemRegion(_finger, _region_limit);
3352
3353 assert(!_curr_region->is_humongous() || mr.start() == _curr_region->bottom(),
3354 "humongous regions should go around loop once only");
3355
3356 // Some special cases:
3357 // If the memory region is empty, we can just give up the region.
3358 // If the current region is humongous then we only need to check
3359 // the bitmap for the bit associated with the start of the object,
3360 // scan the object if it's live, and give up the region.
3361 // Otherwise, let's iterate over the bitmap of the part of the region
3362 // that is left.
3363 // If the iteration is successful, give up the region.
3364 if (mr.is_empty()) {
3365 giveup_current_region();
3366 regular_clock_call();
3367 } else if (_curr_region->is_humongous() && mr.start() == _curr_region->bottom()) {
3368 if (_nextMarkBitMap->isMarked(mr.start())) {
3369 // The object is marked - apply the closure
3370 BitMap::idx_t offset = _nextMarkBitMap->heapWordToOffset(mr.start());
3371 bitmap_closure.do_bit(offset);
3372 }
3373 // Even if this task aborted while scanning the humongous object
3374 // we can (and should) give up the current region.
3375 giveup_current_region();
3376 regular_clock_call();
3377 } else if (_nextMarkBitMap->iterate(&bitmap_closure, mr)) {
3378 giveup_current_region();
3379 regular_clock_call();
3380 } else {
3381 assert(has_aborted(), "currently the only way to do so");
3382 // The only way to abort the bitmap iteration is to return
3383 // false from the do_bit() method. However, inside the
3384 // do_bit() method we move the _finger to point to the
3385 // object currently being looked at. So, if we bail out, we
3386 // have definitely set _finger to something non-null.
3387 assert(_finger != NULL, "invariant");
3388
3389 // Region iteration was actually aborted. So now _finger
3390 // points to the address of the object we last scanned. If we
3391 // leave it there, when we restart this task, we will rescan
3392 // the object. It is easy to avoid this. We move the finger by
3393 // enough to point to the next possible object header (the
3394 // bitmap knows by how much we need to move it as it knows its
3395 // granularity).
3396 assert(_finger < _region_limit, "invariant");
3397 HeapWord* new_finger = _nextMarkBitMap->nextObject(_finger);
3398 // Check if bitmap iteration was aborted while scanning the last object
3399 if (new_finger >= _region_limit) {
3400 giveup_current_region();
3401 } else {
3402 move_finger_to(new_finger);
3403 }
3404 }
3405 }
3406 // At this point we have either completed iterating over the
3407 // region we were holding on to, or we have aborted.
3408
3409 // We then partially drain the local queue and the global stack.
3410 // (Do we really need this?)
3411 drain_local_queue(true);
3412 drain_global_stack(true);
3413
3414 // Read the note on the claim_region() method on why it might
3415 // return NULL with potentially more regions available for
3416 // claiming and why we have to check out_of_regions() to determine
3417 // whether we're done or not.
3418 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
3419 // We are going to try to claim a new region. We should have
3420 // given up on the previous one.
3421 // Separated the asserts so that we know which one fires.
3422 assert(_curr_region == NULL, "invariant");
3423 assert(_finger == NULL, "invariant");
3424 assert(_region_limit == NULL, "invariant");
3425 HeapRegion* claimed_region = _cm->claim_region(_worker_id);
3426 if (claimed_region != NULL) {
3427 // Yes, we managed to claim one
3428 setup_for_region(claimed_region);
3429 assert(_curr_region == claimed_region, "invariant");
3430 }
3431 // It is important to call the regular clock here. It might take
3432 // a while to claim a region if, for example, we hit a large
3433 // block of empty regions. So we need to call the regular clock
3434 // method once round the loop to make sure it's called
3435 // frequently enough.
3436 regular_clock_call();
3437 }
3438
3439 if (!has_aborted() && _curr_region == NULL) {
3440 assert(_cm->out_of_regions(),
3441 "at this point we should be out of regions");
3442 }
3443 } while ( _curr_region != NULL && !has_aborted());
3444
3445 if (!has_aborted()) {
3446 // We cannot check whether the global stack is empty, since other
3447 // tasks might be pushing objects to it concurrently.
3448 assert(_cm->out_of_regions(),
3449 "at this point we should be out of regions");
3450 // Try to reduce the number of available SATB buffers so that
3451 // remark has less work to do.
3452 drain_satb_buffers();
3453 }
3454
3455 // Since we've done everything else, we can now totally drain the
3456 // local queue and global stack.
3457 drain_local_queue(false);
3458 drain_global_stack(false);
3459
3460 // Attempt at work stealing from other task's queues.
3461 if (do_stealing && !has_aborted()) {
3462 // We have not aborted. This means that we have finished all that
3463 // we could. Let's try to do some stealing...
3464
3465 // We cannot check whether the global stack is empty, since other
3466 // tasks might be pushing objects to it concurrently.
3467 assert(_cm->out_of_regions() && _task_queue->size() == 0,
3468 "only way to reach here");
3469 while (!has_aborted()) {
3470 oop obj;
3471 if (_cm->try_stealing(_worker_id, &_hash_seed, obj)) {
3472 assert(_nextMarkBitMap->isMarked((HeapWord*) obj),
3473 "any stolen object should be marked");
3474 scan_object(obj);
3475
3476 // And since we're towards the end, let's totally drain the
3477 // local queue and global stack.
3478 drain_local_queue(false);
3479 drain_global_stack(false);
3480 } else {
3481 break;
3482 }
3483 }
3484 }
3485
3486 // If we are about to wrap up and go into termination, check if we
3487 // should raise the overflow flag.
3488 if (do_termination && !has_aborted()) {
3489 if (_cm->force_overflow()->should_force()) {
3490 _cm->set_has_overflown();
3491 regular_clock_call();
3492 }
3493 }
3494
3495 // We still haven't aborted. Now, let's try to get into the
3496 // termination protocol.
3497 if (do_termination && !has_aborted()) {
3498 // We cannot check whether the global stack is empty, since other
3499 // tasks might be concurrently pushing objects on it.
3500 // Separated the asserts so that we know which one fires.
3501 assert(_cm->out_of_regions(), "only way to reach here");
3502 assert(_task_queue->size() == 0, "only way to reach here");
3503 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
3504
3505 // The CMTask class also extends the TerminatorTerminator class,
3506 // hence its should_exit_termination() method will also decide
3507 // whether to exit the termination protocol or not.
3508 bool finished = (is_serial ||
3509 _cm->terminator()->offer_termination(this));
3510 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
3511 _termination_time_ms +=
3512 termination_end_time_ms - _termination_start_time_ms;
3513
3514 if (finished) {
3515 // We're all done.
3516
3517 if (_worker_id == 0) {
3518 // let's allow task 0 to do this
3519 if (concurrent()) {
3520 assert(_cm->concurrent_marking_in_progress(), "invariant");
3521 // we need to set this to false before the next
3522 // safepoint. This way we ensure that the marking phase
3523 // doesn't observe any more heap expansions.
3524 _cm->clear_concurrent_marking_in_progress();
3525 }
3526 }
3527
3528 // We can now guarantee that the global stack is empty, since
3529 // all other tasks have finished. We separated the guarantees so
3530 // that, if a condition is false, we can immediately find out
3531 // which one.
3532 guarantee(_cm->out_of_regions(), "only way to reach here");
3533 guarantee(_cm->mark_stack_empty(), "only way to reach here");
3534 guarantee(_task_queue->size() == 0, "only way to reach here");
3535 guarantee(!_cm->has_overflown(), "only way to reach here");
3536 guarantee(!_cm->mark_stack_overflow(), "only way to reach here");
3537 } else {
3538 // Apparently there's more work to do. Let's abort this task. It
3539 // will restart it and we can hopefully find more things to do.
3540 set_has_aborted();
3541 }
3542 }
3543
3544 // Mainly for debugging purposes to make sure that a pointer to the
3545 // closure which was statically allocated in this frame doesn't
3546 // escape it by accident.
3547 set_cm_oop_closure(NULL);
3548 double end_time_ms = os::elapsedVTime() * 1000.0;
3549 double elapsed_time_ms = end_time_ms - _start_time_ms;
3550 // Update the step history.
3551 _step_times_ms.add(elapsed_time_ms);
3552
3553 if (has_aborted()) {
3554 // The task was aborted for some reason.
3555 if (_has_timed_out) {
3556 double diff_ms = elapsed_time_ms - _time_target_ms;
3557 // Keep statistics of how well we did with respect to hitting
3558 // our target only if we actually timed out (if we aborted for
3559 // other reasons, then the results might get skewed).
3560 _marking_step_diffs_ms.add(diff_ms);
3561 }
3562
3563 if (_cm->has_overflown()) {
3564 // This is the interesting one. We aborted because a global
3565 // overflow was raised. This means we have to restart the
3566 // marking phase and start iterating over regions. However, in
3567 // order to do this we have to make sure that all tasks stop
3568 // what they are doing and re-initialize in a safe manner. We
3569 // will achieve this with the use of two barrier sync points.
3570
3571 if (!is_serial) {
3572 // We only need to enter the sync barrier if being called
3573 // from a parallel context
3574 _cm->enter_first_sync_barrier(_worker_id);
3575
3576 // When we exit this sync barrier we know that all tasks have
3577 // stopped doing marking work. So, it's now safe to
3578 // re-initialize our data structures. At the end of this method,
3579 // task 0 will clear the global data structures.
3580 }
3581
3582 // We clear the local state of this task...
3583 clear_region_fields();
3584
3585 if (!is_serial) {
3586 // ...and enter the second barrier.
3587 _cm->enter_second_sync_barrier(_worker_id);
3588 }
3589 // At this point, if we're during the concurrent phase of
3590 // marking, everything has been re-initialized and we're
3591 // ready to restart.
3592 }
3593 }
3594
3595 _claimed = false;
3596 }
3597
3598 CMTask::CMTask(uint worker_id,
3599 ConcurrentMark* cm,
3600 size_t* marked_bytes,
3601 BitMap* card_bm,
3602 CMTaskQueue* task_queue,
3603 CMTaskQueueSet* task_queues)
3604 : _g1h(G1CollectedHeap::heap()),
3605 _worker_id(worker_id), _cm(cm),
3606 _claimed(false),
3607 _nextMarkBitMap(NULL), _hash_seed(17),
3608 _task_queue(task_queue),
3609 _task_queues(task_queues),
3610 _cm_oop_closure(NULL),
3611 _marked_bytes_array(marked_bytes),
3612 _card_bm(card_bm) {
3613 guarantee(task_queue != NULL, "invariant");
3614 guarantee(task_queues != NULL, "invariant");
3615
3616 _marking_step_diffs_ms.add(0.5);
3617 }
3618
3619 // These are formatting macros that are used below to ensure
3620 // consistent formatting. The *_H_* versions are used to format the
3621 // header for a particular value and they should be kept consistent
3622 // with the corresponding macro. Also note that most of the macros add
3623 // the necessary white space (as a prefix) which makes them a bit
3624 // easier to compose.
3625
3626 // All the output lines are prefixed with this string to be able to
3627 // identify them easily in a large log file.
3628 #define G1PPRL_LINE_PREFIX "###"
3629
3630 #define G1PPRL_ADDR_BASE_FORMAT " " PTR_FORMAT "-" PTR_FORMAT
3631 #ifdef _LP64
3632 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
3633 #else // _LP64
3634 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
3635 #endif // _LP64
3636
3637 // For per-region info
3638 #define G1PPRL_TYPE_FORMAT " %-4s"
3639 #define G1PPRL_TYPE_H_FORMAT " %4s"
3640 #define G1PPRL_BYTE_FORMAT " " SIZE_FORMAT_W(9)
3641 #define G1PPRL_BYTE_H_FORMAT " %9s"
3642 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
3643 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
3644
3645 // For summary info
3646 #define G1PPRL_SUM_ADDR_FORMAT(tag) " " tag ":" G1PPRL_ADDR_BASE_FORMAT
3647 #define G1PPRL_SUM_BYTE_FORMAT(tag) " " tag ": " SIZE_FORMAT
3648 #define G1PPRL_SUM_MB_FORMAT(tag) " " tag ": %1.2f MB"
3649 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag) " / %1.2f %%"
3650
3651 G1PrintRegionLivenessInfoClosure::
3652 G1PrintRegionLivenessInfoClosure(const char* phase_name)
3653 : _total_used_bytes(0), _total_capacity_bytes(0),
3654 _total_prev_live_bytes(0), _total_next_live_bytes(0),
3655 _hum_used_bytes(0), _hum_capacity_bytes(0),
3656 _hum_prev_live_bytes(0), _hum_next_live_bytes(0),
3657 _total_remset_bytes(0), _total_strong_code_roots_bytes(0) {
3658 G1CollectedHeap* g1h = G1CollectedHeap::heap();
3659 MemRegion g1_reserved = g1h->g1_reserved();
3660 double now = os::elapsedTime();
3661
3662 // Print the header of the output.
3663 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
3664 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX" HEAP"
3665 G1PPRL_SUM_ADDR_FORMAT("reserved")
3666 G1PPRL_SUM_BYTE_FORMAT("region-size"),
3667 p2i(g1_reserved.start()), p2i(g1_reserved.end()),
3668 HeapRegion::GrainBytes);
3669 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
3670 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3671 G1PPRL_TYPE_H_FORMAT
3672 G1PPRL_ADDR_BASE_H_FORMAT
3673 G1PPRL_BYTE_H_FORMAT
3674 G1PPRL_BYTE_H_FORMAT
3675 G1PPRL_BYTE_H_FORMAT
3676 G1PPRL_DOUBLE_H_FORMAT
3677 G1PPRL_BYTE_H_FORMAT
3678 G1PPRL_BYTE_H_FORMAT,
3679 "type", "address-range",
3680 "used", "prev-live", "next-live", "gc-eff",
3681 "remset", "code-roots");
3682 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3683 G1PPRL_TYPE_H_FORMAT
3684 G1PPRL_ADDR_BASE_H_FORMAT
3685 G1PPRL_BYTE_H_FORMAT
3686 G1PPRL_BYTE_H_FORMAT
3687 G1PPRL_BYTE_H_FORMAT
3688 G1PPRL_DOUBLE_H_FORMAT
3689 G1PPRL_BYTE_H_FORMAT
3690 G1PPRL_BYTE_H_FORMAT,
3691 "", "",
3692 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)",
3693 "(bytes)", "(bytes)");
3694 }
3695
3696 // It takes as a parameter a reference to one of the _hum_* fields, it
3697 // deduces the corresponding value for a region in a humongous region
3698 // series (either the region size, or what's left if the _hum_* field
3699 // is < the region size), and updates the _hum_* field accordingly.
3700 size_t G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* hum_bytes) {
3701 size_t bytes = 0;
3702 // The > 0 check is to deal with the prev and next live bytes which
3703 // could be 0.
3704 if (*hum_bytes > 0) {
3705 bytes = MIN2(HeapRegion::GrainBytes, *hum_bytes);
3706 *hum_bytes -= bytes;
3707 }
3708 return bytes;
3709 }
3710
3711 // It deduces the values for a region in a humongous region series
3712 // from the _hum_* fields and updates those accordingly. It assumes
3713 // that that _hum_* fields have already been set up from the "starts
3714 // humongous" region and we visit the regions in address order.
3715 void G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* used_bytes,
3716 size_t* capacity_bytes,
3717 size_t* prev_live_bytes,
3718 size_t* next_live_bytes) {
3719 assert(_hum_used_bytes > 0 && _hum_capacity_bytes > 0, "pre-condition");
3720 *used_bytes = get_hum_bytes(&_hum_used_bytes);
3721 *capacity_bytes = get_hum_bytes(&_hum_capacity_bytes);
3722 *prev_live_bytes = get_hum_bytes(&_hum_prev_live_bytes);
3723 *next_live_bytes = get_hum_bytes(&_hum_next_live_bytes);
3724 }
3725
3726 bool G1PrintRegionLivenessInfoClosure::doHeapRegion(HeapRegion* r) {
3727 const char* type = r->get_type_str();
3728 HeapWord* bottom = r->bottom();
3729 HeapWord* end = r->end();
3730 size_t capacity_bytes = r->capacity();
3731 size_t used_bytes = r->used();
3732 size_t prev_live_bytes = r->live_bytes();
3733 size_t next_live_bytes = r->next_live_bytes();
3734 double gc_eff = r->gc_efficiency();
3735 size_t remset_bytes = r->rem_set()->mem_size();
3736 size_t strong_code_roots_bytes = r->rem_set()->strong_code_roots_mem_size();
3737
3738 if (r->is_starts_humongous()) {
3739 assert(_hum_used_bytes == 0 && _hum_capacity_bytes == 0 &&
3740 _hum_prev_live_bytes == 0 && _hum_next_live_bytes == 0,
3741 "they should have been zeroed after the last time we used them");
3742 // Set up the _hum_* fields.
3743 _hum_capacity_bytes = capacity_bytes;
3744 _hum_used_bytes = used_bytes;
3745 _hum_prev_live_bytes = prev_live_bytes;
3746 _hum_next_live_bytes = next_live_bytes;
3747 get_hum_bytes(&used_bytes, &capacity_bytes,
3748 &prev_live_bytes, &next_live_bytes);
3749 end = bottom + HeapRegion::GrainWords;
3750 } else if (r->is_continues_humongous()) {
3751 get_hum_bytes(&used_bytes, &capacity_bytes,
3752 &prev_live_bytes, &next_live_bytes);
3753 assert(end == bottom + HeapRegion::GrainWords, "invariant");
3754 }
3755
3756 _total_used_bytes += used_bytes;
3757 _total_capacity_bytes += capacity_bytes;
3758 _total_prev_live_bytes += prev_live_bytes;
3759 _total_next_live_bytes += next_live_bytes;
3760 _total_remset_bytes += remset_bytes;
3761 _total_strong_code_roots_bytes += strong_code_roots_bytes;
3762
3763 // Print a line for this particular region.
3764 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3765 G1PPRL_TYPE_FORMAT
3766 G1PPRL_ADDR_BASE_FORMAT
3767 G1PPRL_BYTE_FORMAT
3768 G1PPRL_BYTE_FORMAT
3769 G1PPRL_BYTE_FORMAT
3770 G1PPRL_DOUBLE_FORMAT
3771 G1PPRL_BYTE_FORMAT
3772 G1PPRL_BYTE_FORMAT,
3773 type, p2i(bottom), p2i(end),
3774 used_bytes, prev_live_bytes, next_live_bytes, gc_eff,
3775 remset_bytes, strong_code_roots_bytes);
3776
3777 return false;
3778 }
3779
3780 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
3781 // add static memory usages to remembered set sizes
3782 _total_remset_bytes += HeapRegionRemSet::fl_mem_size() + HeapRegionRemSet::static_mem_size();
3783 // Print the footer of the output.
3784 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX);
3785 log_trace(gc, liveness)(G1PPRL_LINE_PREFIX
3786 " SUMMARY"
3787 G1PPRL_SUM_MB_FORMAT("capacity")
3788 G1PPRL_SUM_MB_PERC_FORMAT("used")
3789 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
3790 G1PPRL_SUM_MB_PERC_FORMAT("next-live")
3791 G1PPRL_SUM_MB_FORMAT("remset")
3792 G1PPRL_SUM_MB_FORMAT("code-roots"),
3793 bytes_to_mb(_total_capacity_bytes),
3794 bytes_to_mb(_total_used_bytes),
3795 perc(_total_used_bytes, _total_capacity_bytes),
3796 bytes_to_mb(_total_prev_live_bytes),
3797 perc(_total_prev_live_bytes, _total_capacity_bytes),
3798 bytes_to_mb(_total_next_live_bytes),
3799 perc(_total_next_live_bytes, _total_capacity_bytes),
3800 bytes_to_mb(_total_remset_bytes),
3801 bytes_to_mb(_total_strong_code_roots_bytes));
3802 }