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/g1ErgoVerbose.hpp"
35 #include "gc/g1/g1Log.hpp"
36 #include "gc/g1/g1OopClosures.inline.hpp"
37 #include "gc/g1/g1RemSet.hpp"
38 #include "gc/g1/g1StringDedup.hpp"
39 #include "gc/g1/heapRegion.inline.hpp"
40 #include "gc/g1/heapRegionManager.inline.hpp"
41 #include "gc/g1/heapRegionRemSet.hpp"
42 #include "gc/g1/heapRegionSet.inline.hpp"
43 #include "gc/g1/suspendibleThreadSet.hpp"
44 #include "gc/shared/gcTimer.hpp"
45 #include "gc/shared/gcTrace.hpp"
46 #include "gc/shared/gcTraceTime.hpp"
47 #include "gc/shared/genOopClosures.inline.hpp"
48 #include "gc/shared/referencePolicy.hpp"
49 #include "gc/shared/strongRootsScope.hpp"
50 #include "gc/shared/taskqueue.inline.hpp"
51 #include "gc/shared/vmGCOperations.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 #ifdef ASSERT
249 , _drain_in_progress(false)
250 , _drain_in_progress_yields(false)
251 #endif
252 {}
253
254 bool CMMarkStack::allocate(size_t capacity) {
255 // allocate a stack of the requisite depth
256 ReservedSpace rs(ReservedSpace::allocation_align_size_up(capacity * sizeof(oop)));
257 if (!rs.is_reserved()) {
258 warning("ConcurrentMark MarkStack allocation failure");
259 return false;
260 }
261 MemTracker::record_virtual_memory_type((address)rs.base(), mtGC);
262 if (!_virtual_space.initialize(rs, rs.size())) {
263 warning("ConcurrentMark MarkStack backing store failure");
264 // Release the virtual memory reserved for the marking stack
265 rs.release();
266 return false;
267 }
268 assert(_virtual_space.committed_size() == rs.size(),
269 "Didn't reserve backing store for all of ConcurrentMark stack?");
270 _base = (oop*) _virtual_space.low();
271 setEmpty();
272 _capacity = (jint) capacity;
273 _saved_index = -1;
274 _should_expand = false;
275 return true;
276 }
277
278 void CMMarkStack::expand() {
279 // Called, during remark, if we've overflown the marking stack during marking.
280 assert(isEmpty(), "stack should been emptied while handling overflow");
281 assert(_capacity <= (jint) MarkStackSizeMax, "stack bigger than permitted");
282 // Clear expansion flag
283 _should_expand = false;
284 if (_capacity == (jint) MarkStackSizeMax) {
285 if (PrintGCDetails && Verbose) {
286 gclog_or_tty->print_cr(" (benign) Can't expand marking stack capacity, at max size limit");
287 }
288 return;
289 }
290 // Double capacity if possible
291 jint new_capacity = MIN2(_capacity*2, (jint) MarkStackSizeMax);
292 // Do not give up existing stack until we have managed to
293 // get the double capacity that we desired.
294 ReservedSpace rs(ReservedSpace::allocation_align_size_up(new_capacity *
295 sizeof(oop)));
296 if (rs.is_reserved()) {
297 // Release the backing store associated with old stack
298 _virtual_space.release();
299 // Reinitialize virtual space for new stack
300 if (!_virtual_space.initialize(rs, rs.size())) {
301 fatal("Not enough swap for expanded marking stack capacity");
302 }
303 _base = (oop*)(_virtual_space.low());
304 _index = 0;
305 _capacity = new_capacity;
306 } else {
307 if (PrintGCDetails && Verbose) {
308 // Failed to double capacity, continue;
309 gclog_or_tty->print(" (benign) Failed to expand marking stack capacity from "
310 SIZE_FORMAT"K to " SIZE_FORMAT"K",
311 _capacity / K, new_capacity / K);
312 }
313 }
314 }
315
316 void CMMarkStack::set_should_expand() {
317 // If we're resetting the marking state because of an
318 // marking stack overflow, record that we should, if
319 // possible, expand the stack.
320 _should_expand = _cm->has_overflown();
321 }
322
323 CMMarkStack::~CMMarkStack() {
324 if (_base != NULL) {
325 _base = NULL;
326 _virtual_space.release();
327 }
328 }
329
330 void CMMarkStack::par_push_arr(oop* ptr_arr, int n) {
331 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
332 jint start = _index;
333 jint next_index = start + n;
334 if (next_index > _capacity) {
335 _overflow = true;
336 return;
337 }
338 // Otherwise.
339 _index = next_index;
340 for (int i = 0; i < n; i++) {
341 int ind = start + i;
342 assert(ind < _capacity, "By overflow test above.");
343 _base[ind] = ptr_arr[i];
344 }
345 }
346
347 bool CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) {
348 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
349 jint index = _index;
350 if (index == 0) {
351 *n = 0;
352 return false;
353 } else {
354 int k = MIN2(max, index);
355 jint new_ind = index - k;
356 for (int j = 0; j < k; j++) {
357 ptr_arr[j] = _base[new_ind + j];
358 }
359 _index = new_ind;
360 *n = k;
361 return true;
362 }
363 }
364
365 template<class OopClosureClass>
366 bool CMMarkStack::drain(OopClosureClass* cl, CMBitMap* bm, bool yield_after) {
367 assert(!_drain_in_progress || !_drain_in_progress_yields || yield_after
368 || SafepointSynchronize::is_at_safepoint(),
369 "Drain recursion must be yield-safe.");
370 bool res = true;
371 debug_only(_drain_in_progress = true);
372 debug_only(_drain_in_progress_yields = yield_after);
373 while (!isEmpty()) {
374 oop newOop = pop();
375 assert(G1CollectedHeap::heap()->is_in_reserved(newOop), "Bad pop");
376 assert(newOop->is_oop(), "Expected an oop");
377 assert(bm == NULL || bm->isMarked((HeapWord*)newOop),
378 "only grey objects on this stack");
379 newOop->oop_iterate(cl);
380 if (yield_after && _cm->do_yield_check()) {
381 res = false;
382 break;
383 }
384 }
385 debug_only(_drain_in_progress = false);
386 return res;
387 }
388
389 void CMMarkStack::note_start_of_gc() {
390 assert(_saved_index == -1,
391 "note_start_of_gc()/end_of_gc() bracketed incorrectly");
392 _saved_index = _index;
393 }
394
395 void CMMarkStack::note_end_of_gc() {
396 // This is intentionally a guarantee, instead of an assert. If we
397 // accidentally add something to the mark stack during GC, it
398 // will be a correctness issue so it's better if we crash. we'll
399 // only check this once per GC anyway, so it won't be a performance
400 // issue in any way.
401 guarantee(_saved_index == _index,
402 err_msg("saved index: %d index: %d", _saved_index, _index));
403 _saved_index = -1;
404 }
405
406 void CMMarkStack::oops_do(OopClosure* f) {
407 assert(_saved_index == _index,
408 err_msg("saved index: %d index: %d", _saved_index, _index));
409 for (int i = 0; i < _index; i += 1) {
410 f->do_oop(&_base[i]);
411 }
412 }
413
414 CMRootRegions::CMRootRegions() :
415 _young_list(NULL), _cm(NULL), _scan_in_progress(false),
416 _should_abort(false), _next_survivor(NULL) { }
417
418 void CMRootRegions::init(G1CollectedHeap* g1h, ConcurrentMark* cm) {
419 _young_list = g1h->young_list();
420 _cm = cm;
421 }
422
423 void CMRootRegions::prepare_for_scan() {
424 assert(!scan_in_progress(), "pre-condition");
425
426 // Currently, only survivors can be root regions.
427 assert(_next_survivor == NULL, "pre-condition");
428 _next_survivor = _young_list->first_survivor_region();
429 _scan_in_progress = (_next_survivor != NULL);
430 _should_abort = false;
431 }
432
433 HeapRegion* CMRootRegions::claim_next() {
434 if (_should_abort) {
435 // If someone has set the should_abort flag, we return NULL to
436 // force the caller to bail out of their loop.
437 return NULL;
438 }
439
440 // Currently, only survivors can be root regions.
441 HeapRegion* res = _next_survivor;
442 if (res != NULL) {
443 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
444 // Read it again in case it changed while we were waiting for the lock.
445 res = _next_survivor;
446 if (res != NULL) {
447 if (res == _young_list->last_survivor_region()) {
448 // We just claimed the last survivor so store NULL to indicate
449 // that we're done.
450 _next_survivor = NULL;
451 } else {
452 _next_survivor = res->get_next_young_region();
453 }
454 } else {
455 // Someone else claimed the last survivor while we were trying
456 // to take the lock so nothing else to do.
457 }
458 }
459 assert(res == NULL || res->is_survivor(), "post-condition");
460
461 return res;
462 }
463
464 void CMRootRegions::scan_finished() {
465 assert(scan_in_progress(), "pre-condition");
466
467 // Currently, only survivors can be root regions.
468 if (!_should_abort) {
469 assert(_next_survivor == NULL, "we should have claimed all survivors");
470 }
471 _next_survivor = NULL;
472
473 {
474 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
475 _scan_in_progress = false;
476 RootRegionScan_lock->notify_all();
477 }
478 }
479
480 bool CMRootRegions::wait_until_scan_finished() {
481 if (!scan_in_progress()) return false;
482
483 {
484 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
485 while (scan_in_progress()) {
486 RootRegionScan_lock->wait(Mutex::_no_safepoint_check_flag);
487 }
488 }
489 return true;
490 }
491
492 #ifdef _MSC_VER // the use of 'this' below gets a warning, make it go away
493 #pragma warning( disable:4355 ) // 'this' : used in base member initializer list
494 #endif // _MSC_VER
495
496 uint ConcurrentMark::scale_parallel_threads(uint n_par_threads) {
497 return MAX2((n_par_threads + 2) / 4, 1U);
498 }
499
500 ConcurrentMark::ConcurrentMark(G1CollectedHeap* g1h, G1RegionToSpaceMapper* prev_bitmap_storage, G1RegionToSpaceMapper* next_bitmap_storage) :
501 _g1h(g1h),
502 _markBitMap1(),
503 _markBitMap2(),
504 _parallel_marking_threads(0),
505 _max_parallel_marking_threads(0),
506 _sleep_factor(0.0),
507 _marking_task_overhead(1.0),
508 _cleanup_sleep_factor(0.0),
509 _cleanup_task_overhead(1.0),
510 _cleanup_list("Cleanup List"),
511 _region_bm((BitMap::idx_t)(g1h->max_regions()), false /* in_resource_area*/),
512 _card_bm((g1h->reserved_region().byte_size() + CardTableModRefBS::card_size - 1) >>
513 CardTableModRefBS::card_shift,
514 false /* in_resource_area*/),
515
516 _prevMarkBitMap(&_markBitMap1),
517 _nextMarkBitMap(&_markBitMap2),
518
519 _markStack(this),
520 // _finger set in set_non_marking_state
521
522 _max_worker_id(ParallelGCThreads),
523 // _active_tasks set in set_non_marking_state
524 // _tasks set inside the constructor
525 _task_queues(new CMTaskQueueSet((int) _max_worker_id)),
526 _terminator(ParallelTaskTerminator((int) _max_worker_id, _task_queues)),
527
528 _has_overflown(false),
529 _concurrent(false),
530 _has_aborted(false),
531 _aborted_gc_id(GCId::undefined()),
532 _restart_for_overflow(false),
533 _concurrent_marking_in_progress(false),
534
535 // _verbose_level set below
536
537 _init_times(),
538 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(),
539 _cleanup_times(),
540 _total_counting_time(0.0),
541 _total_rs_scrub_time(0.0),
542
543 _parallel_workers(NULL),
544
545 _count_card_bitmaps(NULL),
546 _count_marked_bytes(NULL),
547 _completed_initialization(false) {
548 CMVerboseLevel verbose_level = (CMVerboseLevel) G1MarkingVerboseLevel;
549 if (verbose_level < no_verbose) {
550 verbose_level = no_verbose;
551 }
552 if (verbose_level > high_verbose) {
553 verbose_level = high_verbose;
554 }
555 _verbose_level = verbose_level;
556
557 if (verbose_low()) {
558 gclog_or_tty->print_cr("[global] init, heap start = "PTR_FORMAT", "
559 "heap end = " PTR_FORMAT, p2i(_heap_start), p2i(_heap_end));
560 }
561
562 _markBitMap1.initialize(g1h->reserved_region(), prev_bitmap_storage);
563 _markBitMap2.initialize(g1h->reserved_region(), next_bitmap_storage);
564
565 // Create & start a ConcurrentMark thread.
566 _cmThread = new ConcurrentMarkThread(this);
567 assert(cmThread() != NULL, "CM Thread should have been created");
568 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm");
569 if (_cmThread->osthread() == NULL) {
570 vm_shutdown_during_initialization("Could not create ConcurrentMarkThread");
571 }
572
573 assert(CGC_lock != NULL, "Where's the CGC_lock?");
574 assert(_markBitMap1.covers(g1h->reserved_region()), "_markBitMap1 inconsistency");
575 assert(_markBitMap2.covers(g1h->reserved_region()), "_markBitMap2 inconsistency");
576
577 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
578 satb_qs.set_buffer_size(G1SATBBufferSize);
579
580 _root_regions.init(_g1h, this);
581
582 if (ConcGCThreads > ParallelGCThreads) {
583 warning("Can't have more ConcGCThreads (%u) "
584 "than ParallelGCThreads (%u).",
585 ConcGCThreads, ParallelGCThreads);
586 return;
587 }
588 if (!FLAG_IS_DEFAULT(ConcGCThreads) && ConcGCThreads > 0) {
589 // Note: ConcGCThreads has precedence over G1MarkingOverheadPercent
590 // if both are set
591 _sleep_factor = 0.0;
592 _marking_task_overhead = 1.0;
593 } else if (G1MarkingOverheadPercent > 0) {
594 // We will calculate the number of parallel marking threads based
595 // on a target overhead with respect to the soft real-time goal
596 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
597 double overall_cm_overhead =
598 (double) MaxGCPauseMillis * marking_overhead /
599 (double) GCPauseIntervalMillis;
600 double cpu_ratio = 1.0 / (double) os::processor_count();
601 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio);
602 double marking_task_overhead =
603 overall_cm_overhead / marking_thread_num *
604 (double) os::processor_count();
605 double sleep_factor =
606 (1.0 - marking_task_overhead) / marking_task_overhead;
607
608 FLAG_SET_ERGO(uint, ConcGCThreads, (uint) marking_thread_num);
609 _sleep_factor = sleep_factor;
610 _marking_task_overhead = marking_task_overhead;
611 } else {
612 // Calculate the number of parallel marking threads by scaling
613 // the number of parallel GC threads.
614 uint marking_thread_num = scale_parallel_threads(ParallelGCThreads);
615 FLAG_SET_ERGO(uint, ConcGCThreads, marking_thread_num);
616 _sleep_factor = 0.0;
617 _marking_task_overhead = 1.0;
618 }
619
620 assert(ConcGCThreads > 0, "Should have been set");
621 _parallel_marking_threads = ConcGCThreads;
622 _max_parallel_marking_threads = _parallel_marking_threads;
623
624 if (parallel_marking_threads() > 1) {
625 _cleanup_task_overhead = 1.0;
626 } else {
627 _cleanup_task_overhead = marking_task_overhead();
628 }
629 _cleanup_sleep_factor =
630 (1.0 - cleanup_task_overhead()) / cleanup_task_overhead();
631
632 #if 0
633 gclog_or_tty->print_cr("Marking Threads %d", parallel_marking_threads());
634 gclog_or_tty->print_cr("CM Marking Task Overhead %1.4lf", marking_task_overhead());
635 gclog_or_tty->print_cr("CM Sleep Factor %1.4lf", sleep_factor());
636 gclog_or_tty->print_cr("CL Marking Task Overhead %1.4lf", cleanup_task_overhead());
637 gclog_or_tty->print_cr("CL Sleep Factor %1.4lf", cleanup_sleep_factor());
638 #endif
639
640 _parallel_workers = new FlexibleWorkGang("G1 Marker",
641 _max_parallel_marking_threads, false, true);
642 if (_parallel_workers == NULL) {
643 vm_exit_during_initialization("Failed necessary allocation.");
644 } else {
645 _parallel_workers->initialize_workers();
646 }
647
648 if (FLAG_IS_DEFAULT(MarkStackSize)) {
649 size_t mark_stack_size =
650 MIN2(MarkStackSizeMax,
651 MAX2(MarkStackSize, (size_t) (parallel_marking_threads() * TASKQUEUE_SIZE)));
652 // Verify that the calculated value for MarkStackSize is in range.
653 // It would be nice to use the private utility routine from Arguments.
654 if (!(mark_stack_size >= 1 && mark_stack_size <= MarkStackSizeMax)) {
655 warning("Invalid value calculated for MarkStackSize (" SIZE_FORMAT "): "
656 "must be between 1 and " SIZE_FORMAT,
657 mark_stack_size, MarkStackSizeMax);
658 return;
659 }
660 FLAG_SET_ERGO(size_t, MarkStackSize, mark_stack_size);
661 } else {
662 // Verify MarkStackSize is in range.
663 if (FLAG_IS_CMDLINE(MarkStackSize)) {
664 if (FLAG_IS_DEFAULT(MarkStackSizeMax)) {
665 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
666 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT "): "
667 "must be between 1 and " SIZE_FORMAT,
668 MarkStackSize, MarkStackSizeMax);
669 return;
670 }
671 } else if (FLAG_IS_CMDLINE(MarkStackSizeMax)) {
672 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
673 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT ")"
674 " or for MarkStackSizeMax (" SIZE_FORMAT ")",
675 MarkStackSize, MarkStackSizeMax);
676 return;
677 }
678 }
679 }
680 }
681
682 if (!_markStack.allocate(MarkStackSize)) {
683 warning("Failed to allocate CM marking stack");
684 return;
685 }
686
687 _tasks = NEW_C_HEAP_ARRAY(CMTask*, _max_worker_id, mtGC);
688 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_worker_id, mtGC);
689
690 _count_card_bitmaps = NEW_C_HEAP_ARRAY(BitMap, _max_worker_id, mtGC);
691 _count_marked_bytes = NEW_C_HEAP_ARRAY(size_t*, _max_worker_id, mtGC);
692
693 BitMap::idx_t card_bm_size = _card_bm.size();
694
695 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
696 _active_tasks = _max_worker_id;
697
698 uint max_regions = _g1h->max_regions();
699 for (uint i = 0; i < _max_worker_id; ++i) {
700 CMTaskQueue* task_queue = new CMTaskQueue();
701 task_queue->initialize();
702 _task_queues->register_queue(i, task_queue);
703
704 _count_card_bitmaps[i] = BitMap(card_bm_size, false);
705 _count_marked_bytes[i] = NEW_C_HEAP_ARRAY(size_t, max_regions, mtGC);
706
707 _tasks[i] = new CMTask(i, this,
708 _count_marked_bytes[i],
709 &_count_card_bitmaps[i],
710 task_queue, _task_queues);
711
712 _accum_task_vtime[i] = 0.0;
713 }
714
715 // Calculate the card number for the bottom of the heap. Used
716 // in biasing indexes into the accounting card bitmaps.
717 _heap_bottom_card_num =
718 intptr_t(uintptr_t(_g1h->reserved_region().start()) >>
719 CardTableModRefBS::card_shift);
720
721 // Clear all the liveness counting data
722 clear_all_count_data();
723
724 // so that the call below can read a sensible value
725 _heap_start = g1h->reserved_region().start();
726 set_non_marking_state();
727 _completed_initialization = true;
728 }
729
730 void ConcurrentMark::reset() {
731 // Starting values for these two. This should be called in a STW
732 // phase.
733 MemRegion reserved = _g1h->g1_reserved();
734 _heap_start = reserved.start();
735 _heap_end = reserved.end();
736
737 // Separated the asserts so that we know which one fires.
738 assert(_heap_start != NULL, "heap bounds should look ok");
739 assert(_heap_end != NULL, "heap bounds should look ok");
740 assert(_heap_start < _heap_end, "heap bounds should look ok");
741
742 // Reset all the marking data structures and any necessary flags
743 reset_marking_state();
744
745 if (verbose_low()) {
746 gclog_or_tty->print_cr("[global] resetting");
747 }
748
749 // We do reset all of them, since different phases will use
750 // different number of active threads. So, it's easiest to have all
751 // of them ready.
752 for (uint i = 0; i < _max_worker_id; ++i) {
753 _tasks[i]->reset(_nextMarkBitMap);
754 }
755
756 // we need this to make sure that the flag is on during the evac
757 // pause with initial mark piggy-backed
758 set_concurrent_marking_in_progress();
759 }
760
761
762 void ConcurrentMark::reset_marking_state(bool clear_overflow) {
763 _markStack.set_should_expand();
764 _markStack.setEmpty(); // Also clears the _markStack overflow flag
765 if (clear_overflow) {
766 clear_has_overflown();
767 } else {
768 assert(has_overflown(), "pre-condition");
769 }
770 _finger = _heap_start;
771
772 for (uint i = 0; i < _max_worker_id; ++i) {
773 CMTaskQueue* queue = _task_queues->queue(i);
774 queue->set_empty();
775 }
776 }
777
778 void ConcurrentMark::set_concurrency(uint active_tasks) {
779 assert(active_tasks <= _max_worker_id, "we should not have more");
780
781 _active_tasks = active_tasks;
782 // Need to update the three data structures below according to the
783 // number of active threads for this phase.
784 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
785 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
786 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
787 }
788
789 void ConcurrentMark::set_concurrency_and_phase(uint active_tasks, bool concurrent) {
790 set_concurrency(active_tasks);
791
792 _concurrent = concurrent;
793 // We propagate this to all tasks, not just the active ones.
794 for (uint i = 0; i < _max_worker_id; ++i)
795 _tasks[i]->set_concurrent(concurrent);
796
797 if (concurrent) {
798 set_concurrent_marking_in_progress();
799 } else {
800 // We currently assume that the concurrent flag has been set to
801 // false before we start remark. At this point we should also be
802 // in a STW phase.
803 assert(!concurrent_marking_in_progress(), "invariant");
804 assert(out_of_regions(),
805 err_msg("only way to get here: _finger: "PTR_FORMAT", _heap_end: "PTR_FORMAT,
806 p2i(_finger), p2i(_heap_end)));
807 }
808 }
809
810 void ConcurrentMark::set_non_marking_state() {
811 // We set the global marking state to some default values when we're
812 // not doing marking.
813 reset_marking_state();
814 _active_tasks = 0;
815 clear_concurrent_marking_in_progress();
816 }
817
818 ConcurrentMark::~ConcurrentMark() {
819 // The ConcurrentMark instance is never freed.
820 ShouldNotReachHere();
821 }
822
823 void ConcurrentMark::clearNextBitmap() {
824 G1CollectedHeap* g1h = G1CollectedHeap::heap();
825
826 // Make sure that the concurrent mark thread looks to still be in
827 // the current cycle.
828 guarantee(cmThread()->during_cycle(), "invariant");
829
830 // We are finishing up the current cycle by clearing the next
831 // marking bitmap and getting it ready for the next cycle. During
832 // this time no other cycle can start. So, let's make sure that this
833 // is the case.
834 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
835
836 ClearBitmapHRClosure cl(this, _nextMarkBitMap, true /* may_yield */);
837 ParClearNextMarkBitmapTask task(&cl, parallel_marking_threads(), true);
838 _parallel_workers->run_task(&task);
839
840 // Clear the liveness counting data. If the marking has been aborted, the abort()
841 // call already did that.
842 if (cl.complete()) {
843 clear_all_count_data();
844 }
845
846 // Repeat the asserts from above.
847 guarantee(cmThread()->during_cycle(), "invariant");
848 guarantee(!g1h->collector_state()->mark_in_progress(), "invariant");
849 }
850
851 class CheckBitmapClearHRClosure : public HeapRegionClosure {
852 CMBitMap* _bitmap;
853 bool _error;
854 public:
855 CheckBitmapClearHRClosure(CMBitMap* bitmap) : _bitmap(bitmap) {
856 }
857
858 virtual bool doHeapRegion(HeapRegion* r) {
859 // This closure can be called concurrently to the mutator, so we must make sure
860 // that the result of the getNextMarkedWordAddress() call is compared to the
861 // value passed to it as limit to detect any found bits.
862 // We can use the region's orig_end() for the limit and the comparison value
863 // as it always contains the "real" end of the region that never changes and
864 // has no side effects.
865 // Due to the latter, there can also be no problem with the compiler generating
866 // reloads of the orig_end() call.
867 HeapWord* end = r->orig_end();
868 return _bitmap->getNextMarkedWordAddress(r->bottom(), end) != end;
869 }
870 };
871
872 bool ConcurrentMark::nextMarkBitmapIsClear() {
873 CheckBitmapClearHRClosure cl(_nextMarkBitMap);
874 _g1h->heap_region_iterate(&cl);
875 return cl.complete();
876 }
877
878 class NoteStartOfMarkHRClosure: public HeapRegionClosure {
879 public:
880 bool doHeapRegion(HeapRegion* r) {
881 if (!r->is_continues_humongous()) {
882 r->note_start_of_marking();
883 }
884 return false;
885 }
886 };
887
888 void ConcurrentMark::checkpointRootsInitialPre() {
889 G1CollectedHeap* g1h = G1CollectedHeap::heap();
890 G1CollectorPolicy* g1p = g1h->g1_policy();
891
892 _has_aborted = false;
893
894 // Initialize marking structures. This has to be done in a STW phase.
895 reset();
896
897 // For each region note start of marking.
898 NoteStartOfMarkHRClosure startcl;
899 g1h->heap_region_iterate(&startcl);
900 }
901
902
903 void ConcurrentMark::checkpointRootsInitialPost() {
904 G1CollectedHeap* g1h = G1CollectedHeap::heap();
905
906 // If we force an overflow during remark, the remark operation will
907 // actually abort and we'll restart concurrent marking. If we always
908 // force an overflow during remark we'll never actually complete the
909 // marking phase. So, we initialize this here, at the start of the
910 // cycle, so that at the remaining overflow number will decrease at
911 // every remark and we'll eventually not need to cause one.
912 force_overflow_stw()->init();
913
914 // Start Concurrent Marking weak-reference discovery.
915 ReferenceProcessor* rp = g1h->ref_processor_cm();
916 // enable ("weak") refs discovery
917 rp->enable_discovery();
918 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
919
920 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
921 // This is the start of the marking cycle, we're expected all
922 // threads to have SATB queues with active set to false.
923 satb_mq_set.set_active_all_threads(true, /* new active value */
924 false /* expected_active */);
925
926 _root_regions.prepare_for_scan();
927
928 // update_g1_committed() will be called at the end of an evac pause
929 // when marking is on. So, it's also called at the end of the
930 // initial-mark pause to update the heap end, if the heap expands
931 // during it. No need to call it here.
932 }
933
934 /*
935 * Notice that in the next two methods, we actually leave the STS
936 * during the barrier sync and join it immediately afterwards. If we
937 * do not do this, the following deadlock can occur: one thread could
938 * be in the barrier sync code, waiting for the other thread to also
939 * sync up, whereas another one could be trying to yield, while also
940 * waiting for the other threads to sync up too.
941 *
942 * Note, however, that this code is also used during remark and in
943 * this case we should not attempt to leave / enter the STS, otherwise
944 * we'll either hit an assert (debug / fastdebug) or deadlock
945 * (product). So we should only leave / enter the STS if we are
946 * operating concurrently.
947 *
948 * Because the thread that does the sync barrier has left the STS, it
949 * is possible to be suspended for a Full GC or an evacuation pause
950 * could occur. This is actually safe, since the entering the sync
951 * barrier is one of the last things do_marking_step() does, and it
952 * doesn't manipulate any data structures afterwards.
953 */
954
955 void ConcurrentMark::enter_first_sync_barrier(uint worker_id) {
956 bool barrier_aborted;
957
958 if (verbose_low()) {
959 gclog_or_tty->print_cr("[%u] entering first barrier", worker_id);
960 }
961
962 {
963 SuspendibleThreadSetLeaver sts_leave(concurrent());
964 barrier_aborted = !_first_overflow_barrier_sync.enter();
965 }
966
967 // at this point everyone should have synced up and not be doing any
968 // more work
969
970 if (verbose_low()) {
971 if (barrier_aborted) {
972 gclog_or_tty->print_cr("[%u] aborted first barrier", worker_id);
973 } else {
974 gclog_or_tty->print_cr("[%u] leaving first barrier", worker_id);
975 }
976 }
977
978 if (barrier_aborted) {
979 // If the barrier aborted we ignore the overflow condition and
980 // just abort the whole marking phase as quickly as possible.
981 return;
982 }
983
984 // If we're executing the concurrent phase of marking, reset the marking
985 // state; otherwise the marking state is reset after reference processing,
986 // during the remark pause.
987 // If we reset here as a result of an overflow during the remark we will
988 // see assertion failures from any subsequent set_concurrency_and_phase()
989 // calls.
990 if (concurrent()) {
991 // let the task associated with with worker 0 do this
992 if (worker_id == 0) {
993 // task 0 is responsible for clearing the global data structures
994 // We should be here because of an overflow. During STW we should
995 // not clear the overflow flag since we rely on it being true when
996 // we exit this method to abort the pause and restart concurrent
997 // marking.
998 reset_marking_state(true /* clear_overflow */);
999 force_overflow()->update();
1000
1001 if (G1Log::fine()) {
1002 gclog_or_tty->gclog_stamp(concurrent_gc_id());
1003 gclog_or_tty->print_cr("[GC concurrent-mark-reset-for-overflow]");
1004 }
1005 }
1006 }
1007
1008 // after this, each task should reset its own data structures then
1009 // then go into the second barrier
1010 }
1011
1012 void ConcurrentMark::enter_second_sync_barrier(uint worker_id) {
1013 bool barrier_aborted;
1014
1015 if (verbose_low()) {
1016 gclog_or_tty->print_cr("[%u] entering second barrier", worker_id);
1017 }
1018
1019 {
1020 SuspendibleThreadSetLeaver sts_leave(concurrent());
1021 barrier_aborted = !_second_overflow_barrier_sync.enter();
1022 }
1023
1024 // at this point everything should be re-initialized and ready to go
1025
1026 if (verbose_low()) {
1027 if (barrier_aborted) {
1028 gclog_or_tty->print_cr("[%u] aborted second barrier", worker_id);
1029 } else {
1030 gclog_or_tty->print_cr("[%u] leaving second barrier", worker_id);
1031 }
1032 }
1033 }
1034
1035 #ifndef PRODUCT
1036 void ForceOverflowSettings::init() {
1037 _num_remaining = G1ConcMarkForceOverflow;
1038 _force = false;
1039 update();
1040 }
1041
1042 void ForceOverflowSettings::update() {
1043 if (_num_remaining > 0) {
1044 _num_remaining -= 1;
1045 _force = true;
1046 } else {
1047 _force = false;
1048 }
1049 }
1050
1051 bool ForceOverflowSettings::should_force() {
1052 if (_force) {
1053 _force = false;
1054 return true;
1055 } else {
1056 return false;
1057 }
1058 }
1059 #endif // !PRODUCT
1060
1061 class CMConcurrentMarkingTask: public AbstractGangTask {
1062 private:
1063 ConcurrentMark* _cm;
1064 ConcurrentMarkThread* _cmt;
1065
1066 public:
1067 void work(uint worker_id) {
1068 assert(Thread::current()->is_ConcurrentGC_thread(),
1069 "this should only be done by a conc GC thread");
1070 ResourceMark rm;
1071
1072 double start_vtime = os::elapsedVTime();
1073
1074 {
1075 SuspendibleThreadSetJoiner sts_join;
1076
1077 assert(worker_id < _cm->active_tasks(), "invariant");
1078 CMTask* the_task = _cm->task(worker_id);
1079 the_task->record_start_time();
1080 if (!_cm->has_aborted()) {
1081 do {
1082 double start_vtime_sec = os::elapsedVTime();
1083 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1084
1085 the_task->do_marking_step(mark_step_duration_ms,
1086 true /* do_termination */,
1087 false /* is_serial*/);
1088
1089 double end_vtime_sec = os::elapsedVTime();
1090 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec;
1091 _cm->clear_has_overflown();
1092
1093 _cm->do_yield_check(worker_id);
1094
1095 jlong sleep_time_ms;
1096 if (!_cm->has_aborted() && the_task->has_aborted()) {
1097 sleep_time_ms =
1098 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0);
1099 {
1100 SuspendibleThreadSetLeaver sts_leave;
1101 os::sleep(Thread::current(), sleep_time_ms, false);
1102 }
1103 }
1104 } while (!_cm->has_aborted() && the_task->has_aborted());
1105 }
1106 the_task->record_end_time();
1107 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant");
1108 }
1109
1110 double end_vtime = os::elapsedVTime();
1111 _cm->update_accum_task_vtime(worker_id, end_vtime - start_vtime);
1112 }
1113
1114 CMConcurrentMarkingTask(ConcurrentMark* cm,
1115 ConcurrentMarkThread* cmt) :
1116 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { }
1117
1118 ~CMConcurrentMarkingTask() { }
1119 };
1120
1121 // Calculates the number of active workers for a concurrent
1122 // phase.
1123 uint ConcurrentMark::calc_parallel_marking_threads() {
1124 uint n_conc_workers = 0;
1125 if (!UseDynamicNumberOfGCThreads ||
1126 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
1127 !ForceDynamicNumberOfGCThreads)) {
1128 n_conc_workers = max_parallel_marking_threads();
1129 } else {
1130 n_conc_workers =
1131 AdaptiveSizePolicy::calc_default_active_workers(
1132 max_parallel_marking_threads(),
1133 1, /* Minimum workers */
1134 parallel_marking_threads(),
1135 Threads::number_of_non_daemon_threads());
1136 // Don't scale down "n_conc_workers" by scale_parallel_threads() because
1137 // that scaling has already gone into "_max_parallel_marking_threads".
1138 }
1139 assert(n_conc_workers > 0, "Always need at least 1");
1140 return n_conc_workers;
1141 }
1142
1143 void ConcurrentMark::scanRootRegion(HeapRegion* hr, uint worker_id) {
1144 // Currently, only survivors can be root regions.
1145 assert(hr->next_top_at_mark_start() == hr->bottom(), "invariant");
1146 G1RootRegionScanClosure cl(_g1h, this, worker_id);
1147
1148 const uintx interval = PrefetchScanIntervalInBytes;
1149 HeapWord* curr = hr->bottom();
1150 const HeapWord* end = hr->top();
1151 while (curr < end) {
1152 Prefetch::read(curr, interval);
1153 oop obj = oop(curr);
1154 int size = obj->oop_iterate(&cl);
1155 assert(size == obj->size(), "sanity");
1156 curr += size;
1157 }
1158 }
1159
1160 class CMRootRegionScanTask : public AbstractGangTask {
1161 private:
1162 ConcurrentMark* _cm;
1163
1164 public:
1165 CMRootRegionScanTask(ConcurrentMark* cm) :
1166 AbstractGangTask("Root Region Scan"), _cm(cm) { }
1167
1168 void work(uint worker_id) {
1169 assert(Thread::current()->is_ConcurrentGC_thread(),
1170 "this should only be done by a conc GC thread");
1171
1172 CMRootRegions* root_regions = _cm->root_regions();
1173 HeapRegion* hr = root_regions->claim_next();
1174 while (hr != NULL) {
1175 _cm->scanRootRegion(hr, worker_id);
1176 hr = root_regions->claim_next();
1177 }
1178 }
1179 };
1180
1181 void ConcurrentMark::scanRootRegions() {
1182 // Start of concurrent marking.
1183 ClassLoaderDataGraph::clear_claimed_marks();
1184
1185 // scan_in_progress() will have been set to true only if there was
1186 // at least one root region to scan. So, if it's false, we
1187 // should not attempt to do any further work.
1188 if (root_regions()->scan_in_progress()) {
1189 _parallel_marking_threads = calc_parallel_marking_threads();
1190 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1191 "Maximum number of marking threads exceeded");
1192 uint active_workers = MAX2(1U, parallel_marking_threads());
1193
1194 CMRootRegionScanTask task(this);
1195 _parallel_workers->set_active_workers(active_workers);
1196 _parallel_workers->run_task(&task);
1197
1198 // It's possible that has_aborted() is true here without actually
1199 // aborting the survivor scan earlier. This is OK as it's
1200 // mainly used for sanity checking.
1201 root_regions()->scan_finished();
1202 }
1203 }
1204
1205 void ConcurrentMark::markFromRoots() {
1206 // we might be tempted to assert that:
1207 // assert(asynch == !SafepointSynchronize::is_at_safepoint(),
1208 // "inconsistent argument?");
1209 // However that wouldn't be right, because it's possible that
1210 // a safepoint is indeed in progress as a younger generation
1211 // stop-the-world GC happens even as we mark in this generation.
1212
1213 _restart_for_overflow = false;
1214 force_overflow_conc()->init();
1215
1216 // _g1h has _n_par_threads
1217 _parallel_marking_threads = calc_parallel_marking_threads();
1218 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1219 "Maximum number of marking threads exceeded");
1220
1221 uint active_workers = MAX2(1U, parallel_marking_threads());
1222 assert(active_workers > 0, "Should have been set");
1223
1224 // Parallel task terminator is set in "set_concurrency_and_phase()"
1225 set_concurrency_and_phase(active_workers, true /* concurrent */);
1226
1227 CMConcurrentMarkingTask markingTask(this, cmThread());
1228 _parallel_workers->set_active_workers(active_workers);
1229 _parallel_workers->run_task(&markingTask);
1230 print_stats();
1231 }
1232
1233 // Helper class to get rid of some boilerplate code.
1234 class G1CMTraceTime : public GCTraceTime {
1235 static bool doit_and_prepend(bool doit) {
1236 if (doit) {
1237 gclog_or_tty->put(' ');
1238 }
1239 return doit;
1240 }
1241
1242 public:
1243 G1CMTraceTime(const char* title, bool doit)
1244 : GCTraceTime(title, doit_and_prepend(doit), false, G1CollectedHeap::heap()->gc_timer_cm(),
1245 G1CollectedHeap::heap()->concurrent_mark()->concurrent_gc_id()) {
1246 }
1247 };
1248
1249 void ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) {
1250 // world is stopped at this checkpoint
1251 assert(SafepointSynchronize::is_at_safepoint(),
1252 "world should be stopped");
1253
1254 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1255
1256 // If a full collection has happened, we shouldn't do this.
1257 if (has_aborted()) {
1258 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1259 return;
1260 }
1261
1262 SvcGCMarker sgcm(SvcGCMarker::OTHER);
1263
1264 if (VerifyDuringGC) {
1265 HandleMark hm; // handle scope
1266 g1h->prepare_for_verify();
1267 Universe::verify(VerifyOption_G1UsePrevMarking,
1268 " VerifyDuringGC:(before)");
1269 }
1270 g1h->check_bitmaps("Remark Start");
1271
1272 G1CollectorPolicy* g1p = g1h->g1_policy();
1273 g1p->record_concurrent_mark_remark_start();
1274
1275 double start = os::elapsedTime();
1276
1277 checkpointRootsFinalWork();
1278
1279 double mark_work_end = os::elapsedTime();
1280
1281 weakRefsWork(clear_all_soft_refs);
1282
1283 if (has_overflown()) {
1284 // Oops. We overflowed. Restart concurrent marking.
1285 _restart_for_overflow = true;
1286 if (G1TraceMarkStackOverflow) {
1287 gclog_or_tty->print_cr("\nRemark led to restart for overflow.");
1288 }
1289
1290 // Verify the heap w.r.t. the previous marking bitmap.
1291 if (VerifyDuringGC) {
1292 HandleMark hm; // handle scope
1293 g1h->prepare_for_verify();
1294 Universe::verify(VerifyOption_G1UsePrevMarking,
1295 " VerifyDuringGC:(overflow)");
1296 }
1297
1298 // Clear the marking state because we will be restarting
1299 // marking due to overflowing the global mark stack.
1300 reset_marking_state();
1301 } else {
1302 {
1303 G1CMTraceTime trace("GC aggregate-data", G1Log::finer());
1304
1305 // Aggregate the per-task counting data that we have accumulated
1306 // while marking.
1307 aggregate_count_data();
1308 }
1309
1310 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1311 // We're done with marking.
1312 // This is the end of the marking cycle, we're expected all
1313 // threads to have SATB queues with active set to true.
1314 satb_mq_set.set_active_all_threads(false, /* new active value */
1315 true /* expected_active */);
1316
1317 if (VerifyDuringGC) {
1318 HandleMark hm; // handle scope
1319 g1h->prepare_for_verify();
1320 Universe::verify(VerifyOption_G1UseNextMarking,
1321 " VerifyDuringGC:(after)");
1322 }
1323 g1h->check_bitmaps("Remark End");
1324 assert(!restart_for_overflow(), "sanity");
1325 // Completely reset the marking state since marking completed
1326 set_non_marking_state();
1327 }
1328
1329 // Expand the marking stack, if we have to and if we can.
1330 if (_markStack.should_expand()) {
1331 _markStack.expand();
1332 }
1333
1334 // Statistics
1335 double now = os::elapsedTime();
1336 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1337 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1338 _remark_times.add((now - start) * 1000.0);
1339
1340 g1p->record_concurrent_mark_remark_end();
1341
1342 G1CMIsAliveClosure is_alive(g1h);
1343 g1h->gc_tracer_cm()->report_object_count_after_gc(&is_alive);
1344 }
1345
1346 // Base class of the closures that finalize and verify the
1347 // liveness counting data.
1348 class CMCountDataClosureBase: public HeapRegionClosure {
1349 protected:
1350 G1CollectedHeap* _g1h;
1351 ConcurrentMark* _cm;
1352 CardTableModRefBS* _ct_bs;
1353
1354 BitMap* _region_bm;
1355 BitMap* _card_bm;
1356
1357 // Takes a region that's not empty (i.e., it has at least one
1358 // live object in it and sets its corresponding bit on the region
1359 // bitmap to 1. If the region is "starts humongous" it will also set
1360 // to 1 the bits on the region bitmap that correspond to its
1361 // associated "continues humongous" regions.
1362 void set_bit_for_region(HeapRegion* hr) {
1363 assert(!hr->is_continues_humongous(), "should have filtered those out");
1364
1365 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1366 if (!hr->is_starts_humongous()) {
1367 // Normal (non-humongous) case: just set the bit.
1368 _region_bm->par_at_put(index, true);
1369 } else {
1370 // Starts humongous case: calculate how many regions are part of
1371 // this humongous region and then set the bit range.
1372 BitMap::idx_t end_index = (BitMap::idx_t) hr->last_hc_index();
1373 _region_bm->par_at_put_range(index, end_index, true);
1374 }
1375 }
1376
1377 public:
1378 CMCountDataClosureBase(G1CollectedHeap* g1h,
1379 BitMap* region_bm, BitMap* card_bm):
1380 _g1h(g1h), _cm(g1h->concurrent_mark()),
1381 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
1382 _region_bm(region_bm), _card_bm(card_bm) { }
1383 };
1384
1385 // Closure that calculates the # live objects per region. Used
1386 // for verification purposes during the cleanup pause.
1387 class CalcLiveObjectsClosure: public CMCountDataClosureBase {
1388 CMBitMapRO* _bm;
1389 size_t _region_marked_bytes;
1390
1391 public:
1392 CalcLiveObjectsClosure(CMBitMapRO *bm, G1CollectedHeap* g1h,
1393 BitMap* region_bm, BitMap* card_bm) :
1394 CMCountDataClosureBase(g1h, region_bm, card_bm),
1395 _bm(bm), _region_marked_bytes(0) { }
1396
1397 bool doHeapRegion(HeapRegion* hr) {
1398
1399 if (hr->is_continues_humongous()) {
1400 // We will ignore these here and process them when their
1401 // associated "starts humongous" region is processed (see
1402 // set_bit_for_heap_region()). Note that we cannot rely on their
1403 // associated "starts humongous" region to have their bit set to
1404 // 1 since, due to the region chunking in the parallel region
1405 // iteration, a "continues humongous" region might be visited
1406 // before its associated "starts humongous".
1407 return false;
1408 }
1409
1410 HeapWord* ntams = hr->next_top_at_mark_start();
1411 HeapWord* start = hr->bottom();
1412
1413 assert(start <= hr->end() && start <= ntams && ntams <= hr->end(),
1414 err_msg("Preconditions not met - "
1415 "start: "PTR_FORMAT", ntams: "PTR_FORMAT", end: "PTR_FORMAT,
1416 p2i(start), p2i(ntams), p2i(hr->end())));
1417
1418 // Find the first marked object at or after "start".
1419 start = _bm->getNextMarkedWordAddress(start, ntams);
1420
1421 size_t marked_bytes = 0;
1422
1423 while (start < ntams) {
1424 oop obj = oop(start);
1425 int obj_sz = obj->size();
1426 HeapWord* obj_end = start + obj_sz;
1427
1428 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
1429 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(obj_end);
1430
1431 // Note: if we're looking at the last region in heap - obj_end
1432 // could be actually just beyond the end of the heap; end_idx
1433 // will then correspond to a (non-existent) card that is also
1434 // just beyond the heap.
1435 if (_g1h->is_in_g1_reserved(obj_end) && !_ct_bs->is_card_aligned(obj_end)) {
1436 // end of object is not card aligned - increment to cover
1437 // all the cards spanned by the object
1438 end_idx += 1;
1439 }
1440
1441 // Set the bits in the card BM for the cards spanned by this object.
1442 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1443
1444 // Add the size of this object to the number of marked bytes.
1445 marked_bytes += (size_t)obj_sz * HeapWordSize;
1446
1447 // Find the next marked object after this one.
1448 start = _bm->getNextMarkedWordAddress(obj_end, ntams);
1449 }
1450
1451 // Mark the allocated-since-marking portion...
1452 HeapWord* top = hr->top();
1453 if (ntams < top) {
1454 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1455 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1456
1457 // Note: if we're looking at the last region in heap - top
1458 // could be actually just beyond the end of the heap; end_idx
1459 // will then correspond to a (non-existent) card that is also
1460 // just beyond the heap.
1461 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1462 // end of object is not card aligned - increment to cover
1463 // all the cards spanned by the object
1464 end_idx += 1;
1465 }
1466 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1467
1468 // This definitely means the region has live objects.
1469 set_bit_for_region(hr);
1470 }
1471
1472 // Update the live region bitmap.
1473 if (marked_bytes > 0) {
1474 set_bit_for_region(hr);
1475 }
1476
1477 // Set the marked bytes for the current region so that
1478 // it can be queried by a calling verification routine
1479 _region_marked_bytes = marked_bytes;
1480
1481 return false;
1482 }
1483
1484 size_t region_marked_bytes() const { return _region_marked_bytes; }
1485 };
1486
1487 // Heap region closure used for verifying the counting data
1488 // that was accumulated concurrently and aggregated during
1489 // the remark pause. This closure is applied to the heap
1490 // regions during the STW cleanup pause.
1491
1492 class VerifyLiveObjectDataHRClosure: public HeapRegionClosure {
1493 G1CollectedHeap* _g1h;
1494 ConcurrentMark* _cm;
1495 CalcLiveObjectsClosure _calc_cl;
1496 BitMap* _region_bm; // Region BM to be verified
1497 BitMap* _card_bm; // Card BM to be verified
1498 bool _verbose; // verbose output?
1499
1500 BitMap* _exp_region_bm; // Expected Region BM values
1501 BitMap* _exp_card_bm; // Expected card BM values
1502
1503 int _failures;
1504
1505 public:
1506 VerifyLiveObjectDataHRClosure(G1CollectedHeap* g1h,
1507 BitMap* region_bm,
1508 BitMap* card_bm,
1509 BitMap* exp_region_bm,
1510 BitMap* exp_card_bm,
1511 bool verbose) :
1512 _g1h(g1h), _cm(g1h->concurrent_mark()),
1513 _calc_cl(_cm->nextMarkBitMap(), g1h, exp_region_bm, exp_card_bm),
1514 _region_bm(region_bm), _card_bm(card_bm), _verbose(verbose),
1515 _exp_region_bm(exp_region_bm), _exp_card_bm(exp_card_bm),
1516 _failures(0) { }
1517
1518 int failures() const { return _failures; }
1519
1520 bool doHeapRegion(HeapRegion* hr) {
1521 if (hr->is_continues_humongous()) {
1522 // We will ignore these here and process them when their
1523 // associated "starts humongous" region is processed (see
1524 // set_bit_for_heap_region()). Note that we cannot rely on their
1525 // associated "starts humongous" region to have their bit set to
1526 // 1 since, due to the region chunking in the parallel region
1527 // iteration, a "continues humongous" region might be visited
1528 // before its associated "starts humongous".
1529 return false;
1530 }
1531
1532 int failures = 0;
1533
1534 // Call the CalcLiveObjectsClosure to walk the marking bitmap for
1535 // this region and set the corresponding bits in the expected region
1536 // and card bitmaps.
1537 bool res = _calc_cl.doHeapRegion(hr);
1538 assert(res == false, "should be continuing");
1539
1540 MutexLockerEx x((_verbose ? ParGCRareEvent_lock : NULL),
1541 Mutex::_no_safepoint_check_flag);
1542
1543 // Verify the marked bytes for this region.
1544 size_t exp_marked_bytes = _calc_cl.region_marked_bytes();
1545 size_t act_marked_bytes = hr->next_marked_bytes();
1546
1547 // We're not OK if expected marked bytes > actual marked bytes. It means
1548 // we have missed accounting some objects during the actual marking.
1549 if (exp_marked_bytes > act_marked_bytes) {
1550 if (_verbose) {
1551 gclog_or_tty->print_cr("Region %u: marked bytes mismatch: "
1552 "expected: " SIZE_FORMAT ", actual: " SIZE_FORMAT,
1553 hr->hrm_index(), exp_marked_bytes, act_marked_bytes);
1554 }
1555 failures += 1;
1556 }
1557
1558 // Verify the bit, for this region, in the actual and expected
1559 // (which was just calculated) region bit maps.
1560 // We're not OK if the bit in the calculated expected region
1561 // bitmap is set and the bit in the actual region bitmap is not.
1562 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1563
1564 bool expected = _exp_region_bm->at(index);
1565 bool actual = _region_bm->at(index);
1566 if (expected && !actual) {
1567 if (_verbose) {
1568 gclog_or_tty->print_cr("Region %u: region bitmap mismatch: "
1569 "expected: %s, actual: %s",
1570 hr->hrm_index(),
1571 BOOL_TO_STR(expected), BOOL_TO_STR(actual));
1572 }
1573 failures += 1;
1574 }
1575
1576 // Verify that the card bit maps for the cards spanned by the current
1577 // region match. We have an error if we have a set bit in the expected
1578 // bit map and the corresponding bit in the actual bitmap is not set.
1579
1580 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(hr->bottom());
1581 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(hr->top());
1582
1583 for (BitMap::idx_t i = start_idx; i < end_idx; i+=1) {
1584 expected = _exp_card_bm->at(i);
1585 actual = _card_bm->at(i);
1586
1587 if (expected && !actual) {
1588 if (_verbose) {
1589 gclog_or_tty->print_cr("Region %u: card bitmap mismatch at " SIZE_FORMAT ": "
1590 "expected: %s, actual: %s",
1591 hr->hrm_index(), i,
1592 BOOL_TO_STR(expected), BOOL_TO_STR(actual));
1593 }
1594 failures += 1;
1595 }
1596 }
1597
1598 if (failures > 0 && _verbose) {
1599 gclog_or_tty->print_cr("Region " HR_FORMAT ", ntams: " PTR_FORMAT ", "
1600 "marked_bytes: calc/actual " SIZE_FORMAT "/" SIZE_FORMAT,
1601 HR_FORMAT_PARAMS(hr), p2i(hr->next_top_at_mark_start()),
1602 _calc_cl.region_marked_bytes(), hr->next_marked_bytes());
1603 }
1604
1605 _failures += failures;
1606
1607 // We could stop iteration over the heap when we
1608 // find the first violating region by returning true.
1609 return false;
1610 }
1611 };
1612
1613 class G1ParVerifyFinalCountTask: public AbstractGangTask {
1614 protected:
1615 G1CollectedHeap* _g1h;
1616 ConcurrentMark* _cm;
1617 BitMap* _actual_region_bm;
1618 BitMap* _actual_card_bm;
1619
1620 uint _n_workers;
1621
1622 BitMap* _expected_region_bm;
1623 BitMap* _expected_card_bm;
1624
1625 int _failures;
1626 bool _verbose;
1627
1628 HeapRegionClaimer _hrclaimer;
1629
1630 public:
1631 G1ParVerifyFinalCountTask(G1CollectedHeap* g1h,
1632 BitMap* region_bm, BitMap* card_bm,
1633 BitMap* expected_region_bm, BitMap* expected_card_bm)
1634 : AbstractGangTask("G1 verify final counting"),
1635 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1636 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1637 _expected_region_bm(expected_region_bm), _expected_card_bm(expected_card_bm),
1638 _failures(0), _verbose(false),
1639 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1640 assert(VerifyDuringGC, "don't call this otherwise");
1641 assert(_expected_card_bm->size() == _actual_card_bm->size(), "sanity");
1642 assert(_expected_region_bm->size() == _actual_region_bm->size(), "sanity");
1643
1644 _verbose = _cm->verbose_medium();
1645 }
1646
1647 void work(uint worker_id) {
1648 assert(worker_id < _n_workers, "invariant");
1649
1650 VerifyLiveObjectDataHRClosure verify_cl(_g1h,
1651 _actual_region_bm, _actual_card_bm,
1652 _expected_region_bm,
1653 _expected_card_bm,
1654 _verbose);
1655
1656 _g1h->heap_region_par_iterate(&verify_cl, worker_id, &_hrclaimer);
1657
1658 Atomic::add(verify_cl.failures(), &_failures);
1659 }
1660
1661 int failures() const { return _failures; }
1662 };
1663
1664 // Closure that finalizes the liveness counting data.
1665 // Used during the cleanup pause.
1666 // Sets the bits corresponding to the interval [NTAMS, top]
1667 // (which contains the implicitly live objects) in the
1668 // card liveness bitmap. Also sets the bit for each region,
1669 // containing live data, in the region liveness bitmap.
1670
1671 class FinalCountDataUpdateClosure: public CMCountDataClosureBase {
1672 public:
1673 FinalCountDataUpdateClosure(G1CollectedHeap* g1h,
1674 BitMap* region_bm,
1675 BitMap* card_bm) :
1676 CMCountDataClosureBase(g1h, region_bm, card_bm) { }
1677
1678 bool doHeapRegion(HeapRegion* hr) {
1679
1680 if (hr->is_continues_humongous()) {
1681 // We will ignore these here and process them when their
1682 // associated "starts humongous" region is processed (see
1683 // set_bit_for_heap_region()). Note that we cannot rely on their
1684 // associated "starts humongous" region to have their bit set to
1685 // 1 since, due to the region chunking in the parallel region
1686 // iteration, a "continues humongous" region might be visited
1687 // before its associated "starts humongous".
1688 return false;
1689 }
1690
1691 HeapWord* ntams = hr->next_top_at_mark_start();
1692 HeapWord* top = hr->top();
1693
1694 assert(hr->bottom() <= ntams && ntams <= hr->end(), "Preconditions.");
1695
1696 // Mark the allocated-since-marking portion...
1697 if (ntams < top) {
1698 // This definitely means the region has live objects.
1699 set_bit_for_region(hr);
1700
1701 // Now set the bits in the card bitmap for [ntams, top)
1702 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1703 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1704
1705 // Note: if we're looking at the last region in heap - top
1706 // could be actually just beyond the end of the heap; end_idx
1707 // will then correspond to a (non-existent) card that is also
1708 // just beyond the heap.
1709 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1710 // end of object is not card aligned - increment to cover
1711 // all the cards spanned by the object
1712 end_idx += 1;
1713 }
1714
1715 assert(end_idx <= _card_bm->size(),
1716 err_msg("oob: end_idx= "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
1717 end_idx, _card_bm->size()));
1718 assert(start_idx < _card_bm->size(),
1719 err_msg("oob: start_idx= "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
1720 start_idx, _card_bm->size()));
1721
1722 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1723 }
1724
1725 // Set the bit for the region if it contains live data
1726 if (hr->next_marked_bytes() > 0) {
1727 set_bit_for_region(hr);
1728 }
1729
1730 return false;
1731 }
1732 };
1733
1734 class G1ParFinalCountTask: public AbstractGangTask {
1735 protected:
1736 G1CollectedHeap* _g1h;
1737 ConcurrentMark* _cm;
1738 BitMap* _actual_region_bm;
1739 BitMap* _actual_card_bm;
1740
1741 uint _n_workers;
1742 HeapRegionClaimer _hrclaimer;
1743
1744 public:
1745 G1ParFinalCountTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm)
1746 : AbstractGangTask("G1 final counting"),
1747 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1748 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1749 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1750 }
1751
1752 void work(uint worker_id) {
1753 assert(worker_id < _n_workers, "invariant");
1754
1755 FinalCountDataUpdateClosure final_update_cl(_g1h,
1756 _actual_region_bm,
1757 _actual_card_bm);
1758
1759 _g1h->heap_region_par_iterate(&final_update_cl, worker_id, &_hrclaimer);
1760 }
1761 };
1762
1763 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure {
1764 G1CollectedHeap* _g1;
1765 size_t _freed_bytes;
1766 FreeRegionList* _local_cleanup_list;
1767 HeapRegionSetCount _old_regions_removed;
1768 HeapRegionSetCount _humongous_regions_removed;
1769 HRRSCleanupTask* _hrrs_cleanup_task;
1770
1771 public:
1772 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1,
1773 FreeRegionList* local_cleanup_list,
1774 HRRSCleanupTask* hrrs_cleanup_task) :
1775 _g1(g1),
1776 _freed_bytes(0),
1777 _local_cleanup_list(local_cleanup_list),
1778 _old_regions_removed(),
1779 _humongous_regions_removed(),
1780 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1781
1782 size_t freed_bytes() { return _freed_bytes; }
1783 const HeapRegionSetCount& old_regions_removed() { return _old_regions_removed; }
1784 const HeapRegionSetCount& humongous_regions_removed() { return _humongous_regions_removed; }
1785
1786 bool doHeapRegion(HeapRegion *hr) {
1787 if (hr->is_continues_humongous() || hr->is_archive()) {
1788 return false;
1789 }
1790 // We use a claim value of zero here because all regions
1791 // were claimed with value 1 in the FinalCount task.
1792 _g1->reset_gc_time_stamps(hr);
1793 hr->note_end_of_marking();
1794
1795 if (hr->used() > 0 && hr->max_live_bytes() == 0 && !hr->is_young()) {
1796 _freed_bytes += hr->used();
1797 hr->set_containing_set(NULL);
1798 if (hr->is_humongous()) {
1799 assert(hr->is_starts_humongous(), "we should only see starts humongous");
1800 _humongous_regions_removed.increment(1u, hr->capacity());
1801 _g1->free_humongous_region(hr, _local_cleanup_list, true);
1802 } else {
1803 _old_regions_removed.increment(1u, hr->capacity());
1804 _g1->free_region(hr, _local_cleanup_list, true);
1805 }
1806 } else {
1807 hr->rem_set()->do_cleanup_work(_hrrs_cleanup_task);
1808 }
1809
1810 return false;
1811 }
1812 };
1813
1814 class G1ParNoteEndTask: public AbstractGangTask {
1815 friend class G1NoteEndOfConcMarkClosure;
1816
1817 protected:
1818 G1CollectedHeap* _g1h;
1819 FreeRegionList* _cleanup_list;
1820 HeapRegionClaimer _hrclaimer;
1821
1822 public:
1823 G1ParNoteEndTask(G1CollectedHeap* g1h, FreeRegionList* cleanup_list, uint n_workers) :
1824 AbstractGangTask("G1 note end"), _g1h(g1h), _cleanup_list(cleanup_list), _hrclaimer(n_workers) {
1825 }
1826
1827 void work(uint worker_id) {
1828 FreeRegionList local_cleanup_list("Local Cleanup List");
1829 HRRSCleanupTask hrrs_cleanup_task;
1830 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, &local_cleanup_list,
1831 &hrrs_cleanup_task);
1832 _g1h->heap_region_par_iterate(&g1_note_end, worker_id, &_hrclaimer);
1833 assert(g1_note_end.complete(), "Shouldn't have yielded!");
1834
1835 // Now update the lists
1836 _g1h->remove_from_old_sets(g1_note_end.old_regions_removed(), g1_note_end.humongous_regions_removed());
1837 {
1838 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1839 _g1h->decrement_summary_bytes(g1_note_end.freed_bytes());
1840
1841 // If we iterate over the global cleanup list at the end of
1842 // cleanup to do this printing we will not guarantee to only
1843 // generate output for the newly-reclaimed regions (the list
1844 // might not be empty at the beginning of cleanup; we might
1845 // still be working on its previous contents). So we do the
1846 // printing here, before we append the new regions to the global
1847 // cleanup list.
1848
1849 G1HRPrinter* hr_printer = _g1h->hr_printer();
1850 if (hr_printer->is_active()) {
1851 FreeRegionListIterator iter(&local_cleanup_list);
1852 while (iter.more_available()) {
1853 HeapRegion* hr = iter.get_next();
1854 hr_printer->cleanup(hr);
1855 }
1856 }
1857
1858 _cleanup_list->add_ordered(&local_cleanup_list);
1859 assert(local_cleanup_list.is_empty(), "post-condition");
1860
1861 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1862 }
1863 }
1864 };
1865
1866 class G1ParScrubRemSetTask: public AbstractGangTask {
1867 protected:
1868 G1RemSet* _g1rs;
1869 BitMap* _region_bm;
1870 BitMap* _card_bm;
1871 HeapRegionClaimer _hrclaimer;
1872
1873 public:
1874 G1ParScrubRemSetTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm, uint n_workers) :
1875 AbstractGangTask("G1 ScrubRS"), _g1rs(g1h->g1_rem_set()), _region_bm(region_bm), _card_bm(card_bm), _hrclaimer(n_workers) {
1876 }
1877
1878 void work(uint worker_id) {
1879 _g1rs->scrub(_region_bm, _card_bm, worker_id, &_hrclaimer);
1880 }
1881
1882 };
1883
1884 void ConcurrentMark::cleanup() {
1885 // world is stopped at this checkpoint
1886 assert(SafepointSynchronize::is_at_safepoint(),
1887 "world should be stopped");
1888 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1889
1890 // If a full collection has happened, we shouldn't do this.
1891 if (has_aborted()) {
1892 g1h->collector_state()->set_mark_in_progress(false); // So bitmap clearing isn't confused
1893 return;
1894 }
1895
1896 g1h->verify_region_sets_optional();
1897
1898 if (VerifyDuringGC) {
1899 HandleMark hm; // handle scope
1900 g1h->prepare_for_verify();
1901 Universe::verify(VerifyOption_G1UsePrevMarking,
1902 " VerifyDuringGC:(before)");
1903 }
1904 g1h->check_bitmaps("Cleanup Start");
1905
1906 G1CollectorPolicy* g1p = g1h->g1_policy();
1907 g1p->record_concurrent_mark_cleanup_start();
1908
1909 double start = os::elapsedTime();
1910
1911 HeapRegionRemSet::reset_for_cleanup_tasks();
1912
1913 // Do counting once more with the world stopped for good measure.
1914 G1ParFinalCountTask g1_par_count_task(g1h, &_region_bm, &_card_bm);
1915
1916 g1h->workers()->run_task(&g1_par_count_task);
1917
1918 if (VerifyDuringGC) {
1919 // Verify that the counting data accumulated during marking matches
1920 // that calculated by walking the marking bitmap.
1921
1922 // Bitmaps to hold expected values
1923 BitMap expected_region_bm(_region_bm.size(), true);
1924 BitMap expected_card_bm(_card_bm.size(), true);
1925
1926 G1ParVerifyFinalCountTask g1_par_verify_task(g1h,
1927 &_region_bm,
1928 &_card_bm,
1929 &expected_region_bm,
1930 &expected_card_bm);
1931
1932 g1h->workers()->run_task(&g1_par_verify_task);
1933
1934 guarantee(g1_par_verify_task.failures() == 0, "Unexpected accounting failures");
1935 }
1936
1937 size_t start_used_bytes = g1h->used();
1938 g1h->collector_state()->set_mark_in_progress(false);
1939
1940 double count_end = os::elapsedTime();
1941 double this_final_counting_time = (count_end - start);
1942 _total_counting_time += this_final_counting_time;
1943
1944 if (G1PrintRegionLivenessInfo) {
1945 G1PrintRegionLivenessInfoClosure cl(gclog_or_tty, "Post-Marking");
1946 _g1h->heap_region_iterate(&cl);
1947 }
1948
1949 // Install newly created mark bitMap as "prev".
1950 swapMarkBitMaps();
1951
1952 g1h->reset_gc_time_stamp();
1953
1954 uint n_workers = _g1h->workers()->active_workers();
1955
1956 // Note end of marking in all heap regions.
1957 G1ParNoteEndTask g1_par_note_end_task(g1h, &_cleanup_list, n_workers);
1958 g1h->workers()->run_task(&g1_par_note_end_task);
1959 g1h->check_gc_time_stamps();
1960
1961 if (!cleanup_list_is_empty()) {
1962 // The cleanup list is not empty, so we'll have to process it
1963 // concurrently. Notify anyone else that might be wanting free
1964 // regions that there will be more free regions coming soon.
1965 g1h->set_free_regions_coming();
1966 }
1967
1968 // call below, since it affects the metric by which we sort the heap
1969 // regions.
1970 if (G1ScrubRemSets) {
1971 double rs_scrub_start = os::elapsedTime();
1972 G1ParScrubRemSetTask g1_par_scrub_rs_task(g1h, &_region_bm, &_card_bm, n_workers);
1973 g1h->workers()->run_task(&g1_par_scrub_rs_task);
1974
1975 double rs_scrub_end = os::elapsedTime();
1976 double this_rs_scrub_time = (rs_scrub_end - rs_scrub_start);
1977 _total_rs_scrub_time += this_rs_scrub_time;
1978 }
1979
1980 // this will also free any regions totally full of garbage objects,
1981 // and sort the regions.
1982 g1h->g1_policy()->record_concurrent_mark_cleanup_end();
1983
1984 // Statistics.
1985 double end = os::elapsedTime();
1986 _cleanup_times.add((end - start) * 1000.0);
1987
1988 if (G1Log::fine()) {
1989 g1h->g1_policy()->print_heap_transition(start_used_bytes);
1990 }
1991
1992 // Clean up will have freed any regions completely full of garbage.
1993 // Update the soft reference policy with the new heap occupancy.
1994 Universe::update_heap_info_at_gc();
1995
1996 if (VerifyDuringGC) {
1997 HandleMark hm; // handle scope
1998 g1h->prepare_for_verify();
1999 Universe::verify(VerifyOption_G1UsePrevMarking,
2000 " VerifyDuringGC:(after)");
2001 }
2002
2003 g1h->check_bitmaps("Cleanup End");
2004
2005 g1h->verify_region_sets_optional();
2006
2007 // We need to make this be a "collection" so any collection pause that
2008 // races with it goes around and waits for completeCleanup to finish.
2009 g1h->increment_total_collections();
2010
2011 // Clean out dead classes and update Metaspace sizes.
2012 if (ClassUnloadingWithConcurrentMark) {
2013 ClassLoaderDataGraph::purge();
2014 }
2015 MetaspaceGC::compute_new_size();
2016
2017 // We reclaimed old regions so we should calculate the sizes to make
2018 // sure we update the old gen/space data.
2019 g1h->g1mm()->update_sizes();
2020 g1h->allocation_context_stats().update_after_mark();
2021
2022 g1h->trace_heap_after_concurrent_cycle();
2023 }
2024
2025 void ConcurrentMark::completeCleanup() {
2026 if (has_aborted()) return;
2027
2028 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2029
2030 _cleanup_list.verify_optional();
2031 FreeRegionList tmp_free_list("Tmp Free List");
2032
2033 if (G1ConcRegionFreeingVerbose) {
2034 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
2035 "cleanup list has %u entries",
2036 _cleanup_list.length());
2037 }
2038
2039 // No one else should be accessing the _cleanup_list at this point,
2040 // so it is not necessary to take any locks
2041 while (!_cleanup_list.is_empty()) {
2042 HeapRegion* hr = _cleanup_list.remove_region(true /* from_head */);
2043 assert(hr != NULL, "Got NULL from a non-empty list");
2044 hr->par_clear();
2045 tmp_free_list.add_ordered(hr);
2046
2047 // Instead of adding one region at a time to the secondary_free_list,
2048 // we accumulate them in the local list and move them a few at a
2049 // time. This also cuts down on the number of notify_all() calls
2050 // we do during this process. We'll also append the local list when
2051 // _cleanup_list is empty (which means we just removed the last
2052 // region from the _cleanup_list).
2053 if ((tmp_free_list.length() % G1SecondaryFreeListAppendLength == 0) ||
2054 _cleanup_list.is_empty()) {
2055 if (G1ConcRegionFreeingVerbose) {
2056 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
2057 "appending %u entries to the secondary_free_list, "
2058 "cleanup list still has %u entries",
2059 tmp_free_list.length(),
2060 _cleanup_list.length());
2061 }
2062
2063 {
2064 MutexLockerEx x(SecondaryFreeList_lock, Mutex::_no_safepoint_check_flag);
2065 g1h->secondary_free_list_add(&tmp_free_list);
2066 SecondaryFreeList_lock->notify_all();
2067 }
2068 #ifndef PRODUCT
2069 if (G1StressConcRegionFreeing) {
2070 for (uintx i = 0; i < G1StressConcRegionFreeingDelayMillis; ++i) {
2071 os::sleep(Thread::current(), (jlong) 1, false);
2072 }
2073 }
2074 #endif
2075 }
2076 }
2077 assert(tmp_free_list.is_empty(), "post-condition");
2078 }
2079
2080 // Supporting Object and Oop closures for reference discovery
2081 // and processing in during marking
2082
2083 bool G1CMIsAliveClosure::do_object_b(oop obj) {
2084 HeapWord* addr = (HeapWord*)obj;
2085 return addr != NULL &&
2086 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj));
2087 }
2088
2089 // 'Keep Alive' oop closure used by both serial parallel reference processing.
2090 // Uses the CMTask associated with a worker thread (for serial reference
2091 // processing the CMTask for worker 0 is used) to preserve (mark) and
2092 // trace referent objects.
2093 //
2094 // Using the CMTask and embedded local queues avoids having the worker
2095 // threads operating on the global mark stack. This reduces the risk
2096 // of overflowing the stack - which we would rather avoid at this late
2097 // state. Also using the tasks' local queues removes the potential
2098 // of the workers interfering with each other that could occur if
2099 // operating on the global stack.
2100
2101 class G1CMKeepAliveAndDrainClosure: public OopClosure {
2102 ConcurrentMark* _cm;
2103 CMTask* _task;
2104 int _ref_counter_limit;
2105 int _ref_counter;
2106 bool _is_serial;
2107 public:
2108 G1CMKeepAliveAndDrainClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
2109 _cm(cm), _task(task), _is_serial(is_serial),
2110 _ref_counter_limit(G1RefProcDrainInterval) {
2111 assert(_ref_counter_limit > 0, "sanity");
2112 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
2113 _ref_counter = _ref_counter_limit;
2114 }
2115
2116 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
2117 virtual void do_oop( oop* p) { do_oop_work(p); }
2118
2119 template <class T> void do_oop_work(T* p) {
2120 if (!_cm->has_overflown()) {
2121 oop obj = oopDesc::load_decode_heap_oop(p);
2122 if (_cm->verbose_high()) {
2123 gclog_or_tty->print_cr("\t[%u] we're looking at location "
2124 "*"PTR_FORMAT" = "PTR_FORMAT,
2125 _task->worker_id(), p2i(p), p2i((void*) obj));
2126 }
2127
2128 _task->deal_with_reference(obj);
2129 _ref_counter--;
2130
2131 if (_ref_counter == 0) {
2132 // We have dealt with _ref_counter_limit references, pushing them
2133 // and objects reachable from them on to the local stack (and
2134 // possibly the global stack). Call CMTask::do_marking_step() to
2135 // process these entries.
2136 //
2137 // We call CMTask::do_marking_step() in a loop, which we'll exit if
2138 // there's nothing more to do (i.e. we're done with the entries that
2139 // were pushed as a result of the CMTask::deal_with_reference() calls
2140 // above) or we overflow.
2141 //
2142 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
2143 // flag while there may still be some work to do. (See the comment at
2144 // the beginning of CMTask::do_marking_step() for those conditions -
2145 // one of which is reaching the specified time target.) It is only
2146 // when CMTask::do_marking_step() returns without setting the
2147 // has_aborted() flag that the marking step has completed.
2148 do {
2149 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
2150 _task->do_marking_step(mark_step_duration_ms,
2151 false /* do_termination */,
2152 _is_serial);
2153 } while (_task->has_aborted() && !_cm->has_overflown());
2154 _ref_counter = _ref_counter_limit;
2155 }
2156 } else {
2157 if (_cm->verbose_high()) {
2158 gclog_or_tty->print_cr("\t[%u] CM Overflow", _task->worker_id());
2159 }
2160 }
2161 }
2162 };
2163
2164 // 'Drain' oop closure used by both serial and parallel reference processing.
2165 // Uses the CMTask associated with a given worker thread (for serial
2166 // reference processing the CMtask for worker 0 is used). Calls the
2167 // do_marking_step routine, with an unbelievably large timeout value,
2168 // to drain the marking data structures of the remaining entries
2169 // added by the 'keep alive' oop closure above.
2170
2171 class G1CMDrainMarkingStackClosure: public VoidClosure {
2172 ConcurrentMark* _cm;
2173 CMTask* _task;
2174 bool _is_serial;
2175 public:
2176 G1CMDrainMarkingStackClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
2177 _cm(cm), _task(task), _is_serial(is_serial) {
2178 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
2179 }
2180
2181 void do_void() {
2182 do {
2183 if (_cm->verbose_high()) {
2184 gclog_or_tty->print_cr("\t[%u] Drain: Calling do_marking_step - serial: %s",
2185 _task->worker_id(), BOOL_TO_STR(_is_serial));
2186 }
2187
2188 // We call CMTask::do_marking_step() to completely drain the local
2189 // and global marking stacks of entries pushed by the 'keep alive'
2190 // oop closure (an instance of G1CMKeepAliveAndDrainClosure above).
2191 //
2192 // CMTask::do_marking_step() is called in a loop, which we'll exit
2193 // if there's nothing more to do (i.e. we've completely drained the
2194 // entries that were pushed as a a result of applying the 'keep alive'
2195 // closure to the entries on the discovered ref lists) or we overflow
2196 // the global marking stack.
2197 //
2198 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
2199 // flag while there may still be some work to do. (See the comment at
2200 // the beginning of CMTask::do_marking_step() for those conditions -
2201 // one of which is reaching the specified time target.) It is only
2202 // when CMTask::do_marking_step() returns without setting the
2203 // has_aborted() flag that the marking step has completed.
2204
2205 _task->do_marking_step(1000000000.0 /* something very large */,
2206 true /* do_termination */,
2207 _is_serial);
2208 } while (_task->has_aborted() && !_cm->has_overflown());
2209 }
2210 };
2211
2212 // Implementation of AbstractRefProcTaskExecutor for parallel
2213 // reference processing at the end of G1 concurrent marking
2214
2215 class G1CMRefProcTaskExecutor: public AbstractRefProcTaskExecutor {
2216 private:
2217 G1CollectedHeap* _g1h;
2218 ConcurrentMark* _cm;
2219 WorkGang* _workers;
2220 uint _active_workers;
2221
2222 public:
2223 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
2224 ConcurrentMark* cm,
2225 WorkGang* workers,
2226 uint n_workers) :
2227 _g1h(g1h), _cm(cm),
2228 _workers(workers), _active_workers(n_workers) { }
2229
2230 // Executes the given task using concurrent marking worker threads.
2231 virtual void execute(ProcessTask& task);
2232 virtual void execute(EnqueueTask& task);
2233 };
2234
2235 class G1CMRefProcTaskProxy: public AbstractGangTask {
2236 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
2237 ProcessTask& _proc_task;
2238 G1CollectedHeap* _g1h;
2239 ConcurrentMark* _cm;
2240
2241 public:
2242 G1CMRefProcTaskProxy(ProcessTask& proc_task,
2243 G1CollectedHeap* g1h,
2244 ConcurrentMark* cm) :
2245 AbstractGangTask("Process reference objects in parallel"),
2246 _proc_task(proc_task), _g1h(g1h), _cm(cm) {
2247 ReferenceProcessor* rp = _g1h->ref_processor_cm();
2248 assert(rp->processing_is_mt(), "shouldn't be here otherwise");
2249 }
2250
2251 virtual void work(uint worker_id) {
2252 ResourceMark rm;
2253 HandleMark hm;
2254 CMTask* task = _cm->task(worker_id);
2255 G1CMIsAliveClosure g1_is_alive(_g1h);
2256 G1CMKeepAliveAndDrainClosure g1_par_keep_alive(_cm, task, false /* is_serial */);
2257 G1CMDrainMarkingStackClosure g1_par_drain(_cm, task, false /* is_serial */);
2258
2259 _proc_task.work(worker_id, g1_is_alive, g1_par_keep_alive, g1_par_drain);
2260 }
2261 };
2262
2263 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
2264 assert(_workers != NULL, "Need parallel worker threads.");
2265 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2266
2267 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
2268
2269 // We need to reset the concurrency level before each
2270 // proxy task execution, so that the termination protocol
2271 // and overflow handling in CMTask::do_marking_step() knows
2272 // how many workers to wait for.
2273 _cm->set_concurrency(_active_workers);
2274 _workers->run_task(&proc_task_proxy);
2275 }
2276
2277 class G1CMRefEnqueueTaskProxy: public AbstractGangTask {
2278 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
2279 EnqueueTask& _enq_task;
2280
2281 public:
2282 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
2283 AbstractGangTask("Enqueue reference objects in parallel"),
2284 _enq_task(enq_task) { }
2285
2286 virtual void work(uint worker_id) {
2287 _enq_task.work(worker_id);
2288 }
2289 };
2290
2291 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
2292 assert(_workers != NULL, "Need parallel worker threads.");
2293 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2294
2295 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
2296
2297 // Not strictly necessary but...
2298 //
2299 // We need to reset the concurrency level before each
2300 // proxy task execution, so that the termination protocol
2301 // and overflow handling in CMTask::do_marking_step() knows
2302 // how many workers to wait for.
2303 _cm->set_concurrency(_active_workers);
2304 _workers->run_task(&enq_task_proxy);
2305 }
2306
2307 void ConcurrentMark::weakRefsWorkParallelPart(BoolObjectClosure* is_alive, bool purged_classes) {
2308 G1CollectedHeap::heap()->parallel_cleaning(is_alive, true, true, purged_classes);
2309 }
2310
2311 void ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) {
2312 if (has_overflown()) {
2313 // Skip processing the discovered references if we have
2314 // overflown the global marking stack. Reference objects
2315 // only get discovered once so it is OK to not
2316 // de-populate the discovered reference lists. We could have,
2317 // but the only benefit would be that, when marking restarts,
2318 // less reference objects are discovered.
2319 return;
2320 }
2321
2322 ResourceMark rm;
2323 HandleMark hm;
2324
2325 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2326
2327 // Is alive closure.
2328 G1CMIsAliveClosure g1_is_alive(g1h);
2329
2330 // Inner scope to exclude the cleaning of the string and symbol
2331 // tables from the displayed time.
2332 {
2333 G1CMTraceTime t("GC ref-proc", G1Log::finer());
2334
2335 ReferenceProcessor* rp = g1h->ref_processor_cm();
2336
2337 // See the comment in G1CollectedHeap::ref_processing_init()
2338 // about how reference processing currently works in G1.
2339
2340 // Set the soft reference policy
2341 rp->setup_policy(clear_all_soft_refs);
2342 assert(_markStack.isEmpty(), "mark stack should be empty");
2343
2344 // Instances of the 'Keep Alive' and 'Complete GC' closures used
2345 // in serial reference processing. Note these closures are also
2346 // used for serially processing (by the the current thread) the
2347 // JNI references during parallel reference processing.
2348 //
2349 // These closures do not need to synchronize with the worker
2350 // threads involved in parallel reference processing as these
2351 // instances are executed serially by the current thread (e.g.
2352 // reference processing is not multi-threaded and is thus
2353 // performed by the current thread instead of a gang worker).
2354 //
2355 // The gang tasks involved in parallel reference processing create
2356 // their own instances of these closures, which do their own
2357 // synchronization among themselves.
2358 G1CMKeepAliveAndDrainClosure g1_keep_alive(this, task(0), true /* is_serial */);
2359 G1CMDrainMarkingStackClosure g1_drain_mark_stack(this, task(0), true /* is_serial */);
2360
2361 // We need at least one active thread. If reference processing
2362 // is not multi-threaded we use the current (VMThread) thread,
2363 // otherwise we use the work gang from the G1CollectedHeap and
2364 // we utilize all the worker threads we can.
2365 bool processing_is_mt = rp->processing_is_mt();
2366 uint active_workers = (processing_is_mt ? g1h->workers()->active_workers() : 1U);
2367 active_workers = MAX2(MIN2(active_workers, _max_worker_id), 1U);
2368
2369 // Parallel processing task executor.
2370 G1CMRefProcTaskExecutor par_task_executor(g1h, this,
2371 g1h->workers(), active_workers);
2372 AbstractRefProcTaskExecutor* executor = (processing_is_mt ? &par_task_executor : NULL);
2373
2374 // Set the concurrency level. The phase was already set prior to
2375 // executing the remark task.
2376 set_concurrency(active_workers);
2377
2378 // Set the degree of MT processing here. If the discovery was done MT,
2379 // the number of threads involved during discovery could differ from
2380 // the number of active workers. This is OK as long as the discovered
2381 // Reference lists are balanced (see balance_all_queues() and balance_queues()).
2382 rp->set_active_mt_degree(active_workers);
2383
2384 // Process the weak references.
2385 const ReferenceProcessorStats& stats =
2386 rp->process_discovered_references(&g1_is_alive,
2387 &g1_keep_alive,
2388 &g1_drain_mark_stack,
2389 executor,
2390 g1h->gc_timer_cm(),
2391 concurrent_gc_id());
2392 g1h->gc_tracer_cm()->report_gc_reference_stats(stats);
2393
2394 // The do_oop work routines of the keep_alive and drain_marking_stack
2395 // oop closures will set the has_overflown flag if we overflow the
2396 // global marking stack.
2397
2398 assert(_markStack.overflow() || _markStack.isEmpty(),
2399 "mark stack should be empty (unless it overflowed)");
2400
2401 if (_markStack.overflow()) {
2402 // This should have been done already when we tried to push an
2403 // entry on to the global mark stack. But let's do it again.
2404 set_has_overflown();
2405 }
2406
2407 assert(rp->num_q() == active_workers, "why not");
2408
2409 rp->enqueue_discovered_references(executor);
2410
2411 rp->verify_no_references_recorded();
2412 assert(!rp->discovery_enabled(), "Post condition");
2413 }
2414
2415 if (has_overflown()) {
2416 // We can not trust g1_is_alive if the marking stack overflowed
2417 return;
2418 }
2419
2420 assert(_markStack.isEmpty(), "Marking should have completed");
2421
2422 // Unload Klasses, String, Symbols, Code Cache, etc.
2423 {
2424 G1CMTraceTime trace("Unloading", G1Log::finer());
2425
2426 if (ClassUnloadingWithConcurrentMark) {
2427 bool purged_classes;
2428
2429 {
2430 G1CMTraceTime trace("System Dictionary Unloading", G1Log::finest());
2431 purged_classes = SystemDictionary::do_unloading(&g1_is_alive, false /* Defer klass cleaning */);
2432 }
2433
2434 {
2435 G1CMTraceTime trace("Parallel Unloading", G1Log::finest());
2436 weakRefsWorkParallelPart(&g1_is_alive, purged_classes);
2437 }
2438 }
2439
2440 if (G1StringDedup::is_enabled()) {
2441 G1CMTraceTime trace("String Deduplication Unlink", G1Log::finest());
2442 G1StringDedup::unlink(&g1_is_alive);
2443 }
2444 }
2445 }
2446
2447 void ConcurrentMark::swapMarkBitMaps() {
2448 CMBitMapRO* temp = _prevMarkBitMap;
2449 _prevMarkBitMap = (CMBitMapRO*)_nextMarkBitMap;
2450 _nextMarkBitMap = (CMBitMap*) temp;
2451 }
2452
2453 // Closure for marking entries in SATB buffers.
2454 class CMSATBBufferClosure : public SATBBufferClosure {
2455 private:
2456 CMTask* _task;
2457 G1CollectedHeap* _g1h;
2458
2459 // This is very similar to CMTask::deal_with_reference, but with
2460 // more relaxed requirements for the argument, so this must be more
2461 // circumspect about treating the argument as an object.
2462 void do_entry(void* entry) const {
2463 _task->increment_refs_reached();
2464 HeapRegion* hr = _g1h->heap_region_containing_raw(entry);
2465 if (entry < hr->next_top_at_mark_start()) {
2466 // Until we get here, we don't know whether entry refers to a valid
2467 // object; it could instead have been a stale reference.
2468 oop obj = static_cast<oop>(entry);
2469 assert(obj->is_oop(true /* ignore mark word */),
2470 err_msg("Invalid oop in SATB buffer: " PTR_FORMAT, p2i(obj)));
2471 _task->make_reference_grey(obj, hr);
2472 }
2473 }
2474
2475 public:
2476 CMSATBBufferClosure(CMTask* task, G1CollectedHeap* g1h)
2477 : _task(task), _g1h(g1h) { }
2478
2479 virtual void do_buffer(void** buffer, size_t size) {
2480 for (size_t i = 0; i < size; ++i) {
2481 do_entry(buffer[i]);
2482 }
2483 }
2484 };
2485
2486 class G1RemarkThreadsClosure : public ThreadClosure {
2487 CMSATBBufferClosure _cm_satb_cl;
2488 G1CMOopClosure _cm_cl;
2489 MarkingCodeBlobClosure _code_cl;
2490 int _thread_parity;
2491
2492 public:
2493 G1RemarkThreadsClosure(G1CollectedHeap* g1h, CMTask* task) :
2494 _cm_satb_cl(task, g1h),
2495 _cm_cl(g1h, g1h->concurrent_mark(), task),
2496 _code_cl(&_cm_cl, !CodeBlobToOopClosure::FixRelocations),
2497 _thread_parity(Threads::thread_claim_parity()) {}
2498
2499 void do_thread(Thread* thread) {
2500 if (thread->is_Java_thread()) {
2501 if (thread->claim_oops_do(true, _thread_parity)) {
2502 JavaThread* jt = (JavaThread*)thread;
2503
2504 // In theory it should not be neccessary to explicitly walk the nmethods to find roots for concurrent marking
2505 // however the liveness of oops reachable from nmethods have very complex lifecycles:
2506 // * Alive if on the stack of an executing method
2507 // * Weakly reachable otherwise
2508 // Some objects reachable from nmethods, such as the class loader (or klass_holder) of the receiver should be
2509 // live by the SATB invariant but other oops recorded in nmethods may behave differently.
2510 jt->nmethods_do(&_code_cl);
2511
2512 jt->satb_mark_queue().apply_closure_and_empty(&_cm_satb_cl);
2513 }
2514 } else if (thread->is_VM_thread()) {
2515 if (thread->claim_oops_do(true, _thread_parity)) {
2516 JavaThread::satb_mark_queue_set().shared_satb_queue()->apply_closure_and_empty(&_cm_satb_cl);
2517 }
2518 }
2519 }
2520 };
2521
2522 class CMRemarkTask: public AbstractGangTask {
2523 private:
2524 ConcurrentMark* _cm;
2525 public:
2526 void work(uint worker_id) {
2527 // Since all available tasks are actually started, we should
2528 // only proceed if we're supposed to be active.
2529 if (worker_id < _cm->active_tasks()) {
2530 CMTask* task = _cm->task(worker_id);
2531 task->record_start_time();
2532 {
2533 ResourceMark rm;
2534 HandleMark hm;
2535
2536 G1RemarkThreadsClosure threads_f(G1CollectedHeap::heap(), task);
2537 Threads::threads_do(&threads_f);
2538 }
2539
2540 do {
2541 task->do_marking_step(1000000000.0 /* something very large */,
2542 true /* do_termination */,
2543 false /* is_serial */);
2544 } while (task->has_aborted() && !_cm->has_overflown());
2545 // If we overflow, then we do not want to restart. We instead
2546 // want to abort remark and do concurrent marking again.
2547 task->record_end_time();
2548 }
2549 }
2550
2551 CMRemarkTask(ConcurrentMark* cm, uint active_workers) :
2552 AbstractGangTask("Par Remark"), _cm(cm) {
2553 _cm->terminator()->reset_for_reuse(active_workers);
2554 }
2555 };
2556
2557 void ConcurrentMark::checkpointRootsFinalWork() {
2558 ResourceMark rm;
2559 HandleMark hm;
2560 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2561
2562 G1CMTraceTime trace("Finalize Marking", G1Log::finer());
2563
2564 g1h->ensure_parsability(false);
2565
2566 // this is remark, so we'll use up all active threads
2567 uint active_workers = g1h->workers()->active_workers();
2568 set_concurrency_and_phase(active_workers, false /* concurrent */);
2569 // Leave _parallel_marking_threads at it's
2570 // value originally calculated in the ConcurrentMark
2571 // constructor and pass values of the active workers
2572 // through the gang in the task.
2573
2574 {
2575 StrongRootsScope srs(active_workers);
2576
2577 CMRemarkTask remarkTask(this, active_workers);
2578 // We will start all available threads, even if we decide that the
2579 // active_workers will be fewer. The extra ones will just bail out
2580 // immediately.
2581 g1h->workers()->run_task(&remarkTask);
2582 }
2583
2584 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2585 guarantee(has_overflown() ||
2586 satb_mq_set.completed_buffers_num() == 0,
2587 err_msg("Invariant: has_overflown = %s, num buffers = %d",
2588 BOOL_TO_STR(has_overflown()),
2589 satb_mq_set.completed_buffers_num()));
2590
2591 print_stats();
2592 }
2593
2594 void ConcurrentMark::clearRangePrevBitmap(MemRegion mr) {
2595 // Note we are overriding the read-only view of the prev map here, via
2596 // the cast.
2597 ((CMBitMap*)_prevMarkBitMap)->clearRange(mr);
2598 }
2599
2600 void ConcurrentMark::clearRangeNextBitmap(MemRegion mr) {
2601 _nextMarkBitMap->clearRange(mr);
2602 }
2603
2604 HeapRegion*
2605 ConcurrentMark::claim_region(uint worker_id) {
2606 // "checkpoint" the finger
2607 HeapWord* finger = _finger;
2608
2609 // _heap_end will not change underneath our feet; it only changes at
2610 // yield points.
2611 while (finger < _heap_end) {
2612 assert(_g1h->is_in_g1_reserved(finger), "invariant");
2613
2614 // Note on how this code handles humongous regions. In the
2615 // normal case the finger will reach the start of a "starts
2616 // humongous" (SH) region. Its end will either be the end of the
2617 // last "continues humongous" (CH) region in the sequence, or the
2618 // standard end of the SH region (if the SH is the only region in
2619 // the sequence). That way claim_region() will skip over the CH
2620 // regions. However, there is a subtle race between a CM thread
2621 // executing this method and a mutator thread doing a humongous
2622 // object allocation. The two are not mutually exclusive as the CM
2623 // thread does not need to hold the Heap_lock when it gets
2624 // here. So there is a chance that claim_region() will come across
2625 // a free region that's in the progress of becoming a SH or a CH
2626 // region. In the former case, it will either
2627 // a) Miss the update to the region's end, in which case it will
2628 // visit every subsequent CH region, will find their bitmaps
2629 // empty, and do nothing, or
2630 // b) Will observe the update of the region's end (in which case
2631 // it will skip the subsequent CH regions).
2632 // If it comes across a region that suddenly becomes CH, the
2633 // scenario will be similar to b). So, the race between
2634 // claim_region() and a humongous object allocation might force us
2635 // to do a bit of unnecessary work (due to some unnecessary bitmap
2636 // iterations) but it should not introduce and correctness issues.
2637 HeapRegion* curr_region = _g1h->heap_region_containing_raw(finger);
2638
2639 // Above heap_region_containing_raw may return NULL as we always scan claim
2640 // until the end of the heap. In this case, just jump to the next region.
2641 HeapWord* end = curr_region != NULL ? curr_region->end() : finger + HeapRegion::GrainWords;
2642
2643 // Is the gap between reading the finger and doing the CAS too long?
2644 HeapWord* res = (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger);
2645 if (res == finger && curr_region != NULL) {
2646 // we succeeded
2647 HeapWord* bottom = curr_region->bottom();
2648 HeapWord* limit = curr_region->next_top_at_mark_start();
2649
2650 if (verbose_low()) {
2651 gclog_or_tty->print_cr("[%u] curr_region = "PTR_FORMAT" "
2652 "["PTR_FORMAT", "PTR_FORMAT"), "
2653 "limit = "PTR_FORMAT,
2654 worker_id, p2i(curr_region), p2i(bottom), p2i(end), p2i(limit));
2655 }
2656
2657 // notice that _finger == end cannot be guaranteed here since,
2658 // someone else might have moved the finger even further
2659 assert(_finger >= end, "the finger should have moved forward");
2660
2661 if (verbose_low()) {
2662 gclog_or_tty->print_cr("[%u] we were successful with region = "
2663 PTR_FORMAT, worker_id, p2i(curr_region));
2664 }
2665
2666 if (limit > bottom) {
2667 if (verbose_low()) {
2668 gclog_or_tty->print_cr("[%u] region "PTR_FORMAT" is not empty, "
2669 "returning it ", worker_id, p2i(curr_region));
2670 }
2671 return curr_region;
2672 } else {
2673 assert(limit == bottom,
2674 "the region limit should be at bottom");
2675 if (verbose_low()) {
2676 gclog_or_tty->print_cr("[%u] region "PTR_FORMAT" is empty, "
2677 "returning NULL", worker_id, p2i(curr_region));
2678 }
2679 // we return NULL and the caller should try calling
2680 // claim_region() again.
2681 return NULL;
2682 }
2683 } else {
2684 assert(_finger > finger, "the finger should have moved forward");
2685 if (verbose_low()) {
2686 if (curr_region == NULL) {
2687 gclog_or_tty->print_cr("[%u] found uncommitted region, moving finger, "
2688 "global finger = "PTR_FORMAT", "
2689 "our finger = "PTR_FORMAT,
2690 worker_id, p2i(_finger), p2i(finger));
2691 } else {
2692 gclog_or_tty->print_cr("[%u] somebody else moved the finger, "
2693 "global finger = "PTR_FORMAT", "
2694 "our finger = "PTR_FORMAT,
2695 worker_id, p2i(_finger), p2i(finger));
2696 }
2697 }
2698
2699 // read it again
2700 finger = _finger;
2701 }
2702 }
2703
2704 return NULL;
2705 }
2706
2707 #ifndef PRODUCT
2708 enum VerifyNoCSetOopsPhase {
2709 VerifyNoCSetOopsStack,
2710 VerifyNoCSetOopsQueues
2711 };
2712
2713 class VerifyNoCSetOopsClosure : public OopClosure, public ObjectClosure {
2714 private:
2715 G1CollectedHeap* _g1h;
2716 VerifyNoCSetOopsPhase _phase;
2717 int _info;
2718
2719 const char* phase_str() {
2720 switch (_phase) {
2721 case VerifyNoCSetOopsStack: return "Stack";
2722 case VerifyNoCSetOopsQueues: return "Queue";
2723 default: ShouldNotReachHere();
2724 }
2725 return NULL;
2726 }
2727
2728 void do_object_work(oop obj) {
2729 guarantee(!_g1h->obj_in_cs(obj),
2730 err_msg("obj: "PTR_FORMAT" in CSet, phase: %s, info: %d",
2731 p2i((void*) obj), phase_str(), _info));
2732 }
2733
2734 public:
2735 VerifyNoCSetOopsClosure() : _g1h(G1CollectedHeap::heap()) { }
2736
2737 void set_phase(VerifyNoCSetOopsPhase phase, int info = -1) {
2738 _phase = phase;
2739 _info = info;
2740 }
2741
2742 virtual void do_oop(oop* p) {
2743 oop obj = oopDesc::load_decode_heap_oop(p);
2744 do_object_work(obj);
2745 }
2746
2747 virtual void do_oop(narrowOop* p) {
2748 // We should not come across narrow oops while scanning marking
2749 // stacks
2750 ShouldNotReachHere();
2751 }
2752
2753 virtual void do_object(oop obj) {
2754 do_object_work(obj);
2755 }
2756 };
2757
2758 void ConcurrentMark::verify_no_cset_oops() {
2759 assert(SafepointSynchronize::is_at_safepoint(), "should be at a safepoint");
2760 if (!G1CollectedHeap::heap()->collector_state()->mark_in_progress()) {
2761 return;
2762 }
2763
2764 VerifyNoCSetOopsClosure cl;
2765
2766 // Verify entries on the global mark stack
2767 cl.set_phase(VerifyNoCSetOopsStack);
2768 _markStack.oops_do(&cl);
2769
2770 // Verify entries on the task queues
2771 for (uint i = 0; i < _max_worker_id; i += 1) {
2772 cl.set_phase(VerifyNoCSetOopsQueues, i);
2773 CMTaskQueue* queue = _task_queues->queue(i);
2774 queue->oops_do(&cl);
2775 }
2776
2777 // Verify the global finger
2778 HeapWord* global_finger = finger();
2779 if (global_finger != NULL && global_finger < _heap_end) {
2780 // The global finger always points to a heap region boundary. We
2781 // use heap_region_containing_raw() to get the containing region
2782 // given that the global finger could be pointing to a free region
2783 // which subsequently becomes continues humongous. If that
2784 // happens, heap_region_containing() will return the bottom of the
2785 // corresponding starts humongous region and the check below will
2786 // not hold any more.
2787 // Since we always iterate over all regions, we might get a NULL HeapRegion
2788 // here.
2789 HeapRegion* global_hr = _g1h->heap_region_containing_raw(global_finger);
2790 guarantee(global_hr == NULL || global_finger == global_hr->bottom(),
2791 err_msg("global finger: "PTR_FORMAT" region: "HR_FORMAT,
2792 p2i(global_finger), HR_FORMAT_PARAMS(global_hr)));
2793 }
2794
2795 // Verify the task fingers
2796 assert(parallel_marking_threads() <= _max_worker_id, "sanity");
2797 for (int i = 0; i < (int) parallel_marking_threads(); i += 1) {
2798 CMTask* task = _tasks[i];
2799 HeapWord* task_finger = task->finger();
2800 if (task_finger != NULL && task_finger < _heap_end) {
2801 // See above note on the global finger verification.
2802 HeapRegion* task_hr = _g1h->heap_region_containing_raw(task_finger);
2803 guarantee(task_hr == NULL || task_finger == task_hr->bottom() ||
2804 !task_hr->in_collection_set(),
2805 err_msg("task finger: "PTR_FORMAT" region: "HR_FORMAT,
2806 p2i(task_finger), HR_FORMAT_PARAMS(task_hr)));
2807 }
2808 }
2809 }
2810 #endif // PRODUCT
2811
2812 // Aggregate the counting data that was constructed concurrently
2813 // with marking.
2814 class AggregateCountDataHRClosure: public HeapRegionClosure {
2815 G1CollectedHeap* _g1h;
2816 ConcurrentMark* _cm;
2817 CardTableModRefBS* _ct_bs;
2818 BitMap* _cm_card_bm;
2819 uint _max_worker_id;
2820
2821 public:
2822 AggregateCountDataHRClosure(G1CollectedHeap* g1h,
2823 BitMap* cm_card_bm,
2824 uint max_worker_id) :
2825 _g1h(g1h), _cm(g1h->concurrent_mark()),
2826 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
2827 _cm_card_bm(cm_card_bm), _max_worker_id(max_worker_id) { }
2828
2829 bool doHeapRegion(HeapRegion* hr) {
2830 if (hr->is_continues_humongous()) {
2831 // We will ignore these here and process them when their
2832 // associated "starts humongous" region is processed.
2833 // Note that we cannot rely on their associated
2834 // "starts humongous" region to have their bit set to 1
2835 // since, due to the region chunking in the parallel region
2836 // iteration, a "continues humongous" region might be visited
2837 // before its associated "starts humongous".
2838 return false;
2839 }
2840
2841 HeapWord* start = hr->bottom();
2842 HeapWord* limit = hr->next_top_at_mark_start();
2843 HeapWord* end = hr->end();
2844
2845 assert(start <= limit && limit <= hr->top() && hr->top() <= hr->end(),
2846 err_msg("Preconditions not met - "
2847 "start: "PTR_FORMAT", limit: "PTR_FORMAT", "
2848 "top: "PTR_FORMAT", end: "PTR_FORMAT,
2849 p2i(start), p2i(limit), p2i(hr->top()), p2i(hr->end())));
2850
2851 assert(hr->next_marked_bytes() == 0, "Precondition");
2852
2853 if (start == limit) {
2854 // NTAMS of this region has not been set so nothing to do.
2855 return false;
2856 }
2857
2858 // 'start' should be in the heap.
2859 assert(_g1h->is_in_g1_reserved(start) && _ct_bs->is_card_aligned(start), "sanity");
2860 // 'end' *may* be just beyond the end of the heap (if hr is the last region)
2861 assert(!_g1h->is_in_g1_reserved(end) || _ct_bs->is_card_aligned(end), "sanity");
2862
2863 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
2864 BitMap::idx_t limit_idx = _cm->card_bitmap_index_for(limit);
2865 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(end);
2866
2867 // If ntams is not card aligned then we bump card bitmap index
2868 // for limit so that we get the all the cards spanned by
2869 // the object ending at ntams.
2870 // Note: if this is the last region in the heap then ntams
2871 // could be actually just beyond the end of the the heap;
2872 // limit_idx will then correspond to a (non-existent) card
2873 // that is also outside the heap.
2874 if (_g1h->is_in_g1_reserved(limit) && !_ct_bs->is_card_aligned(limit)) {
2875 limit_idx += 1;
2876 }
2877
2878 assert(limit_idx <= end_idx, "or else use atomics");
2879
2880 // Aggregate the "stripe" in the count data associated with hr.
2881 uint hrm_index = hr->hrm_index();
2882 size_t marked_bytes = 0;
2883
2884 for (uint i = 0; i < _max_worker_id; i += 1) {
2885 size_t* marked_bytes_array = _cm->count_marked_bytes_array_for(i);
2886 BitMap* task_card_bm = _cm->count_card_bitmap_for(i);
2887
2888 // Fetch the marked_bytes in this region for task i and
2889 // add it to the running total for this region.
2890 marked_bytes += marked_bytes_array[hrm_index];
2891
2892 // Now union the bitmaps[0,max_worker_id)[start_idx..limit_idx)
2893 // into the global card bitmap.
2894 BitMap::idx_t scan_idx = task_card_bm->get_next_one_offset(start_idx, limit_idx);
2895
2896 while (scan_idx < limit_idx) {
2897 assert(task_card_bm->at(scan_idx) == true, "should be");
2898 _cm_card_bm->set_bit(scan_idx);
2899 assert(_cm_card_bm->at(scan_idx) == true, "should be");
2900
2901 // BitMap::get_next_one_offset() can handle the case when
2902 // its left_offset parameter is greater than its right_offset
2903 // parameter. It does, however, have an early exit if
2904 // left_offset == right_offset. So let's limit the value
2905 // passed in for left offset here.
2906 BitMap::idx_t next_idx = MIN2(scan_idx + 1, limit_idx);
2907 scan_idx = task_card_bm->get_next_one_offset(next_idx, limit_idx);
2908 }
2909 }
2910
2911 // Update the marked bytes for this region.
2912 hr->add_to_marked_bytes(marked_bytes);
2913
2914 // Next heap region
2915 return false;
2916 }
2917 };
2918
2919 class G1AggregateCountDataTask: public AbstractGangTask {
2920 protected:
2921 G1CollectedHeap* _g1h;
2922 ConcurrentMark* _cm;
2923 BitMap* _cm_card_bm;
2924 uint _max_worker_id;
2925 uint _active_workers;
2926 HeapRegionClaimer _hrclaimer;
2927
2928 public:
2929 G1AggregateCountDataTask(G1CollectedHeap* g1h,
2930 ConcurrentMark* cm,
2931 BitMap* cm_card_bm,
2932 uint max_worker_id,
2933 uint n_workers) :
2934 AbstractGangTask("Count Aggregation"),
2935 _g1h(g1h), _cm(cm), _cm_card_bm(cm_card_bm),
2936 _max_worker_id(max_worker_id),
2937 _active_workers(n_workers),
2938 _hrclaimer(_active_workers) {
2939 }
2940
2941 void work(uint worker_id) {
2942 AggregateCountDataHRClosure cl(_g1h, _cm_card_bm, _max_worker_id);
2943
2944 _g1h->heap_region_par_iterate(&cl, worker_id, &_hrclaimer);
2945 }
2946 };
2947
2948
2949 void ConcurrentMark::aggregate_count_data() {
2950 uint n_workers = _g1h->workers()->active_workers();
2951
2952 G1AggregateCountDataTask g1_par_agg_task(_g1h, this, &_card_bm,
2953 _max_worker_id, n_workers);
2954
2955 _g1h->workers()->run_task(&g1_par_agg_task);
2956 }
2957
2958 // Clear the per-worker arrays used to store the per-region counting data
2959 void ConcurrentMark::clear_all_count_data() {
2960 // Clear the global card bitmap - it will be filled during
2961 // liveness count aggregation (during remark) and the
2962 // final counting task.
2963 _card_bm.clear();
2964
2965 // Clear the global region bitmap - it will be filled as part
2966 // of the final counting task.
2967 _region_bm.clear();
2968
2969 uint max_regions = _g1h->max_regions();
2970 assert(_max_worker_id > 0, "uninitialized");
2971
2972 for (uint i = 0; i < _max_worker_id; i += 1) {
2973 BitMap* task_card_bm = count_card_bitmap_for(i);
2974 size_t* marked_bytes_array = count_marked_bytes_array_for(i);
2975
2976 assert(task_card_bm->size() == _card_bm.size(), "size mismatch");
2977 assert(marked_bytes_array != NULL, "uninitialized");
2978
2979 memset(marked_bytes_array, 0, (size_t) max_regions * sizeof(size_t));
2980 task_card_bm->clear();
2981 }
2982 }
2983
2984 void ConcurrentMark::print_stats() {
2985 if (verbose_stats()) {
2986 gclog_or_tty->print_cr("---------------------------------------------------------------------");
2987 for (size_t i = 0; i < _active_tasks; ++i) {
2988 _tasks[i]->print_stats();
2989 gclog_or_tty->print_cr("---------------------------------------------------------------------");
2990 }
2991 }
2992 }
2993
2994 // abandon current marking iteration due to a Full GC
2995 void ConcurrentMark::abort() {
2996 if (!cmThread()->during_cycle() || _has_aborted) {
2997 // We haven't started a concurrent cycle or we have already aborted it. No need to do anything.
2998 return;
2999 }
3000
3001 // Clear all marks in the next bitmap for the next marking cycle. This will allow us to skip the next
3002 // concurrent bitmap clearing.
3003 _nextMarkBitMap->clearAll();
3004
3005 // Note we cannot clear the previous marking bitmap here
3006 // since VerifyDuringGC verifies the objects marked during
3007 // a full GC against the previous bitmap.
3008
3009 // Clear the liveness counting data
3010 clear_all_count_data();
3011 // Empty mark stack
3012 reset_marking_state();
3013 for (uint i = 0; i < _max_worker_id; ++i) {
3014 _tasks[i]->clear_region_fields();
3015 }
3016 _first_overflow_barrier_sync.abort();
3017 _second_overflow_barrier_sync.abort();
3018 _aborted_gc_id = _g1h->gc_tracer_cm()->gc_id();
3019 assert(!_aborted_gc_id.is_undefined(), "abort() executed more than once?");
3020 _has_aborted = true;
3021
3022 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3023 satb_mq_set.abandon_partial_marking();
3024 // This can be called either during or outside marking, we'll read
3025 // the expected_active value from the SATB queue set.
3026 satb_mq_set.set_active_all_threads(
3027 false, /* new active value */
3028 satb_mq_set.is_active() /* expected_active */);
3029
3030 _g1h->trace_heap_after_concurrent_cycle();
3031 _g1h->register_concurrent_cycle_end();
3032 }
3033
3034 const GCId& ConcurrentMark::concurrent_gc_id() {
3035 if (has_aborted()) {
3036 return _aborted_gc_id;
3037 }
3038 return _g1h->gc_tracer_cm()->gc_id();
3039 }
3040
3041 static void print_ms_time_info(const char* prefix, const char* name,
3042 NumberSeq& ns) {
3043 gclog_or_tty->print_cr("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
3044 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
3045 if (ns.num() > 0) {
3046 gclog_or_tty->print_cr("%s [std. dev = %8.2f ms, max = %8.2f ms]",
3047 prefix, ns.sd(), ns.maximum());
3048 }
3049 }
3050
3051 void ConcurrentMark::print_summary_info() {
3052 gclog_or_tty->print_cr(" Concurrent marking:");
3053 print_ms_time_info(" ", "init marks", _init_times);
3054 print_ms_time_info(" ", "remarks", _remark_times);
3055 {
3056 print_ms_time_info(" ", "final marks", _remark_mark_times);
3057 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
3058
3059 }
3060 print_ms_time_info(" ", "cleanups", _cleanup_times);
3061 gclog_or_tty->print_cr(" Final counting total time = %8.2f s (avg = %8.2f ms).",
3062 _total_counting_time,
3063 (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 /
3064 (double)_cleanup_times.num()
3065 : 0.0));
3066 if (G1ScrubRemSets) {
3067 gclog_or_tty->print_cr(" RS scrub total time = %8.2f s (avg = %8.2f ms).",
3068 _total_rs_scrub_time,
3069 (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 /
3070 (double)_cleanup_times.num()
3071 : 0.0));
3072 }
3073 gclog_or_tty->print_cr(" Total stop_world time = %8.2f s.",
3074 (_init_times.sum() + _remark_times.sum() +
3075 _cleanup_times.sum())/1000.0);
3076 gclog_or_tty->print_cr(" Total concurrent time = %8.2f s "
3077 "(%8.2f s marking).",
3078 cmThread()->vtime_accum(),
3079 cmThread()->vtime_mark_accum());
3080 }
3081
3082 void ConcurrentMark::print_worker_threads_on(outputStream* st) const {
3083 _parallel_workers->print_worker_threads_on(st);
3084 }
3085
3086 void ConcurrentMark::print_on_error(outputStream* st) const {
3087 st->print_cr("Marking Bits (Prev, Next): (CMBitMap*) " PTR_FORMAT ", (CMBitMap*) " PTR_FORMAT,
3088 p2i(_prevMarkBitMap), p2i(_nextMarkBitMap));
3089 _prevMarkBitMap->print_on_error(st, " Prev Bits: ");
3090 _nextMarkBitMap->print_on_error(st, " Next Bits: ");
3091 }
3092
3093 // We take a break if someone is trying to stop the world.
3094 bool ConcurrentMark::do_yield_check(uint worker_id) {
3095 if (SuspendibleThreadSet::should_yield()) {
3096 if (worker_id == 0) {
3097 _g1h->g1_policy()->record_concurrent_pause();
3098 }
3099 SuspendibleThreadSet::yield();
3100 return true;
3101 } else {
3102 return false;
3103 }
3104 }
3105
3106 #ifndef PRODUCT
3107 // for debugging purposes
3108 void ConcurrentMark::print_finger() {
3109 gclog_or_tty->print_cr("heap ["PTR_FORMAT", "PTR_FORMAT"), global finger = "PTR_FORMAT,
3110 p2i(_heap_start), p2i(_heap_end), p2i(_finger));
3111 for (uint i = 0; i < _max_worker_id; ++i) {
3112 gclog_or_tty->print(" %u: " PTR_FORMAT, i, p2i(_tasks[i]->finger()));
3113 }
3114 gclog_or_tty->cr();
3115 }
3116 #endif
3117
3118 template<bool scan>
3119 inline void CMTask::process_grey_object(oop obj) {
3120 assert(scan || obj->is_typeArray(), "Skipping scan of grey non-typeArray");
3121 assert(_nextMarkBitMap->isMarked((HeapWord*) obj), "invariant");
3122
3123 if (_cm->verbose_high()) {
3124 gclog_or_tty->print_cr("[%u] processing grey object " PTR_FORMAT,
3125 _worker_id, p2i((void*) obj));
3126 }
3127
3128 size_t obj_size = obj->size();
3129 _words_scanned += obj_size;
3130
3131 if (scan) {
3132 obj->oop_iterate(_cm_oop_closure);
3133 }
3134 statsOnly( ++_objs_scanned );
3135 check_limits();
3136 }
3137
3138 template void CMTask::process_grey_object<true>(oop);
3139 template void CMTask::process_grey_object<false>(oop);
3140
3141 // Closure for iteration over bitmaps
3142 class CMBitMapClosure : public BitMapClosure {
3143 private:
3144 // the bitmap that is being iterated over
3145 CMBitMap* _nextMarkBitMap;
3146 ConcurrentMark* _cm;
3147 CMTask* _task;
3148
3149 public:
3150 CMBitMapClosure(CMTask *task, ConcurrentMark* cm, CMBitMap* nextMarkBitMap) :
3151 _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { }
3152
3153 bool do_bit(size_t offset) {
3154 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset);
3155 assert(_nextMarkBitMap->isMarked(addr), "invariant");
3156 assert( addr < _cm->finger(), "invariant");
3157
3158 statsOnly( _task->increase_objs_found_on_bitmap() );
3159 assert(addr >= _task->finger(), "invariant");
3160
3161 // We move that task's local finger along.
3162 _task->move_finger_to(addr);
3163
3164 _task->scan_object(oop(addr));
3165 // we only partially drain the local queue and global stack
3166 _task->drain_local_queue(true);
3167 _task->drain_global_stack(true);
3168
3169 // if the has_aborted flag has been raised, we need to bail out of
3170 // the iteration
3171 return !_task->has_aborted();
3172 }
3173 };
3174
3175 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
3176 ConcurrentMark* cm,
3177 CMTask* task)
3178 : _g1h(g1h), _cm(cm), _task(task) {
3179 assert(_ref_processor == NULL, "should be initialized to NULL");
3180
3181 if (G1UseConcMarkReferenceProcessing) {
3182 _ref_processor = g1h->ref_processor_cm();
3183 assert(_ref_processor != NULL, "should not be NULL");
3184 }
3185 }
3186
3187 void CMTask::setup_for_region(HeapRegion* hr) {
3188 assert(hr != NULL,
3189 "claim_region() should have filtered out NULL regions");
3190 assert(!hr->is_continues_humongous(),
3191 "claim_region() should have filtered out continues humongous regions");
3192
3193 if (_cm->verbose_low()) {
3194 gclog_or_tty->print_cr("[%u] setting up for region "PTR_FORMAT,
3195 _worker_id, p2i(hr));
3196 }
3197
3198 _curr_region = hr;
3199 _finger = hr->bottom();
3200 update_region_limit();
3201 }
3202
3203 void CMTask::update_region_limit() {
3204 HeapRegion* hr = _curr_region;
3205 HeapWord* bottom = hr->bottom();
3206 HeapWord* limit = hr->next_top_at_mark_start();
3207
3208 if (limit == bottom) {
3209 if (_cm->verbose_low()) {
3210 gclog_or_tty->print_cr("[%u] found an empty region "
3211 "["PTR_FORMAT", "PTR_FORMAT")",
3212 _worker_id, p2i(bottom), p2i(limit));
3213 }
3214 // The region was collected underneath our feet.
3215 // We set the finger to bottom to ensure that the bitmap
3216 // iteration that will follow this will not do anything.
3217 // (this is not a condition that holds when we set the region up,
3218 // as the region is not supposed to be empty in the first place)
3219 _finger = bottom;
3220 } else if (limit >= _region_limit) {
3221 assert(limit >= _finger, "peace of mind");
3222 } else {
3223 assert(limit < _region_limit, "only way to get here");
3224 // This can happen under some pretty unusual circumstances. An
3225 // evacuation pause empties the region underneath our feet (NTAMS
3226 // at bottom). We then do some allocation in the region (NTAMS
3227 // stays at bottom), followed by the region being used as a GC
3228 // alloc region (NTAMS will move to top() and the objects
3229 // originally below it will be grayed). All objects now marked in
3230 // the region are explicitly grayed, if below the global finger,
3231 // and we do not need in fact to scan anything else. So, we simply
3232 // set _finger to be limit to ensure that the bitmap iteration
3233 // doesn't do anything.
3234 _finger = limit;
3235 }
3236
3237 _region_limit = limit;
3238 }
3239
3240 void CMTask::giveup_current_region() {
3241 assert(_curr_region != NULL, "invariant");
3242 if (_cm->verbose_low()) {
3243 gclog_or_tty->print_cr("[%u] giving up region "PTR_FORMAT,
3244 _worker_id, p2i(_curr_region));
3245 }
3246 clear_region_fields();
3247 }
3248
3249 void CMTask::clear_region_fields() {
3250 // Values for these three fields that indicate that we're not
3251 // holding on to a region.
3252 _curr_region = NULL;
3253 _finger = NULL;
3254 _region_limit = NULL;
3255 }
3256
3257 void CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
3258 if (cm_oop_closure == NULL) {
3259 assert(_cm_oop_closure != NULL, "invariant");
3260 } else {
3261 assert(_cm_oop_closure == NULL, "invariant");
3262 }
3263 _cm_oop_closure = cm_oop_closure;
3264 }
3265
3266 void CMTask::reset(CMBitMap* nextMarkBitMap) {
3267 guarantee(nextMarkBitMap != NULL, "invariant");
3268
3269 if (_cm->verbose_low()) {
3270 gclog_or_tty->print_cr("[%u] resetting", _worker_id);
3271 }
3272
3273 _nextMarkBitMap = nextMarkBitMap;
3274 clear_region_fields();
3275
3276 _calls = 0;
3277 _elapsed_time_ms = 0.0;
3278 _termination_time_ms = 0.0;
3279 _termination_start_time_ms = 0.0;
3280
3281 #if _MARKING_STATS_
3282 _aborted = 0;
3283 _aborted_overflow = 0;
3284 _aborted_cm_aborted = 0;
3285 _aborted_yield = 0;
3286 _aborted_timed_out = 0;
3287 _aborted_satb = 0;
3288 _aborted_termination = 0;
3289 _steal_attempts = 0;
3290 _steals = 0;
3291 _local_pushes = 0;
3292 _local_pops = 0;
3293 _local_max_size = 0;
3294 _objs_scanned = 0;
3295 _global_pushes = 0;
3296 _global_pops = 0;
3297 _global_max_size = 0;
3298 _global_transfers_to = 0;
3299 _global_transfers_from = 0;
3300 _regions_claimed = 0;
3301 _objs_found_on_bitmap = 0;
3302 _satb_buffers_processed = 0;
3303 #endif // _MARKING_STATS_
3304 }
3305
3306 bool CMTask::should_exit_termination() {
3307 regular_clock_call();
3308 // This is called when we are in the termination protocol. We should
3309 // quit if, for some reason, this task wants to abort or the global
3310 // stack is not empty (this means that we can get work from it).
3311 return !_cm->mark_stack_empty() || has_aborted();
3312 }
3313
3314 void CMTask::reached_limit() {
3315 assert(_words_scanned >= _words_scanned_limit ||
3316 _refs_reached >= _refs_reached_limit ,
3317 "shouldn't have been called otherwise");
3318 regular_clock_call();
3319 }
3320
3321 void CMTask::regular_clock_call() {
3322 if (has_aborted()) return;
3323
3324 // First, we need to recalculate the words scanned and refs reached
3325 // limits for the next clock call.
3326 recalculate_limits();
3327
3328 // During the regular clock call we do the following
3329
3330 // (1) If an overflow has been flagged, then we abort.
3331 if (_cm->has_overflown()) {
3332 set_has_aborted();
3333 return;
3334 }
3335
3336 // If we are not concurrent (i.e. we're doing remark) we don't need
3337 // to check anything else. The other steps are only needed during
3338 // the concurrent marking phase.
3339 if (!concurrent()) return;
3340
3341 // (2) If marking has been aborted for Full GC, then we also abort.
3342 if (_cm->has_aborted()) {
3343 set_has_aborted();
3344 statsOnly( ++_aborted_cm_aborted );
3345 return;
3346 }
3347
3348 double curr_time_ms = os::elapsedVTime() * 1000.0;
3349
3350 // (3) If marking stats are enabled, then we update the step history.
3351 #if _MARKING_STATS_
3352 if (_words_scanned >= _words_scanned_limit) {
3353 ++_clock_due_to_scanning;
3354 }
3355 if (_refs_reached >= _refs_reached_limit) {
3356 ++_clock_due_to_marking;
3357 }
3358
3359 double last_interval_ms = curr_time_ms - _interval_start_time_ms;
3360 _interval_start_time_ms = curr_time_ms;
3361 _all_clock_intervals_ms.add(last_interval_ms);
3362
3363 if (_cm->verbose_medium()) {
3364 gclog_or_tty->print_cr("[%u] regular clock, interval = %1.2lfms, "
3365 "scanned = "SIZE_FORMAT"%s, refs reached = "SIZE_FORMAT"%s",
3366 _worker_id, last_interval_ms,
3367 _words_scanned,
3368 (_words_scanned >= _words_scanned_limit) ? " (*)" : "",
3369 _refs_reached,
3370 (_refs_reached >= _refs_reached_limit) ? " (*)" : "");
3371 }
3372 #endif // _MARKING_STATS_
3373
3374 // (4) We check whether we should yield. If we have to, then we abort.
3375 if (SuspendibleThreadSet::should_yield()) {
3376 // We should yield. To do this we abort the task. The caller is
3377 // responsible for yielding.
3378 set_has_aborted();
3379 statsOnly( ++_aborted_yield );
3380 return;
3381 }
3382
3383 // (5) We check whether we've reached our time quota. If we have,
3384 // then we abort.
3385 double elapsed_time_ms = curr_time_ms - _start_time_ms;
3386 if (elapsed_time_ms > _time_target_ms) {
3387 set_has_aborted();
3388 _has_timed_out = true;
3389 statsOnly( ++_aborted_timed_out );
3390 return;
3391 }
3392
3393 // (6) Finally, we check whether there are enough completed STAB
3394 // buffers available for processing. If there are, we abort.
3395 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3396 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
3397 if (_cm->verbose_low()) {
3398 gclog_or_tty->print_cr("[%u] aborting to deal with pending SATB buffers",
3399 _worker_id);
3400 }
3401 // we do need to process SATB buffers, we'll abort and restart
3402 // the marking task to do so
3403 set_has_aborted();
3404 statsOnly( ++_aborted_satb );
3405 return;
3406 }
3407 }
3408
3409 void CMTask::recalculate_limits() {
3410 _real_words_scanned_limit = _words_scanned + words_scanned_period;
3411 _words_scanned_limit = _real_words_scanned_limit;
3412
3413 _real_refs_reached_limit = _refs_reached + refs_reached_period;
3414 _refs_reached_limit = _real_refs_reached_limit;
3415 }
3416
3417 void CMTask::decrease_limits() {
3418 // This is called when we believe that we're going to do an infrequent
3419 // operation which will increase the per byte scanned cost (i.e. move
3420 // entries to/from the global stack). It basically tries to decrease the
3421 // scanning limit so that the clock is called earlier.
3422
3423 if (_cm->verbose_medium()) {
3424 gclog_or_tty->print_cr("[%u] decreasing limits", _worker_id);
3425 }
3426
3427 _words_scanned_limit = _real_words_scanned_limit -
3428 3 * words_scanned_period / 4;
3429 _refs_reached_limit = _real_refs_reached_limit -
3430 3 * refs_reached_period / 4;
3431 }
3432
3433 void CMTask::move_entries_to_global_stack() {
3434 // local array where we'll store the entries that will be popped
3435 // from the local queue
3436 oop buffer[global_stack_transfer_size];
3437
3438 int n = 0;
3439 oop obj;
3440 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) {
3441 buffer[n] = obj;
3442 ++n;
3443 }
3444
3445 if (n > 0) {
3446 // we popped at least one entry from the local queue
3447
3448 statsOnly( ++_global_transfers_to; _local_pops += n );
3449
3450 if (!_cm->mark_stack_push(buffer, n)) {
3451 if (_cm->verbose_low()) {
3452 gclog_or_tty->print_cr("[%u] aborting due to global stack overflow",
3453 _worker_id);
3454 }
3455 set_has_aborted();
3456 } else {
3457 // the transfer was successful
3458
3459 if (_cm->verbose_medium()) {
3460 gclog_or_tty->print_cr("[%u] pushed %d entries to the global stack",
3461 _worker_id, n);
3462 }
3463 statsOnly( size_t tmp_size = _cm->mark_stack_size();
3464 if (tmp_size > _global_max_size) {
3465 _global_max_size = tmp_size;
3466 }
3467 _global_pushes += n );
3468 }
3469 }
3470
3471 // this operation was quite expensive, so decrease the limits
3472 decrease_limits();
3473 }
3474
3475 void CMTask::get_entries_from_global_stack() {
3476 // local array where we'll store the entries that will be popped
3477 // from the global stack.
3478 oop buffer[global_stack_transfer_size];
3479 int n;
3480 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n);
3481 assert(n <= global_stack_transfer_size,
3482 "we should not pop more than the given limit");
3483 if (n > 0) {
3484 // yes, we did actually pop at least one entry
3485
3486 statsOnly( ++_global_transfers_from; _global_pops += n );
3487 if (_cm->verbose_medium()) {
3488 gclog_or_tty->print_cr("[%u] popped %d entries from the global stack",
3489 _worker_id, n);
3490 }
3491 for (int i = 0; i < n; ++i) {
3492 bool success = _task_queue->push(buffer[i]);
3493 // We only call this when the local queue is empty or under a
3494 // given target limit. So, we do not expect this push to fail.
3495 assert(success, "invariant");
3496 }
3497
3498 statsOnly( size_t tmp_size = (size_t)_task_queue->size();
3499 if (tmp_size > _local_max_size) {
3500 _local_max_size = tmp_size;
3501 }
3502 _local_pushes += n );
3503 }
3504
3505 // this operation was quite expensive, so decrease the limits
3506 decrease_limits();
3507 }
3508
3509 void CMTask::drain_local_queue(bool partially) {
3510 if (has_aborted()) return;
3511
3512 // Decide what the target size is, depending whether we're going to
3513 // drain it partially (so that other tasks can steal if they run out
3514 // of things to do) or totally (at the very end).
3515 size_t target_size;
3516 if (partially) {
3517 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
3518 } else {
3519 target_size = 0;
3520 }
3521
3522 if (_task_queue->size() > target_size) {
3523 if (_cm->verbose_high()) {
3524 gclog_or_tty->print_cr("[%u] draining local queue, target size = " SIZE_FORMAT,
3525 _worker_id, target_size);
3526 }
3527
3528 oop obj;
3529 bool ret = _task_queue->pop_local(obj);
3530 while (ret) {
3531 statsOnly( ++_local_pops );
3532
3533 if (_cm->verbose_high()) {
3534 gclog_or_tty->print_cr("[%u] popped "PTR_FORMAT, _worker_id,
3535 p2i((void*) obj));
3536 }
3537
3538 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" );
3539 assert(!_g1h->is_on_master_free_list(
3540 _g1h->heap_region_containing((HeapWord*) obj)), "invariant");
3541
3542 scan_object(obj);
3543
3544 if (_task_queue->size() <= target_size || has_aborted()) {
3545 ret = false;
3546 } else {
3547 ret = _task_queue->pop_local(obj);
3548 }
3549 }
3550
3551 if (_cm->verbose_high()) {
3552 gclog_or_tty->print_cr("[%u] drained local queue, size = %u",
3553 _worker_id, _task_queue->size());
3554 }
3555 }
3556 }
3557
3558 void CMTask::drain_global_stack(bool partially) {
3559 if (has_aborted()) return;
3560
3561 // We have a policy to drain the local queue before we attempt to
3562 // drain the global stack.
3563 assert(partially || _task_queue->size() == 0, "invariant");
3564
3565 // Decide what the target size is, depending whether we're going to
3566 // drain it partially (so that other tasks can steal if they run out
3567 // of things to do) or totally (at the very end). Notice that,
3568 // because we move entries from the global stack in chunks or
3569 // because another task might be doing the same, we might in fact
3570 // drop below the target. But, this is not a problem.
3571 size_t target_size;
3572 if (partially) {
3573 target_size = _cm->partial_mark_stack_size_target();
3574 } else {
3575 target_size = 0;
3576 }
3577
3578 if (_cm->mark_stack_size() > target_size) {
3579 if (_cm->verbose_low()) {
3580 gclog_or_tty->print_cr("[%u] draining global_stack, target size " SIZE_FORMAT,
3581 _worker_id, target_size);
3582 }
3583
3584 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
3585 get_entries_from_global_stack();
3586 drain_local_queue(partially);
3587 }
3588
3589 if (_cm->verbose_low()) {
3590 gclog_or_tty->print_cr("[%u] drained global stack, size = " SIZE_FORMAT,
3591 _worker_id, _cm->mark_stack_size());
3592 }
3593 }
3594 }
3595
3596 // SATB Queue has several assumptions on whether to call the par or
3597 // non-par versions of the methods. this is why some of the code is
3598 // replicated. We should really get rid of the single-threaded version
3599 // of the code to simplify things.
3600 void CMTask::drain_satb_buffers() {
3601 if (has_aborted()) return;
3602
3603 // We set this so that the regular clock knows that we're in the
3604 // middle of draining buffers and doesn't set the abort flag when it
3605 // notices that SATB buffers are available for draining. It'd be
3606 // very counter productive if it did that. :-)
3607 _draining_satb_buffers = true;
3608
3609 CMSATBBufferClosure satb_cl(this, _g1h);
3610 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3611
3612 // This keeps claiming and applying the closure to completed buffers
3613 // until we run out of buffers or we need to abort.
3614 while (!has_aborted() &&
3615 satb_mq_set.apply_closure_to_completed_buffer(&satb_cl)) {
3616 if (_cm->verbose_medium()) {
3617 gclog_or_tty->print_cr("[%u] processed an SATB buffer", _worker_id);
3618 }
3619 statsOnly( ++_satb_buffers_processed );
3620 regular_clock_call();
3621 }
3622
3623 _draining_satb_buffers = false;
3624
3625 assert(has_aborted() ||
3626 concurrent() ||
3627 satb_mq_set.completed_buffers_num() == 0, "invariant");
3628
3629 // again, this was a potentially expensive operation, decrease the
3630 // limits to get the regular clock call early
3631 decrease_limits();
3632 }
3633
3634 void CMTask::print_stats() {
3635 gclog_or_tty->print_cr("Marking Stats, task = %u, calls = %d",
3636 _worker_id, _calls);
3637 gclog_or_tty->print_cr(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
3638 _elapsed_time_ms, _termination_time_ms);
3639 gclog_or_tty->print_cr(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3640 _step_times_ms.num(), _step_times_ms.avg(),
3641 _step_times_ms.sd());
3642 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
3643 _step_times_ms.maximum(), _step_times_ms.sum());
3644
3645 #if _MARKING_STATS_
3646 gclog_or_tty->print_cr(" Clock Intervals (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3647 _all_clock_intervals_ms.num(), _all_clock_intervals_ms.avg(),
3648 _all_clock_intervals_ms.sd());
3649 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
3650 _all_clock_intervals_ms.maximum(),
3651 _all_clock_intervals_ms.sum());
3652 gclog_or_tty->print_cr(" Clock Causes (cum): scanning = " SIZE_FORMAT ", marking = " SIZE_FORMAT,
3653 _clock_due_to_scanning, _clock_due_to_marking);
3654 gclog_or_tty->print_cr(" Objects: scanned = " SIZE_FORMAT ", found on the bitmap = " SIZE_FORMAT,
3655 _objs_scanned, _objs_found_on_bitmap);
3656 gclog_or_tty->print_cr(" Local Queue: pushes = " SIZE_FORMAT ", pops = " SIZE_FORMAT ", max size = " SIZE_FORMAT,
3657 _local_pushes, _local_pops, _local_max_size);
3658 gclog_or_tty->print_cr(" Global Stack: pushes = " SIZE_FORMAT ", pops = " SIZE_FORMAT ", max size = " SIZE_FORMAT,
3659 _global_pushes, _global_pops, _global_max_size);
3660 gclog_or_tty->print_cr(" transfers to = " SIZE_FORMAT ", transfers from = " SIZE_FORMAT,
3661 _global_transfers_to,_global_transfers_from);
3662 gclog_or_tty->print_cr(" Regions: claimed = " SIZE_FORMAT, _regions_claimed);
3663 gclog_or_tty->print_cr(" SATB buffers: processed = " SIZE_FORMAT, _satb_buffers_processed);
3664 gclog_or_tty->print_cr(" Steals: attempts = " SIZE_FORMAT ", successes = " SIZE_FORMAT,
3665 _steal_attempts, _steals);
3666 gclog_or_tty->print_cr(" Aborted: " SIZE_FORMAT ", due to", _aborted);
3667 gclog_or_tty->print_cr(" overflow: " SIZE_FORMAT ", global abort: " SIZE_FORMAT ", yield: " SIZE_FORMAT,
3668 _aborted_overflow, _aborted_cm_aborted, _aborted_yield);
3669 gclog_or_tty->print_cr(" time out: " SIZE_FORMAT ", SATB: " SIZE_FORMAT ", termination: " SIZE_FORMAT,
3670 _aborted_timed_out, _aborted_satb, _aborted_termination);
3671 #endif // _MARKING_STATS_
3672 }
3673
3674 bool ConcurrentMark::try_stealing(uint worker_id, int* hash_seed, oop& obj) {
3675 return _task_queues->steal(worker_id, hash_seed, obj);
3676 }
3677
3678 /*****************************************************************************
3679
3680 The do_marking_step(time_target_ms, ...) method is the building
3681 block of the parallel marking framework. It can be called in parallel
3682 with other invocations of do_marking_step() on different tasks
3683 (but only one per task, obviously) and concurrently with the
3684 mutator threads, or during remark, hence it eliminates the need
3685 for two versions of the code. When called during remark, it will
3686 pick up from where the task left off during the concurrent marking
3687 phase. Interestingly, tasks are also claimable during evacuation
3688 pauses too, since do_marking_step() ensures that it aborts before
3689 it needs to yield.
3690
3691 The data structures that it uses to do marking work are the
3692 following:
3693
3694 (1) Marking Bitmap. If there are gray objects that appear only
3695 on the bitmap (this happens either when dealing with an overflow
3696 or when the initial marking phase has simply marked the roots
3697 and didn't push them on the stack), then tasks claim heap
3698 regions whose bitmap they then scan to find gray objects. A
3699 global finger indicates where the end of the last claimed region
3700 is. A local finger indicates how far into the region a task has
3701 scanned. The two fingers are used to determine how to gray an
3702 object (i.e. whether simply marking it is OK, as it will be
3703 visited by a task in the future, or whether it needs to be also
3704 pushed on a stack).
3705
3706 (2) Local Queue. The local queue of the task which is accessed
3707 reasonably efficiently by the task. Other tasks can steal from
3708 it when they run out of work. Throughout the marking phase, a
3709 task attempts to keep its local queue short but not totally
3710 empty, so that entries are available for stealing by other
3711 tasks. Only when there is no more work, a task will totally
3712 drain its local queue.
3713
3714 (3) Global Mark Stack. This handles local queue overflow. During
3715 marking only sets of entries are moved between it and the local
3716 queues, as access to it requires a mutex and more fine-grain
3717 interaction with it which might cause contention. If it
3718 overflows, then the marking phase should restart and iterate
3719 over the bitmap to identify gray objects. Throughout the marking
3720 phase, tasks attempt to keep the global mark stack at a small
3721 length but not totally empty, so that entries are available for
3722 popping by other tasks. Only when there is no more work, tasks
3723 will totally drain the global mark stack.
3724
3725 (4) SATB Buffer Queue. This is where completed SATB buffers are
3726 made available. Buffers are regularly removed from this queue
3727 and scanned for roots, so that the queue doesn't get too
3728 long. During remark, all completed buffers are processed, as
3729 well as the filled in parts of any uncompleted buffers.
3730
3731 The do_marking_step() method tries to abort when the time target
3732 has been reached. There are a few other cases when the
3733 do_marking_step() method also aborts:
3734
3735 (1) When the marking phase has been aborted (after a Full GC).
3736
3737 (2) When a global overflow (on the global stack) has been
3738 triggered. Before the task aborts, it will actually sync up with
3739 the other tasks to ensure that all the marking data structures
3740 (local queues, stacks, fingers etc.) are re-initialized so that
3741 when do_marking_step() completes, the marking phase can
3742 immediately restart.
3743
3744 (3) When enough completed SATB buffers are available. The
3745 do_marking_step() method only tries to drain SATB buffers right
3746 at the beginning. So, if enough buffers are available, the
3747 marking step aborts and the SATB buffers are processed at
3748 the beginning of the next invocation.
3749
3750 (4) To yield. when we have to yield then we abort and yield
3751 right at the end of do_marking_step(). This saves us from a lot
3752 of hassle as, by yielding we might allow a Full GC. If this
3753 happens then objects will be compacted underneath our feet, the
3754 heap might shrink, etc. We save checking for this by just
3755 aborting and doing the yield right at the end.
3756
3757 From the above it follows that the do_marking_step() method should
3758 be called in a loop (or, otherwise, regularly) until it completes.
3759
3760 If a marking step completes without its has_aborted() flag being
3761 true, it means it has completed the current marking phase (and
3762 also all other marking tasks have done so and have all synced up).
3763
3764 A method called regular_clock_call() is invoked "regularly" (in
3765 sub ms intervals) throughout marking. It is this clock method that
3766 checks all the abort conditions which were mentioned above and
3767 decides when the task should abort. A work-based scheme is used to
3768 trigger this clock method: when the number of object words the
3769 marking phase has scanned or the number of references the marking
3770 phase has visited reach a given limit. Additional invocations to
3771 the method clock have been planted in a few other strategic places
3772 too. The initial reason for the clock method was to avoid calling
3773 vtime too regularly, as it is quite expensive. So, once it was in
3774 place, it was natural to piggy-back all the other conditions on it
3775 too and not constantly check them throughout the code.
3776
3777 If do_termination is true then do_marking_step will enter its
3778 termination protocol.
3779
3780 The value of is_serial must be true when do_marking_step is being
3781 called serially (i.e. by the VMThread) and do_marking_step should
3782 skip any synchronization in the termination and overflow code.
3783 Examples include the serial remark code and the serial reference
3784 processing closures.
3785
3786 The value of is_serial must be false when do_marking_step is
3787 being called by any of the worker threads in a work gang.
3788 Examples include the concurrent marking code (CMMarkingTask),
3789 the MT remark code, and the MT reference processing closures.
3790
3791 *****************************************************************************/
3792
3793 void CMTask::do_marking_step(double time_target_ms,
3794 bool do_termination,
3795 bool is_serial) {
3796 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
3797 assert(concurrent() == _cm->concurrent(), "they should be the same");
3798
3799 G1CollectorPolicy* g1_policy = _g1h->g1_policy();
3800 assert(_task_queues != NULL, "invariant");
3801 assert(_task_queue != NULL, "invariant");
3802 assert(_task_queues->queue(_worker_id) == _task_queue, "invariant");
3803
3804 assert(!_claimed,
3805 "only one thread should claim this task at any one time");
3806
3807 // OK, this doesn't safeguard again all possible scenarios, as it is
3808 // possible for two threads to set the _claimed flag at the same
3809 // time. But it is only for debugging purposes anyway and it will
3810 // catch most problems.
3811 _claimed = true;
3812
3813 _start_time_ms = os::elapsedVTime() * 1000.0;
3814 statsOnly( _interval_start_time_ms = _start_time_ms );
3815
3816 // If do_stealing is true then do_marking_step will attempt to
3817 // steal work from the other CMTasks. It only makes sense to
3818 // enable stealing when the termination protocol is enabled
3819 // and do_marking_step() is not being called serially.
3820 bool do_stealing = do_termination && !is_serial;
3821
3822 double diff_prediction_ms =
3823 g1_policy->get_new_prediction(&_marking_step_diffs_ms);
3824 _time_target_ms = time_target_ms - diff_prediction_ms;
3825
3826 // set up the variables that are used in the work-based scheme to
3827 // call the regular clock method
3828 _words_scanned = 0;
3829 _refs_reached = 0;
3830 recalculate_limits();
3831
3832 // clear all flags
3833 clear_has_aborted();
3834 _has_timed_out = false;
3835 _draining_satb_buffers = false;
3836
3837 ++_calls;
3838
3839 if (_cm->verbose_low()) {
3840 gclog_or_tty->print_cr("[%u] >>>>>>>>>> START, call = %d, "
3841 "target = %1.2lfms >>>>>>>>>>",
3842 _worker_id, _calls, _time_target_ms);
3843 }
3844
3845 // Set up the bitmap and oop closures. Anything that uses them is
3846 // eventually called from this method, so it is OK to allocate these
3847 // statically.
3848 CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap);
3849 G1CMOopClosure cm_oop_closure(_g1h, _cm, this);
3850 set_cm_oop_closure(&cm_oop_closure);
3851
3852 if (_cm->has_overflown()) {
3853 // This can happen if the mark stack overflows during a GC pause
3854 // and this task, after a yield point, restarts. We have to abort
3855 // as we need to get into the overflow protocol which happens
3856 // right at the end of this task.
3857 set_has_aborted();
3858 }
3859
3860 // First drain any available SATB buffers. After this, we will not
3861 // look at SATB buffers before the next invocation of this method.
3862 // If enough completed SATB buffers are queued up, the regular clock
3863 // will abort this task so that it restarts.
3864 drain_satb_buffers();
3865 // ...then partially drain the local queue and the global stack
3866 drain_local_queue(true);
3867 drain_global_stack(true);
3868
3869 do {
3870 if (!has_aborted() && _curr_region != NULL) {
3871 // This means that we're already holding on to a region.
3872 assert(_finger != NULL, "if region is not NULL, then the finger "
3873 "should not be NULL either");
3874
3875 // We might have restarted this task after an evacuation pause
3876 // which might have evacuated the region we're holding on to
3877 // underneath our feet. Let's read its limit again to make sure
3878 // that we do not iterate over a region of the heap that
3879 // contains garbage (update_region_limit() will also move
3880 // _finger to the start of the region if it is found empty).
3881 update_region_limit();
3882 // We will start from _finger not from the start of the region,
3883 // as we might be restarting this task after aborting half-way
3884 // through scanning this region. In this case, _finger points to
3885 // the address where we last found a marked object. If this is a
3886 // fresh region, _finger points to start().
3887 MemRegion mr = MemRegion(_finger, _region_limit);
3888
3889 if (_cm->verbose_low()) {
3890 gclog_or_tty->print_cr("[%u] we're scanning part "
3891 "["PTR_FORMAT", "PTR_FORMAT") "
3892 "of region "HR_FORMAT,
3893 _worker_id, p2i(_finger), p2i(_region_limit),
3894 HR_FORMAT_PARAMS(_curr_region));
3895 }
3896
3897 assert(!_curr_region->is_humongous() || mr.start() == _curr_region->bottom(),
3898 "humongous regions should go around loop once only");
3899
3900 // Some special cases:
3901 // If the memory region is empty, we can just give up the region.
3902 // If the current region is humongous then we only need to check
3903 // the bitmap for the bit associated with the start of the object,
3904 // scan the object if it's live, and give up the region.
3905 // Otherwise, let's iterate over the bitmap of the part of the region
3906 // that is left.
3907 // If the iteration is successful, give up the region.
3908 if (mr.is_empty()) {
3909 giveup_current_region();
3910 regular_clock_call();
3911 } else if (_curr_region->is_humongous() && mr.start() == _curr_region->bottom()) {
3912 if (_nextMarkBitMap->isMarked(mr.start())) {
3913 // The object is marked - apply the closure
3914 BitMap::idx_t offset = _nextMarkBitMap->heapWordToOffset(mr.start());
3915 bitmap_closure.do_bit(offset);
3916 }
3917 // Even if this task aborted while scanning the humongous object
3918 // we can (and should) give up the current region.
3919 giveup_current_region();
3920 regular_clock_call();
3921 } else if (_nextMarkBitMap->iterate(&bitmap_closure, mr)) {
3922 giveup_current_region();
3923 regular_clock_call();
3924 } else {
3925 assert(has_aborted(), "currently the only way to do so");
3926 // The only way to abort the bitmap iteration is to return
3927 // false from the do_bit() method. However, inside the
3928 // do_bit() method we move the _finger to point to the
3929 // object currently being looked at. So, if we bail out, we
3930 // have definitely set _finger to something non-null.
3931 assert(_finger != NULL, "invariant");
3932
3933 // Region iteration was actually aborted. So now _finger
3934 // points to the address of the object we last scanned. If we
3935 // leave it there, when we restart this task, we will rescan
3936 // the object. It is easy to avoid this. We move the finger by
3937 // enough to point to the next possible object header (the
3938 // bitmap knows by how much we need to move it as it knows its
3939 // granularity).
3940 assert(_finger < _region_limit, "invariant");
3941 HeapWord* new_finger = _nextMarkBitMap->nextObject(_finger);
3942 // Check if bitmap iteration was aborted while scanning the last object
3943 if (new_finger >= _region_limit) {
3944 giveup_current_region();
3945 } else {
3946 move_finger_to(new_finger);
3947 }
3948 }
3949 }
3950 // At this point we have either completed iterating over the
3951 // region we were holding on to, or we have aborted.
3952
3953 // We then partially drain the local queue and the global stack.
3954 // (Do we really need this?)
3955 drain_local_queue(true);
3956 drain_global_stack(true);
3957
3958 // Read the note on the claim_region() method on why it might
3959 // return NULL with potentially more regions available for
3960 // claiming and why we have to check out_of_regions() to determine
3961 // whether we're done or not.
3962 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
3963 // We are going to try to claim a new region. We should have
3964 // given up on the previous one.
3965 // Separated the asserts so that we know which one fires.
3966 assert(_curr_region == NULL, "invariant");
3967 assert(_finger == NULL, "invariant");
3968 assert(_region_limit == NULL, "invariant");
3969 if (_cm->verbose_low()) {
3970 gclog_or_tty->print_cr("[%u] trying to claim a new region", _worker_id);
3971 }
3972 HeapRegion* claimed_region = _cm->claim_region(_worker_id);
3973 if (claimed_region != NULL) {
3974 // Yes, we managed to claim one
3975 statsOnly( ++_regions_claimed );
3976
3977 if (_cm->verbose_low()) {
3978 gclog_or_tty->print_cr("[%u] we successfully claimed "
3979 "region "PTR_FORMAT,
3980 _worker_id, p2i(claimed_region));
3981 }
3982
3983 setup_for_region(claimed_region);
3984 assert(_curr_region == claimed_region, "invariant");
3985 }
3986 // It is important to call the regular clock here. It might take
3987 // a while to claim a region if, for example, we hit a large
3988 // block of empty regions. So we need to call the regular clock
3989 // method once round the loop to make sure it's called
3990 // frequently enough.
3991 regular_clock_call();
3992 }
3993
3994 if (!has_aborted() && _curr_region == NULL) {
3995 assert(_cm->out_of_regions(),
3996 "at this point we should be out of regions");
3997 }
3998 } while ( _curr_region != NULL && !has_aborted());
3999
4000 if (!has_aborted()) {
4001 // We cannot check whether the global stack is empty, since other
4002 // tasks might be pushing objects to it concurrently.
4003 assert(_cm->out_of_regions(),
4004 "at this point we should be out of regions");
4005
4006 if (_cm->verbose_low()) {
4007 gclog_or_tty->print_cr("[%u] all regions claimed", _worker_id);
4008 }
4009
4010 // Try to reduce the number of available SATB buffers so that
4011 // remark has less work to do.
4012 drain_satb_buffers();
4013 }
4014
4015 // Since we've done everything else, we can now totally drain the
4016 // local queue and global stack.
4017 drain_local_queue(false);
4018 drain_global_stack(false);
4019
4020 // Attempt at work stealing from other task's queues.
4021 if (do_stealing && !has_aborted()) {
4022 // We have not aborted. This means that we have finished all that
4023 // we could. Let's try to do some stealing...
4024
4025 // We cannot check whether the global stack is empty, since other
4026 // tasks might be pushing objects to it concurrently.
4027 assert(_cm->out_of_regions() && _task_queue->size() == 0,
4028 "only way to reach here");
4029
4030 if (_cm->verbose_low()) {
4031 gclog_or_tty->print_cr("[%u] starting to steal", _worker_id);
4032 }
4033
4034 while (!has_aborted()) {
4035 oop obj;
4036 statsOnly( ++_steal_attempts );
4037
4038 if (_cm->try_stealing(_worker_id, &_hash_seed, obj)) {
4039 if (_cm->verbose_medium()) {
4040 gclog_or_tty->print_cr("[%u] stolen "PTR_FORMAT" successfully",
4041 _worker_id, p2i((void*) obj));
4042 }
4043
4044 statsOnly( ++_steals );
4045
4046 assert(_nextMarkBitMap->isMarked((HeapWord*) obj),
4047 "any stolen object should be marked");
4048 scan_object(obj);
4049
4050 // And since we're towards the end, let's totally drain the
4051 // local queue and global stack.
4052 drain_local_queue(false);
4053 drain_global_stack(false);
4054 } else {
4055 break;
4056 }
4057 }
4058 }
4059
4060 // If we are about to wrap up and go into termination, check if we
4061 // should raise the overflow flag.
4062 if (do_termination && !has_aborted()) {
4063 if (_cm->force_overflow()->should_force()) {
4064 _cm->set_has_overflown();
4065 regular_clock_call();
4066 }
4067 }
4068
4069 // We still haven't aborted. Now, let's try to get into the
4070 // termination protocol.
4071 if (do_termination && !has_aborted()) {
4072 // We cannot check whether the global stack is empty, since other
4073 // tasks might be concurrently pushing objects on it.
4074 // Separated the asserts so that we know which one fires.
4075 assert(_cm->out_of_regions(), "only way to reach here");
4076 assert(_task_queue->size() == 0, "only way to reach here");
4077
4078 if (_cm->verbose_low()) {
4079 gclog_or_tty->print_cr("[%u] starting termination protocol", _worker_id);
4080 }
4081
4082 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
4083
4084 // The CMTask class also extends the TerminatorTerminator class,
4085 // hence its should_exit_termination() method will also decide
4086 // whether to exit the termination protocol or not.
4087 bool finished = (is_serial ||
4088 _cm->terminator()->offer_termination(this));
4089 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
4090 _termination_time_ms +=
4091 termination_end_time_ms - _termination_start_time_ms;
4092
4093 if (finished) {
4094 // We're all done.
4095
4096 if (_worker_id == 0) {
4097 // let's allow task 0 to do this
4098 if (concurrent()) {
4099 assert(_cm->concurrent_marking_in_progress(), "invariant");
4100 // we need to set this to false before the next
4101 // safepoint. This way we ensure that the marking phase
4102 // doesn't observe any more heap expansions.
4103 _cm->clear_concurrent_marking_in_progress();
4104 }
4105 }
4106
4107 // We can now guarantee that the global stack is empty, since
4108 // all other tasks have finished. We separated the guarantees so
4109 // that, if a condition is false, we can immediately find out
4110 // which one.
4111 guarantee(_cm->out_of_regions(), "only way to reach here");
4112 guarantee(_cm->mark_stack_empty(), "only way to reach here");
4113 guarantee(_task_queue->size() == 0, "only way to reach here");
4114 guarantee(!_cm->has_overflown(), "only way to reach here");
4115 guarantee(!_cm->mark_stack_overflow(), "only way to reach here");
4116
4117 if (_cm->verbose_low()) {
4118 gclog_or_tty->print_cr("[%u] all tasks terminated", _worker_id);
4119 }
4120 } else {
4121 // Apparently there's more work to do. Let's abort this task. It
4122 // will restart it and we can hopefully find more things to do.
4123
4124 if (_cm->verbose_low()) {
4125 gclog_or_tty->print_cr("[%u] apparently there is more work to do",
4126 _worker_id);
4127 }
4128
4129 set_has_aborted();
4130 statsOnly( ++_aborted_termination );
4131 }
4132 }
4133
4134 // Mainly for debugging purposes to make sure that a pointer to the
4135 // closure which was statically allocated in this frame doesn't
4136 // escape it by accident.
4137 set_cm_oop_closure(NULL);
4138 double end_time_ms = os::elapsedVTime() * 1000.0;
4139 double elapsed_time_ms = end_time_ms - _start_time_ms;
4140 // Update the step history.
4141 _step_times_ms.add(elapsed_time_ms);
4142
4143 if (has_aborted()) {
4144 // The task was aborted for some reason.
4145
4146 statsOnly( ++_aborted );
4147
4148 if (_has_timed_out) {
4149 double diff_ms = elapsed_time_ms - _time_target_ms;
4150 // Keep statistics of how well we did with respect to hitting
4151 // our target only if we actually timed out (if we aborted for
4152 // other reasons, then the results might get skewed).
4153 _marking_step_diffs_ms.add(diff_ms);
4154 }
4155
4156 if (_cm->has_overflown()) {
4157 // This is the interesting one. We aborted because a global
4158 // overflow was raised. This means we have to restart the
4159 // marking phase and start iterating over regions. However, in
4160 // order to do this we have to make sure that all tasks stop
4161 // what they are doing and re-initialize in a safe manner. We
4162 // will achieve this with the use of two barrier sync points.
4163
4164 if (_cm->verbose_low()) {
4165 gclog_or_tty->print_cr("[%u] detected overflow", _worker_id);
4166 }
4167
4168 if (!is_serial) {
4169 // We only need to enter the sync barrier if being called
4170 // from a parallel context
4171 _cm->enter_first_sync_barrier(_worker_id);
4172
4173 // When we exit this sync barrier we know that all tasks have
4174 // stopped doing marking work. So, it's now safe to
4175 // re-initialize our data structures. At the end of this method,
4176 // task 0 will clear the global data structures.
4177 }
4178
4179 statsOnly( ++_aborted_overflow );
4180
4181 // We clear the local state of this task...
4182 clear_region_fields();
4183
4184 if (!is_serial) {
4185 // ...and enter the second barrier.
4186 _cm->enter_second_sync_barrier(_worker_id);
4187 }
4188 // At this point, if we're during the concurrent phase of
4189 // marking, everything has been re-initialized and we're
4190 // ready to restart.
4191 }
4192
4193 if (_cm->verbose_low()) {
4194 gclog_or_tty->print_cr("[%u] <<<<<<<<<< ABORTING, target = %1.2lfms, "
4195 "elapsed = %1.2lfms <<<<<<<<<<",
4196 _worker_id, _time_target_ms, elapsed_time_ms);
4197 if (_cm->has_aborted()) {
4198 gclog_or_tty->print_cr("[%u] ========== MARKING ABORTED ==========",
4199 _worker_id);
4200 }
4201 }
4202 } else {
4203 if (_cm->verbose_low()) {
4204 gclog_or_tty->print_cr("[%u] <<<<<<<<<< FINISHED, target = %1.2lfms, "
4205 "elapsed = %1.2lfms <<<<<<<<<<",
4206 _worker_id, _time_target_ms, elapsed_time_ms);
4207 }
4208 }
4209
4210 _claimed = false;
4211 }
4212
4213 CMTask::CMTask(uint worker_id,
4214 ConcurrentMark* cm,
4215 size_t* marked_bytes,
4216 BitMap* card_bm,
4217 CMTaskQueue* task_queue,
4218 CMTaskQueueSet* task_queues)
4219 : _g1h(G1CollectedHeap::heap()),
4220 _worker_id(worker_id), _cm(cm),
4221 _claimed(false),
4222 _nextMarkBitMap(NULL), _hash_seed(17),
4223 _task_queue(task_queue),
4224 _task_queues(task_queues),
4225 _cm_oop_closure(NULL),
4226 _marked_bytes_array(marked_bytes),
4227 _card_bm(card_bm) {
4228 guarantee(task_queue != NULL, "invariant");
4229 guarantee(task_queues != NULL, "invariant");
4230
4231 statsOnly( _clock_due_to_scanning = 0;
4232 _clock_due_to_marking = 0 );
4233
4234 _marking_step_diffs_ms.add(0.5);
4235 }
4236
4237 // These are formatting macros that are used below to ensure
4238 // consistent formatting. The *_H_* versions are used to format the
4239 // header for a particular value and they should be kept consistent
4240 // with the corresponding macro. Also note that most of the macros add
4241 // the necessary white space (as a prefix) which makes them a bit
4242 // easier to compose.
4243
4244 // All the output lines are prefixed with this string to be able to
4245 // identify them easily in a large log file.
4246 #define G1PPRL_LINE_PREFIX "###"
4247
4248 #define G1PPRL_ADDR_BASE_FORMAT " "PTR_FORMAT"-"PTR_FORMAT
4249 #ifdef _LP64
4250 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
4251 #else // _LP64
4252 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
4253 #endif // _LP64
4254
4255 // For per-region info
4256 #define G1PPRL_TYPE_FORMAT " %-4s"
4257 #define G1PPRL_TYPE_H_FORMAT " %4s"
4258 #define G1PPRL_BYTE_FORMAT " "SIZE_FORMAT_W(9)
4259 #define G1PPRL_BYTE_H_FORMAT " %9s"
4260 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
4261 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
4262
4263 // For summary info
4264 #define G1PPRL_SUM_ADDR_FORMAT(tag) " "tag":"G1PPRL_ADDR_BASE_FORMAT
4265 #define G1PPRL_SUM_BYTE_FORMAT(tag) " "tag": "SIZE_FORMAT
4266 #define G1PPRL_SUM_MB_FORMAT(tag) " "tag": %1.2f MB"
4267 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag)" / %1.2f %%"
4268
4269 G1PrintRegionLivenessInfoClosure::
4270 G1PrintRegionLivenessInfoClosure(outputStream* out, const char* phase_name)
4271 : _out(out),
4272 _total_used_bytes(0), _total_capacity_bytes(0),
4273 _total_prev_live_bytes(0), _total_next_live_bytes(0),
4274 _hum_used_bytes(0), _hum_capacity_bytes(0),
4275 _hum_prev_live_bytes(0), _hum_next_live_bytes(0),
4276 _total_remset_bytes(0), _total_strong_code_roots_bytes(0) {
4277 G1CollectedHeap* g1h = G1CollectedHeap::heap();
4278 MemRegion g1_reserved = g1h->g1_reserved();
4279 double now = os::elapsedTime();
4280
4281 // Print the header of the output.
4282 _out->cr();
4283 _out->print_cr(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
4284 _out->print_cr(G1PPRL_LINE_PREFIX" HEAP"
4285 G1PPRL_SUM_ADDR_FORMAT("reserved")
4286 G1PPRL_SUM_BYTE_FORMAT("region-size"),
4287 p2i(g1_reserved.start()), p2i(g1_reserved.end()),
4288 HeapRegion::GrainBytes);
4289 _out->print_cr(G1PPRL_LINE_PREFIX);
4290 _out->print_cr(G1PPRL_LINE_PREFIX
4291 G1PPRL_TYPE_H_FORMAT
4292 G1PPRL_ADDR_BASE_H_FORMAT
4293 G1PPRL_BYTE_H_FORMAT
4294 G1PPRL_BYTE_H_FORMAT
4295 G1PPRL_BYTE_H_FORMAT
4296 G1PPRL_DOUBLE_H_FORMAT
4297 G1PPRL_BYTE_H_FORMAT
4298 G1PPRL_BYTE_H_FORMAT,
4299 "type", "address-range",
4300 "used", "prev-live", "next-live", "gc-eff",
4301 "remset", "code-roots");
4302 _out->print_cr(G1PPRL_LINE_PREFIX
4303 G1PPRL_TYPE_H_FORMAT
4304 G1PPRL_ADDR_BASE_H_FORMAT
4305 G1PPRL_BYTE_H_FORMAT
4306 G1PPRL_BYTE_H_FORMAT
4307 G1PPRL_BYTE_H_FORMAT
4308 G1PPRL_DOUBLE_H_FORMAT
4309 G1PPRL_BYTE_H_FORMAT
4310 G1PPRL_BYTE_H_FORMAT,
4311 "", "",
4312 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)",
4313 "(bytes)", "(bytes)");
4314 }
4315
4316 // It takes as a parameter a reference to one of the _hum_* fields, it
4317 // deduces the corresponding value for a region in a humongous region
4318 // series (either the region size, or what's left if the _hum_* field
4319 // is < the region size), and updates the _hum_* field accordingly.
4320 size_t G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* hum_bytes) {
4321 size_t bytes = 0;
4322 // The > 0 check is to deal with the prev and next live bytes which
4323 // could be 0.
4324 if (*hum_bytes > 0) {
4325 bytes = MIN2(HeapRegion::GrainBytes, *hum_bytes);
4326 *hum_bytes -= bytes;
4327 }
4328 return bytes;
4329 }
4330
4331 // It deduces the values for a region in a humongous region series
4332 // from the _hum_* fields and updates those accordingly. It assumes
4333 // that that _hum_* fields have already been set up from the "starts
4334 // humongous" region and we visit the regions in address order.
4335 void G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* used_bytes,
4336 size_t* capacity_bytes,
4337 size_t* prev_live_bytes,
4338 size_t* next_live_bytes) {
4339 assert(_hum_used_bytes > 0 && _hum_capacity_bytes > 0, "pre-condition");
4340 *used_bytes = get_hum_bytes(&_hum_used_bytes);
4341 *capacity_bytes = get_hum_bytes(&_hum_capacity_bytes);
4342 *prev_live_bytes = get_hum_bytes(&_hum_prev_live_bytes);
4343 *next_live_bytes = get_hum_bytes(&_hum_next_live_bytes);
4344 }
4345
4346 bool G1PrintRegionLivenessInfoClosure::doHeapRegion(HeapRegion* r) {
4347 const char* type = r->get_type_str();
4348 HeapWord* bottom = r->bottom();
4349 HeapWord* end = r->end();
4350 size_t capacity_bytes = r->capacity();
4351 size_t used_bytes = r->used();
4352 size_t prev_live_bytes = r->live_bytes();
4353 size_t next_live_bytes = r->next_live_bytes();
4354 double gc_eff = r->gc_efficiency();
4355 size_t remset_bytes = r->rem_set()->mem_size();
4356 size_t strong_code_roots_bytes = r->rem_set()->strong_code_roots_mem_size();
4357
4358 if (r->is_starts_humongous()) {
4359 assert(_hum_used_bytes == 0 && _hum_capacity_bytes == 0 &&
4360 _hum_prev_live_bytes == 0 && _hum_next_live_bytes == 0,
4361 "they should have been zeroed after the last time we used them");
4362 // Set up the _hum_* fields.
4363 _hum_capacity_bytes = capacity_bytes;
4364 _hum_used_bytes = used_bytes;
4365 _hum_prev_live_bytes = prev_live_bytes;
4366 _hum_next_live_bytes = next_live_bytes;
4367 get_hum_bytes(&used_bytes, &capacity_bytes,
4368 &prev_live_bytes, &next_live_bytes);
4369 end = bottom + HeapRegion::GrainWords;
4370 } else if (r->is_continues_humongous()) {
4371 get_hum_bytes(&used_bytes, &capacity_bytes,
4372 &prev_live_bytes, &next_live_bytes);
4373 assert(end == bottom + HeapRegion::GrainWords, "invariant");
4374 }
4375
4376 _total_used_bytes += used_bytes;
4377 _total_capacity_bytes += capacity_bytes;
4378 _total_prev_live_bytes += prev_live_bytes;
4379 _total_next_live_bytes += next_live_bytes;
4380 _total_remset_bytes += remset_bytes;
4381 _total_strong_code_roots_bytes += strong_code_roots_bytes;
4382
4383 // Print a line for this particular region.
4384 _out->print_cr(G1PPRL_LINE_PREFIX
4385 G1PPRL_TYPE_FORMAT
4386 G1PPRL_ADDR_BASE_FORMAT
4387 G1PPRL_BYTE_FORMAT
4388 G1PPRL_BYTE_FORMAT
4389 G1PPRL_BYTE_FORMAT
4390 G1PPRL_DOUBLE_FORMAT
4391 G1PPRL_BYTE_FORMAT
4392 G1PPRL_BYTE_FORMAT,
4393 type, p2i(bottom), p2i(end),
4394 used_bytes, prev_live_bytes, next_live_bytes, gc_eff,
4395 remset_bytes, strong_code_roots_bytes);
4396
4397 return false;
4398 }
4399
4400 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
4401 // add static memory usages to remembered set sizes
4402 _total_remset_bytes += HeapRegionRemSet::fl_mem_size() + HeapRegionRemSet::static_mem_size();
4403 // Print the footer of the output.
4404 _out->print_cr(G1PPRL_LINE_PREFIX);
4405 _out->print_cr(G1PPRL_LINE_PREFIX
4406 " SUMMARY"
4407 G1PPRL_SUM_MB_FORMAT("capacity")
4408 G1PPRL_SUM_MB_PERC_FORMAT("used")
4409 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
4410 G1PPRL_SUM_MB_PERC_FORMAT("next-live")
4411 G1PPRL_SUM_MB_FORMAT("remset")
4412 G1PPRL_SUM_MB_FORMAT("code-roots"),
4413 bytes_to_mb(_total_capacity_bytes),
4414 bytes_to_mb(_total_used_bytes),
4415 perc(_total_used_bytes, _total_capacity_bytes),
4416 bytes_to_mb(_total_prev_live_bytes),
4417 perc(_total_prev_live_bytes, _total_capacity_bytes),
4418 bytes_to_mb(_total_next_live_bytes),
4419 perc(_total_next_live_bytes, _total_capacity_bytes),
4420 bytes_to_mb(_total_remset_bytes),
4421 bytes_to_mb(_total_strong_code_roots_bytes));
4422 _out->cr();
4423 }