rev 2891 : 7120038: G1: ParallelGCThreads==0 is broken
Summary: Running G1 with ParallelGCThreads==0 results in various crashes and asserts. Most of these are caused by unguarded references to the worker threads array or an incorrect number of active workers.
Reviewed-by:
1 /*
2 * Copyright (c) 2001, 2011, 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/symbolTable.hpp"
27 #include "gc_implementation/g1/concurrentMark.inline.hpp"
28 #include "gc_implementation/g1/concurrentMarkThread.inline.hpp"
29 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp"
30 #include "gc_implementation/g1/g1CollectorPolicy.hpp"
31 #include "gc_implementation/g1/g1ErgoVerbose.hpp"
32 #include "gc_implementation/g1/g1OopClosures.inline.hpp"
33 #include "gc_implementation/g1/g1RemSet.hpp"
34 #include "gc_implementation/g1/heapRegionRemSet.hpp"
35 #include "gc_implementation/g1/heapRegionSeq.inline.hpp"
36 #include "gc_implementation/shared/vmGCOperations.hpp"
37 #include "memory/genOopClosures.inline.hpp"
38 #include "memory/referencePolicy.hpp"
39 #include "memory/resourceArea.hpp"
40 #include "oops/oop.inline.hpp"
41 #include "runtime/handles.inline.hpp"
42 #include "runtime/java.hpp"
43
44 //
45 // CMS Bit Map Wrapper
46
47 CMBitMapRO::CMBitMapRO(ReservedSpace rs, int shifter) :
48 _bm((uintptr_t*)NULL,0),
49 _shifter(shifter) {
50 _bmStartWord = (HeapWord*)(rs.base());
51 _bmWordSize = rs.size()/HeapWordSize; // rs.size() is in bytes
52 ReservedSpace brs(ReservedSpace::allocation_align_size_up(
53 (_bmWordSize >> (_shifter + LogBitsPerByte)) + 1));
54
55 guarantee(brs.is_reserved(), "couldn't allocate CMS bit map");
56 // For now we'll just commit all of the bit map up fromt.
57 // Later on we'll try to be more parsimonious with swap.
58 guarantee(_virtual_space.initialize(brs, brs.size()),
59 "couldn't reseve backing store for CMS bit map");
60 assert(_virtual_space.committed_size() == brs.size(),
61 "didn't reserve backing store for all of CMS bit map?");
62 _bm.set_map((uintptr_t*)_virtual_space.low());
63 assert(_virtual_space.committed_size() << (_shifter + LogBitsPerByte) >=
64 _bmWordSize, "inconsistency in bit map sizing");
65 _bm.set_size(_bmWordSize >> _shifter);
66 }
67
68 HeapWord* CMBitMapRO::getNextMarkedWordAddress(HeapWord* addr,
69 HeapWord* limit) const {
70 // First we must round addr *up* to a possible object boundary.
71 addr = (HeapWord*)align_size_up((intptr_t)addr,
72 HeapWordSize << _shifter);
73 size_t addrOffset = heapWordToOffset(addr);
74 if (limit == NULL) {
75 limit = _bmStartWord + _bmWordSize;
76 }
77 size_t limitOffset = heapWordToOffset(limit);
78 size_t nextOffset = _bm.get_next_one_offset(addrOffset, limitOffset);
79 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
80 assert(nextAddr >= addr, "get_next_one postcondition");
81 assert(nextAddr == limit || isMarked(nextAddr),
82 "get_next_one postcondition");
83 return nextAddr;
84 }
85
86 HeapWord* CMBitMapRO::getNextUnmarkedWordAddress(HeapWord* addr,
87 HeapWord* limit) const {
88 size_t addrOffset = heapWordToOffset(addr);
89 if (limit == NULL) {
90 limit = _bmStartWord + _bmWordSize;
91 }
92 size_t limitOffset = heapWordToOffset(limit);
93 size_t nextOffset = _bm.get_next_zero_offset(addrOffset, limitOffset);
94 HeapWord* nextAddr = offsetToHeapWord(nextOffset);
95 assert(nextAddr >= addr, "get_next_one postcondition");
96 assert(nextAddr == limit || !isMarked(nextAddr),
97 "get_next_one postcondition");
98 return nextAddr;
99 }
100
101 int CMBitMapRO::heapWordDiffToOffsetDiff(size_t diff) const {
102 assert((diff & ((1 << _shifter) - 1)) == 0, "argument check");
103 return (int) (diff >> _shifter);
104 }
105
106 bool CMBitMapRO::iterate(BitMapClosure* cl, MemRegion mr) {
107 HeapWord* left = MAX2(_bmStartWord, mr.start());
108 HeapWord* right = MIN2(_bmStartWord + _bmWordSize, mr.end());
109 if (right > left) {
110 // Right-open interval [leftOffset, rightOffset).
111 return _bm.iterate(cl, heapWordToOffset(left), heapWordToOffset(right));
112 } else {
113 return true;
114 }
115 }
116
117 void CMBitMapRO::mostly_disjoint_range_union(BitMap* from_bitmap,
118 size_t from_start_index,
119 HeapWord* to_start_word,
120 size_t word_num) {
121 _bm.mostly_disjoint_range_union(from_bitmap,
122 from_start_index,
123 heapWordToOffset(to_start_word),
124 word_num);
125 }
126
127 #ifndef PRODUCT
128 bool CMBitMapRO::covers(ReservedSpace rs) const {
129 // assert(_bm.map() == _virtual_space.low(), "map inconsistency");
130 assert(((size_t)_bm.size() * (size_t)(1 << _shifter)) == _bmWordSize,
131 "size inconsistency");
132 return _bmStartWord == (HeapWord*)(rs.base()) &&
133 _bmWordSize == rs.size()>>LogHeapWordSize;
134 }
135 #endif
136
137 void CMBitMap::clearAll() {
138 _bm.clear();
139 return;
140 }
141
142 void CMBitMap::markRange(MemRegion mr) {
143 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
144 assert(!mr.is_empty(), "unexpected empty region");
145 assert((offsetToHeapWord(heapWordToOffset(mr.end())) ==
146 ((HeapWord *) mr.end())),
147 "markRange memory region end is not card aligned");
148 // convert address range into offset range
149 _bm.at_put_range(heapWordToOffset(mr.start()),
150 heapWordToOffset(mr.end()), true);
151 }
152
153 void CMBitMap::clearRange(MemRegion mr) {
154 mr.intersection(MemRegion(_bmStartWord, _bmWordSize));
155 assert(!mr.is_empty(), "unexpected empty region");
156 // convert address range into offset range
157 _bm.at_put_range(heapWordToOffset(mr.start()),
158 heapWordToOffset(mr.end()), false);
159 }
160
161 MemRegion CMBitMap::getAndClearMarkedRegion(HeapWord* addr,
162 HeapWord* end_addr) {
163 HeapWord* start = getNextMarkedWordAddress(addr);
164 start = MIN2(start, end_addr);
165 HeapWord* end = getNextUnmarkedWordAddress(start);
166 end = MIN2(end, end_addr);
167 assert(start <= end, "Consistency check");
168 MemRegion mr(start, end);
169 if (!mr.is_empty()) {
170 clearRange(mr);
171 }
172 return mr;
173 }
174
175 CMMarkStack::CMMarkStack(ConcurrentMark* cm) :
176 _base(NULL), _cm(cm)
177 #ifdef ASSERT
178 , _drain_in_progress(false)
179 , _drain_in_progress_yields(false)
180 #endif
181 {}
182
183 void CMMarkStack::allocate(size_t size) {
184 _base = NEW_C_HEAP_ARRAY(oop, size);
185 if (_base == NULL) {
186 vm_exit_during_initialization("Failed to allocate "
187 "CM region mark stack");
188 }
189 _index = 0;
190 _capacity = (jint) size;
191 _oops_do_bound = -1;
192 NOT_PRODUCT(_max_depth = 0);
193 }
194
195 CMMarkStack::~CMMarkStack() {
196 if (_base != NULL) {
197 FREE_C_HEAP_ARRAY(oop, _base);
198 }
199 }
200
201 void CMMarkStack::par_push(oop ptr) {
202 while (true) {
203 if (isFull()) {
204 _overflow = true;
205 return;
206 }
207 // Otherwise...
208 jint index = _index;
209 jint next_index = index+1;
210 jint res = Atomic::cmpxchg(next_index, &_index, index);
211 if (res == index) {
212 _base[index] = ptr;
213 // Note that we don't maintain this atomically. We could, but it
214 // doesn't seem necessary.
215 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index));
216 return;
217 }
218 // Otherwise, we need to try again.
219 }
220 }
221
222 void CMMarkStack::par_adjoin_arr(oop* ptr_arr, int n) {
223 while (true) {
224 if (isFull()) {
225 _overflow = true;
226 return;
227 }
228 // Otherwise...
229 jint index = _index;
230 jint next_index = index + n;
231 if (next_index > _capacity) {
232 _overflow = true;
233 return;
234 }
235 jint res = Atomic::cmpxchg(next_index, &_index, index);
236 if (res == index) {
237 for (int i = 0; i < n; i++) {
238 int ind = index + i;
239 assert(ind < _capacity, "By overflow test above.");
240 _base[ind] = ptr_arr[i];
241 }
242 NOT_PRODUCT(_max_depth = MAX2(_max_depth, next_index));
243 return;
244 }
245 // Otherwise, we need to try again.
246 }
247 }
248
249
250 void CMMarkStack::par_push_arr(oop* ptr_arr, int n) {
251 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
252 jint start = _index;
253 jint next_index = start + n;
254 if (next_index > _capacity) {
255 _overflow = true;
256 return;
257 }
258 // Otherwise.
259 _index = next_index;
260 for (int i = 0; i < n; i++) {
261 int ind = start + i;
262 assert(ind < _capacity, "By overflow test above.");
263 _base[ind] = ptr_arr[i];
264 }
265 }
266
267
268 bool CMMarkStack::par_pop_arr(oop* ptr_arr, int max, int* n) {
269 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
270 jint index = _index;
271 if (index == 0) {
272 *n = 0;
273 return false;
274 } else {
275 int k = MIN2(max, index);
276 jint new_ind = index - k;
277 for (int j = 0; j < k; j++) {
278 ptr_arr[j] = _base[new_ind + j];
279 }
280 _index = new_ind;
281 *n = k;
282 return true;
283 }
284 }
285
286
287 CMRegionStack::CMRegionStack() : _base(NULL) {}
288
289 void CMRegionStack::allocate(size_t size) {
290 _base = NEW_C_HEAP_ARRAY(MemRegion, size);
291 if (_base == NULL) {
292 vm_exit_during_initialization("Failed to allocate CM region mark stack");
293 }
294 _index = 0;
295 _capacity = (jint) size;
296 }
297
298 CMRegionStack::~CMRegionStack() {
299 if (_base != NULL) {
300 FREE_C_HEAP_ARRAY(oop, _base);
301 }
302 }
303
304 void CMRegionStack::push_lock_free(MemRegion mr) {
305 assert(mr.word_size() > 0, "Precondition");
306 while (true) {
307 jint index = _index;
308
309 if (index >= _capacity) {
310 _overflow = true;
311 return;
312 }
313 // Otherwise...
314 jint next_index = index+1;
315 jint res = Atomic::cmpxchg(next_index, &_index, index);
316 if (res == index) {
317 _base[index] = mr;
318 return;
319 }
320 // Otherwise, we need to try again.
321 }
322 }
323
324 // Lock-free pop of the region stack. Called during the concurrent
325 // marking / remark phases. Should only be called in tandem with
326 // other lock-free pops.
327 MemRegion CMRegionStack::pop_lock_free() {
328 while (true) {
329 jint index = _index;
330
331 if (index == 0) {
332 return MemRegion();
333 }
334 // Otherwise...
335 jint next_index = index-1;
336 jint res = Atomic::cmpxchg(next_index, &_index, index);
337 if (res == index) {
338 MemRegion mr = _base[next_index];
339 if (mr.start() != NULL) {
340 assert(mr.end() != NULL, "invariant");
341 assert(mr.word_size() > 0, "invariant");
342 return mr;
343 } else {
344 // that entry was invalidated... let's skip it
345 assert(mr.end() == NULL, "invariant");
346 }
347 }
348 // Otherwise, we need to try again.
349 }
350 }
351
352 #if 0
353 // The routines that manipulate the region stack with a lock are
354 // not currently used. They should be retained, however, as a
355 // diagnostic aid.
356
357 void CMRegionStack::push_with_lock(MemRegion mr) {
358 assert(mr.word_size() > 0, "Precondition");
359 MutexLockerEx x(CMRegionStack_lock, Mutex::_no_safepoint_check_flag);
360
361 if (isFull()) {
362 _overflow = true;
363 return;
364 }
365
366 _base[_index] = mr;
367 _index += 1;
368 }
369
370 MemRegion CMRegionStack::pop_with_lock() {
371 MutexLockerEx x(CMRegionStack_lock, Mutex::_no_safepoint_check_flag);
372
373 while (true) {
374 if (_index == 0) {
375 return MemRegion();
376 }
377 _index -= 1;
378
379 MemRegion mr = _base[_index];
380 if (mr.start() != NULL) {
381 assert(mr.end() != NULL, "invariant");
382 assert(mr.word_size() > 0, "invariant");
383 return mr;
384 } else {
385 // that entry was invalidated... let's skip it
386 assert(mr.end() == NULL, "invariant");
387 }
388 }
389 }
390 #endif
391
392 bool CMRegionStack::invalidate_entries_into_cset() {
393 bool result = false;
394 G1CollectedHeap* g1h = G1CollectedHeap::heap();
395 for (int i = 0; i < _oops_do_bound; ++i) {
396 MemRegion mr = _base[i];
397 if (mr.start() != NULL) {
398 assert(mr.end() != NULL, "invariant");
399 assert(mr.word_size() > 0, "invariant");
400 HeapRegion* hr = g1h->heap_region_containing(mr.start());
401 assert(hr != NULL, "invariant");
402 if (hr->in_collection_set()) {
403 // The region points into the collection set
404 _base[i] = MemRegion();
405 result = true;
406 }
407 } else {
408 // that entry was invalidated... let's skip it
409 assert(mr.end() == NULL, "invariant");
410 }
411 }
412 return result;
413 }
414
415 template<class OopClosureClass>
416 bool CMMarkStack::drain(OopClosureClass* cl, CMBitMap* bm, bool yield_after) {
417 assert(!_drain_in_progress || !_drain_in_progress_yields || yield_after
418 || SafepointSynchronize::is_at_safepoint(),
419 "Drain recursion must be yield-safe.");
420 bool res = true;
421 debug_only(_drain_in_progress = true);
422 debug_only(_drain_in_progress_yields = yield_after);
423 while (!isEmpty()) {
424 oop newOop = pop();
425 assert(G1CollectedHeap::heap()->is_in_reserved(newOop), "Bad pop");
426 assert(newOop->is_oop(), "Expected an oop");
427 assert(bm == NULL || bm->isMarked((HeapWord*)newOop),
428 "only grey objects on this stack");
429 // iterate over the oops in this oop, marking and pushing
430 // the ones in CMS generation.
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::oops_do(OopClosure* f) {
442 if (_index == 0) return;
443 assert(_oops_do_bound != -1 && _oops_do_bound <= _index,
444 "Bound must be set.");
445 for (int i = 0; i < _oops_do_bound; i++) {
446 f->do_oop(&_base[i]);
447 }
448 _oops_do_bound = -1;
449 }
450
451 bool ConcurrentMark::not_yet_marked(oop obj) const {
452 return (_g1h->is_obj_ill(obj)
453 || (_g1h->is_in_permanent(obj)
454 && !nextMarkBitMap()->isMarked((HeapWord*)obj)));
455 }
456
457 #ifdef _MSC_VER // the use of 'this' below gets a warning, make it go away
458 #pragma warning( disable:4355 ) // 'this' : used in base member initializer list
459 #endif // _MSC_VER
460
461 size_t ConcurrentMark::scale_parallel_threads(size_t n_par_threads) {
462 return MAX2((n_par_threads + 2) / 4, (size_t)1);
463 }
464
465 ConcurrentMark::ConcurrentMark(ReservedSpace rs,
466 int max_regions) :
467 _markBitMap1(rs, MinObjAlignment - 1),
468 _markBitMap2(rs, MinObjAlignment - 1),
469
470 _parallel_marking_threads(0),
471 _max_parallel_marking_threads(0),
472 _sleep_factor(0.0),
473 _marking_task_overhead(1.0),
474 _cleanup_sleep_factor(0.0),
475 _cleanup_task_overhead(1.0),
476 _cleanup_list("Cleanup List"),
477 _region_bm(max_regions, false /* in_resource_area*/),
478 _card_bm((rs.size() + CardTableModRefBS::card_size - 1) >>
479 CardTableModRefBS::card_shift,
480 false /* in_resource_area*/),
481 _prevMarkBitMap(&_markBitMap1),
482 _nextMarkBitMap(&_markBitMap2),
483 _at_least_one_mark_complete(false),
484
485 _markStack(this),
486 _regionStack(),
487 // _finger set in set_non_marking_state
488
489 _max_task_num(MAX2(ParallelGCThreads, (size_t)1)),
490 // _active_tasks set in set_non_marking_state
491 // _tasks set inside the constructor
492 _task_queues(new CMTaskQueueSet((int) _max_task_num)),
493 _terminator(ParallelTaskTerminator((int) _max_task_num, _task_queues)),
494
495 _has_overflown(false),
496 _concurrent(false),
497 _has_aborted(false),
498 _restart_for_overflow(false),
499 _concurrent_marking_in_progress(false),
500 _should_gray_objects(false),
501
502 // _verbose_level set below
503
504 _init_times(),
505 _remark_times(), _remark_mark_times(), _remark_weak_ref_times(),
506 _cleanup_times(),
507 _total_counting_time(0.0),
508 _total_rs_scrub_time(0.0),
509
510 _parallel_workers(NULL) {
511 CMVerboseLevel verbose_level = (CMVerboseLevel) G1MarkingVerboseLevel;
512 if (verbose_level < no_verbose) {
513 verbose_level = no_verbose;
514 }
515 if (verbose_level > high_verbose) {
516 verbose_level = high_verbose;
517 }
518 _verbose_level = verbose_level;
519
520 if (verbose_low()) {
521 gclog_or_tty->print_cr("[global] init, heap start = "PTR_FORMAT", "
522 "heap end = "PTR_FORMAT, _heap_start, _heap_end);
523 }
524
525 _markStack.allocate(MarkStackSize);
526 _regionStack.allocate(G1MarkRegionStackSize);
527
528 // Create & start a ConcurrentMark thread.
529 _cmThread = new ConcurrentMarkThread(this);
530 assert(cmThread() != NULL, "CM Thread should have been created");
531 assert(cmThread()->cm() != NULL, "CM Thread should refer to this cm");
532
533 _g1h = G1CollectedHeap::heap();
534 assert(CGC_lock != NULL, "Where's the CGC_lock?");
535 assert(_markBitMap1.covers(rs), "_markBitMap1 inconsistency");
536 assert(_markBitMap2.covers(rs), "_markBitMap2 inconsistency");
537
538 SATBMarkQueueSet& satb_qs = JavaThread::satb_mark_queue_set();
539 satb_qs.set_buffer_size(G1SATBBufferSize);
540
541 _tasks = NEW_C_HEAP_ARRAY(CMTask*, _max_task_num);
542 _accum_task_vtime = NEW_C_HEAP_ARRAY(double, _max_task_num);
543
544 // so that the assertion in MarkingTaskQueue::task_queue doesn't fail
545 _active_tasks = _max_task_num;
546 for (int i = 0; i < (int) _max_task_num; ++i) {
547 CMTaskQueue* task_queue = new CMTaskQueue();
548 task_queue->initialize();
549 _task_queues->register_queue(i, task_queue);
550
551 _tasks[i] = new CMTask(i, this, task_queue, _task_queues);
552 _accum_task_vtime[i] = 0.0;
553 }
554
555 if (ConcGCThreads > ParallelGCThreads) {
556 vm_exit_during_initialization("Can't have more ConcGCThreads "
557 "than ParallelGCThreads.");
558 }
559 if (ParallelGCThreads == 0) {
560 // if we are not running with any parallel GC threads we will not
561 // spawn any marking threads either
562 _parallel_marking_threads = 0;
563 _max_parallel_marking_threads = 0;
564 _sleep_factor = 0.0;
565 _marking_task_overhead = 1.0;
566 } else {
567 if (ConcGCThreads > 0) {
568 // notice that ConcGCThreads overwrites G1MarkingOverheadPercent
569 // if both are set
570
571 _parallel_marking_threads = ConcGCThreads;
572 _max_parallel_marking_threads = _parallel_marking_threads;
573 _sleep_factor = 0.0;
574 _marking_task_overhead = 1.0;
575 } else if (G1MarkingOverheadPercent > 0) {
576 // we will calculate the number of parallel marking threads
577 // based on a target overhead with respect to the soft real-time
578 // goal
579
580 double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
581 double overall_cm_overhead =
582 (double) MaxGCPauseMillis * marking_overhead /
583 (double) GCPauseIntervalMillis;
584 double cpu_ratio = 1.0 / (double) os::processor_count();
585 double marking_thread_num = ceil(overall_cm_overhead / cpu_ratio);
586 double marking_task_overhead =
587 overall_cm_overhead / marking_thread_num *
588 (double) os::processor_count();
589 double sleep_factor =
590 (1.0 - marking_task_overhead) / marking_task_overhead;
591
592 _parallel_marking_threads = (size_t) marking_thread_num;
593 _max_parallel_marking_threads = _parallel_marking_threads;
594 _sleep_factor = sleep_factor;
595 _marking_task_overhead = marking_task_overhead;
596 } else {
597 _parallel_marking_threads = scale_parallel_threads(ParallelGCThreads);
598 _max_parallel_marking_threads = _parallel_marking_threads;
599 _sleep_factor = 0.0;
600 _marking_task_overhead = 1.0;
601 }
602
603 if (parallel_marking_threads() > 1) {
604 _cleanup_task_overhead = 1.0;
605 } else {
606 _cleanup_task_overhead = marking_task_overhead();
607 }
608 _cleanup_sleep_factor =
609 (1.0 - cleanup_task_overhead()) / cleanup_task_overhead();
610
611 #if 0
612 gclog_or_tty->print_cr("Marking Threads %d", parallel_marking_threads());
613 gclog_or_tty->print_cr("CM Marking Task Overhead %1.4lf", marking_task_overhead());
614 gclog_or_tty->print_cr("CM Sleep Factor %1.4lf", sleep_factor());
615 gclog_or_tty->print_cr("CL Marking Task Overhead %1.4lf", cleanup_task_overhead());
616 gclog_or_tty->print_cr("CL Sleep Factor %1.4lf", cleanup_sleep_factor());
617 #endif
618
619 guarantee(parallel_marking_threads() > 0, "peace of mind");
620 _parallel_workers = new FlexibleWorkGang("G1 Parallel Marking Threads",
621 (int) _max_parallel_marking_threads, false, true);
622 if (_parallel_workers == NULL) {
623 vm_exit_during_initialization("Failed necessary allocation.");
624 } else {
625 _parallel_workers->initialize_workers();
626 }
627 }
628
629 // so that the call below can read a sensible value
630 _heap_start = (HeapWord*) rs.base();
631 set_non_marking_state();
632 }
633
634 void ConcurrentMark::update_g1_committed(bool force) {
635 // If concurrent marking is not in progress, then we do not need to
636 // update _heap_end. This has a subtle and important
637 // side-effect. Imagine that two evacuation pauses happen between
638 // marking completion and remark. The first one can grow the
639 // heap (hence now the finger is below the heap end). Then, the
640 // second one could unnecessarily push regions on the region
641 // stack. This causes the invariant that the region stack is empty
642 // at the beginning of remark to be false. By ensuring that we do
643 // not observe heap expansions after marking is complete, then we do
644 // not have this problem.
645 if (!concurrent_marking_in_progress() && !force) return;
646
647 MemRegion committed = _g1h->g1_committed();
648 assert(committed.start() == _heap_start, "start shouldn't change");
649 HeapWord* new_end = committed.end();
650 if (new_end > _heap_end) {
651 // The heap has been expanded.
652
653 _heap_end = new_end;
654 }
655 // Notice that the heap can also shrink. However, this only happens
656 // during a Full GC (at least currently) and the entire marking
657 // phase will bail out and the task will not be restarted. So, let's
658 // do nothing.
659 }
660
661 void ConcurrentMark::reset() {
662 // Starting values for these two. This should be called in a STW
663 // phase. CM will be notified of any future g1_committed expansions
664 // will be at the end of evacuation pauses, when tasks are
665 // inactive.
666 MemRegion committed = _g1h->g1_committed();
667 _heap_start = committed.start();
668 _heap_end = committed.end();
669
670 // Separated the asserts so that we know which one fires.
671 assert(_heap_start != NULL, "heap bounds should look ok");
672 assert(_heap_end != NULL, "heap bounds should look ok");
673 assert(_heap_start < _heap_end, "heap bounds should look ok");
674
675 // reset all the marking data structures and any necessary flags
676 clear_marking_state();
677
678 if (verbose_low()) {
679 gclog_or_tty->print_cr("[global] resetting");
680 }
681
682 // We do reset all of them, since different phases will use
683 // different number of active threads. So, it's easiest to have all
684 // of them ready.
685 for (int i = 0; i < (int) _max_task_num; ++i) {
686 _tasks[i]->reset(_nextMarkBitMap);
687 }
688
689 // we need this to make sure that the flag is on during the evac
690 // pause with initial mark piggy-backed
691 set_concurrent_marking_in_progress();
692 }
693
694 void ConcurrentMark::set_phase(size_t active_tasks, bool concurrent) {
695 assert(active_tasks <= _max_task_num, "we should not have more");
696
697 _active_tasks = active_tasks;
698 // Need to update the three data structures below according to the
699 // number of active threads for this phase.
700 _terminator = ParallelTaskTerminator((int) active_tasks, _task_queues);
701 _first_overflow_barrier_sync.set_n_workers((int) active_tasks);
702 _second_overflow_barrier_sync.set_n_workers((int) active_tasks);
703
704 _concurrent = concurrent;
705 // We propagate this to all tasks, not just the active ones.
706 for (int i = 0; i < (int) _max_task_num; ++i)
707 _tasks[i]->set_concurrent(concurrent);
708
709 if (concurrent) {
710 set_concurrent_marking_in_progress();
711 } else {
712 // We currently assume that the concurrent flag has been set to
713 // false before we start remark. At this point we should also be
714 // in a STW phase.
715 assert(!concurrent_marking_in_progress(), "invariant");
716 assert(_finger == _heap_end, "only way to get here");
717 update_g1_committed(true);
718 }
719 }
720
721 void ConcurrentMark::set_non_marking_state() {
722 // We set the global marking state to some default values when we're
723 // not doing marking.
724 clear_marking_state();
725 _active_tasks = 0;
726 clear_concurrent_marking_in_progress();
727 }
728
729 ConcurrentMark::~ConcurrentMark() {
730 for (int i = 0; i < (int) _max_task_num; ++i) {
731 delete _task_queues->queue(i);
732 delete _tasks[i];
733 }
734 delete _task_queues;
735 FREE_C_HEAP_ARRAY(CMTask*, _max_task_num);
736 }
737
738 // This closure is used to mark refs into the g1 generation
739 // from external roots in the CMS bit map.
740 // Called at the first checkpoint.
741 //
742
743 void ConcurrentMark::clearNextBitmap() {
744 G1CollectedHeap* g1h = G1CollectedHeap::heap();
745 G1CollectorPolicy* g1p = g1h->g1_policy();
746
747 // Make sure that the concurrent mark thread looks to still be in
748 // the current cycle.
749 guarantee(cmThread()->during_cycle(), "invariant");
750
751 // We are finishing up the current cycle by clearing the next
752 // marking bitmap and getting it ready for the next cycle. During
753 // this time no other cycle can start. So, let's make sure that this
754 // is the case.
755 guarantee(!g1h->mark_in_progress(), "invariant");
756
757 // clear the mark bitmap (no grey objects to start with).
758 // We need to do this in chunks and offer to yield in between
759 // each chunk.
760 HeapWord* start = _nextMarkBitMap->startWord();
761 HeapWord* end = _nextMarkBitMap->endWord();
762 HeapWord* cur = start;
763 size_t chunkSize = M;
764 while (cur < end) {
765 HeapWord* next = cur + chunkSize;
766 if (next > end) {
767 next = end;
768 }
769 MemRegion mr(cur,next);
770 _nextMarkBitMap->clearRange(mr);
771 cur = next;
772 do_yield_check();
773
774 // Repeat the asserts from above. We'll do them as asserts here to
775 // minimize their overhead on the product. However, we'll have
776 // them as guarantees at the beginning / end of the bitmap
777 // clearing to get some checking in the product.
778 assert(cmThread()->during_cycle(), "invariant");
779 assert(!g1h->mark_in_progress(), "invariant");
780 }
781
782 // Repeat the asserts from above.
783 guarantee(cmThread()->during_cycle(), "invariant");
784 guarantee(!g1h->mark_in_progress(), "invariant");
785 }
786
787 class NoteStartOfMarkHRClosure: public HeapRegionClosure {
788 public:
789 bool doHeapRegion(HeapRegion* r) {
790 if (!r->continuesHumongous()) {
791 r->note_start_of_marking(true);
792 }
793 return false;
794 }
795 };
796
797 void ConcurrentMark::checkpointRootsInitialPre() {
798 G1CollectedHeap* g1h = G1CollectedHeap::heap();
799 G1CollectorPolicy* g1p = g1h->g1_policy();
800
801 _has_aborted = false;
802
803 #ifndef PRODUCT
804 if (G1PrintReachableAtInitialMark) {
805 print_reachable("at-cycle-start",
806 VerifyOption_G1UsePrevMarking, true /* all */);
807 }
808 #endif
809
810 // Initialise marking structures. This has to be done in a STW phase.
811 reset();
812 }
813
814
815 void ConcurrentMark::checkpointRootsInitialPost() {
816 G1CollectedHeap* g1h = G1CollectedHeap::heap();
817
818 // If we force an overflow during remark, the remark operation will
819 // actually abort and we'll restart concurrent marking. If we always
820 // force an oveflow during remark we'll never actually complete the
821 // marking phase. So, we initilize this here, at the start of the
822 // cycle, so that at the remaining overflow number will decrease at
823 // every remark and we'll eventually not need to cause one.
824 force_overflow_stw()->init();
825
826 // For each region note start of marking.
827 NoteStartOfMarkHRClosure startcl;
828 g1h->heap_region_iterate(&startcl);
829
830 // Start Concurrent Marking weak-reference discovery.
831 ReferenceProcessor* rp = g1h->ref_processor_cm();
832 // enable ("weak") refs discovery
833 rp->enable_discovery(true /*verify_disabled*/, true /*verify_no_refs*/);
834 rp->setup_policy(false); // snapshot the soft ref policy to be used in this cycle
835
836 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
837 // This is the start of the marking cycle, we're expected all
838 // threads to have SATB queues with active set to false.
839 satb_mq_set.set_active_all_threads(true, /* new active value */
840 false /* expected_active */);
841
842 // update_g1_committed() will be called at the end of an evac pause
843 // when marking is on. So, it's also called at the end of the
844 // initial-mark pause to update the heap end, if the heap expands
845 // during it. No need to call it here.
846 }
847
848 /*
849 * Notice that in the next two methods, we actually leave the STS
850 * during the barrier sync and join it immediately afterwards. If we
851 * do not do this, the following deadlock can occur: one thread could
852 * be in the barrier sync code, waiting for the other thread to also
853 * sync up, whereas another one could be trying to yield, while also
854 * waiting for the other threads to sync up too.
855 *
856 * Note, however, that this code is also used during remark and in
857 * this case we should not attempt to leave / enter the STS, otherwise
858 * we'll either hit an asseert (debug / fastdebug) or deadlock
859 * (product). So we should only leave / enter the STS if we are
860 * operating concurrently.
861 *
862 * Because the thread that does the sync barrier has left the STS, it
863 * is possible to be suspended for a Full GC or an evacuation pause
864 * could occur. This is actually safe, since the entering the sync
865 * barrier is one of the last things do_marking_step() does, and it
866 * doesn't manipulate any data structures afterwards.
867 */
868
869 void ConcurrentMark::enter_first_sync_barrier(int task_num) {
870 if (verbose_low()) {
871 gclog_or_tty->print_cr("[%d] entering first barrier", task_num);
872 }
873
874 if (concurrent()) {
875 ConcurrentGCThread::stsLeave();
876 }
877 _first_overflow_barrier_sync.enter();
878 if (concurrent()) {
879 ConcurrentGCThread::stsJoin();
880 }
881 // at this point everyone should have synced up and not be doing any
882 // more work
883
884 if (verbose_low()) {
885 gclog_or_tty->print_cr("[%d] leaving first barrier", task_num);
886 }
887
888 // let task 0 do this
889 if (task_num == 0) {
890 // task 0 is responsible for clearing the global data structures
891 // We should be here because of an overflow. During STW we should
892 // not clear the overflow flag since we rely on it being true when
893 // we exit this method to abort the pause and restart concurent
894 // marking.
895 clear_marking_state(concurrent() /* clear_overflow */);
896 force_overflow()->update();
897
898 if (PrintGC) {
899 gclog_or_tty->date_stamp(PrintGCDateStamps);
900 gclog_or_tty->stamp(PrintGCTimeStamps);
901 gclog_or_tty->print_cr("[GC concurrent-mark-reset-for-overflow]");
902 }
903 }
904
905 // after this, each task should reset its own data structures then
906 // then go into the second barrier
907 }
908
909 void ConcurrentMark::enter_second_sync_barrier(int task_num) {
910 if (verbose_low()) {
911 gclog_or_tty->print_cr("[%d] entering second barrier", task_num);
912 }
913
914 if (concurrent()) {
915 ConcurrentGCThread::stsLeave();
916 }
917 _second_overflow_barrier_sync.enter();
918 if (concurrent()) {
919 ConcurrentGCThread::stsJoin();
920 }
921 // at this point everything should be re-initialised and ready to go
922
923 if (verbose_low()) {
924 gclog_or_tty->print_cr("[%d] leaving second barrier", task_num);
925 }
926 }
927
928 #ifndef PRODUCT
929 void ForceOverflowSettings::init() {
930 _num_remaining = G1ConcMarkForceOverflow;
931 _force = false;
932 update();
933 }
934
935 void ForceOverflowSettings::update() {
936 if (_num_remaining > 0) {
937 _num_remaining -= 1;
938 _force = true;
939 } else {
940 _force = false;
941 }
942 }
943
944 bool ForceOverflowSettings::should_force() {
945 if (_force) {
946 _force = false;
947 return true;
948 } else {
949 return false;
950 }
951 }
952 #endif // !PRODUCT
953
954 void ConcurrentMark::grayRoot(oop p) {
955 HeapWord* addr = (HeapWord*) p;
956 // We can't really check against _heap_start and _heap_end, since it
957 // is possible during an evacuation pause with piggy-backed
958 // initial-mark that the committed space is expanded during the
959 // pause without CM observing this change. So the assertions below
960 // is a bit conservative; but better than nothing.
961 assert(_g1h->g1_committed().contains(addr),
962 "address should be within the heap bounds");
963
964 if (!_nextMarkBitMap->isMarked(addr)) {
965 _nextMarkBitMap->parMark(addr);
966 }
967 }
968
969 void ConcurrentMark::grayRegionIfNecessary(MemRegion mr) {
970 // The objects on the region have already been marked "in bulk" by
971 // the caller. We only need to decide whether to push the region on
972 // the region stack or not.
973
974 if (!concurrent_marking_in_progress() || !_should_gray_objects) {
975 // We're done with marking and waiting for remark. We do not need to
976 // push anything else on the region stack.
977 return;
978 }
979
980 HeapWord* finger = _finger;
981
982 if (verbose_low()) {
983 gclog_or_tty->print_cr("[global] attempting to push "
984 "region ["PTR_FORMAT", "PTR_FORMAT"), finger is at "
985 PTR_FORMAT, mr.start(), mr.end(), finger);
986 }
987
988 if (mr.start() < finger) {
989 // The finger is always heap region aligned and it is not possible
990 // for mr to span heap regions.
991 assert(mr.end() <= finger, "invariant");
992
993 // Separated the asserts so that we know which one fires.
994 assert(mr.start() <= mr.end(),
995 "region boundaries should fall within the committed space");
996 assert(_heap_start <= mr.start(),
997 "region boundaries should fall within the committed space");
998 assert(mr.end() <= _heap_end,
999 "region boundaries should fall within the committed space");
1000 if (verbose_low()) {
1001 gclog_or_tty->print_cr("[global] region ["PTR_FORMAT", "PTR_FORMAT") "
1002 "below the finger, pushing it",
1003 mr.start(), mr.end());
1004 }
1005
1006 if (!region_stack_push_lock_free(mr)) {
1007 if (verbose_low()) {
1008 gclog_or_tty->print_cr("[global] region stack has overflown.");
1009 }
1010 }
1011 }
1012 }
1013
1014 void ConcurrentMark::markAndGrayObjectIfNecessary(oop p) {
1015 // The object is not marked by the caller. We need to at least mark
1016 // it and maybe push in on the stack.
1017
1018 HeapWord* addr = (HeapWord*)p;
1019 if (!_nextMarkBitMap->isMarked(addr)) {
1020 // We definitely need to mark it, irrespective whether we bail out
1021 // because we're done with marking.
1022 if (_nextMarkBitMap->parMark(addr)) {
1023 if (!concurrent_marking_in_progress() || !_should_gray_objects) {
1024 // If we're done with concurrent marking and we're waiting for
1025 // remark, then we're not pushing anything on the stack.
1026 return;
1027 }
1028
1029 // No OrderAccess:store_load() is needed. It is implicit in the
1030 // CAS done in parMark(addr) above
1031 HeapWord* finger = _finger;
1032
1033 if (addr < finger) {
1034 if (!mark_stack_push(oop(addr))) {
1035 if (verbose_low()) {
1036 gclog_or_tty->print_cr("[global] global stack overflow "
1037 "during parMark");
1038 }
1039 }
1040 }
1041 }
1042 }
1043 }
1044
1045 class CMConcurrentMarkingTask: public AbstractGangTask {
1046 private:
1047 ConcurrentMark* _cm;
1048 ConcurrentMarkThread* _cmt;
1049
1050 public:
1051 void work(int worker_i) {
1052 assert(Thread::current()->is_ConcurrentGC_thread(),
1053 "this should only be done by a conc GC thread");
1054 ResourceMark rm;
1055
1056 double start_vtime = os::elapsedVTime();
1057
1058 ConcurrentGCThread::stsJoin();
1059
1060 assert((size_t) worker_i < _cm->active_tasks(), "invariant");
1061 CMTask* the_task = _cm->task(worker_i);
1062 the_task->record_start_time();
1063 if (!_cm->has_aborted()) {
1064 do {
1065 double start_vtime_sec = os::elapsedVTime();
1066 double start_time_sec = os::elapsedTime();
1067 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
1068
1069 the_task->do_marking_step(mark_step_duration_ms,
1070 true /* do_stealing */,
1071 true /* do_termination */);
1072
1073 double end_time_sec = os::elapsedTime();
1074 double end_vtime_sec = os::elapsedVTime();
1075 double elapsed_vtime_sec = end_vtime_sec - start_vtime_sec;
1076 double elapsed_time_sec = end_time_sec - start_time_sec;
1077 _cm->clear_has_overflown();
1078
1079 bool ret = _cm->do_yield_check(worker_i);
1080
1081 jlong sleep_time_ms;
1082 if (!_cm->has_aborted() && the_task->has_aborted()) {
1083 sleep_time_ms =
1084 (jlong) (elapsed_vtime_sec * _cm->sleep_factor() * 1000.0);
1085 ConcurrentGCThread::stsLeave();
1086 os::sleep(Thread::current(), sleep_time_ms, false);
1087 ConcurrentGCThread::stsJoin();
1088 }
1089 double end_time2_sec = os::elapsedTime();
1090 double elapsed_time2_sec = end_time2_sec - start_time_sec;
1091
1092 #if 0
1093 gclog_or_tty->print_cr("CM: elapsed %1.4lf ms, sleep %1.4lf ms, "
1094 "overhead %1.4lf",
1095 elapsed_vtime_sec * 1000.0, (double) sleep_time_ms,
1096 the_task->conc_overhead(os::elapsedTime()) * 8.0);
1097 gclog_or_tty->print_cr("elapsed time %1.4lf ms, time 2: %1.4lf ms",
1098 elapsed_time_sec * 1000.0, elapsed_time2_sec * 1000.0);
1099 #endif
1100 } while (!_cm->has_aborted() && the_task->has_aborted());
1101 }
1102 the_task->record_end_time();
1103 guarantee(!the_task->has_aborted() || _cm->has_aborted(), "invariant");
1104
1105 ConcurrentGCThread::stsLeave();
1106
1107 double end_vtime = os::elapsedVTime();
1108 _cm->update_accum_task_vtime(worker_i, end_vtime - start_vtime);
1109 }
1110
1111 CMConcurrentMarkingTask(ConcurrentMark* cm,
1112 ConcurrentMarkThread* cmt) :
1113 AbstractGangTask("Concurrent Mark"), _cm(cm), _cmt(cmt) { }
1114
1115 ~CMConcurrentMarkingTask() { }
1116 };
1117
1118 // Calculates the number of active workers for a concurrent
1119 // phase.
1120 int ConcurrentMark::calc_parallel_marking_threads() {
1121
1122 size_t n_conc_workers;
1123 if (!G1CollectedHeap::use_parallel_gc_threads()) {
1124 n_conc_workers = 1;
1125 } else {
1126 if (!UseDynamicNumberOfGCThreads ||
1127 (!FLAG_IS_DEFAULT(ConcGCThreads) &&
1128 !ForceDynamicNumberOfGCThreads)) {
1129 n_conc_workers = max_parallel_marking_threads();
1130 } else {
1131 n_conc_workers =
1132 AdaptiveSizePolicy::calc_default_active_workers(
1133 max_parallel_marking_threads(),
1134 1, /* Minimum workers */
1135 parallel_marking_threads(),
1136 Threads::number_of_non_daemon_threads());
1137 // Don't scale down "n_conc_workers" by scale_parallel_threads() because
1138 // that scaling has already gone into "_max_parallel_marking_threads".
1139 }
1140 }
1141 assert(n_conc_workers > 0, "Always need at least 1");
1142 return (int) MAX2(n_conc_workers, (size_t) 1);
1143 }
1144
1145 void ConcurrentMark::markFromRoots() {
1146 // we might be tempted to assert that:
1147 // assert(asynch == !SafepointSynchronize::is_at_safepoint(),
1148 // "inconsistent argument?");
1149 // However that wouldn't be right, because it's possible that
1150 // a safepoint is indeed in progress as a younger generation
1151 // stop-the-world GC happens even as we mark in this generation.
1152
1153 _restart_for_overflow = false;
1154
1155 // Parallel task terminator is set in "set_phase()".
1156 force_overflow_conc()->init();
1157
1158 // _g1h has _n_par_threads
1159
1160 _parallel_marking_threads = calc_parallel_marking_threads();
1161 assert(parallel_marking_threads() <= max_parallel_marking_threads(),
1162 "Maximum number of marking threads exceeded");
1163 _parallel_workers->set_active_workers((int)_parallel_marking_threads);
1164 // Don't set _n_par_threads because it affects MT in proceess_strong_roots()
1165 // and the decisions on that MT processing is made elsewhere.
1166
1167 assert( _parallel_workers->active_workers() > 0, "Should have been set");
1168 set_phase(_parallel_workers->active_workers(), true /* concurrent */);
1169
1170 CMConcurrentMarkingTask markingTask(this, cmThread());
1171 if (parallel_marking_threads() > 0) {
1172 _parallel_workers->run_task(&markingTask);
1173 } else {
1174 markingTask.work(0);
1175 }
1176 print_stats();
1177 }
1178
1179 void ConcurrentMark::checkpointRootsFinal(bool clear_all_soft_refs) {
1180 // world is stopped at this checkpoint
1181 assert(SafepointSynchronize::is_at_safepoint(),
1182 "world should be stopped");
1183
1184 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1185
1186 // If a full collection has happened, we shouldn't do this.
1187 if (has_aborted()) {
1188 g1h->set_marking_complete(); // So bitmap clearing isn't confused
1189 return;
1190 }
1191
1192 SvcGCMarker sgcm(SvcGCMarker::OTHER);
1193
1194 if (VerifyDuringGC) {
1195 HandleMark hm; // handle scope
1196 gclog_or_tty->print(" VerifyDuringGC:(before)");
1197 Universe::heap()->prepare_for_verify();
1198 Universe::verify(/* allow dirty */ true,
1199 /* silent */ false,
1200 /* option */ VerifyOption_G1UsePrevMarking);
1201 }
1202
1203 G1CollectorPolicy* g1p = g1h->g1_policy();
1204 g1p->record_concurrent_mark_remark_start();
1205
1206 double start = os::elapsedTime();
1207
1208 checkpointRootsFinalWork();
1209
1210 double mark_work_end = os::elapsedTime();
1211
1212 weakRefsWork(clear_all_soft_refs);
1213
1214 if (has_overflown()) {
1215 // Oops. We overflowed. Restart concurrent marking.
1216 _restart_for_overflow = true;
1217 // Clear the flag. We do not need it any more.
1218 clear_has_overflown();
1219 if (G1TraceMarkStackOverflow) {
1220 gclog_or_tty->print_cr("\nRemark led to restart for overflow.");
1221 }
1222 } else {
1223 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
1224 // We're done with marking.
1225 // This is the end of the marking cycle, we're expected all
1226 // threads to have SATB queues with active set to true.
1227 satb_mq_set.set_active_all_threads(false, /* new active value */
1228 true /* expected_active */);
1229
1230 if (VerifyDuringGC) {
1231
1232 HandleMark hm; // handle scope
1233 gclog_or_tty->print(" VerifyDuringGC:(after)");
1234 Universe::heap()->prepare_for_verify();
1235 Universe::verify(/* allow dirty */ true,
1236 /* silent */ false,
1237 /* option */ VerifyOption_G1UseNextMarking);
1238 }
1239 assert(!restart_for_overflow(), "sanity");
1240 }
1241
1242 // Reset the marking state if marking completed
1243 if (!restart_for_overflow()) {
1244 set_non_marking_state();
1245 }
1246
1247 #if VERIFY_OBJS_PROCESSED
1248 _scan_obj_cl.objs_processed = 0;
1249 ThreadLocalObjQueue::objs_enqueued = 0;
1250 #endif
1251
1252 // Statistics
1253 double now = os::elapsedTime();
1254 _remark_mark_times.add((mark_work_end - start) * 1000.0);
1255 _remark_weak_ref_times.add((now - mark_work_end) * 1000.0);
1256 _remark_times.add((now - start) * 1000.0);
1257
1258 g1p->record_concurrent_mark_remark_end();
1259 }
1260
1261 #define CARD_BM_TEST_MODE 0
1262
1263 class CalcLiveObjectsClosure: public HeapRegionClosure {
1264
1265 CMBitMapRO* _bm;
1266 ConcurrentMark* _cm;
1267 bool _changed;
1268 bool _yield;
1269 size_t _words_done;
1270 size_t _tot_live;
1271 size_t _tot_used;
1272 size_t _regions_done;
1273 double _start_vtime_sec;
1274
1275 BitMap* _region_bm;
1276 BitMap* _card_bm;
1277 intptr_t _bottom_card_num;
1278 bool _final;
1279
1280 void mark_card_num_range(intptr_t start_card_num, intptr_t last_card_num) {
1281 for (intptr_t i = start_card_num; i <= last_card_num; i++) {
1282 #if CARD_BM_TEST_MODE
1283 guarantee(_card_bm->at(i - _bottom_card_num), "Should already be set.");
1284 #else
1285 _card_bm->par_at_put(i - _bottom_card_num, 1);
1286 #endif
1287 }
1288 }
1289
1290 public:
1291 CalcLiveObjectsClosure(bool final,
1292 CMBitMapRO *bm, ConcurrentMark *cm,
1293 BitMap* region_bm, BitMap* card_bm) :
1294 _bm(bm), _cm(cm), _changed(false), _yield(true),
1295 _words_done(0), _tot_live(0), _tot_used(0),
1296 _region_bm(region_bm), _card_bm(card_bm),_final(final),
1297 _regions_done(0), _start_vtime_sec(0.0)
1298 {
1299 _bottom_card_num =
1300 intptr_t(uintptr_t(G1CollectedHeap::heap()->reserved_region().start()) >>
1301 CardTableModRefBS::card_shift);
1302 }
1303
1304 // It takes a region that's not empty (i.e., it has at least one
1305 // live object in it and sets its corresponding bit on the region
1306 // bitmap to 1. If the region is "starts humongous" it will also set
1307 // to 1 the bits on the region bitmap that correspond to its
1308 // associated "continues humongous" regions.
1309 void set_bit_for_region(HeapRegion* hr) {
1310 assert(!hr->continuesHumongous(), "should have filtered those out");
1311
1312 size_t index = hr->hrs_index();
1313 if (!hr->startsHumongous()) {
1314 // Normal (non-humongous) case: just set the bit.
1315 _region_bm->par_at_put((BitMap::idx_t) index, true);
1316 } else {
1317 // Starts humongous case: calculate how many regions are part of
1318 // this humongous region and then set the bit range. It might
1319 // have been a bit more efficient to look at the object that
1320 // spans these humongous regions to calculate their number from
1321 // the object's size. However, it's a good idea to calculate
1322 // this based on the metadata itself, and not the region
1323 // contents, so that this code is not aware of what goes into
1324 // the humongous regions (in case this changes in the future).
1325 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1326 size_t end_index = index + 1;
1327 while (end_index < g1h->n_regions()) {
1328 HeapRegion* chr = g1h->region_at(end_index);
1329 if (!chr->continuesHumongous()) break;
1330 end_index += 1;
1331 }
1332 _region_bm->par_at_put_range((BitMap::idx_t) index,
1333 (BitMap::idx_t) end_index, true);
1334 }
1335 }
1336
1337 bool doHeapRegion(HeapRegion* hr) {
1338 if (!_final && _regions_done == 0) {
1339 _start_vtime_sec = os::elapsedVTime();
1340 }
1341
1342 if (hr->continuesHumongous()) {
1343 // We will ignore these here and process them when their
1344 // associated "starts humongous" region is processed (see
1345 // set_bit_for_heap_region()). Note that we cannot rely on their
1346 // associated "starts humongous" region to have their bit set to
1347 // 1 since, due to the region chunking in the parallel region
1348 // iteration, a "continues humongous" region might be visited
1349 // before its associated "starts humongous".
1350 return false;
1351 }
1352
1353 HeapWord* nextTop = hr->next_top_at_mark_start();
1354 HeapWord* start = hr->top_at_conc_mark_count();
1355 assert(hr->bottom() <= start && start <= hr->end() &&
1356 hr->bottom() <= nextTop && nextTop <= hr->end() &&
1357 start <= nextTop,
1358 "Preconditions.");
1359 // Otherwise, record the number of word's we'll examine.
1360 size_t words_done = (nextTop - start);
1361 // Find the first marked object at or after "start".
1362 start = _bm->getNextMarkedWordAddress(start, nextTop);
1363 size_t marked_bytes = 0;
1364
1365 // Below, the term "card num" means the result of shifting an address
1366 // by the card shift -- address 0 corresponds to card number 0. One
1367 // must subtract the card num of the bottom of the heap to obtain a
1368 // card table index.
1369 // The first card num of the sequence of live cards currently being
1370 // constructed. -1 ==> no sequence.
1371 intptr_t start_card_num = -1;
1372 // The last card num of the sequence of live cards currently being
1373 // constructed. -1 ==> no sequence.
1374 intptr_t last_card_num = -1;
1375
1376 while (start < nextTop) {
1377 if (_yield && _cm->do_yield_check()) {
1378 // We yielded. It might be for a full collection, in which case
1379 // all bets are off; terminate the traversal.
1380 if (_cm->has_aborted()) {
1381 _changed = false;
1382 return true;
1383 } else {
1384 // Otherwise, it might be a collection pause, and the region
1385 // we're looking at might be in the collection set. We'll
1386 // abandon this region.
1387 return false;
1388 }
1389 }
1390 oop obj = oop(start);
1391 int obj_sz = obj->size();
1392 // The card num of the start of the current object.
1393 intptr_t obj_card_num =
1394 intptr_t(uintptr_t(start) >> CardTableModRefBS::card_shift);
1395
1396 HeapWord* obj_last = start + obj_sz - 1;
1397 intptr_t obj_last_card_num =
1398 intptr_t(uintptr_t(obj_last) >> CardTableModRefBS::card_shift);
1399
1400 if (obj_card_num != last_card_num) {
1401 if (start_card_num == -1) {
1402 assert(last_card_num == -1, "Both or neither.");
1403 start_card_num = obj_card_num;
1404 } else {
1405 assert(last_card_num != -1, "Both or neither.");
1406 assert(obj_card_num >= last_card_num, "Inv");
1407 if ((obj_card_num - last_card_num) > 1) {
1408 // Mark the last run, and start a new one.
1409 mark_card_num_range(start_card_num, last_card_num);
1410 start_card_num = obj_card_num;
1411 }
1412 }
1413 #if CARD_BM_TEST_MODE
1414 /*
1415 gclog_or_tty->print_cr("Setting bits from %d/%d.",
1416 obj_card_num - _bottom_card_num,
1417 obj_last_card_num - _bottom_card_num);
1418 */
1419 for (intptr_t j = obj_card_num; j <= obj_last_card_num; j++) {
1420 _card_bm->par_at_put(j - _bottom_card_num, 1);
1421 }
1422 #endif
1423 }
1424 // In any case, we set the last card num.
1425 last_card_num = obj_last_card_num;
1426
1427 marked_bytes += (size_t)obj_sz * HeapWordSize;
1428 // Find the next marked object after this one.
1429 start = _bm->getNextMarkedWordAddress(start + 1, nextTop);
1430 _changed = true;
1431 }
1432 // Handle the last range, if any.
1433 if (start_card_num != -1) {
1434 mark_card_num_range(start_card_num, last_card_num);
1435 }
1436 if (_final) {
1437 // Mark the allocated-since-marking portion...
1438 HeapWord* tp = hr->top();
1439 if (nextTop < tp) {
1440 start_card_num =
1441 intptr_t(uintptr_t(nextTop) >> CardTableModRefBS::card_shift);
1442 last_card_num =
1443 intptr_t(uintptr_t(tp) >> CardTableModRefBS::card_shift);
1444 mark_card_num_range(start_card_num, last_card_num);
1445 // This definitely means the region has live objects.
1446 set_bit_for_region(hr);
1447 }
1448 }
1449
1450 hr->add_to_marked_bytes(marked_bytes);
1451 // Update the live region bitmap.
1452 if (marked_bytes > 0) {
1453 set_bit_for_region(hr);
1454 }
1455 hr->set_top_at_conc_mark_count(nextTop);
1456 _tot_live += hr->next_live_bytes();
1457 _tot_used += hr->used();
1458 _words_done = words_done;
1459
1460 if (!_final) {
1461 ++_regions_done;
1462 if (_regions_done % 10 == 0) {
1463 double end_vtime_sec = os::elapsedVTime();
1464 double elapsed_vtime_sec = end_vtime_sec - _start_vtime_sec;
1465 if (elapsed_vtime_sec > (10.0 / 1000.0)) {
1466 jlong sleep_time_ms =
1467 (jlong) (elapsed_vtime_sec * _cm->cleanup_sleep_factor() * 1000.0);
1468 os::sleep(Thread::current(), sleep_time_ms, false);
1469 _start_vtime_sec = end_vtime_sec;
1470 }
1471 }
1472 }
1473
1474 return false;
1475 }
1476
1477 bool changed() { return _changed; }
1478 void reset() { _changed = false; _words_done = 0; }
1479 void no_yield() { _yield = false; }
1480 size_t words_done() { return _words_done; }
1481 size_t tot_live() { return _tot_live; }
1482 size_t tot_used() { return _tot_used; }
1483 };
1484
1485
1486 void ConcurrentMark::calcDesiredRegions() {
1487 _region_bm.clear();
1488 _card_bm.clear();
1489 CalcLiveObjectsClosure calccl(false /*final*/,
1490 nextMarkBitMap(), this,
1491 &_region_bm, &_card_bm);
1492 G1CollectedHeap *g1h = G1CollectedHeap::heap();
1493 g1h->heap_region_iterate(&calccl);
1494
1495 do {
1496 calccl.reset();
1497 g1h->heap_region_iterate(&calccl);
1498 } while (calccl.changed());
1499 }
1500
1501 class G1ParFinalCountTask: public AbstractGangTask {
1502 protected:
1503 G1CollectedHeap* _g1h;
1504 CMBitMap* _bm;
1505 size_t _n_workers;
1506 size_t *_live_bytes;
1507 size_t *_used_bytes;
1508 BitMap* _region_bm;
1509 BitMap* _card_bm;
1510 public:
1511 G1ParFinalCountTask(G1CollectedHeap* g1h, CMBitMap* bm,
1512 BitMap* region_bm, BitMap* card_bm)
1513 : AbstractGangTask("G1 final counting"), _g1h(g1h),
1514 _bm(bm), _region_bm(region_bm), _card_bm(card_bm),
1515 _n_workers(0)
1516 {
1517 // Use the value already set as the number of active threads
1518 // in the call to run_task(). Needed for the allocation of
1519 // _live_bytes and _used_bytes.
1520 if (G1CollectedHeap::use_parallel_gc_threads()) {
1521 assert( _g1h->workers()->active_workers() > 0,
1522 "Should have been previously set");
1523 _n_workers = _g1h->workers()->active_workers();
1524 } else {
1525 _n_workers = 1;
1526 }
1527
1528 _live_bytes = NEW_C_HEAP_ARRAY(size_t, _n_workers);
1529 _used_bytes = NEW_C_HEAP_ARRAY(size_t, _n_workers);
1530 }
1531
1532 ~G1ParFinalCountTask() {
1533 FREE_C_HEAP_ARRAY(size_t, _live_bytes);
1534 FREE_C_HEAP_ARRAY(size_t, _used_bytes);
1535 }
1536
1537 void work(int i) {
1538 CalcLiveObjectsClosure calccl(true /*final*/,
1539 _bm, _g1h->concurrent_mark(),
1540 _region_bm, _card_bm);
1541 calccl.no_yield();
1542 if (G1CollectedHeap::use_parallel_gc_threads()) {
1543 _g1h->heap_region_par_iterate_chunked(&calccl, i,
1544 (int) _n_workers,
1545 HeapRegion::FinalCountClaimValue);
1546 } else {
1547 _g1h->heap_region_iterate(&calccl);
1548 }
1549 assert(calccl.complete(), "Shouldn't have yielded!");
1550
1551 assert((size_t) i < _n_workers, "invariant");
1552 _live_bytes[i] = calccl.tot_live();
1553 _used_bytes[i] = calccl.tot_used();
1554 }
1555 size_t live_bytes() {
1556 size_t live_bytes = 0;
1557 for (size_t i = 0; i < _n_workers; ++i)
1558 live_bytes += _live_bytes[i];
1559 return live_bytes;
1560 }
1561 size_t used_bytes() {
1562 size_t used_bytes = 0;
1563 for (size_t i = 0; i < _n_workers; ++i)
1564 used_bytes += _used_bytes[i];
1565 return used_bytes;
1566 }
1567 };
1568
1569 class G1ParNoteEndTask;
1570
1571 class G1NoteEndOfConcMarkClosure : public HeapRegionClosure {
1572 G1CollectedHeap* _g1;
1573 int _worker_num;
1574 size_t _max_live_bytes;
1575 size_t _regions_claimed;
1576 size_t _freed_bytes;
1577 FreeRegionList* _local_cleanup_list;
1578 OldRegionSet* _old_proxy_set;
1579 HumongousRegionSet* _humongous_proxy_set;
1580 HRRSCleanupTask* _hrrs_cleanup_task;
1581 double _claimed_region_time;
1582 double _max_region_time;
1583
1584 public:
1585 G1NoteEndOfConcMarkClosure(G1CollectedHeap* g1,
1586 int worker_num,
1587 FreeRegionList* local_cleanup_list,
1588 OldRegionSet* old_proxy_set,
1589 HumongousRegionSet* humongous_proxy_set,
1590 HRRSCleanupTask* hrrs_cleanup_task) :
1591 _g1(g1), _worker_num(worker_num),
1592 _max_live_bytes(0), _regions_claimed(0),
1593 _freed_bytes(0),
1594 _claimed_region_time(0.0), _max_region_time(0.0),
1595 _local_cleanup_list(local_cleanup_list),
1596 _old_proxy_set(old_proxy_set),
1597 _humongous_proxy_set(humongous_proxy_set),
1598 _hrrs_cleanup_task(hrrs_cleanup_task) { }
1599
1600 size_t freed_bytes() { return _freed_bytes; }
1601
1602 bool doHeapRegion(HeapRegion *hr) {
1603 // We use a claim value of zero here because all regions
1604 // were claimed with value 1 in the FinalCount task.
1605 hr->reset_gc_time_stamp();
1606 if (!hr->continuesHumongous()) {
1607 double start = os::elapsedTime();
1608 _regions_claimed++;
1609 hr->note_end_of_marking();
1610 _max_live_bytes += hr->max_live_bytes();
1611 _g1->free_region_if_empty(hr,
1612 &_freed_bytes,
1613 _local_cleanup_list,
1614 _old_proxy_set,
1615 _humongous_proxy_set,
1616 _hrrs_cleanup_task,
1617 true /* par */);
1618 double region_time = (os::elapsedTime() - start);
1619 _claimed_region_time += region_time;
1620 if (region_time > _max_region_time) {
1621 _max_region_time = region_time;
1622 }
1623 }
1624 return false;
1625 }
1626
1627 size_t max_live_bytes() { return _max_live_bytes; }
1628 size_t regions_claimed() { return _regions_claimed; }
1629 double claimed_region_time_sec() { return _claimed_region_time; }
1630 double max_region_time_sec() { return _max_region_time; }
1631 };
1632
1633 class G1ParNoteEndTask: public AbstractGangTask {
1634 friend class G1NoteEndOfConcMarkClosure;
1635
1636 protected:
1637 G1CollectedHeap* _g1h;
1638 size_t _max_live_bytes;
1639 size_t _freed_bytes;
1640 FreeRegionList* _cleanup_list;
1641
1642 public:
1643 G1ParNoteEndTask(G1CollectedHeap* g1h,
1644 FreeRegionList* cleanup_list) :
1645 AbstractGangTask("G1 note end"), _g1h(g1h),
1646 _max_live_bytes(0), _freed_bytes(0), _cleanup_list(cleanup_list) { }
1647
1648 void work(int i) {
1649 double start = os::elapsedTime();
1650 FreeRegionList local_cleanup_list("Local Cleanup List");
1651 OldRegionSet old_proxy_set("Local Cleanup Old Proxy Set");
1652 HumongousRegionSet humongous_proxy_set("Local Cleanup Humongous Proxy Set");
1653 HRRSCleanupTask hrrs_cleanup_task;
1654 G1NoteEndOfConcMarkClosure g1_note_end(_g1h, i, &local_cleanup_list,
1655 &old_proxy_set,
1656 &humongous_proxy_set,
1657 &hrrs_cleanup_task);
1658 if (G1CollectedHeap::use_parallel_gc_threads()) {
1659 _g1h->heap_region_par_iterate_chunked(&g1_note_end, i,
1660 _g1h->workers()->active_workers(),
1661 HeapRegion::NoteEndClaimValue);
1662 } else {
1663 _g1h->heap_region_iterate(&g1_note_end);
1664 }
1665 assert(g1_note_end.complete(), "Shouldn't have yielded!");
1666
1667 // Now update the lists
1668 _g1h->update_sets_after_freeing_regions(g1_note_end.freed_bytes(),
1669 NULL /* free_list */,
1670 &old_proxy_set,
1671 &humongous_proxy_set,
1672 true /* par */);
1673 {
1674 MutexLockerEx x(ParGCRareEvent_lock, Mutex::_no_safepoint_check_flag);
1675 _max_live_bytes += g1_note_end.max_live_bytes();
1676 _freed_bytes += g1_note_end.freed_bytes();
1677
1678 // If we iterate over the global cleanup list at the end of
1679 // cleanup to do this printing we will not guarantee to only
1680 // generate output for the newly-reclaimed regions (the list
1681 // might not be empty at the beginning of cleanup; we might
1682 // still be working on its previous contents). So we do the
1683 // printing here, before we append the new regions to the global
1684 // cleanup list.
1685
1686 G1HRPrinter* hr_printer = _g1h->hr_printer();
1687 if (hr_printer->is_active()) {
1688 HeapRegionLinkedListIterator iter(&local_cleanup_list);
1689 while (iter.more_available()) {
1690 HeapRegion* hr = iter.get_next();
1691 hr_printer->cleanup(hr);
1692 }
1693 }
1694
1695 _cleanup_list->add_as_tail(&local_cleanup_list);
1696 assert(local_cleanup_list.is_empty(), "post-condition");
1697
1698 HeapRegionRemSet::finish_cleanup_task(&hrrs_cleanup_task);
1699 }
1700 double end = os::elapsedTime();
1701 if (G1PrintParCleanupStats) {
1702 gclog_or_tty->print(" Worker thread %d [%8.3f..%8.3f = %8.3f ms] "
1703 "claimed %d regions (tot = %8.3f ms, max = %8.3f ms).\n",
1704 i, start, end, (end-start)*1000.0,
1705 g1_note_end.regions_claimed(),
1706 g1_note_end.claimed_region_time_sec()*1000.0,
1707 g1_note_end.max_region_time_sec()*1000.0);
1708 }
1709 }
1710 size_t max_live_bytes() { return _max_live_bytes; }
1711 size_t freed_bytes() { return _freed_bytes; }
1712 };
1713
1714 class G1ParScrubRemSetTask: public AbstractGangTask {
1715 protected:
1716 G1RemSet* _g1rs;
1717 BitMap* _region_bm;
1718 BitMap* _card_bm;
1719 public:
1720 G1ParScrubRemSetTask(G1CollectedHeap* g1h,
1721 BitMap* region_bm, BitMap* card_bm) :
1722 AbstractGangTask("G1 ScrubRS"), _g1rs(g1h->g1_rem_set()),
1723 _region_bm(region_bm), _card_bm(card_bm)
1724 {}
1725
1726 void work(int i) {
1727 if (G1CollectedHeap::use_parallel_gc_threads()) {
1728 _g1rs->scrub_par(_region_bm, _card_bm, i,
1729 HeapRegion::ScrubRemSetClaimValue);
1730 } else {
1731 _g1rs->scrub(_region_bm, _card_bm);
1732 }
1733 }
1734
1735 };
1736
1737 void ConcurrentMark::cleanup() {
1738 // world is stopped at this checkpoint
1739 assert(SafepointSynchronize::is_at_safepoint(),
1740 "world should be stopped");
1741 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1742
1743 // If a full collection has happened, we shouldn't do this.
1744 if (has_aborted()) {
1745 g1h->set_marking_complete(); // So bitmap clearing isn't confused
1746 return;
1747 }
1748
1749 HRSPhaseSetter x(HRSPhaseCleanup);
1750 g1h->verify_region_sets_optional();
1751
1752 if (VerifyDuringGC) {
1753 HandleMark hm; // handle scope
1754 gclog_or_tty->print(" VerifyDuringGC:(before)");
1755 Universe::heap()->prepare_for_verify();
1756 Universe::verify(/* allow dirty */ true,
1757 /* silent */ false,
1758 /* option */ VerifyOption_G1UsePrevMarking);
1759 }
1760
1761 G1CollectorPolicy* g1p = G1CollectedHeap::heap()->g1_policy();
1762 g1p->record_concurrent_mark_cleanup_start();
1763
1764 double start = os::elapsedTime();
1765
1766 HeapRegionRemSet::reset_for_cleanup_tasks();
1767
1768 g1h->set_par_threads();
1769 size_t n_workers = g1h->n_par_threads();
1770
1771 // Do counting once more with the world stopped for good measure.
1772 G1ParFinalCountTask g1_par_count_task(g1h, nextMarkBitMap(),
1773 &_region_bm, &_card_bm);
1774 if (G1CollectedHeap::use_parallel_gc_threads()) {
1775 assert(g1h->check_heap_region_claim_values(
1776 HeapRegion::InitialClaimValue),
1777 "sanity check");
1778
1779 assert(g1h->n_par_threads() == (int) n_workers,
1780 "Should not have been reset");
1781 g1h->workers()->run_task(&g1_par_count_task);
1782 // Done with the parallel phase so reset to 0.
1783 g1h->set_par_threads(0);
1784
1785 assert(g1h->check_heap_region_claim_values(
1786 HeapRegion::FinalCountClaimValue),
1787 "sanity check");
1788 } else {
1789 g1_par_count_task.work(0);
1790 }
1791
1792 size_t known_garbage_bytes =
1793 g1_par_count_task.used_bytes() - g1_par_count_task.live_bytes();
1794 g1p->set_known_garbage_bytes(known_garbage_bytes);
1795
1796 size_t start_used_bytes = g1h->used();
1797 _at_least_one_mark_complete = true;
1798 g1h->set_marking_complete();
1799
1800 ergo_verbose4(ErgoConcCycles,
1801 "finish cleanup",
1802 ergo_format_byte("occupancy")
1803 ergo_format_byte("capacity")
1804 ergo_format_byte_perc("known garbage"),
1805 start_used_bytes, g1h->capacity(),
1806 known_garbage_bytes,
1807 ((double) known_garbage_bytes / (double) g1h->capacity()) * 100.0);
1808
1809 double count_end = os::elapsedTime();
1810 double this_final_counting_time = (count_end - start);
1811 if (G1PrintParCleanupStats) {
1812 gclog_or_tty->print_cr("Cleanup:");
1813 gclog_or_tty->print_cr(" Finalize counting: %8.3f ms",
1814 this_final_counting_time*1000.0);
1815 }
1816 _total_counting_time += this_final_counting_time;
1817
1818 if (G1PrintRegionLivenessInfo) {
1819 G1PrintRegionLivenessInfoClosure cl(gclog_or_tty, "Post-Marking");
1820 _g1h->heap_region_iterate(&cl);
1821 }
1822
1823 // Install newly created mark bitMap as "prev".
1824 swapMarkBitMaps();
1825
1826 g1h->reset_gc_time_stamp();
1827
1828 // Note end of marking in all heap regions.
1829 double note_end_start = os::elapsedTime();
1830 G1ParNoteEndTask g1_par_note_end_task(g1h, &_cleanup_list);
1831 if (G1CollectedHeap::use_parallel_gc_threads()) {
1832 g1h->set_par_threads((int)n_workers);
1833 g1h->workers()->run_task(&g1_par_note_end_task);
1834 g1h->set_par_threads(0);
1835
1836 assert(g1h->check_heap_region_claim_values(HeapRegion::NoteEndClaimValue),
1837 "sanity check");
1838 } else {
1839 g1_par_note_end_task.work(0);
1840 }
1841
1842 if (!cleanup_list_is_empty()) {
1843 // The cleanup list is not empty, so we'll have to process it
1844 // concurrently. Notify anyone else that might be wanting free
1845 // regions that there will be more free regions coming soon.
1846 g1h->set_free_regions_coming();
1847 }
1848 double note_end_end = os::elapsedTime();
1849 if (G1PrintParCleanupStats) {
1850 gclog_or_tty->print_cr(" note end of marking: %8.3f ms.",
1851 (note_end_end - note_end_start)*1000.0);
1852 }
1853
1854
1855 // call below, since it affects the metric by which we sort the heap
1856 // regions.
1857 if (G1ScrubRemSets) {
1858 double rs_scrub_start = os::elapsedTime();
1859 G1ParScrubRemSetTask g1_par_scrub_rs_task(g1h, &_region_bm, &_card_bm);
1860 if (G1CollectedHeap::use_parallel_gc_threads()) {
1861 g1h->set_par_threads((int)n_workers);
1862 g1h->workers()->run_task(&g1_par_scrub_rs_task);
1863 g1h->set_par_threads(0);
1864
1865 assert(g1h->check_heap_region_claim_values(
1866 HeapRegion::ScrubRemSetClaimValue),
1867 "sanity check");
1868 } else {
1869 g1_par_scrub_rs_task.work(0);
1870 }
1871
1872 double rs_scrub_end = os::elapsedTime();
1873 double this_rs_scrub_time = (rs_scrub_end - rs_scrub_start);
1874 _total_rs_scrub_time += this_rs_scrub_time;
1875 }
1876
1877 // this will also free any regions totally full of garbage objects,
1878 // and sort the regions.
1879 g1h->g1_policy()->record_concurrent_mark_cleanup_end((int)n_workers);
1880
1881 // Statistics.
1882 double end = os::elapsedTime();
1883 _cleanup_times.add((end - start) * 1000.0);
1884
1885 // G1CollectedHeap::heap()->print();
1886 // gclog_or_tty->print_cr("HEAP GC TIME STAMP : %d",
1887 // G1CollectedHeap::heap()->get_gc_time_stamp());
1888
1889 if (PrintGC || PrintGCDetails) {
1890 g1h->print_size_transition(gclog_or_tty,
1891 start_used_bytes,
1892 g1h->used(),
1893 g1h->capacity());
1894 }
1895
1896 size_t cleaned_up_bytes = start_used_bytes - g1h->used();
1897 g1p->decrease_known_garbage_bytes(cleaned_up_bytes);
1898
1899 // Clean up will have freed any regions completely full of garbage.
1900 // Update the soft reference policy with the new heap occupancy.
1901 Universe::update_heap_info_at_gc();
1902
1903 // We need to make this be a "collection" so any collection pause that
1904 // races with it goes around and waits for completeCleanup to finish.
1905 g1h->increment_total_collections();
1906
1907 if (VerifyDuringGC) {
1908 HandleMark hm; // handle scope
1909 gclog_or_tty->print(" VerifyDuringGC:(after)");
1910 Universe::heap()->prepare_for_verify();
1911 Universe::verify(/* allow dirty */ true,
1912 /* silent */ false,
1913 /* option */ VerifyOption_G1UsePrevMarking);
1914 }
1915
1916 g1h->verify_region_sets_optional();
1917 }
1918
1919 void ConcurrentMark::completeCleanup() {
1920 if (has_aborted()) return;
1921
1922 G1CollectedHeap* g1h = G1CollectedHeap::heap();
1923
1924 _cleanup_list.verify_optional();
1925 FreeRegionList tmp_free_list("Tmp Free List");
1926
1927 if (G1ConcRegionFreeingVerbose) {
1928 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
1929 "cleanup list has "SIZE_FORMAT" entries",
1930 _cleanup_list.length());
1931 }
1932
1933 // Noone else should be accessing the _cleanup_list at this point,
1934 // so it's not necessary to take any locks
1935 while (!_cleanup_list.is_empty()) {
1936 HeapRegion* hr = _cleanup_list.remove_head();
1937 assert(hr != NULL, "the list was not empty");
1938 hr->par_clear();
1939 tmp_free_list.add_as_tail(hr);
1940
1941 // Instead of adding one region at a time to the secondary_free_list,
1942 // we accumulate them in the local list and move them a few at a
1943 // time. This also cuts down on the number of notify_all() calls
1944 // we do during this process. We'll also append the local list when
1945 // _cleanup_list is empty (which means we just removed the last
1946 // region from the _cleanup_list).
1947 if ((tmp_free_list.length() % G1SecondaryFreeListAppendLength == 0) ||
1948 _cleanup_list.is_empty()) {
1949 if (G1ConcRegionFreeingVerbose) {
1950 gclog_or_tty->print_cr("G1ConcRegionFreeing [complete cleanup] : "
1951 "appending "SIZE_FORMAT" entries to the "
1952 "secondary_free_list, clean list still has "
1953 SIZE_FORMAT" entries",
1954 tmp_free_list.length(),
1955 _cleanup_list.length());
1956 }
1957
1958 {
1959 MutexLockerEx x(SecondaryFreeList_lock, Mutex::_no_safepoint_check_flag);
1960 g1h->secondary_free_list_add_as_tail(&tmp_free_list);
1961 SecondaryFreeList_lock->notify_all();
1962 }
1963
1964 if (G1StressConcRegionFreeing) {
1965 for (uintx i = 0; i < G1StressConcRegionFreeingDelayMillis; ++i) {
1966 os::sleep(Thread::current(), (jlong) 1, false);
1967 }
1968 }
1969 }
1970 }
1971 assert(tmp_free_list.is_empty(), "post-condition");
1972 }
1973
1974 // Support closures for reference procssing in G1
1975
1976 bool G1CMIsAliveClosure::do_object_b(oop obj) {
1977 HeapWord* addr = (HeapWord*)obj;
1978 return addr != NULL &&
1979 (!_g1->is_in_g1_reserved(addr) || !_g1->is_obj_ill(obj));
1980 }
1981
1982 class G1CMKeepAliveClosure: public OopClosure {
1983 G1CollectedHeap* _g1;
1984 ConcurrentMark* _cm;
1985 CMBitMap* _bitMap;
1986 public:
1987 G1CMKeepAliveClosure(G1CollectedHeap* g1, ConcurrentMark* cm,
1988 CMBitMap* bitMap) :
1989 _g1(g1), _cm(cm),
1990 _bitMap(bitMap) {}
1991
1992 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
1993 virtual void do_oop( oop* p) { do_oop_work(p); }
1994
1995 template <class T> void do_oop_work(T* p) {
1996 oop obj = oopDesc::load_decode_heap_oop(p);
1997 HeapWord* addr = (HeapWord*)obj;
1998
1999 if (_cm->verbose_high()) {
2000 gclog_or_tty->print_cr("\t[0] we're looking at location "
2001 "*"PTR_FORMAT" = "PTR_FORMAT,
2002 p, (void*) obj);
2003 }
2004
2005 if (_g1->is_in_g1_reserved(addr) && _g1->is_obj_ill(obj)) {
2006 _bitMap->mark(addr);
2007 _cm->mark_stack_push(obj);
2008 }
2009 }
2010 };
2011
2012 class G1CMDrainMarkingStackClosure: public VoidClosure {
2013 CMMarkStack* _markStack;
2014 CMBitMap* _bitMap;
2015 G1CMKeepAliveClosure* _oopClosure;
2016 public:
2017 G1CMDrainMarkingStackClosure(CMBitMap* bitMap, CMMarkStack* markStack,
2018 G1CMKeepAliveClosure* oopClosure) :
2019 _bitMap(bitMap),
2020 _markStack(markStack),
2021 _oopClosure(oopClosure)
2022 {}
2023
2024 void do_void() {
2025 _markStack->drain((OopClosure*)_oopClosure, _bitMap, false);
2026 }
2027 };
2028
2029 // 'Keep Alive' closure used by parallel reference processing.
2030 // An instance of this closure is used in the parallel reference processing
2031 // code rather than an instance of G1CMKeepAliveClosure. We could have used
2032 // the G1CMKeepAliveClosure as it is MT-safe. Also reference objects are
2033 // placed on to discovered ref lists once so we can mark and push with no
2034 // need to check whether the object has already been marked. Using the
2035 // G1CMKeepAliveClosure would mean, however, having all the worker threads
2036 // operating on the global mark stack. This means that an individual
2037 // worker would be doing lock-free pushes while it processes its own
2038 // discovered ref list followed by drain call. If the discovered ref lists
2039 // are unbalanced then this could cause interference with the other
2040 // workers. Using a CMTask (and its embedded local data structures)
2041 // avoids that potential interference.
2042 class G1CMParKeepAliveAndDrainClosure: public OopClosure {
2043 ConcurrentMark* _cm;
2044 CMTask* _task;
2045 int _ref_counter_limit;
2046 int _ref_counter;
2047 public:
2048 G1CMParKeepAliveAndDrainClosure(ConcurrentMark* cm, CMTask* task) :
2049 _cm(cm), _task(task),
2050 _ref_counter_limit(G1RefProcDrainInterval) {
2051 assert(_ref_counter_limit > 0, "sanity");
2052 _ref_counter = _ref_counter_limit;
2053 }
2054
2055 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
2056 virtual void do_oop( oop* p) { do_oop_work(p); }
2057
2058 template <class T> void do_oop_work(T* p) {
2059 if (!_cm->has_overflown()) {
2060 oop obj = oopDesc::load_decode_heap_oop(p);
2061 if (_cm->verbose_high()) {
2062 gclog_or_tty->print_cr("\t[%d] we're looking at location "
2063 "*"PTR_FORMAT" = "PTR_FORMAT,
2064 _task->task_id(), p, (void*) obj);
2065 }
2066
2067 _task->deal_with_reference(obj);
2068 _ref_counter--;
2069
2070 if (_ref_counter == 0) {
2071 // We have dealt with _ref_counter_limit references, pushing them and objects
2072 // reachable from them on to the local stack (and possibly the global stack).
2073 // Call do_marking_step() to process these entries. We call the routine in a
2074 // loop, which we'll exit if there's nothing more to do (i.e. we're done
2075 // with the entries that we've pushed as a result of the deal_with_reference
2076 // calls above) or we overflow.
2077 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted() flag
2078 // while there may still be some work to do. (See the comment at the
2079 // beginning of CMTask::do_marking_step() for those conditions - one of which
2080 // is reaching the specified time target.) It is only when
2081 // CMTask::do_marking_step() returns without setting the has_aborted() flag
2082 // that the marking has completed.
2083 do {
2084 double mark_step_duration_ms = G1ConcMarkStepDurationMillis;
2085 _task->do_marking_step(mark_step_duration_ms,
2086 false /* do_stealing */,
2087 false /* do_termination */);
2088 } while (_task->has_aborted() && !_cm->has_overflown());
2089 _ref_counter = _ref_counter_limit;
2090 }
2091 } else {
2092 if (_cm->verbose_high()) {
2093 gclog_or_tty->print_cr("\t[%d] CM Overflow", _task->task_id());
2094 }
2095 }
2096 }
2097 };
2098
2099 class G1CMParDrainMarkingStackClosure: public VoidClosure {
2100 ConcurrentMark* _cm;
2101 CMTask* _task;
2102 public:
2103 G1CMParDrainMarkingStackClosure(ConcurrentMark* cm, CMTask* task) :
2104 _cm(cm), _task(task)
2105 {}
2106
2107 void do_void() {
2108 do {
2109 if (_cm->verbose_high()) {
2110 gclog_or_tty->print_cr("\t[%d] Drain: Calling do marking_step",
2111 _task->task_id());
2112 }
2113
2114 // We call CMTask::do_marking_step() to completely drain the local and
2115 // global marking stacks. The routine is called in a loop, which we'll
2116 // exit if there's nothing more to do (i.e. we'completely drained the
2117 // entries that were pushed as a result of applying the
2118 // G1CMParKeepAliveAndDrainClosure to the entries on the discovered ref
2119 // lists above) or we overflow the global marking stack.
2120 // Note: CMTask::do_marking_step() can set the CMTask::has_aborted() flag
2121 // while there may still be some work to do. (See the comment at the
2122 // beginning of CMTask::do_marking_step() for those conditions - one of which
2123 // is reaching the specified time target.) It is only when
2124 // CMTask::do_marking_step() returns without setting the has_aborted() flag
2125 // that the marking has completed.
2126
2127 _task->do_marking_step(1000000000.0 /* something very large */,
2128 true /* do_stealing */,
2129 true /* do_termination */);
2130 } while (_task->has_aborted() && !_cm->has_overflown());
2131 }
2132 };
2133
2134 // Implementation of AbstractRefProcTaskExecutor for parallel
2135 // reference processing at the end of G1 concurrent marking
2136
2137 class G1CMRefProcTaskExecutor: public AbstractRefProcTaskExecutor {
2138 private:
2139 G1CollectedHeap* _g1h;
2140 ConcurrentMark* _cm;
2141 WorkGang* _workers;
2142 int _active_workers;
2143
2144 public:
2145 G1CMRefProcTaskExecutor(G1CollectedHeap* g1h,
2146 ConcurrentMark* cm,
2147 WorkGang* workers,
2148 int n_workers) :
2149 _g1h(g1h), _cm(cm),
2150 _workers(workers), _active_workers(n_workers) { }
2151
2152 // Executes the given task using concurrent marking worker threads.
2153 virtual void execute(ProcessTask& task);
2154 virtual void execute(EnqueueTask& task);
2155 };
2156
2157 class G1CMRefProcTaskProxy: public AbstractGangTask {
2158 typedef AbstractRefProcTaskExecutor::ProcessTask ProcessTask;
2159 ProcessTask& _proc_task;
2160 G1CollectedHeap* _g1h;
2161 ConcurrentMark* _cm;
2162
2163 public:
2164 G1CMRefProcTaskProxy(ProcessTask& proc_task,
2165 G1CollectedHeap* g1h,
2166 ConcurrentMark* cm) :
2167 AbstractGangTask("Process reference objects in parallel"),
2168 _proc_task(proc_task), _g1h(g1h), _cm(cm) { }
2169
2170 virtual void work(int i) {
2171 CMTask* marking_task = _cm->task(i);
2172 G1CMIsAliveClosure g1_is_alive(_g1h);
2173 G1CMParKeepAliveAndDrainClosure g1_par_keep_alive(_cm, marking_task);
2174 G1CMParDrainMarkingStackClosure g1_par_drain(_cm, marking_task);
2175
2176 _proc_task.work(i, g1_is_alive, g1_par_keep_alive, g1_par_drain);
2177 }
2178 };
2179
2180 void G1CMRefProcTaskExecutor::execute(ProcessTask& proc_task) {
2181 assert(_workers != NULL, "Need parallel worker threads.");
2182
2183 G1CMRefProcTaskProxy proc_task_proxy(proc_task, _g1h, _cm);
2184
2185 // We need to reset the phase for each task execution so that
2186 // the termination protocol of CMTask::do_marking_step works.
2187 _cm->set_phase(_active_workers, false /* concurrent */);
2188 _g1h->set_par_threads(_active_workers);
2189 _workers->run_task(&proc_task_proxy);
2190 _g1h->set_par_threads(0);
2191 }
2192
2193 class G1CMRefEnqueueTaskProxy: public AbstractGangTask {
2194 typedef AbstractRefProcTaskExecutor::EnqueueTask EnqueueTask;
2195 EnqueueTask& _enq_task;
2196
2197 public:
2198 G1CMRefEnqueueTaskProxy(EnqueueTask& enq_task) :
2199 AbstractGangTask("Enqueue reference objects in parallel"),
2200 _enq_task(enq_task) { }
2201
2202 virtual void work(int i) {
2203 _enq_task.work(i);
2204 }
2205 };
2206
2207 void G1CMRefProcTaskExecutor::execute(EnqueueTask& enq_task) {
2208 assert(_workers != NULL, "Need parallel worker threads.");
2209
2210 G1CMRefEnqueueTaskProxy enq_task_proxy(enq_task);
2211
2212 _g1h->set_par_threads(_active_workers);
2213 _workers->run_task(&enq_task_proxy);
2214 _g1h->set_par_threads(0);
2215 }
2216
2217 void ConcurrentMark::weakRefsWork(bool clear_all_soft_refs) {
2218 ResourceMark rm;
2219 HandleMark hm;
2220
2221 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2222
2223 // Is alive closure.
2224 G1CMIsAliveClosure g1_is_alive(g1h);
2225
2226 // Inner scope to exclude the cleaning of the string and symbol
2227 // tables from the displayed time.
2228 {
2229 bool verbose = PrintGC && PrintGCDetails;
2230 if (verbose) {
2231 gclog_or_tty->put(' ');
2232 }
2233 TraceTime t("GC ref-proc", verbose, false, gclog_or_tty);
2234
2235 ReferenceProcessor* rp = g1h->ref_processor_cm();
2236
2237 // See the comment in G1CollectedHeap::ref_processing_init()
2238 // about how reference processing currently works in G1.
2239
2240 // Process weak references.
2241 rp->setup_policy(clear_all_soft_refs);
2242 assert(_markStack.isEmpty(), "mark stack should be empty");
2243
2244 G1CMKeepAliveClosure g1_keep_alive(g1h, this, nextMarkBitMap());
2245 G1CMDrainMarkingStackClosure
2246 g1_drain_mark_stack(nextMarkBitMap(), &_markStack, &g1_keep_alive);
2247
2248 // We use the work gang from the G1CollectedHeap and we utilize all
2249 // the worker threads.
2250 int active_workers = g1h->workers() ? g1h->workers()->active_workers() : 1;
2251 active_workers = MAX2(MIN2(active_workers, (int)_max_task_num), 1);
2252
2253 G1CMRefProcTaskExecutor par_task_executor(g1h, this,
2254 g1h->workers(), active_workers);
2255
2256 if (rp->processing_is_mt()) {
2257 // Set the degree of MT here. If the discovery is done MT, there
2258 // may have been a different number of threads doing the discovery
2259 // and a different number of discovered lists may have Ref objects.
2260 // That is OK as long as the Reference lists are balanced (see
2261 // balance_all_queues() and balance_queues()).
2262 rp->set_active_mt_degree(active_workers);
2263
2264 rp->process_discovered_references(&g1_is_alive,
2265 &g1_keep_alive,
2266 &g1_drain_mark_stack,
2267 &par_task_executor);
2268
2269 // The work routines of the parallel keep_alive and drain_marking_stack
2270 // will set the has_overflown flag if we overflow the global marking
2271 // stack.
2272 } else {
2273 rp->process_discovered_references(&g1_is_alive,
2274 &g1_keep_alive,
2275 &g1_drain_mark_stack,
2276 NULL);
2277 }
2278
2279 assert(_markStack.overflow() || _markStack.isEmpty(),
2280 "mark stack should be empty (unless it overflowed)");
2281 if (_markStack.overflow()) {
2282 // Should have been done already when we tried to push an
2283 // entry on to the global mark stack. But let's do it again.
2284 set_has_overflown();
2285 }
2286
2287 if (rp->processing_is_mt()) {
2288 assert(rp->num_q() == active_workers, "why not");
2289 rp->enqueue_discovered_references(&par_task_executor);
2290 } else {
2291 rp->enqueue_discovered_references();
2292 }
2293
2294 rp->verify_no_references_recorded();
2295 assert(!rp->discovery_enabled(), "Post condition");
2296 }
2297
2298 // Now clean up stale oops in StringTable
2299 StringTable::unlink(&g1_is_alive);
2300 // Clean up unreferenced symbols in symbol table.
2301 SymbolTable::unlink();
2302 }
2303
2304 void ConcurrentMark::swapMarkBitMaps() {
2305 CMBitMapRO* temp = _prevMarkBitMap;
2306 _prevMarkBitMap = (CMBitMapRO*)_nextMarkBitMap;
2307 _nextMarkBitMap = (CMBitMap*) temp;
2308 }
2309
2310 class CMRemarkTask: public AbstractGangTask {
2311 private:
2312 ConcurrentMark *_cm;
2313
2314 public:
2315 void work(int worker_i) {
2316 // Since all available tasks are actually started, we should
2317 // only proceed if we're supposed to be actived.
2318 if ((size_t)worker_i < _cm->active_tasks()) {
2319 CMTask* task = _cm->task(worker_i);
2320 task->record_start_time();
2321 do {
2322 task->do_marking_step(1000000000.0 /* something very large */,
2323 true /* do_stealing */,
2324 true /* do_termination */);
2325 } while (task->has_aborted() && !_cm->has_overflown());
2326 // If we overflow, then we do not want to restart. We instead
2327 // want to abort remark and do concurrent marking again.
2328 task->record_end_time();
2329 }
2330 }
2331
2332 CMRemarkTask(ConcurrentMark* cm) :
2333 AbstractGangTask("Par Remark"), _cm(cm) {
2334 _cm->terminator()->reset_for_reuse(cm->_g1h->workers()->active_workers());
2335 }
2336 };
2337
2338 void ConcurrentMark::checkpointRootsFinalWork() {
2339 ResourceMark rm;
2340 HandleMark hm;
2341 G1CollectedHeap* g1h = G1CollectedHeap::heap();
2342
2343 g1h->ensure_parsability(false);
2344
2345 if (G1CollectedHeap::use_parallel_gc_threads()) {
2346 G1CollectedHeap::StrongRootsScope srs(g1h);
2347 // this is remark, so we'll use up all active threads
2348 int active_workers = g1h->workers()->active_workers();
2349 if (active_workers == 0) {
2350 assert(active_workers > 0, "Should have been set earlier");
2351 active_workers = ParallelGCThreads;
2352 g1h->workers()->set_active_workers(active_workers);
2353 }
2354 set_phase(active_workers, false /* concurrent */);
2355 // Leave _parallel_marking_threads at it's
2356 // value originally calculated in the ConcurrentMark
2357 // constructor and pass values of the active workers
2358 // through the gang in the task.
2359
2360 CMRemarkTask remarkTask(this);
2361 g1h->set_par_threads(active_workers);
2362 g1h->workers()->run_task(&remarkTask);
2363 g1h->set_par_threads(0);
2364 } else {
2365 G1CollectedHeap::StrongRootsScope srs(g1h);
2366 // this is remark, so we'll use up all available threads
2367 int active_workers = 1;
2368 set_phase(active_workers, false /* concurrent */);
2369
2370 CMRemarkTask remarkTask(this);
2371 // We will start all available threads, even if we decide that the
2372 // active_workers will be fewer. The extra ones will just bail out
2373 // immediately.
2374 remarkTask.work(0);
2375 }
2376 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2377 guarantee(satb_mq_set.completed_buffers_num() == 0, "invariant");
2378
2379 print_stats();
2380
2381 #if VERIFY_OBJS_PROCESSED
2382 if (_scan_obj_cl.objs_processed != ThreadLocalObjQueue::objs_enqueued) {
2383 gclog_or_tty->print_cr("Processed = %d, enqueued = %d.",
2384 _scan_obj_cl.objs_processed,
2385 ThreadLocalObjQueue::objs_enqueued);
2386 guarantee(_scan_obj_cl.objs_processed ==
2387 ThreadLocalObjQueue::objs_enqueued,
2388 "Different number of objs processed and enqueued.");
2389 }
2390 #endif
2391 }
2392
2393 #ifndef PRODUCT
2394
2395 class PrintReachableOopClosure: public OopClosure {
2396 private:
2397 G1CollectedHeap* _g1h;
2398 outputStream* _out;
2399 VerifyOption _vo;
2400 bool _all;
2401
2402 public:
2403 PrintReachableOopClosure(outputStream* out,
2404 VerifyOption vo,
2405 bool all) :
2406 _g1h(G1CollectedHeap::heap()),
2407 _out(out), _vo(vo), _all(all) { }
2408
2409 void do_oop(narrowOop* p) { do_oop_work(p); }
2410 void do_oop( oop* p) { do_oop_work(p); }
2411
2412 template <class T> void do_oop_work(T* p) {
2413 oop obj = oopDesc::load_decode_heap_oop(p);
2414 const char* str = NULL;
2415 const char* str2 = "";
2416
2417 if (obj == NULL) {
2418 str = "";
2419 } else if (!_g1h->is_in_g1_reserved(obj)) {
2420 str = " O";
2421 } else {
2422 HeapRegion* hr = _g1h->heap_region_containing(obj);
2423 guarantee(hr != NULL, "invariant");
2424 bool over_tams = false;
2425 bool marked = false;
2426
2427 switch (_vo) {
2428 case VerifyOption_G1UsePrevMarking:
2429 over_tams = hr->obj_allocated_since_prev_marking(obj);
2430 marked = _g1h->isMarkedPrev(obj);
2431 break;
2432 case VerifyOption_G1UseNextMarking:
2433 over_tams = hr->obj_allocated_since_next_marking(obj);
2434 marked = _g1h->isMarkedNext(obj);
2435 break;
2436 case VerifyOption_G1UseMarkWord:
2437 marked = obj->is_gc_marked();
2438 break;
2439 default:
2440 ShouldNotReachHere();
2441 }
2442
2443 if (over_tams) {
2444 str = " >";
2445 if (marked) {
2446 str2 = " AND MARKED";
2447 }
2448 } else if (marked) {
2449 str = " M";
2450 } else {
2451 str = " NOT";
2452 }
2453 }
2454
2455 _out->print_cr(" "PTR_FORMAT": "PTR_FORMAT"%s%s",
2456 p, (void*) obj, str, str2);
2457 }
2458 };
2459
2460 class PrintReachableObjectClosure : public ObjectClosure {
2461 private:
2462 G1CollectedHeap* _g1h;
2463 outputStream* _out;
2464 VerifyOption _vo;
2465 bool _all;
2466 HeapRegion* _hr;
2467
2468 public:
2469 PrintReachableObjectClosure(outputStream* out,
2470 VerifyOption vo,
2471 bool all,
2472 HeapRegion* hr) :
2473 _g1h(G1CollectedHeap::heap()),
2474 _out(out), _vo(vo), _all(all), _hr(hr) { }
2475
2476 void do_object(oop o) {
2477 bool over_tams = false;
2478 bool marked = false;
2479
2480 switch (_vo) {
2481 case VerifyOption_G1UsePrevMarking:
2482 over_tams = _hr->obj_allocated_since_prev_marking(o);
2483 marked = _g1h->isMarkedPrev(o);
2484 break;
2485 case VerifyOption_G1UseNextMarking:
2486 over_tams = _hr->obj_allocated_since_next_marking(o);
2487 marked = _g1h->isMarkedNext(o);
2488 break;
2489 case VerifyOption_G1UseMarkWord:
2490 marked = o->is_gc_marked();
2491 break;
2492 default:
2493 ShouldNotReachHere();
2494 }
2495 bool print_it = _all || over_tams || marked;
2496
2497 if (print_it) {
2498 _out->print_cr(" "PTR_FORMAT"%s",
2499 o, (over_tams) ? " >" : (marked) ? " M" : "");
2500 PrintReachableOopClosure oopCl(_out, _vo, _all);
2501 o->oop_iterate(&oopCl);
2502 }
2503 }
2504 };
2505
2506 class PrintReachableRegionClosure : public HeapRegionClosure {
2507 private:
2508 outputStream* _out;
2509 VerifyOption _vo;
2510 bool _all;
2511
2512 public:
2513 bool doHeapRegion(HeapRegion* hr) {
2514 HeapWord* b = hr->bottom();
2515 HeapWord* e = hr->end();
2516 HeapWord* t = hr->top();
2517 HeapWord* p = NULL;
2518
2519 switch (_vo) {
2520 case VerifyOption_G1UsePrevMarking:
2521 p = hr->prev_top_at_mark_start();
2522 break;
2523 case VerifyOption_G1UseNextMarking:
2524 p = hr->next_top_at_mark_start();
2525 break;
2526 case VerifyOption_G1UseMarkWord:
2527 // When we are verifying marking using the mark word
2528 // TAMS has no relevance.
2529 assert(p == NULL, "post-condition");
2530 break;
2531 default:
2532 ShouldNotReachHere();
2533 }
2534 _out->print_cr("** ["PTR_FORMAT", "PTR_FORMAT"] top: "PTR_FORMAT" "
2535 "TAMS: "PTR_FORMAT, b, e, t, p);
2536 _out->cr();
2537
2538 HeapWord* from = b;
2539 HeapWord* to = t;
2540
2541 if (to > from) {
2542 _out->print_cr("Objects in ["PTR_FORMAT", "PTR_FORMAT"]", from, to);
2543 _out->cr();
2544 PrintReachableObjectClosure ocl(_out, _vo, _all, hr);
2545 hr->object_iterate_mem_careful(MemRegion(from, to), &ocl);
2546 _out->cr();
2547 }
2548
2549 return false;
2550 }
2551
2552 PrintReachableRegionClosure(outputStream* out,
2553 VerifyOption vo,
2554 bool all) :
2555 _out(out), _vo(vo), _all(all) { }
2556 };
2557
2558 static const char* verify_option_to_tams(VerifyOption vo) {
2559 switch (vo) {
2560 case VerifyOption_G1UsePrevMarking:
2561 return "PTAMS";
2562 case VerifyOption_G1UseNextMarking:
2563 return "NTAMS";
2564 default:
2565 return "NONE";
2566 }
2567 }
2568
2569 void ConcurrentMark::print_reachable(const char* str,
2570 VerifyOption vo,
2571 bool all) {
2572 gclog_or_tty->cr();
2573 gclog_or_tty->print_cr("== Doing heap dump... ");
2574
2575 if (G1PrintReachableBaseFile == NULL) {
2576 gclog_or_tty->print_cr(" #### error: no base file defined");
2577 return;
2578 }
2579
2580 if (strlen(G1PrintReachableBaseFile) + 1 + strlen(str) >
2581 (JVM_MAXPATHLEN - 1)) {
2582 gclog_or_tty->print_cr(" #### error: file name too long");
2583 return;
2584 }
2585
2586 char file_name[JVM_MAXPATHLEN];
2587 sprintf(file_name, "%s.%s", G1PrintReachableBaseFile, str);
2588 gclog_or_tty->print_cr(" dumping to file %s", file_name);
2589
2590 fileStream fout(file_name);
2591 if (!fout.is_open()) {
2592 gclog_or_tty->print_cr(" #### error: could not open file");
2593 return;
2594 }
2595
2596 outputStream* out = &fout;
2597 out->print_cr("-- USING %s", verify_option_to_tams(vo));
2598 out->cr();
2599
2600 out->print_cr("--- ITERATING OVER REGIONS");
2601 out->cr();
2602 PrintReachableRegionClosure rcl(out, vo, all);
2603 _g1h->heap_region_iterate(&rcl);
2604 out->cr();
2605
2606 gclog_or_tty->print_cr(" done");
2607 gclog_or_tty->flush();
2608 }
2609
2610 #endif // PRODUCT
2611
2612 // This note is for drainAllSATBBuffers and the code in between.
2613 // In the future we could reuse a task to do this work during an
2614 // evacuation pause (since now tasks are not active and can be claimed
2615 // during an evacuation pause). This was a late change to the code and
2616 // is currently not being taken advantage of.
2617
2618 class CMGlobalObjectClosure : public ObjectClosure {
2619 private:
2620 ConcurrentMark* _cm;
2621
2622 public:
2623 void do_object(oop obj) {
2624 _cm->deal_with_reference(obj);
2625 }
2626
2627 CMGlobalObjectClosure(ConcurrentMark* cm) : _cm(cm) { }
2628 };
2629
2630 void ConcurrentMark::deal_with_reference(oop obj) {
2631 if (verbose_high()) {
2632 gclog_or_tty->print_cr("[global] we're dealing with reference "PTR_FORMAT,
2633 (void*) obj);
2634 }
2635
2636 HeapWord* objAddr = (HeapWord*) obj;
2637 assert(obj->is_oop_or_null(true /* ignore mark word */), "Error");
2638 if (_g1h->is_in_g1_reserved(objAddr)) {
2639 assert(obj != NULL, "null check is implicit");
2640 if (!_nextMarkBitMap->isMarked(objAddr)) {
2641 // Only get the containing region if the object is not marked on the
2642 // bitmap (otherwise, it's a waste of time since we won't do
2643 // anything with it).
2644 HeapRegion* hr = _g1h->heap_region_containing_raw(obj);
2645 if (!hr->obj_allocated_since_next_marking(obj)) {
2646 if (verbose_high()) {
2647 gclog_or_tty->print_cr("[global] "PTR_FORMAT" is not considered "
2648 "marked", (void*) obj);
2649 }
2650
2651 // we need to mark it first
2652 if (_nextMarkBitMap->parMark(objAddr)) {
2653 // No OrderAccess:store_load() is needed. It is implicit in the
2654 // CAS done in parMark(objAddr) above
2655 HeapWord* finger = _finger;
2656 if (objAddr < finger) {
2657 if (verbose_high()) {
2658 gclog_or_tty->print_cr("[global] below the global finger "
2659 "("PTR_FORMAT"), pushing it", finger);
2660 }
2661 if (!mark_stack_push(obj)) {
2662 if (verbose_low()) {
2663 gclog_or_tty->print_cr("[global] global stack overflow during "
2664 "deal_with_reference");
2665 }
2666 }
2667 }
2668 }
2669 }
2670 }
2671 }
2672 }
2673
2674 void ConcurrentMark::drainAllSATBBuffers() {
2675 CMGlobalObjectClosure oc(this);
2676 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
2677 satb_mq_set.set_closure(&oc);
2678
2679 while (satb_mq_set.apply_closure_to_completed_buffer()) {
2680 if (verbose_medium()) {
2681 gclog_or_tty->print_cr("[global] processed an SATB buffer");
2682 }
2683 }
2684
2685 // no need to check whether we should do this, as this is only
2686 // called during an evacuation pause
2687 satb_mq_set.iterate_closure_all_threads();
2688
2689 satb_mq_set.set_closure(NULL);
2690 assert(satb_mq_set.completed_buffers_num() == 0, "invariant");
2691 }
2692
2693 void ConcurrentMark::markPrev(oop p) {
2694 // Note we are overriding the read-only view of the prev map here, via
2695 // the cast.
2696 ((CMBitMap*)_prevMarkBitMap)->mark((HeapWord*)p);
2697 }
2698
2699 void ConcurrentMark::clear(oop p) {
2700 assert(p != NULL && p->is_oop(), "expected an oop");
2701 HeapWord* addr = (HeapWord*)p;
2702 assert(addr >= _nextMarkBitMap->startWord() ||
2703 addr < _nextMarkBitMap->endWord(), "in a region");
2704
2705 _nextMarkBitMap->clear(addr);
2706 }
2707
2708 void ConcurrentMark::clearRangeBothMaps(MemRegion mr) {
2709 // Note we are overriding the read-only view of the prev map here, via
2710 // the cast.
2711 ((CMBitMap*)_prevMarkBitMap)->clearRange(mr);
2712 _nextMarkBitMap->clearRange(mr);
2713 }
2714
2715 HeapRegion*
2716 ConcurrentMark::claim_region(int task_num) {
2717 // "checkpoint" the finger
2718 HeapWord* finger = _finger;
2719
2720 // _heap_end will not change underneath our feet; it only changes at
2721 // yield points.
2722 while (finger < _heap_end) {
2723 assert(_g1h->is_in_g1_reserved(finger), "invariant");
2724
2725 // Note on how this code handles humongous regions. In the
2726 // normal case the finger will reach the start of a "starts
2727 // humongous" (SH) region. Its end will either be the end of the
2728 // last "continues humongous" (CH) region in the sequence, or the
2729 // standard end of the SH region (if the SH is the only region in
2730 // the sequence). That way claim_region() will skip over the CH
2731 // regions. However, there is a subtle race between a CM thread
2732 // executing this method and a mutator thread doing a humongous
2733 // object allocation. The two are not mutually exclusive as the CM
2734 // thread does not need to hold the Heap_lock when it gets
2735 // here. So there is a chance that claim_region() will come across
2736 // a free region that's in the progress of becoming a SH or a CH
2737 // region. In the former case, it will either
2738 // a) Miss the update to the region's end, in which case it will
2739 // visit every subsequent CH region, will find their bitmaps
2740 // empty, and do nothing, or
2741 // b) Will observe the update of the region's end (in which case
2742 // it will skip the subsequent CH regions).
2743 // If it comes across a region that suddenly becomes CH, the
2744 // scenario will be similar to b). So, the race between
2745 // claim_region() and a humongous object allocation might force us
2746 // to do a bit of unnecessary work (due to some unnecessary bitmap
2747 // iterations) but it should not introduce and correctness issues.
2748 HeapRegion* curr_region = _g1h->heap_region_containing_raw(finger);
2749 HeapWord* bottom = curr_region->bottom();
2750 HeapWord* end = curr_region->end();
2751 HeapWord* limit = curr_region->next_top_at_mark_start();
2752
2753 if (verbose_low()) {
2754 gclog_or_tty->print_cr("[%d] curr_region = "PTR_FORMAT" "
2755 "["PTR_FORMAT", "PTR_FORMAT"), "
2756 "limit = "PTR_FORMAT,
2757 task_num, curr_region, bottom, end, limit);
2758 }
2759
2760 // Is the gap between reading the finger and doing the CAS too long?
2761 HeapWord* res = (HeapWord*) Atomic::cmpxchg_ptr(end, &_finger, finger);
2762 if (res == finger) {
2763 // we succeeded
2764
2765 // notice that _finger == end cannot be guaranteed here since,
2766 // someone else might have moved the finger even further
2767 assert(_finger >= end, "the finger should have moved forward");
2768
2769 if (verbose_low()) {
2770 gclog_or_tty->print_cr("[%d] we were successful with region = "
2771 PTR_FORMAT, task_num, curr_region);
2772 }
2773
2774 if (limit > bottom) {
2775 if (verbose_low()) {
2776 gclog_or_tty->print_cr("[%d] region "PTR_FORMAT" is not empty, "
2777 "returning it ", task_num, curr_region);
2778 }
2779 return curr_region;
2780 } else {
2781 assert(limit == bottom,
2782 "the region limit should be at bottom");
2783 if (verbose_low()) {
2784 gclog_or_tty->print_cr("[%d] region "PTR_FORMAT" is empty, "
2785 "returning NULL", task_num, curr_region);
2786 }
2787 // we return NULL and the caller should try calling
2788 // claim_region() again.
2789 return NULL;
2790 }
2791 } else {
2792 assert(_finger > finger, "the finger should have moved forward");
2793 if (verbose_low()) {
2794 gclog_or_tty->print_cr("[%d] somebody else moved the finger, "
2795 "global finger = "PTR_FORMAT", "
2796 "our finger = "PTR_FORMAT,
2797 task_num, _finger, finger);
2798 }
2799
2800 // read it again
2801 finger = _finger;
2802 }
2803 }
2804
2805 return NULL;
2806 }
2807
2808 bool ConcurrentMark::invalidate_aborted_regions_in_cset() {
2809 bool result = false;
2810 for (int i = 0; i < (int)_max_task_num; ++i) {
2811 CMTask* the_task = _tasks[i];
2812 MemRegion mr = the_task->aborted_region();
2813 if (mr.start() != NULL) {
2814 assert(mr.end() != NULL, "invariant");
2815 assert(mr.word_size() > 0, "invariant");
2816 HeapRegion* hr = _g1h->heap_region_containing(mr.start());
2817 assert(hr != NULL, "invariant");
2818 if (hr->in_collection_set()) {
2819 // The region points into the collection set
2820 the_task->set_aborted_region(MemRegion());
2821 result = true;
2822 }
2823 }
2824 }
2825 return result;
2826 }
2827
2828 bool ConcurrentMark::has_aborted_regions() {
2829 for (int i = 0; i < (int)_max_task_num; ++i) {
2830 CMTask* the_task = _tasks[i];
2831 MemRegion mr = the_task->aborted_region();
2832 if (mr.start() != NULL) {
2833 assert(mr.end() != NULL, "invariant");
2834 assert(mr.word_size() > 0, "invariant");
2835 return true;
2836 }
2837 }
2838 return false;
2839 }
2840
2841 void ConcurrentMark::oops_do(OopClosure* cl) {
2842 if (_markStack.size() > 0 && verbose_low()) {
2843 gclog_or_tty->print_cr("[global] scanning the global marking stack, "
2844 "size = %d", _markStack.size());
2845 }
2846 // we first iterate over the contents of the mark stack...
2847 _markStack.oops_do(cl);
2848
2849 for (int i = 0; i < (int)_max_task_num; ++i) {
2850 OopTaskQueue* queue = _task_queues->queue((int)i);
2851
2852 if (queue->size() > 0 && verbose_low()) {
2853 gclog_or_tty->print_cr("[global] scanning task queue of task %d, "
2854 "size = %d", i, queue->size());
2855 }
2856
2857 // ...then over the contents of the all the task queues.
2858 queue->oops_do(cl);
2859 }
2860
2861 // Invalidate any entries, that are in the region stack, that
2862 // point into the collection set
2863 if (_regionStack.invalidate_entries_into_cset()) {
2864 // otherwise, any gray objects copied during the evacuation pause
2865 // might not be visited.
2866 assert(_should_gray_objects, "invariant");
2867 }
2868
2869 // Invalidate any aborted regions, recorded in the individual CM
2870 // tasks, that point into the collection set.
2871 if (invalidate_aborted_regions_in_cset()) {
2872 // otherwise, any gray objects copied during the evacuation pause
2873 // might not be visited.
2874 assert(_should_gray_objects, "invariant");
2875 }
2876
2877 }
2878
2879 void ConcurrentMark::clear_marking_state(bool clear_overflow) {
2880 _markStack.setEmpty();
2881 _markStack.clear_overflow();
2882 _regionStack.setEmpty();
2883 _regionStack.clear_overflow();
2884 if (clear_overflow) {
2885 clear_has_overflown();
2886 } else {
2887 assert(has_overflown(), "pre-condition");
2888 }
2889 _finger = _heap_start;
2890
2891 for (int i = 0; i < (int)_max_task_num; ++i) {
2892 OopTaskQueue* queue = _task_queues->queue(i);
2893 queue->set_empty();
2894 // Clear any partial regions from the CMTasks
2895 _tasks[i]->clear_aborted_region();
2896 }
2897 }
2898
2899 void ConcurrentMark::print_stats() {
2900 if (verbose_stats()) {
2901 gclog_or_tty->print_cr("---------------------------------------------------------------------");
2902 for (size_t i = 0; i < _active_tasks; ++i) {
2903 _tasks[i]->print_stats();
2904 gclog_or_tty->print_cr("---------------------------------------------------------------------");
2905 }
2906 }
2907 }
2908
2909 // Closures used by ConcurrentMark::complete_marking_in_collection_set().
2910
2911 class CSetMarkOopClosure: public OopClosure {
2912 friend class CSetMarkBitMapClosure;
2913
2914 G1CollectedHeap* _g1h;
2915 CMBitMap* _bm;
2916 ConcurrentMark* _cm;
2917 oop* _ms;
2918 jint* _array_ind_stack;
2919 int _ms_size;
2920 int _ms_ind;
2921 int _array_increment;
2922 int _worker_i;
2923
2924 bool push(oop obj, int arr_ind = 0) {
2925 if (_ms_ind == _ms_size) {
2926 gclog_or_tty->print_cr("Mark stack is full.");
2927 return false;
2928 }
2929 _ms[_ms_ind] = obj;
2930 if (obj->is_objArray()) {
2931 _array_ind_stack[_ms_ind] = arr_ind;
2932 }
2933 _ms_ind++;
2934 return true;
2935 }
2936
2937 oop pop() {
2938 if (_ms_ind == 0) {
2939 return NULL;
2940 } else {
2941 _ms_ind--;
2942 return _ms[_ms_ind];
2943 }
2944 }
2945
2946 template <class T> bool drain() {
2947 while (_ms_ind > 0) {
2948 oop obj = pop();
2949 assert(obj != NULL, "Since index was non-zero.");
2950 if (obj->is_objArray()) {
2951 jint arr_ind = _array_ind_stack[_ms_ind];
2952 objArrayOop aobj = objArrayOop(obj);
2953 jint len = aobj->length();
2954 jint next_arr_ind = arr_ind + _array_increment;
2955 if (next_arr_ind < len) {
2956 push(obj, next_arr_ind);
2957 }
2958 // Now process this portion of this one.
2959 int lim = MIN2(next_arr_ind, len);
2960 for (int j = arr_ind; j < lim; j++) {
2961 do_oop(aobj->objArrayOopDesc::obj_at_addr<T>(j));
2962 }
2963 } else {
2964 obj->oop_iterate(this);
2965 }
2966 if (abort()) return false;
2967 }
2968 return true;
2969 }
2970
2971 public:
2972 CSetMarkOopClosure(ConcurrentMark* cm, int ms_size, int worker_i) :
2973 _g1h(G1CollectedHeap::heap()),
2974 _cm(cm),
2975 _bm(cm->nextMarkBitMap()),
2976 _ms_size(ms_size), _ms_ind(0),
2977 _ms(NEW_C_HEAP_ARRAY(oop, ms_size)),
2978 _array_ind_stack(NEW_C_HEAP_ARRAY(jint, ms_size)),
2979 _array_increment(MAX2(ms_size/8, 16)),
2980 _worker_i(worker_i) { }
2981
2982 ~CSetMarkOopClosure() {
2983 FREE_C_HEAP_ARRAY(oop, _ms);
2984 FREE_C_HEAP_ARRAY(jint, _array_ind_stack);
2985 }
2986
2987 virtual void do_oop(narrowOop* p) { do_oop_work(p); }
2988 virtual void do_oop( oop* p) { do_oop_work(p); }
2989
2990 template <class T> void do_oop_work(T* p) {
2991 T heap_oop = oopDesc::load_heap_oop(p);
2992 if (oopDesc::is_null(heap_oop)) return;
2993 oop obj = oopDesc::decode_heap_oop_not_null(heap_oop);
2994 if (obj->is_forwarded()) {
2995 // If the object has already been forwarded, we have to make sure
2996 // that it's marked. So follow the forwarding pointer. Note that
2997 // this does the right thing for self-forwarding pointers in the
2998 // evacuation failure case.
2999 obj = obj->forwardee();
3000 }
3001 HeapRegion* hr = _g1h->heap_region_containing(obj);
3002 if (hr != NULL) {
3003 if (hr->in_collection_set()) {
3004 if (_g1h->is_obj_ill(obj)) {
3005 if (_bm->parMark((HeapWord*)obj)) {
3006 if (!push(obj)) {
3007 gclog_or_tty->print_cr("Setting abort in CSetMarkOopClosure because push failed.");
3008 set_abort();
3009 }
3010 }
3011 }
3012 } else {
3013 // Outside the collection set; we need to gray it
3014 _cm->deal_with_reference(obj);
3015 }
3016 }
3017 }
3018 };
3019
3020 class CSetMarkBitMapClosure: public BitMapClosure {
3021 G1CollectedHeap* _g1h;
3022 CMBitMap* _bitMap;
3023 ConcurrentMark* _cm;
3024 CSetMarkOopClosure _oop_cl;
3025 int _worker_i;
3026
3027 public:
3028 CSetMarkBitMapClosure(ConcurrentMark* cm, int ms_size, int worker_i) :
3029 _g1h(G1CollectedHeap::heap()),
3030 _bitMap(cm->nextMarkBitMap()),
3031 _oop_cl(cm, ms_size, worker_i),
3032 _worker_i(worker_i) { }
3033
3034 bool do_bit(size_t offset) {
3035 // convert offset into a HeapWord*
3036 HeapWord* addr = _bitMap->offsetToHeapWord(offset);
3037 assert(_bitMap->endWord() && addr < _bitMap->endWord(),
3038 "address out of range");
3039 assert(_bitMap->isMarked(addr), "tautology");
3040 oop obj = oop(addr);
3041 if (!obj->is_forwarded()) {
3042 if (!_oop_cl.push(obj)) return false;
3043 if (UseCompressedOops) {
3044 if (!_oop_cl.drain<narrowOop>()) return false;
3045 } else {
3046 if (!_oop_cl.drain<oop>()) return false;
3047 }
3048 }
3049 // Otherwise...
3050 return true;
3051 }
3052 };
3053
3054 class CompleteMarkingInCSetHRClosure: public HeapRegionClosure {
3055 CMBitMap* _bm;
3056 CSetMarkBitMapClosure _bit_cl;
3057 int _worker_i;
3058
3059 enum SomePrivateConstants {
3060 MSSize = 1000
3061 };
3062
3063 public:
3064 CompleteMarkingInCSetHRClosure(ConcurrentMark* cm, int worker_i) :
3065 _bm(cm->nextMarkBitMap()),
3066 _bit_cl(cm, MSSize, worker_i),
3067 _worker_i(worker_i) { }
3068
3069 bool doHeapRegion(HeapRegion* hr) {
3070 if (hr->claimHeapRegion(HeapRegion::CompleteMarkCSetClaimValue)) {
3071 // The current worker has successfully claimed the region.
3072 if (!hr->evacuation_failed()) {
3073 MemRegion mr = MemRegion(hr->bottom(), hr->next_top_at_mark_start());
3074 if (!mr.is_empty()) {
3075 bool done = false;
3076 while (!done) {
3077 done = _bm->iterate(&_bit_cl, mr);
3078 }
3079 }
3080 }
3081 }
3082 return false;
3083 }
3084 };
3085
3086 class SetClaimValuesInCSetHRClosure: public HeapRegionClosure {
3087 jint _claim_value;
3088
3089 public:
3090 SetClaimValuesInCSetHRClosure(jint claim_value) :
3091 _claim_value(claim_value) { }
3092
3093 bool doHeapRegion(HeapRegion* hr) {
3094 hr->set_claim_value(_claim_value);
3095 return false;
3096 }
3097 };
3098
3099 class G1ParCompleteMarkInCSetTask: public AbstractGangTask {
3100 protected:
3101 G1CollectedHeap* _g1h;
3102 ConcurrentMark* _cm;
3103
3104 public:
3105 G1ParCompleteMarkInCSetTask(G1CollectedHeap* g1h,
3106 ConcurrentMark* cm) :
3107 AbstractGangTask("Complete Mark in CSet"),
3108 _g1h(g1h), _cm(cm) { }
3109
3110 void work(int worker_i) {
3111 CompleteMarkingInCSetHRClosure cmplt(_cm, worker_i);
3112 HeapRegion* hr = _g1h->start_cset_region_for_worker(worker_i);
3113 _g1h->collection_set_iterate_from(hr, &cmplt);
3114 }
3115 };
3116
3117 void ConcurrentMark::complete_marking_in_collection_set() {
3118 G1CollectedHeap* g1h = G1CollectedHeap::heap();
3119
3120 if (!g1h->mark_in_progress()) {
3121 g1h->g1_policy()->record_mark_closure_time(0.0);
3122 return;
3123 }
3124
3125 double start = os::elapsedTime();
3126 int n_workers = g1h->workers()->total_workers();
3127
3128 G1ParCompleteMarkInCSetTask complete_mark_task(g1h, this);
3129
3130 assert(g1h->check_cset_heap_region_claim_values(HeapRegion::InitialClaimValue), "sanity");
3131
3132 if (G1CollectedHeap::use_parallel_gc_threads()) {
3133 g1h->set_par_threads(n_workers);
3134 g1h->workers()->run_task(&complete_mark_task);
3135 g1h->set_par_threads(0);
3136 } else {
3137 complete_mark_task.work(0);
3138 }
3139
3140 assert(g1h->check_cset_heap_region_claim_values(HeapRegion::CompleteMarkCSetClaimValue), "sanity");
3141
3142 // Now reset the claim values in the regions in the collection set.
3143 SetClaimValuesInCSetHRClosure set_cv_cl(HeapRegion::InitialClaimValue);
3144 g1h->collection_set_iterate(&set_cv_cl);
3145
3146 assert(g1h->check_cset_heap_region_claim_values(HeapRegion::InitialClaimValue), "sanity");
3147
3148 double end_time = os::elapsedTime();
3149 double elapsed_time_ms = (end_time - start) * 1000.0;
3150 g1h->g1_policy()->record_mark_closure_time(elapsed_time_ms);
3151 }
3152
3153 // The next two methods deal with the following optimisation. Some
3154 // objects are gray by being marked and located above the finger. If
3155 // they are copied, during an evacuation pause, below the finger then
3156 // the need to be pushed on the stack. The observation is that, if
3157 // there are no regions in the collection set located above the
3158 // finger, then the above cannot happen, hence we do not need to
3159 // explicitly gray any objects when copying them to below the
3160 // finger. The global stack will be scanned to ensure that, if it
3161 // points to objects being copied, it will update their
3162 // location. There is a tricky situation with the gray objects in
3163 // region stack that are being coped, however. See the comment in
3164 // newCSet().
3165
3166 void ConcurrentMark::newCSet() {
3167 if (!concurrent_marking_in_progress()) {
3168 // nothing to do if marking is not in progress
3169 return;
3170 }
3171
3172 // find what the lowest finger is among the global and local fingers
3173 _min_finger = _finger;
3174 for (int i = 0; i < (int)_max_task_num; ++i) {
3175 CMTask* task = _tasks[i];
3176 HeapWord* task_finger = task->finger();
3177 if (task_finger != NULL && task_finger < _min_finger) {
3178 _min_finger = task_finger;
3179 }
3180 }
3181
3182 _should_gray_objects = false;
3183
3184 // This fixes a very subtle and fustrating bug. It might be the case
3185 // that, during en evacuation pause, heap regions that contain
3186 // objects that are gray (by being in regions contained in the
3187 // region stack) are included in the collection set. Since such gray
3188 // objects will be moved, and because it's not easy to redirect
3189 // region stack entries to point to a new location (because objects
3190 // in one region might be scattered to multiple regions after they
3191 // are copied), one option is to ensure that all marked objects
3192 // copied during a pause are pushed on the stack. Notice, however,
3193 // that this problem can only happen when the region stack is not
3194 // empty during an evacuation pause. So, we make the fix a bit less
3195 // conservative and ensure that regions are pushed on the stack,
3196 // irrespective whether all collection set regions are below the
3197 // finger, if the region stack is not empty. This is expected to be
3198 // a rare case, so I don't think it's necessary to be smarted about it.
3199 if (!region_stack_empty() || has_aborted_regions()) {
3200 _should_gray_objects = true;
3201 }
3202 }
3203
3204 void ConcurrentMark::registerCSetRegion(HeapRegion* hr) {
3205 if (!concurrent_marking_in_progress()) return;
3206
3207 HeapWord* region_end = hr->end();
3208 if (region_end > _min_finger) {
3209 _should_gray_objects = true;
3210 }
3211 }
3212
3213 // Resets the region fields of active CMTasks whose values point
3214 // into the collection set.
3215 void ConcurrentMark::reset_active_task_region_fields_in_cset() {
3216 assert(SafepointSynchronize::is_at_safepoint(), "should be in STW");
3217 assert(parallel_marking_threads() <= _max_task_num, "sanity");
3218
3219 for (int i = 0; i < (int)parallel_marking_threads(); i += 1) {
3220 CMTask* task = _tasks[i];
3221 HeapWord* task_finger = task->finger();
3222 if (task_finger != NULL) {
3223 assert(_g1h->is_in_g1_reserved(task_finger), "not in heap");
3224 HeapRegion* finger_region = _g1h->heap_region_containing(task_finger);
3225 if (finger_region->in_collection_set()) {
3226 // The task's current region is in the collection set.
3227 // This region will be evacuated in the current GC and
3228 // the region fields in the task will be stale.
3229 task->giveup_current_region();
3230 }
3231 }
3232 }
3233 }
3234
3235 // abandon current marking iteration due to a Full GC
3236 void ConcurrentMark::abort() {
3237 // Clear all marks to force marking thread to do nothing
3238 _nextMarkBitMap->clearAll();
3239 // Empty mark stack
3240 clear_marking_state();
3241 for (int i = 0; i < (int)_max_task_num; ++i) {
3242 _tasks[i]->clear_region_fields();
3243 }
3244 _has_aborted = true;
3245
3246 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3247 satb_mq_set.abandon_partial_marking();
3248 // This can be called either during or outside marking, we'll read
3249 // the expected_active value from the SATB queue set.
3250 satb_mq_set.set_active_all_threads(
3251 false, /* new active value */
3252 satb_mq_set.is_active() /* expected_active */);
3253 }
3254
3255 static void print_ms_time_info(const char* prefix, const char* name,
3256 NumberSeq& ns) {
3257 gclog_or_tty->print_cr("%s%5d %12s: total time = %8.2f s (avg = %8.2f ms).",
3258 prefix, ns.num(), name, ns.sum()/1000.0, ns.avg());
3259 if (ns.num() > 0) {
3260 gclog_or_tty->print_cr("%s [std. dev = %8.2f ms, max = %8.2f ms]",
3261 prefix, ns.sd(), ns.maximum());
3262 }
3263 }
3264
3265 void ConcurrentMark::print_summary_info() {
3266 gclog_or_tty->print_cr(" Concurrent marking:");
3267 print_ms_time_info(" ", "init marks", _init_times);
3268 print_ms_time_info(" ", "remarks", _remark_times);
3269 {
3270 print_ms_time_info(" ", "final marks", _remark_mark_times);
3271 print_ms_time_info(" ", "weak refs", _remark_weak_ref_times);
3272
3273 }
3274 print_ms_time_info(" ", "cleanups", _cleanup_times);
3275 gclog_or_tty->print_cr(" Final counting total time = %8.2f s (avg = %8.2f ms).",
3276 _total_counting_time,
3277 (_cleanup_times.num() > 0 ? _total_counting_time * 1000.0 /
3278 (double)_cleanup_times.num()
3279 : 0.0));
3280 if (G1ScrubRemSets) {
3281 gclog_or_tty->print_cr(" RS scrub total time = %8.2f s (avg = %8.2f ms).",
3282 _total_rs_scrub_time,
3283 (_cleanup_times.num() > 0 ? _total_rs_scrub_time * 1000.0 /
3284 (double)_cleanup_times.num()
3285 : 0.0));
3286 }
3287 gclog_or_tty->print_cr(" Total stop_world time = %8.2f s.",
3288 (_init_times.sum() + _remark_times.sum() +
3289 _cleanup_times.sum())/1000.0);
3290 gclog_or_tty->print_cr(" Total concurrent time = %8.2f s "
3291 "(%8.2f s marking, %8.2f s counting).",
3292 cmThread()->vtime_accum(),
3293 cmThread()->vtime_mark_accum(),
3294 cmThread()->vtime_count_accum());
3295 }
3296
3297 void ConcurrentMark::print_worker_threads_on(outputStream* st) const {
3298 _parallel_workers->print_worker_threads_on(st);
3299 }
3300
3301 // Closures
3302 // XXX: there seems to be a lot of code duplication here;
3303 // should refactor and consolidate the shared code.
3304
3305 // This closure is used to mark refs into the CMS generation in
3306 // the CMS bit map. Called at the first checkpoint.
3307
3308 // We take a break if someone is trying to stop the world.
3309 bool ConcurrentMark::do_yield_check(int worker_i) {
3310 if (should_yield()) {
3311 if (worker_i == 0) {
3312 _g1h->g1_policy()->record_concurrent_pause();
3313 }
3314 cmThread()->yield();
3315 if (worker_i == 0) {
3316 _g1h->g1_policy()->record_concurrent_pause_end();
3317 }
3318 return true;
3319 } else {
3320 return false;
3321 }
3322 }
3323
3324 bool ConcurrentMark::should_yield() {
3325 return cmThread()->should_yield();
3326 }
3327
3328 bool ConcurrentMark::containing_card_is_marked(void* p) {
3329 size_t offset = pointer_delta(p, _g1h->reserved_region().start(), 1);
3330 return _card_bm.at(offset >> CardTableModRefBS::card_shift);
3331 }
3332
3333 bool ConcurrentMark::containing_cards_are_marked(void* start,
3334 void* last) {
3335 return containing_card_is_marked(start) &&
3336 containing_card_is_marked(last);
3337 }
3338
3339 #ifndef PRODUCT
3340 // for debugging purposes
3341 void ConcurrentMark::print_finger() {
3342 gclog_or_tty->print_cr("heap ["PTR_FORMAT", "PTR_FORMAT"), global finger = "PTR_FORMAT,
3343 _heap_start, _heap_end, _finger);
3344 for (int i = 0; i < (int) _max_task_num; ++i) {
3345 gclog_or_tty->print(" %d: "PTR_FORMAT, i, _tasks[i]->finger());
3346 }
3347 gclog_or_tty->print_cr("");
3348 }
3349 #endif
3350
3351 void CMTask::scan_object(oop obj) {
3352 assert(_nextMarkBitMap->isMarked((HeapWord*) obj), "invariant");
3353
3354 if (_cm->verbose_high()) {
3355 gclog_or_tty->print_cr("[%d] we're scanning object "PTR_FORMAT,
3356 _task_id, (void*) obj);
3357 }
3358
3359 size_t obj_size = obj->size();
3360 _words_scanned += obj_size;
3361
3362 obj->oop_iterate(_cm_oop_closure);
3363 statsOnly( ++_objs_scanned );
3364 check_limits();
3365 }
3366
3367 // Closure for iteration over bitmaps
3368 class CMBitMapClosure : public BitMapClosure {
3369 private:
3370 // the bitmap that is being iterated over
3371 CMBitMap* _nextMarkBitMap;
3372 ConcurrentMark* _cm;
3373 CMTask* _task;
3374 // true if we're scanning a heap region claimed by the task (so that
3375 // we move the finger along), false if we're not, i.e. currently when
3376 // scanning a heap region popped from the region stack (so that we
3377 // do not move the task finger along; it'd be a mistake if we did so).
3378 bool _scanning_heap_region;
3379
3380 public:
3381 CMBitMapClosure(CMTask *task,
3382 ConcurrentMark* cm,
3383 CMBitMap* nextMarkBitMap)
3384 : _task(task), _cm(cm), _nextMarkBitMap(nextMarkBitMap) { }
3385
3386 void set_scanning_heap_region(bool scanning_heap_region) {
3387 _scanning_heap_region = scanning_heap_region;
3388 }
3389
3390 bool do_bit(size_t offset) {
3391 HeapWord* addr = _nextMarkBitMap->offsetToHeapWord(offset);
3392 assert(_nextMarkBitMap->isMarked(addr), "invariant");
3393 assert( addr < _cm->finger(), "invariant");
3394
3395 if (_scanning_heap_region) {
3396 statsOnly( _task->increase_objs_found_on_bitmap() );
3397 assert(addr >= _task->finger(), "invariant");
3398 // We move that task's local finger along.
3399 _task->move_finger_to(addr);
3400 } else {
3401 // We move the task's region finger along.
3402 _task->move_region_finger_to(addr);
3403 }
3404
3405 _task->scan_object(oop(addr));
3406 // we only partially drain the local queue and global stack
3407 _task->drain_local_queue(true);
3408 _task->drain_global_stack(true);
3409
3410 // if the has_aborted flag has been raised, we need to bail out of
3411 // the iteration
3412 return !_task->has_aborted();
3413 }
3414 };
3415
3416 // Closure for iterating over objects, currently only used for
3417 // processing SATB buffers.
3418 class CMObjectClosure : public ObjectClosure {
3419 private:
3420 CMTask* _task;
3421
3422 public:
3423 void do_object(oop obj) {
3424 _task->deal_with_reference(obj);
3425 }
3426
3427 CMObjectClosure(CMTask* task) : _task(task) { }
3428 };
3429
3430 G1CMOopClosure::G1CMOopClosure(G1CollectedHeap* g1h,
3431 ConcurrentMark* cm,
3432 CMTask* task)
3433 : _g1h(g1h), _cm(cm), _task(task) {
3434 assert(_ref_processor == NULL, "should be initialized to NULL");
3435
3436 if (G1UseConcMarkReferenceProcessing) {
3437 _ref_processor = g1h->ref_processor_cm();
3438 assert(_ref_processor != NULL, "should not be NULL");
3439 }
3440 }
3441
3442 void CMTask::setup_for_region(HeapRegion* hr) {
3443 // Separated the asserts so that we know which one fires.
3444 assert(hr != NULL,
3445 "claim_region() should have filtered out continues humongous regions");
3446 assert(!hr->continuesHumongous(),
3447 "claim_region() should have filtered out continues humongous regions");
3448
3449 if (_cm->verbose_low()) {
3450 gclog_or_tty->print_cr("[%d] setting up for region "PTR_FORMAT,
3451 _task_id, hr);
3452 }
3453
3454 _curr_region = hr;
3455 _finger = hr->bottom();
3456 update_region_limit();
3457 }
3458
3459 void CMTask::update_region_limit() {
3460 HeapRegion* hr = _curr_region;
3461 HeapWord* bottom = hr->bottom();
3462 HeapWord* limit = hr->next_top_at_mark_start();
3463
3464 if (limit == bottom) {
3465 if (_cm->verbose_low()) {
3466 gclog_or_tty->print_cr("[%d] found an empty region "
3467 "["PTR_FORMAT", "PTR_FORMAT")",
3468 _task_id, bottom, limit);
3469 }
3470 // The region was collected underneath our feet.
3471 // We set the finger to bottom to ensure that the bitmap
3472 // iteration that will follow this will not do anything.
3473 // (this is not a condition that holds when we set the region up,
3474 // as the region is not supposed to be empty in the first place)
3475 _finger = bottom;
3476 } else if (limit >= _region_limit) {
3477 assert(limit >= _finger, "peace of mind");
3478 } else {
3479 assert(limit < _region_limit, "only way to get here");
3480 // This can happen under some pretty unusual circumstances. An
3481 // evacuation pause empties the region underneath our feet (NTAMS
3482 // at bottom). We then do some allocation in the region (NTAMS
3483 // stays at bottom), followed by the region being used as a GC
3484 // alloc region (NTAMS will move to top() and the objects
3485 // originally below it will be grayed). All objects now marked in
3486 // the region are explicitly grayed, if below the global finger,
3487 // and we do not need in fact to scan anything else. So, we simply
3488 // set _finger to be limit to ensure that the bitmap iteration
3489 // doesn't do anything.
3490 _finger = limit;
3491 }
3492
3493 _region_limit = limit;
3494 }
3495
3496 void CMTask::giveup_current_region() {
3497 assert(_curr_region != NULL, "invariant");
3498 if (_cm->verbose_low()) {
3499 gclog_or_tty->print_cr("[%d] giving up region "PTR_FORMAT,
3500 _task_id, _curr_region);
3501 }
3502 clear_region_fields();
3503 }
3504
3505 void CMTask::clear_region_fields() {
3506 // Values for these three fields that indicate that we're not
3507 // holding on to a region.
3508 _curr_region = NULL;
3509 _finger = NULL;
3510 _region_limit = NULL;
3511
3512 _region_finger = NULL;
3513 }
3514
3515 void CMTask::set_cm_oop_closure(G1CMOopClosure* cm_oop_closure) {
3516 if (cm_oop_closure == NULL) {
3517 assert(_cm_oop_closure != NULL, "invariant");
3518 } else {
3519 assert(_cm_oop_closure == NULL, "invariant");
3520 }
3521 _cm_oop_closure = cm_oop_closure;
3522 }
3523
3524 void CMTask::reset(CMBitMap* nextMarkBitMap) {
3525 guarantee(nextMarkBitMap != NULL, "invariant");
3526
3527 if (_cm->verbose_low()) {
3528 gclog_or_tty->print_cr("[%d] resetting", _task_id);
3529 }
3530
3531 _nextMarkBitMap = nextMarkBitMap;
3532 clear_region_fields();
3533 assert(_aborted_region.is_empty(), "should have been cleared");
3534
3535 _calls = 0;
3536 _elapsed_time_ms = 0.0;
3537 _termination_time_ms = 0.0;
3538 _termination_start_time_ms = 0.0;
3539
3540 #if _MARKING_STATS_
3541 _local_pushes = 0;
3542 _local_pops = 0;
3543 _local_max_size = 0;
3544 _objs_scanned = 0;
3545 _global_pushes = 0;
3546 _global_pops = 0;
3547 _global_max_size = 0;
3548 _global_transfers_to = 0;
3549 _global_transfers_from = 0;
3550 _region_stack_pops = 0;
3551 _regions_claimed = 0;
3552 _objs_found_on_bitmap = 0;
3553 _satb_buffers_processed = 0;
3554 _steal_attempts = 0;
3555 _steals = 0;
3556 _aborted = 0;
3557 _aborted_overflow = 0;
3558 _aborted_cm_aborted = 0;
3559 _aborted_yield = 0;
3560 _aborted_timed_out = 0;
3561 _aborted_satb = 0;
3562 _aborted_termination = 0;
3563 #endif // _MARKING_STATS_
3564 }
3565
3566 bool CMTask::should_exit_termination() {
3567 regular_clock_call();
3568 // This is called when we are in the termination protocol. We should
3569 // quit if, for some reason, this task wants to abort or the global
3570 // stack is not empty (this means that we can get work from it).
3571 return !_cm->mark_stack_empty() || has_aborted();
3572 }
3573
3574 void CMTask::reached_limit() {
3575 assert(_words_scanned >= _words_scanned_limit ||
3576 _refs_reached >= _refs_reached_limit ,
3577 "shouldn't have been called otherwise");
3578 regular_clock_call();
3579 }
3580
3581 void CMTask::regular_clock_call() {
3582 if (has_aborted()) return;
3583
3584 // First, we need to recalculate the words scanned and refs reached
3585 // limits for the next clock call.
3586 recalculate_limits();
3587
3588 // During the regular clock call we do the following
3589
3590 // (1) If an overflow has been flagged, then we abort.
3591 if (_cm->has_overflown()) {
3592 set_has_aborted();
3593 return;
3594 }
3595
3596 // If we are not concurrent (i.e. we're doing remark) we don't need
3597 // to check anything else. The other steps are only needed during
3598 // the concurrent marking phase.
3599 if (!concurrent()) return;
3600
3601 // (2) If marking has been aborted for Full GC, then we also abort.
3602 if (_cm->has_aborted()) {
3603 set_has_aborted();
3604 statsOnly( ++_aborted_cm_aborted );
3605 return;
3606 }
3607
3608 double curr_time_ms = os::elapsedVTime() * 1000.0;
3609
3610 // (3) If marking stats are enabled, then we update the step history.
3611 #if _MARKING_STATS_
3612 if (_words_scanned >= _words_scanned_limit) {
3613 ++_clock_due_to_scanning;
3614 }
3615 if (_refs_reached >= _refs_reached_limit) {
3616 ++_clock_due_to_marking;
3617 }
3618
3619 double last_interval_ms = curr_time_ms - _interval_start_time_ms;
3620 _interval_start_time_ms = curr_time_ms;
3621 _all_clock_intervals_ms.add(last_interval_ms);
3622
3623 if (_cm->verbose_medium()) {
3624 gclog_or_tty->print_cr("[%d] regular clock, interval = %1.2lfms, "
3625 "scanned = %d%s, refs reached = %d%s",
3626 _task_id, last_interval_ms,
3627 _words_scanned,
3628 (_words_scanned >= _words_scanned_limit) ? " (*)" : "",
3629 _refs_reached,
3630 (_refs_reached >= _refs_reached_limit) ? " (*)" : "");
3631 }
3632 #endif // _MARKING_STATS_
3633
3634 // (4) We check whether we should yield. If we have to, then we abort.
3635 if (_cm->should_yield()) {
3636 // We should yield. To do this we abort the task. The caller is
3637 // responsible for yielding.
3638 set_has_aborted();
3639 statsOnly( ++_aborted_yield );
3640 return;
3641 }
3642
3643 // (5) We check whether we've reached our time quota. If we have,
3644 // then we abort.
3645 double elapsed_time_ms = curr_time_ms - _start_time_ms;
3646 if (elapsed_time_ms > _time_target_ms) {
3647 set_has_aborted();
3648 _has_timed_out = true;
3649 statsOnly( ++_aborted_timed_out );
3650 return;
3651 }
3652
3653 // (6) Finally, we check whether there are enough completed STAB
3654 // buffers available for processing. If there are, we abort.
3655 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3656 if (!_draining_satb_buffers && satb_mq_set.process_completed_buffers()) {
3657 if (_cm->verbose_low()) {
3658 gclog_or_tty->print_cr("[%d] aborting to deal with pending SATB buffers",
3659 _task_id);
3660 }
3661 // we do need to process SATB buffers, we'll abort and restart
3662 // the marking task to do so
3663 set_has_aborted();
3664 statsOnly( ++_aborted_satb );
3665 return;
3666 }
3667 }
3668
3669 void CMTask::recalculate_limits() {
3670 _real_words_scanned_limit = _words_scanned + words_scanned_period;
3671 _words_scanned_limit = _real_words_scanned_limit;
3672
3673 _real_refs_reached_limit = _refs_reached + refs_reached_period;
3674 _refs_reached_limit = _real_refs_reached_limit;
3675 }
3676
3677 void CMTask::decrease_limits() {
3678 // This is called when we believe that we're going to do an infrequent
3679 // operation which will increase the per byte scanned cost (i.e. move
3680 // entries to/from the global stack). It basically tries to decrease the
3681 // scanning limit so that the clock is called earlier.
3682
3683 if (_cm->verbose_medium()) {
3684 gclog_or_tty->print_cr("[%d] decreasing limits", _task_id);
3685 }
3686
3687 _words_scanned_limit = _real_words_scanned_limit -
3688 3 * words_scanned_period / 4;
3689 _refs_reached_limit = _real_refs_reached_limit -
3690 3 * refs_reached_period / 4;
3691 }
3692
3693 void CMTask::move_entries_to_global_stack() {
3694 // local array where we'll store the entries that will be popped
3695 // from the local queue
3696 oop buffer[global_stack_transfer_size];
3697
3698 int n = 0;
3699 oop obj;
3700 while (n < global_stack_transfer_size && _task_queue->pop_local(obj)) {
3701 buffer[n] = obj;
3702 ++n;
3703 }
3704
3705 if (n > 0) {
3706 // we popped at least one entry from the local queue
3707
3708 statsOnly( ++_global_transfers_to; _local_pops += n );
3709
3710 if (!_cm->mark_stack_push(buffer, n)) {
3711 if (_cm->verbose_low()) {
3712 gclog_or_tty->print_cr("[%d] aborting due to global stack overflow",
3713 _task_id);
3714 }
3715 set_has_aborted();
3716 } else {
3717 // the transfer was successful
3718
3719 if (_cm->verbose_medium()) {
3720 gclog_or_tty->print_cr("[%d] pushed %d entries to the global stack",
3721 _task_id, n);
3722 }
3723 statsOnly( int tmp_size = _cm->mark_stack_size();
3724 if (tmp_size > _global_max_size) {
3725 _global_max_size = tmp_size;
3726 }
3727 _global_pushes += n );
3728 }
3729 }
3730
3731 // this operation was quite expensive, so decrease the limits
3732 decrease_limits();
3733 }
3734
3735 void CMTask::get_entries_from_global_stack() {
3736 // local array where we'll store the entries that will be popped
3737 // from the global stack.
3738 oop buffer[global_stack_transfer_size];
3739 int n;
3740 _cm->mark_stack_pop(buffer, global_stack_transfer_size, &n);
3741 assert(n <= global_stack_transfer_size,
3742 "we should not pop more than the given limit");
3743 if (n > 0) {
3744 // yes, we did actually pop at least one entry
3745
3746 statsOnly( ++_global_transfers_from; _global_pops += n );
3747 if (_cm->verbose_medium()) {
3748 gclog_or_tty->print_cr("[%d] popped %d entries from the global stack",
3749 _task_id, n);
3750 }
3751 for (int i = 0; i < n; ++i) {
3752 bool success = _task_queue->push(buffer[i]);
3753 // We only call this when the local queue is empty or under a
3754 // given target limit. So, we do not expect this push to fail.
3755 assert(success, "invariant");
3756 }
3757
3758 statsOnly( int tmp_size = _task_queue->size();
3759 if (tmp_size > _local_max_size) {
3760 _local_max_size = tmp_size;
3761 }
3762 _local_pushes += n );
3763 }
3764
3765 // this operation was quite expensive, so decrease the limits
3766 decrease_limits();
3767 }
3768
3769 void CMTask::drain_local_queue(bool partially) {
3770 if (has_aborted()) return;
3771
3772 // Decide what the target size is, depending whether we're going to
3773 // drain it partially (so that other tasks can steal if they run out
3774 // of things to do) or totally (at the very end).
3775 size_t target_size;
3776 if (partially) {
3777 target_size = MIN2((size_t)_task_queue->max_elems()/3, GCDrainStackTargetSize);
3778 } else {
3779 target_size = 0;
3780 }
3781
3782 if (_task_queue->size() > target_size) {
3783 if (_cm->verbose_high()) {
3784 gclog_or_tty->print_cr("[%d] draining local queue, target size = %d",
3785 _task_id, target_size);
3786 }
3787
3788 oop obj;
3789 bool ret = _task_queue->pop_local(obj);
3790 while (ret) {
3791 statsOnly( ++_local_pops );
3792
3793 if (_cm->verbose_high()) {
3794 gclog_or_tty->print_cr("[%d] popped "PTR_FORMAT, _task_id,
3795 (void*) obj);
3796 }
3797
3798 assert(_g1h->is_in_g1_reserved((HeapWord*) obj), "invariant" );
3799 assert(!_g1h->is_on_master_free_list(
3800 _g1h->heap_region_containing((HeapWord*) obj)), "invariant");
3801
3802 scan_object(obj);
3803
3804 if (_task_queue->size() <= target_size || has_aborted()) {
3805 ret = false;
3806 } else {
3807 ret = _task_queue->pop_local(obj);
3808 }
3809 }
3810
3811 if (_cm->verbose_high()) {
3812 gclog_or_tty->print_cr("[%d] drained local queue, size = %d",
3813 _task_id, _task_queue->size());
3814 }
3815 }
3816 }
3817
3818 void CMTask::drain_global_stack(bool partially) {
3819 if (has_aborted()) return;
3820
3821 // We have a policy to drain the local queue before we attempt to
3822 // drain the global stack.
3823 assert(partially || _task_queue->size() == 0, "invariant");
3824
3825 // Decide what the target size is, depending whether we're going to
3826 // drain it partially (so that other tasks can steal if they run out
3827 // of things to do) or totally (at the very end). Notice that,
3828 // because we move entries from the global stack in chunks or
3829 // because another task might be doing the same, we might in fact
3830 // drop below the target. But, this is not a problem.
3831 size_t target_size;
3832 if (partially) {
3833 target_size = _cm->partial_mark_stack_size_target();
3834 } else {
3835 target_size = 0;
3836 }
3837
3838 if (_cm->mark_stack_size() > target_size) {
3839 if (_cm->verbose_low()) {
3840 gclog_or_tty->print_cr("[%d] draining global_stack, target size %d",
3841 _task_id, target_size);
3842 }
3843
3844 while (!has_aborted() && _cm->mark_stack_size() > target_size) {
3845 get_entries_from_global_stack();
3846 drain_local_queue(partially);
3847 }
3848
3849 if (_cm->verbose_low()) {
3850 gclog_or_tty->print_cr("[%d] drained global stack, size = %d",
3851 _task_id, _cm->mark_stack_size());
3852 }
3853 }
3854 }
3855
3856 // SATB Queue has several assumptions on whether to call the par or
3857 // non-par versions of the methods. this is why some of the code is
3858 // replicated. We should really get rid of the single-threaded version
3859 // of the code to simplify things.
3860 void CMTask::drain_satb_buffers() {
3861 if (has_aborted()) return;
3862
3863 // We set this so that the regular clock knows that we're in the
3864 // middle of draining buffers and doesn't set the abort flag when it
3865 // notices that SATB buffers are available for draining. It'd be
3866 // very counter productive if it did that. :-)
3867 _draining_satb_buffers = true;
3868
3869 CMObjectClosure oc(this);
3870 SATBMarkQueueSet& satb_mq_set = JavaThread::satb_mark_queue_set();
3871 if (G1CollectedHeap::use_parallel_gc_threads()) {
3872 satb_mq_set.set_par_closure(_task_id, &oc);
3873 } else {
3874 satb_mq_set.set_closure(&oc);
3875 }
3876
3877 // This keeps claiming and applying the closure to completed buffers
3878 // until we run out of buffers or we need to abort.
3879 if (G1CollectedHeap::use_parallel_gc_threads()) {
3880 while (!has_aborted() &&
3881 satb_mq_set.par_apply_closure_to_completed_buffer(_task_id)) {
3882 if (_cm->verbose_medium()) {
3883 gclog_or_tty->print_cr("[%d] processed an SATB buffer", _task_id);
3884 }
3885 statsOnly( ++_satb_buffers_processed );
3886 regular_clock_call();
3887 }
3888 } else {
3889 while (!has_aborted() &&
3890 satb_mq_set.apply_closure_to_completed_buffer()) {
3891 if (_cm->verbose_medium()) {
3892 gclog_or_tty->print_cr("[%d] processed an SATB buffer", _task_id);
3893 }
3894 statsOnly( ++_satb_buffers_processed );
3895 regular_clock_call();
3896 }
3897 }
3898
3899 if (!concurrent() && !has_aborted()) {
3900 // We should only do this during remark.
3901 if (G1CollectedHeap::use_parallel_gc_threads()) {
3902 satb_mq_set.par_iterate_closure_all_threads(_task_id);
3903 } else {
3904 satb_mq_set.iterate_closure_all_threads();
3905 }
3906 }
3907
3908 _draining_satb_buffers = false;
3909
3910 assert(has_aborted() ||
3911 concurrent() ||
3912 satb_mq_set.completed_buffers_num() == 0, "invariant");
3913
3914 if (G1CollectedHeap::use_parallel_gc_threads()) {
3915 satb_mq_set.set_par_closure(_task_id, NULL);
3916 } else {
3917 satb_mq_set.set_closure(NULL);
3918 }
3919
3920 // again, this was a potentially expensive operation, decrease the
3921 // limits to get the regular clock call early
3922 decrease_limits();
3923 }
3924
3925 void CMTask::drain_region_stack(BitMapClosure* bc) {
3926 if (has_aborted()) return;
3927
3928 assert(_region_finger == NULL,
3929 "it should be NULL when we're not scanning a region");
3930
3931 if (!_cm->region_stack_empty() || !_aborted_region.is_empty()) {
3932 if (_cm->verbose_low()) {
3933 gclog_or_tty->print_cr("[%d] draining region stack, size = %d",
3934 _task_id, _cm->region_stack_size());
3935 }
3936
3937 MemRegion mr;
3938
3939 if (!_aborted_region.is_empty()) {
3940 mr = _aborted_region;
3941 _aborted_region = MemRegion();
3942
3943 if (_cm->verbose_low()) {
3944 gclog_or_tty->print_cr("[%d] scanning aborted region "
3945 "[ " PTR_FORMAT ", " PTR_FORMAT " )",
3946 _task_id, mr.start(), mr.end());
3947 }
3948 } else {
3949 mr = _cm->region_stack_pop_lock_free();
3950 // it returns MemRegion() if the pop fails
3951 statsOnly(if (mr.start() != NULL) ++_region_stack_pops );
3952 }
3953
3954 while (mr.start() != NULL) {
3955 if (_cm->verbose_medium()) {
3956 gclog_or_tty->print_cr("[%d] we are scanning region "
3957 "["PTR_FORMAT", "PTR_FORMAT")",
3958 _task_id, mr.start(), mr.end());
3959 }
3960
3961 assert(mr.end() <= _cm->finger(),
3962 "otherwise the region shouldn't be on the stack");
3963 assert(!mr.is_empty(), "Only non-empty regions live on the region stack");
3964 if (_nextMarkBitMap->iterate(bc, mr)) {
3965 assert(!has_aborted(),
3966 "cannot abort the task without aborting the bitmap iteration");
3967
3968 // We finished iterating over the region without aborting.
3969 regular_clock_call();
3970 if (has_aborted()) {
3971 mr = MemRegion();
3972 } else {
3973 mr = _cm->region_stack_pop_lock_free();
3974 // it returns MemRegion() if the pop fails
3975 statsOnly(if (mr.start() != NULL) ++_region_stack_pops );
3976 }
3977 } else {
3978 assert(has_aborted(), "currently the only way to do so");
3979
3980 // The only way to abort the bitmap iteration is to return
3981 // false from the do_bit() method. However, inside the
3982 // do_bit() method we move the _region_finger to point to the
3983 // object currently being looked at. So, if we bail out, we
3984 // have definitely set _region_finger to something non-null.
3985 assert(_region_finger != NULL, "invariant");
3986
3987 // Make sure that any previously aborted region has been
3988 // cleared.
3989 assert(_aborted_region.is_empty(), "aborted region not cleared");
3990
3991 // The iteration was actually aborted. So now _region_finger
3992 // points to the address of the object we last scanned. If we
3993 // leave it there, when we restart this task, we will rescan
3994 // the object. It is easy to avoid this. We move the finger by
3995 // enough to point to the next possible object header (the
3996 // bitmap knows by how much we need to move it as it knows its
3997 // granularity).
3998 MemRegion newRegion =
3999 MemRegion(_nextMarkBitMap->nextWord(_region_finger), mr.end());
4000
4001 if (!newRegion.is_empty()) {
4002 if (_cm->verbose_low()) {
4003 gclog_or_tty->print_cr("[%d] recording unscanned region"
4004 "[" PTR_FORMAT "," PTR_FORMAT ") in CMTask",
4005 _task_id,
4006 newRegion.start(), newRegion.end());
4007 }
4008 // Now record the part of the region we didn't scan to
4009 // make sure this task scans it later.
4010 _aborted_region = newRegion;
4011 }
4012 // break from while
4013 mr = MemRegion();
4014 }
4015 _region_finger = NULL;
4016 }
4017
4018 if (_cm->verbose_low()) {
4019 gclog_or_tty->print_cr("[%d] drained region stack, size = %d",
4020 _task_id, _cm->region_stack_size());
4021 }
4022 }
4023 }
4024
4025 void CMTask::print_stats() {
4026 gclog_or_tty->print_cr("Marking Stats, task = %d, calls = %d",
4027 _task_id, _calls);
4028 gclog_or_tty->print_cr(" Elapsed time = %1.2lfms, Termination time = %1.2lfms",
4029 _elapsed_time_ms, _termination_time_ms);
4030 gclog_or_tty->print_cr(" Step Times (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
4031 _step_times_ms.num(), _step_times_ms.avg(),
4032 _step_times_ms.sd());
4033 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
4034 _step_times_ms.maximum(), _step_times_ms.sum());
4035
4036 #if _MARKING_STATS_
4037 gclog_or_tty->print_cr(" Clock Intervals (cum): num = %d, avg = %1.2lfms, sd = %1.2lfms",
4038 _all_clock_intervals_ms.num(), _all_clock_intervals_ms.avg(),
4039 _all_clock_intervals_ms.sd());
4040 gclog_or_tty->print_cr(" max = %1.2lfms, total = %1.2lfms",
4041 _all_clock_intervals_ms.maximum(),
4042 _all_clock_intervals_ms.sum());
4043 gclog_or_tty->print_cr(" Clock Causes (cum): scanning = %d, marking = %d",
4044 _clock_due_to_scanning, _clock_due_to_marking);
4045 gclog_or_tty->print_cr(" Objects: scanned = %d, found on the bitmap = %d",
4046 _objs_scanned, _objs_found_on_bitmap);
4047 gclog_or_tty->print_cr(" Local Queue: pushes = %d, pops = %d, max size = %d",
4048 _local_pushes, _local_pops, _local_max_size);
4049 gclog_or_tty->print_cr(" Global Stack: pushes = %d, pops = %d, max size = %d",
4050 _global_pushes, _global_pops, _global_max_size);
4051 gclog_or_tty->print_cr(" transfers to = %d, transfers from = %d",
4052 _global_transfers_to,_global_transfers_from);
4053 gclog_or_tty->print_cr(" Regions: claimed = %d, Region Stack: pops = %d",
4054 _regions_claimed, _region_stack_pops);
4055 gclog_or_tty->print_cr(" SATB buffers: processed = %d", _satb_buffers_processed);
4056 gclog_or_tty->print_cr(" Steals: attempts = %d, successes = %d",
4057 _steal_attempts, _steals);
4058 gclog_or_tty->print_cr(" Aborted: %d, due to", _aborted);
4059 gclog_or_tty->print_cr(" overflow: %d, global abort: %d, yield: %d",
4060 _aborted_overflow, _aborted_cm_aborted, _aborted_yield);
4061 gclog_or_tty->print_cr(" time out: %d, SATB: %d, termination: %d",
4062 _aborted_timed_out, _aborted_satb, _aborted_termination);
4063 #endif // _MARKING_STATS_
4064 }
4065
4066 /*****************************************************************************
4067
4068 The do_marking_step(time_target_ms) method is the building block
4069 of the parallel marking framework. It can be called in parallel
4070 with other invocations of do_marking_step() on different tasks
4071 (but only one per task, obviously) and concurrently with the
4072 mutator threads, or during remark, hence it eliminates the need
4073 for two versions of the code. When called during remark, it will
4074 pick up from where the task left off during the concurrent marking
4075 phase. Interestingly, tasks are also claimable during evacuation
4076 pauses too, since do_marking_step() ensures that it aborts before
4077 it needs to yield.
4078
4079 The data structures that is uses to do marking work are the
4080 following:
4081
4082 (1) Marking Bitmap. If there are gray objects that appear only
4083 on the bitmap (this happens either when dealing with an overflow
4084 or when the initial marking phase has simply marked the roots
4085 and didn't push them on the stack), then tasks claim heap
4086 regions whose bitmap they then scan to find gray objects. A
4087 global finger indicates where the end of the last claimed region
4088 is. A local finger indicates how far into the region a task has
4089 scanned. The two fingers are used to determine how to gray an
4090 object (i.e. whether simply marking it is OK, as it will be
4091 visited by a task in the future, or whether it needs to be also
4092 pushed on a stack).
4093
4094 (2) Local Queue. The local queue of the task which is accessed
4095 reasonably efficiently by the task. Other tasks can steal from
4096 it when they run out of work. Throughout the marking phase, a
4097 task attempts to keep its local queue short but not totally
4098 empty, so that entries are available for stealing by other
4099 tasks. Only when there is no more work, a task will totally
4100 drain its local queue.
4101
4102 (3) Global Mark Stack. This handles local queue overflow. During
4103 marking only sets of entries are moved between it and the local
4104 queues, as access to it requires a mutex and more fine-grain
4105 interaction with it which might cause contention. If it
4106 overflows, then the marking phase should restart and iterate
4107 over the bitmap to identify gray objects. Throughout the marking
4108 phase, tasks attempt to keep the global mark stack at a small
4109 length but not totally empty, so that entries are available for
4110 popping by other tasks. Only when there is no more work, tasks
4111 will totally drain the global mark stack.
4112
4113 (4) Global Region Stack. Entries on it correspond to areas of
4114 the bitmap that need to be scanned since they contain gray
4115 objects. Pushes on the region stack only happen during
4116 evacuation pauses and typically correspond to areas covered by
4117 GC LABS. If it overflows, then the marking phase should restart
4118 and iterate over the bitmap to identify gray objects. Tasks will
4119 try to totally drain the region stack as soon as possible.
4120
4121 (5) SATB Buffer Queue. This is where completed SATB buffers are
4122 made available. Buffers are regularly removed from this queue
4123 and scanned for roots, so that the queue doesn't get too
4124 long. During remark, all completed buffers are processed, as
4125 well as the filled in parts of any uncompleted buffers.
4126
4127 The do_marking_step() method tries to abort when the time target
4128 has been reached. There are a few other cases when the
4129 do_marking_step() method also aborts:
4130
4131 (1) When the marking phase has been aborted (after a Full GC).
4132
4133 (2) When a global overflow (either on the global stack or the
4134 region stack) has been triggered. Before the task aborts, it
4135 will actually sync up with the other tasks to ensure that all
4136 the marking data structures (local queues, stacks, fingers etc.)
4137 are re-initialised so that when do_marking_step() completes,
4138 the marking phase can immediately restart.
4139
4140 (3) When enough completed SATB buffers are available. The
4141 do_marking_step() method only tries to drain SATB buffers right
4142 at the beginning. So, if enough buffers are available, the
4143 marking step aborts and the SATB buffers are processed at
4144 the beginning of the next invocation.
4145
4146 (4) To yield. when we have to yield then we abort and yield
4147 right at the end of do_marking_step(). This saves us from a lot
4148 of hassle as, by yielding we might allow a Full GC. If this
4149 happens then objects will be compacted underneath our feet, the
4150 heap might shrink, etc. We save checking for this by just
4151 aborting and doing the yield right at the end.
4152
4153 From the above it follows that the do_marking_step() method should
4154 be called in a loop (or, otherwise, regularly) until it completes.
4155
4156 If a marking step completes without its has_aborted() flag being
4157 true, it means it has completed the current marking phase (and
4158 also all other marking tasks have done so and have all synced up).
4159
4160 A method called regular_clock_call() is invoked "regularly" (in
4161 sub ms intervals) throughout marking. It is this clock method that
4162 checks all the abort conditions which were mentioned above and
4163 decides when the task should abort. A work-based scheme is used to
4164 trigger this clock method: when the number of object words the
4165 marking phase has scanned or the number of references the marking
4166 phase has visited reach a given limit. Additional invocations to
4167 the method clock have been planted in a few other strategic places
4168 too. The initial reason for the clock method was to avoid calling
4169 vtime too regularly, as it is quite expensive. So, once it was in
4170 place, it was natural to piggy-back all the other conditions on it
4171 too and not constantly check them throughout the code.
4172
4173 *****************************************************************************/
4174
4175 void CMTask::do_marking_step(double time_target_ms,
4176 bool do_stealing,
4177 bool do_termination) {
4178 assert(time_target_ms >= 1.0, "minimum granularity is 1ms");
4179 assert(concurrent() == _cm->concurrent(), "they should be the same");
4180
4181 assert(concurrent() || _cm->region_stack_empty(),
4182 "the region stack should have been cleared before remark");
4183 assert(concurrent() || !_cm->has_aborted_regions(),
4184 "aborted regions should have been cleared before remark");
4185 assert(_region_finger == NULL,
4186 "this should be non-null only when a region is being scanned");
4187
4188 G1CollectorPolicy* g1_policy = _g1h->g1_policy();
4189 assert(_task_queues != NULL, "invariant");
4190 assert(_task_queue != NULL, "invariant");
4191 assert(_task_queues->queue(_task_id) == _task_queue, "invariant");
4192
4193 assert(!_claimed,
4194 "only one thread should claim this task at any one time");
4195
4196 // OK, this doesn't safeguard again all possible scenarios, as it is
4197 // possible for two threads to set the _claimed flag at the same
4198 // time. But it is only for debugging purposes anyway and it will
4199 // catch most problems.
4200 _claimed = true;
4201
4202 _start_time_ms = os::elapsedVTime() * 1000.0;
4203 statsOnly( _interval_start_time_ms = _start_time_ms );
4204
4205 double diff_prediction_ms =
4206 g1_policy->get_new_prediction(&_marking_step_diffs_ms);
4207 _time_target_ms = time_target_ms - diff_prediction_ms;
4208
4209 // set up the variables that are used in the work-based scheme to
4210 // call the regular clock method
4211 _words_scanned = 0;
4212 _refs_reached = 0;
4213 recalculate_limits();
4214
4215 // clear all flags
4216 clear_has_aborted();
4217 _has_timed_out = false;
4218 _draining_satb_buffers = false;
4219
4220 ++_calls;
4221
4222 if (_cm->verbose_low()) {
4223 gclog_or_tty->print_cr("[%d] >>>>>>>>>> START, call = %d, "
4224 "target = %1.2lfms >>>>>>>>>>",
4225 _task_id, _calls, _time_target_ms);
4226 }
4227
4228 // Set up the bitmap and oop closures. Anything that uses them is
4229 // eventually called from this method, so it is OK to allocate these
4230 // statically.
4231 CMBitMapClosure bitmap_closure(this, _cm, _nextMarkBitMap);
4232 G1CMOopClosure cm_oop_closure(_g1h, _cm, this);
4233 set_cm_oop_closure(&cm_oop_closure);
4234
4235 if (_cm->has_overflown()) {
4236 // This can happen if the region stack or the mark stack overflows
4237 // during a GC pause and this task, after a yield point,
4238 // restarts. We have to abort as we need to get into the overflow
4239 // protocol which happens right at the end of this task.
4240 set_has_aborted();
4241 }
4242
4243 // First drain any available SATB buffers. After this, we will not
4244 // look at SATB buffers before the next invocation of this method.
4245 // If enough completed SATB buffers are queued up, the regular clock
4246 // will abort this task so that it restarts.
4247 drain_satb_buffers();
4248 // ...then partially drain the local queue and the global stack
4249 drain_local_queue(true);
4250 drain_global_stack(true);
4251
4252 // Then totally drain the region stack. We will not look at
4253 // it again before the next invocation of this method. Entries on
4254 // the region stack are only added during evacuation pauses, for
4255 // which we have to yield. When we do, we abort the task anyway so
4256 // it will look at the region stack again when it restarts.
4257 bitmap_closure.set_scanning_heap_region(false);
4258 drain_region_stack(&bitmap_closure);
4259 // ...then partially drain the local queue and the global stack
4260 drain_local_queue(true);
4261 drain_global_stack(true);
4262
4263 do {
4264 if (!has_aborted() && _curr_region != NULL) {
4265 // This means that we're already holding on to a region.
4266 assert(_finger != NULL, "if region is not NULL, then the finger "
4267 "should not be NULL either");
4268
4269 // We might have restarted this task after an evacuation pause
4270 // which might have evacuated the region we're holding on to
4271 // underneath our feet. Let's read its limit again to make sure
4272 // that we do not iterate over a region of the heap that
4273 // contains garbage (update_region_limit() will also move
4274 // _finger to the start of the region if it is found empty).
4275 update_region_limit();
4276 // We will start from _finger not from the start of the region,
4277 // as we might be restarting this task after aborting half-way
4278 // through scanning this region. In this case, _finger points to
4279 // the address where we last found a marked object. If this is a
4280 // fresh region, _finger points to start().
4281 MemRegion mr = MemRegion(_finger, _region_limit);
4282
4283 if (_cm->verbose_low()) {
4284 gclog_or_tty->print_cr("[%d] we're scanning part "
4285 "["PTR_FORMAT", "PTR_FORMAT") "
4286 "of region "PTR_FORMAT,
4287 _task_id, _finger, _region_limit, _curr_region);
4288 }
4289
4290 // Let's iterate over the bitmap of the part of the
4291 // region that is left.
4292 bitmap_closure.set_scanning_heap_region(true);
4293 if (mr.is_empty() ||
4294 _nextMarkBitMap->iterate(&bitmap_closure, mr)) {
4295 // We successfully completed iterating over the region. Now,
4296 // let's give up the region.
4297 giveup_current_region();
4298 regular_clock_call();
4299 } else {
4300 assert(has_aborted(), "currently the only way to do so");
4301 // The only way to abort the bitmap iteration is to return
4302 // false from the do_bit() method. However, inside the
4303 // do_bit() method we move the _finger to point to the
4304 // object currently being looked at. So, if we bail out, we
4305 // have definitely set _finger to something non-null.
4306 assert(_finger != NULL, "invariant");
4307
4308 // Region iteration was actually aborted. So now _finger
4309 // points to the address of the object we last scanned. If we
4310 // leave it there, when we restart this task, we will rescan
4311 // the object. It is easy to avoid this. We move the finger by
4312 // enough to point to the next possible object header (the
4313 // bitmap knows by how much we need to move it as it knows its
4314 // granularity).
4315 assert(_finger < _region_limit, "invariant");
4316 HeapWord* new_finger = _nextMarkBitMap->nextWord(_finger);
4317 // Check if bitmap iteration was aborted while scanning the last object
4318 if (new_finger >= _region_limit) {
4319 giveup_current_region();
4320 } else {
4321 move_finger_to(new_finger);
4322 }
4323 }
4324 }
4325 // At this point we have either completed iterating over the
4326 // region we were holding on to, or we have aborted.
4327
4328 // We then partially drain the local queue and the global stack.
4329 // (Do we really need this?)
4330 drain_local_queue(true);
4331 drain_global_stack(true);
4332
4333 // Read the note on the claim_region() method on why it might
4334 // return NULL with potentially more regions available for
4335 // claiming and why we have to check out_of_regions() to determine
4336 // whether we're done or not.
4337 while (!has_aborted() && _curr_region == NULL && !_cm->out_of_regions()) {
4338 // We are going to try to claim a new region. We should have
4339 // given up on the previous one.
4340 // Separated the asserts so that we know which one fires.
4341 assert(_curr_region == NULL, "invariant");
4342 assert(_finger == NULL, "invariant");
4343 assert(_region_limit == NULL, "invariant");
4344 if (_cm->verbose_low()) {
4345 gclog_or_tty->print_cr("[%d] trying to claim a new region", _task_id);
4346 }
4347 HeapRegion* claimed_region = _cm->claim_region(_task_id);
4348 if (claimed_region != NULL) {
4349 // Yes, we managed to claim one
4350 statsOnly( ++_regions_claimed );
4351
4352 if (_cm->verbose_low()) {
4353 gclog_or_tty->print_cr("[%d] we successfully claimed "
4354 "region "PTR_FORMAT,
4355 _task_id, claimed_region);
4356 }
4357
4358 setup_for_region(claimed_region);
4359 assert(_curr_region == claimed_region, "invariant");
4360 }
4361 // It is important to call the regular clock here. It might take
4362 // a while to claim a region if, for example, we hit a large
4363 // block of empty regions. So we need to call the regular clock
4364 // method once round the loop to make sure it's called
4365 // frequently enough.
4366 regular_clock_call();
4367 }
4368
4369 if (!has_aborted() && _curr_region == NULL) {
4370 assert(_cm->out_of_regions(),
4371 "at this point we should be out of regions");
4372 }
4373 } while ( _curr_region != NULL && !has_aborted());
4374
4375 if (!has_aborted()) {
4376 // We cannot check whether the global stack is empty, since other
4377 // tasks might be pushing objects to it concurrently. We also cannot
4378 // check if the region stack is empty because if a thread is aborting
4379 // it can push a partially done region back.
4380 assert(_cm->out_of_regions(),
4381 "at this point we should be out of regions");
4382
4383 if (_cm->verbose_low()) {
4384 gclog_or_tty->print_cr("[%d] all regions claimed", _task_id);
4385 }
4386
4387 // Try to reduce the number of available SATB buffers so that
4388 // remark has less work to do.
4389 drain_satb_buffers();
4390 }
4391
4392 // Since we've done everything else, we can now totally drain the
4393 // local queue and global stack.
4394 drain_local_queue(false);
4395 drain_global_stack(false);
4396
4397 // Attempt at work stealing from other task's queues.
4398 if (do_stealing && !has_aborted()) {
4399 // We have not aborted. This means that we have finished all that
4400 // we could. Let's try to do some stealing...
4401
4402 // We cannot check whether the global stack is empty, since other
4403 // tasks might be pushing objects to it concurrently. We also cannot
4404 // check if the region stack is empty because if a thread is aborting
4405 // it can push a partially done region back.
4406 assert(_cm->out_of_regions() && _task_queue->size() == 0,
4407 "only way to reach here");
4408
4409 if (_cm->verbose_low()) {
4410 gclog_or_tty->print_cr("[%d] starting to steal", _task_id);
4411 }
4412
4413 while (!has_aborted()) {
4414 oop obj;
4415 statsOnly( ++_steal_attempts );
4416
4417 if (_cm->try_stealing(_task_id, &_hash_seed, obj)) {
4418 if (_cm->verbose_medium()) {
4419 gclog_or_tty->print_cr("[%d] stolen "PTR_FORMAT" successfully",
4420 _task_id, (void*) obj);
4421 }
4422
4423 statsOnly( ++_steals );
4424
4425 assert(_nextMarkBitMap->isMarked((HeapWord*) obj),
4426 "any stolen object should be marked");
4427 scan_object(obj);
4428
4429 // And since we're towards the end, let's totally drain the
4430 // local queue and global stack.
4431 drain_local_queue(false);
4432 drain_global_stack(false);
4433 } else {
4434 break;
4435 }
4436 }
4437 }
4438
4439 // If we are about to wrap up and go into termination, check if we
4440 // should raise the overflow flag.
4441 if (do_termination && !has_aborted()) {
4442 if (_cm->force_overflow()->should_force()) {
4443 _cm->set_has_overflown();
4444 regular_clock_call();
4445 }
4446 }
4447
4448 // We still haven't aborted. Now, let's try to get into the
4449 // termination protocol.
4450 if (do_termination && !has_aborted()) {
4451 // We cannot check whether the global stack is empty, since other
4452 // tasks might be concurrently pushing objects on it. We also cannot
4453 // check if the region stack is empty because if a thread is aborting
4454 // it can push a partially done region back.
4455 // Separated the asserts so that we know which one fires.
4456 assert(_cm->out_of_regions(), "only way to reach here");
4457 assert(_task_queue->size() == 0, "only way to reach here");
4458
4459 if (_cm->verbose_low()) {
4460 gclog_or_tty->print_cr("[%d] starting termination protocol", _task_id);
4461 }
4462
4463 _termination_start_time_ms = os::elapsedVTime() * 1000.0;
4464 // The CMTask class also extends the TerminatorTerminator class,
4465 // hence its should_exit_termination() method will also decide
4466 // whether to exit the termination protocol or not.
4467 bool finished = _cm->terminator()->offer_termination(this);
4468 double termination_end_time_ms = os::elapsedVTime() * 1000.0;
4469 _termination_time_ms +=
4470 termination_end_time_ms - _termination_start_time_ms;
4471
4472 if (finished) {
4473 // We're all done.
4474
4475 if (_task_id == 0) {
4476 // let's allow task 0 to do this
4477 if (concurrent()) {
4478 assert(_cm->concurrent_marking_in_progress(), "invariant");
4479 // we need to set this to false before the next
4480 // safepoint. This way we ensure that the marking phase
4481 // doesn't observe any more heap expansions.
4482 _cm->clear_concurrent_marking_in_progress();
4483 }
4484 }
4485
4486 // We can now guarantee that the global stack is empty, since
4487 // all other tasks have finished. We separated the guarantees so
4488 // that, if a condition is false, we can immediately find out
4489 // which one.
4490 guarantee(_cm->out_of_regions(), "only way to reach here");
4491 guarantee(_aborted_region.is_empty(), "only way to reach here");
4492 guarantee(_cm->region_stack_empty(), "only way to reach here");
4493 guarantee(_cm->mark_stack_empty(), "only way to reach here");
4494 guarantee(_task_queue->size() == 0, "only way to reach here");
4495 guarantee(!_cm->has_overflown(), "only way to reach here");
4496 guarantee(!_cm->mark_stack_overflow(), "only way to reach here");
4497 guarantee(!_cm->region_stack_overflow(), "only way to reach here");
4498
4499 if (_cm->verbose_low()) {
4500 gclog_or_tty->print_cr("[%d] all tasks terminated", _task_id);
4501 }
4502 } else {
4503 // Apparently there's more work to do. Let's abort this task. It
4504 // will restart it and we can hopefully find more things to do.
4505
4506 if (_cm->verbose_low()) {
4507 gclog_or_tty->print_cr("[%d] apparently there is more work to do",
4508 _task_id);
4509 }
4510
4511 set_has_aborted();
4512 statsOnly( ++_aborted_termination );
4513 }
4514 }
4515
4516 // Mainly for debugging purposes to make sure that a pointer to the
4517 // closure which was statically allocated in this frame doesn't
4518 // escape it by accident.
4519 set_cm_oop_closure(NULL);
4520 double end_time_ms = os::elapsedVTime() * 1000.0;
4521 double elapsed_time_ms = end_time_ms - _start_time_ms;
4522 // Update the step history.
4523 _step_times_ms.add(elapsed_time_ms);
4524
4525 if (has_aborted()) {
4526 // The task was aborted for some reason.
4527
4528 statsOnly( ++_aborted );
4529
4530 if (_has_timed_out) {
4531 double diff_ms = elapsed_time_ms - _time_target_ms;
4532 // Keep statistics of how well we did with respect to hitting
4533 // our target only if we actually timed out (if we aborted for
4534 // other reasons, then the results might get skewed).
4535 _marking_step_diffs_ms.add(diff_ms);
4536 }
4537
4538 if (_cm->has_overflown()) {
4539 // This is the interesting one. We aborted because a global
4540 // overflow was raised. This means we have to restart the
4541 // marking phase and start iterating over regions. However, in
4542 // order to do this we have to make sure that all tasks stop
4543 // what they are doing and re-initialise in a safe manner. We
4544 // will achieve this with the use of two barrier sync points.
4545
4546 if (_cm->verbose_low()) {
4547 gclog_or_tty->print_cr("[%d] detected overflow", _task_id);
4548 }
4549
4550 _cm->enter_first_sync_barrier(_task_id);
4551 // When we exit this sync barrier we know that all tasks have
4552 // stopped doing marking work. So, it's now safe to
4553 // re-initialise our data structures. At the end of this method,
4554 // task 0 will clear the global data structures.
4555
4556 statsOnly( ++_aborted_overflow );
4557
4558 // We clear the local state of this task...
4559 clear_region_fields();
4560
4561 // ...and enter the second barrier.
4562 _cm->enter_second_sync_barrier(_task_id);
4563 // At this point everything has bee re-initialised and we're
4564 // ready to restart.
4565 }
4566
4567 if (_cm->verbose_low()) {
4568 gclog_or_tty->print_cr("[%d] <<<<<<<<<< ABORTING, target = %1.2lfms, "
4569 "elapsed = %1.2lfms <<<<<<<<<<",
4570 _task_id, _time_target_ms, elapsed_time_ms);
4571 if (_cm->has_aborted()) {
4572 gclog_or_tty->print_cr("[%d] ========== MARKING ABORTED ==========",
4573 _task_id);
4574 }
4575 }
4576 } else {
4577 if (_cm->verbose_low()) {
4578 gclog_or_tty->print_cr("[%d] <<<<<<<<<< FINISHED, target = %1.2lfms, "
4579 "elapsed = %1.2lfms <<<<<<<<<<",
4580 _task_id, _time_target_ms, elapsed_time_ms);
4581 }
4582 }
4583
4584 _claimed = false;
4585 }
4586
4587 CMTask::CMTask(int task_id,
4588 ConcurrentMark* cm,
4589 CMTaskQueue* task_queue,
4590 CMTaskQueueSet* task_queues)
4591 : _g1h(G1CollectedHeap::heap()),
4592 _task_id(task_id), _cm(cm),
4593 _claimed(false),
4594 _nextMarkBitMap(NULL), _hash_seed(17),
4595 _task_queue(task_queue),
4596 _task_queues(task_queues),
4597 _cm_oop_closure(NULL),
4598 _aborted_region(MemRegion()) {
4599 guarantee(task_queue != NULL, "invariant");
4600 guarantee(task_queues != NULL, "invariant");
4601
4602 statsOnly( _clock_due_to_scanning = 0;
4603 _clock_due_to_marking = 0 );
4604
4605 _marking_step_diffs_ms.add(0.5);
4606 }
4607
4608 // These are formatting macros that are used below to ensure
4609 // consistent formatting. The *_H_* versions are used to format the
4610 // header for a particular value and they should be kept consistent
4611 // with the corresponding macro. Also note that most of the macros add
4612 // the necessary white space (as a prefix) which makes them a bit
4613 // easier to compose.
4614
4615 // All the output lines are prefixed with this string to be able to
4616 // identify them easily in a large log file.
4617 #define G1PPRL_LINE_PREFIX "###"
4618
4619 #define G1PPRL_ADDR_BASE_FORMAT " "PTR_FORMAT"-"PTR_FORMAT
4620 #ifdef _LP64
4621 #define G1PPRL_ADDR_BASE_H_FORMAT " %37s"
4622 #else // _LP64
4623 #define G1PPRL_ADDR_BASE_H_FORMAT " %21s"
4624 #endif // _LP64
4625
4626 // For per-region info
4627 #define G1PPRL_TYPE_FORMAT " %-4s"
4628 #define G1PPRL_TYPE_H_FORMAT " %4s"
4629 #define G1PPRL_BYTE_FORMAT " "SIZE_FORMAT_W(9)
4630 #define G1PPRL_BYTE_H_FORMAT " %9s"
4631 #define G1PPRL_DOUBLE_FORMAT " %14.1f"
4632 #define G1PPRL_DOUBLE_H_FORMAT " %14s"
4633
4634 // For summary info
4635 #define G1PPRL_SUM_ADDR_FORMAT(tag) " "tag":"G1PPRL_ADDR_BASE_FORMAT
4636 #define G1PPRL_SUM_BYTE_FORMAT(tag) " "tag": "SIZE_FORMAT
4637 #define G1PPRL_SUM_MB_FORMAT(tag) " "tag": %1.2f MB"
4638 #define G1PPRL_SUM_MB_PERC_FORMAT(tag) G1PPRL_SUM_MB_FORMAT(tag)" / %1.2f %%"
4639
4640 G1PrintRegionLivenessInfoClosure::
4641 G1PrintRegionLivenessInfoClosure(outputStream* out, const char* phase_name)
4642 : _out(out),
4643 _total_used_bytes(0), _total_capacity_bytes(0),
4644 _total_prev_live_bytes(0), _total_next_live_bytes(0),
4645 _hum_used_bytes(0), _hum_capacity_bytes(0),
4646 _hum_prev_live_bytes(0), _hum_next_live_bytes(0) {
4647 G1CollectedHeap* g1h = G1CollectedHeap::heap();
4648 MemRegion g1_committed = g1h->g1_committed();
4649 MemRegion g1_reserved = g1h->g1_reserved();
4650 double now = os::elapsedTime();
4651
4652 // Print the header of the output.
4653 _out->cr();
4654 _out->print_cr(G1PPRL_LINE_PREFIX" PHASE %s @ %1.3f", phase_name, now);
4655 _out->print_cr(G1PPRL_LINE_PREFIX" HEAP"
4656 G1PPRL_SUM_ADDR_FORMAT("committed")
4657 G1PPRL_SUM_ADDR_FORMAT("reserved")
4658 G1PPRL_SUM_BYTE_FORMAT("region-size"),
4659 g1_committed.start(), g1_committed.end(),
4660 g1_reserved.start(), g1_reserved.end(),
4661 HeapRegion::GrainBytes);
4662 _out->print_cr(G1PPRL_LINE_PREFIX);
4663 _out->print_cr(G1PPRL_LINE_PREFIX
4664 G1PPRL_TYPE_H_FORMAT
4665 G1PPRL_ADDR_BASE_H_FORMAT
4666 G1PPRL_BYTE_H_FORMAT
4667 G1PPRL_BYTE_H_FORMAT
4668 G1PPRL_BYTE_H_FORMAT
4669 G1PPRL_DOUBLE_H_FORMAT,
4670 "type", "address-range",
4671 "used", "prev-live", "next-live", "gc-eff");
4672 _out->print_cr(G1PPRL_LINE_PREFIX
4673 G1PPRL_TYPE_H_FORMAT
4674 G1PPRL_ADDR_BASE_H_FORMAT
4675 G1PPRL_BYTE_H_FORMAT
4676 G1PPRL_BYTE_H_FORMAT
4677 G1PPRL_BYTE_H_FORMAT
4678 G1PPRL_DOUBLE_H_FORMAT,
4679 "", "",
4680 "(bytes)", "(bytes)", "(bytes)", "(bytes/ms)");
4681 }
4682
4683 // It takes as a parameter a reference to one of the _hum_* fields, it
4684 // deduces the corresponding value for a region in a humongous region
4685 // series (either the region size, or what's left if the _hum_* field
4686 // is < the region size), and updates the _hum_* field accordingly.
4687 size_t G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* hum_bytes) {
4688 size_t bytes = 0;
4689 // The > 0 check is to deal with the prev and next live bytes which
4690 // could be 0.
4691 if (*hum_bytes > 0) {
4692 bytes = MIN2(HeapRegion::GrainBytes, *hum_bytes);
4693 *hum_bytes -= bytes;
4694 }
4695 return bytes;
4696 }
4697
4698 // It deduces the values for a region in a humongous region series
4699 // from the _hum_* fields and updates those accordingly. It assumes
4700 // that that _hum_* fields have already been set up from the "starts
4701 // humongous" region and we visit the regions in address order.
4702 void G1PrintRegionLivenessInfoClosure::get_hum_bytes(size_t* used_bytes,
4703 size_t* capacity_bytes,
4704 size_t* prev_live_bytes,
4705 size_t* next_live_bytes) {
4706 assert(_hum_used_bytes > 0 && _hum_capacity_bytes > 0, "pre-condition");
4707 *used_bytes = get_hum_bytes(&_hum_used_bytes);
4708 *capacity_bytes = get_hum_bytes(&_hum_capacity_bytes);
4709 *prev_live_bytes = get_hum_bytes(&_hum_prev_live_bytes);
4710 *next_live_bytes = get_hum_bytes(&_hum_next_live_bytes);
4711 }
4712
4713 bool G1PrintRegionLivenessInfoClosure::doHeapRegion(HeapRegion* r) {
4714 const char* type = "";
4715 HeapWord* bottom = r->bottom();
4716 HeapWord* end = r->end();
4717 size_t capacity_bytes = r->capacity();
4718 size_t used_bytes = r->used();
4719 size_t prev_live_bytes = r->live_bytes();
4720 size_t next_live_bytes = r->next_live_bytes();
4721 double gc_eff = r->gc_efficiency();
4722 if (r->used() == 0) {
4723 type = "FREE";
4724 } else if (r->is_survivor()) {
4725 type = "SURV";
4726 } else if (r->is_young()) {
4727 type = "EDEN";
4728 } else if (r->startsHumongous()) {
4729 type = "HUMS";
4730
4731 assert(_hum_used_bytes == 0 && _hum_capacity_bytes == 0 &&
4732 _hum_prev_live_bytes == 0 && _hum_next_live_bytes == 0,
4733 "they should have been zeroed after the last time we used them");
4734 // Set up the _hum_* fields.
4735 _hum_capacity_bytes = capacity_bytes;
4736 _hum_used_bytes = used_bytes;
4737 _hum_prev_live_bytes = prev_live_bytes;
4738 _hum_next_live_bytes = next_live_bytes;
4739 get_hum_bytes(&used_bytes, &capacity_bytes,
4740 &prev_live_bytes, &next_live_bytes);
4741 end = bottom + HeapRegion::GrainWords;
4742 } else if (r->continuesHumongous()) {
4743 type = "HUMC";
4744 get_hum_bytes(&used_bytes, &capacity_bytes,
4745 &prev_live_bytes, &next_live_bytes);
4746 assert(end == bottom + HeapRegion::GrainWords, "invariant");
4747 } else {
4748 type = "OLD";
4749 }
4750
4751 _total_used_bytes += used_bytes;
4752 _total_capacity_bytes += capacity_bytes;
4753 _total_prev_live_bytes += prev_live_bytes;
4754 _total_next_live_bytes += next_live_bytes;
4755
4756 // Print a line for this particular region.
4757 _out->print_cr(G1PPRL_LINE_PREFIX
4758 G1PPRL_TYPE_FORMAT
4759 G1PPRL_ADDR_BASE_FORMAT
4760 G1PPRL_BYTE_FORMAT
4761 G1PPRL_BYTE_FORMAT
4762 G1PPRL_BYTE_FORMAT
4763 G1PPRL_DOUBLE_FORMAT,
4764 type, bottom, end,
4765 used_bytes, prev_live_bytes, next_live_bytes, gc_eff);
4766
4767 return false;
4768 }
4769
4770 G1PrintRegionLivenessInfoClosure::~G1PrintRegionLivenessInfoClosure() {
4771 // Print the footer of the output.
4772 _out->print_cr(G1PPRL_LINE_PREFIX);
4773 _out->print_cr(G1PPRL_LINE_PREFIX
4774 " SUMMARY"
4775 G1PPRL_SUM_MB_FORMAT("capacity")
4776 G1PPRL_SUM_MB_PERC_FORMAT("used")
4777 G1PPRL_SUM_MB_PERC_FORMAT("prev-live")
4778 G1PPRL_SUM_MB_PERC_FORMAT("next-live"),
4779 bytes_to_mb(_total_capacity_bytes),
4780 bytes_to_mb(_total_used_bytes),
4781 perc(_total_used_bytes, _total_capacity_bytes),
4782 bytes_to_mb(_total_prev_live_bytes),
4783 perc(_total_prev_live_bytes, _total_capacity_bytes),
4784 bytes_to_mb(_total_next_live_bytes),
4785 perc(_total_next_live_bytes, _total_capacity_bytes));
4786 _out->cr();
4787 }
--- EOF ---