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