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_implementation/g1/concurrentMark.inline.hpp"
30 #include "gc_implementation/g1/concurrentMarkThread.inline.hpp"
31 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp"
32 #include "gc_implementation/g1/g1CollectorPolicy.hpp"
33 #include "gc_implementation/g1/g1ErgoVerbose.hpp"
34 #include "gc_implementation/g1/g1Log.hpp"
35 #include "gc_implementation/g1/g1OopClosures.inline.hpp"
36 #include "gc_implementation/g1/g1RemSet.hpp"
37 #include "gc_implementation/g1/g1StringDedup.hpp"
38 #include "gc_implementation/g1/heapRegion.inline.hpp"
39 #include "gc_implementation/g1/heapRegionManager.inline.hpp"
40 #include "gc_implementation/g1/heapRegionRemSet.hpp"
41 #include "gc_implementation/g1/heapRegionSet.inline.hpp"
42 #include "gc_implementation/shared/vmGCOperations.hpp"
43 #include "gc_implementation/shared/gcTimer.hpp"
44 #include "gc_implementation/shared/gcTrace.hpp"
45 #include "gc_implementation/shared/gcTraceTime.hpp"
46 #include "memory/allocation.hpp"
47 #include "memory/genOopClosures.inline.hpp"
48 #include "memory/referencePolicy.hpp"
49 #include "memory/resourceArea.hpp"
50 #include "memory/strongRootsScope.hpp"
51 #include "oops/oop.inline.hpp"
52 #include "runtime/handles.inline.hpp"
53 #include "runtime/java.hpp"
54 #include "runtime/atomic.inline.hpp"
55 #include "runtime/prefetch.inline.hpp"
56 #include "services/memTracker.hpp"
57
58 // Concurrent marking bit map wrapper
59
60 CMBitMapRO::CMBitMapRO(int shifter) :
61 _bm(),
62 _shifter(shifter) {
63 _bmStartWord = 0;
64 _bmWordSize = 0;
65 }
66
67 HeapWord* CMBitMapRO::getNextMarkedWordAddress(const HeapWord* addr,
68 const HeapWord* limit) const {
69 // First we must round addr *up* to a possible object boundary.
70 addr = (HeapWord*)align_size_up((intptr_t)addr,
71 HeapWordSize << _shifter);
72 size_t addrOffset = heapWordToOffset(addr);
73 if (limit == NULL) {
74 limit = _bmStartWord + _bmWordSize;
75 }
76 size_t limitOffset = heapWordToOffset(limit);
77 size_t nextOffset = _bm.get_next_one_offset(addrOffset, limitOffset);
78 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
79 assert(nextAddr >= addr, "get_next_one postcondition");
80 assert(nextAddr == limit || isMarked(nextAddr),
81 "get_next_one postcondition");
82 return nextAddr;
83 }
84
85 HeapWord* CMBitMapRO::getNextUnmarkedWordAddress(const HeapWord* addr,
86 const HeapWord* limit) const {
87 size_t addrOffset = heapWordToOffset(addr);
88 if (limit == NULL) {
89 limit = _bmStartWord + _bmWordSize;
90 }
91 size_t limitOffset = heapWordToOffset(limit);
92 size_t nextOffset = _bm.get_next_zero_offset(addrOffset, limitOffset);
93 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
94 assert(nextAddr >= addr, "get_next_one postcondition");
95 assert(nextAddr == limit || !isMarked(nextAddr),
96 "get_next_one postcondition");
97 return nextAddr;
98 }
99
100 int CMBitMapRO::heapWordDiffToOffsetDiff(size_t diff) const {
101 assert((diff & ((1 << _shifter) - 1)) == 0, "argument check");
102 return (int) (diff >> _shifter);
103 }
104
105 #ifndef PRODUCT
106 bool CMBitMapRO::covers(MemRegion heap_rs) const {
107 // assert(_bm.map() == _virtual_space.low(), "map inconsistency");
108 assert(((size_t)_bm.size() * ((size_t)1 << _shifter)) == _bmWordSize,
109 "size inconsistency");
110 return _bmStartWord == (HeapWord*)(heap_rs.start()) &&
111 _bmWordSize == heap_rs.word_size();
112 }
113 #endif
114
115 void CMBitMapRO::print_on_error(outputStream* st, const char* prefix) const {
116 _bm.print_on_error(st, prefix);
117 }
118
119 size_t CMBitMap::compute_size(size_t heap_size) {
120 return ReservedSpace::allocation_align_size_up(heap_size / mark_distance());
121 }
122
123 size_t CMBitMap::mark_distance() {
124 return MinObjAlignmentInBytes * BitsPerByte;
125 }
126
127 void CMBitMap::initialize(MemRegion heap, G1RegionToSpaceMapper* storage) {
128 _bmStartWord = heap.start();
129 _bmWordSize = heap.word_size();
130
131 _bm.set_map((BitMap::bm_word_t*) storage->reserved().start());
132 _bm.set_size(_bmWordSize >> _shifter);
133
134 storage->set_mapping_changed_listener(&_listener);
135 }
136
137 void CMBitMapMappingChangedListener::on_commit(uint start_region, size_t num_regions, bool zero_filled) {
138 if (zero_filled) {
139 return;
140 }
141 // We need to clear the bitmap on commit, removing any existing information.
142 MemRegion mr(G1CollectedHeap::heap()->bottom_addr_for_region(start_region), num_regions * HeapRegion::GrainWords);
143 _bm->clearRange(mr);
144 }
145
146 // Closure used for clearing the given mark bitmap.
147 class ClearBitmapHRClosure : public HeapRegionClosure {
148 private:
149 ConcurrentMark* _cm;
150 CMBitMap* _bitmap;
151 bool _may_yield; // The closure may yield during iteration. If yielded, abort the iteration.
152 public:
153 ClearBitmapHRClosure(ConcurrentMark* cm, CMBitMap* bitmap, bool may_yield) : HeapRegionClosure(), _cm(cm), _bitmap(bitmap), _may_yield(may_yield) {
154 assert(!may_yield || cm != NULL, "CM must be non-NULL if this closure is expected to yield.");
155 }
156
157 virtual bool doHeapRegion(HeapRegion* r) {
158 size_t const chunk_size_in_words = M / HeapWordSize;
159
160 HeapWord* cur = r->bottom();
161 HeapWord* const end = r->end();
162
163 while (cur < end) {
164 MemRegion mr(cur, MIN2(cur + chunk_size_in_words, end));
165 _bitmap->clearRange(mr);
166
167 cur += chunk_size_in_words;
168
169 // Abort iteration if after yielding the marking has been aborted.
170 if (_may_yield && _cm->do_yield_check() && _cm->has_aborted()) {
171 return true;
172 }
173 // Repeat the asserts from before the start of the closure. We will do them
174 // as asserts here to minimize their overhead on the product. However, we
175 // will have them as guarantees at the beginning / end of the bitmap
176 // clearing to get some checking in the product.
177 assert(!_may_yield || _cm->cmThread()->during_cycle(), "invariant");
178 assert(!_may_yield || !G1CollectedHeap::heap()->mark_in_progress(), "invariant");
179 }
180
181 return false;
182 }
183 };
184
185 class ParClearNextMarkBitmapTask : public AbstractGangTask {
186 ClearBitmapHRClosure* _cl;
187 HeapRegionClaimer _hrclaimer;
188 bool _suspendible; // If the task is suspendible, workers must join the STS.
189
190 public:
191 ParClearNextMarkBitmapTask(ClearBitmapHRClosure *cl, uint n_workers, bool suspendible) :
192 _cl(cl), _suspendible(suspendible), AbstractGangTask("Parallel Clear Bitmap Task"), _hrclaimer(n_workers) {}
193
194 void work(uint worker_id) {
195 if (_suspendible) {
196 SuspendibleThreadSet::join();
197 }
198 G1CollectedHeap::heap()->heap_region_par_iterate(_cl, worker_id, &_hrclaimer, true);
199 if (_suspendible) {
200 SuspendibleThreadSet::leave();
201 }
202 }
203 };
204
205 void CMBitMap::clearAll() {
206 G1CollectedHeap* g1h = G1CollectedHeap::heap();
207 ClearBitmapHRClosure cl(NULL, this, false /* may_yield */);
208 uint n_workers = g1h->workers()->active_workers();
209 ParClearNextMarkBitmapTask task(&cl, n_workers, false);
210 g1h->workers()->run_task(&task);
211 guarantee(cl.complete(), "Must have completed iteration.");
212 return;
213 }
214
215 void CMBitMap::markRange(MemRegion mr) {
216 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
217 assert(!mr.is_empty(), "unexpected empty region");
218 assert((offsetToHeapWord(heapWordToOffset(mr.end())) ==
219 ((HeapWord *) mr.end())),
220 "markRange memory region end is not card aligned");
221 // convert address range into offset range
222 _bm.at_put_range(heapWordToOffset(mr.start()),
223 heapWordToOffset(mr.end()), true);
224 }
225
226 void CMBitMap::clearRange(MemRegion mr) {
227 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
228 assert(!mr.is_empty(), "unexpected empty region");
229 // convert address range into offset range
230 _bm.at_put_range(heapWordToOffset(mr.start()),
231 heapWordToOffset(mr.end()), false);
232 }
233
234 MemRegion CMBitMap::getAndClearMarkedRegion(HeapWord* addr,
235 HeapWord* end_addr) {
236 HeapWord* start = getNextMarkedWordAddress(addr);
237 start = MIN2(start, end_addr);
238 HeapWord* end = getNextUnmarkedWordAddress(start);
239 end = MIN2(end, end_addr);
240 assert(start <= end, "Consistency check");
241 MemRegion mr(start, end);
242 if (!mr.is_empty()) {
243 clearRange(mr);
244 }
245 return mr;
246 }
247
248 CMMarkStack::CMMarkStack(ConcurrentMark* cm) :
249 _base(NULL), _cm(cm)
250 #ifdef ASSERT
251 , _drain_in_progress(false)
252 , _drain_in_progress_yields(false)
253 #endif
254 {}
255
256 bool CMMarkStack::allocate(size_t capacity) {
257 // allocate a stack of the requisite depth
258 ReservedSpace rs(ReservedSpace::allocation_align_size_up(capacity * sizeof(oop)));
259 if (!rs.is_reserved()) {
260 warning("ConcurrentMark MarkStack allocation failure");
261 return false;
262 }
263 MemTracker::record_virtual_memory_type((address)rs.base(), mtGC);
264 if (!_virtual_space.initialize(rs, rs.size())) {
265 warning("ConcurrentMark MarkStack backing store failure");
266 // Release the virtual memory reserved for the marking stack
267 rs.release();
268 return false;
269 }
270 assert(_virtual_space.committed_size() == rs.size(),
271 "Didn't reserve backing store for all of ConcurrentMark stack?");
272 _base = (oop*) _virtual_space.low();
273 setEmpty();
274 _capacity = (jint) capacity;
275 _saved_index = -1;
276 _should_expand = false;
277 NOT_PRODUCT(_max_depth = 0);
278 return true;
279 }
280
281 void CMMarkStack::expand() {
282 // Called, during remark, if we've overflown the marking stack during marking.
283 assert(isEmpty(), "stack should been emptied while handling overflow");
284 assert(_capacity <= (jint) MarkStackSizeMax, "stack bigger than permitted");
285 // Clear expansion flag
286 _should_expand = false;
287 if (_capacity == (jint) MarkStackSizeMax) {
288 if (PrintGCDetails && Verbose) {
289 gclog_or_tty->print_cr(" (benign) Can't expand marking stack capacity, at max size limit");
290 }
291 return;
292 }
293 // Double capacity if possible
294 jint new_capacity = MIN2(_capacity*2, (jint) MarkStackSizeMax);
295 // Do not give up existing stack until we have managed to
296 // get the double capacity that we desired.
297 ReservedSpace rs(ReservedSpace::allocation_align_size_up(new_capacity *
298 sizeof(oop)));
299 if (rs.is_reserved()) {
300 // Release the backing store associated with old stack
301 _virtual_space.release();
302 // Reinitialize virtual space for new stack
303 if (!_virtual_space.initialize(rs, rs.size())) {
304 fatal("Not enough swap for expanded marking stack capacity");
305 }
306 _base = (oop*)(_virtual_space.low());
307 _index = 0;
308 _capacity = new_capacity;
309 } else {
310 if (PrintGCDetails && Verbose) {
311 // Failed to double capacity, continue;
312 gclog_or_tty->print(" (benign) Failed to expand marking stack capacity from "
313 SIZE_FORMAT"K to " SIZE_FORMAT"K",
314 _capacity / K, new_capacity / K);
315 }
316 }
317 }
318
319 void CMMarkStack::set_should_expand() {
320 // If we're resetting the marking state because of an
321 // marking stack overflow, record that we should, if
322 // possible, expand the stack.
323 _should_expand = _cm->has_overflown();
324 }
325
326 CMMarkStack::~CMMarkStack() {
327 if (_base != NULL) {
328 _base = NULL;
329 _virtual_space.release();
330 }
331 }
332
333 void CMMarkStack::par_push(oop ptr) {
334 while (true) {
335 if (isFull()) {
336 _overflow = true;
337 return;
338 }
339 // Otherwise...
340 jint index = _index;
341 jint next_index = index+1;
342 jint res = Atomic::cmpxchg(next_index, &_index, index);
343 if (res == index) {
344 _base[index] = ptr;
345 // Note that we don't maintain this atomically. We could, but it
346 // doesn't seem necessary.
347 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index));
348 return;
349 }
350 // Otherwise, we need to try again.
351 }
352 }
353
354 void CMMarkStack::par_adjoin_arr(oop* ptr_arr, int n) {
355 while (true) {
356 if (isFull()) {
357 _overflow = true;
358 return;
359 }
360 // Otherwise...
361 jint index = _index;
362 jint next_index = index + n;
363 if (next_index > _capacity) {
364 _overflow = true;
365 return;
366 }
367 jint res = Atomic::cmpxchg(next_index, &_index, index);
368 if (res == index) {
369 for (int i = 0; i < n; i++) {
370 int ind = index + i;
371 assert(ind < _capacity, "By overflow test above.");
372 _base[ind] = ptr_arr[i];
373 }
374 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index));
375 return;
376 }
377 // Otherwise, we need to try again.
378 }
379 }
380
381 void CMMarkStack::par_push_arr(oop* ptr_arr, int n) {
382 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
383 jint start = _index;
384 jint next_index = start + n;
385 if (next_index > _capacity) {
386 _overflow = true;
387 return;
388 }
389 // Otherwise.
390 _index = next_index;
391 for (int i = 0; i < n; i++) {
392 int ind = start + i;
393 assert(ind < _capacity, "By overflow test above.");
394 _base[ind] = ptr_arr[i];
395 }
396 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index));
397 }
398
399 bool CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) {
400 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
401 jint index = _index;
402 if (index == 0) {
403 *n = 0;
404 return false;
405 } else {
406 int k = MIN2(max, index);
407 jint new_ind = index - k;
408 for (int j = 0; j < k; j++) {
409 ptr_arr[j] = _base[new_ind + j];
410 }
411 _index = new_ind;
412 *n = k;
413 return true;
414 }
415 }
416
417 template<class OopClosureClass>
418 bool CMMarkStack::drain(OopClosureClass* cl, CMBitMap* bm, bool yield_after) {
419 assert(!_drain_in_progress || !_drain_in_progress_yields || yield_after
420 || SafepointSynchronize::is_at_safepoint(),
421 "Drain recursion must be yield-safe.");
422 bool res = true;
423 debug_only(_drain_in_progress = true);
424 debug_only(_drain_in_progress_yields = yield_after);
425 while (!isEmpty()) {
426 oop newOop = pop();
427 assert(G1CollectedHeap::heap()->is_in_reserved(newOop), "Bad pop");
428 assert(newOop->is_oop(), "Expected an oop");
429 assert(bm == NULL || bm->isMarked((HeapWord*)newOop),
430 "only grey objects on this stack");
431 newOop->oop_iterate(cl);
432 if (yield_after && _cm->do_yield_check()) {
433 res = false;
434 break;
435 }
436 }
437 debug_only(_drain_in_progress = false);
438 return res;
439 }
440
441 void CMMarkStack::note_start_of_gc() {
442 assert(_saved_index == -1,
443 "note_start_of_gc()/end_of_gc() bracketed incorrectly");
444 _saved_index = _index;
445 }
446
447 void CMMarkStack::note_end_of_gc() {
448 // This is intentionally a guarantee, instead of an assert. If we
449 // accidentally add something to the mark stack during GC, it
450 // will be a correctness issue so it's better if we crash. we'll
451 // only check this once per GC anyway, so it won't be a performance
452 // issue in any way.
453 guarantee(_saved_index == _index,
454 err_msg("saved index: %d index: %d", _saved_index, _index));
455 _saved_index = -1;
456 }
457
458 void CMMarkStack::oops_do(OopClosure* f) {
459 assert(_saved_index == _index,
460 err_msg("saved index: %d index: %d", _saved_index, _index));
461 for (int i = 0; i < _index; i += 1) {
462 f->do_oop(&_base[i]);
463 }
464 }
465
466 CMRootRegions::CMRootRegions() :
467 _young_list(NULL), _cm(NULL), _scan_in_progress(false),
468 _should_abort(false), _next_survivor(NULL) { }
469
470 void CMRootRegions::init(G1CollectedHeap* g1h, ConcurrentMark* cm) {
471 _young_list = g1h->young_list();
472 _cm = cm;
473 }
474
475 void CMRootRegions::prepare_for_scan() {
476 assert(!scan_in_progress(), "pre-condition");
477
478 // Currently, only survivors can be root regions.
479 assert(_next_survivor == NULL, "pre-condition");
480 _next_survivor = _young_list->first_survivor_region();
481 _scan_in_progress = (_next_survivor != NULL);
482 _should_abort = false;
483 }
484
485 HeapRegion* CMRootRegions::claim_next() {
486 if (_should_abort) {
487 // If someone has set the should_abort flag, we return NULL to
488 // force the caller to bail out of their loop.
489 return NULL;
490 }
491
492 // Currently, only survivors can be root regions.
493 HeapRegion* res = _next_survivor;
494 if (res != NULL) {
495 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
496 // Read it again in case it changed while we were waiting for the lock.
497 res = _next_survivor;
498 if (res != NULL) {
499 if (res == _young_list->last_survivor_region()) {
500 // We just claimed the last survivor so store NULL to indicate
501 // that we're done.
502 _next_survivor = NULL;
503 } else {
504 _next_survivor = res->get_next_young_region();
505 }
506 } else {
507 // Someone else claimed the last survivor while we were trying
508 // to take the lock so nothing else to do.
509 }
510 }
511 assert(res == NULL || res->is_survivor(), "post-condition");
512
513 return res;
514 }
515
516 void CMRootRegions::scan_finished() {
517 assert(scan_in_progress(), "pre-condition");
518
519 // Currently, only survivors can be root regions.
520 if (!_should_abort) {
521 assert(_next_survivor == NULL, "we should have claimed all survivors");
522 }
523 _next_survivor = NULL;
524
525 {
526 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
527 _scan_in_progress = false;
528 RootRegionScan_lock->notify_all();
529 }
530 }
531
532 bool CMRootRegions::wait_until_scan_finished() {
533 if (!scan_in_progress()) return false;
534
535 {
536 MutexLockerEx x(RootRegionScan_lock, Mutex::_no_safepoint_check_flag);
537 while (scan_in_progress()) {
538 RootRegionScan_lock->wait(Mutex::_no_safepoint_check_flag);
539 }
540 }
541 return true;
542 }
543
544 #ifdef _MSC_VER // the use of 'this' below gets a warning, make it go away
545 #pragma warning( disable:4355 ) // 'this' : used in base member initializer list
546 #endif // _MSC_VER
547
548 uint ConcurrentMark::scale_parallel_threads(uint n_par_threads) {
549 return MAX2((n_par_threads + 2) / 4, 1U);
550 }
551
552 ConcurrentMark::ConcurrentMark(G1CollectedHeap* g1h, G1RegionToSpaceMapper* prev_bitmap_storage, G1RegionToSpaceMapper* next_bitmap_storage) :
553 _g1h(g1h),
554 _markBitMap1(),
555 _markBitMap2(),
556 _parallel_marking_threads(0),
557 _max_parallel_marking_threads(0),
558 _sleep_factor(0.0),
559 _marking_task_overhead(1.0),
560 _cleanup_sleep_factor(0.0),
561 _cleanup_task_overhead(1.0),
562 _cleanup_list("Cleanup List"),
563 _region_bm((BitMap::idx_t)(g1h->max_regions()), false /* in_resource_area*/),
564 _card_bm((g1h->reserved_region().byte_size() + CardTableModRefBS::card_size - 1) >>
565 CardTableModRefBS::card_shift,
566 false /* in_resource_area*/),
567
568 _prevMarkBitMap(&_markBitMap1),
569 _nextMarkBitMap(&_markBitMap2),
570
571 _markStack(this),
572 // _finger set in set_non_marking_state
573
574 _max_worker_id(MAX2((uint)ParallelGCThreads, 1U)),
575 // _active_tasks set in set_non_marking_state
576 // _tasks set inside the constructor
577 _task_queues(new CMTaskQueueSet((int) _max_worker_id)),
578 _terminator(ParallelTaskTerminator((int) _max_worker_id, _task_queues)),
579
580 _has_overflown(false),
581 _concurrent(false),
582 _has_aborted(false),
583 _aborted_gc_id(GCId::undefined()),
584 _restart_for_overflow(false),
585 _concurrent_marking_in_progress(false),
586
587 // _verbose_level set below
588
589 _init_times(),
590 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(),
591 _cleanup_times(),
592 _total_counting_time(0.0),
593 _total_rs_scrub_time(0.0),
594
595 _parallel_workers(NULL),
596
597 _count_card_bitmaps(NULL),
598 _count_marked_bytes(NULL),
599 _completed_initialization(false) {
600 CMVerboseLevel verbose_level = (CMVerboseLevel) G1MarkingVerboseLevel;
601 if (verbose_level < no_verbose) {
602 verbose_level = no_verbose;
603 }
604 if (verbose_level > high_verbose) {
605 verbose_level = high_verbose;
606 }
607 _verbose_level = verbose_level;
608
609 if (verbose_low()) {
610 gclog_or_tty->print_cr("[global] init, heap start = "PTR_FORMAT", "
611 "heap end = " PTR_FORMAT, p2i(_heap_start), p2i(_heap_end));
612 }
613
614 _markBitMap1.initialize(g1h->reserved_region(), prev_bitmap_storage);
615 _markBitMap2.initialize(g1h->reserved_region(), next_bitmap_storage);
616
617 // Create & start a ConcurrentMark thread.
618 _cmThread = new ConcurrentMarkThread(this);
619 assert(cmThread() != NULL, "CM Thread should have been created");
620 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm");
621 if (_cmThread->osthread() == NULL) {
622 vm_shutdown_during_initialization("Could not create ConcurrentMarkThread");
623 }
624
625 assert(CGC_lock != NULL, "Where's the CGC_lock?");
626 assert(_markBitMap1.covers(g1h->reserved_region()), "_markBitMap1 inconsistency");
627 assert(_markBitMap2.covers(g1h->reserved_region()), "_markBitMap2 inconsistency");
628
629 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
630 satb_qs.set_buffer_size(G1SATBBufferSize);
631
632 _root_regions.init(_g1h, this);
633
634 if (ConcGCThreads > ParallelGCThreads) {
635 warning("Can't have more ConcGCThreads (" UINTX_FORMAT ") "
636 "than ParallelGCThreads (" UINTX_FORMAT ").",
637 ConcGCThreads, ParallelGCThreads);
638 return;
639 }
640 if (!FLAG_IS_DEFAULT(ConcGCThreads) && ConcGCThreads > 0) {
641 // Note: ConcGCThreads has precedence over G1MarkingOverheadPercent
642 // if both are set
643 _sleep_factor = 0.0;
644 _marking_task_overhead = 1.0;
645 } else if (G1MarkingOverheadPercent > 0) {
646 // We will calculate the number of parallel marking threads based
647 // on a target overhead with respect to the soft real-time goal
648 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
649 double overall_cm_overhead =
650 (double) MaxGCPauseMillis * marking_overhead /
651 (double) GCPauseIntervalMillis;
652 double cpu_ratio = 1.0 / (double) os::processor_count();
653 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio);
654 double marking_task_overhead =
655 overall_cm_overhead / marking_thread_num *
656 (double) os::processor_count();
657 double sleep_factor =
658 (1.0 - marking_task_overhead) / marking_task_overhead;
659
660 FLAG_SET_ERGO(uintx, ConcGCThreads, (uint) marking_thread_num);
661 _sleep_factor = sleep_factor;
662 _marking_task_overhead = marking_task_overhead;
663 } else {
664 // Calculate the number of parallel marking threads by scaling
665 // the number of parallel GC threads.
666 uint marking_thread_num = scale_parallel_threads((uint) ParallelGCThreads);
667 FLAG_SET_ERGO(uintx, ConcGCThreads, marking_thread_num);
668 _sleep_factor = 0.0;
669 _marking_task_overhead = 1.0;
670 }
671
672 assert(ConcGCThreads > 0, "Should have been set");
673 _parallel_marking_threads = (uint) ConcGCThreads;
674 _max_parallel_marking_threads = _parallel_marking_threads;
675
676 if (parallel_marking_threads() > 1) {
677 _cleanup_task_overhead = 1.0;
678 } else {
679 _cleanup_task_overhead = marking_task_overhead();
680 }
681 _cleanup_sleep_factor =
682 (1.0 - cleanup_task_overhead()) / cleanup_task_overhead();
683
684 #if 0
685 gclog_or_tty->print_cr("Marking Threads %d", parallel_marking_threads());
686 gclog_or_tty->print_cr("CM Marking Task Overhead %1.4lf", marking_task_overhead());
687 gclog_or_tty->print_cr("CM Sleep Factor %1.4lf", sleep_factor());
688 gclog_or_tty->print_cr("CL Marking Task Overhead %1.4lf", cleanup_task_overhead());
689 gclog_or_tty->print_cr("CL Sleep Factor %1.4lf", cleanup_sleep_factor());
690 #endif
691
692 _parallel_workers = new FlexibleWorkGang("G1 Marker",
693 _max_parallel_marking_threads, false, true);
694 if (_parallel_workers == NULL) {
695 vm_exit_during_initialization("Failed necessary allocation.");
696 } else {
697 _parallel_workers->initialize_workers();
698 }
699
700 if (FLAG_IS_DEFAULT(MarkStackSize)) {
701 size_t mark_stack_size =
702 MIN2(MarkStackSizeMax,
703 MAX2(MarkStackSize, (size_t) (parallel_marking_threads() * TASKQUEUE_SIZE)));
704 // Verify that the calculated value for MarkStackSize is in range.
705 // It would be nice to use the private utility routine from Arguments.
706 if (!(mark_stack_size >= 1 && mark_stack_size <= MarkStackSizeMax)) {
707 warning("Invalid value calculated for MarkStackSize (" SIZE_FORMAT "): "
708 "must be between 1 and " SIZE_FORMAT,
709 mark_stack_size, MarkStackSizeMax);
710 return;
711 }
712 FLAG_SET_ERGO(size_t, MarkStackSize, mark_stack_size);
713 } else {
714 // Verify MarkStackSize is in range.
715 if (FLAG_IS_CMDLINE(MarkStackSize)) {
716 if (FLAG_IS_DEFAULT(MarkStackSizeMax)) {
717 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
718 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT "): "
719 "must be between 1 and " SIZE_FORMAT,
720 MarkStackSize, MarkStackSizeMax);
721 return;
722 }
723 } else if (FLAG_IS_CMDLINE(MarkStackSizeMax)) {
724 if (!(MarkStackSize >= 1 && MarkStackSize <= MarkStackSizeMax)) {
725 warning("Invalid value specified for MarkStackSize (" SIZE_FORMAT ")"
726 " or for MarkStackSizeMax (" SIZE_FORMAT ")",
727 MarkStackSize, MarkStackSizeMax);
728 return;
729 }
730 }
731 }
732 }
733
734 if (!_markStack.allocate(MarkStackSize)) {
735 warning("Failed to allocate CM marking stack");
736 return;
737 }
738
739 _tasks = NEW_C_HEAP_ARRAY(CMTask*, _max_worker_id, mtGC);
740 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_worker_id, mtGC);
741
742 _count_card_bitmaps = NEW_C_HEAP_ARRAY(BitMap, _max_worker_id, mtGC);
743 _count_marked_bytes = NEW_C_HEAP_ARRAY(size_t*, _max_worker_id, mtGC);
744
745 BitMap::idx_t card_bm_size = _card_bm.size();
746
747 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
748 _active_tasks = _max_worker_id;
749
750 uint max_regions = _g1h->max_regions();
751 for (uint i = 0; i < _max_worker_id; ++i) {
752 CMTaskQueue* task_queue = new CMTaskQueue();
753 task_queue->initialize();
754 _task_queues->register_queue(i, task_queue);
755
756 _count_card_bitmaps[i] = BitMap(card_bm_size, false);
757 _count_marked_bytes[i] = NEW_C_HEAP_ARRAY(size_t, max_regions, mtGC);
758
759 _tasks[i] = new CMTask(i, this,
760 _count_marked_bytes[i],
761 &_count_card_bitmaps[i],
762 task_queue, _task_queues);
763
764 _accum_task_vtime[i] = 0.0;
765 }
766
767 // Calculate the card number for the bottom of the heap. Used
768 // in biasing indexes into the accounting card bitmaps.
769 _heap_bottom_card_num =
770 intptr_t(uintptr_t(_g1h->reserved_region().start()) >>
771 CardTableModRefBS::card_shift);
772
773 // Clear all the liveness counting data
774 clear_all_count_data();
775
776 // so that the call below can read a sensible value
777 _heap_start = g1h->reserved_region().start();
778 set_non_marking_state();
779 _completed_initialization = true;
780 }
781
782 void ConcurrentMark::reset() {
783 // Starting values for these two. This should be called in a STW
784 // phase.
785 MemRegion reserved = _g1h->g1_reserved();
786 _heap_start = reserved.start();
787 _heap_end = reserved.end();
788
789 // Separated the asserts so that we know which one fires.
790 assert(_heap_start != NULL, "heap bounds should look ok");
791 assert(_heap_end != NULL, "heap bounds should look ok");
792 assert(_heap_start < _heap_end, "heap bounds should look ok");
793
794 // Reset all the marking data structures and any necessary flags
795 reset_marking_state();
796
797 if (verbose_low()) {
798 gclog_or_tty->print_cr("[global] resetting");
799 }
800
801 // We do reset all of them, since different phases will use
802 // different number of active threads. So, it's easiest to have all
803 // of them ready.
804 for (uint i = 0; i < _max_worker_id; ++i) {
805 _tasks[i]->reset(_nextMarkBitMap);
806 }
807
808 // we need this to make sure that the flag is on during the evac
809 // pause with initial mark piggy-backed
810 set_concurrent_marking_in_progress();
811 }
812
813
814 void ConcurrentMark::reset_marking_state(bool clear_overflow) {
815 _markStack.set_should_expand();
816 _markStack.setEmpty(); // Also clears the _markStack overflow flag
817 if (clear_overflow) {
818 clear_has_overflown();
819 } else {
820 assert(has_overflown(), "pre-condition");
821 }
822 _finger = _heap_start;
823
824 for (uint i = 0; i < _max_worker_id; ++i) {
825 CMTaskQueue* queue = _task_queues->queue(i);
826 queue->set_empty();
827 }
828 }
829
830 void ConcurrentMark::set_concurrency(uint active_tasks) {
831 assert(active_tasks <= _max_worker_id, "we should not have more");
832
833 _active_tasks = active_tasks;
834 // Need to update the three data structures below according to the
835 // number of active threads for this phase.
836 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
837 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
838 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
839 }
840
841 void ConcurrentMark::set_concurrency_and_phase(uint active_tasks, bool concurrent) {
842 set_concurrency(active_tasks);
843
844 _concurrent = concurrent;
845 // We propagate this to all tasks, not just the active ones.
846 for (uint i = 0; i < _max_worker_id; ++i)
847 _tasks[i]->set_concurrent(concurrent);
848
849 if (concurrent) {
850 set_concurrent_marking_in_progress();
851 } else {
852 // We currently assume that the concurrent flag has been set to
853 // false before we start remark. At this point we should also be
854 // in a STW phase.
855 assert(!concurrent_marking_in_progress(), "invariant");
856 assert(out_of_regions(),
857 err_msg("only way to get here: _finger: "PTR_FORMAT", _heap_end: "PTR_FORMAT,
858 p2i(_finger), p2i(_heap_end)));
859 }
860 }
861
862 void ConcurrentMark::set_non_marking_state() {
863 // We set the global marking state to some default values when we're
864 // not doing marking.
865 reset_marking_state();
866 _active_tasks = 0;
867 clear_concurrent_marking_in_progress();
868 }
869
870 ConcurrentMark::~ConcurrentMark() {
871 // The ConcurrentMark instance is never freed.
872 ShouldNotReachHere();
873 }
874
875 void ConcurrentMark::clearNextBitmap() {
876 G1CollectedHeap* g1h = G1CollectedHeap::heap();
877
878 // Make sure that the concurrent mark thread looks to still be in
879 // the current cycle.
880 guarantee(cmThread()->during_cycle(), "invariant");
881
882 // We are finishing up the current cycle by clearing the next
883 // marking bitmap and getting it ready for the next cycle. During
884 // this time no other cycle can start. So, let's make sure that this
885 // is the case.
886 guarantee(!g1h->mark_in_progress(), "invariant");
887
888 ClearBitmapHRClosure cl(this, _nextMarkBitMap, true /* may_yield */);
889 ParClearNextMarkBitmapTask task(&cl, parallel_marking_threads(), true);
890 _parallel_workers->run_task(&task);
891
892 // Clear the liveness counting data. If the marking has been aborted, the abort()
893 // call already did that.
894 if (cl.complete()) {
895 clear_all_count_data();
896 }
897
898 // Repeat the asserts from above.
899 guarantee(cmThread()->during_cycle(), "invariant");
900 guarantee(!g1h->mark_in_progress(), "invariant");
901 }
902
903 class CheckBitmapClearHRClosure : public HeapRegionClosure {
904 CMBitMap* _bitmap;
905 bool _error;
906 public:
907 CheckBitmapClearHRClosure(CMBitMap* bitmap) : _bitmap(bitmap) {
908 }
909
910 virtual bool doHeapRegion(HeapRegion* r) {
911 // This closure can be called concurrently to the mutator, so we must make sure
912 // that the result of the getNextMarkedWordAddress() call is compared to the
913 // value passed to it as limit to detect any found bits.
914 // We can use the region's orig_end() for the limit and the comparison value
915 // as it always contains the "real" end of the region that never changes and
916 // has no side effects.
917 // Due to the latter, there can also be no problem with the compiler generating
918 // reloads of the orig_end() call.
919 HeapWord* end = r->orig_end();
920 return _bitmap->getNextMarkedWordAddress(r->bottom(), end) != end;
921 }
922 };
923
924 bool ConcurrentMark::nextMarkBitmapIsClear() {
925 CheckBitmapClearHRClosure cl(_nextMarkBitMap);
926 _g1h->heap_region_iterate(&cl);
927 return cl.complete();
928 }
929
930 class NoteStartOfMarkHRClosure: public HeapRegionClosure {
931 public:
932 bool doHeapRegion(HeapRegion* r) {
933 if (!r->is_continues_humongous()) {
934 r->note_start_of_marking();
935 }
936 return false;
937 }
938 };
939
940 void ConcurrentMark::checkpointRootsInitialPre() {
941 G1CollectedHeap* g1h = G1CollectedHeap::heap();
942 G1CollectorPolicy* g1p = g1h->g1_policy();
943
944 _has_aborted = false;
945
946 #ifndef PRODUCT
947 if (G1PrintReachableAtInitialMark) {
948 print_reachable("at-cycle-start",
949 VerifyOption_G1UsePrevMarking, true /* all */);
950 }
951 #endif
952
953 // Initialize marking structures. This has to be done in a STW phase.
954 reset();
955
956 // For each region note start of marking.
957 NoteStartOfMarkHRClosure startcl;
958 g1h->heap_region_iterate(&startcl);
959 }
960
961
962 void ConcurrentMark::checkpointRootsInitialPost() {
963 G1CollectedHeap* g1h = G1CollectedHeap::heap();
964
965 // If we force an overflow during remark, the remark operation will
966 // actually abort and we'll restart concurrent marking. If we always
967 // force an overflow during remark we'll never actually complete the
968 // marking phase. So, we initialize this here, at the start of the
969 // cycle, so that at the remaining overflow number will decrease at
970 // every remark and we'll eventually not need to cause one.
971 force_overflow_stw()->init();
972
973 // Start Concurrent Marking weak-reference discovery.
974 ReferenceProcessor* rp = g1h->ref_processor_cm();
975 // enable ("weak") refs discovery
976 rp->enable_discovery();
977 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
978
979 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
980 // This is the start of the marking cycle, we're expected all
981 // threads to have SATB queues with active set to false.
982 satb_mq_set.set_active_all_threads(true, /* new active value */
983 false /* expected_active */);
984
985 _root_regions.prepare_for_scan();
986
987 // update_g1_committed() will be called at the end of an evac pause
988 // when marking is on. So, it's also called at the end of the
989 // initial-mark pause to update the heap end, if the heap expands
990 // during it. No need to call it here.
991 }
992
993 /*
994 * Notice that in the next two methods, we actually leave the STS
995 * during the barrier sync and join it immediately afterwards. If we
996 * do not do this, the following deadlock can occur: one thread could
997 * be in the barrier sync code, waiting for the other thread to also
998 * sync up, whereas another one could be trying to yield, while also
999 * waiting for the other threads to sync up too.
1000 *
1001 * Note, however, that this code is also used during remark and in
1002 * this case we should not attempt to leave / enter the STS, otherwise
1003 * we'll either hit an assert (debug / fastdebug) or deadlock
1004 * (product). So we should only leave / enter the STS if we are
1005 * operating concurrently.
1006 *
1007 * Because the thread that does the sync barrier has left the STS, it
1008 * is possible to be suspended for a Full GC or an evacuation pause
1009 * could occur. This is actually safe, since the entering the sync
1010 * barrier is one of the last things do_marking_step() does, and it
1011 * doesn't manipulate any data structures afterwards.
1012 */
1013
1014 void ConcurrentMark::enter_first_sync_barrier(uint worker_id) {
1015 if (verbose_low()) {
1016 gclog_or_tty->print_cr("[%u] entering first barrier", worker_id);
1017 }
1018
1019 if (concurrent()) {
1020 SuspendibleThreadSet::leave();
1021 }
1022
1023 bool barrier_aborted = !_first_overflow_barrier_sync.enter();
1024
1025 if (concurrent()) {
1026 SuspendibleThreadSet::join();
1027 }
1028 // at this point everyone should have synced up and not be doing any
1029 // more work
1030
1031 if (verbose_low()) {
1032 if (barrier_aborted) {
1033 gclog_or_tty->print_cr("[%u] aborted first barrier", worker_id);
1034 } else {
1035 gclog_or_tty->print_cr("[%u] leaving first barrier", worker_id);
1036 }
1037 }
1038
1039 if (barrier_aborted) {
1040 // If the barrier aborted we ignore the overflow condition and
1041 // just abort the whole marking phase as quickly as possible.
1042 return;
1043 }
1044
1045 // If we're executing the concurrent phase of marking, reset the marking
1046 // state; otherwise the marking state is reset after reference processing,
1047 // during the remark pause.
1048 // If we reset here as a result of an overflow during the remark we will
1049 // see assertion failures from any subsequent set_concurrency_and_phase()
1050 // calls.
1051 if (concurrent()) {
1052 // let the task associated with with worker 0 do this
1053 if (worker_id == 0) {
1054 // task 0 is responsible for clearing the global data structures
1055 // We should be here because of an overflow. During STW we should
1056 // not clear the overflow flag since we rely on it being true when
1057 // we exit this method to abort the pause and restart concurrent
1058 // marking.
1059 reset_marking_state(true /* clear_overflow */);
1060 force_overflow()->update();
1061
1062 if (G1Log::fine()) {
1063 gclog_or_tty->gclog_stamp(concurrent_gc_id());
1064 gclog_or_tty->print_cr("[GC concurrent-mark-reset-for-overflow]");
1065 }
1066 }
1067 }
1068
1069 // after this, each task should reset its own data structures then
1070 // then go into the second barrier
1071 }
1072
1073 void ConcurrentMark::enter_second_sync_barrier(uint worker_id) {
1074 if (verbose_low()) {
1075 gclog_or_tty->print_cr("[%u] entering second barrier", worker_id);
1076 }
1077
1078 if (concurrent()) {
1079 SuspendibleThreadSet::leave();
1080 }
1081
1082 bool barrier_aborted = !_second_overflow_barrier_sync.enter();
1083
1084 if (concurrent()) {
1085 SuspendibleThreadSet::join();
1086 }
1087 // at this point everything should be re-initialized and ready to go
1088
1089 if (verbose_low()) {
1090 if (barrier_aborted) {
1091 gclog_or_tty->print_cr("[%u] aborted second barrier", worker_id);
1092 } else {
1093 gclog_or_tty->print_cr("[%u] leaving second barrier", worker_id);
1094 }
1095 }
1096 }
1097
1098 #ifndef PRODUCT
1099 void ForceOverflowSettings::init() {
1100 _num_remaining = G1ConcMarkForceOverflow;
1101 _force = false;
1102 update();
1103 }
1104
1105 void ForceOverflowSettings::update() {
1106 if (_num_remaining > 0) {
1107 _num_remaining -= 1;
1108 _force = true;
1109 } else {
1110 _force = false;
1111 }
1112 }
1113
1114 bool ForceOverflowSettings::should_force() {
1115 if (_force) {
1116 _force = false;
1117 return true;
1118 } else {
1119 return false;
1120 }
1121 }
1122 #endif // !PRODUCT
1123
1124 class CMConcurrentMarkingTask: public AbstractGangTask {
1125 private:
1126 ConcurrentMark* _cm;
1127 ConcurrentMarkThread* _cmt;
1128
1129 public:
1130 void work(uint worker_id) {
1131 assert(Thread::current()->is_ConcurrentGC_thread(),
1132 "this should only be done by a conc GC thread");
1133 ResourceMark rm;
1134
1135 double start_vtime = os::elapsedVTime();
1136
1137 SuspendibleThreadSet::join();
1138
1139 assert(worker_id < _cm->active_tasks(), "invariant");
1140 CMTask* the_task = _cm->task(worker_id);
1141 the_task->record_start_time();
1142 if (!_cm->has_aborted()) {
1143 do {
1144 double start_vtime_sec = os::elapsedVTime();
1145 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1146
1147 the_task->do_marking_step(mark_step_duration_ms,
1148 true /* do_termination */,
1149 false /* is_serial*/);
1150
1151 double end_vtime_sec = os::elapsedVTime();
1152 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec;
1153 _cm->clear_has_overflown();
1154
1155 _cm->do_yield_check(worker_id);
1156
1157 jlong sleep_time_ms;
1158 if (!_cm->has_aborted() && the_task->has_aborted()) {
1159 sleep_time_ms =
1160 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0);
1161 SuspendibleThreadSet::leave();
1162 os::sleep(Thread::current(), sleep_time_ms, false);
1163 SuspendibleThreadSet::join();
1164 }
1165 } while (!_cm->has_aborted() && the_task->has_aborted());
1166 }
1167 the_task->record_end_time();
1168 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant");
1169
1170 SuspendibleThreadSet::leave();
1171
1172 double end_vtime = os::elapsedVTime();
1173 _cm->update_accum_task_vtime(worker_id, end_vtime - start_vtime);
1174 }
1175
1176 CMConcurrentMarkingTask(ConcurrentMark* cm,
1177 ConcurrentMarkThread* cmt) :
1178 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { }
1179
1180 ~CMConcurrentMarkingTask() { }
1181 };
1182
1183 // Calculates the number of active workers for a concurrent
1184 // phase.
1185 uint ConcurrentMark::calc_parallel_marking_threads() {
1186 uint n_conc_workers = 0;
1187 if (!UseDynamicNumberOfGCThreads ||
1188 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
1189 !ForceDynamicNumberOfGCThreads)) {
1190 n_conc_workers = max_parallel_marking_threads();
1191 } else {
1192 n_conc_workers =
1193 AdaptiveSizePolicy::calc_default_active_workers(
1194 max_parallel_marking_threads(),
1195 1, /* Minimum workers */
1196 parallel_marking_threads(),
1197 Threads::number_of_non_daemon_threads());
1198 // Don't scale down "n_conc_workers" by scale_parallel_threads() because
1199 // that scaling has already gone into "_max_parallel_marking_threads".
1200 }
1201 assert(n_conc_workers > 0, "Always need at least 1");
1202 return n_conc_workers;
1203 }
1204
1205 void ConcurrentMark::scanRootRegion(HeapRegion* hr, uint worker_id) {
1206 // Currently, only survivors can be root regions.
1207 assert(hr->next_top_at_mark_start() == hr->bottom(), "invariant");
1208 G1RootRegionScanClosure cl(_g1h, this, worker_id);
1209
1210 const uintx interval = PrefetchScanIntervalInBytes;
1211 HeapWord* curr = hr->bottom();
1212 const HeapWord* end = hr->top();
1213 while (curr < end) {
1214 Prefetch::read(curr, interval);
1215 oop obj = oop(curr);
1216 int size = obj->oop_iterate(&cl);
1217 assert(size == obj->size(), "sanity");
1218 curr += size;
1219 }
1220 }
1221
1222 class CMRootRegionScanTask : public AbstractGangTask {
1223 private:
1224 ConcurrentMark* _cm;
1225
1226 public:
1227 CMRootRegionScanTask(ConcurrentMark* cm) :
1228 AbstractGangTask("Root Region Scan"), _cm(cm) { }
1229
1230 void work(uint worker_id) {
1231 assert(Thread::current()->is_ConcurrentGC_thread(),
1232 "this should only be done by a conc GC thread");
1233
1234 CMRootRegions* root_regions = _cm->root_regions();
1235 HeapRegion* hr = root_regions->claim_next();
1236 while (hr != NULL) {
1237 _cm->scanRootRegion(hr, worker_id);
1238 hr = root_regions->claim_next();
1239 }
1240 }
1241 };
1242
1243 void ConcurrentMark::scanRootRegions() {
1244 // Start of concurrent marking.
1245 ClassLoaderDataGraph::clear_claimed_marks();
1246
1247 // scan_in_progress() will have been set to true only if there was
1248 // at least one root region to scan. So, if it's false, we
1249 // should not attempt to do any further work.
1250 if (root_regions()->scan_in_progress()) {
1251 _parallel_marking_threads = calc_parallel_marking_threads();
1252 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1253 "Maximum number of marking threads exceeded");
1254 uint active_workers = MAX2(1U, parallel_marking_threads());
1255
1256 CMRootRegionScanTask task(this);
1257 _parallel_workers->set_active_workers(active_workers);
1258 _parallel_workers->run_task(&task);
1259
1260 // It's possible that has_aborted() is true here without actually
1261 // aborting the survivor scan earlier. This is OK as it's
1262 // mainly used for sanity checking.
1263 root_regions()->scan_finished();
1264 }
1265 }
1266
1267 void ConcurrentMark::markFromRoots() {
1268 // we might be tempted to assert that:
1269 // assert(asynch == !SafepointSynchronize::is_at_safepoint(),
1270 // "inconsistent argument?");
1271 // However that wouldn't be right, because it's possible that
1272 // a safepoint is indeed in progress as a younger generation
1273 // stop-the-world GC happens even as we mark in this generation.
1274
1275 _restart_for_overflow = false;
1276 force_overflow_conc()->init();
1277
1278 // _g1h has _n_par_threads
1279 _parallel_marking_threads = calc_parallel_marking_threads();
1280 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1281 "Maximum number of marking threads exceeded");
1282
1283 uint active_workers = MAX2(1U, parallel_marking_threads());
1284
1285 // Parallel task terminator is set in "set_concurrency_and_phase()"
1286 set_concurrency_and_phase(active_workers, true /* concurrent */);
1287
1288 CMConcurrentMarkingTask markingTask(this, cmThread());
1289 _parallel_workers->set_active_workers(active_workers);
1290 // Don't set _n_par_threads because it affects MT in process_roots()
1291 // and the decisions on that MT processing is made elsewhere.
1292 assert(_parallel_workers->active_workers() > 0, "Should have been set");
1293 _parallel_workers->run_task(&markingTask);
1294 print_stats();
1295 }
1296
1297 // Helper class to get rid of some boilerplate code.
1298 class G1CMTraceTime : public GCTraceTime {
1299 static bool doit_and_prepend(bool doit) {
1300 if (doit) {
1301 gclog_or_tty->put(' ');
1302 }
1303 return doit;
1304 }
1305
1306 public:
1307 G1CMTraceTime(const char* title, bool doit)
1308 : GCTraceTime(title, doit_and_prepend(doit), false, G1CollectedHeap::heap()->gc_timer_cm(),
1309 G1CollectedHeap::heap()->concurrent_mark()->concurrent_gc_id()) {
1310 }
1311 };
1312
1313 void ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) {
1314 // world is stopped at this checkpoint
1315 assert(SafepointSynchronize::is_at_safepoint(),
1316 "world should be stopped");
1317
1318 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1319
1320 // If a full collection has happened, we shouldn't do this.
1321 if (has_aborted()) {
1322 g1h->set_marking_complete(); // So bitmap clearing isn't confused
1323 return;
1324 }
1325
1326 SvcGCMarker sgcm(SvcGCMarker::OTHER);
1327
1328 if (VerifyDuringGC) {
1329 HandleMark hm; // handle scope
1330 g1h->prepare_for_verify();
1331 Universe::verify(VerifyOption_G1UsePrevMarking,
1332 " VerifyDuringGC:(before)");
1333 }
1334 g1h->check_bitmaps("Remark Start");
1335
1336 G1CollectorPolicy* g1p = g1h->g1_policy();
1337 g1p->record_concurrent_mark_remark_start();
1338
1339 double start = os::elapsedTime();
1340
1341 checkpointRootsFinalWork();
1342
1343 double mark_work_end = os::elapsedTime();
1344
1345 weakRefsWork(clear_all_soft_refs);
1346
1347 if (has_overflown()) {
1348 // Oops. We overflowed. Restart concurrent marking.
1349 _restart_for_overflow = true;
1350 if (G1TraceMarkStackOverflow) {
1351 gclog_or_tty->print_cr("\nRemark led to restart for overflow.");
1352 }
1353
1354 // Verify the heap w.r.t. the previous marking bitmap.
1355 if (VerifyDuringGC) {
1356 HandleMark hm; // handle scope
1357 g1h->prepare_for_verify();
1358 Universe::verify(VerifyOption_G1UsePrevMarking,
1359 " VerifyDuringGC:(overflow)");
1360 }
1361
1362 // Clear the marking state because we will be restarting
1363 // marking due to overflowing the global mark stack.
1364 reset_marking_state();
1365 } else {
1366 {
1367 G1CMTraceTime trace("GC aggregate-data", G1Log::finer());
1368
1369 // Aggregate the per-task counting data that we have accumulated
1370 // while marking.
1371 aggregate_count_data();
1372 }
1373
1374 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1375 // We're done with marking.
1376 // This is the end of the marking cycle, we're expected all
1377 // threads to have SATB queues with active set to true.
1378 satb_mq_set.set_active_all_threads(false, /* new active value */
1379 true /* expected_active */);
1380
1381 if (VerifyDuringGC) {
1382 HandleMark hm; // handle scope
1383 g1h->prepare_for_verify();
1384 Universe::verify(VerifyOption_G1UseNextMarking,
1385 " VerifyDuringGC:(after)");
1386 }
1387 g1h->check_bitmaps("Remark End");
1388 assert(!restart_for_overflow(), "sanity");
1389 // Completely reset the marking state since marking completed
1390 set_non_marking_state();
1391 }
1392
1393 // Expand the marking stack, if we have to and if we can.
1394 if (_markStack.should_expand()) {
1395 _markStack.expand();
1396 }
1397
1398 // Statistics
1399 double now = os::elapsedTime();
1400 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1401 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1402 _remark_times.add((now - start) * 1000.0);
1403
1404 g1p->record_concurrent_mark_remark_end();
1405
1406 G1CMIsAliveClosure is_alive(g1h);
1407 g1h->gc_tracer_cm()->report_object_count_after_gc(&is_alive);
1408 }
1409
1410 // Base class of the closures that finalize and verify the
1411 // liveness counting data.
1412 class CMCountDataClosureBase: public HeapRegionClosure {
1413 protected:
1414 G1CollectedHeap* _g1h;
1415 ConcurrentMark* _cm;
1416 CardTableModRefBS* _ct_bs;
1417
1418 BitMap* _region_bm;
1419 BitMap* _card_bm;
1420
1421 // Takes a region that's not empty (i.e., it has at least one
1422 // live object in it and sets its corresponding bit on the region
1423 // bitmap to 1. If the region is "starts humongous" it will also set
1424 // to 1 the bits on the region bitmap that correspond to its
1425 // associated "continues humongous" regions.
1426 void set_bit_for_region(HeapRegion* hr) {
1427 assert(!hr->is_continues_humongous(), "should have filtered those out");
1428
1429 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1430 if (!hr->is_starts_humongous()) {
1431 // Normal (non-humongous) case: just set the bit.
1432 _region_bm->par_at_put(index, true);
1433 } else {
1434 // Starts humongous case: calculate how many regions are part of
1435 // this humongous region and then set the bit range.
1436 BitMap::idx_t end_index = (BitMap::idx_t) hr->last_hc_index();
1437 _region_bm->par_at_put_range(index, end_index, true);
1438 }
1439 }
1440
1441 public:
1442 CMCountDataClosureBase(G1CollectedHeap* g1h,
1443 BitMap* region_bm, BitMap* card_bm):
1444 _g1h(g1h), _cm(g1h->concurrent_mark()),
1445 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
1446 _region_bm(region_bm), _card_bm(card_bm) { }
1447 };
1448
1449 // Closure that calculates the # live objects per region. Used
1450 // for verification purposes during the cleanup pause.
1451 class CalcLiveObjectsClosure: public CMCountDataClosureBase {
1452 CMBitMapRO* _bm;
1453 size_t _region_marked_bytes;
1454
1455 public:
1456 CalcLiveObjectsClosure(CMBitMapRO *bm, G1CollectedHeap* g1h,
1457 BitMap* region_bm, BitMap* card_bm) :
1458 CMCountDataClosureBase(g1h, region_bm, card_bm),
1459 _bm(bm), _region_marked_bytes(0) { }
1460
1461 bool doHeapRegion(HeapRegion* hr) {
1462
1463 if (hr->is_continues_humongous()) {
1464 // We will ignore these here and process them when their
1465 // associated "starts humongous" region is processed (see
1466 // set_bit_for_heap_region()). Note that we cannot rely on their
1467 // associated "starts humongous" region to have their bit set to
1468 // 1 since, due to the region chunking in the parallel region
1469 // iteration, a "continues humongous" region might be visited
1470 // before its associated "starts humongous".
1471 return false;
1472 }
1473
1474 HeapWord* ntams = hr->next_top_at_mark_start();
1475 HeapWord* start = hr->bottom();
1476
1477 assert(start <= hr->end() && start <= ntams && ntams <= hr->end(),
1478 err_msg("Preconditions not met - "
1479 "start: "PTR_FORMAT", ntams: "PTR_FORMAT", end: "PTR_FORMAT,
1480 p2i(start), p2i(ntams), p2i(hr->end())));
1481
1482 // Find the first marked object at or after "start".
1483 start = _bm->getNextMarkedWordAddress(start, ntams);
1484
1485 size_t marked_bytes = 0;
1486
1487 while (start < ntams) {
1488 oop obj = oop(start);
1489 int obj_sz = obj->size();
1490 HeapWord* obj_end = start + obj_sz;
1491
1492 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
1493 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(obj_end);
1494
1495 // Note: if we're looking at the last region in heap - obj_end
1496 // could be actually just beyond the end of the heap; end_idx
1497 // will then correspond to a (non-existent) card that is also
1498 // just beyond the heap.
1499 if (_g1h->is_in_g1_reserved(obj_end) && !_ct_bs->is_card_aligned(obj_end)) {
1500 // end of object is not card aligned - increment to cover
1501 // all the cards spanned by the object
1502 end_idx += 1;
1503 }
1504
1505 // Set the bits in the card BM for the cards spanned by this object.
1506 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1507
1508 // Add the size of this object to the number of marked bytes.
1509 marked_bytes += (size_t)obj_sz * HeapWordSize;
1510
1511 // Find the next marked object after this one.
1512 start = _bm->getNextMarkedWordAddress(obj_end, ntams);
1513 }
1514
1515 // Mark the allocated-since-marking portion...
1516 HeapWord* top = hr->top();
1517 if (ntams < top) {
1518 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1519 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1520
1521 // Note: if we're looking at the last region in heap - top
1522 // could be actually just beyond the end of the heap; end_idx
1523 // will then correspond to a (non-existent) card that is also
1524 // just beyond the heap.
1525 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1526 // end of object is not card aligned - increment to cover
1527 // all the cards spanned by the object
1528 end_idx += 1;
1529 }
1530 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1531
1532 // This definitely means the region has live objects.
1533 set_bit_for_region(hr);
1534 }
1535
1536 // Update the live region bitmap.
1537 if (marked_bytes > 0) {
1538 set_bit_for_region(hr);
1539 }
1540
1541 // Set the marked bytes for the current region so that
1542 // it can be queried by a calling verification routine
1543 _region_marked_bytes = marked_bytes;
1544
1545 return false;
1546 }
1547
1548 size_t region_marked_bytes() const { return _region_marked_bytes; }
1549 };
1550
1551 // Heap region closure used for verifying the counting data
1552 // that was accumulated concurrently and aggregated during
1553 // the remark pause. This closure is applied to the heap
1554 // regions during the STW cleanup pause.
1555
1556 class VerifyLiveObjectDataHRClosure: public HeapRegionClosure {
1557 G1CollectedHeap* _g1h;
1558 ConcurrentMark* _cm;
1559 CalcLiveObjectsClosure _calc_cl;
1560 BitMap* _region_bm; // Region BM to be verified
1561 BitMap* _card_bm; // Card BM to be verified
1562 bool _verbose; // verbose output?
1563
1564 BitMap* _exp_region_bm; // Expected Region BM values
1565 BitMap* _exp_card_bm; // Expected card BM values
1566
1567 int _failures;
1568
1569 public:
1570 VerifyLiveObjectDataHRClosure(G1CollectedHeap* g1h,
1571 BitMap* region_bm,
1572 BitMap* card_bm,
1573 BitMap* exp_region_bm,
1574 BitMap* exp_card_bm,
1575 bool verbose) :
1576 _g1h(g1h), _cm(g1h->concurrent_mark()),
1577 _calc_cl(_cm->nextMarkBitMap(), g1h, exp_region_bm, exp_card_bm),
1578 _region_bm(region_bm), _card_bm(card_bm), _verbose(verbose),
1579 _exp_region_bm(exp_region_bm), _exp_card_bm(exp_card_bm),
1580 _failures(0) { }
1581
1582 int failures() const { return _failures; }
1583
1584 bool doHeapRegion(HeapRegion* hr) {
1585 if (hr->is_continues_humongous()) {
1586 // We will ignore these here and process them when their
1587 // associated "starts humongous" region is processed (see
1588 // set_bit_for_heap_region()). Note that we cannot rely on their
1589 // associated "starts humongous" region to have their bit set to
1590 // 1 since, due to the region chunking in the parallel region
1591 // iteration, a "continues humongous" region might be visited
1592 // before its associated "starts humongous".
1593 return false;
1594 }
1595
1596 int failures = 0;
1597
1598 // Call the CalcLiveObjectsClosure to walk the marking bitmap for
1599 // this region and set the corresponding bits in the expected region
1600 // and card bitmaps.
1601 bool res = _calc_cl.doHeapRegion(hr);
1602 assert(res == false, "should be continuing");
1603
1604 MutexLockerEx x((_verbose ? ParGCRareEvent_lock : NULL),
1605 Mutex::_no_safepoint_check_flag);
1606
1607 // Verify the marked bytes for this region.
1608 size_t exp_marked_bytes = _calc_cl.region_marked_bytes();
1609 size_t act_marked_bytes = hr->next_marked_bytes();
1610
1611 // We're not OK if expected marked bytes > actual marked bytes. It means
1612 // we have missed accounting some objects during the actual marking.
1613 if (exp_marked_bytes > act_marked_bytes) {
1614 if (_verbose) {
1615 gclog_or_tty->print_cr("Region %u: marked bytes mismatch: "
1616 "expected: " SIZE_FORMAT ", actual: " SIZE_FORMAT,
1617 hr->hrm_index(), exp_marked_bytes, act_marked_bytes);
1618 }
1619 failures += 1;
1620 }
1621
1622 // Verify the bit, for this region, in the actual and expected
1623 // (which was just calculated) region bit maps.
1624 // We're not OK if the bit in the calculated expected region
1625 // bitmap is set and the bit in the actual region bitmap is not.
1626 BitMap::idx_t index = (BitMap::idx_t) hr->hrm_index();
1627
1628 bool expected = _exp_region_bm->at(index);
1629 bool actual = _region_bm->at(index);
1630 if (expected && !actual) {
1631 if (_verbose) {
1632 gclog_or_tty->print_cr("Region %u: region bitmap mismatch: "
1633 "expected: %s, actual: %s",
1634 hr->hrm_index(),
1635 BOOL_TO_STR(expected), BOOL_TO_STR(actual));
1636 }
1637 failures += 1;
1638 }
1639
1640 // Verify that the card bit maps for the cards spanned by the current
1641 // region match. We have an error if we have a set bit in the expected
1642 // bit map and the corresponding bit in the actual bitmap is not set.
1643
1644 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(hr->bottom());
1645 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(hr->top());
1646
1647 for (BitMap::idx_t i = start_idx; i < end_idx; i+=1) {
1648 expected = _exp_card_bm->at(i);
1649 actual = _card_bm->at(i);
1650
1651 if (expected && !actual) {
1652 if (_verbose) {
1653 gclog_or_tty->print_cr("Region %u: card bitmap mismatch at " SIZE_FORMAT ": "
1654 "expected: %s, actual: %s",
1655 hr->hrm_index(), i,
1656 BOOL_TO_STR(expected), BOOL_TO_STR(actual));
1657 }
1658 failures += 1;
1659 }
1660 }
1661
1662 if (failures > 0 && _verbose) {
1663 gclog_or_tty->print_cr("Region " HR_FORMAT ", ntams: " PTR_FORMAT ", "
1664 "marked_bytes: calc/actual " SIZE_FORMAT "/" SIZE_FORMAT,
1665 HR_FORMAT_PARAMS(hr), p2i(hr->next_top_at_mark_start()),
1666 _calc_cl.region_marked_bytes(), hr->next_marked_bytes());
1667 }
1668
1669 _failures += failures;
1670
1671 // We could stop iteration over the heap when we
1672 // find the first violating region by returning true.
1673 return false;
1674 }
1675 };
1676
1677 class G1ParVerifyFinalCountTask: public AbstractGangTask {
1678 protected:
1679 G1CollectedHeap* _g1h;
1680 ConcurrentMark* _cm;
1681 BitMap* _actual_region_bm;
1682 BitMap* _actual_card_bm;
1683
1684 uint _n_workers;
1685
1686 BitMap* _expected_region_bm;
1687 BitMap* _expected_card_bm;
1688
1689 int _failures;
1690 bool _verbose;
1691
1692 HeapRegionClaimer _hrclaimer;
1693
1694 public:
1695 G1ParVerifyFinalCountTask(G1CollectedHeap* g1h,
1696 BitMap* region_bm, BitMap* card_bm,
1697 BitMap* expected_region_bm, BitMap* expected_card_bm)
1698 : AbstractGangTask("G1 verify final counting"),
1699 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1700 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1701 _expected_region_bm(expected_region_bm), _expected_card_bm(expected_card_bm),
1702 _failures(0), _verbose(false),
1703 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1704 assert(VerifyDuringGC, "don't call this otherwise");
1705 assert(_expected_card_bm->size() == _actual_card_bm->size(), "sanity");
1706 assert(_expected_region_bm->size() == _actual_region_bm->size(), "sanity");
1707
1708 _verbose = _cm->verbose_medium();
1709 }
1710
1711 void work(uint worker_id) {
1712 assert(worker_id < _n_workers, "invariant");
1713
1714 VerifyLiveObjectDataHRClosure verify_cl(_g1h,
1715 _actual_region_bm, _actual_card_bm,
1716 _expected_region_bm,
1717 _expected_card_bm,
1718 _verbose);
1719
1720 _g1h->heap_region_par_iterate(&verify_cl, worker_id, &_hrclaimer);
1721
1722 Atomic::add(verify_cl.failures(), &_failures);
1723 }
1724
1725 int failures() const { return _failures; }
1726 };
1727
1728 // Closure that finalizes the liveness counting data.
1729 // Used during the cleanup pause.
1730 // Sets the bits corresponding to the interval [NTAMS, top]
1731 // (which contains the implicitly live objects) in the
1732 // card liveness bitmap. Also sets the bit for each region,
1733 // containing live data, in the region liveness bitmap.
1734
1735 class FinalCountDataUpdateClosure: public CMCountDataClosureBase {
1736 public:
1737 FinalCountDataUpdateClosure(G1CollectedHeap* g1h,
1738 BitMap* region_bm,
1739 BitMap* card_bm) :
1740 CMCountDataClosureBase(g1h, region_bm, card_bm) { }
1741
1742 bool doHeapRegion(HeapRegion* hr) {
1743
1744 if (hr->is_continues_humongous()) {
1745 // We will ignore these here and process them when their
1746 // associated "starts humongous" region is processed (see
1747 // set_bit_for_heap_region()). Note that we cannot rely on their
1748 // associated "starts humongous" region to have their bit set to
1749 // 1 since, due to the region chunking in the parallel region
1750 // iteration, a "continues humongous" region might be visited
1751 // before its associated "starts humongous".
1752 return false;
1753 }
1754
1755 HeapWord* ntams = hr->next_top_at_mark_start();
1756 HeapWord* top = hr->top();
1757
1758 assert(hr->bottom() <= ntams && ntams <= hr->end(), "Preconditions.");
1759
1760 // Mark the allocated-since-marking portion...
1761 if (ntams < top) {
1762 // This definitely means the region has live objects.
1763 set_bit_for_region(hr);
1764
1765 // Now set the bits in the card bitmap for [ntams, top)
1766 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
1767 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
1768
1769 // Note: if we're looking at the last region in heap - top
1770 // could be actually just beyond the end of the heap; end_idx
1771 // will then correspond to a (non-existent) card that is also
1772 // just beyond the heap.
1773 if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
1774 // end of object is not card aligned - increment to cover
1775 // all the cards spanned by the object
1776 end_idx += 1;
1777 }
1778
1779 assert(end_idx <= _card_bm->size(),
1780 err_msg("oob: end_idx= "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
1781 end_idx, _card_bm->size()));
1782 assert(start_idx < _card_bm->size(),
1783 err_msg("oob: start_idx= "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
1784 start_idx, _card_bm->size()));
1785
1786 _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
1787 }
1788
1789 // Set the bit for the region if it contains live data
1790 if (hr->next_marked_bytes() > 0) {
1791 set_bit_for_region(hr);
1792 }
1793
1794 return false;
1795 }
1796 };
1797
1798 class G1ParFinalCountTask: public AbstractGangTask {
1799 protected:
1800 G1CollectedHeap* _g1h;
1801 ConcurrentMark* _cm;
1802 BitMap* _actual_region_bm;
1803 BitMap* _actual_card_bm;
1804
1805 uint _n_workers;
1806 HeapRegionClaimer _hrclaimer;
1807
1808 public:
1809 G1ParFinalCountTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm)
1810 : AbstractGangTask("G1 final counting"),
1811 _g1h(g1h), _cm(_g1h->concurrent_mark()),
1812 _actual_region_bm(region_bm), _actual_card_bm(card_bm),
1813 _n_workers(_g1h->workers()->active_workers()), _hrclaimer(_n_workers) {
1814 }
1815
1816 void work(uint worker_id) {
1817 assert(worker_id < _n_workers, "invariant");
1818
1819 FinalCountDataUpdateClosure final_update_cl(_g1h,
1820 _actual_region_bm,
1821 _actual_card_bm);
1822
1823 _g1h->heap_region_par_iterate(&final_update_cl, worker_id, &_hrclaimer);
1824 }
1825 };
1826
1827 class G1ParNoteEndTask;
1828
1829 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure {
1830 G1CollectedHeap* _g1;
1831 size_t _max_live_bytes;
1832 uint _regions_claimed;
1833 size_t _freed_bytes;
1834 FreeRegionList* _local_cleanup_list;
1835 HeapRegionSetCount _old_regions_removed;
1836 HeapRegionSetCount _humongous_regions_removed;
1837 HRRSCleanupTask* _hrrs_cleanup_task;
1838 double _claimed_region_time;
1839 double _max_region_time;
1840
1841 public:
1842 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1,
1843 FreeRegionList* local_cleanup_list,
1844 HRRSCleanupTask* hrrs_cleanup_task) :
1845 _g1(g1),
1846 _max_live_bytes(0), _regions_claimed(0),
1847 _freed_bytes(0),
1848 _claimed_region_time(0.0), _max_region_time(0.0),
1849 _local_cleanup_list(local_cleanup_list),
1850 _old_regions_removed(),
1851 _humongous_regions_removed(),
1852 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1853
1854 size_t freed_bytes() { return _freed_bytes; }
1855 const HeapRegionSetCount& old_regions_removed() { return _old_regions_removed; }
1856 const HeapRegionSetCount& humongous_regions_removed() { return _humongous_regions_removed; }
1857
1858 bool doHeapRegion(HeapRegion *hr) {
1859 if (hr->is_continues_humongous()) {
1860 return false;
1861 }
1862 // We use a claim value of zero here because all regions
1863 // were claimed with value 1 in the FinalCount task.
1864 _g1->reset_gc_time_stamps(hr);
1865 double start = os::elapsedTime();
1866 _regions_claimed++;
1867 hr->note_end_of_marking();
1868 _max_live_bytes += hr->max_live_bytes();
1869
1870 if (hr->used() > 0 && hr->max_live_bytes() == 0 && !hr->is_young()) {
1871 _freed_bytes += hr->used();
1872 hr->set_containing_set(NULL);
1873 if (hr->is_humongous()) {
1874 assert(hr->is_starts_humongous(), "we should only see starts humongous");
1875 _humongous_regions_removed.increment(1u, hr->capacity());
1876 _g1->free_humongous_region(hr, _local_cleanup_list, true);
1877 } else {
1878 _old_regions_removed.increment(1u, hr->capacity());
1879 _g1->free_region(hr, _local_cleanup_list, true);
1880 }
1881 } else {
1882 hr->rem_set()->do_cleanup_work(_hrrs_cleanup_task);
1883 }
1884
1885 double region_time = (os::elapsedTime() - start);
1886 _claimed_region_time += region_time;
1887 if (region_time > _max_region_time) {
1888 _max_region_time = region_time;
1889 }
1890 return false;
1891 }
1892
1893 size_t max_live_bytes() { return _max_live_bytes; }
1894 uint regions_claimed() { return _regions_claimed; }
1895 double claimed_region_time_sec() { return _claimed_region_time; }
1896 double max_region_time_sec() { return _max_region_time; }
1897 };
1898
1899 class G1ParNoteEndTask: public AbstractGangTask {
1900 friend class G1NoteEndOfConcMarkClosure;
1901
1902 protected:
1903 G1CollectedHeap* _g1h;
1904 size_t _max_live_bytes;
1905 size_t _freed_bytes;
1906 FreeRegionList* _cleanup_list;
1907 HeapRegionClaimer _hrclaimer;
1908
1909 public:
1910 G1ParNoteEndTask(G1CollectedHeap* g1h, FreeRegionList* cleanup_list, uint n_workers) :
1911 AbstractGangTask("G1 note end"), _g1h(g1h), _max_live_bytes(0), _freed_bytes(0), _cleanup_list(cleanup_list), _hrclaimer(n_workers) {
1912 }
1913
1914 void work(uint worker_id) {
1915 FreeRegionList local_cleanup_list("Local Cleanup List");
1916 HRRSCleanupTask hrrs_cleanup_task;
1917 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, &local_cleanup_list,
1918 &hrrs_cleanup_task);
1919 _g1h->heap_region_par_iterate(&g1_note_end, worker_id, &_hrclaimer);
1920 assert(g1_note_end.complete(), "Shouldn't have yielded!");
1921
1922 // Now update the lists
1923 _g1h->remove_from_old_sets(g1_note_end.old_regions_removed(), g1_note_end.humongous_regions_removed());
1924 {
1925 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1926 _g1h->decrement_summary_bytes(g1_note_end.freed_bytes());
1927 _max_live_bytes += g1_note_end.max_live_bytes();
1928 _freed_bytes += g1_note_end.freed_bytes();
1929
1930 // If we iterate over the global cleanup list at the end of
1931 // cleanup to do this printing we will not guarantee to only
1932 // generate output for the newly-reclaimed regions (the list
1933 // might not be empty at the beginning of cleanup; we might
1934 // still be working on its previous contents). So we do the
1935 // printing here, before we append the new regions to the global
1936 // cleanup list.
1937
1938 G1HRPrinter* hr_printer = _g1h->hr_printer();
1939 if (hr_printer->is_active()) {
1940 FreeRegionListIterator iter(&local_cleanup_list);
1941 while (iter.more_available()) {
1942 HeapRegion* hr = iter.get_next();
1943 hr_printer->cleanup(hr);
1944 }
1945 }
1946
1947 _cleanup_list->add_ordered(&local_cleanup_list);
1948 assert(local_cleanup_list.is_empty(), "post-condition");
1949
1950 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1951 }
1952 }
1953 size_t max_live_bytes() { return _max_live_bytes; }
1954 size_t freed_bytes() { return _freed_bytes; }
1955 };
1956
1957 class G1ParScrubRemSetTask: public AbstractGangTask {
1958 protected:
1959 G1RemSet* _g1rs;
1960 BitMap* _region_bm;
1961 BitMap* _card_bm;
1962 HeapRegionClaimer _hrclaimer;
1963
1964 public:
1965 G1ParScrubRemSetTask(G1CollectedHeap* g1h, BitMap* region_bm, BitMap* card_bm, uint n_workers) :
1966 AbstractGangTask("G1 ScrubRS"), _g1rs(g1h->g1_rem_set()), _region_bm(region_bm), _card_bm(card_bm), _hrclaimer(n_workers) {
1967 }
1968
1969 void work(uint worker_id) {
1970 _g1rs->scrub(_region_bm, _card_bm, worker_id, &_hrclaimer);
1971 }
1972
1973 };
1974
1975 void ConcurrentMark::cleanup() {
1976 // world is stopped at this checkpoint
1977 assert(SafepointSynchronize::is_at_safepoint(),
1978 "world should be stopped");
1979 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1980
1981 // If a full collection has happened, we shouldn't do this.
1982 if (has_aborted()) {
1983 g1h->set_marking_complete(); // So bitmap clearing isn't confused
1984 return;
1985 }
1986
1987 g1h->verify_region_sets_optional();
1988
1989 if (VerifyDuringGC) {
1990 HandleMark hm; // handle scope
1991 g1h->prepare_for_verify();
1992 Universe::verify(VerifyOption_G1UsePrevMarking,
1993 " VerifyDuringGC:(before)");
1994 }
1995 g1h->check_bitmaps("Cleanup Start");
1996
1997 G1CollectorPolicy* g1p = g1h->g1_policy();
1998 g1p->record_concurrent_mark_cleanup_start();
1999
2000 double start = os::elapsedTime();
2001
2002 HeapRegionRemSet::reset_for_cleanup_tasks();
2003
2004 uint n_workers;
2005
2006 // Do counting once more with the world stopped for good measure.
2007 G1ParFinalCountTask g1_par_count_task(g1h, &_region_bm, &_card_bm);
2008
2009 g1h->set_par_threads();
2010 n_workers = g1h->n_par_threads();
2011 assert(g1h->n_par_threads() == n_workers,
2012 "Should not have been reset");
2013 g1h->workers()->run_task(&g1_par_count_task);
2014 // Done with the parallel phase so reset to 0.
2015 g1h->set_par_threads(0);
2016
2017 if (VerifyDuringGC) {
2018 // Verify that the counting data accumulated during marking matches
2019 // that calculated by walking the marking bitmap.
2020
2021 // Bitmaps to hold expected values
2022 BitMap expected_region_bm(_region_bm.size(), true);
2023 BitMap expected_card_bm(_card_bm.size(), true);
2024
2025 G1ParVerifyFinalCountTask g1_par_verify_task(g1h,
2026 &_region_bm,
2027 &_card_bm,
2028 &expected_region_bm,
2029 &expected_card_bm);
2030
2031 g1h->set_par_threads((int)n_workers);
2032 g1h->workers()->run_task(&g1_par_verify_task);
2033 // Done with the parallel phase so reset to 0.
2034 g1h->set_par_threads(0);
2035
2036 guarantee(g1_par_verify_task.failures() == 0, "Unexpected accounting failures");
2037 }
2038
2039 size_t start_used_bytes = g1h->used();
2040 g1h->set_marking_complete();
2041
2042 double count_end = os::elapsedTime();
2043 double this_final_counting_time = (count_end - start);
2044 _total_counting_time += this_final_counting_time;
2045
2046 if (G1PrintRegionLivenessInfo) {
2047 G1PrintRegionLivenessInfoClosure cl(gclog_or_tty, "Post-Marking");
2048 _g1h->heap_region_iterate(&cl);
2049 }
2050
2051 // Install newly created mark bitMap as "prev".
2052 swapMarkBitMaps();
2053
2054 g1h->reset_gc_time_stamp();
2055
2056 // Note end of marking in all heap regions.
2057 G1ParNoteEndTask g1_par_note_end_task(g1h, &_cleanup_list, n_workers);
2058 g1h->set_par_threads((int)n_workers);
2059 g1h->workers()->run_task(&g1_par_note_end_task);
2060 g1h->set_par_threads(0);
2061 g1h->check_gc_time_stamps();
2062
2063 if (!cleanup_list_is_empty()) {
2064 // The cleanup list is not empty, so we'll have to process it
2065 // concurrently. Notify anyone else that might be wanting free
2066 // regions that there will be more free regions coming soon.
2067 g1h->set_free_regions_coming();
2068 }
2069
2070 // call below, since it affects the metric by which we sort the heap
2071 // regions.
2072 if (G1ScrubRemSets) {
2073 double rs_scrub_start = os::elapsedTime();
2074 G1ParScrubRemSetTask g1_par_scrub_rs_task(g1h, &_region_bm, &_card_bm, n_workers);
2075 g1h->set_par_threads((int)n_workers);
2076 g1h->workers()->run_task(&g1_par_scrub_rs_task);
2077 g1h->set_par_threads(0);
2078
2079 double rs_scrub_end = os::elapsedTime();
2080 double this_rs_scrub_time = (rs_scrub_end - rs_scrub_start);
2081 _total_rs_scrub_time += this_rs_scrub_time;
2082 }
2083
2084 // this will also free any regions totally full of garbage objects,
2085 // and sort the regions.
2086 g1h->g1_policy()->record_concurrent_mark_cleanup_end((int)n_workers);
2087
2088 // Statistics.
2089 double end = os::elapsedTime();
2090 _cleanup_times.add((end - start) * 1000.0);
2091
2092 if (G1Log::fine()) {
2093 g1h->g1_policy()->print_heap_transition(start_used_bytes);
2094 }
2095
2096 // Clean up will have freed any regions completely full of garbage.
2097 // Update the soft reference policy with the new heap occupancy.
2098 Universe::update_heap_info_at_gc();
2099
2100 if (VerifyDuringGC) {
2101 HandleMark hm; // handle scope
2102 g1h->prepare_for_verify();
2103 Universe::verify(VerifyOption_G1UsePrevMarking,
2104 " VerifyDuringGC:(after)");
2105 }
2106
2107 g1h->check_bitmaps("Cleanup End");
2108
2109 g1h->verify_region_sets_optional();
2110
2111 // We need to make this be a "collection" so any collection pause that
2112 // races with it goes around and waits for completeCleanup to finish.
2113 g1h->increment_total_collections();
2114
2115 // Clean out dead classes and update Metaspace sizes.
2116 if (ClassUnloadingWithConcurrentMark) {
2117 ClassLoaderDataGraph::purge();
2118 }
2119 MetaspaceGC::compute_new_size();
2120
2121 // We reclaimed old regions so we should calculate the sizes to make
2122 // sure we update the old gen/space data.
2123 g1h->g1mm()->update_sizes();
2124 g1h->allocation_context_stats().update_after_mark();
2125
2126 g1h->trace_heap_after_concurrent_cycle();
2127 }
2128
2129 void ConcurrentMark::completeCleanup() {
2130 if (has_aborted()) return;
2131
2132 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2133
2134 _cleanup_list.verify_optional();
2135 FreeRegionList tmp_free_list("Tmp Free List");
2136
2137 if (G1ConcRegionFreeingVerbose) {
2138 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
2139 "cleanup list has %u entries",
2140 _cleanup_list.length());
2141 }
2142
2143 // No one else should be accessing the _cleanup_list at this point,
2144 // so it is not necessary to take any locks
2145 while (!_cleanup_list.is_empty()) {
2146 HeapRegion* hr = _cleanup_list.remove_region(true /* from_head */);
2147 assert(hr != NULL, "Got NULL from a non-empty list");
2148 hr->par_clear();
2149 tmp_free_list.add_ordered(hr);
2150
2151 // Instead of adding one region at a time to the secondary_free_list,
2152 // we accumulate them in the local list and move them a few at a
2153 // time. This also cuts down on the number of notify_all() calls
2154 // we do during this process. We'll also append the local list when
2155 // _cleanup_list is empty (which means we just removed the last
2156 // region from the _cleanup_list).
2157 if ((tmp_free_list.length() % G1SecondaryFreeListAppendLength == 0) ||
2158 _cleanup_list.is_empty()) {
2159 if (G1ConcRegionFreeingVerbose) {
2160 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
2161 "appending %u entries to the secondary_free_list, "
2162 "cleanup list still has %u entries",
2163 tmp_free_list.length(),
2164 _cleanup_list.length());
2165 }
2166
2167 {
2168 MutexLockerEx x(SecondaryFreeList_lock, Mutex::_no_safepoint_check_flag);
2169 g1h->secondary_free_list_add(&tmp_free_list);
2170 SecondaryFreeList_lock->notify_all();
2171 }
2172 #ifndef PRODUCT
2173 if (G1StressConcRegionFreeing) {
2174 for (uintx i = 0; i < G1StressConcRegionFreeingDelayMillis; ++i) {
2175 os::sleep(Thread::current(), (jlong) 1, false);
2176 }
2177 }
2178 #endif
2179 }
2180 }
2181 assert(tmp_free_list.is_empty(), "post-condition");
2182 }
2183
2184 // Supporting Object and Oop closures for reference discovery
2185 // and processing in during marking
2186
2187 bool G1CMIsAliveClosure::do_object_b(oop obj) {
2188 HeapWord* addr = (HeapWord*)obj;
2189 return addr != NULL &&
2190 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj));
2191 }
2192
2193 // 'Keep Alive' oop closure used by both serial parallel reference processing.
2194 // Uses the CMTask associated with a worker thread (for serial reference
2195 // processing the CMTask for worker 0 is used) to preserve (mark) and
2196 // trace referent objects.
2197 //
2198 // Using the CMTask and embedded local queues avoids having the worker
2199 // threads operating on the global mark stack. This reduces the risk
2200 // of overflowing the stack - which we would rather avoid at this late
2201 // state. Also using the tasks' local queues removes the potential
2202 // of the workers interfering with each other that could occur if
2203 // operating on the global stack.
2204
2205 class G1CMKeepAliveAndDrainClosure: public OopClosure {
2206 ConcurrentMark* _cm;
2207 CMTask* _task;
2208 int _ref_counter_limit;
2209 int _ref_counter;
2210 bool _is_serial;
2211 public:
2212 G1CMKeepAliveAndDrainClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
2213 _cm(cm), _task(task), _is_serial(is_serial),
2214 _ref_counter_limit(G1RefProcDrainInterval) {
2215 assert(_ref_counter_limit > 0, "sanity");
2216 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
2217 _ref_counter = _ref_counter_limit;
2218 }
2219
2220 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
2221 virtual void do_oop( oop* p) { do_oop_work(p); }
2222
2223 template <class T> void do_oop_work(T* p) {
2224 if (!_cm->has_overflown()) {
2225 oop obj = oopDesc::load_decode_heap_oop(p);
2226 if (_cm->verbose_high()) {
2227 gclog_or_tty->print_cr("\t[%u] we're looking at location "
2228 "*"PTR_FORMAT" = "PTR_FORMAT,
2229 _task->worker_id(), p2i(p), p2i((void*) obj));
2230 }
2231
2232 _task->deal_with_reference(obj);
2233 _ref_counter--;
2234
2235 if (_ref_counter == 0) {
2236 // We have dealt with _ref_counter_limit references, pushing them
2237 // and objects reachable from them on to the local stack (and
2238 // possibly the global stack). Call CMTask::do_marking_step() to
2239 // process these entries.
2240 //
2241 // We call CMTask::do_marking_step() in a loop, which we'll exit if
2242 // there's nothing more to do (i.e. we're done with the entries that
2243 // were pushed as a result of the CMTask::deal_with_reference() calls
2244 // above) or we overflow.
2245 //
2246 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
2247 // flag while there may still be some work to do. (See the comment at
2248 // the beginning of CMTask::do_marking_step() for those conditions -
2249 // one of which is reaching the specified time target.) It is only
2250 // when CMTask::do_marking_step() returns without setting the
2251 // has_aborted() flag that the marking step has completed.
2252 do {
2253 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
2254 _task->do_marking_step(mark_step_duration_ms,
2255 false /* do_termination */,
2256 _is_serial);
2257 } while (_task->has_aborted() && !_cm->has_overflown());
2258 _ref_counter = _ref_counter_limit;
2259 }
2260 } else {
2261 if (_cm->verbose_high()) {
2262 gclog_or_tty->print_cr("\t[%u] CM Overflow", _task->worker_id());
2263 }
2264 }
2265 }
2266 };
2267
2268 // 'Drain' oop closure used by both serial and parallel reference processing.
2269 // Uses the CMTask associated with a given worker thread (for serial
2270 // reference processing the CMtask for worker 0 is used). Calls the
2271 // do_marking_step routine, with an unbelievably large timeout value,
2272 // to drain the marking data structures of the remaining entries
2273 // added by the 'keep alive' oop closure above.
2274
2275 class G1CMDrainMarkingStackClosure: public VoidClosure {
2276 ConcurrentMark* _cm;
2277 CMTask* _task;
2278 bool _is_serial;
2279 public:
2280 G1CMDrainMarkingStackClosure(ConcurrentMark* cm, CMTask* task, bool is_serial) :
2281 _cm(cm), _task(task), _is_serial(is_serial) {
2282 assert(!_is_serial || _task->worker_id() == 0, "only task 0 for serial code");
2283 }
2284
2285 void do_void() {
2286 do {
2287 if (_cm->verbose_high()) {
2288 gclog_or_tty->print_cr("\t[%u] Drain: Calling do_marking_step - serial: %s",
2289 _task->worker_id(), BOOL_TO_STR(_is_serial));
2290 }
2291
2292 // We call CMTask::do_marking_step() to completely drain the local
2293 // and global marking stacks of entries pushed by the 'keep alive'
2294 // oop closure (an instance of G1CMKeepAliveAndDrainClosure above).
2295 //
2296 // CMTask::do_marking_step() is called in a loop, which we'll exit
2297 // if there's nothing more to do (i.e. we've completely drained the
2298 // entries that were pushed as a a result of applying the 'keep alive'
2299 // closure to the entries on the discovered ref lists) or we overflow
2300 // the global marking stack.
2301 //
2302 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted()
2303 // flag while there may still be some work to do. (See the comment at
2304 // the beginning of CMTask::do_marking_step() for those conditions -
2305 // one of which is reaching the specified time target.) It is only
2306 // when CMTask::do_marking_step() returns without setting the
2307 // has_aborted() flag that the marking step has completed.
2308
2309 _task->do_marking_step(1000000000.0 /* something very large */,
2310 true /* do_termination */,
2311 _is_serial);
2312 } while (_task->has_aborted() && !_cm->has_overflown());
2313 }
2314 };
2315
2316 // Implementation of AbstractRefProcTaskExecutor for parallel
2317 // reference processing at the end of G1 concurrent marking
2318
2319 class G1CMRefProcTaskExecutor: public AbstractRefProcTaskExecutor {
2320 private:
2321 G1CollectedHeap* _g1h;
2322 ConcurrentMark* _cm;
2323 WorkGang* _workers;
2324 int _active_workers;
2325
2326 public:
2327 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
2328 ConcurrentMark* cm,
2329 WorkGang* workers,
2330 int n_workers) :
2331 _g1h(g1h), _cm(cm),
2332 _workers(workers), _active_workers(n_workers) { }
2333
2334 // Executes the given task using concurrent marking worker threads.
2335 virtual void execute(ProcessTask& task);
2336 virtual void execute(EnqueueTask& task);
2337 };
2338
2339 class G1CMRefProcTaskProxy: public AbstractGangTask {
2340 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
2341 ProcessTask& _proc_task;
2342 G1CollectedHeap* _g1h;
2343 ConcurrentMark* _cm;
2344
2345 public:
2346 G1CMRefProcTaskProxy(ProcessTask& proc_task,
2347 G1CollectedHeap* g1h,
2348 ConcurrentMark* cm) :
2349 AbstractGangTask("Process reference objects in parallel"),
2350 _proc_task(proc_task), _g1h(g1h), _cm(cm) {
2351 ReferenceProcessor* rp = _g1h->ref_processor_cm();
2352 assert(rp->processing_is_mt(), "shouldn't be here otherwise");
2353 }
2354
2355 virtual void work(uint worker_id) {
2356 ResourceMark rm;
2357 HandleMark hm;
2358 CMTask* task = _cm->task(worker_id);
2359 G1CMIsAliveClosure g1_is_alive(_g1h);
2360 G1CMKeepAliveAndDrainClosure g1_par_keep_alive(_cm, task, false /* is_serial */);
2361 G1CMDrainMarkingStackClosure g1_par_drain(_cm, task, false /* is_serial */);
2362
2363 _proc_task.work(worker_id, g1_is_alive, g1_par_keep_alive, g1_par_drain);
2364 }
2365 };
2366
2367 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
2368 assert(_workers != NULL, "Need parallel worker threads.");
2369 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2370
2371 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
2372
2373 // We need to reset the concurrency level before each
2374 // proxy task execution, so that the termination protocol
2375 // and overflow handling in CMTask::do_marking_step() knows
2376 // how many workers to wait for.
2377 _cm->set_concurrency(_active_workers);
2378 _g1h->set_par_threads(_active_workers);
2379 _workers->run_task(&proc_task_proxy);
2380 _g1h->set_par_threads(0);
2381 }
2382
2383 class G1CMRefEnqueueTaskProxy: public AbstractGangTask {
2384 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
2385 EnqueueTask& _enq_task;
2386
2387 public:
2388 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
2389 AbstractGangTask("Enqueue reference objects in parallel"),
2390 _enq_task(enq_task) { }
2391
2392 virtual void work(uint worker_id) {
2393 _enq_task.work(worker_id);
2394 }
2395 };
2396
2397 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
2398 assert(_workers != NULL, "Need parallel worker threads.");
2399 assert(_g1h->ref_processor_cm()->processing_is_mt(), "processing is not MT");
2400
2401 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
2402
2403 // Not strictly necessary but...
2404 //
2405 // We need to reset the concurrency level before each
2406 // proxy task execution, so that the termination protocol
2407 // and overflow handling in CMTask::do_marking_step() knows
2408 // how many workers to wait for.
2409 _cm->set_concurrency(_active_workers);
2410 _g1h->set_par_threads(_active_workers);
2411 _workers->run_task(&enq_task_proxy);
2412 _g1h->set_par_threads(0);
2413 }
2414
2415 void ConcurrentMark::weakRefsWorkParallelPart(BoolObjectClosure* is_alive, bool purged_classes) {
2416 G1CollectedHeap::heap()->parallel_cleaning(is_alive, true, true, purged_classes);
2417 }
2418
2419 void ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) {
2420 if (has_overflown()) {
2421 // Skip processing the discovered references if we have
2422 // overflown the global marking stack. Reference objects
2423 // only get discovered once so it is OK to not
2424 // de-populate the discovered reference lists. We could have,
2425 // but the only benefit would be that, when marking restarts,
2426 // less reference objects are discovered.
2427 return;
2428 }
2429
2430 ResourceMark rm;
2431 HandleMark hm;
2432
2433 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2434
2435 // Is alive closure.
2436 G1CMIsAliveClosure g1_is_alive(g1h);
2437
2438 // Inner scope to exclude the cleaning of the string and symbol
2439 // tables from the displayed time.
2440 {
2441 G1CMTraceTime t("GC ref-proc", G1Log::finer());
2442
2443 ReferenceProcessor* rp = g1h->ref_processor_cm();
2444
2445 // See the comment in G1CollectedHeap::ref_processing_init()
2446 // about how reference processing currently works in G1.
2447
2448 // Set the soft reference policy
2449 rp->setup_policy(clear_all_soft_refs);
2450 assert(_markStack.isEmpty(), "mark stack should be empty");
2451
2452 // Instances of the 'Keep Alive' and 'Complete GC' closures used
2453 // in serial reference processing. Note these closures are also
2454 // used for serially processing (by the the current thread) the
2455 // JNI references during parallel reference processing.
2456 //
2457 // These closures do not need to synchronize with the worker
2458 // threads involved in parallel reference processing as these
2459 // instances are executed serially by the current thread (e.g.
2460 // reference processing is not multi-threaded and is thus
2461 // performed by the current thread instead of a gang worker).
2462 //
2463 // The gang tasks involved in parallel reference processing create
2464 // their own instances of these closures, which do their own
2465 // synchronization among themselves.
2466 G1CMKeepAliveAndDrainClosure g1_keep_alive(this, task(0), true /* is_serial */);
2467 G1CMDrainMarkingStackClosure g1_drain_mark_stack(this, task(0), true /* is_serial */);
2468
2469 // We need at least one active thread. If reference processing
2470 // is not multi-threaded we use the current (VMThread) thread,
2471 // otherwise we use the work gang from the G1CollectedHeap and
2472 // we utilize all the worker threads we can.
2473 bool processing_is_mt = rp->processing_is_mt();
2474 uint active_workers = (processing_is_mt ? g1h->workers()->active_workers() : 1U);
2475 active_workers = MAX2(MIN2(active_workers, _max_worker_id), 1U);
2476
2477 // Parallel processing task executor.
2478 G1CMRefProcTaskExecutor par_task_executor(g1h, this,
2479 g1h->workers(), active_workers);
2480 AbstractRefProcTaskExecutor* executor = (processing_is_mt ? &par_task_executor : NULL);
2481
2482 // Set the concurrency level. The phase was already set prior to
2483 // executing the remark task.
2484 set_concurrency(active_workers);
2485
2486 // Set the degree of MT processing here. If the discovery was done MT,
2487 // the number of threads involved during discovery could differ from
2488 // the number of active workers. This is OK as long as the discovered
2489 // Reference lists are balanced (see balance_all_queues() and balance_queues()).
2490 rp->set_active_mt_degree(active_workers);
2491
2492 // Process the weak references.
2493 const ReferenceProcessorStats& stats =
2494 rp->process_discovered_references(&g1_is_alive,
2495 &g1_keep_alive,
2496 &g1_drain_mark_stack,
2497 executor,
2498 g1h->gc_timer_cm(),
2499 concurrent_gc_id());
2500 g1h->gc_tracer_cm()->report_gc_reference_stats(stats);
2501
2502 // The do_oop work routines of the keep_alive and drain_marking_stack
2503 // oop closures will set the has_overflown flag if we overflow the
2504 // global marking stack.
2505
2506 assert(_markStack.overflow() || _markStack.isEmpty(),
2507 "mark stack should be empty (unless it overflowed)");
2508
2509 if (_markStack.overflow()) {
2510 // This should have been done already when we tried to push an
2511 // entry on to the global mark stack. But let's do it again.
2512 set_has_overflown();
2513 }
2514
2515 assert(rp->num_q() == active_workers, "why not");
2516
2517 rp->enqueue_discovered_references(executor);
2518
2519 rp->verify_no_references_recorded();
2520 assert(!rp->discovery_enabled(), "Post condition");
2521 }
2522
2523 if (has_overflown()) {
2524 // We can not trust g1_is_alive if the marking stack overflowed
2525 return;
2526 }
2527
2528 assert(_markStack.isEmpty(), "Marking should have completed");
2529
2530 // Unload Klasses, String, Symbols, Code Cache, etc.
2531 {
2532 G1CMTraceTime trace("Unloading", G1Log::finer());
2533
2534 if (ClassUnloadingWithConcurrentMark) {
2535 bool purged_classes;
2536
2537 {
2538 G1CMTraceTime trace("System Dictionary Unloading", G1Log::finest());
2539 purged_classes = SystemDictionary::do_unloading(&g1_is_alive, false /* Defer klass cleaning */);
2540 }
2541
2542 {
2543 G1CMTraceTime trace("Parallel Unloading", G1Log::finest());
2544 weakRefsWorkParallelPart(&g1_is_alive, purged_classes);
2545 }
2546 }
2547
2548 if (G1StringDedup::is_enabled()) {
2549 G1CMTraceTime trace("String Deduplication Unlink", G1Log::finest());
2550 G1StringDedup::unlink(&g1_is_alive);
2551 }
2552 }
2553 }
2554
2555 void ConcurrentMark::swapMarkBitMaps() {
2556 CMBitMapRO* temp = _prevMarkBitMap;
2557 _prevMarkBitMap = (CMBitMapRO*)_nextMarkBitMap;
2558 _nextMarkBitMap = (CMBitMap*) temp;
2559 }
2560
2561 class CMObjectClosure;
2562
2563 // Closure for iterating over objects, currently only used for
2564 // processing SATB buffers.
2565 class CMObjectClosure : public ObjectClosure {
2566 private:
2567 CMTask* _task;
2568
2569 public:
2570 void do_object(oop obj) {
2571 _task->deal_with_reference(obj);
2572 }
2573
2574 CMObjectClosure(CMTask* task) : _task(task) { }
2575 };
2576
2577 class G1RemarkThreadsClosure : public ThreadClosure {
2578 CMObjectClosure _cm_obj;
2579 G1CMOopClosure _cm_cl;
2580 MarkingCodeBlobClosure _code_cl;
2581 int _thread_parity;
2582
2583 public:
2584 G1RemarkThreadsClosure(G1CollectedHeap* g1h, CMTask* task) :
2585 _cm_obj(task), _cm_cl(g1h, g1h->concurrent_mark(), task), _code_cl(&_cm_cl, !CodeBlobToOopClosure::FixRelocations),
2586 _thread_parity(Threads::thread_claim_parity()) {}
2587
2588 void do_thread(Thread* thread) {
2589 if (thread->is_Java_thread()) {
2590 if (thread->claim_oops_do(true, _thread_parity)) {
2591 JavaThread* jt = (JavaThread*)thread;
2592
2593 // In theory it should not be neccessary to explicitly walk the nmethods to find roots for concurrent marking
2594 // however the liveness of oops reachable from nmethods have very complex lifecycles:
2595 // * Alive if on the stack of an executing method
2596 // * Weakly reachable otherwise
2597 // Some objects reachable from nmethods, such as the class loader (or klass_holder) of the receiver should be
2598 // live by the SATB invariant but other oops recorded in nmethods may behave differently.
2599 jt->nmethods_do(&_code_cl);
2600
2601 jt->satb_mark_queue().apply_closure_and_empty(&_cm_obj);
2602 }
2603 } else if (thread->is_VM_thread()) {
2604 if (thread->claim_oops_do(true, _thread_parity)) {
2605 JavaThread::satb_mark_queue_set().shared_satb_queue()->apply_closure_and_empty(&_cm_obj);
2606 }
2607 }
2608 }
2609 };
2610
2611 class CMRemarkTask: public AbstractGangTask {
2612 private:
2613 ConcurrentMark* _cm;
2614 public:
2615 void work(uint worker_id) {
2616 // Since all available tasks are actually started, we should
2617 // only proceed if we're supposed to be active.
2618 if (worker_id < _cm->active_tasks()) {
2619 CMTask* task = _cm->task(worker_id);
2620 task->record_start_time();
2621 {
2622 ResourceMark rm;
2623 HandleMark hm;
2624
2625 G1RemarkThreadsClosure threads_f(G1CollectedHeap::heap(), task);
2626 Threads::threads_do(&threads_f);
2627 }
2628
2629 do {
2630 task->do_marking_step(1000000000.0 /* something very large */,
2631 true /* do_termination */,
2632 false /* is_serial */);
2633 } while (task->has_aborted() && !_cm->has_overflown());
2634 // If we overflow, then we do not want to restart. We instead
2635 // want to abort remark and do concurrent marking again.
2636 task->record_end_time();
2637 }
2638 }
2639
2640 CMRemarkTask(ConcurrentMark* cm, int active_workers) :
2641 AbstractGangTask("Par Remark"), _cm(cm) {
2642 _cm->terminator()->reset_for_reuse(active_workers);
2643 }
2644 };
2645
2646 void ConcurrentMark::checkpointRootsFinalWork() {
2647 ResourceMark rm;
2648 HandleMark hm;
2649 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2650
2651 G1CMTraceTime trace("Finalize Marking", G1Log::finer());
2652
2653 g1h->ensure_parsability(false);
2654
2655 StrongRootsScope srs;
2656 // this is remark, so we'll use up all active threads
2657 uint active_workers = g1h->workers()->active_workers();
2658 if (active_workers == 0) {
2659 assert(active_workers > 0, "Should have been set earlier");
2660 active_workers = (uint) ParallelGCThreads;
2661 g1h->workers()->set_active_workers(active_workers);
2662 }
2663 set_concurrency_and_phase(active_workers, false /* concurrent */);
2664 // Leave _parallel_marking_threads at it's
2665 // value originally calculated in the ConcurrentMark
2666 // constructor and pass values of the active workers
2667 // through the gang in the task.
2668
2669 CMRemarkTask remarkTask(this, active_workers);
2670 // We will start all available threads, even if we decide that the
2671 // active_workers will be fewer. The extra ones will just bail out
2672 // immediately.
2673 g1h->set_par_threads(active_workers);
2674 g1h->workers()->run_task(&remarkTask);
2675 g1h->set_par_threads(0);
2676
2677 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2678 guarantee(has_overflown() ||
2679 satb_mq_set.completed_buffers_num() == 0,
2680 err_msg("Invariant: has_overflown = %s, num buffers = %d",
2681 BOOL_TO_STR(has_overflown()),
2682 satb_mq_set.completed_buffers_num()));
2683
2684 print_stats();
2685 }
2686
2687 #ifndef PRODUCT
2688
2689 class PrintReachableOopClosure: public OopClosure {
2690 private:
2691 G1CollectedHeap* _g1h;
2692 outputStream* _out;
2693 VerifyOption _vo;
2694 bool _all;
2695
2696 public:
2697 PrintReachableOopClosure(outputStream* out,
2698 VerifyOption vo,
2699 bool all) :
2700 _g1h(G1CollectedHeap::heap()),
2701 _out(out), _vo(vo), _all(all) { }
2702
2703 void do_oop(narrowOop* p) { do_oop_work(p); }
2704 void do_oop( oop* p) { do_oop_work(p); }
2705
2706 template <class T> void do_oop_work(T* p) {
2707 oop obj = oopDesc::load_decode_heap_oop(p);
2708 const char* str = NULL;
2709 const char* str2 = "";
2710
2711 if (obj == NULL) {
2712 str = "";
2713 } else if (!_g1h->is_in_g1_reserved(obj)) {
2714 str = " O";
2715 } else {
2716 HeapRegion* hr = _g1h->heap_region_containing(obj);
2717 bool over_tams = _g1h->allocated_since_marking(obj, hr, _vo);
2718 bool marked = _g1h->is_marked(obj, _vo);
2719
2720 if (over_tams) {
2721 str = " >";
2722 if (marked) {
2723 str2 = " AND MARKED";
2724 }
2725 } else if (marked) {
2726 str = " M";
2727 } else {
2728 str = " NOT";
2729 }
2730 }
2731
2732 _out->print_cr(" "PTR_FORMAT": "PTR_FORMAT"%s%s",
2733 p2i(p), p2i((void*) obj), str, str2);
2734 }
2735 };
2736
2737 class PrintReachableObjectClosure : public ObjectClosure {
2738 private:
2739 G1CollectedHeap* _g1h;
2740 outputStream* _out;
2741 VerifyOption _vo;
2742 bool _all;
2743 HeapRegion* _hr;
2744
2745 public:
2746 PrintReachableObjectClosure(outputStream* out,
2747 VerifyOption vo,
2748 bool all,
2749 HeapRegion* hr) :
2750 _g1h(G1CollectedHeap::heap()),
2751 _out(out), _vo(vo), _all(all), _hr(hr) { }
2752
2753 void do_object(oop o) {
2754 bool over_tams = _g1h->allocated_since_marking(o, _hr, _vo);
2755 bool marked = _g1h->is_marked(o, _vo);
2756 bool print_it = _all || over_tams || marked;
2757
2758 if (print_it) {
2759 _out->print_cr(" "PTR_FORMAT"%s",
2760 p2i((void *)o), (over_tams) ? " >" : (marked) ? " M" : "");
2761 PrintReachableOopClosure oopCl(_out, _vo, _all);
2762 o->oop_iterate_no_header(&oopCl);
2763 }
2764 }
2765 };
2766
2767 class PrintReachableRegionClosure : public HeapRegionClosure {
2768 private:
2769 G1CollectedHeap* _g1h;
2770 outputStream* _out;
2771 VerifyOption _vo;
2772 bool _all;
2773
2774 public:
2775 bool doHeapRegion(HeapRegion* hr) {
2776 HeapWord* b = hr->bottom();
2777 HeapWord* e = hr->end();
2778 HeapWord* t = hr->top();
2779 HeapWord* p = _g1h->top_at_mark_start(hr, _vo);
2780 _out->print_cr("** ["PTR_FORMAT", "PTR_FORMAT"] top: "PTR_FORMAT" "
2781 "TAMS: " PTR_FORMAT, p2i(b), p2i(e), p2i(t), p2i(p));
2782 _out->cr();
2783
2784 HeapWord* from = b;
2785 HeapWord* to = t;
2786
2787 if (to > from) {
2788 _out->print_cr("Objects in [" PTR_FORMAT ", " PTR_FORMAT "]", p2i(from), p2i(to));
2789 _out->cr();
2790 PrintReachableObjectClosure ocl(_out, _vo, _all, hr);
2791 hr->object_iterate_mem_careful(MemRegion(from, to), &ocl);
2792 _out->cr();
2793 }
2794
2795 return false;
2796 }
2797
2798 PrintReachableRegionClosure(outputStream* out,
2799 VerifyOption vo,
2800 bool all) :
2801 _g1h(G1CollectedHeap::heap()), _out(out), _vo(vo), _all(all) { }
2802 };
2803
2804 void ConcurrentMark::print_reachable(const char* str,
2805 VerifyOption vo,
2806 bool all) {
2807 gclog_or_tty->cr();
2808 gclog_or_tty->print_cr("== Doing heap dump... ");
2809
2810 if (G1PrintReachableBaseFile == NULL) {
2811 gclog_or_tty->print_cr(" #### error: no base file defined");
2812 return;
2813 }
2814
2815 if (strlen(G1PrintReachableBaseFile) + 1 + strlen(str) >
2816 (JVM_MAXPATHLEN - 1)) {
2817 gclog_or_tty->print_cr(" #### error: file name too long");
2818 return;
2819 }
2820
2821 char file_name[JVM_MAXPATHLEN];
2822 sprintf(file_name, "%s.%s", G1PrintReachableBaseFile, str);
2823 gclog_or_tty->print_cr(" dumping to file %s", file_name);
2824
2825 fileStream fout(file_name);
2826 if (!fout.is_open()) {
2827 gclog_or_tty->print_cr(" #### error: could not open file");
2828 return;
2829 }
2830
2831 outputStream* out = &fout;
2832 out->print_cr("-- USING %s", _g1h->top_at_mark_start_str(vo));
2833 out->cr();
2834
2835 out->print_cr("--- ITERATING OVER REGIONS");
2836 out->cr();
2837 PrintReachableRegionClosure rcl(out, vo, all);
2838 _g1h->heap_region_iterate(&rcl);
2839 out->cr();
2840
2841 gclog_or_tty->print_cr(" done");
2842 gclog_or_tty->flush();
2843 }
2844
2845 #endif // PRODUCT
2846
2847 void ConcurrentMark::clearRangePrevBitmap(MemRegion mr) {
2848 // Note we are overriding the read-only view of the prev map here, via
2849 // the cast.
2850 ((CMBitMap*)_prevMarkBitMap)->clearRange(mr);
2851 }
2852
2853 void ConcurrentMark::clearRangeNextBitmap(MemRegion mr) {
2854 _nextMarkBitMap->clearRange(mr);
2855 }
2856
2857 HeapRegion*
2858 ConcurrentMark::claim_region(uint worker_id) {
2859 // "checkpoint" the finger
2860 HeapWord* finger = _finger;
2861
2862 // _heap_end will not change underneath our feet; it only changes at
2863 // yield points.
2864 while (finger < _heap_end) {
2865 assert(_g1h->is_in_g1_reserved(finger), "invariant");
2866
2867 // Note on how this code handles humongous regions. In the
2868 // normal case the finger will reach the start of a "starts
2869 // humongous" (SH) region. Its end will either be the end of the
2870 // last "continues humongous" (CH) region in the sequence, or the
2871 // standard end of the SH region (if the SH is the only region in
2872 // the sequence). That way claim_region() will skip over the CH
2873 // regions. However, there is a subtle race between a CM thread
2874 // executing this method and a mutator thread doing a humongous
2875 // object allocation. The two are not mutually exclusive as the CM
2876 // thread does not need to hold the Heap_lock when it gets
2877 // here. So there is a chance that claim_region() will come across
2878 // a free region that's in the progress of becoming a SH or a CH
2879 // region. In the former case, it will either
2880 // a) Miss the update to the region's end, in which case it will
2881 // visit every subsequent CH region, will find their bitmaps
2882 // empty, and do nothing, or
2883 // b) Will observe the update of the region's end (in which case
2884 // it will skip the subsequent CH regions).
2885 // If it comes across a region that suddenly becomes CH, the
2886 // scenario will be similar to b). So, the race between
2887 // claim_region() and a humongous object allocation might force us
2888 // to do a bit of unnecessary work (due to some unnecessary bitmap
2889 // iterations) but it should not introduce and correctness issues.
2890 HeapRegion* curr_region = _g1h->heap_region_containing_raw(finger);
2891
2892 // Above heap_region_containing_raw may return NULL as we always scan claim
2893 // until the end of the heap. In this case, just jump to the next region.
2894 HeapWord* end = curr_region != NULL ? curr_region->end() : finger + HeapRegion::GrainWords;
2895
2896 // Is the gap between reading the finger and doing the CAS too long?
2897 HeapWord* res = (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger);
2898 if (res == finger && curr_region != NULL) {
2899 // we succeeded
2900 HeapWord* bottom = curr_region->bottom();
2901 HeapWord* limit = curr_region->next_top_at_mark_start();
2902
2903 if (verbose_low()) {
2904 gclog_or_tty->print_cr("[%u] curr_region = "PTR_FORMAT" "
2905 "["PTR_FORMAT", "PTR_FORMAT"), "
2906 "limit = "PTR_FORMAT,
2907 worker_id, p2i(curr_region), p2i(bottom), p2i(end), p2i(limit));
2908 }
2909
2910 // notice that _finger == end cannot be guaranteed here since,
2911 // someone else might have moved the finger even further
2912 assert(_finger >= end, "the finger should have moved forward");
2913
2914 if (verbose_low()) {
2915 gclog_or_tty->print_cr("[%u] we were successful with region = "
2916 PTR_FORMAT, worker_id, p2i(curr_region));
2917 }
2918
2919 if (limit > bottom) {
2920 if (verbose_low()) {
2921 gclog_or_tty->print_cr("[%u] region "PTR_FORMAT" is not empty, "
2922 "returning it ", worker_id, p2i(curr_region));
2923 }
2924 return curr_region;
2925 } else {
2926 assert(limit == bottom,
2927 "the region limit should be at bottom");
2928 if (verbose_low()) {
2929 gclog_or_tty->print_cr("[%u] region "PTR_FORMAT" is empty, "
2930 "returning NULL", worker_id, p2i(curr_region));
2931 }
2932 // we return NULL and the caller should try calling
2933 // claim_region() again.
2934 return NULL;
2935 }
2936 } else {
2937 assert(_finger > finger, "the finger should have moved forward");
2938 if (verbose_low()) {
2939 if (curr_region == NULL) {
2940 gclog_or_tty->print_cr("[%u] found uncommitted region, moving finger, "
2941 "global finger = "PTR_FORMAT", "
2942 "our finger = "PTR_FORMAT,
2943 worker_id, p2i(_finger), p2i(finger));
2944 } else {
2945 gclog_or_tty->print_cr("[%u] somebody else moved the finger, "
2946 "global finger = "PTR_FORMAT", "
2947 "our finger = "PTR_FORMAT,
2948 worker_id, p2i(_finger), p2i(finger));
2949 }
2950 }
2951
2952 // read it again
2953 finger = _finger;
2954 }
2955 }
2956
2957 return NULL;
2958 }
2959
2960 #ifndef PRODUCT
2961 enum VerifyNoCSetOopsPhase {
2962 VerifyNoCSetOopsStack,
2963 VerifyNoCSetOopsQueues,
2964 VerifyNoCSetOopsSATBCompleted,
2965 VerifyNoCSetOopsSATBThread
2966 };
2967
2968 class VerifyNoCSetOopsClosure : public OopClosure, public ObjectClosure {
2969 private:
2970 G1CollectedHeap* _g1h;
2971 VerifyNoCSetOopsPhase _phase;
2972 int _info;
2973
2974 const char* phase_str() {
2975 switch (_phase) {
2976 case VerifyNoCSetOopsStack: return "Stack";
2977 case VerifyNoCSetOopsQueues: return "Queue";
2978 case VerifyNoCSetOopsSATBCompleted: return "Completed SATB Buffers";
2979 case VerifyNoCSetOopsSATBThread: return "Thread SATB Buffers";
2980 default: ShouldNotReachHere();
2981 }
2982 return NULL;
2983 }
2984
2985 void do_object_work(oop obj) {
2986 guarantee(!_g1h->obj_in_cs(obj),
2987 err_msg("obj: "PTR_FORMAT" in CSet, phase: %s, info: %d",
2988 p2i((void*) obj), phase_str(), _info));
2989 }
2990
2991 public:
2992 VerifyNoCSetOopsClosure() : _g1h(G1CollectedHeap::heap()) { }
2993
2994 void set_phase(VerifyNoCSetOopsPhase phase, int info = -1) {
2995 _phase = phase;
2996 _info = info;
2997 }
2998
2999 virtual void do_oop(oop* p) {
3000 oop obj = oopDesc::load_decode_heap_oop(p);
3001 do_object_work(obj);
3002 }
3003
3004 virtual void do_oop(narrowOop* p) {
3005 // We should not come across narrow oops while scanning marking
3006 // stacks and SATB buffers.
3007 ShouldNotReachHere();
3008 }
3009
3010 virtual void do_object(oop obj) {
3011 do_object_work(obj);
3012 }
3013 };
3014
3015 void ConcurrentMark::verify_no_cset_oops(bool verify_stacks,
3016 bool verify_enqueued_buffers,
3017 bool verify_thread_buffers,
3018 bool verify_fingers) {
3019 assert(SafepointSynchronize::is_at_safepoint(), "should be at a safepoint");
3020 if (!G1CollectedHeap::heap()->mark_in_progress()) {
3021 return;
3022 }
3023
3024 VerifyNoCSetOopsClosure cl;
3025
3026 if (verify_stacks) {
3027 // Verify entries on the global mark stack
3028 cl.set_phase(VerifyNoCSetOopsStack);
3029 _markStack.oops_do(&cl);
3030
3031 // Verify entries on the task queues
3032 for (uint i = 0; i < _max_worker_id; i += 1) {
3033 cl.set_phase(VerifyNoCSetOopsQueues, i);
3034 CMTaskQueue* queue = _task_queues->queue(i);
3035 queue->oops_do(&cl);
3036 }
3037 }
3038
3039 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
3040
3041 // Verify entries on the enqueued SATB buffers
3042 if (verify_enqueued_buffers) {
3043 cl.set_phase(VerifyNoCSetOopsSATBCompleted);
3044 satb_qs.iterate_completed_buffers_read_only(&cl);
3045 }
3046
3047 // Verify entries on the per-thread SATB buffers
3048 if (verify_thread_buffers) {
3049 cl.set_phase(VerifyNoCSetOopsSATBThread);
3050 satb_qs.iterate_thread_buffers_read_only(&cl);
3051 }
3052
3053 if (verify_fingers) {
3054 // Verify the global finger
3055 HeapWord* global_finger = finger();
3056 if (global_finger != NULL && global_finger < _heap_end) {
3057 // The global finger always points to a heap region boundary. We
3058 // use heap_region_containing_raw() to get the containing region
3059 // given that the global finger could be pointing to a free region
3060 // which subsequently becomes continues humongous. If that
3061 // happens, heap_region_containing() will return the bottom of the
3062 // corresponding starts humongous region and the check below will
3063 // not hold any more.
3064 // Since we always iterate over all regions, we might get a NULL HeapRegion
3065 // here.
3066 HeapRegion* global_hr = _g1h->heap_region_containing_raw(global_finger);
3067 guarantee(global_hr == NULL || global_finger == global_hr->bottom(),
3068 err_msg("global finger: "PTR_FORMAT" region: "HR_FORMAT,
3069 p2i(global_finger), HR_FORMAT_PARAMS(global_hr)));
3070 }
3071
3072 // Verify the task fingers
3073 assert(parallel_marking_threads() <= _max_worker_id, "sanity");
3074 for (int i = 0; i < (int) parallel_marking_threads(); i += 1) {
3075 CMTask* task = _tasks[i];
3076 HeapWord* task_finger = task->finger();
3077 if (task_finger != NULL && task_finger < _heap_end) {
3078 // See above note on the global finger verification.
3079 HeapRegion* task_hr = _g1h->heap_region_containing_raw(task_finger);
3080 guarantee(task_hr == NULL || task_finger == task_hr->bottom() ||
3081 !task_hr->in_collection_set(),
3082 err_msg("task finger: "PTR_FORMAT" region: "HR_FORMAT,
3083 p2i(task_finger), HR_FORMAT_PARAMS(task_hr)));
3084 }
3085 }
3086 }
3087 }
3088 #endif // PRODUCT
3089
3090 // Aggregate the counting data that was constructed concurrently
3091 // with marking.
3092 class AggregateCountDataHRClosure: public HeapRegionClosure {
3093 G1CollectedHeap* _g1h;
3094 ConcurrentMark* _cm;
3095 CardTableModRefBS* _ct_bs;
3096 BitMap* _cm_card_bm;
3097 uint _max_worker_id;
3098
3099 public:
3100 AggregateCountDataHRClosure(G1CollectedHeap* g1h,
3101 BitMap* cm_card_bm,
3102 uint max_worker_id) :
3103 _g1h(g1h), _cm(g1h->concurrent_mark()),
3104 _ct_bs(barrier_set_cast<CardTableModRefBS>(g1h->barrier_set())),
3105 _cm_card_bm(cm_card_bm), _max_worker_id(max_worker_id) { }
3106
3107 bool doHeapRegion(HeapRegion* hr) {
3108 if (hr->is_continues_humongous()) {
3109 // We will ignore these here and process them when their
3110 // associated "starts humongous" region is processed.
3111 // Note that we cannot rely on their associated
3112 // "starts humongous" region to have their bit set to 1
3113 // since, due to the region chunking in the parallel region
3114 // iteration, a "continues humongous" region might be visited
3115 // before its associated "starts humongous".
3116 return false;
3117 }
3118
3119 HeapWord* start = hr->bottom();
3120 HeapWord* limit = hr->next_top_at_mark_start();
3121 HeapWord* end = hr->end();
3122
3123 assert(start <= limit && limit <= hr->top() && hr->top() <= hr->end(),
3124 err_msg("Preconditions not met - "
3125 "start: "PTR_FORMAT", limit: "PTR_FORMAT", "
3126 "top: "PTR_FORMAT", end: "PTR_FORMAT,
3127 p2i(start), p2i(limit), p2i(hr->top()), p2i(hr->end())));
3128
3129 assert(hr->next_marked_bytes() == 0, "Precondition");
3130
3131 if (start == limit) {
3132 // NTAMS of this region has not been set so nothing to do.
3133 return false;
3134 }
3135
3136 // 'start' should be in the heap.
3137 assert(_g1h->is_in_g1_reserved(start) && _ct_bs->is_card_aligned(start), "sanity");
3138 // 'end' *may* be just beyond the end of the heap (if hr is the last region)
3139 assert(!_g1h->is_in_g1_reserved(end) || _ct_bs->is_card_aligned(end), "sanity");
3140
3141 BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
3142 BitMap::idx_t limit_idx = _cm->card_bitmap_index_for(limit);
3143 BitMap::idx_t end_idx = _cm->card_bitmap_index_for(end);
3144
3145 // If ntams is not card aligned then we bump card bitmap index
3146 // for limit so that we get the all the cards spanned by
3147 // the object ending at ntams.
3148 // Note: if this is the last region in the heap then ntams
3149 // could be actually just beyond the end of the the heap;
3150 // limit_idx will then correspond to a (non-existent) card
3151 // that is also outside the heap.
3152 if (_g1h->is_in_g1_reserved(limit) && !_ct_bs->is_card_aligned(limit)) {
3153 limit_idx += 1;
3154 }
3155
3156 assert(limit_idx <= end_idx, "or else use atomics");
3157
3158 // Aggregate the "stripe" in the count data associated with hr.
3159 uint hrm_index = hr->hrm_index();
3160 size_t marked_bytes = 0;
3161
3162 for (uint i = 0; i < _max_worker_id; i += 1) {
3163 size_t* marked_bytes_array = _cm->count_marked_bytes_array_for(i);
3164 BitMap* task_card_bm = _cm->count_card_bitmap_for(i);
3165
3166 // Fetch the marked_bytes in this region for task i and
3167 // add it to the running total for this region.
3168 marked_bytes += marked_bytes_array[hrm_index];
3169
3170 // Now union the bitmaps[0,max_worker_id)[start_idx..limit_idx)
3171 // into the global card bitmap.
3172 BitMap::idx_t scan_idx = task_card_bm->get_next_one_offset(start_idx, limit_idx);
3173
3174 while (scan_idx < limit_idx) {
3175 assert(task_card_bm->at(scan_idx) == true, "should be");
3176 _cm_card_bm->set_bit(scan_idx);
3177 assert(_cm_card_bm->at(scan_idx) == true, "should be");
3178
3179 // BitMap::get_next_one_offset() can handle the case when
3180 // its left_offset parameter is greater than its right_offset
3181 // parameter. It does, however, have an early exit if
3182 // left_offset == right_offset. So let's limit the value
3183 // passed in for left offset here.
3184 BitMap::idx_t next_idx = MIN2(scan_idx + 1, limit_idx);
3185 scan_idx = task_card_bm->get_next_one_offset(next_idx, limit_idx);
3186 }
3187 }
3188
3189 // Update the marked bytes for this region.
3190 hr->add_to_marked_bytes(marked_bytes);
3191
3192 // Next heap region
3193 return false;
3194 }
3195 };
3196
3197 class G1AggregateCountDataTask: public AbstractGangTask {
3198 protected:
3199 G1CollectedHeap* _g1h;
3200 ConcurrentMark* _cm;
3201 BitMap* _cm_card_bm;
3202 uint _max_worker_id;
3203 int _active_workers;
3204 HeapRegionClaimer _hrclaimer;
3205
3206 public:
3207 G1AggregateCountDataTask(G1CollectedHeap* g1h,
3208 ConcurrentMark* cm,
3209 BitMap* cm_card_bm,
3210 uint max_worker_id,
3211 int n_workers) :
3212 AbstractGangTask("Count Aggregation"),
3213 _g1h(g1h), _cm(cm), _cm_card_bm(cm_card_bm),
3214 _max_worker_id(max_worker_id),
3215 _active_workers(n_workers),
3216 _hrclaimer(_active_workers) {
3217 }
3218
3219 void work(uint worker_id) {
3220 AggregateCountDataHRClosure cl(_g1h, _cm_card_bm, _max_worker_id);
3221
3222 _g1h->heap_region_par_iterate(&cl, worker_id, &_hrclaimer);
3223 }
3224 };
3225
3226
3227 void ConcurrentMark::aggregate_count_data() {
3228 int n_workers = _g1h->workers()->active_workers();
3229
3230 G1AggregateCountDataTask g1_par_agg_task(_g1h, this, &_card_bm,
3231 _max_worker_id, n_workers);
3232
3233 _g1h->set_par_threads(n_workers);
3234 _g1h->workers()->run_task(&g1_par_agg_task);
3235 _g1h->set_par_threads(0);
3236 }
3237
3238 // Clear the per-worker arrays used to store the per-region counting data
3239 void ConcurrentMark::clear_all_count_data() {
3240 // Clear the global card bitmap - it will be filled during
3241 // liveness count aggregation (during remark) and the
3242 // final counting task.
3243 _card_bm.clear();
3244
3245 // Clear the global region bitmap - it will be filled as part
3246 // of the final counting task.
3247 _region_bm.clear();
3248
3249 uint max_regions = _g1h->max_regions();
3250 assert(_max_worker_id > 0, "uninitialized");
3251
3252 for (uint i = 0; i < _max_worker_id; i += 1) {
3253 BitMap* task_card_bm = count_card_bitmap_for(i);
3254 size_t* marked_bytes_array = count_marked_bytes_array_for(i);
3255
3256 assert(task_card_bm->size() == _card_bm.size(), "size mismatch");
3257 assert(marked_bytes_array != NULL, "uninitialized");
3258
3259 memset(marked_bytes_array, 0, (size_t) max_regions * sizeof(size_t));
3260 task_card_bm->clear();
3261 }
3262 }
3263
3264 void ConcurrentMark::print_stats() {
3265 if (verbose_stats()) {
3266 gclog_or_tty->print_cr("---------------------------------------------------------------------");
3267 for (size_t i = 0; i < _active_tasks; ++i) {
3268 _tasks[i]->print_stats();
3269 gclog_or_tty->print_cr("---------------------------------------------------------------------");
3270 }
3271 }
3272 }
3273
3274 // abandon current marking iteration due to a Full GC
3275 void ConcurrentMark::abort() {
3276 // Clear all marks in the next bitmap for the next marking cycle. This will allow us to skip the next
3277 // concurrent bitmap clearing.
3278 _nextMarkBitMap->clearAll();
3279
3280 // Note we cannot clear the previous marking bitmap here
3281 // since VerifyDuringGC verifies the objects marked during
3282 // a full GC against the previous bitmap.
3283
3284 // Clear the liveness counting data
3285 clear_all_count_data();
3286 // Empty mark stack
3287 reset_marking_state();
3288 for (uint i = 0; i < _max_worker_id; ++i) {
3289 _tasks[i]->clear_region_fields();
3290 }
3291 _first_overflow_barrier_sync.abort();
3292 _second_overflow_barrier_sync.abort();
3293 const GCId& gc_id = _g1h->gc_tracer_cm()->gc_id();
3294 if (!gc_id.is_undefined()) {
3295 // We can do multiple full GCs before ConcurrentMarkThread::run() gets a chance
3296 // to detect that it was aborted. Only keep track of the first GC id that we aborted.
3297 _aborted_gc_id = gc_id;
3298 }
3299 _has_aborted = true;
3300
3301 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3302 satb_mq_set.abandon_partial_marking();
3303 // This can be called either during or outside marking, we'll read
3304 // the expected_active value from the SATB queue set.
3305 satb_mq_set.set_active_all_threads(
3306 false, /* new active value */
3307 satb_mq_set.is_active() /* expected_active */);
3308
3309 _g1h->trace_heap_after_concurrent_cycle();
3310 _g1h->register_concurrent_cycle_end();
3311 }
3312
3313 const GCId& ConcurrentMark::concurrent_gc_id() {
3314 if (has_aborted()) {
3315 return _aborted_gc_id;
3316 }
3317 return _g1h->gc_tracer_cm()->gc_id();
3318 }
3319
3320 static void print_ms_time_info(const char* prefix, const char* name,
3321 NumberSeq& ns) {
3322 gclog_or_tty->print_cr("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
3323 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
3324 if (ns.num() > 0) {
3325 gclog_or_tty->print_cr("%s [std. dev = %8.2f ms, max = %8.2f ms]",
3326 prefix, ns.sd(), ns.maximum());
3327 }
3328 }
3329
3330 void ConcurrentMark::print_summary_info() {
3331 gclog_or_tty->print_cr(" Concurrent marking:");
3332 print_ms_time_info(" ", "init marks", _init_times);
3333 print_ms_time_info(" ", "remarks", _remark_times);
3334 {
3335 print_ms_time_info(" ", "final marks", _remark_mark_times);
3336 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
3337
3338 }
3339 print_ms_time_info(" ", "cleanups", _cleanup_times);
3340 gclog_or_tty->print_cr(" Final counting total time = %8.2f s (avg = %8.2f ms).",
3341 _total_counting_time,
3342 (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 /
3343 (double)_cleanup_times.num()
3344 : 0.0));
3345 if (G1ScrubRemSets) {
3346 gclog_or_tty->print_cr(" RS scrub total time = %8.2f s (avg = %8.2f ms).",
3347 _total_rs_scrub_time,
3348 (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 /
3349 (double)_cleanup_times.num()
3350 : 0.0));
3351 }
3352 gclog_or_tty->print_cr(" Total stop_world time = %8.2f s.",
3353 (_init_times.sum() + _remark_times.sum() +
3354 _cleanup_times.sum())/1000.0);
3355 gclog_or_tty->print_cr(" Total concurrent time = %8.2f s "
3356 "(%8.2f s marking).",
3357 cmThread()->vtime_accum(),
3358 cmThread()->vtime_mark_accum());
3359 }
3360
3361 void ConcurrentMark::print_worker_threads_on(outputStream* st) const {
3362 _parallel_workers->print_worker_threads_on(st);
3363 }
3364
3365 void ConcurrentMark::print_on_error(outputStream* st) const {
3366 st->print_cr("Marking Bits (Prev, Next): (CMBitMap*) " PTR_FORMAT ", (CMBitMap*) " PTR_FORMAT,
3367 p2i(_prevMarkBitMap), p2i(_nextMarkBitMap));
3368 _prevMarkBitMap->print_on_error(st, " Prev Bits: ");
3369 _nextMarkBitMap->print_on_error(st, " Next Bits: ");
3370 }
3371
3372 // We take a break if someone is trying to stop the world.
3373 bool ConcurrentMark::do_yield_check(uint worker_id) {
3374 if (SuspendibleThreadSet::should_yield()) {
3375 if (worker_id == 0) {
3376 _g1h->g1_policy()->record_concurrent_pause();
3377 }
3378 SuspendibleThreadSet::yield();
3379 return true;
3380 } else {
3381 return false;
3382 }
3383 }
3384
3385 #ifndef PRODUCT
3386 // for debugging purposes
3387 void ConcurrentMark::print_finger() {
3388 gclog_or_tty->print_cr("heap ["PTR_FORMAT", "PTR_FORMAT"), global finger = "PTR_FORMAT,
3389 p2i(_heap_start), p2i(_heap_end), p2i(_finger));
3390 for (uint i = 0; i < _max_worker_id; ++i) {
3391 gclog_or_tty->print(" %u: " PTR_FORMAT, i, p2i(_tasks[i]->finger()));
3392 }
3393 gclog_or_tty->cr();
3394 }
3395 #endif
3396
3397 void CMTask::scan_object(oop obj) {
3398 assert(_nextMarkBitMap->isMarked((HeapWord*) obj), "invariant");
3399
3400 if (_cm->verbose_high()) {
3401 gclog_or_tty->print_cr("[%u] we're scanning object "PTR_FORMAT,
3402 _worker_id, p2i((void*) obj));
3403 }
3404
3405 size_t obj_size = obj->size();
3406 _words_scanned += obj_size;
3407
3408 obj->oop_iterate(_cm_oop_closure);
3409 statsOnly( ++_objs_scanned );
3410 check_limits();
3411 }
3412
3413 // Closure for iteration over bitmaps
3414 class CMBitMapClosure : public BitMapClosure {
3415 private:
3416 // the bitmap that is being iterated over
3417 CMBitMap* _nextMarkBitMap;
3418 ConcurrentMark* _cm;
3419 CMTask* _task;
3420
3421 public:
3422 CMBitMapClosure(CMTask *task, ConcurrentMark* cm, CMBitMap* nextMarkBitMap) :
3423 _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { }
3424
3425 bool do_bit(size_t offset) {
3426 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset);
3427 assert(_nextMarkBitMap->isMarked(addr), "invariant");
3428 assert( addr < _cm->finger(), "invariant");
3429
3430 statsOnly( _task->increase_objs_found_on_bitmap() );
3431 assert(addr >= _task->finger(), "invariant");
3432
3433 // We move that task's local finger along.
3434 _task->move_finger_to(addr);
3435
3436 _task->scan_object(oop(addr));
3437 // we only partially drain the local queue and global stack
3438 _task->drain_local_queue(true);
3439 _task->drain_global_stack(true);
3440
3441 // if the has_aborted flag has been raised, we need to bail out of
3442 // the iteration
3443 return !_task->has_aborted();
3444 }
3445 };
3446
3447 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
3448 ConcurrentMark* cm,
3449 CMTask* task)
3450 : _g1h(g1h), _cm(cm), _task(task) {
3451 assert(_ref_processor == NULL, "should be initialized to NULL");
3452
3453 if (G1UseConcMarkReferenceProcessing) {
3454 _ref_processor = g1h->ref_processor_cm();
3455 assert(_ref_processor != NULL, "should not be NULL");
3456 }
3457 }
3458
3459 void CMTask::setup_for_region(HeapRegion* hr) {
3460 assert(hr != NULL,
3461 "claim_region() should have filtered out NULL regions");
3462 assert(!hr->is_continues_humongous(),
3463 "claim_region() should have filtered out continues humongous regions");
3464
3465 if (_cm->verbose_low()) {
3466 gclog_or_tty->print_cr("[%u] setting up for region "PTR_FORMAT,
3467 _worker_id, p2i(hr));
3468 }
3469
3470 _curr_region = hr;
3471 _finger = hr->bottom();
3472 update_region_limit();
3473 }
3474
3475 void CMTask::update_region_limit() {
3476 HeapRegion* hr = _curr_region;
3477 HeapWord* bottom = hr->bottom();
3478 HeapWord* limit = hr->next_top_at_mark_start();
3479
3480 if (limit == bottom) {
3481 if (_cm->verbose_low()) {
3482 gclog_or_tty->print_cr("[%u] found an empty region "
3483 "["PTR_FORMAT", "PTR_FORMAT")",
3484 _worker_id, p2i(bottom), p2i(limit));
3485 }
3486 // The region was collected underneath our feet.
3487 // We set the finger to bottom to ensure that the bitmap
3488 // iteration that will follow this will not do anything.
3489 // (this is not a condition that holds when we set the region up,
3490 // as the region is not supposed to be empty in the first place)
3491 _finger = bottom;
3492 } else if (limit >= _region_limit) {
3493 assert(limit >= _finger, "peace of mind");
3494 } else {
3495 assert(limit < _region_limit, "only way to get here");
3496 // This can happen under some pretty unusual circumstances. An
3497 // evacuation pause empties the region underneath our feet (NTAMS
3498 // at bottom). We then do some allocation in the region (NTAMS
3499 // stays at bottom), followed by the region being used as a GC
3500 // alloc region (NTAMS will move to top() and the objects
3501 // originally below it will be grayed). All objects now marked in
3502 // the region are explicitly grayed, if below the global finger,
3503 // and we do not need in fact to scan anything else. So, we simply
3504 // set _finger to be limit to ensure that the bitmap iteration
3505 // doesn't do anything.
3506 _finger = limit;
3507 }
3508
3509 _region_limit = limit;
3510 }
3511
3512 void CMTask::giveup_current_region() {
3513 assert(_curr_region != NULL, "invariant");
3514 if (_cm->verbose_low()) {
3515 gclog_or_tty->print_cr("[%u] giving up region "PTR_FORMAT,
3516 _worker_id, p2i(_curr_region));
3517 }
3518 clear_region_fields();
3519 }
3520
3521 void CMTask::clear_region_fields() {
3522 // Values for these three fields that indicate that we're not
3523 // holding on to a region.
3524 _curr_region = NULL;
3525 _finger = NULL;
3526 _region_limit = NULL;
3527 }
3528
3529 void CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
3530 if (cm_oop_closure == NULL) {
3531 assert(_cm_oop_closure != NULL, "invariant");
3532 } else {
3533 assert(_cm_oop_closure == NULL, "invariant");
3534 }
3535 _cm_oop_closure = cm_oop_closure;
3536 }
3537
3538 void CMTask::reset(CMBitMap* nextMarkBitMap) {
3539 guarantee(nextMarkBitMap != NULL, "invariant");
3540
3541 if (_cm->verbose_low()) {
3542 gclog_or_tty->print_cr("[%u] resetting", _worker_id);
3543 }
3544
3545 _nextMarkBitMap = nextMarkBitMap;
3546 clear_region_fields();
3547
3548 _calls = 0;
3549 _elapsed_time_ms = 0.0;
3550 _termination_time_ms = 0.0;
3551 _termination_start_time_ms = 0.0;
3552
3553 #if _MARKING_STATS_
3554 _aborted = 0;
3555 _aborted_overflow = 0;
3556 _aborted_cm_aborted = 0;
3557 _aborted_yield = 0;
3558 _aborted_timed_out = 0;
3559 _aborted_satb = 0;
3560 _aborted_termination = 0;
3561 _steal_attempts = 0;
3562 _steals = 0;
3563 _local_pushes = 0;
3564 _local_pops = 0;
3565 _local_max_size = 0;
3566 _objs_scanned = 0;
3567 _global_pushes = 0;
3568 _global_pops = 0;
3569 _global_max_size = 0;
3570 _global_transfers_to = 0;
3571 _global_transfers_from = 0;
3572 _regions_claimed = 0;
3573 _objs_found_on_bitmap = 0;
3574 _satb_buffers_processed = 0;
3575 #endif // _MARKING_STATS_
3576 }
3577
3578 bool CMTask::should_exit_termination() {
3579 regular_clock_call();
3580 // This is called when we are in the termination protocol. We should
3581 // quit if, for some reason, this task wants to abort or the global
3582 // stack is not empty (this means that we can get work from it).
3583 return !_cm->mark_stack_empty() || has_aborted();
3584 }
3585
3586 void CMTask::reached_limit() {
3587 assert(_words_scanned >= _words_scanned_limit ||
3588 _refs_reached >= _refs_reached_limit ,
3589 "shouldn't have been called otherwise");
3590 regular_clock_call();
3591 }
3592
3593 void CMTask::regular_clock_call() {
3594 if (has_aborted()) return;
3595
3596 // First, we need to recalculate the words scanned and refs reached
3597 // limits for the next clock call.
3598 recalculate_limits();
3599
3600 // During the regular clock call we do the following
3601
3602 // (1) If an overflow has been flagged, then we abort.
3603 if (_cm->has_overflown()) {
3604 set_has_aborted();
3605 return;
3606 }
3607
3608 // If we are not concurrent (i.e. we're doing remark) we don't need
3609 // to check anything else. The other steps are only needed during
3610 // the concurrent marking phase.
3611 if (!concurrent()) return;
3612
3613 // (2) If marking has been aborted for Full GC, then we also abort.
3614 if (_cm->has_aborted()) {
3615 set_has_aborted();
3616 statsOnly( ++_aborted_cm_aborted );
3617 return;
3618 }
3619
3620 double curr_time_ms = os::elapsedVTime() * 1000.0;
3621
3622 // (3) If marking stats are enabled, then we update the step history.
3623 #if _MARKING_STATS_
3624 if (_words_scanned >= _words_scanned_limit) {
3625 ++_clock_due_to_scanning;
3626 }
3627 if (_refs_reached >= _refs_reached_limit) {
3628 ++_clock_due_to_marking;
3629 }
3630
3631 double last_interval_ms = curr_time_ms - _interval_start_time_ms;
3632 _interval_start_time_ms = curr_time_ms;
3633 _all_clock_intervals_ms.add(last_interval_ms);
3634
3635 if (_cm->verbose_medium()) {
3636 gclog_or_tty->print_cr("[%u] regular clock, interval = %1.2lfms, "
3637 "scanned = "SIZE_FORMAT"%s, refs reached = "SIZE_FORMAT"%s",
3638 _worker_id, last_interval_ms,
3639 _words_scanned,
3640 (_words_scanned >= _words_scanned_limit) ? " (*)" : "",
3641 _refs_reached,
3642 (_refs_reached >= _refs_reached_limit) ? " (*)" : "");
3643 }
3644 #endif // _MARKING_STATS_
3645
3646 // (4) We check whether we should yield. If we have to, then we abort.
3647 if (SuspendibleThreadSet::should_yield()) {
3648 // We should yield. To do this we abort the task. The caller is
3649 // responsible for yielding.
3650 set_has_aborted();
3651 statsOnly( ++_aborted_yield );
3652 return;
3653 }
3654
3655 // (5) We check whether we've reached our time quota. If we have,
3656 // then we abort.
3657 double elapsed_time_ms = curr_time_ms - _start_time_ms;
3658 if (elapsed_time_ms > _time_target_ms) {
3659 set_has_aborted();
3660 _has_timed_out = true;
3661 statsOnly( ++_aborted_timed_out );
3662 return;
3663 }
3664
3665 // (6) Finally, we check whether there are enough completed STAB
3666 // buffers available for processing. If there are, we abort.
3667 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3668 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
3669 if (_cm->verbose_low()) {
3670 gclog_or_tty->print_cr("[%u] aborting to deal with pending SATB buffers",
3671 _worker_id);
3672 }
3673 // we do need to process SATB buffers, we'll abort and restart
3674 // the marking task to do so
3675 set_has_aborted();
3676 statsOnly( ++_aborted_satb );
3677 return;
3678 }
3679 }
3680
3681 void CMTask::recalculate_limits() {
3682 _real_words_scanned_limit = _words_scanned + words_scanned_period;
3683 _words_scanned_limit = _real_words_scanned_limit;
3684
3685 _real_refs_reached_limit = _refs_reached + refs_reached_period;
3686 _refs_reached_limit = _real_refs_reached_limit;
3687 }
3688
3689 void CMTask::decrease_limits() {
3690 // This is called when we believe that we're going to do an infrequent
3691 // operation which will increase the per byte scanned cost (i.e. move
3692 // entries to/from the global stack). It basically tries to decrease the
3693 // scanning limit so that the clock is called earlier.
3694
3695 if (_cm->verbose_medium()) {
3696 gclog_or_tty->print_cr("[%u] decreasing limits", _worker_id);
3697 }
3698
3699 _words_scanned_limit = _real_words_scanned_limit -
3700 3 * words_scanned_period / 4;
3701 _refs_reached_limit = _real_refs_reached_limit -
3702 3 * refs_reached_period / 4;
3703 }
3704
3705 void CMTask::move_entries_to_global_stack() {
3706 // local array where we'll store the entries that will be popped
3707 // from the local queue
3708 oop buffer[global_stack_transfer_size];
3709
3710 int n = 0;
3711 oop obj;
3712 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) {
3713 buffer[n] = obj;
3714 ++n;
3715 }
3716
3717 if (n > 0) {
3718 // we popped at least one entry from the local queue
3719
3720 statsOnly( ++_global_transfers_to; _local_pops += n );
3721
3722 if (!_cm->mark_stack_push(buffer, n)) {
3723 if (_cm->verbose_low()) {
3724 gclog_or_tty->print_cr("[%u] aborting due to global stack overflow",
3725 _worker_id);
3726 }
3727 set_has_aborted();
3728 } else {
3729 // the transfer was successful
3730
3731 if (_cm->verbose_medium()) {
3732 gclog_or_tty->print_cr("[%u] pushed %d entries to the global stack",
3733 _worker_id, n);
3734 }
3735 statsOnly( size_t tmp_size = _cm->mark_stack_size();
3736 if (tmp_size > _global_max_size) {
3737 _global_max_size = tmp_size;
3738 }
3739 _global_pushes += n );
3740 }
3741 }
3742
3743 // this operation was quite expensive, so decrease the limits
3744 decrease_limits();
3745 }
3746
3747 void CMTask::get_entries_from_global_stack() {
3748 // local array where we'll store the entries that will be popped
3749 // from the global stack.
3750 oop buffer[global_stack_transfer_size];
3751 int n;
3752 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n);
3753 assert(n <= global_stack_transfer_size,
3754 "we should not pop more than the given limit");
3755 if (n > 0) {
3756 // yes, we did actually pop at least one entry
3757
3758 statsOnly( ++_global_transfers_from; _global_pops += n );
3759 if (_cm->verbose_medium()) {
3760 gclog_or_tty->print_cr("[%u] popped %d entries from the global stack",
3761 _worker_id, n);
3762 }
3763 for (int i = 0; i < n; ++i) {
3764 bool success = _task_queue->push(buffer[i]);
3765 // We only call this when the local queue is empty or under a
3766 // given target limit. So, we do not expect this push to fail.
3767 assert(success, "invariant");
3768 }
3769
3770 statsOnly( size_t tmp_size = (size_t)_task_queue->size();
3771 if (tmp_size > _local_max_size) {
3772 _local_max_size = tmp_size;
3773 }
3774 _local_pushes += n );
3775 }
3776
3777 // this operation was quite expensive, so decrease the limits
3778 decrease_limits();
3779 }
3780
3781 void CMTask::drain_local_queue(bool partially) {
3782 if (has_aborted()) return;
3783
3784 // Decide what the target size is, depending whether we're going to
3785 // drain it partially (so that other tasks can steal if they run out
3786 // of things to do) or totally (at the very end).
3787 size_t target_size;
3788 if (partially) {
3789 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
3790 } else {
3791 target_size = 0;
3792 }
3793
3794 if (_task_queue->size() > target_size) {
3795 if (_cm->verbose_high()) {
3796 gclog_or_tty->print_cr("[%u] draining local queue, target size = " SIZE_FORMAT,
3797 _worker_id, target_size);
3798 }
3799
3800 oop obj;
3801 bool ret = _task_queue->pop_local(obj);
3802 while (ret) {
3803 statsOnly( ++_local_pops );
3804
3805 if (_cm->verbose_high()) {
3806 gclog_or_tty->print_cr("[%u] popped "PTR_FORMAT, _worker_id,
3807 p2i((void*) obj));
3808 }
3809
3810 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" );
3811 assert(!_g1h->is_on_master_free_list(
3812 _g1h->heap_region_containing((HeapWord*) obj)), "invariant");
3813
3814 scan_object(obj);
3815
3816 if (_task_queue->size() <= target_size || has_aborted()) {
3817 ret = false;
3818 } else {
3819 ret = _task_queue->pop_local(obj);
3820 }
3821 }
3822
3823 if (_cm->verbose_high()) {
3824 gclog_or_tty->print_cr("[%u] drained local queue, size = %u",
3825 _worker_id, _task_queue->size());
3826 }
3827 }
3828 }
3829
3830 void CMTask::drain_global_stack(bool partially) {
3831 if (has_aborted()) return;
3832
3833 // We have a policy to drain the local queue before we attempt to
3834 // drain the global stack.
3835 assert(partially || _task_queue->size() == 0, "invariant");
3836
3837 // Decide what the target size is, depending whether we're going to
3838 // drain it partially (so that other tasks can steal if they run out
3839 // of things to do) or totally (at the very end). Notice that,
3840 // because we move entries from the global stack in chunks or
3841 // because another task might be doing the same, we might in fact
3842 // drop below the target. But, this is not a problem.
3843 size_t target_size;
3844 if (partially) {
3845 target_size = _cm->partial_mark_stack_size_target();
3846 } else {
3847 target_size = 0;
3848 }
3849
3850 if (_cm->mark_stack_size() > target_size) {
3851 if (_cm->verbose_low()) {
3852 gclog_or_tty->print_cr("[%u] draining global_stack, target size " SIZE_FORMAT,
3853 _worker_id, target_size);
3854 }
3855
3856 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
3857 get_entries_from_global_stack();
3858 drain_local_queue(partially);
3859 }
3860
3861 if (_cm->verbose_low()) {
3862 gclog_or_tty->print_cr("[%u] drained global stack, size = " SIZE_FORMAT,
3863 _worker_id, _cm->mark_stack_size());
3864 }
3865 }
3866 }
3867
3868 // SATB Queue has several assumptions on whether to call the par or
3869 // non-par versions of the methods. this is why some of the code is
3870 // replicated. We should really get rid of the single-threaded version
3871 // of the code to simplify things.
3872 void CMTask::drain_satb_buffers() {
3873 if (has_aborted()) return;
3874
3875 // We set this so that the regular clock knows that we're in the
3876 // middle of draining buffers and doesn't set the abort flag when it
3877 // notices that SATB buffers are available for draining. It'd be
3878 // very counter productive if it did that. :-)
3879 _draining_satb_buffers = true;
3880
3881 CMObjectClosure oc(this);
3882 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3883 satb_mq_set.set_closure(_worker_id, &oc);
3884
3885 // This keeps claiming and applying the closure to completed buffers
3886 // until we run out of buffers or we need to abort.
3887 while (!has_aborted() &&
3888 satb_mq_set.apply_closure_to_completed_buffer(_worker_id)) {
3889 if (_cm->verbose_medium()) {
3890 gclog_or_tty->print_cr("[%u] processed an SATB buffer", _worker_id);
3891 }
3892 statsOnly( ++_satb_buffers_processed );
3893 regular_clock_call();
3894 }
3895
3896 _draining_satb_buffers = false;
3897
3898 assert(has_aborted() ||
3899 concurrent() ||
3900 satb_mq_set.completed_buffers_num() == 0, "invariant");
3901
3902 satb_mq_set.set_closure(_worker_id, NULL);
3903
3904 // again, this was a potentially expensive operation, decrease the
3905 // limits to get the regular clock call early
3906 decrease_limits();
3907 }
3908
3909 void CMTask::print_stats() {
3910 gclog_or_tty->print_cr("Marking Stats, task = %u, calls = %d",
3911 _worker_id, _calls);
3912 gclog_or_tty->print_cr(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
3913 _elapsed_time_ms, _termination_time_ms);
3914 gclog_or_tty->print_cr(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3915 _step_times_ms.num(), _step_times_ms.avg(),
3916 _step_times_ms.sd());
3917 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
3918 _step_times_ms.maximum(), _step_times_ms.sum());
3919
3920 #if _MARKING_STATS_
3921 gclog_or_tty->print_cr(" Clock Intervals (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
3922 _all_clock_intervals_ms.num(), _all_clock_intervals_ms.avg(),
3923 _all_clock_intervals_ms.sd());
3924 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
3925 _all_clock_intervals_ms.maximum(),
3926 _all_clock_intervals_ms.sum());
3927 gclog_or_tty->print_cr(" Clock Causes (cum): scanning = " SIZE_FORMAT ", marking = " SIZE_FORMAT,
3928 _clock_due_to_scanning, _clock_due_to_marking);
3929 gclog_or_tty->print_cr(" Objects: scanned = " SIZE_FORMAT ", found on the bitmap = " SIZE_FORMAT,
3930 _objs_scanned, _objs_found_on_bitmap);
3931 gclog_or_tty->print_cr(" Local Queue: pushes = " SIZE_FORMAT ", pops = " SIZE_FORMAT ", max size = " SIZE_FORMAT,
3932 _local_pushes, _local_pops, _local_max_size);
3933 gclog_or_tty->print_cr(" Global Stack: pushes = " SIZE_FORMAT ", pops = " SIZE_FORMAT ", max size = " SIZE_FORMAT,
3934 _global_pushes, _global_pops, _global_max_size);
3935 gclog_or_tty->print_cr(" transfers to = " SIZE_FORMAT ", transfers from = " SIZE_FORMAT,
3936 _global_transfers_to,_global_transfers_from);
3937 gclog_or_tty->print_cr(" Regions: claimed = " SIZE_FORMAT, _regions_claimed);
3938 gclog_or_tty->print_cr(" SATB buffers: processed = " SIZE_FORMAT, _satb_buffers_processed);
3939 gclog_or_tty->print_cr(" Steals: attempts = " SIZE_FORMAT ", successes = " SIZE_FORMAT,
3940 _steal_attempts, _steals);
3941 gclog_or_tty->print_cr(" Aborted: " SIZE_FORMAT ", due to", _aborted);
3942 gclog_or_tty->print_cr(" overflow: " SIZE_FORMAT ", global abort: " SIZE_FORMAT ", yield: " SIZE_FORMAT,
3943 _aborted_overflow, _aborted_cm_aborted, _aborted_yield);
3944 gclog_or_tty->print_cr(" time out: " SIZE_FORMAT ", SATB: " SIZE_FORMAT ", termination: " SIZE_FORMAT,
3945 _aborted_timed_out, _aborted_satb, _aborted_termination);
3946 #endif // _MARKING_STATS_
3947 }
3948
3949 /*****************************************************************************
3950
3951 The do_marking_step(time_target_ms, ...) method is the building
3952 block of the parallel marking framework. It can be called in parallel
3953 with other invocations of do_marking_step() on different tasks
3954 (but only one per task, obviously) and concurrently with the
3955 mutator threads, or during remark, hence it eliminates the need
3956 for two versions of the code. When called during remark, it will
3957 pick up from where the task left off during the concurrent marking
3958 phase. Interestingly, tasks are also claimable during evacuation
3959 pauses too, since do_marking_step() ensures that it aborts before
3960 it needs to yield.
3961
3962 The data structures that it uses to do marking work are the
3963 following:
3964
3965 (1) Marking Bitmap. If there are gray objects that appear only
3966 on the bitmap (this happens either when dealing with an overflow
3967 or when the initial marking phase has simply marked the roots
3968 and didn't push them on the stack), then tasks claim heap
3969 regions whose bitmap they then scan to find gray objects. A
3970 global finger indicates where the end of the last claimed region
3971 is. A local finger indicates how far into the region a task has
3972 scanned. The two fingers are used to determine how to gray an
3973 object (i.e. whether simply marking it is OK, as it will be
3974 visited by a task in the future, or whether it needs to be also
3975 pushed on a stack).
3976
3977 (2) Local Queue. The local queue of the task which is accessed
3978 reasonably efficiently by the task. Other tasks can steal from
3979 it when they run out of work. Throughout the marking phase, a
3980 task attempts to keep its local queue short but not totally
3981 empty, so that entries are available for stealing by other
3982 tasks. Only when there is no more work, a task will totally
3983 drain its local queue.
3984
3985 (3) Global Mark Stack. This handles local queue overflow. During
3986 marking only sets of entries are moved between it and the local
3987 queues, as access to it requires a mutex and more fine-grain
3988 interaction with it which might cause contention. If it
3989 overflows, then the marking phase should restart and iterate
3990 over the bitmap to identify gray objects. Throughout the marking
3991 phase, tasks attempt to keep the global mark stack at a small
3992 length but not totally empty, so that entries are available for
3993 popping by other tasks. Only when there is no more work, tasks
3994 will totally drain the global mark stack.
3995
3996 (4) SATB Buffer Queue. This is where completed SATB buffers are
3997 made available. Buffers are regularly removed from this queue
3998 and scanned for roots, so that the queue doesn't get too
3999 long. During remark, all completed buffers are processed, as
4000 well as the filled in parts of any uncompleted buffers.
4001
4002 The do_marking_step() method tries to abort when the time target
4003 has been reached. There are a few other cases when the
4004 do_marking_step() method also aborts:
4005
4006 (1) When the marking phase has been aborted (after a Full GC).
4007
4008 (2) When a global overflow (on the global stack) has been
4009 triggered. Before the task aborts, it will actually sync up with
4010 the other tasks to ensure that all the marking data structures
4011 (local queues, stacks, fingers etc.) are re-initialized so that
4012 when do_marking_step() completes, the marking phase can
4013 immediately restart.
4014
4015 (3) When enough completed SATB buffers are available. The
4016 do_marking_step() method only tries to drain SATB buffers right
4017 at the beginning. So, if enough buffers are available, the
4018 marking step aborts and the SATB buffers are processed at
4019 the beginning of the next invocation.
4020
4021 (4) To yield. when we have to yield then we abort and yield
4022 right at the end of do_marking_step(). This saves us from a lot
4023 of hassle as, by yielding we might allow a Full GC. If this
4024 happens then objects will be compacted underneath our feet, the
4025 heap might shrink, etc. We save checking for this by just
4026 aborting and doing the yield right at the end.
4027
4028 From the above it follows that the do_marking_step() method should
4029 be called in a loop (or, otherwise, regularly) until it completes.
4030
4031 If a marking step completes without its has_aborted() flag being
4032 true, it means it has completed the current marking phase (and
4033 also all other marking tasks have done so and have all synced up).
4034
4035 A method called regular_clock_call() is invoked "regularly" (in
4036 sub ms intervals) throughout marking. It is this clock method that
4037 checks all the abort conditions which were mentioned above and
4038 decides when the task should abort. A work-based scheme is used to
4039 trigger this clock method: when the number of object words the
4040 marking phase has scanned or the number of references the marking
4041 phase has visited reach a given limit. Additional invocations to
4042 the method clock have been planted in a few other strategic places
4043 too. The initial reason for the clock method was to avoid calling
4044 vtime too regularly, as it is quite expensive. So, once it was in
4045 place, it was natural to piggy-back all the other conditions on it
4046 too and not constantly check them throughout the code.
4047
4048 If do_termination is true then do_marking_step will enter its
4049 termination protocol.
4050
4051 The value of is_serial must be true when do_marking_step is being
4052 called serially (i.e. by the VMThread) and do_marking_step should
4053 skip any synchronization in the termination and overflow code.
4054 Examples include the serial remark code and the serial reference
4055 processing closures.
4056
4057 The value of is_serial must be false when do_marking_step is
4058 being called by any of the worker threads in a work gang.
4059 Examples include the concurrent marking code (CMMarkingTask),
4060 the MT remark code, and the MT reference processing closures.
4061
4062 *****************************************************************************/
4063
4064 void CMTask::do_marking_step(double time_target_ms,
4065 bool do_termination,
4066 bool is_serial) {
4067 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
4068 assert(concurrent() == _cm->concurrent(), "they should be the same");
4069
4070 G1CollectorPolicy* g1_policy = _g1h->g1_policy();
4071 assert(_task_queues != NULL, "invariant");
4072 assert(_task_queue != NULL, "invariant");
4073 assert(_task_queues->queue(_worker_id) == _task_queue, "invariant");
4074
4075 assert(!_claimed,
4076 "only one thread should claim this task at any one time");
4077
4078 // OK, this doesn't safeguard again all possible scenarios, as it is
4079 // possible for two threads to set the _claimed flag at the same
4080 // time. But it is only for debugging purposes anyway and it will
4081 // catch most problems.
4082 _claimed = true;
4083
4084 _start_time_ms = os::elapsedVTime() * 1000.0;
4085 statsOnly( _interval_start_time_ms = _start_time_ms );
4086
4087 // If do_stealing is true then do_marking_step will attempt to
4088 // steal work from the other CMTasks. It only makes sense to
4089 // enable stealing when the termination protocol is enabled
4090 // and do_marking_step() is not being called serially.
4091 bool do_stealing = do_termination && !is_serial;
4092
4093 double diff_prediction_ms =
4094 g1_policy->get_new_prediction(&_marking_step_diffs_ms);
4095 _time_target_ms = time_target_ms - diff_prediction_ms;
4096
4097 // set up the variables that are used in the work-based scheme to
4098 // call the regular clock method
4099 _words_scanned = 0;
4100 _refs_reached = 0;
4101 recalculate_limits();
4102
4103 // clear all flags
4104 clear_has_aborted();
4105 _has_timed_out = false;
4106 _draining_satb_buffers = false;
4107
4108 ++_calls;
4109
4110 if (_cm->verbose_low()) {
4111 gclog_or_tty->print_cr("[%u] >>>>>>>>>> START, call = %d, "
4112 "target = %1.2lfms >>>>>>>>>>",
4113 _worker_id, _calls, _time_target_ms);
4114 }
4115
4116 // Set up the bitmap and oop closures. Anything that uses them is
4117 // eventually called from this method, so it is OK to allocate these
4118 // statically.
4119 CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap);
4120 G1CMOopClosure cm_oop_closure(_g1h, _cm, this);
4121 set_cm_oop_closure(&cm_oop_closure);
4122
4123 if (_cm->has_overflown()) {
4124 // This can happen if the mark stack overflows during a GC pause
4125 // and this task, after a yield point, restarts. We have to abort
4126 // as we need to get into the overflow protocol which happens
4127 // right at the end of this task.
4128 set_has_aborted();
4129 }
4130
4131 // First drain any available SATB buffers. After this, we will not
4132 // look at SATB buffers before the next invocation of this method.
4133 // If enough completed SATB buffers are queued up, the regular clock
4134 // will abort this task so that it restarts.
4135 drain_satb_buffers();
4136 // ...then partially drain the local queue and the global stack
4137 drain_local_queue(true);
4138 drain_global_stack(true);
4139
4140 do {
4141 if (!has_aborted() && _curr_region != NULL) {
4142 // This means that we're already holding on to a region.
4143 assert(_finger != NULL, "if region is not NULL, then the finger "
4144 "should not be NULL either");
4145
4146 // We might have restarted this task after an evacuation pause
4147 // which might have evacuated the region we're holding on to
4148 // underneath our feet. Let's read its limit again to make sure
4149 // that we do not iterate over a region of the heap that
4150 // contains garbage (update_region_limit() will also move
4151 // _finger to the start of the region if it is found empty).
4152 update_region_limit();
4153 // We will start from _finger not from the start of the region,
4154 // as we might be restarting this task after aborting half-way
4155 // through scanning this region. In this case, _finger points to
4156 // the address where we last found a marked object. If this is a
4157 // fresh region, _finger points to start().
4158 MemRegion mr = MemRegion(_finger, _region_limit);
4159
4160 if (_cm->verbose_low()) {
4161 gclog_or_tty->print_cr("[%u] we're scanning part "
4162 "["PTR_FORMAT", "PTR_FORMAT") "
4163 "of region "HR_FORMAT,
4164 _worker_id, p2i(_finger), p2i(_region_limit),
4165 HR_FORMAT_PARAMS(_curr_region));
4166 }
4167
4168 assert(!_curr_region->is_humongous() || mr.start() == _curr_region->bottom(),
4169 "humongous regions should go around loop once only");
4170
4171 // Some special cases:
4172 // If the memory region is empty, we can just give up the region.
4173 // If the current region is humongous then we only need to check
4174 // the bitmap for the bit associated with the start of the object,
4175 // scan the object if it's live, and give up the region.
4176 // Otherwise, let's iterate over the bitmap of the part of the region
4177 // that is left.
4178 // If the iteration is successful, give up the region.
4179 if (mr.is_empty()) {
4180 giveup_current_region();
4181 regular_clock_call();
4182 } else if (_curr_region->is_humongous() && mr.start() == _curr_region->bottom()) {
4183 if (_nextMarkBitMap->isMarked(mr.start())) {
4184 // The object is marked - apply the closure
4185 BitMap::idx_t offset = _nextMarkBitMap->heapWordToOffset(mr.start());
4186 bitmap_closure.do_bit(offset);
4187 }
4188 // Even if this task aborted while scanning the humongous object
4189 // we can (and should) give up the current region.
4190 giveup_current_region();
4191 regular_clock_call();
4192 } else if (_nextMarkBitMap->iterate(&bitmap_closure, mr)) {
4193 giveup_current_region();
4194 regular_clock_call();
4195 } else {
4196 assert(has_aborted(), "currently the only way to do so");
4197 // The only way to abort the bitmap iteration is to return
4198 // false from the do_bit() method. However, inside the
4199 // do_bit() method we move the _finger to point to the
4200 // object currently being looked at. So, if we bail out, we
4201 // have definitely set _finger to something non-null.
4202 assert(_finger != NULL, "invariant");
4203
4204 // Region iteration was actually aborted. So now _finger
4205 // points to the address of the object we last scanned. If we
4206 // leave it there, when we restart this task, we will rescan
4207 // the object. It is easy to avoid this. We move the finger by
4208 // enough to point to the next possible object header (the
4209 // bitmap knows by how much we need to move it as it knows its
4210 // granularity).
4211 assert(_finger < _region_limit, "invariant");
4212 HeapWord* new_finger = _nextMarkBitMap->nextObject(_finger);
4213 // Check if bitmap iteration was aborted while scanning the last object
4214 if (new_finger >= _region_limit) {
4215 giveup_current_region();
4216 } else {
4217 move_finger_to(new_finger);
4218 }
4219 }
4220 }
4221 // At this point we have either completed iterating over the
4222 // region we were holding on to, or we have aborted.
4223
4224 // We then partially drain the local queue and the global stack.
4225 // (Do we really need this?)
4226 drain_local_queue(true);
4227 drain_global_stack(true);
4228
4229 // Read the note on the claim_region() method on why it might
4230 // return NULL with potentially more regions available for
4231 // claiming and why we have to check out_of_regions() to determine
4232 // whether we're done or not.
4233 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
4234 // We are going to try to claim a new region. We should have
4235 // given up on the previous one.
4236 // Separated the asserts so that we know which one fires.
4237 assert(_curr_region == NULL, "invariant");
4238 assert(_finger == NULL, "invariant");
4239 assert(_region_limit == NULL, "invariant");
4240 if (_cm->verbose_low()) {
4241 gclog_or_tty->print_cr("[%u] trying to claim a new region", _worker_id);
4242 }
4243 HeapRegion* claimed_region = _cm->claim_region(_worker_id);
4244 if (claimed_region != NULL) {
4245 // Yes, we managed to claim one
4246 statsOnly( ++_regions_claimed );
4247
4248 if (_cm->verbose_low()) {
4249 gclog_or_tty->print_cr("[%u] we successfully claimed "
4250 "region "PTR_FORMAT,
4251 _worker_id, p2i(claimed_region));
4252 }
4253
4254 setup_for_region(claimed_region);
4255 assert(_curr_region == claimed_region, "invariant");
4256 }
4257 // It is important to call the regular clock here. It might take
4258 // a while to claim a region if, for example, we hit a large
4259 // block of empty regions. So we need to call the regular clock
4260 // method once round the loop to make sure it's called
4261 // frequently enough.
4262 regular_clock_call();
4263 }
4264
4265 if (!has_aborted() && _curr_region == NULL) {
4266 assert(_cm->out_of_regions(),
4267 "at this point we should be out of regions");
4268 }
4269 } while ( _curr_region != NULL && !has_aborted());
4270
4271 if (!has_aborted()) {
4272 // We cannot check whether the global stack is empty, since other
4273 // tasks might be pushing objects to it concurrently.
4274 assert(_cm->out_of_regions(),
4275 "at this point we should be out of regions");
4276
4277 if (_cm->verbose_low()) {
4278 gclog_or_tty->print_cr("[%u] all regions claimed", _worker_id);
4279 }
4280
4281 // Try to reduce the number of available SATB buffers so that
4282 // remark has less work to do.
4283 drain_satb_buffers();
4284 }
4285
4286 // Since we've done everything else, we can now totally drain the
4287 // local queue and global stack.
4288 drain_local_queue(false);
4289 drain_global_stack(false);
4290
4291 // Attempt at work stealing from other task's queues.
4292 if (do_stealing && !has_aborted()) {
4293 // We have not aborted. This means that we have finished all that
4294 // we could. Let's try to do some stealing...
4295
4296 // We cannot check whether the global stack is empty, since other
4297 // tasks might be pushing objects to it concurrently.
4298 assert(_cm->out_of_regions() && _task_queue->size() == 0,
4299 "only way to reach here");
4300
4301 if (_cm->verbose_low()) {
4302 gclog_or_tty->print_cr("[%u] starting to steal", _worker_id);
4303 }
4304
4305 while (!has_aborted()) {
4306 oop obj;
4307 statsOnly( ++_steal_attempts );
4308
4309 if (_cm->try_stealing(_worker_id, &_hash_seed, obj)) {
4310 if (_cm->verbose_medium()) {
4311 gclog_or_tty->print_cr("[%u] stolen "PTR_FORMAT" successfully",
4312 _worker_id, p2i((void*) obj));
4313 }
4314
4315 statsOnly( ++_steals );
4316
4317 assert(_nextMarkBitMap->isMarked((HeapWord*) obj),
4318 "any stolen object should be marked");
4319 scan_object(obj);
4320
4321 // And since we're towards the end, let's totally drain the
4322 // local queue and global stack.
4323 drain_local_queue(false);
4324 drain_global_stack(false);
4325 } else {
4326 break;
4327 }
4328 }
4329 }
4330
4331 // If we are about to wrap up and go into termination, check if we
4332 // should raise the overflow flag.
4333 if (do_termination && !has_aborted()) {
4334 if (_cm->force_overflow()->should_force()) {
4335 _cm->set_has_overflown();
4336 regular_clock_call();
4337 }
4338 }
4339
4340 // We still haven't aborted. Now, let's try to get into the
4341 // termination protocol.
4342 if (do_termination && !has_aborted()) {
4343 // We cannot check whether the global stack is empty, since other
4344 // tasks might be concurrently pushing objects on it.
4345 // Separated the asserts so that we know which one fires.
4346 assert(_cm->out_of_regions(), "only way to reach here");
4347 assert(_task_queue->size() == 0, "only way to reach here");
4348
4349 if (_cm->verbose_low()) {
4350 gclog_or_tty->print_cr("[%u] starting termination protocol", _worker_id);
4351 }
4352
4353 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
4354
4355 // The CMTask class also extends the TerminatorTerminator class,
4356 // hence its should_exit_termination() method will also decide
4357 // whether to exit the termination protocol or not.
4358 bool finished = (is_serial ||
4359 _cm->terminator()->offer_termination(this));
4360 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
4361 _termination_time_ms +=
4362 termination_end_time_ms - _termination_start_time_ms;
4363
4364 if (finished) {
4365 // We're all done.
4366
4367 if (_worker_id == 0) {
4368 // let's allow task 0 to do this
4369 if (concurrent()) {
4370 assert(_cm->concurrent_marking_in_progress(), "invariant");
4371 // we need to set this to false before the next
4372 // safepoint. This way we ensure that the marking phase
4373 // doesn't observe any more heap expansions.
4374 _cm->clear_concurrent_marking_in_progress();
4375 }
4376 }
4377
4378 // We can now guarantee that the global stack is empty, since
4379 // all other tasks have finished. We separated the guarantees so
4380 // that, if a condition is false, we can immediately find out
4381 // which one.
4382 guarantee(_cm->out_of_regions(), "only way to reach here");
4383 guarantee(_cm->mark_stack_empty(), "only way to reach here");
4384 guarantee(_task_queue->size() == 0, "only way to reach here");
4385 guarantee(!_cm->has_overflown(), "only way to reach here");
4386 guarantee(!_cm->mark_stack_overflow(), "only way to reach here");
4387
4388 if (_cm->verbose_low()) {
4389 gclog_or_tty->print_cr("[%u] all tasks terminated", _worker_id);
4390 }
4391 } else {
4392 // Apparently there's more work to do. Let's abort this task. It
4393 // will restart it and we can hopefully find more things to do.
4394
4395 if (_cm->verbose_low()) {
4396 gclog_or_tty->print_cr("[%u] apparently there is more work to do",
4397 _worker_id);
4398 }
4399
4400 set_has_aborted();
4401 statsOnly( ++_aborted_termination );
4402 }
4403 }
4404
4405 // Mainly for debugging purposes to make sure that a pointer to the
4406 // closure which was statically allocated in this frame doesn't
4407 // escape it by accident.
4408 set_cm_oop_closure(NULL);
4409 double end_time_ms = os::elapsedVTime() * 1000.0;
4410 double elapsed_time_ms = end_time_ms - _start_time_ms;
4411 // Update the step history.
4412 _step_times_ms.add(elapsed_time_ms);
4413
4414 if (has_aborted()) {
4415 // The task was aborted for some reason.
4416
4417 statsOnly( ++_aborted );
4418
4419 if (_has_timed_out) {
4420 double diff_ms = elapsed_time_ms - _time_target_ms;
4421 // Keep statistics of how well we did with respect to hitting
4422 // our target only if we actually timed out (if we aborted for
4423 // other reasons, then the results might get skewed).
4424 _marking_step_diffs_ms.add(diff_ms);
4425 }
4426
4427 if (_cm->has_overflown()) {
4428 // This is the interesting one. We aborted because a global
4429 // overflow was raised. This means we have to restart the
4430 // marking phase and start iterating over regions. However, in
4431 // order to do this we have to make sure that all tasks stop
4432 // what they are doing and re-initialize in a safe manner. We
4433 // will achieve this with the use of two barrier sync points.
4434
4435 if (_cm->verbose_low()) {
4436 gclog_or_tty->print_cr("[%u] detected overflow", _worker_id);
4437 }
4438
4439 if (!is_serial) {
4440 // We only need to enter the sync barrier if being called
4441 // from a parallel context
4442 _cm->enter_first_sync_barrier(_worker_id);
4443
4444 // When we exit this sync barrier we know that all tasks have
4445 // stopped doing marking work. So, it's now safe to
4446 // re-initialize our data structures. At the end of this method,
4447 // task 0 will clear the global data structures.
4448 }
4449
4450 statsOnly( ++_aborted_overflow );
4451
4452 // We clear the local state of this task...
4453 clear_region_fields();
4454
4455 if (!is_serial) {
4456 // ...and enter the second barrier.
4457 _cm->enter_second_sync_barrier(_worker_id);
4458 }
4459 // At this point, if we're during the concurrent phase of
4460 // marking, everything has been re-initialized and we're
4461 // ready to restart.
4462 }
4463
4464 if (_cm->verbose_low()) {
4465 gclog_or_tty->print_cr("[%u] <<<<<<<<<< ABORTING, target = %1.2lfms, "
4466 "elapsed = %1.2lfms <<<<<<<<<<",
4467 _worker_id, _time_target_ms, elapsed_time_ms);
4468 if (_cm->has_aborted()) {
4469 gclog_or_tty->print_cr("[%u] ========== MARKING ABORTED ==========",
4470 _worker_id);
4471 }
4472 }
4473 } else {
4474 if (_cm->verbose_low()) {
4475 gclog_or_tty->print_cr("[%u] <<<<<<<<<< FINISHED, target = %1.2lfms, "
4476 "elapsed = %1.2lfms <<<<<<<<<<",
4477 _worker_id, _time_target_ms, elapsed_time_ms);
4478 }
4479 }
4480
4481 _claimed = false;
4482 }
4483
4484 CMTask::CMTask(uint worker_id,
4485 ConcurrentMark* cm,
4486 size_t* marked_bytes,
4487 BitMap* card_bm,
4488 CMTaskQueue* task_queue,
4489 CMTaskQueueSet* task_queues)
4490 : _g1h(G1CollectedHeap::heap()),
4491 _worker_id(worker_id), _cm(cm),
4492 _claimed(false),
4493 _nextMarkBitMap(NULL), _hash_seed(17),
4494 _task_queue(task_queue),
4495 _task_queues(task_queues),
4496 _cm_oop_closure(NULL),
4497 _marked_bytes_array(marked_bytes),
4498 _card_bm(card_bm) {
4499 guarantee(task_queue != NULL, "invariant");
4500 guarantee(task_queues != NULL, "invariant");
4501
4502 statsOnly( _clock_due_to_scanning = 0;
4503 _clock_due_to_marking = 0 );
4504
4505 _marking_step_diffs_ms.add(0.5);
4506 }
4507
4508 // These are formatting macros that are used below to ensure
4509 // consistent formatting. The *_H_* versions are used to format the
4510 // header for a particular value and they should be kept consistent
4511 // with the corresponding macro. Also note that most of the macros add
4512 // the necessary white space (as a prefix) which makes them a bit
4513 // easier to compose.
4514
4515 // All the output lines are prefixed with this string to be able to
4516 // identify them easily in a large log file.
4517 #define G1PPRL_LINE_PREFIX "###"
4518
4519 #define G1PPRL_ADDR_BASE_FORMAT " "PTR_FORMAT"-"PTR_FORMAT
4520 #ifdef _LP64
4521 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
4522 #else // _LP64
4523 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
4524 #endif // _LP64
4525
4526 // For per-region info
4527 #define G1PPRL_TYPE_FORMAT " %-4s"
4528 #define G1PPRL_TYPE_H_FORMAT " %4s"
4529 #define G1PPRL_BYTE_FORMAT " "SIZE_FORMAT_W(9)
4530 #define G1PPRL_BYTE_H_FORMAT " %9s"
4531 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
4532 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
4533
4534 // For summary info
4535 #define G1PPRL_SUM_ADDR_FORMAT(tag) " "tag":"G1PPRL_ADDR_BASE_FORMAT
4536 #define G1PPRL_SUM_BYTE_FORMAT(tag) " "tag": "SIZE_FORMAT
4537 #define G1PPRL_SUM_MB_FORMAT(tag) " "tag": %1.2f MB"
4538 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag)" / %1.2f %%"
4539
4540 G1PrintRegionLivenessInfoClosure::
4541 G1PrintRegionLivenessInfoClosure(outputStream* out, const char* phase_name)
4542 : _out(out),
4543 _total_used_bytes(0), _total_capacity_bytes(0),
4544 _total_prev_live_bytes(0), _total_next_live_bytes(0),
4545 _hum_used_bytes(0), _hum_capacity_bytes(0),
4546 _hum_prev_live_bytes(0), _hum_next_live_bytes(0),
4547 _total_remset_bytes(0), _total_strong_code_roots_bytes(0) {
4548 G1CollectedHeap* g1h = G1CollectedHeap::heap();
4549 MemRegion g1_reserved = g1h->g1_reserved();
4550 double now = os::elapsedTime();
4551
4552 // Print the header of the output.
4553 _out->cr();
4554 _out->print_cr(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
4555 _out->print_cr(G1PPRL_LINE_PREFIX" HEAP"
4556 G1PPRL_SUM_ADDR_FORMAT("reserved")
4557 G1PPRL_SUM_BYTE_FORMAT("region-size"),
4558 p2i(g1_reserved.start()), p2i(g1_reserved.end()),
4559 HeapRegion::GrainBytes);
4560 _out->print_cr(G1PPRL_LINE_PREFIX);
4561 _out->print_cr(G1PPRL_LINE_PREFIX
4562 G1PPRL_TYPE_H_FORMAT
4563 G1PPRL_ADDR_BASE_H_FORMAT
4564 G1PPRL_BYTE_H_FORMAT
4565 G1PPRL_BYTE_H_FORMAT
4566 G1PPRL_BYTE_H_FORMAT
4567 G1PPRL_DOUBLE_H_FORMAT
4568 G1PPRL_BYTE_H_FORMAT
4569 G1PPRL_BYTE_H_FORMAT,
4570 "type", "address-range",
4571 "used", "prev-live", "next-live", "gc-eff",
4572 "remset", "code-roots");
4573 _out->print_cr(G1PPRL_LINE_PREFIX
4574 G1PPRL_TYPE_H_FORMAT
4575 G1PPRL_ADDR_BASE_H_FORMAT
4576 G1PPRL_BYTE_H_FORMAT
4577 G1PPRL_BYTE_H_FORMAT
4578 G1PPRL_BYTE_H_FORMAT
4579 G1PPRL_DOUBLE_H_FORMAT
4580 G1PPRL_BYTE_H_FORMAT
4581 G1PPRL_BYTE_H_FORMAT,
4582 "", "",
4583 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)",
4584 "(bytes)", "(bytes)");
4585 }
4586
4587 // It takes as a parameter a reference to one of the _hum_* fields, it
4588 // deduces the corresponding value for a region in a humongous region
4589 // series (either the region size, or what's left if the _hum_* field
4590 // is < the region size), and updates the _hum_* field accordingly.
4591 size_t G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* hum_bytes) {
4592 size_t bytes = 0;
4593 // The > 0 check is to deal with the prev and next live bytes which
4594 // could be 0.
4595 if (*hum_bytes > 0) {
4596 bytes = MIN2(HeapRegion::GrainBytes, *hum_bytes);
4597 *hum_bytes -= bytes;
4598 }
4599 return bytes;
4600 }
4601
4602 // It deduces the values for a region in a humongous region series
4603 // from the _hum_* fields and updates those accordingly. It assumes
4604 // that that _hum_* fields have already been set up from the "starts
4605 // humongous" region and we visit the regions in address order.
4606 void G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* used_bytes,
4607 size_t* capacity_bytes,
4608 size_t* prev_live_bytes,
4609 size_t* next_live_bytes) {
4610 assert(_hum_used_bytes > 0 && _hum_capacity_bytes > 0, "pre-condition");
4611 *used_bytes = get_hum_bytes(&_hum_used_bytes);
4612 *capacity_bytes = get_hum_bytes(&_hum_capacity_bytes);
4613 *prev_live_bytes = get_hum_bytes(&_hum_prev_live_bytes);
4614 *next_live_bytes = get_hum_bytes(&_hum_next_live_bytes);
4615 }
4616
4617 bool G1PrintRegionLivenessInfoClosure::doHeapRegion(HeapRegion* r) {
4618 const char* type = r->get_type_str();
4619 HeapWord* bottom = r->bottom();
4620 HeapWord* end = r->end();
4621 size_t capacity_bytes = r->capacity();
4622 size_t used_bytes = r->used();
4623 size_t prev_live_bytes = r->live_bytes();
4624 size_t next_live_bytes = r->next_live_bytes();
4625 double gc_eff = r->gc_efficiency();
4626 size_t remset_bytes = r->rem_set()->mem_size();
4627 size_t strong_code_roots_bytes = r->rem_set()->strong_code_roots_mem_size();
4628
4629 if (r->is_starts_humongous()) {
4630 assert(_hum_used_bytes == 0 && _hum_capacity_bytes == 0 &&
4631 _hum_prev_live_bytes == 0 && _hum_next_live_bytes == 0,
4632 "they should have been zeroed after the last time we used them");
4633 // Set up the _hum_* fields.
4634 _hum_capacity_bytes = capacity_bytes;
4635 _hum_used_bytes = used_bytes;
4636 _hum_prev_live_bytes = prev_live_bytes;
4637 _hum_next_live_bytes = next_live_bytes;
4638 get_hum_bytes(&used_bytes, &capacity_bytes,
4639 &prev_live_bytes, &next_live_bytes);
4640 end = bottom + HeapRegion::GrainWords;
4641 } else if (r->is_continues_humongous()) {
4642 get_hum_bytes(&used_bytes, &capacity_bytes,
4643 &prev_live_bytes, &next_live_bytes);
4644 assert(end == bottom + HeapRegion::GrainWords, "invariant");
4645 }
4646
4647 _total_used_bytes += used_bytes;
4648 _total_capacity_bytes += capacity_bytes;
4649 _total_prev_live_bytes += prev_live_bytes;
4650 _total_next_live_bytes += next_live_bytes;
4651 _total_remset_bytes += remset_bytes;
4652 _total_strong_code_roots_bytes += strong_code_roots_bytes;
4653
4654 // Print a line for this particular region.
4655 _out->print_cr(G1PPRL_LINE_PREFIX
4656 G1PPRL_TYPE_FORMAT
4657 G1PPRL_ADDR_BASE_FORMAT
4658 G1PPRL_BYTE_FORMAT
4659 G1PPRL_BYTE_FORMAT
4660 G1PPRL_BYTE_FORMAT
4661 G1PPRL_DOUBLE_FORMAT
4662 G1PPRL_BYTE_FORMAT
4663 G1PPRL_BYTE_FORMAT,
4664 type, p2i(bottom), p2i(end),
4665 used_bytes, prev_live_bytes, next_live_bytes, gc_eff,
4666 remset_bytes, strong_code_roots_bytes);
4667
4668 return false;
4669 }
4670
4671 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
4672 // add static memory usages to remembered set sizes
4673 _total_remset_bytes += HeapRegionRemSet::fl_mem_size() + HeapRegionRemSet::static_mem_size();
4674 // Print the footer of the output.
4675 _out->print_cr(G1PPRL_LINE_PREFIX);
4676 _out->print_cr(G1PPRL_LINE_PREFIX
4677 " SUMMARY"
4678 G1PPRL_SUM_MB_FORMAT("capacity")
4679 G1PPRL_SUM_MB_PERC_FORMAT("used")
4680 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
4681 G1PPRL_SUM_MB_PERC_FORMAT("next-live")
4682 G1PPRL_SUM_MB_FORMAT("remset")
4683 G1PPRL_SUM_MB_FORMAT("code-roots"),
4684 bytes_to_mb(_total_capacity_bytes),
4685 bytes_to_mb(_total_used_bytes),
4686 perc(_total_used_bytes, _total_capacity_bytes),
4687 bytes_to_mb(_total_prev_live_bytes),
4688 perc(_total_prev_live_bytes, _total_capacity_bytes),
4689 bytes_to_mb(_total_next_live_bytes),
4690 perc(_total_next_live_bytes, _total_capacity_bytes),
4691 bytes_to_mb(_total_remset_bytes),
4692 bytes_to_mb(_total_strong_code_roots_bytes));
4693 _out->cr();
4694 }