1 /*
2 * Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25 #include "precompiled.hpp"
26 #include "gc_implementation/concurrentMarkSweep/cmsLockVerifier.hpp"
27 #include "gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.hpp"
28 #include "gc_implementation/concurrentMarkSweep/concurrentMarkSweepGeneration.inline.hpp"
29 #include "gc_implementation/concurrentMarkSweep/concurrentMarkSweepThread.hpp"
30 #include "gc_implementation/shared/liveRange.hpp"
31 #include "gc_implementation/shared/spaceDecorator.hpp"
32 #include "gc_interface/collectedHeap.inline.hpp"
33 #include "memory/allocation.inline.hpp"
34 #include "memory/blockOffsetTable.inline.hpp"
35 #include "memory/genCollectedHeap.hpp"
36 #include "memory/resourceArea.hpp"
37 #include "memory/space.inline.hpp"
38 #include "memory/universe.inline.hpp"
39 #include "oops/oop.inline.hpp"
40 #include "runtime/globals.hpp"
41 #include "runtime/handles.inline.hpp"
42 #include "runtime/init.hpp"
43 #include "runtime/java.hpp"
44 #include "runtime/orderAccess.inline.hpp"
45 #include "runtime/vmThread.hpp"
46 #include "utilities/copy.hpp"
47
48 /////////////////////////////////////////////////////////////////////////
49 //// CompactibleFreeListSpace
50 /////////////////////////////////////////////////////////////////////////
51
52 // highest ranked free list lock rank
53 int CompactibleFreeListSpace::_lockRank = Mutex::leaf + 3;
54
55 // Defaults are 0 so things will break badly if incorrectly initialized.
56 size_t CompactibleFreeListSpace::IndexSetStart = 0;
57 size_t CompactibleFreeListSpace::IndexSetStride = 0;
58
59 size_t MinChunkSize = 0;
60
61 void CompactibleFreeListSpace::set_cms_values() {
62 // Set CMS global values
63 assert(MinChunkSize == 0, "already set");
64
65 // MinChunkSize should be a multiple of MinObjAlignment and be large enough
66 // for chunks to contain a FreeChunk.
67 size_t min_chunk_size_in_bytes = align_size_up(sizeof(FreeChunk), MinObjAlignmentInBytes);
68 MinChunkSize = min_chunk_size_in_bytes / BytesPerWord;
69
70 assert(IndexSetStart == 0 && IndexSetStride == 0, "already set");
71 IndexSetStart = MinChunkSize;
72 IndexSetStride = MinObjAlignment;
73 }
74
75 // Constructor
76 CompactibleFreeListSpace::CompactibleFreeListSpace(BlockOffsetSharedArray* bs,
77 MemRegion mr, bool use_adaptive_freelists,
78 FreeBlockDictionary<FreeChunk>::DictionaryChoice dictionaryChoice) :
79 _dictionaryChoice(dictionaryChoice),
80 _adaptive_freelists(use_adaptive_freelists),
81 _bt(bs, mr),
82 // free list locks are in the range of values taken by _lockRank
83 // This range currently is [_leaf+2, _leaf+3]
84 // Note: this requires that CFLspace c'tors
85 // are called serially in the order in which the locks are
86 // are acquired in the program text. This is true today.
87 _freelistLock(_lockRank--, "CompactibleFreeListSpace._lock", true,
88 Monitor::_safepoint_check_sometimes),
89 _parDictionaryAllocLock(Mutex::leaf - 1, // == rank(ExpandHeap_lock) - 1
90 "CompactibleFreeListSpace._dict_par_lock", true,
91 Monitor::_safepoint_check_never),
92 _rescan_task_size(CardTableModRefBS::card_size_in_words * BitsPerWord *
93 CMSRescanMultiple),
94 _marking_task_size(CardTableModRefBS::card_size_in_words * BitsPerWord *
95 CMSConcMarkMultiple),
96 _collector(NULL),
97 _preconsumptionDirtyCardClosure(NULL)
98 {
99 assert(sizeof(FreeChunk) / BytesPerWord <= MinChunkSize,
100 "FreeChunk is larger than expected");
101 _bt.set_space(this);
102 initialize(mr, SpaceDecorator::Clear, SpaceDecorator::Mangle);
103 // We have all of "mr", all of which we place in the dictionary
104 // as one big chunk. We'll need to decide here which of several
105 // possible alternative dictionary implementations to use. For
106 // now the choice is easy, since we have only one working
107 // implementation, namely, the simple binary tree (splaying
108 // temporarily disabled).
109 switch (dictionaryChoice) {
110 case FreeBlockDictionary<FreeChunk>::dictionaryBinaryTree:
111 _dictionary = new AFLBinaryTreeDictionary(mr);
112 break;
113 case FreeBlockDictionary<FreeChunk>::dictionarySplayTree:
114 case FreeBlockDictionary<FreeChunk>::dictionarySkipList:
115 default:
116 warning("dictionaryChoice: selected option not understood; using"
117 " default BinaryTreeDictionary implementation instead.");
118 }
119 assert(_dictionary != NULL, "CMS dictionary initialization");
120 // The indexed free lists are initially all empty and are lazily
121 // filled in on demand. Initialize the array elements to NULL.
122 initializeIndexedFreeListArray();
123
124 // Not using adaptive free lists assumes that allocation is first
125 // from the linAB's. Also a cms perm gen which can be compacted
126 // has to have the klass's klassKlass allocated at a lower
127 // address in the heap than the klass so that the klassKlass is
128 // moved to its new location before the klass is moved.
129 // Set the _refillSize for the linear allocation blocks
130 if (!use_adaptive_freelists) {
131 FreeChunk* fc = _dictionary->get_chunk(mr.word_size(),
132 FreeBlockDictionary<FreeChunk>::atLeast);
133 // The small linAB initially has all the space and will allocate
134 // a chunk of any size.
135 HeapWord* addr = (HeapWord*) fc;
136 _smallLinearAllocBlock.set(addr, fc->size() ,
137 1024*SmallForLinearAlloc, fc->size());
138 // Note that _unallocated_block is not updated here.
139 // Allocations from the linear allocation block should
140 // update it.
141 } else {
142 _smallLinearAllocBlock.set(0, 0, 1024*SmallForLinearAlloc,
143 SmallForLinearAlloc);
144 }
145 // CMSIndexedFreeListReplenish should be at least 1
146 CMSIndexedFreeListReplenish = MAX2((uintx)1, CMSIndexedFreeListReplenish);
147 _promoInfo.setSpace(this);
148 if (UseCMSBestFit) {
149 _fitStrategy = FreeBlockBestFitFirst;
150 } else {
151 _fitStrategy = FreeBlockStrategyNone;
152 }
153 check_free_list_consistency();
154
155 // Initialize locks for parallel case.
156 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
157 _indexedFreeListParLocks[i] = new Mutex(Mutex::leaf - 1, // == ExpandHeap_lock - 1
158 "a freelist par lock", true, Mutex::_safepoint_check_sometimes);
159 DEBUG_ONLY(
160 _indexedFreeList[i].set_protecting_lock(_indexedFreeListParLocks[i]);
161 )
162 }
163 _dictionary->set_par_lock(&_parDictionaryAllocLock);
164 }
165
166 // Like CompactibleSpace forward() but always calls cross_threshold() to
167 // update the block offset table. Removed initialize_threshold call because
168 // CFLS does not use a block offset array for contiguous spaces.
169 HeapWord* CompactibleFreeListSpace::forward(oop q, size_t size,
170 CompactPoint* cp, HeapWord* compact_top) {
171 // q is alive
172 // First check if we should switch compaction space
173 assert(this == cp->space, "'this' should be current compaction space.");
174 size_t compaction_max_size = pointer_delta(end(), compact_top);
175 assert(adjustObjectSize(size) == cp->space->adjust_object_size_v(size),
176 "virtual adjustObjectSize_v() method is not correct");
177 size_t adjusted_size = adjustObjectSize(size);
178 assert(compaction_max_size >= MinChunkSize || compaction_max_size == 0,
179 "no small fragments allowed");
180 assert(minimum_free_block_size() == MinChunkSize,
181 "for de-virtualized reference below");
182 // Can't leave a nonzero size, residual fragment smaller than MinChunkSize
183 if (adjusted_size + MinChunkSize > compaction_max_size &&
184 adjusted_size != compaction_max_size) {
185 do {
186 // switch to next compaction space
187 cp->space->set_compaction_top(compact_top);
188 cp->space = cp->space->next_compaction_space();
189 if (cp->space == NULL) {
190 cp->gen = GenCollectedHeap::heap()->young_gen();
191 assert(cp->gen != NULL, "compaction must succeed");
192 cp->space = cp->gen->first_compaction_space();
193 assert(cp->space != NULL, "generation must have a first compaction space");
194 }
195 compact_top = cp->space->bottom();
196 cp->space->set_compaction_top(compact_top);
197 // The correct adjusted_size may not be the same as that for this method
198 // (i.e., cp->space may no longer be "this" so adjust the size again.
199 // Use the virtual method which is not used above to save the virtual
200 // dispatch.
201 adjusted_size = cp->space->adjust_object_size_v(size);
202 compaction_max_size = pointer_delta(cp->space->end(), compact_top);
203 assert(cp->space->minimum_free_block_size() == 0, "just checking");
204 } while (adjusted_size > compaction_max_size);
205 }
206
207 // store the forwarding pointer into the mark word
208 if ((HeapWord*)q != compact_top) {
209 q->forward_to(oop(compact_top));
210 assert(q->is_gc_marked(), "encoding the pointer should preserve the mark");
211 } else {
212 // if the object isn't moving we can just set the mark to the default
213 // mark and handle it specially later on.
214 q->init_mark();
215 assert(q->forwardee() == NULL, "should be forwarded to NULL");
216 }
217
218 compact_top += adjusted_size;
219
220 // we need to update the offset table so that the beginnings of objects can be
221 // found during scavenge. Note that we are updating the offset table based on
222 // where the object will be once the compaction phase finishes.
223
224 // Always call cross_threshold(). A contiguous space can only call it when
225 // the compaction_top exceeds the current threshold but not for an
226 // non-contiguous space.
227 cp->threshold =
228 cp->space->cross_threshold(compact_top - adjusted_size, compact_top);
229 return compact_top;
230 }
231
232 // A modified copy of OffsetTableContigSpace::cross_threshold() with _offsets -> _bt
233 // and use of single_block instead of alloc_block. The name here is not really
234 // appropriate - maybe a more general name could be invented for both the
235 // contiguous and noncontiguous spaces.
236
237 HeapWord* CompactibleFreeListSpace::cross_threshold(HeapWord* start, HeapWord* the_end) {
238 _bt.single_block(start, the_end);
239 return end();
240 }
241
242 // Initialize them to NULL.
243 void CompactibleFreeListSpace::initializeIndexedFreeListArray() {
244 for (size_t i = 0; i < IndexSetSize; i++) {
245 // Note that on platforms where objects are double word aligned,
246 // the odd array elements are not used. It is convenient, however,
247 // to map directly from the object size to the array element.
248 _indexedFreeList[i].reset(IndexSetSize);
249 _indexedFreeList[i].set_size(i);
250 assert(_indexedFreeList[i].count() == 0, "reset check failed");
251 assert(_indexedFreeList[i].head() == NULL, "reset check failed");
252 assert(_indexedFreeList[i].tail() == NULL, "reset check failed");
253 assert(_indexedFreeList[i].hint() == IndexSetSize, "reset check failed");
254 }
255 }
256
257 void CompactibleFreeListSpace::resetIndexedFreeListArray() {
258 for (size_t i = 1; i < IndexSetSize; i++) {
259 assert(_indexedFreeList[i].size() == (size_t) i,
260 "Indexed free list sizes are incorrect");
261 _indexedFreeList[i].reset(IndexSetSize);
262 assert(_indexedFreeList[i].count() == 0, "reset check failed");
263 assert(_indexedFreeList[i].head() == NULL, "reset check failed");
264 assert(_indexedFreeList[i].tail() == NULL, "reset check failed");
265 assert(_indexedFreeList[i].hint() == IndexSetSize, "reset check failed");
266 }
267 }
268
269 void CompactibleFreeListSpace::reset(MemRegion mr) {
270 resetIndexedFreeListArray();
271 dictionary()->reset();
272 if (BlockOffsetArrayUseUnallocatedBlock) {
273 assert(end() == mr.end(), "We are compacting to the bottom of CMS gen");
274 // Everything's allocated until proven otherwise.
275 _bt.set_unallocated_block(end());
276 }
277 if (!mr.is_empty()) {
278 assert(mr.word_size() >= MinChunkSize, "Chunk size is too small");
279 _bt.single_block(mr.start(), mr.word_size());
280 FreeChunk* fc = (FreeChunk*) mr.start();
281 fc->set_size(mr.word_size());
282 if (mr.word_size() >= IndexSetSize ) {
283 returnChunkToDictionary(fc);
284 } else {
285 _bt.verify_not_unallocated((HeapWord*)fc, fc->size());
286 _indexedFreeList[mr.word_size()].return_chunk_at_head(fc);
287 }
288 coalBirth(mr.word_size());
289 }
290 _promoInfo.reset();
291 _smallLinearAllocBlock._ptr = NULL;
292 _smallLinearAllocBlock._word_size = 0;
293 }
294
295 void CompactibleFreeListSpace::reset_after_compaction() {
296 // Reset the space to the new reality - one free chunk.
297 MemRegion mr(compaction_top(), end());
298 reset(mr);
299 // Now refill the linear allocation block(s) if possible.
300 if (_adaptive_freelists) {
301 refillLinearAllocBlocksIfNeeded();
302 } else {
303 // Place as much of mr in the linAB as we can get,
304 // provided it was big enough to go into the dictionary.
305 FreeChunk* fc = dictionary()->find_largest_dict();
306 if (fc != NULL) {
307 assert(fc->size() == mr.word_size(),
308 "Why was the chunk broken up?");
309 removeChunkFromDictionary(fc);
310 HeapWord* addr = (HeapWord*) fc;
311 _smallLinearAllocBlock.set(addr, fc->size() ,
312 1024*SmallForLinearAlloc, fc->size());
313 // Note that _unallocated_block is not updated here.
314 }
315 }
316 }
317
318 // Walks the entire dictionary, returning a coterminal
319 // chunk, if it exists. Use with caution since it involves
320 // a potentially complete walk of a potentially large tree.
321 FreeChunk* CompactibleFreeListSpace::find_chunk_at_end() {
322
323 assert_lock_strong(&_freelistLock);
324
325 return dictionary()->find_chunk_ends_at(end());
326 }
327
328
329 #ifndef PRODUCT
330 void CompactibleFreeListSpace::initializeIndexedFreeListArrayReturnedBytes() {
331 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
332 _indexedFreeList[i].allocation_stats()->set_returned_bytes(0);
333 }
334 }
335
336 size_t CompactibleFreeListSpace::sumIndexedFreeListArrayReturnedBytes() {
337 size_t sum = 0;
338 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
339 sum += _indexedFreeList[i].allocation_stats()->returned_bytes();
340 }
341 return sum;
342 }
343
344 size_t CompactibleFreeListSpace::totalCountInIndexedFreeLists() const {
345 size_t count = 0;
346 for (size_t i = IndexSetStart; i < IndexSetSize; i++) {
347 debug_only(
348 ssize_t total_list_count = 0;
349 for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL;
350 fc = fc->next()) {
351 total_list_count++;
352 }
353 assert(total_list_count == _indexedFreeList[i].count(),
354 "Count in list is incorrect");
355 )
356 count += _indexedFreeList[i].count();
357 }
358 return count;
359 }
360
361 size_t CompactibleFreeListSpace::totalCount() {
362 size_t num = totalCountInIndexedFreeLists();
363 num += dictionary()->total_count();
364 if (_smallLinearAllocBlock._word_size != 0) {
365 num++;
366 }
367 return num;
368 }
369 #endif
370
371 bool CompactibleFreeListSpace::is_free_block(const HeapWord* p) const {
372 FreeChunk* fc = (FreeChunk*) p;
373 return fc->is_free();
374 }
375
376 size_t CompactibleFreeListSpace::used() const {
377 return capacity() - free();
378 }
379
380 size_t CompactibleFreeListSpace::free() const {
381 // "MT-safe, but not MT-precise"(TM), if you will: i.e.
382 // if you do this while the structures are in flux you
383 // may get an approximate answer only; for instance
384 // because there is concurrent allocation either
385 // directly by mutators or for promotion during a GC.
386 // It's "MT-safe", however, in the sense that you are guaranteed
387 // not to crash and burn, for instance, because of walking
388 // pointers that could disappear as you were walking them.
389 // The approximation is because the various components
390 // that are read below are not read atomically (and
391 // further the computation of totalSizeInIndexedFreeLists()
392 // is itself a non-atomic computation. The normal use of
393 // this is during a resize operation at the end of GC
394 // and at that time you are guaranteed to get the
395 // correct actual value. However, for instance, this is
396 // also read completely asynchronously by the "perf-sampler"
397 // that supports jvmstat, and you are apt to see the values
398 // flicker in such cases.
399 assert(_dictionary != NULL, "No _dictionary?");
400 return (_dictionary->total_chunk_size(DEBUG_ONLY(freelistLock())) +
401 totalSizeInIndexedFreeLists() +
402 _smallLinearAllocBlock._word_size) * HeapWordSize;
403 }
404
405 size_t CompactibleFreeListSpace::max_alloc_in_words() const {
406 assert(_dictionary != NULL, "No _dictionary?");
407 assert_locked();
408 size_t res = _dictionary->max_chunk_size();
409 res = MAX2(res, MIN2(_smallLinearAllocBlock._word_size,
410 (size_t) SmallForLinearAlloc - 1));
411 // XXX the following could potentially be pretty slow;
412 // should one, pessimistically for the rare cases when res
413 // calculated above is less than IndexSetSize,
414 // just return res calculated above? My reasoning was that
415 // those cases will be so rare that the extra time spent doesn't
416 // really matter....
417 // Note: do not change the loop test i >= res + IndexSetStride
418 // to i > res below, because i is unsigned and res may be zero.
419 for (size_t i = IndexSetSize - 1; i >= res + IndexSetStride;
420 i -= IndexSetStride) {
421 if (_indexedFreeList[i].head() != NULL) {
422 assert(_indexedFreeList[i].count() != 0, "Inconsistent FreeList");
423 return i;
424 }
425 }
426 return res;
427 }
428
429 void LinearAllocBlock::print_on(outputStream* st) const {
430 st->print_cr(" LinearAllocBlock: ptr = " PTR_FORMAT ", word_size = " SIZE_FORMAT
431 ", refillsize = " SIZE_FORMAT ", allocation_size_limit = " SIZE_FORMAT,
432 p2i(_ptr), _word_size, _refillSize, _allocation_size_limit);
433 }
434
435 void CompactibleFreeListSpace::print_on(outputStream* st) const {
436 st->print_cr("COMPACTIBLE FREELIST SPACE");
437 st->print_cr(" Space:");
438 Space::print_on(st);
439
440 st->print_cr("promoInfo:");
441 _promoInfo.print_on(st);
442
443 st->print_cr("_smallLinearAllocBlock");
444 _smallLinearAllocBlock.print_on(st);
445
446 // dump_memory_block(_smallLinearAllocBlock->_ptr, 128);
447
448 st->print_cr(" _fitStrategy = %s, _adaptive_freelists = %s",
449 _fitStrategy?"true":"false", _adaptive_freelists?"true":"false");
450 }
451
452 void CompactibleFreeListSpace::print_indexed_free_lists(outputStream* st)
453 const {
454 reportIndexedFreeListStatistics();
455 gclog_or_tty->print_cr("Layout of Indexed Freelists");
456 gclog_or_tty->print_cr("---------------------------");
457 AdaptiveFreeList<FreeChunk>::print_labels_on(st, "size");
458 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
459 _indexedFreeList[i].print_on(gclog_or_tty);
460 for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL;
461 fc = fc->next()) {
462 gclog_or_tty->print_cr("\t[" PTR_FORMAT "," PTR_FORMAT ") %s",
463 p2i(fc), p2i((HeapWord*)fc + i),
464 fc->cantCoalesce() ? "\t CC" : "");
465 }
466 }
467 }
468
469 void CompactibleFreeListSpace::print_promo_info_blocks(outputStream* st)
470 const {
471 _promoInfo.print_on(st);
472 }
473
474 void CompactibleFreeListSpace::print_dictionary_free_lists(outputStream* st)
475 const {
476 _dictionary->report_statistics();
477 st->print_cr("Layout of Freelists in Tree");
478 st->print_cr("---------------------------");
479 _dictionary->print_free_lists(st);
480 }
481
482 class BlkPrintingClosure: public BlkClosure {
483 const CMSCollector* _collector;
484 const CompactibleFreeListSpace* _sp;
485 const CMSBitMap* _live_bit_map;
486 const bool _post_remark;
487 outputStream* _st;
488 public:
489 BlkPrintingClosure(const CMSCollector* collector,
490 const CompactibleFreeListSpace* sp,
491 const CMSBitMap* live_bit_map,
492 outputStream* st):
493 _collector(collector),
494 _sp(sp),
495 _live_bit_map(live_bit_map),
496 _post_remark(collector->abstract_state() > CMSCollector::FinalMarking),
497 _st(st) { }
498 size_t do_blk(HeapWord* addr);
499 };
500
501 size_t BlkPrintingClosure::do_blk(HeapWord* addr) {
502 size_t sz = _sp->block_size_no_stall(addr, _collector);
503 assert(sz != 0, "Should always be able to compute a size");
504 if (_sp->block_is_obj(addr)) {
505 const bool dead = _post_remark && !_live_bit_map->isMarked(addr);
506 _st->print_cr(PTR_FORMAT ": %s object of size " SIZE_FORMAT "%s",
507 p2i(addr),
508 dead ? "dead" : "live",
509 sz,
510 (!dead && CMSPrintObjectsInDump) ? ":" : ".");
511 if (CMSPrintObjectsInDump && !dead) {
512 oop(addr)->print_on(_st);
513 _st->print_cr("--------------------------------------");
514 }
515 } else { // free block
516 _st->print_cr(PTR_FORMAT ": free block of size " SIZE_FORMAT "%s",
517 p2i(addr), sz, CMSPrintChunksInDump ? ":" : ".");
518 if (CMSPrintChunksInDump) {
519 ((FreeChunk*)addr)->print_on(_st);
520 _st->print_cr("--------------------------------------");
521 }
522 }
523 return sz;
524 }
525
526 void CompactibleFreeListSpace::dump_at_safepoint_with_locks(CMSCollector* c,
527 outputStream* st) {
528 st->print_cr("\n=========================");
529 st->print_cr("Block layout in CMS Heap:");
530 st->print_cr("=========================");
531 BlkPrintingClosure bpcl(c, this, c->markBitMap(), st);
532 blk_iterate(&bpcl);
533
534 st->print_cr("\n=======================================");
535 st->print_cr("Order & Layout of Promotion Info Blocks");
536 st->print_cr("=======================================");
537 print_promo_info_blocks(st);
538
539 st->print_cr("\n===========================");
540 st->print_cr("Order of Indexed Free Lists");
541 st->print_cr("=========================");
542 print_indexed_free_lists(st);
543
544 st->print_cr("\n=================================");
545 st->print_cr("Order of Free Lists in Dictionary");
546 st->print_cr("=================================");
547 print_dictionary_free_lists(st);
548 }
549
550
551 void CompactibleFreeListSpace::reportFreeListStatistics() const {
552 assert_lock_strong(&_freelistLock);
553 assert(PrintFLSStatistics != 0, "Reporting error");
554 _dictionary->report_statistics();
555 if (PrintFLSStatistics > 1) {
556 reportIndexedFreeListStatistics();
557 size_t total_size = totalSizeInIndexedFreeLists() +
558 _dictionary->total_chunk_size(DEBUG_ONLY(freelistLock()));
559 gclog_or_tty->print(" free=" SIZE_FORMAT " frag=%1.4f\n", total_size, flsFrag());
560 }
561 }
562
563 void CompactibleFreeListSpace::reportIndexedFreeListStatistics() const {
564 assert_lock_strong(&_freelistLock);
565 gclog_or_tty->print("Statistics for IndexedFreeLists:\n"
566 "--------------------------------\n");
567 size_t total_size = totalSizeInIndexedFreeLists();
568 size_t free_blocks = numFreeBlocksInIndexedFreeLists();
569 gclog_or_tty->print("Total Free Space: " SIZE_FORMAT "\n", total_size);
570 gclog_or_tty->print("Max Chunk Size: " SIZE_FORMAT "\n", maxChunkSizeInIndexedFreeLists());
571 gclog_or_tty->print("Number of Blocks: " SIZE_FORMAT "\n", free_blocks);
572 if (free_blocks != 0) {
573 gclog_or_tty->print("Av. Block Size: " SIZE_FORMAT "\n", total_size/free_blocks);
574 }
575 }
576
577 size_t CompactibleFreeListSpace::numFreeBlocksInIndexedFreeLists() const {
578 size_t res = 0;
579 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
580 debug_only(
581 ssize_t recount = 0;
582 for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL;
583 fc = fc->next()) {
584 recount += 1;
585 }
586 assert(recount == _indexedFreeList[i].count(),
587 "Incorrect count in list");
588 )
589 res += _indexedFreeList[i].count();
590 }
591 return res;
592 }
593
594 size_t CompactibleFreeListSpace::maxChunkSizeInIndexedFreeLists() const {
595 for (size_t i = IndexSetSize - 1; i != 0; i -= IndexSetStride) {
596 if (_indexedFreeList[i].head() != NULL) {
597 assert(_indexedFreeList[i].count() != 0, "Inconsistent FreeList");
598 return (size_t)i;
599 }
600 }
601 return 0;
602 }
603
604 void CompactibleFreeListSpace::set_end(HeapWord* value) {
605 HeapWord* prevEnd = end();
606 assert(prevEnd != value, "unnecessary set_end call");
607 assert(prevEnd == NULL || !BlockOffsetArrayUseUnallocatedBlock || value >= unallocated_block(),
608 "New end is below unallocated block");
609 _end = value;
610 if (prevEnd != NULL) {
611 // Resize the underlying block offset table.
612 _bt.resize(pointer_delta(value, bottom()));
613 if (value <= prevEnd) {
614 assert(!BlockOffsetArrayUseUnallocatedBlock || value >= unallocated_block(),
615 "New end is below unallocated block");
616 } else {
617 // Now, take this new chunk and add it to the free blocks.
618 // Note that the BOT has not yet been updated for this block.
619 size_t newFcSize = pointer_delta(value, prevEnd);
620 // XXX This is REALLY UGLY and should be fixed up. XXX
621 if (!_adaptive_freelists && _smallLinearAllocBlock._ptr == NULL) {
622 // Mark the boundary of the new block in BOT
623 _bt.mark_block(prevEnd, value);
624 // put it all in the linAB
625 MutexLockerEx x(parDictionaryAllocLock(),
626 Mutex::_no_safepoint_check_flag);
627 _smallLinearAllocBlock._ptr = prevEnd;
628 _smallLinearAllocBlock._word_size = newFcSize;
629 repairLinearAllocBlock(&_smallLinearAllocBlock);
630 // Births of chunks put into a LinAB are not recorded. Births
631 // of chunks as they are allocated out of a LinAB are.
632 } else {
633 // Add the block to the free lists, if possible coalescing it
634 // with the last free block, and update the BOT and census data.
635 addChunkToFreeListsAtEndRecordingStats(prevEnd, newFcSize);
636 }
637 }
638 }
639 }
640
641 class FreeListSpace_DCTOC : public Filtering_DCTOC {
642 CompactibleFreeListSpace* _cfls;
643 CMSCollector* _collector;
644 protected:
645 // Override.
646 #define walk_mem_region_with_cl_DECL(ClosureType) \
647 virtual void walk_mem_region_with_cl(MemRegion mr, \
648 HeapWord* bottom, HeapWord* top, \
649 ClosureType* cl); \
650 void walk_mem_region_with_cl_par(MemRegion mr, \
651 HeapWord* bottom, HeapWord* top, \
652 ClosureType* cl); \
653 void walk_mem_region_with_cl_nopar(MemRegion mr, \
654 HeapWord* bottom, HeapWord* top, \
655 ClosureType* cl)
656 walk_mem_region_with_cl_DECL(ExtendedOopClosure);
657 walk_mem_region_with_cl_DECL(FilteringClosure);
658
659 public:
660 FreeListSpace_DCTOC(CompactibleFreeListSpace* sp,
661 CMSCollector* collector,
662 ExtendedOopClosure* cl,
663 CardTableModRefBS::PrecisionStyle precision,
664 HeapWord* boundary) :
665 Filtering_DCTOC(sp, cl, precision, boundary),
666 _cfls(sp), _collector(collector) {}
667 };
668
669 // We de-virtualize the block-related calls below, since we know that our
670 // space is a CompactibleFreeListSpace.
671
672 #define FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(ClosureType) \
673 void FreeListSpace_DCTOC::walk_mem_region_with_cl(MemRegion mr, \
674 HeapWord* bottom, \
675 HeapWord* top, \
676 ClosureType* cl) { \
677 bool is_par = GenCollectedHeap::heap()->n_par_threads() > 0; \
678 if (is_par) { \
679 assert(GenCollectedHeap::heap()->n_par_threads() == \
680 GenCollectedHeap::heap()->workers()->active_workers(), "Mismatch"); \
681 walk_mem_region_with_cl_par(mr, bottom, top, cl); \
682 } else { \
683 walk_mem_region_with_cl_nopar(mr, bottom, top, cl); \
684 } \
685 } \
686 void FreeListSpace_DCTOC::walk_mem_region_with_cl_par(MemRegion mr, \
687 HeapWord* bottom, \
688 HeapWord* top, \
689 ClosureType* cl) { \
690 /* Skip parts that are before "mr", in case "block_start" sent us \
691 back too far. */ \
692 HeapWord* mr_start = mr.start(); \
693 size_t bot_size = _cfls->CompactibleFreeListSpace::block_size(bottom); \
694 HeapWord* next = bottom + bot_size; \
695 while (next < mr_start) { \
696 bottom = next; \
697 bot_size = _cfls->CompactibleFreeListSpace::block_size(bottom); \
698 next = bottom + bot_size; \
699 } \
700 \
701 while (bottom < top) { \
702 if (_cfls->CompactibleFreeListSpace::block_is_obj(bottom) && \
703 !_cfls->CompactibleFreeListSpace::obj_allocated_since_save_marks( \
704 oop(bottom)) && \
705 !_collector->CMSCollector::is_dead_obj(oop(bottom))) { \
706 size_t word_sz = oop(bottom)->oop_iterate(cl, mr); \
707 bottom += _cfls->adjustObjectSize(word_sz); \
708 } else { \
709 bottom += _cfls->CompactibleFreeListSpace::block_size(bottom); \
710 } \
711 } \
712 } \
713 void FreeListSpace_DCTOC::walk_mem_region_with_cl_nopar(MemRegion mr, \
714 HeapWord* bottom, \
715 HeapWord* top, \
716 ClosureType* cl) { \
717 /* Skip parts that are before "mr", in case "block_start" sent us \
718 back too far. */ \
719 HeapWord* mr_start = mr.start(); \
720 size_t bot_size = _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \
721 HeapWord* next = bottom + bot_size; \
722 while (next < mr_start) { \
723 bottom = next; \
724 bot_size = _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \
725 next = bottom + bot_size; \
726 } \
727 \
728 while (bottom < top) { \
729 if (_cfls->CompactibleFreeListSpace::block_is_obj_nopar(bottom) && \
730 !_cfls->CompactibleFreeListSpace::obj_allocated_since_save_marks( \
731 oop(bottom)) && \
732 !_collector->CMSCollector::is_dead_obj(oop(bottom))) { \
733 size_t word_sz = oop(bottom)->oop_iterate(cl, mr); \
734 bottom += _cfls->adjustObjectSize(word_sz); \
735 } else { \
736 bottom += _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \
737 } \
738 } \
739 }
740
741 // (There are only two of these, rather than N, because the split is due
742 // only to the introduction of the FilteringClosure, a local part of the
743 // impl of this abstraction.)
744 FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(ExtendedOopClosure)
745 FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(FilteringClosure)
746
747 DirtyCardToOopClosure*
748 CompactibleFreeListSpace::new_dcto_cl(ExtendedOopClosure* cl,
749 CardTableModRefBS::PrecisionStyle precision,
750 HeapWord* boundary) {
751 return new FreeListSpace_DCTOC(this, _collector, cl, precision, boundary);
752 }
753
754
755 // Note on locking for the space iteration functions:
756 // since the collector's iteration activities are concurrent with
757 // allocation activities by mutators, absent a suitable mutual exclusion
758 // mechanism the iterators may go awry. For instance a block being iterated
759 // may suddenly be allocated or divided up and part of it allocated and
760 // so on.
761
762 // Apply the given closure to each block in the space.
763 void CompactibleFreeListSpace::blk_iterate_careful(BlkClosureCareful* cl) {
764 assert_lock_strong(freelistLock());
765 HeapWord *cur, *limit;
766 for (cur = bottom(), limit = end(); cur < limit;
767 cur += cl->do_blk_careful(cur));
768 }
769
770 // Apply the given closure to each block in the space.
771 void CompactibleFreeListSpace::blk_iterate(BlkClosure* cl) {
772 assert_lock_strong(freelistLock());
773 HeapWord *cur, *limit;
774 for (cur = bottom(), limit = end(); cur < limit;
775 cur += cl->do_blk(cur));
776 }
777
778 // Apply the given closure to each oop in the space.
779 void CompactibleFreeListSpace::oop_iterate(ExtendedOopClosure* cl) {
780 assert_lock_strong(freelistLock());
781 HeapWord *cur, *limit;
782 size_t curSize;
783 for (cur = bottom(), limit = end(); cur < limit;
784 cur += curSize) {
785 curSize = block_size(cur);
786 if (block_is_obj(cur)) {
787 oop(cur)->oop_iterate(cl);
788 }
789 }
790 }
791
792 // NOTE: In the following methods, in order to safely be able to
793 // apply the closure to an object, we need to be sure that the
794 // object has been initialized. We are guaranteed that an object
795 // is initialized if we are holding the Heap_lock with the
796 // world stopped.
797 void CompactibleFreeListSpace::verify_objects_initialized() const {
798 if (is_init_completed()) {
799 assert_locked_or_safepoint(Heap_lock);
800 if (Universe::is_fully_initialized()) {
801 guarantee(SafepointSynchronize::is_at_safepoint(),
802 "Required for objects to be initialized");
803 }
804 } // else make a concession at vm start-up
805 }
806
807 // Apply the given closure to each object in the space
808 void CompactibleFreeListSpace::object_iterate(ObjectClosure* blk) {
809 assert_lock_strong(freelistLock());
810 NOT_PRODUCT(verify_objects_initialized());
811 HeapWord *cur, *limit;
812 size_t curSize;
813 for (cur = bottom(), limit = end(); cur < limit;
814 cur += curSize) {
815 curSize = block_size(cur);
816 if (block_is_obj(cur)) {
817 blk->do_object(oop(cur));
818 }
819 }
820 }
821
822 // Apply the given closure to each live object in the space
823 // The usage of CompactibleFreeListSpace
824 // by the ConcurrentMarkSweepGeneration for concurrent GC's allows
825 // objects in the space with references to objects that are no longer
826 // valid. For example, an object may reference another object
827 // that has already been sweep up (collected). This method uses
828 // obj_is_alive() to determine whether it is safe to apply the closure to
829 // an object. See obj_is_alive() for details on how liveness of an
830 // object is decided.
831
832 void CompactibleFreeListSpace::safe_object_iterate(ObjectClosure* blk) {
833 assert_lock_strong(freelistLock());
834 NOT_PRODUCT(verify_objects_initialized());
835 HeapWord *cur, *limit;
836 size_t curSize;
837 for (cur = bottom(), limit = end(); cur < limit;
838 cur += curSize) {
839 curSize = block_size(cur);
840 if (block_is_obj(cur) && obj_is_alive(cur)) {
841 blk->do_object(oop(cur));
842 }
843 }
844 }
845
846 void CompactibleFreeListSpace::object_iterate_mem(MemRegion mr,
847 UpwardsObjectClosure* cl) {
848 assert_locked(freelistLock());
849 NOT_PRODUCT(verify_objects_initialized());
850 assert(!mr.is_empty(), "Should be non-empty");
851 // We use MemRegion(bottom(), end()) rather than used_region() below
852 // because the two are not necessarily equal for some kinds of
853 // spaces, in particular, certain kinds of free list spaces.
854 // We could use the more complicated but more precise:
855 // MemRegion(used_region().start(), round_to(used_region().end(), CardSize))
856 // but the slight imprecision seems acceptable in the assertion check.
857 assert(MemRegion(bottom(), end()).contains(mr),
858 "Should be within used space");
859 HeapWord* prev = cl->previous(); // max address from last time
860 if (prev >= mr.end()) { // nothing to do
861 return;
862 }
863 // This assert will not work when we go from cms space to perm
864 // space, and use same closure. Easy fix deferred for later. XXX YSR
865 // assert(prev == NULL || contains(prev), "Should be within space");
866
867 bool last_was_obj_array = false;
868 HeapWord *blk_start_addr, *region_start_addr;
869 if (prev > mr.start()) {
870 region_start_addr = prev;
871 blk_start_addr = prev;
872 // The previous invocation may have pushed "prev" beyond the
873 // last allocated block yet there may be still be blocks
874 // in this region due to a particular coalescing policy.
875 // Relax the assertion so that the case where the unallocated
876 // block is maintained and "prev" is beyond the unallocated
877 // block does not cause the assertion to fire.
878 assert((BlockOffsetArrayUseUnallocatedBlock &&
879 (!is_in(prev))) ||
880 (blk_start_addr == block_start(region_start_addr)), "invariant");
881 } else {
882 region_start_addr = mr.start();
883 blk_start_addr = block_start(region_start_addr);
884 }
885 HeapWord* region_end_addr = mr.end();
886 MemRegion derived_mr(region_start_addr, region_end_addr);
887 while (blk_start_addr < region_end_addr) {
888 const size_t size = block_size(blk_start_addr);
889 if (block_is_obj(blk_start_addr)) {
890 last_was_obj_array = cl->do_object_bm(oop(blk_start_addr), derived_mr);
891 } else {
892 last_was_obj_array = false;
893 }
894 blk_start_addr += size;
895 }
896 if (!last_was_obj_array) {
897 assert((bottom() <= blk_start_addr) && (blk_start_addr <= end()),
898 "Should be within (closed) used space");
899 assert(blk_start_addr > prev, "Invariant");
900 cl->set_previous(blk_start_addr); // min address for next time
901 }
902 }
903
904 // Callers of this iterator beware: The closure application should
905 // be robust in the face of uninitialized objects and should (always)
906 // return a correct size so that the next addr + size below gives us a
907 // valid block boundary. [See for instance,
908 // ScanMarkedObjectsAgainCarefullyClosure::do_object_careful()
909 // in ConcurrentMarkSweepGeneration.cpp.]
910 HeapWord*
911 CompactibleFreeListSpace::object_iterate_careful_m(MemRegion mr,
912 ObjectClosureCareful* cl) {
913 assert_lock_strong(freelistLock());
914 // Can't use used_region() below because it may not necessarily
915 // be the same as [bottom(),end()); although we could
916 // use [used_region().start(),round_to(used_region().end(),CardSize)),
917 // that appears too cumbersome, so we just do the simpler check
918 // in the assertion below.
919 assert(!mr.is_empty() && MemRegion(bottom(),end()).contains(mr),
920 "mr should be non-empty and within used space");
921 HeapWord *addr, *end;
922 size_t size;
923 for (addr = block_start_careful(mr.start()), end = mr.end();
924 addr < end; addr += size) {
925 FreeChunk* fc = (FreeChunk*)addr;
926 if (fc->is_free()) {
927 // Since we hold the free list lock, which protects direct
928 // allocation in this generation by mutators, a free object
929 // will remain free throughout this iteration code.
930 size = fc->size();
931 } else {
932 // Note that the object need not necessarily be initialized,
933 // because (for instance) the free list lock does NOT protect
934 // object initialization. The closure application below must
935 // therefore be correct in the face of uninitialized objects.
936 size = cl->do_object_careful_m(oop(addr), mr);
937 if (size == 0) {
938 // An unparsable object found. Signal early termination.
939 return addr;
940 }
941 }
942 }
943 return NULL;
944 }
945
946
947 HeapWord* CompactibleFreeListSpace::block_start_const(const void* p) const {
948 NOT_PRODUCT(verify_objects_initialized());
949 return _bt.block_start(p);
950 }
951
952 HeapWord* CompactibleFreeListSpace::block_start_careful(const void* p) const {
953 return _bt.block_start_careful(p);
954 }
955
956 size_t CompactibleFreeListSpace::block_size(const HeapWord* p) const {
957 NOT_PRODUCT(verify_objects_initialized());
958 // This must be volatile, or else there is a danger that the compiler
959 // will compile the code below into a sometimes-infinite loop, by keeping
960 // the value read the first time in a register.
961 while (true) {
962 // We must do this until we get a consistent view of the object.
963 if (FreeChunk::indicatesFreeChunk(p)) {
964 volatile FreeChunk* fc = (volatile FreeChunk*)p;
965 size_t res = fc->size();
966
967 // Bugfix for systems with weak memory model (PPC64/IA64). The
968 // block's free bit was set and we have read the size of the
969 // block. Acquire and check the free bit again. If the block is
970 // still free, the read size is correct.
971 OrderAccess::acquire();
972
973 // If the object is still a free chunk, return the size, else it
974 // has been allocated so try again.
975 if (FreeChunk::indicatesFreeChunk(p)) {
976 assert(res != 0, "Block size should not be 0");
977 return res;
978 }
979 } else {
980 // must read from what 'p' points to in each loop.
981 Klass* k = ((volatile oopDesc*)p)->klass_or_null();
982 if (k != NULL) {
983 assert(k->is_klass(), "Should really be klass oop.");
984 oop o = (oop)p;
985 assert(o->is_oop(true /* ignore mark word */), "Should be an oop.");
986
987 // Bugfix for systems with weak memory model (PPC64/IA64).
988 // The object o may be an array. Acquire to make sure that the array
989 // size (third word) is consistent.
990 OrderAccess::acquire();
991
992 size_t res = o->size_given_klass(k);
993 res = adjustObjectSize(res);
994 assert(res != 0, "Block size should not be 0");
995 return res;
996 }
997 }
998 }
999 }
1000
1001 // TODO: Now that is_parsable is gone, we should combine these two functions.
1002 // A variant of the above that uses the Printezis bits for
1003 // unparsable but allocated objects. This avoids any possible
1004 // stalls waiting for mutators to initialize objects, and is
1005 // thus potentially faster than the variant above. However,
1006 // this variant may return a zero size for a block that is
1007 // under mutation and for which a consistent size cannot be
1008 // inferred without stalling; see CMSCollector::block_size_if_printezis_bits().
1009 size_t CompactibleFreeListSpace::block_size_no_stall(HeapWord* p,
1010 const CMSCollector* c)
1011 const {
1012 assert(MemRegion(bottom(), end()).contains(p), "p not in space");
1013 // This must be volatile, or else there is a danger that the compiler
1014 // will compile the code below into a sometimes-infinite loop, by keeping
1015 // the value read the first time in a register.
1016 DEBUG_ONLY(uint loops = 0;)
1017 while (true) {
1018 // We must do this until we get a consistent view of the object.
1019 if (FreeChunk::indicatesFreeChunk(p)) {
1020 volatile FreeChunk* fc = (volatile FreeChunk*)p;
1021 size_t res = fc->size();
1022
1023 // Bugfix for systems with weak memory model (PPC64/IA64). The
1024 // free bit of the block was set and we have read the size of
1025 // the block. Acquire and check the free bit again. If the
1026 // block is still free, the read size is correct.
1027 OrderAccess::acquire();
1028
1029 if (FreeChunk::indicatesFreeChunk(p)) {
1030 assert(res != 0, "Block size should not be 0");
1031 assert(loops == 0, "Should be 0");
1032 return res;
1033 }
1034 } else {
1035 // must read from what 'p' points to in each loop.
1036 Klass* k = ((volatile oopDesc*)p)->klass_or_null();
1037 // We trust the size of any object that has a non-NULL
1038 // klass and (for those in the perm gen) is parsable
1039 // -- irrespective of its conc_safe-ty.
1040 if (k != NULL) {
1041 assert(k->is_klass(), "Should really be klass oop.");
1042 oop o = (oop)p;
1043 assert(o->is_oop(), "Should be an oop");
1044
1045 // Bugfix for systems with weak memory model (PPC64/IA64).
1046 // The object o may be an array. Acquire to make sure that the array
1047 // size (third word) is consistent.
1048 OrderAccess::acquire();
1049
1050 size_t res = o->size_given_klass(k);
1051 res = adjustObjectSize(res);
1052 assert(res != 0, "Block size should not be 0");
1053 return res;
1054 } else {
1055 // May return 0 if P-bits not present.
1056 return c->block_size_if_printezis_bits(p);
1057 }
1058 }
1059 assert(loops == 0, "Can loop at most once");
1060 DEBUG_ONLY(loops++;)
1061 }
1062 }
1063
1064 size_t CompactibleFreeListSpace::block_size_nopar(const HeapWord* p) const {
1065 NOT_PRODUCT(verify_objects_initialized());
1066 assert(MemRegion(bottom(), end()).contains(p), "p not in space");
1067 FreeChunk* fc = (FreeChunk*)p;
1068 if (fc->is_free()) {
1069 return fc->size();
1070 } else {
1071 // Ignore mark word because this may be a recently promoted
1072 // object whose mark word is used to chain together grey
1073 // objects (the last one would have a null value).
1074 assert(oop(p)->is_oop(true), "Should be an oop");
1075 return adjustObjectSize(oop(p)->size());
1076 }
1077 }
1078
1079 // This implementation assumes that the property of "being an object" is
1080 // stable. But being a free chunk may not be (because of parallel
1081 // promotion.)
1082 bool CompactibleFreeListSpace::block_is_obj(const HeapWord* p) const {
1083 FreeChunk* fc = (FreeChunk*)p;
1084 assert(is_in_reserved(p), "Should be in space");
1085 if (FreeChunk::indicatesFreeChunk(p)) return false;
1086 Klass* k = oop(p)->klass_or_null();
1087 if (k != NULL) {
1088 // Ignore mark word because it may have been used to
1089 // chain together promoted objects (the last one
1090 // would have a null value).
1091 assert(oop(p)->is_oop(true), "Should be an oop");
1092 return true;
1093 } else {
1094 return false; // Was not an object at the start of collection.
1095 }
1096 }
1097
1098 // Check if the object is alive. This fact is checked either by consulting
1099 // the main marking bitmap in the sweeping phase or, if it's a permanent
1100 // generation and we're not in the sweeping phase, by checking the
1101 // perm_gen_verify_bit_map where we store the "deadness" information if
1102 // we did not sweep the perm gen in the most recent previous GC cycle.
1103 bool CompactibleFreeListSpace::obj_is_alive(const HeapWord* p) const {
1104 assert(SafepointSynchronize::is_at_safepoint() || !is_init_completed(),
1105 "Else races are possible");
1106 assert(block_is_obj(p), "The address should point to an object");
1107
1108 // If we're sweeping, we use object liveness information from the main bit map
1109 // for both perm gen and old gen.
1110 // We don't need to lock the bitmap (live_map or dead_map below), because
1111 // EITHER we are in the middle of the sweeping phase, and the
1112 // main marking bit map (live_map below) is locked,
1113 // OR we're in other phases and perm_gen_verify_bit_map (dead_map below)
1114 // is stable, because it's mutated only in the sweeping phase.
1115 // NOTE: This method is also used by jmap where, if class unloading is
1116 // off, the results can return "false" for legitimate perm objects,
1117 // when we are not in the midst of a sweeping phase, which can result
1118 // in jmap not reporting certain perm gen objects. This will be moot
1119 // if/when the perm gen goes away in the future.
1120 if (_collector->abstract_state() == CMSCollector::Sweeping) {
1121 CMSBitMap* live_map = _collector->markBitMap();
1122 return live_map->par_isMarked((HeapWord*) p);
1123 }
1124 return true;
1125 }
1126
1127 bool CompactibleFreeListSpace::block_is_obj_nopar(const HeapWord* p) const {
1128 FreeChunk* fc = (FreeChunk*)p;
1129 assert(is_in_reserved(p), "Should be in space");
1130 assert(_bt.block_start(p) == p, "Should be a block boundary");
1131 if (!fc->is_free()) {
1132 // Ignore mark word because it may have been used to
1133 // chain together promoted objects (the last one
1134 // would have a null value).
1135 assert(oop(p)->is_oop(true), "Should be an oop");
1136 return true;
1137 }
1138 return false;
1139 }
1140
1141 // "MT-safe but not guaranteed MT-precise" (TM); you may get an
1142 // approximate answer if you don't hold the freelistlock when you call this.
1143 size_t CompactibleFreeListSpace::totalSizeInIndexedFreeLists() const {
1144 size_t size = 0;
1145 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
1146 debug_only(
1147 // We may be calling here without the lock in which case we
1148 // won't do this modest sanity check.
1149 if (freelistLock()->owned_by_self()) {
1150 size_t total_list_size = 0;
1151 for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL;
1152 fc = fc->next()) {
1153 total_list_size += i;
1154 }
1155 assert(total_list_size == i * _indexedFreeList[i].count(),
1156 "Count in list is incorrect");
1157 }
1158 )
1159 size += i * _indexedFreeList[i].count();
1160 }
1161 return size;
1162 }
1163
1164 HeapWord* CompactibleFreeListSpace::par_allocate(size_t size) {
1165 MutexLockerEx x(freelistLock(), Mutex::_no_safepoint_check_flag);
1166 return allocate(size);
1167 }
1168
1169 HeapWord*
1170 CompactibleFreeListSpace::getChunkFromSmallLinearAllocBlockRemainder(size_t size) {
1171 return getChunkFromLinearAllocBlockRemainder(&_smallLinearAllocBlock, size);
1172 }
1173
1174 HeapWord* CompactibleFreeListSpace::allocate(size_t size) {
1175 assert_lock_strong(freelistLock());
1176 HeapWord* res = NULL;
1177 assert(size == adjustObjectSize(size),
1178 "use adjustObjectSize() before calling into allocate()");
1179
1180 if (_adaptive_freelists) {
1181 res = allocate_adaptive_freelists(size);
1182 } else { // non-adaptive free lists
1183 res = allocate_non_adaptive_freelists(size);
1184 }
1185
1186 if (res != NULL) {
1187 // check that res does lie in this space!
1188 assert(is_in_reserved(res), "Not in this space!");
1189 assert(is_aligned((void*)res), "alignment check");
1190
1191 FreeChunk* fc = (FreeChunk*)res;
1192 fc->markNotFree();
1193 assert(!fc->is_free(), "shouldn't be marked free");
1194 assert(oop(fc)->klass_or_null() == NULL, "should look uninitialized");
1195 // Verify that the block offset table shows this to
1196 // be a single block, but not one which is unallocated.
1197 _bt.verify_single_block(res, size);
1198 _bt.verify_not_unallocated(res, size);
1199 // mangle a just allocated object with a distinct pattern.
1200 debug_only(fc->mangleAllocated(size));
1201 }
1202
1203 return res;
1204 }
1205
1206 HeapWord* CompactibleFreeListSpace::allocate_non_adaptive_freelists(size_t size) {
1207 HeapWord* res = NULL;
1208 // try and use linear allocation for smaller blocks
1209 if (size < _smallLinearAllocBlock._allocation_size_limit) {
1210 // if successful, the following also adjusts block offset table
1211 res = getChunkFromSmallLinearAllocBlock(size);
1212 }
1213 // Else triage to indexed lists for smaller sizes
1214 if (res == NULL) {
1215 if (size < SmallForDictionary) {
1216 res = (HeapWord*) getChunkFromIndexedFreeList(size);
1217 } else {
1218 // else get it from the big dictionary; if even this doesn't
1219 // work we are out of luck.
1220 res = (HeapWord*)getChunkFromDictionaryExact(size);
1221 }
1222 }
1223
1224 return res;
1225 }
1226
1227 HeapWord* CompactibleFreeListSpace::allocate_adaptive_freelists(size_t size) {
1228 assert_lock_strong(freelistLock());
1229 HeapWord* res = NULL;
1230 assert(size == adjustObjectSize(size),
1231 "use adjustObjectSize() before calling into allocate()");
1232
1233 // Strategy
1234 // if small
1235 // exact size from small object indexed list if small
1236 // small or large linear allocation block (linAB) as appropriate
1237 // take from lists of greater sized chunks
1238 // else
1239 // dictionary
1240 // small or large linear allocation block if it has the space
1241 // Try allocating exact size from indexTable first
1242 if (size < IndexSetSize) {
1243 res = (HeapWord*) getChunkFromIndexedFreeList(size);
1244 if(res != NULL) {
1245 assert(res != (HeapWord*)_indexedFreeList[size].head(),
1246 "Not removed from free list");
1247 // no block offset table adjustment is necessary on blocks in
1248 // the indexed lists.
1249
1250 // Try allocating from the small LinAB
1251 } else if (size < _smallLinearAllocBlock._allocation_size_limit &&
1252 (res = getChunkFromSmallLinearAllocBlock(size)) != NULL) {
1253 // if successful, the above also adjusts block offset table
1254 // Note that this call will refill the LinAB to
1255 // satisfy the request. This is different that
1256 // evm.
1257 // Don't record chunk off a LinAB? smallSplitBirth(size);
1258 } else {
1259 // Raid the exact free lists larger than size, even if they are not
1260 // overpopulated.
1261 res = (HeapWord*) getChunkFromGreater(size);
1262 }
1263 } else {
1264 // Big objects get allocated directly from the dictionary.
1265 res = (HeapWord*) getChunkFromDictionaryExact(size);
1266 if (res == NULL) {
1267 // Try hard not to fail since an allocation failure will likely
1268 // trigger a synchronous GC. Try to get the space from the
1269 // allocation blocks.
1270 res = getChunkFromSmallLinearAllocBlockRemainder(size);
1271 }
1272 }
1273
1274 return res;
1275 }
1276
1277 // A worst-case estimate of the space required (in HeapWords) to expand the heap
1278 // when promoting obj.
1279 size_t CompactibleFreeListSpace::expansionSpaceRequired(size_t obj_size) const {
1280 // Depending on the object size, expansion may require refilling either a
1281 // bigLAB or a smallLAB plus refilling a PromotionInfo object. MinChunkSize
1282 // is added because the dictionary may over-allocate to avoid fragmentation.
1283 size_t space = obj_size;
1284 if (!_adaptive_freelists) {
1285 space = MAX2(space, _smallLinearAllocBlock._refillSize);
1286 }
1287 space += _promoInfo.refillSize() + 2 * MinChunkSize;
1288 return space;
1289 }
1290
1291 FreeChunk* CompactibleFreeListSpace::getChunkFromGreater(size_t numWords) {
1292 FreeChunk* ret;
1293
1294 assert(numWords >= MinChunkSize, "Size is less than minimum");
1295 assert(linearAllocationWouldFail() || bestFitFirst(),
1296 "Should not be here");
1297
1298 size_t i;
1299 size_t currSize = numWords + MinChunkSize;
1300 assert(currSize % MinObjAlignment == 0, "currSize should be aligned");
1301 for (i = currSize; i < IndexSetSize; i += IndexSetStride) {
1302 AdaptiveFreeList<FreeChunk>* fl = &_indexedFreeList[i];
1303 if (fl->head()) {
1304 ret = getFromListGreater(fl, numWords);
1305 assert(ret == NULL || ret->is_free(), "Should be returning a free chunk");
1306 return ret;
1307 }
1308 }
1309
1310 currSize = MAX2((size_t)SmallForDictionary,
1311 (size_t)(numWords + MinChunkSize));
1312
1313 /* Try to get a chunk that satisfies request, while avoiding
1314 fragmentation that can't be handled. */
1315 {
1316 ret = dictionary()->get_chunk(currSize);
1317 if (ret != NULL) {
1318 assert(ret->size() - numWords >= MinChunkSize,
1319 "Chunk is too small");
1320 _bt.allocated((HeapWord*)ret, ret->size());
1321 /* Carve returned chunk. */
1322 (void) splitChunkAndReturnRemainder(ret, numWords);
1323 /* Label this as no longer a free chunk. */
1324 assert(ret->is_free(), "This chunk should be free");
1325 ret->link_prev(NULL);
1326 }
1327 assert(ret == NULL || ret->is_free(), "Should be returning a free chunk");
1328 return ret;
1329 }
1330 ShouldNotReachHere();
1331 }
1332
1333 bool CompactibleFreeListSpace::verifyChunkInIndexedFreeLists(FreeChunk* fc) const {
1334 assert(fc->size() < IndexSetSize, "Size of chunk is too large");
1335 return _indexedFreeList[fc->size()].verify_chunk_in_free_list(fc);
1336 }
1337
1338 bool CompactibleFreeListSpace::verify_chunk_is_linear_alloc_block(FreeChunk* fc) const {
1339 assert((_smallLinearAllocBlock._ptr != (HeapWord*)fc) ||
1340 (_smallLinearAllocBlock._word_size == fc->size()),
1341 "Linear allocation block shows incorrect size");
1342 return ((_smallLinearAllocBlock._ptr == (HeapWord*)fc) &&
1343 (_smallLinearAllocBlock._word_size == fc->size()));
1344 }
1345
1346 // Check if the purported free chunk is present either as a linear
1347 // allocation block, the size-indexed table of (smaller) free blocks,
1348 // or the larger free blocks kept in the binary tree dictionary.
1349 bool CompactibleFreeListSpace::verify_chunk_in_free_list(FreeChunk* fc) const {
1350 if (verify_chunk_is_linear_alloc_block(fc)) {
1351 return true;
1352 } else if (fc->size() < IndexSetSize) {
1353 return verifyChunkInIndexedFreeLists(fc);
1354 } else {
1355 return dictionary()->verify_chunk_in_free_list(fc);
1356 }
1357 }
1358
1359 #ifndef PRODUCT
1360 void CompactibleFreeListSpace::assert_locked() const {
1361 CMSLockVerifier::assert_locked(freelistLock(), parDictionaryAllocLock());
1362 }
1363
1364 void CompactibleFreeListSpace::assert_locked(const Mutex* lock) const {
1365 CMSLockVerifier::assert_locked(lock);
1366 }
1367 #endif
1368
1369 FreeChunk* CompactibleFreeListSpace::allocateScratch(size_t size) {
1370 // In the parallel case, the main thread holds the free list lock
1371 // on behalf the parallel threads.
1372 FreeChunk* fc;
1373 {
1374 // If GC is parallel, this might be called by several threads.
1375 // This should be rare enough that the locking overhead won't affect
1376 // the sequential code.
1377 MutexLockerEx x(parDictionaryAllocLock(),
1378 Mutex::_no_safepoint_check_flag);
1379 fc = getChunkFromDictionary(size);
1380 }
1381 if (fc != NULL) {
1382 fc->dontCoalesce();
1383 assert(fc->is_free(), "Should be free, but not coalescable");
1384 // Verify that the block offset table shows this to
1385 // be a single block, but not one which is unallocated.
1386 _bt.verify_single_block((HeapWord*)fc, fc->size());
1387 _bt.verify_not_unallocated((HeapWord*)fc, fc->size());
1388 }
1389 return fc;
1390 }
1391
1392 oop CompactibleFreeListSpace::promote(oop obj, size_t obj_size) {
1393 assert(obj_size == (size_t)obj->size(), "bad obj_size passed in");
1394 assert_locked();
1395
1396 // if we are tracking promotions, then first ensure space for
1397 // promotion (including spooling space for saving header if necessary).
1398 // then allocate and copy, then track promoted info if needed.
1399 // When tracking (see PromotionInfo::track()), the mark word may
1400 // be displaced and in this case restoration of the mark word
1401 // occurs in the (oop_since_save_marks_)iterate phase.
1402 if (_promoInfo.tracking() && !_promoInfo.ensure_spooling_space()) {
1403 return NULL;
1404 }
1405 // Call the allocate(size_t, bool) form directly to avoid the
1406 // additional call through the allocate(size_t) form. Having
1407 // the compile inline the call is problematic because allocate(size_t)
1408 // is a virtual method.
1409 HeapWord* res = allocate(adjustObjectSize(obj_size));
1410 if (res != NULL) {
1411 Copy::aligned_disjoint_words((HeapWord*)obj, res, obj_size);
1412 // if we should be tracking promotions, do so.
1413 if (_promoInfo.tracking()) {
1414 _promoInfo.track((PromotedObject*)res);
1415 }
1416 }
1417 return oop(res);
1418 }
1419
1420 HeapWord*
1421 CompactibleFreeListSpace::getChunkFromSmallLinearAllocBlock(size_t size) {
1422 assert_locked();
1423 assert(size >= MinChunkSize, "minimum chunk size");
1424 assert(size < _smallLinearAllocBlock._allocation_size_limit,
1425 "maximum from smallLinearAllocBlock");
1426 return getChunkFromLinearAllocBlock(&_smallLinearAllocBlock, size);
1427 }
1428
1429 HeapWord*
1430 CompactibleFreeListSpace::getChunkFromLinearAllocBlock(LinearAllocBlock *blk,
1431 size_t size) {
1432 assert_locked();
1433 assert(size >= MinChunkSize, "too small");
1434 HeapWord* res = NULL;
1435 // Try to do linear allocation from blk, making sure that
1436 if (blk->_word_size == 0) {
1437 // We have probably been unable to fill this either in the prologue or
1438 // when it was exhausted at the last linear allocation. Bail out until
1439 // next time.
1440 assert(blk->_ptr == NULL, "consistency check");
1441 return NULL;
1442 }
1443 assert(blk->_word_size != 0 && blk->_ptr != NULL, "consistency check");
1444 res = getChunkFromLinearAllocBlockRemainder(blk, size);
1445 if (res != NULL) return res;
1446
1447 // about to exhaust this linear allocation block
1448 if (blk->_word_size == size) { // exactly satisfied
1449 res = blk->_ptr;
1450 _bt.allocated(res, blk->_word_size);
1451 } else if (size + MinChunkSize <= blk->_refillSize) {
1452 size_t sz = blk->_word_size;
1453 // Update _unallocated_block if the size is such that chunk would be
1454 // returned to the indexed free list. All other chunks in the indexed
1455 // free lists are allocated from the dictionary so that _unallocated_block
1456 // has already been adjusted for them. Do it here so that the cost
1457 // for all chunks added back to the indexed free lists.
1458 if (sz < SmallForDictionary) {
1459 _bt.allocated(blk->_ptr, sz);
1460 }
1461 // Return the chunk that isn't big enough, and then refill below.
1462 addChunkToFreeLists(blk->_ptr, sz);
1463 split_birth(sz);
1464 // Don't keep statistics on adding back chunk from a LinAB.
1465 } else {
1466 // A refilled block would not satisfy the request.
1467 return NULL;
1468 }
1469
1470 blk->_ptr = NULL; blk->_word_size = 0;
1471 refillLinearAllocBlock(blk);
1472 assert(blk->_ptr == NULL || blk->_word_size >= size + MinChunkSize,
1473 "block was replenished");
1474 if (res != NULL) {
1475 split_birth(size);
1476 repairLinearAllocBlock(blk);
1477 } else if (blk->_ptr != NULL) {
1478 res = blk->_ptr;
1479 size_t blk_size = blk->_word_size;
1480 blk->_word_size -= size;
1481 blk->_ptr += size;
1482 split_birth(size);
1483 repairLinearAllocBlock(blk);
1484 // Update BOT last so that other (parallel) GC threads see a consistent
1485 // view of the BOT and free blocks.
1486 // Above must occur before BOT is updated below.
1487 OrderAccess::storestore();
1488 _bt.split_block(res, blk_size, size); // adjust block offset table
1489 }
1490 return res;
1491 }
1492
1493 HeapWord* CompactibleFreeListSpace::getChunkFromLinearAllocBlockRemainder(
1494 LinearAllocBlock* blk,
1495 size_t size) {
1496 assert_locked();
1497 assert(size >= MinChunkSize, "too small");
1498
1499 HeapWord* res = NULL;
1500 // This is the common case. Keep it simple.
1501 if (blk->_word_size >= size + MinChunkSize) {
1502 assert(blk->_ptr != NULL, "consistency check");
1503 res = blk->_ptr;
1504 // Note that the BOT is up-to-date for the linAB before allocation. It
1505 // indicates the start of the linAB. The split_block() updates the
1506 // BOT for the linAB after the allocation (indicates the start of the
1507 // next chunk to be allocated).
1508 size_t blk_size = blk->_word_size;
1509 blk->_word_size -= size;
1510 blk->_ptr += size;
1511 split_birth(size);
1512 repairLinearAllocBlock(blk);
1513 // Update BOT last so that other (parallel) GC threads see a consistent
1514 // view of the BOT and free blocks.
1515 // Above must occur before BOT is updated below.
1516 OrderAccess::storestore();
1517 _bt.split_block(res, blk_size, size); // adjust block offset table
1518 _bt.allocated(res, size);
1519 }
1520 return res;
1521 }
1522
1523 FreeChunk*
1524 CompactibleFreeListSpace::getChunkFromIndexedFreeList(size_t size) {
1525 assert_locked();
1526 assert(size < SmallForDictionary, "just checking");
1527 FreeChunk* res;
1528 res = _indexedFreeList[size].get_chunk_at_head();
1529 if (res == NULL) {
1530 res = getChunkFromIndexedFreeListHelper(size);
1531 }
1532 _bt.verify_not_unallocated((HeapWord*) res, size);
1533 assert(res == NULL || res->size() == size, "Incorrect block size");
1534 return res;
1535 }
1536
1537 FreeChunk*
1538 CompactibleFreeListSpace::getChunkFromIndexedFreeListHelper(size_t size,
1539 bool replenish) {
1540 assert_locked();
1541 FreeChunk* fc = NULL;
1542 if (size < SmallForDictionary) {
1543 assert(_indexedFreeList[size].head() == NULL ||
1544 _indexedFreeList[size].surplus() <= 0,
1545 "List for this size should be empty or under populated");
1546 // Try best fit in exact lists before replenishing the list
1547 if (!bestFitFirst() || (fc = bestFitSmall(size)) == NULL) {
1548 // Replenish list.
1549 //
1550 // Things tried that failed.
1551 // Tried allocating out of the two LinAB's first before
1552 // replenishing lists.
1553 // Tried small linAB of size 256 (size in indexed list)
1554 // and replenishing indexed lists from the small linAB.
1555 //
1556 FreeChunk* newFc = NULL;
1557 const size_t replenish_size = CMSIndexedFreeListReplenish * size;
1558 if (replenish_size < SmallForDictionary) {
1559 // Do not replenish from an underpopulated size.
1560 if (_indexedFreeList[replenish_size].surplus() > 0 &&
1561 _indexedFreeList[replenish_size].head() != NULL) {
1562 newFc = _indexedFreeList[replenish_size].get_chunk_at_head();
1563 } else if (bestFitFirst()) {
1564 newFc = bestFitSmall(replenish_size);
1565 }
1566 }
1567 if (newFc == NULL && replenish_size > size) {
1568 assert(CMSIndexedFreeListReplenish > 1, "ctl pt invariant");
1569 newFc = getChunkFromIndexedFreeListHelper(replenish_size, false);
1570 }
1571 // Note: The stats update re split-death of block obtained above
1572 // will be recorded below precisely when we know we are going to
1573 // be actually splitting it into more than one pieces below.
1574 if (newFc != NULL) {
1575 if (replenish || CMSReplenishIntermediate) {
1576 // Replenish this list and return one block to caller.
1577 size_t i;
1578 FreeChunk *curFc, *nextFc;
1579 size_t num_blk = newFc->size() / size;
1580 assert(num_blk >= 1, "Smaller than requested?");
1581 assert(newFc->size() % size == 0, "Should be integral multiple of request");
1582 if (num_blk > 1) {
1583 // we are sure we will be splitting the block just obtained
1584 // into multiple pieces; record the split-death of the original
1585 splitDeath(replenish_size);
1586 }
1587 // carve up and link blocks 0, ..., num_blk - 2
1588 // The last chunk is not added to the lists but is returned as the
1589 // free chunk.
1590 for (curFc = newFc, nextFc = (FreeChunk*)((HeapWord*)curFc + size),
1591 i = 0;
1592 i < (num_blk - 1);
1593 curFc = nextFc, nextFc = (FreeChunk*)((HeapWord*)nextFc + size),
1594 i++) {
1595 curFc->set_size(size);
1596 // Don't record this as a return in order to try and
1597 // determine the "returns" from a GC.
1598 _bt.verify_not_unallocated((HeapWord*) fc, size);
1599 _indexedFreeList[size].return_chunk_at_tail(curFc, false);
1600 _bt.mark_block((HeapWord*)curFc, size);
1601 split_birth(size);
1602 // Don't record the initial population of the indexed list
1603 // as a split birth.
1604 }
1605
1606 // check that the arithmetic was OK above
1607 assert((HeapWord*)nextFc == (HeapWord*)newFc + num_blk*size,
1608 "inconsistency in carving newFc");
1609 curFc->set_size(size);
1610 _bt.mark_block((HeapWord*)curFc, size);
1611 split_birth(size);
1612 fc = curFc;
1613 } else {
1614 // Return entire block to caller
1615 fc = newFc;
1616 }
1617 }
1618 }
1619 } else {
1620 // Get a free chunk from the free chunk dictionary to be returned to
1621 // replenish the indexed free list.
1622 fc = getChunkFromDictionaryExact(size);
1623 }
1624 // assert(fc == NULL || fc->is_free(), "Should be returning a free chunk");
1625 return fc;
1626 }
1627
1628 FreeChunk*
1629 CompactibleFreeListSpace::getChunkFromDictionary(size_t size) {
1630 assert_locked();
1631 FreeChunk* fc = _dictionary->get_chunk(size,
1632 FreeBlockDictionary<FreeChunk>::atLeast);
1633 if (fc == NULL) {
1634 return NULL;
1635 }
1636 _bt.allocated((HeapWord*)fc, fc->size());
1637 if (fc->size() >= size + MinChunkSize) {
1638 fc = splitChunkAndReturnRemainder(fc, size);
1639 }
1640 assert(fc->size() >= size, "chunk too small");
1641 assert(fc->size() < size + MinChunkSize, "chunk too big");
1642 _bt.verify_single_block((HeapWord*)fc, fc->size());
1643 return fc;
1644 }
1645
1646 FreeChunk*
1647 CompactibleFreeListSpace::getChunkFromDictionaryExact(size_t size) {
1648 assert_locked();
1649 FreeChunk* fc = _dictionary->get_chunk(size,
1650 FreeBlockDictionary<FreeChunk>::atLeast);
1651 if (fc == NULL) {
1652 return fc;
1653 }
1654 _bt.allocated((HeapWord*)fc, fc->size());
1655 if (fc->size() == size) {
1656 _bt.verify_single_block((HeapWord*)fc, size);
1657 return fc;
1658 }
1659 assert(fc->size() > size, "get_chunk() guarantee");
1660 if (fc->size() < size + MinChunkSize) {
1661 // Return the chunk to the dictionary and go get a bigger one.
1662 returnChunkToDictionary(fc);
1663 fc = _dictionary->get_chunk(size + MinChunkSize,
1664 FreeBlockDictionary<FreeChunk>::atLeast);
1665 if (fc == NULL) {
1666 return NULL;
1667 }
1668 _bt.allocated((HeapWord*)fc, fc->size());
1669 }
1670 assert(fc->size() >= size + MinChunkSize, "tautology");
1671 fc = splitChunkAndReturnRemainder(fc, size);
1672 assert(fc->size() == size, "chunk is wrong size");
1673 _bt.verify_single_block((HeapWord*)fc, size);
1674 return fc;
1675 }
1676
1677 void
1678 CompactibleFreeListSpace::returnChunkToDictionary(FreeChunk* chunk) {
1679 assert_locked();
1680
1681 size_t size = chunk->size();
1682 _bt.verify_single_block((HeapWord*)chunk, size);
1683 // adjust _unallocated_block downward, as necessary
1684 _bt.freed((HeapWord*)chunk, size);
1685 _dictionary->return_chunk(chunk);
1686 #ifndef PRODUCT
1687 if (CMSCollector::abstract_state() != CMSCollector::Sweeping) {
1688 TreeChunk<FreeChunk, AdaptiveFreeList<FreeChunk> >* tc = TreeChunk<FreeChunk, AdaptiveFreeList<FreeChunk> >::as_TreeChunk(chunk);
1689 TreeList<FreeChunk, AdaptiveFreeList<FreeChunk> >* tl = tc->list();
1690 tl->verify_stats();
1691 }
1692 #endif // PRODUCT
1693 }
1694
1695 void
1696 CompactibleFreeListSpace::returnChunkToFreeList(FreeChunk* fc) {
1697 assert_locked();
1698 size_t size = fc->size();
1699 _bt.verify_single_block((HeapWord*) fc, size);
1700 _bt.verify_not_unallocated((HeapWord*) fc, size);
1701 if (_adaptive_freelists) {
1702 _indexedFreeList[size].return_chunk_at_tail(fc);
1703 } else {
1704 _indexedFreeList[size].return_chunk_at_head(fc);
1705 }
1706 #ifndef PRODUCT
1707 if (CMSCollector::abstract_state() != CMSCollector::Sweeping) {
1708 _indexedFreeList[size].verify_stats();
1709 }
1710 #endif // PRODUCT
1711 }
1712
1713 // Add chunk to end of last block -- if it's the largest
1714 // block -- and update BOT and census data. We would
1715 // of course have preferred to coalesce it with the
1716 // last block, but it's currently less expensive to find the
1717 // largest block than it is to find the last.
1718 void
1719 CompactibleFreeListSpace::addChunkToFreeListsAtEndRecordingStats(
1720 HeapWord* chunk, size_t size) {
1721 // check that the chunk does lie in this space!
1722 assert(chunk != NULL && is_in_reserved(chunk), "Not in this space!");
1723 // One of the parallel gc task threads may be here
1724 // whilst others are allocating.
1725 Mutex* lock = &_parDictionaryAllocLock;
1726 FreeChunk* ec;
1727 {
1728 MutexLockerEx x(lock, Mutex::_no_safepoint_check_flag);
1729 ec = dictionary()->find_largest_dict(); // get largest block
1730 if (ec != NULL && ec->end() == (uintptr_t*) chunk) {
1731 // It's a coterminal block - we can coalesce.
1732 size_t old_size = ec->size();
1733 coalDeath(old_size);
1734 removeChunkFromDictionary(ec);
1735 size += old_size;
1736 } else {
1737 ec = (FreeChunk*)chunk;
1738 }
1739 }
1740 ec->set_size(size);
1741 debug_only(ec->mangleFreed(size));
1742 if (size < SmallForDictionary) {
1743 lock = _indexedFreeListParLocks[size];
1744 }
1745 MutexLockerEx x(lock, Mutex::_no_safepoint_check_flag);
1746 addChunkAndRepairOffsetTable((HeapWord*)ec, size, true);
1747 // record the birth under the lock since the recording involves
1748 // manipulation of the list on which the chunk lives and
1749 // if the chunk is allocated and is the last on the list,
1750 // the list can go away.
1751 coalBirth(size);
1752 }
1753
1754 void
1755 CompactibleFreeListSpace::addChunkToFreeLists(HeapWord* chunk,
1756 size_t size) {
1757 // check that the chunk does lie in this space!
1758 assert(chunk != NULL && is_in_reserved(chunk), "Not in this space!");
1759 assert_locked();
1760 _bt.verify_single_block(chunk, size);
1761
1762 FreeChunk* fc = (FreeChunk*) chunk;
1763 fc->set_size(size);
1764 debug_only(fc->mangleFreed(size));
1765 if (size < SmallForDictionary) {
1766 returnChunkToFreeList(fc);
1767 } else {
1768 returnChunkToDictionary(fc);
1769 }
1770 }
1771
1772 void
1773 CompactibleFreeListSpace::addChunkAndRepairOffsetTable(HeapWord* chunk,
1774 size_t size, bool coalesced) {
1775 assert_locked();
1776 assert(chunk != NULL, "null chunk");
1777 if (coalesced) {
1778 // repair BOT
1779 _bt.single_block(chunk, size);
1780 }
1781 addChunkToFreeLists(chunk, size);
1782 }
1783
1784 // We _must_ find the purported chunk on our free lists;
1785 // we assert if we don't.
1786 void
1787 CompactibleFreeListSpace::removeFreeChunkFromFreeLists(FreeChunk* fc) {
1788 size_t size = fc->size();
1789 assert_locked();
1790 debug_only(verifyFreeLists());
1791 if (size < SmallForDictionary) {
1792 removeChunkFromIndexedFreeList(fc);
1793 } else {
1794 removeChunkFromDictionary(fc);
1795 }
1796 _bt.verify_single_block((HeapWord*)fc, size);
1797 debug_only(verifyFreeLists());
1798 }
1799
1800 void
1801 CompactibleFreeListSpace::removeChunkFromDictionary(FreeChunk* fc) {
1802 size_t size = fc->size();
1803 assert_locked();
1804 assert(fc != NULL, "null chunk");
1805 _bt.verify_single_block((HeapWord*)fc, size);
1806 _dictionary->remove_chunk(fc);
1807 // adjust _unallocated_block upward, as necessary
1808 _bt.allocated((HeapWord*)fc, size);
1809 }
1810
1811 void
1812 CompactibleFreeListSpace::removeChunkFromIndexedFreeList(FreeChunk* fc) {
1813 assert_locked();
1814 size_t size = fc->size();
1815 _bt.verify_single_block((HeapWord*)fc, size);
1816 NOT_PRODUCT(
1817 if (FLSVerifyIndexTable) {
1818 verifyIndexedFreeList(size);
1819 }
1820 )
1821 _indexedFreeList[size].remove_chunk(fc);
1822 NOT_PRODUCT(
1823 if (FLSVerifyIndexTable) {
1824 verifyIndexedFreeList(size);
1825 }
1826 )
1827 }
1828
1829 FreeChunk* CompactibleFreeListSpace::bestFitSmall(size_t numWords) {
1830 /* A hint is the next larger size that has a surplus.
1831 Start search at a size large enough to guarantee that
1832 the excess is >= MIN_CHUNK. */
1833 size_t start = align_object_size(numWords + MinChunkSize);
1834 if (start < IndexSetSize) {
1835 AdaptiveFreeList<FreeChunk>* it = _indexedFreeList;
1836 size_t hint = _indexedFreeList[start].hint();
1837 while (hint < IndexSetSize) {
1838 assert(hint % MinObjAlignment == 0, "hint should be aligned");
1839 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[hint];
1840 if (fl->surplus() > 0 && fl->head() != NULL) {
1841 // Found a list with surplus, reset original hint
1842 // and split out a free chunk which is returned.
1843 _indexedFreeList[start].set_hint(hint);
1844 FreeChunk* res = getFromListGreater(fl, numWords);
1845 assert(res == NULL || res->is_free(),
1846 "Should be returning a free chunk");
1847 return res;
1848 }
1849 hint = fl->hint(); /* keep looking */
1850 }
1851 /* None found. */
1852 it[start].set_hint(IndexSetSize);
1853 }
1854 return NULL;
1855 }
1856
1857 /* Requires fl->size >= numWords + MinChunkSize */
1858 FreeChunk* CompactibleFreeListSpace::getFromListGreater(AdaptiveFreeList<FreeChunk>* fl,
1859 size_t numWords) {
1860 FreeChunk *curr = fl->head();
1861 size_t oldNumWords = curr->size();
1862 assert(numWords >= MinChunkSize, "Word size is too small");
1863 assert(curr != NULL, "List is empty");
1864 assert(oldNumWords >= numWords + MinChunkSize,
1865 "Size of chunks in the list is too small");
1866
1867 fl->remove_chunk(curr);
1868 // recorded indirectly by splitChunkAndReturnRemainder -
1869 // smallSplit(oldNumWords, numWords);
1870 FreeChunk* new_chunk = splitChunkAndReturnRemainder(curr, numWords);
1871 // Does anything have to be done for the remainder in terms of
1872 // fixing the card table?
1873 assert(new_chunk == NULL || new_chunk->is_free(),
1874 "Should be returning a free chunk");
1875 return new_chunk;
1876 }
1877
1878 FreeChunk*
1879 CompactibleFreeListSpace::splitChunkAndReturnRemainder(FreeChunk* chunk,
1880 size_t new_size) {
1881 assert_locked();
1882 size_t size = chunk->size();
1883 assert(size > new_size, "Split from a smaller block?");
1884 assert(is_aligned(chunk), "alignment problem");
1885 assert(size == adjustObjectSize(size), "alignment problem");
1886 size_t rem_sz = size - new_size;
1887 assert(rem_sz == adjustObjectSize(rem_sz), "alignment problem");
1888 assert(rem_sz >= MinChunkSize, "Free chunk smaller than minimum");
1889 FreeChunk* ffc = (FreeChunk*)((HeapWord*)chunk + new_size);
1890 assert(is_aligned(ffc), "alignment problem");
1891 ffc->set_size(rem_sz);
1892 ffc->link_next(NULL);
1893 ffc->link_prev(NULL); // Mark as a free block for other (parallel) GC threads.
1894 // Above must occur before BOT is updated below.
1895 // adjust block offset table
1896 OrderAccess::storestore();
1897 assert(chunk->is_free() && ffc->is_free(), "Error");
1898 _bt.split_block((HeapWord*)chunk, chunk->size(), new_size);
1899 if (rem_sz < SmallForDictionary) {
1900 bool is_par = (GenCollectedHeap::heap()->n_par_threads() > 0);
1901 if (is_par) _indexedFreeListParLocks[rem_sz]->lock();
1902 assert(!is_par ||
1903 (GenCollectedHeap::heap()->n_par_threads() ==
1904 GenCollectedHeap::heap()->workers()->active_workers()), "Mismatch");
1905 returnChunkToFreeList(ffc);
1906 split(size, rem_sz);
1907 if (is_par) _indexedFreeListParLocks[rem_sz]->unlock();
1908 } else {
1909 returnChunkToDictionary(ffc);
1910 split(size, rem_sz);
1911 }
1912 chunk->set_size(new_size);
1913 return chunk;
1914 }
1915
1916 void
1917 CompactibleFreeListSpace::sweep_completed() {
1918 // Now that space is probably plentiful, refill linear
1919 // allocation blocks as needed.
1920 refillLinearAllocBlocksIfNeeded();
1921 }
1922
1923 void
1924 CompactibleFreeListSpace::gc_prologue() {
1925 assert_locked();
1926 if (PrintFLSStatistics != 0) {
1927 gclog_or_tty->print("Before GC:\n");
1928 reportFreeListStatistics();
1929 }
1930 refillLinearAllocBlocksIfNeeded();
1931 }
1932
1933 void
1934 CompactibleFreeListSpace::gc_epilogue() {
1935 assert_locked();
1936 if (PrintGCDetails && Verbose && !_adaptive_freelists) {
1937 if (_smallLinearAllocBlock._word_size == 0)
1938 warning("CompactibleFreeListSpace(epilogue):: Linear allocation failure");
1939 }
1940 assert(_promoInfo.noPromotions(), "_promoInfo inconsistency");
1941 _promoInfo.stopTrackingPromotions();
1942 repairLinearAllocationBlocks();
1943 // Print Space's stats
1944 if (PrintFLSStatistics != 0) {
1945 gclog_or_tty->print("After GC:\n");
1946 reportFreeListStatistics();
1947 }
1948 }
1949
1950 // Iteration support, mostly delegated from a CMS generation
1951
1952 void CompactibleFreeListSpace::save_marks() {
1953 assert(Thread::current()->is_VM_thread(),
1954 "Global variable should only be set when single-threaded");
1955 // Mark the "end" of the used space at the time of this call;
1956 // note, however, that promoted objects from this point
1957 // on are tracked in the _promoInfo below.
1958 set_saved_mark_word(unallocated_block());
1959 #ifdef ASSERT
1960 // Check the sanity of save_marks() etc.
1961 MemRegion ur = used_region();
1962 MemRegion urasm = used_region_at_save_marks();
1963 assert(ur.contains(urasm),
1964 err_msg(" Error at save_marks(): [" PTR_FORMAT "," PTR_FORMAT ")"
1965 " should contain [" PTR_FORMAT "," PTR_FORMAT ")",
1966 p2i(ur.start()), p2i(ur.end()), p2i(urasm.start()), p2i(urasm.end())));
1967 #endif
1968 // inform allocator that promotions should be tracked.
1969 assert(_promoInfo.noPromotions(), "_promoInfo inconsistency");
1970 _promoInfo.startTrackingPromotions();
1971 }
1972
1973 bool CompactibleFreeListSpace::no_allocs_since_save_marks() {
1974 assert(_promoInfo.tracking(), "No preceding save_marks?");
1975 assert(GenCollectedHeap::heap()->n_par_threads() == 0,
1976 "Shouldn't be called if using parallel gc.");
1977 return _promoInfo.noPromotions();
1978 }
1979
1980 #define CFLS_OOP_SINCE_SAVE_MARKS_DEFN(OopClosureType, nv_suffix) \
1981 \
1982 void CompactibleFreeListSpace:: \
1983 oop_since_save_marks_iterate##nv_suffix(OopClosureType* blk) { \
1984 assert(GenCollectedHeap::heap()->n_par_threads() == 0, \
1985 "Shouldn't be called (yet) during parallel part of gc."); \
1986 _promoInfo.promoted_oops_iterate##nv_suffix(blk); \
1987 /* \
1988 * This also restores any displaced headers and removes the elements from \
1989 * the iteration set as they are processed, so that we have a clean slate \
1990 * at the end of the iteration. Note, thus, that if new objects are \
1991 * promoted as a result of the iteration they are iterated over as well. \
1992 */ \
1993 assert(_promoInfo.noPromotions(), "_promoInfo inconsistency"); \
1994 }
1995
1996 ALL_SINCE_SAVE_MARKS_CLOSURES(CFLS_OOP_SINCE_SAVE_MARKS_DEFN)
1997
1998 bool CompactibleFreeListSpace::linearAllocationWouldFail() const {
1999 return _smallLinearAllocBlock._word_size == 0;
2000 }
2001
2002 void CompactibleFreeListSpace::repairLinearAllocationBlocks() {
2003 // Fix up linear allocation blocks to look like free blocks
2004 repairLinearAllocBlock(&_smallLinearAllocBlock);
2005 }
2006
2007 void CompactibleFreeListSpace::repairLinearAllocBlock(LinearAllocBlock* blk) {
2008 assert_locked();
2009 if (blk->_ptr != NULL) {
2010 assert(blk->_word_size != 0 && blk->_word_size >= MinChunkSize,
2011 "Minimum block size requirement");
2012 FreeChunk* fc = (FreeChunk*)(blk->_ptr);
2013 fc->set_size(blk->_word_size);
2014 fc->link_prev(NULL); // mark as free
2015 fc->dontCoalesce();
2016 assert(fc->is_free(), "just marked it free");
2017 assert(fc->cantCoalesce(), "just marked it uncoalescable");
2018 }
2019 }
2020
2021 void CompactibleFreeListSpace::refillLinearAllocBlocksIfNeeded() {
2022 assert_locked();
2023 if (_smallLinearAllocBlock._ptr == NULL) {
2024 assert(_smallLinearAllocBlock._word_size == 0,
2025 "Size of linAB should be zero if the ptr is NULL");
2026 // Reset the linAB refill and allocation size limit.
2027 _smallLinearAllocBlock.set(0, 0, 1024*SmallForLinearAlloc, SmallForLinearAlloc);
2028 }
2029 refillLinearAllocBlockIfNeeded(&_smallLinearAllocBlock);
2030 }
2031
2032 void
2033 CompactibleFreeListSpace::refillLinearAllocBlockIfNeeded(LinearAllocBlock* blk) {
2034 assert_locked();
2035 assert((blk->_ptr == NULL && blk->_word_size == 0) ||
2036 (blk->_ptr != NULL && blk->_word_size >= MinChunkSize),
2037 "blk invariant");
2038 if (blk->_ptr == NULL) {
2039 refillLinearAllocBlock(blk);
2040 }
2041 if (PrintMiscellaneous && Verbose) {
2042 if (blk->_word_size == 0) {
2043 warning("CompactibleFreeListSpace(prologue):: Linear allocation failure");
2044 }
2045 }
2046 }
2047
2048 void
2049 CompactibleFreeListSpace::refillLinearAllocBlock(LinearAllocBlock* blk) {
2050 assert_locked();
2051 assert(blk->_word_size == 0 && blk->_ptr == NULL,
2052 "linear allocation block should be empty");
2053 FreeChunk* fc;
2054 if (blk->_refillSize < SmallForDictionary &&
2055 (fc = getChunkFromIndexedFreeList(blk->_refillSize)) != NULL) {
2056 // A linAB's strategy might be to use small sizes to reduce
2057 // fragmentation but still get the benefits of allocation from a
2058 // linAB.
2059 } else {
2060 fc = getChunkFromDictionary(blk->_refillSize);
2061 }
2062 if (fc != NULL) {
2063 blk->_ptr = (HeapWord*)fc;
2064 blk->_word_size = fc->size();
2065 fc->dontCoalesce(); // to prevent sweeper from sweeping us up
2066 }
2067 }
2068
2069 // Support for concurrent collection policy decisions.
2070 bool CompactibleFreeListSpace::should_concurrent_collect() const {
2071 // In the future we might want to add in fragmentation stats --
2072 // including erosion of the "mountain" into this decision as well.
2073 return !adaptive_freelists() && linearAllocationWouldFail();
2074 }
2075
2076 // Support for compaction
2077 void CompactibleFreeListSpace::prepare_for_compaction(CompactPoint* cp) {
2078 scan_and_forward(this, cp);
2079 // Prepare_for_compaction() uses the space between live objects
2080 // so that later phase can skip dead space quickly. So verification
2081 // of the free lists doesn't work after.
2082 }
2083
2084 void CompactibleFreeListSpace::adjust_pointers() {
2085 // In other versions of adjust_pointers(), a bail out
2086 // based on the amount of live data in the generation
2087 // (i.e., if 0, bail out) may be used.
2088 // Cannot test used() == 0 here because the free lists have already
2089 // been mangled by the compaction.
2090
2091 scan_and_adjust_pointers(this);
2092 // See note about verification in prepare_for_compaction().
2093 }
2094
2095 void CompactibleFreeListSpace::compact() {
2096 scan_and_compact(this);
2097 }
2098
2099 // Fragmentation metric = 1 - [sum of (fbs**2) / (sum of fbs)**2]
2100 // where fbs is free block sizes
2101 double CompactibleFreeListSpace::flsFrag() const {
2102 size_t itabFree = totalSizeInIndexedFreeLists();
2103 double frag = 0.0;
2104 size_t i;
2105
2106 for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
2107 double sz = i;
2108 frag += _indexedFreeList[i].count() * (sz * sz);
2109 }
2110
2111 double totFree = itabFree +
2112 _dictionary->total_chunk_size(DEBUG_ONLY(freelistLock()));
2113 if (totFree > 0) {
2114 frag = ((frag + _dictionary->sum_of_squared_block_sizes()) /
2115 (totFree * totFree));
2116 frag = (double)1.0 - frag;
2117 } else {
2118 assert(frag == 0.0, "Follows from totFree == 0");
2119 }
2120 return frag;
2121 }
2122
2123 void CompactibleFreeListSpace::beginSweepFLCensus(
2124 float inter_sweep_current,
2125 float inter_sweep_estimate,
2126 float intra_sweep_estimate) {
2127 assert_locked();
2128 size_t i;
2129 for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
2130 AdaptiveFreeList<FreeChunk>* fl = &_indexedFreeList[i];
2131 if (PrintFLSStatistics > 1) {
2132 gclog_or_tty->print("size[" SIZE_FORMAT "] : ", i);
2133 }
2134 fl->compute_desired(inter_sweep_current, inter_sweep_estimate, intra_sweep_estimate);
2135 fl->set_coal_desired((ssize_t)((double)fl->desired() * CMSSmallCoalSurplusPercent));
2136 fl->set_before_sweep(fl->count());
2137 fl->set_bfr_surp(fl->surplus());
2138 }
2139 _dictionary->begin_sweep_dict_census(CMSLargeCoalSurplusPercent,
2140 inter_sweep_current,
2141 inter_sweep_estimate,
2142 intra_sweep_estimate);
2143 }
2144
2145 void CompactibleFreeListSpace::setFLSurplus() {
2146 assert_locked();
2147 size_t i;
2148 for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
2149 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[i];
2150 fl->set_surplus(fl->count() -
2151 (ssize_t)((double)fl->desired() * CMSSmallSplitSurplusPercent));
2152 }
2153 }
2154
2155 void CompactibleFreeListSpace::setFLHints() {
2156 assert_locked();
2157 size_t i;
2158 size_t h = IndexSetSize;
2159 for (i = IndexSetSize - 1; i != 0; i -= IndexSetStride) {
2160 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[i];
2161 fl->set_hint(h);
2162 if (fl->surplus() > 0) {
2163 h = i;
2164 }
2165 }
2166 }
2167
2168 void CompactibleFreeListSpace::clearFLCensus() {
2169 assert_locked();
2170 size_t i;
2171 for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
2172 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[i];
2173 fl->set_prev_sweep(fl->count());
2174 fl->set_coal_births(0);
2175 fl->set_coal_deaths(0);
2176 fl->set_split_births(0);
2177 fl->set_split_deaths(0);
2178 }
2179 }
2180
2181 void CompactibleFreeListSpace::endSweepFLCensus(size_t sweep_count) {
2182 if (PrintFLSStatistics > 0) {
2183 HeapWord* largestAddr = (HeapWord*) dictionary()->find_largest_dict();
2184 gclog_or_tty->print_cr("CMS: Large block " PTR_FORMAT,
2185 p2i(largestAddr));
2186 }
2187 setFLSurplus();
2188 setFLHints();
2189 if (PrintGC && PrintFLSCensus > 0) {
2190 printFLCensus(sweep_count);
2191 }
2192 clearFLCensus();
2193 assert_locked();
2194 _dictionary->end_sweep_dict_census(CMSLargeSplitSurplusPercent);
2195 }
2196
2197 bool CompactibleFreeListSpace::coalOverPopulated(size_t size) {
2198 if (size < SmallForDictionary) {
2199 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[size];
2200 return (fl->coal_desired() < 0) ||
2201 ((int)fl->count() > fl->coal_desired());
2202 } else {
2203 return dictionary()->coal_dict_over_populated(size);
2204 }
2205 }
2206
2207 void CompactibleFreeListSpace::smallCoalBirth(size_t size) {
2208 assert(size < SmallForDictionary, "Size too large for indexed list");
2209 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[size];
2210 fl->increment_coal_births();
2211 fl->increment_surplus();
2212 }
2213
2214 void CompactibleFreeListSpace::smallCoalDeath(size_t size) {
2215 assert(size < SmallForDictionary, "Size too large for indexed list");
2216 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[size];
2217 fl->increment_coal_deaths();
2218 fl->decrement_surplus();
2219 }
2220
2221 void CompactibleFreeListSpace::coalBirth(size_t size) {
2222 if (size < SmallForDictionary) {
2223 smallCoalBirth(size);
2224 } else {
2225 dictionary()->dict_census_update(size,
2226 false /* split */,
2227 true /* birth */);
2228 }
2229 }
2230
2231 void CompactibleFreeListSpace::coalDeath(size_t size) {
2232 if(size < SmallForDictionary) {
2233 smallCoalDeath(size);
2234 } else {
2235 dictionary()->dict_census_update(size,
2236 false /* split */,
2237 false /* birth */);
2238 }
2239 }
2240
2241 void CompactibleFreeListSpace::smallSplitBirth(size_t size) {
2242 assert(size < SmallForDictionary, "Size too large for indexed list");
2243 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[size];
2244 fl->increment_split_births();
2245 fl->increment_surplus();
2246 }
2247
2248 void CompactibleFreeListSpace::smallSplitDeath(size_t size) {
2249 assert(size < SmallForDictionary, "Size too large for indexed list");
2250 AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[size];
2251 fl->increment_split_deaths();
2252 fl->decrement_surplus();
2253 }
2254
2255 void CompactibleFreeListSpace::split_birth(size_t size) {
2256 if (size < SmallForDictionary) {
2257 smallSplitBirth(size);
2258 } else {
2259 dictionary()->dict_census_update(size,
2260 true /* split */,
2261 true /* birth */);
2262 }
2263 }
2264
2265 void CompactibleFreeListSpace::splitDeath(size_t size) {
2266 if (size < SmallForDictionary) {
2267 smallSplitDeath(size);
2268 } else {
2269 dictionary()->dict_census_update(size,
2270 true /* split */,
2271 false /* birth */);
2272 }
2273 }
2274
2275 void CompactibleFreeListSpace::split(size_t from, size_t to1) {
2276 size_t to2 = from - to1;
2277 splitDeath(from);
2278 split_birth(to1);
2279 split_birth(to2);
2280 }
2281
2282 void CompactibleFreeListSpace::print() const {
2283 print_on(tty);
2284 }
2285
2286 void CompactibleFreeListSpace::prepare_for_verify() {
2287 assert_locked();
2288 repairLinearAllocationBlocks();
2289 // Verify that the SpoolBlocks look like free blocks of
2290 // appropriate sizes... To be done ...
2291 }
2292
2293 class VerifyAllBlksClosure: public BlkClosure {
2294 private:
2295 const CompactibleFreeListSpace* _sp;
2296 const MemRegion _span;
2297 HeapWord* _last_addr;
2298 size_t _last_size;
2299 bool _last_was_obj;
2300 bool _last_was_live;
2301
2302 public:
2303 VerifyAllBlksClosure(const CompactibleFreeListSpace* sp,
2304 MemRegion span) : _sp(sp), _span(span),
2305 _last_addr(NULL), _last_size(0),
2306 _last_was_obj(false), _last_was_live(false) { }
2307
2308 virtual size_t do_blk(HeapWord* addr) {
2309 size_t res;
2310 bool was_obj = false;
2311 bool was_live = false;
2312 if (_sp->block_is_obj(addr)) {
2313 was_obj = true;
2314 oop p = oop(addr);
2315 guarantee(p->is_oop(), "Should be an oop");
2316 res = _sp->adjustObjectSize(p->size());
2317 if (_sp->obj_is_alive(addr)) {
2318 was_live = true;
2319 p->verify();
2320 }
2321 } else {
2322 FreeChunk* fc = (FreeChunk*)addr;
2323 res = fc->size();
2324 if (FLSVerifyLists && !fc->cantCoalesce()) {
2325 guarantee(_sp->verify_chunk_in_free_list(fc),
2326 "Chunk should be on a free list");
2327 }
2328 }
2329 if (res == 0) {
2330 gclog_or_tty->print_cr("Livelock: no rank reduction!");
2331 gclog_or_tty->print_cr(
2332 " Current: addr = " PTR_FORMAT ", size = " SIZE_FORMAT ", obj = %s, live = %s \n"
2333 " Previous: addr = " PTR_FORMAT ", size = " SIZE_FORMAT ", obj = %s, live = %s \n",
2334 p2i(addr), res, was_obj ?"true":"false", was_live ?"true":"false",
2335 p2i(_last_addr), _last_size, _last_was_obj?"true":"false", _last_was_live?"true":"false");
2336 _sp->print_on(gclog_or_tty);
2337 guarantee(false, "Seppuku!");
2338 }
2339 _last_addr = addr;
2340 _last_size = res;
2341 _last_was_obj = was_obj;
2342 _last_was_live = was_live;
2343 return res;
2344 }
2345 };
2346
2347 class VerifyAllOopsClosure: public OopClosure {
2348 private:
2349 const CMSCollector* _collector;
2350 const CompactibleFreeListSpace* _sp;
2351 const MemRegion _span;
2352 const bool _past_remark;
2353 const CMSBitMap* _bit_map;
2354
2355 protected:
2356 void do_oop(void* p, oop obj) {
2357 if (_span.contains(obj)) { // the interior oop points into CMS heap
2358 if (!_span.contains(p)) { // reference from outside CMS heap
2359 // Should be a valid object; the first disjunct below allows
2360 // us to sidestep an assertion in block_is_obj() that insists
2361 // that p be in _sp. Note that several generations (and spaces)
2362 // are spanned by _span (CMS heap) above.
2363 guarantee(!_sp->is_in_reserved(obj) ||
2364 _sp->block_is_obj((HeapWord*)obj),
2365 "Should be an object");
2366 guarantee(obj->is_oop(), "Should be an oop");
2367 obj->verify();
2368 if (_past_remark) {
2369 // Remark has been completed, the object should be marked
2370 _bit_map->isMarked((HeapWord*)obj);
2371 }
2372 } else { // reference within CMS heap
2373 if (_past_remark) {
2374 // Remark has been completed -- so the referent should have
2375 // been marked, if referring object is.
2376 if (_bit_map->isMarked(_collector->block_start(p))) {
2377 guarantee(_bit_map->isMarked((HeapWord*)obj), "Marking error?");
2378 }
2379 }
2380 }
2381 } else if (_sp->is_in_reserved(p)) {
2382 // the reference is from FLS, and points out of FLS
2383 guarantee(obj->is_oop(), "Should be an oop");
2384 obj->verify();
2385 }
2386 }
2387
2388 template <class T> void do_oop_work(T* p) {
2389 T heap_oop = oopDesc::load_heap_oop(p);
2390 if (!oopDesc::is_null(heap_oop)) {
2391 oop obj = oopDesc::decode_heap_oop_not_null(heap_oop);
2392 do_oop(p, obj);
2393 }
2394 }
2395
2396 public:
2397 VerifyAllOopsClosure(const CMSCollector* collector,
2398 const CompactibleFreeListSpace* sp, MemRegion span,
2399 bool past_remark, CMSBitMap* bit_map) :
2400 _collector(collector), _sp(sp), _span(span),
2401 _past_remark(past_remark), _bit_map(bit_map) { }
2402
2403 virtual void do_oop(oop* p) { VerifyAllOopsClosure::do_oop_work(p); }
2404 virtual void do_oop(narrowOop* p) { VerifyAllOopsClosure::do_oop_work(p); }
2405 };
2406
2407 void CompactibleFreeListSpace::verify() const {
2408 assert_lock_strong(&_freelistLock);
2409 verify_objects_initialized();
2410 MemRegion span = _collector->_span;
2411 bool past_remark = (_collector->abstract_state() ==
2412 CMSCollector::Sweeping);
2413
2414 ResourceMark rm;
2415 HandleMark hm;
2416
2417 // Check integrity of CFL data structures
2418 _promoInfo.verify();
2419 _dictionary->verify();
2420 if (FLSVerifyIndexTable) {
2421 verifyIndexedFreeLists();
2422 }
2423 // Check integrity of all objects and free blocks in space
2424 {
2425 VerifyAllBlksClosure cl(this, span);
2426 ((CompactibleFreeListSpace*)this)->blk_iterate(&cl); // cast off const
2427 }
2428 // Check that all references in the heap to FLS
2429 // are to valid objects in FLS or that references in
2430 // FLS are to valid objects elsewhere in the heap
2431 if (FLSVerifyAllHeapReferences)
2432 {
2433 VerifyAllOopsClosure cl(_collector, this, span, past_remark,
2434 _collector->markBitMap());
2435
2436 // Iterate over all oops in the heap. Uses the _no_header version
2437 // since we are not interested in following the klass pointers.
2438 GenCollectedHeap::heap()->oop_iterate_no_header(&cl);
2439 }
2440
2441 if (VerifyObjectStartArray) {
2442 // Verify the block offset table
2443 _bt.verify();
2444 }
2445 }
2446
2447 #ifndef PRODUCT
2448 void CompactibleFreeListSpace::verifyFreeLists() const {
2449 if (FLSVerifyLists) {
2450 _dictionary->verify();
2451 verifyIndexedFreeLists();
2452 } else {
2453 if (FLSVerifyDictionary) {
2454 _dictionary->verify();
2455 }
2456 if (FLSVerifyIndexTable) {
2457 verifyIndexedFreeLists();
2458 }
2459 }
2460 }
2461 #endif
2462
2463 void CompactibleFreeListSpace::verifyIndexedFreeLists() const {
2464 size_t i = 0;
2465 for (; i < IndexSetStart; i++) {
2466 guarantee(_indexedFreeList[i].head() == NULL, "should be NULL");
2467 }
2468 for (; i < IndexSetSize; i++) {
2469 verifyIndexedFreeList(i);
2470 }
2471 }
2472
2473 void CompactibleFreeListSpace::verifyIndexedFreeList(size_t size) const {
2474 FreeChunk* fc = _indexedFreeList[size].head();
2475 FreeChunk* tail = _indexedFreeList[size].tail();
2476 size_t num = _indexedFreeList[size].count();
2477 size_t n = 0;
2478 guarantee(((size >= IndexSetStart) && (size % IndexSetStride == 0)) || fc == NULL,
2479 "Slot should have been empty");
2480 for (; fc != NULL; fc = fc->next(), n++) {
2481 guarantee(fc->size() == size, "Size inconsistency");
2482 guarantee(fc->is_free(), "!free?");
2483 guarantee(fc->next() == NULL || fc->next()->prev() == fc, "Broken list");
2484 guarantee((fc->next() == NULL) == (fc == tail), "Incorrect tail");
2485 }
2486 guarantee(n == num, "Incorrect count");
2487 }
2488
2489 #ifndef PRODUCT
2490 void CompactibleFreeListSpace::check_free_list_consistency() const {
2491 assert((TreeChunk<FreeChunk, AdaptiveFreeList<FreeChunk> >::min_size() <= IndexSetSize),
2492 "Some sizes can't be allocated without recourse to"
2493 " linear allocation buffers");
2494 assert((TreeChunk<FreeChunk, AdaptiveFreeList<FreeChunk> >::min_size()*HeapWordSize == sizeof(TreeChunk<FreeChunk, AdaptiveFreeList<FreeChunk> >)),
2495 "else MIN_TREE_CHUNK_SIZE is wrong");
2496 assert(IndexSetStart != 0, "IndexSetStart not initialized");
2497 assert(IndexSetStride != 0, "IndexSetStride not initialized");
2498 }
2499 #endif
2500
2501 void CompactibleFreeListSpace::printFLCensus(size_t sweep_count) const {
2502 assert_lock_strong(&_freelistLock);
2503 AdaptiveFreeList<FreeChunk> total;
2504 gclog_or_tty->print("end sweep# " SIZE_FORMAT "\n", sweep_count);
2505 AdaptiveFreeList<FreeChunk>::print_labels_on(gclog_or_tty, "size");
2506 size_t total_free = 0;
2507 for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) {
2508 const AdaptiveFreeList<FreeChunk> *fl = &_indexedFreeList[i];
2509 total_free += fl->count() * fl->size();
2510 if (i % (40*IndexSetStride) == 0) {
2511 AdaptiveFreeList<FreeChunk>::print_labels_on(gclog_or_tty, "size");
2512 }
2513 fl->print_on(gclog_or_tty);
2514 total.set_bfr_surp( total.bfr_surp() + fl->bfr_surp() );
2515 total.set_surplus( total.surplus() + fl->surplus() );
2516 total.set_desired( total.desired() + fl->desired() );
2517 total.set_prev_sweep( total.prev_sweep() + fl->prev_sweep() );
2518 total.set_before_sweep(total.before_sweep() + fl->before_sweep());
2519 total.set_count( total.count() + fl->count() );
2520 total.set_coal_births( total.coal_births() + fl->coal_births() );
2521 total.set_coal_deaths( total.coal_deaths() + fl->coal_deaths() );
2522 total.set_split_births(total.split_births() + fl->split_births());
2523 total.set_split_deaths(total.split_deaths() + fl->split_deaths());
2524 }
2525 total.print_on(gclog_or_tty, "TOTAL");
2526 gclog_or_tty->print_cr("Total free in indexed lists "
2527 SIZE_FORMAT " words", total_free);
2528 gclog_or_tty->print("growth: %8.5f deficit: %8.5f\n",
2529 (double)(total.split_births()+total.coal_births()-total.split_deaths()-total.coal_deaths())/
2530 (total.prev_sweep() != 0 ? (double)total.prev_sweep() : 1.0),
2531 (double)(total.desired() - total.count())/(total.desired() != 0 ? (double)total.desired() : 1.0));
2532 _dictionary->print_dict_census();
2533 }
2534
2535 ///////////////////////////////////////////////////////////////////////////
2536 // CFLS_LAB
2537 ///////////////////////////////////////////////////////////////////////////
2538
2539 #define VECTOR_257(x) \
2540 /* 1 2 3 4 5 6 7 8 9 1x 11 12 13 14 15 16 17 18 19 2x 21 22 23 24 25 26 27 28 29 3x 31 32 */ \
2541 { x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2542 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2543 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2544 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2545 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2546 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2547 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2548 x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, x, \
2549 x }
2550
2551 // Initialize with default setting for CMS, _not_
2552 // generic OldPLABSize, whose static default is different; if overridden at the
2553 // command-line, this will get reinitialized via a call to
2554 // modify_initialization() below.
2555 AdaptiveWeightedAverage CFLS_LAB::_blocks_to_claim[] =
2556 VECTOR_257(AdaptiveWeightedAverage(OldPLABWeight, (float)CFLS_LAB::_default_dynamic_old_plab_size));
2557 size_t CFLS_LAB::_global_num_blocks[] = VECTOR_257(0);
2558 uint CFLS_LAB::_global_num_workers[] = VECTOR_257(0);
2559
2560 CFLS_LAB::CFLS_LAB(CompactibleFreeListSpace* cfls) :
2561 _cfls(cfls)
2562 {
2563 assert(CompactibleFreeListSpace::IndexSetSize == 257, "Modify VECTOR_257() macro above");
2564 for (size_t i = CompactibleFreeListSpace::IndexSetStart;
2565 i < CompactibleFreeListSpace::IndexSetSize;
2566 i += CompactibleFreeListSpace::IndexSetStride) {
2567 _indexedFreeList[i].set_size(i);
2568 _num_blocks[i] = 0;
2569 }
2570 }
2571
2572 static bool _CFLS_LAB_modified = false;
2573
2574 void CFLS_LAB::modify_initialization(size_t n, unsigned wt) {
2575 assert(!_CFLS_LAB_modified, "Call only once");
2576 _CFLS_LAB_modified = true;
2577 for (size_t i = CompactibleFreeListSpace::IndexSetStart;
2578 i < CompactibleFreeListSpace::IndexSetSize;
2579 i += CompactibleFreeListSpace::IndexSetStride) {
2580 _blocks_to_claim[i].modify(n, wt, true /* force */);
2581 }
2582 }
2583
2584 HeapWord* CFLS_LAB::alloc(size_t word_sz) {
2585 FreeChunk* res;
2586 assert(word_sz == _cfls->adjustObjectSize(word_sz), "Error");
2587 if (word_sz >= CompactibleFreeListSpace::IndexSetSize) {
2588 // This locking manages sync with other large object allocations.
2589 MutexLockerEx x(_cfls->parDictionaryAllocLock(),
2590 Mutex::_no_safepoint_check_flag);
2591 res = _cfls->getChunkFromDictionaryExact(word_sz);
2592 if (res == NULL) return NULL;
2593 } else {
2594 AdaptiveFreeList<FreeChunk>* fl = &_indexedFreeList[word_sz];
2595 if (fl->count() == 0) {
2596 // Attempt to refill this local free list.
2597 get_from_global_pool(word_sz, fl);
2598 // If it didn't work, give up.
2599 if (fl->count() == 0) return NULL;
2600 }
2601 res = fl->get_chunk_at_head();
2602 assert(res != NULL, "Why was count non-zero?");
2603 }
2604 res->markNotFree();
2605 assert(!res->is_free(), "shouldn't be marked free");
2606 assert(oop(res)->klass_or_null() == NULL, "should look uninitialized");
2607 // mangle a just allocated object with a distinct pattern.
2608 debug_only(res->mangleAllocated(word_sz));
2609 return (HeapWord*)res;
2610 }
2611
2612 // Get a chunk of blocks of the right size and update related
2613 // book-keeping stats
2614 void CFLS_LAB::get_from_global_pool(size_t word_sz, AdaptiveFreeList<FreeChunk>* fl) {
2615 // Get the #blocks we want to claim
2616 size_t n_blks = (size_t)_blocks_to_claim[word_sz].average();
2617 assert(n_blks > 0, "Error");
2618 assert(ResizeOldPLAB || n_blks == OldPLABSize, "Error");
2619 // In some cases, when the application has a phase change,
2620 // there may be a sudden and sharp shift in the object survival
2621 // profile, and updating the counts at the end of a scavenge
2622 // may not be quick enough, giving rise to large scavenge pauses
2623 // during these phase changes. It is beneficial to detect such
2624 // changes on-the-fly during a scavenge and avoid such a phase-change
2625 // pothole. The following code is a heuristic attempt to do that.
2626 // It is protected by a product flag until we have gained
2627 // enough experience with this heuristic and fine-tuned its behavior.
2628 // WARNING: This might increase fragmentation if we overreact to
2629 // small spikes, so some kind of historical smoothing based on
2630 // previous experience with the greater reactivity might be useful.
2631 // Lacking sufficient experience, CMSOldPLABResizeQuicker is disabled by
2632 // default.
2633 if (ResizeOldPLAB && CMSOldPLABResizeQuicker) {
2634 size_t multiple = _num_blocks[word_sz]/(CMSOldPLABToleranceFactor*CMSOldPLABNumRefills*n_blks);
2635 n_blks += CMSOldPLABReactivityFactor*multiple*n_blks;
2636 n_blks = MIN2(n_blks, CMSOldPLABMax);
2637 }
2638 assert(n_blks > 0, "Error");
2639 _cfls->par_get_chunk_of_blocks(word_sz, n_blks, fl);
2640 // Update stats table entry for this block size
2641 _num_blocks[word_sz] += fl->count();
2642 }
2643
2644 void CFLS_LAB::compute_desired_plab_size() {
2645 for (size_t i = CompactibleFreeListSpace::IndexSetStart;
2646 i < CompactibleFreeListSpace::IndexSetSize;
2647 i += CompactibleFreeListSpace::IndexSetStride) {
2648 assert((_global_num_workers[i] == 0) == (_global_num_blocks[i] == 0),
2649 "Counter inconsistency");
2650 if (_global_num_workers[i] > 0) {
2651 // Need to smooth wrt historical average
2652 if (ResizeOldPLAB) {
2653 _blocks_to_claim[i].sample(
2654 MAX2(CMSOldPLABMin,
2655 MIN2(CMSOldPLABMax,
2656 _global_num_blocks[i]/(_global_num_workers[i]*CMSOldPLABNumRefills))));
2657 }
2658 // Reset counters for next round
2659 _global_num_workers[i] = 0;
2660 _global_num_blocks[i] = 0;
2661 if (PrintOldPLAB) {
2662 gclog_or_tty->print_cr("[" SIZE_FORMAT "]: " SIZE_FORMAT,
2663 i, (size_t)_blocks_to_claim[i].average());
2664 }
2665 }
2666 }
2667 }
2668
2669 // If this is changed in the future to allow parallel
2670 // access, one would need to take the FL locks and,
2671 // depending on how it is used, stagger access from
2672 // parallel threads to reduce contention.
2673 void CFLS_LAB::retire(int tid) {
2674 // We run this single threaded with the world stopped;
2675 // so no need for locks and such.
2676 NOT_PRODUCT(Thread* t = Thread::current();)
2677 assert(Thread::current()->is_VM_thread(), "Error");
2678 for (size_t i = CompactibleFreeListSpace::IndexSetStart;
2679 i < CompactibleFreeListSpace::IndexSetSize;
2680 i += CompactibleFreeListSpace::IndexSetStride) {
2681 assert(_num_blocks[i] >= (size_t)_indexedFreeList[i].count(),
2682 "Can't retire more than what we obtained");
2683 if (_num_blocks[i] > 0) {
2684 size_t num_retire = _indexedFreeList[i].count();
2685 assert(_num_blocks[i] > num_retire, "Should have used at least one");
2686 {
2687 // MutexLockerEx x(_cfls->_indexedFreeListParLocks[i],
2688 // Mutex::_no_safepoint_check_flag);
2689
2690 // Update globals stats for num_blocks used
2691 _global_num_blocks[i] += (_num_blocks[i] - num_retire);
2692 _global_num_workers[i]++;
2693 assert(_global_num_workers[i] <= ParallelGCThreads, "Too big");
2694 if (num_retire > 0) {
2695 _cfls->_indexedFreeList[i].prepend(&_indexedFreeList[i]);
2696 // Reset this list.
2697 _indexedFreeList[i] = AdaptiveFreeList<FreeChunk>();
2698 _indexedFreeList[i].set_size(i);
2699 }
2700 }
2701 if (PrintOldPLAB) {
2702 gclog_or_tty->print_cr("%d[" SIZE_FORMAT "]: " SIZE_FORMAT "/" SIZE_FORMAT "/" SIZE_FORMAT,
2703 tid, i, num_retire, _num_blocks[i], (size_t)_blocks_to_claim[i].average());
2704 }
2705 // Reset stats for next round
2706 _num_blocks[i] = 0;
2707 }
2708 }
2709 }
2710
2711 // Used by par_get_chunk_of_blocks() for the chunks from the
2712 // indexed_free_lists. Looks for a chunk with size that is a multiple
2713 // of "word_sz" and if found, splits it into "word_sz" chunks and add
2714 // to the free list "fl". "n" is the maximum number of chunks to
2715 // be added to "fl".
2716 bool CompactibleFreeListSpace:: par_get_chunk_of_blocks_IFL(size_t word_sz, size_t n, AdaptiveFreeList<FreeChunk>* fl) {
2717
2718 // We'll try all multiples of word_sz in the indexed set, starting with
2719 // word_sz itself and, if CMSSplitIndexedFreeListBlocks, try larger multiples,
2720 // then try getting a big chunk and splitting it.
2721 {
2722 bool found;
2723 int k;
2724 size_t cur_sz;
2725 for (k = 1, cur_sz = k * word_sz, found = false;
2726 (cur_sz < CompactibleFreeListSpace::IndexSetSize) &&
2727 (CMSSplitIndexedFreeListBlocks || k <= 1);
2728 k++, cur_sz = k * word_sz) {
2729 AdaptiveFreeList<FreeChunk> fl_for_cur_sz; // Empty.
2730 fl_for_cur_sz.set_size(cur_sz);
2731 {
2732 MutexLockerEx x(_indexedFreeListParLocks[cur_sz],
2733 Mutex::_no_safepoint_check_flag);
2734 AdaptiveFreeList<FreeChunk>* gfl = &_indexedFreeList[cur_sz];
2735 if (gfl->count() != 0) {
2736 // nn is the number of chunks of size cur_sz that
2737 // we'd need to split k-ways each, in order to create
2738 // "n" chunks of size word_sz each.
2739 const size_t nn = MAX2(n/k, (size_t)1);
2740 gfl->getFirstNChunksFromList(nn, &fl_for_cur_sz);
2741 found = true;
2742 if (k > 1) {
2743 // Update split death stats for the cur_sz-size blocks list:
2744 // we increment the split death count by the number of blocks
2745 // we just took from the cur_sz-size blocks list and which
2746 // we will be splitting below.
2747 ssize_t deaths = gfl->split_deaths() +
2748 fl_for_cur_sz.count();
2749 gfl->set_split_deaths(deaths);
2750 }
2751 }
2752 }
2753 // Now transfer fl_for_cur_sz to fl. Common case, we hope, is k = 1.
2754 if (found) {
2755 if (k == 1) {
2756 fl->prepend(&fl_for_cur_sz);
2757 } else {
2758 // Divide each block on fl_for_cur_sz up k ways.
2759 FreeChunk* fc;
2760 while ((fc = fl_for_cur_sz.get_chunk_at_head()) != NULL) {
2761 // Must do this in reverse order, so that anybody attempting to
2762 // access the main chunk sees it as a single free block until we
2763 // change it.
2764 size_t fc_size = fc->size();
2765 assert(fc->is_free(), "Error");
2766 for (int i = k-1; i >= 0; i--) {
2767 FreeChunk* ffc = (FreeChunk*)((HeapWord*)fc + i * word_sz);
2768 assert((i != 0) ||
2769 ((fc == ffc) && ffc->is_free() &&
2770 (ffc->size() == k*word_sz) && (fc_size == word_sz)),
2771 "Counting error");
2772 ffc->set_size(word_sz);
2773 ffc->link_prev(NULL); // Mark as a free block for other (parallel) GC threads.
2774 ffc->link_next(NULL);
2775 // Above must occur before BOT is updated below.
2776 OrderAccess::storestore();
2777 // splitting from the right, fc_size == i * word_sz
2778 _bt.mark_block((HeapWord*)ffc, word_sz, true /* reducing */);
2779 fc_size -= word_sz;
2780 assert(fc_size == i*word_sz, "Error");
2781 _bt.verify_not_unallocated((HeapWord*)ffc, word_sz);
2782 _bt.verify_single_block((HeapWord*)fc, fc_size);
2783 _bt.verify_single_block((HeapWord*)ffc, word_sz);
2784 // Push this on "fl".
2785 fl->return_chunk_at_head(ffc);
2786 }
2787 // TRAP
2788 assert(fl->tail()->next() == NULL, "List invariant.");
2789 }
2790 }
2791 // Update birth stats for this block size.
2792 size_t num = fl->count();
2793 MutexLockerEx x(_indexedFreeListParLocks[word_sz],
2794 Mutex::_no_safepoint_check_flag);
2795 ssize_t births = _indexedFreeList[word_sz].split_births() + num;
2796 _indexedFreeList[word_sz].set_split_births(births);
2797 return true;
2798 }
2799 }
2800 return found;
2801 }
2802 }
2803
2804 FreeChunk* CompactibleFreeListSpace::get_n_way_chunk_to_split(size_t word_sz, size_t n) {
2805
2806 FreeChunk* fc = NULL;
2807 FreeChunk* rem_fc = NULL;
2808 size_t rem;
2809 {
2810 MutexLockerEx x(parDictionaryAllocLock(),
2811 Mutex::_no_safepoint_check_flag);
2812 while (n > 0) {
2813 fc = dictionary()->get_chunk(MAX2(n * word_sz, _dictionary->min_size()),
2814 FreeBlockDictionary<FreeChunk>::atLeast);
2815 if (fc != NULL) {
2816 break;
2817 } else {
2818 n--;
2819 }
2820 }
2821 if (fc == NULL) return NULL;
2822 // Otherwise, split up that block.
2823 assert((ssize_t)n >= 1, "Control point invariant");
2824 assert(fc->is_free(), "Error: should be a free block");
2825 _bt.verify_single_block((HeapWord*)fc, fc->size());
2826 const size_t nn = fc->size() / word_sz;
2827 n = MIN2(nn, n);
2828 assert((ssize_t)n >= 1, "Control point invariant");
2829 rem = fc->size() - n * word_sz;
2830 // If there is a remainder, and it's too small, allocate one fewer.
2831 if (rem > 0 && rem < MinChunkSize) {
2832 n--; rem += word_sz;
2833 }
2834 // Note that at this point we may have n == 0.
2835 assert((ssize_t)n >= 0, "Control point invariant");
2836
2837 // If n is 0, the chunk fc that was found is not large
2838 // enough to leave a viable remainder. We are unable to
2839 // allocate even one block. Return fc to the
2840 // dictionary and return, leaving "fl" empty.
2841 if (n == 0) {
2842 returnChunkToDictionary(fc);
2843 return NULL;
2844 }
2845
2846 _bt.allocated((HeapWord*)fc, fc->size(), true /* reducing */); // update _unallocated_blk
2847 dictionary()->dict_census_update(fc->size(),
2848 true /*split*/,
2849 false /*birth*/);
2850
2851 // First return the remainder, if any.
2852 // Note that we hold the lock until we decide if we're going to give
2853 // back the remainder to the dictionary, since a concurrent allocation
2854 // may otherwise see the heap as empty. (We're willing to take that
2855 // hit if the block is a small block.)
2856 if (rem > 0) {
2857 size_t prefix_size = n * word_sz;
2858 rem_fc = (FreeChunk*)((HeapWord*)fc + prefix_size);
2859 rem_fc->set_size(rem);
2860 rem_fc->link_prev(NULL); // Mark as a free block for other (parallel) GC threads.
2861 rem_fc->link_next(NULL);
2862 // Above must occur before BOT is updated below.
2863 assert((ssize_t)n > 0 && prefix_size > 0 && rem_fc > fc, "Error");
2864 OrderAccess::storestore();
2865 _bt.split_block((HeapWord*)fc, fc->size(), prefix_size);
2866 assert(fc->is_free(), "Error");
2867 fc->set_size(prefix_size);
2868 if (rem >= IndexSetSize) {
2869 returnChunkToDictionary(rem_fc);
2870 dictionary()->dict_census_update(rem, true /*split*/, true /*birth*/);
2871 rem_fc = NULL;
2872 }
2873 // Otherwise, return it to the small list below.
2874 }
2875 }
2876 if (rem_fc != NULL) {
2877 MutexLockerEx x(_indexedFreeListParLocks[rem],
2878 Mutex::_no_safepoint_check_flag);
2879 _bt.verify_not_unallocated((HeapWord*)rem_fc, rem_fc->size());
2880 _indexedFreeList[rem].return_chunk_at_head(rem_fc);
2881 smallSplitBirth(rem);
2882 }
2883 assert(n * word_sz == fc->size(),
2884 err_msg("Chunk size " SIZE_FORMAT " is not exactly splittable by "
2885 SIZE_FORMAT " sized chunks of size " SIZE_FORMAT,
2886 fc->size(), n, word_sz));
2887 return fc;
2888 }
2889
2890 void CompactibleFreeListSpace:: par_get_chunk_of_blocks_dictionary(size_t word_sz, size_t targetted_number_of_chunks, AdaptiveFreeList<FreeChunk>* fl) {
2891
2892 FreeChunk* fc = get_n_way_chunk_to_split(word_sz, targetted_number_of_chunks);
2893
2894 if (fc == NULL) {
2895 return;
2896 }
2897
2898 size_t n = fc->size() / word_sz;
2899
2900 assert((ssize_t)n > 0, "Consistency");
2901 // Now do the splitting up.
2902 // Must do this in reverse order, so that anybody attempting to
2903 // access the main chunk sees it as a single free block until we
2904 // change it.
2905 size_t fc_size = n * word_sz;
2906 // All but first chunk in this loop
2907 for (ssize_t i = n-1; i > 0; i--) {
2908 FreeChunk* ffc = (FreeChunk*)((HeapWord*)fc + i * word_sz);
2909 ffc->set_size(word_sz);
2910 ffc->link_prev(NULL); // Mark as a free block for other (parallel) GC threads.
2911 ffc->link_next(NULL);
2912 // Above must occur before BOT is updated below.
2913 OrderAccess::storestore();
2914 // splitting from the right, fc_size == (n - i + 1) * wordsize
2915 _bt.mark_block((HeapWord*)ffc, word_sz, true /* reducing */);
2916 fc_size -= word_sz;
2917 _bt.verify_not_unallocated((HeapWord*)ffc, ffc->size());
2918 _bt.verify_single_block((HeapWord*)ffc, ffc->size());
2919 _bt.verify_single_block((HeapWord*)fc, fc_size);
2920 // Push this on "fl".
2921 fl->return_chunk_at_head(ffc);
2922 }
2923 // First chunk
2924 assert(fc->is_free() && fc->size() == n*word_sz, "Error: should still be a free block");
2925 // The blocks above should show their new sizes before the first block below
2926 fc->set_size(word_sz);
2927 fc->link_prev(NULL); // idempotent wrt free-ness, see assert above
2928 fc->link_next(NULL);
2929 _bt.verify_not_unallocated((HeapWord*)fc, fc->size());
2930 _bt.verify_single_block((HeapWord*)fc, fc->size());
2931 fl->return_chunk_at_head(fc);
2932
2933 assert((ssize_t)n > 0 && (ssize_t)n == fl->count(), "Incorrect number of blocks");
2934 {
2935 // Update the stats for this block size.
2936 MutexLockerEx x(_indexedFreeListParLocks[word_sz],
2937 Mutex::_no_safepoint_check_flag);
2938 const ssize_t births = _indexedFreeList[word_sz].split_births() + n;
2939 _indexedFreeList[word_sz].set_split_births(births);
2940 // ssize_t new_surplus = _indexedFreeList[word_sz].surplus() + n;
2941 // _indexedFreeList[word_sz].set_surplus(new_surplus);
2942 }
2943
2944 // TRAP
2945 assert(fl->tail()->next() == NULL, "List invariant.");
2946 }
2947
2948 void CompactibleFreeListSpace:: par_get_chunk_of_blocks(size_t word_sz, size_t n, AdaptiveFreeList<FreeChunk>* fl) {
2949 assert(fl->count() == 0, "Precondition.");
2950 assert(word_sz < CompactibleFreeListSpace::IndexSetSize,
2951 "Precondition");
2952
2953 if (par_get_chunk_of_blocks_IFL(word_sz, n, fl)) {
2954 // Got it
2955 return;
2956 }
2957
2958 // Otherwise, we'll split a block from the dictionary.
2959 par_get_chunk_of_blocks_dictionary(word_sz, n, fl);
2960 }
2961
2962 // Set up the space's par_seq_tasks structure for work claiming
2963 // for parallel rescan. See CMSParRemarkTask where this is currently used.
2964 // XXX Need to suitably abstract and generalize this and the next
2965 // method into one.
2966 void
2967 CompactibleFreeListSpace::
2968 initialize_sequential_subtasks_for_rescan(int n_threads) {
2969 // The "size" of each task is fixed according to rescan_task_size.
2970 assert(n_threads > 0, "Unexpected n_threads argument");
2971 const size_t task_size = rescan_task_size();
2972 size_t n_tasks = (used_region().word_size() + task_size - 1)/task_size;
2973 assert((n_tasks == 0) == used_region().is_empty(), "n_tasks incorrect");
2974 assert(n_tasks == 0 ||
2975 ((used_region().start() + (n_tasks - 1)*task_size < used_region().end()) &&
2976 (used_region().start() + n_tasks*task_size >= used_region().end())),
2977 "n_tasks calculation incorrect");
2978 SequentialSubTasksDone* pst = conc_par_seq_tasks();
2979 assert(!pst->valid(), "Clobbering existing data?");
2980 // Sets the condition for completion of the subtask (how many threads
2981 // need to finish in order to be done).
2982 pst->set_n_threads(n_threads);
2983 pst->set_n_tasks((int)n_tasks);
2984 }
2985
2986 // Set up the space's par_seq_tasks structure for work claiming
2987 // for parallel concurrent marking. See CMSConcMarkTask where this is currently used.
2988 void
2989 CompactibleFreeListSpace::
2990 initialize_sequential_subtasks_for_marking(int n_threads,
2991 HeapWord* low) {
2992 // The "size" of each task is fixed according to rescan_task_size.
2993 assert(n_threads > 0, "Unexpected n_threads argument");
2994 const size_t task_size = marking_task_size();
2995 assert(task_size > CardTableModRefBS::card_size_in_words &&
2996 (task_size % CardTableModRefBS::card_size_in_words == 0),
2997 "Otherwise arithmetic below would be incorrect");
2998 MemRegion span = _gen->reserved();
2999 if (low != NULL) {
3000 if (span.contains(low)) {
3001 // Align low down to a card boundary so that
3002 // we can use block_offset_careful() on span boundaries.
3003 HeapWord* aligned_low = (HeapWord*)align_size_down((uintptr_t)low,
3004 CardTableModRefBS::card_size);
3005 // Clip span prefix at aligned_low
3006 span = span.intersection(MemRegion(aligned_low, span.end()));
3007 } else if (low > span.end()) {
3008 span = MemRegion(low, low); // Null region
3009 } // else use entire span
3010 }
3011 assert(span.is_empty() ||
3012 ((uintptr_t)span.start() % CardTableModRefBS::card_size == 0),
3013 "span should start at a card boundary");
3014 size_t n_tasks = (span.word_size() + task_size - 1)/task_size;
3015 assert((n_tasks == 0) == span.is_empty(), "Inconsistency");
3016 assert(n_tasks == 0 ||
3017 ((span.start() + (n_tasks - 1)*task_size < span.end()) &&
3018 (span.start() + n_tasks*task_size >= span.end())),
3019 "n_tasks calculation incorrect");
3020 SequentialSubTasksDone* pst = conc_par_seq_tasks();
3021 assert(!pst->valid(), "Clobbering existing data?");
3022 // Sets the condition for completion of the subtask (how many threads
3023 // need to finish in order to be done).
3024 pst->set_n_threads(n_threads);
3025 pst->set_n_tasks((int)n_tasks);
3026 }