Mercurial > hg > graal-compiler
view src/share/vm/gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.cpp @ 2828:d6f3dbb4e3b5
merge
author | Lukas Stadler <lukas.stadler@jku.at> |
---|---|
date | Mon, 30 May 2011 18:47:33 +0200 |
parents | 1d1603768966 |
children | 1f4413413144 |
line wrap: on
line source
/* * Copyright (c) 2001, 2011, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #include "precompiled.hpp" #include "gc_implementation/concurrentMarkSweep/cmsLockVerifier.hpp" #include "gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.hpp" #include "gc_implementation/concurrentMarkSweep/concurrentMarkSweepGeneration.inline.hpp" #include "gc_implementation/concurrentMarkSweep/concurrentMarkSweepThread.hpp" #include "gc_implementation/shared/liveRange.hpp" #include "gc_implementation/shared/spaceDecorator.hpp" #include "gc_interface/collectedHeap.hpp" #include "memory/allocation.inline.hpp" #include "memory/blockOffsetTable.inline.hpp" #include "memory/resourceArea.hpp" #include "memory/universe.inline.hpp" #include "oops/oop.inline.hpp" #include "runtime/globals.hpp" #include "runtime/handles.inline.hpp" #include "runtime/init.hpp" #include "runtime/java.hpp" #include "runtime/vmThread.hpp" #include "utilities/copy.hpp" ///////////////////////////////////////////////////////////////////////// //// CompactibleFreeListSpace ///////////////////////////////////////////////////////////////////////// // highest ranked free list lock rank int CompactibleFreeListSpace::_lockRank = Mutex::leaf + 3; // Defaults are 0 so things will break badly if incorrectly initialized. int CompactibleFreeListSpace::IndexSetStart = 0; int CompactibleFreeListSpace::IndexSetStride = 0; size_t MinChunkSize = 0; void CompactibleFreeListSpace::set_cms_values() { // Set CMS global values assert(MinChunkSize == 0, "already set"); #define numQuanta(x,y) ((x+y-1)/y) MinChunkSize = numQuanta(sizeof(FreeChunk), MinObjAlignmentInBytes) * MinObjAlignment; assert(IndexSetStart == 0 && IndexSetStride == 0, "already set"); IndexSetStart = MinObjAlignment; IndexSetStride = MinObjAlignment; } // Constructor CompactibleFreeListSpace::CompactibleFreeListSpace(BlockOffsetSharedArray* bs, MemRegion mr, bool use_adaptive_freelists, FreeBlockDictionary::DictionaryChoice dictionaryChoice) : _dictionaryChoice(dictionaryChoice), _adaptive_freelists(use_adaptive_freelists), _bt(bs, mr), // free list locks are in the range of values taken by _lockRank // This range currently is [_leaf+2, _leaf+3] // Note: this requires that CFLspace c'tors // are called serially in the order in which the locks are // are acquired in the program text. This is true today. _freelistLock(_lockRank--, "CompactibleFreeListSpace._lock", true), _parDictionaryAllocLock(Mutex::leaf - 1, // == rank(ExpandHeap_lock) - 1 "CompactibleFreeListSpace._dict_par_lock", true), _rescan_task_size(CardTableModRefBS::card_size_in_words * BitsPerWord * CMSRescanMultiple), _marking_task_size(CardTableModRefBS::card_size_in_words * BitsPerWord * CMSConcMarkMultiple), _collector(NULL) { _bt.set_space(this); initialize(mr, SpaceDecorator::Clear, SpaceDecorator::Mangle); // We have all of "mr", all of which we place in the dictionary // as one big chunk. We'll need to decide here which of several // possible alternative dictionary implementations to use. For // now the choice is easy, since we have only one working // implementation, namely, the simple binary tree (splaying // temporarily disabled). switch (dictionaryChoice) { case FreeBlockDictionary::dictionarySplayTree: case FreeBlockDictionary::dictionarySkipList: default: warning("dictionaryChoice: selected option not understood; using" " default BinaryTreeDictionary implementation instead."); case FreeBlockDictionary::dictionaryBinaryTree: _dictionary = new BinaryTreeDictionary(mr); break; } assert(_dictionary != NULL, "CMS dictionary initialization"); // The indexed free lists are initially all empty and are lazily // filled in on demand. Initialize the array elements to NULL. initializeIndexedFreeListArray(); // Not using adaptive free lists assumes that allocation is first // from the linAB's. Also a cms perm gen which can be compacted // has to have the klass's klassKlass allocated at a lower // address in the heap than the klass so that the klassKlass is // moved to its new location before the klass is moved. // Set the _refillSize for the linear allocation blocks if (!use_adaptive_freelists) { FreeChunk* fc = _dictionary->getChunk(mr.word_size()); // The small linAB initially has all the space and will allocate // a chunk of any size. HeapWord* addr = (HeapWord*) fc; _smallLinearAllocBlock.set(addr, fc->size() , 1024*SmallForLinearAlloc, fc->size()); // Note that _unallocated_block is not updated here. // Allocations from the linear allocation block should // update it. } else { _smallLinearAllocBlock.set(0, 0, 1024*SmallForLinearAlloc, SmallForLinearAlloc); } // CMSIndexedFreeListReplenish should be at least 1 CMSIndexedFreeListReplenish = MAX2((uintx)1, CMSIndexedFreeListReplenish); _promoInfo.setSpace(this); if (UseCMSBestFit) { _fitStrategy = FreeBlockBestFitFirst; } else { _fitStrategy = FreeBlockStrategyNone; } checkFreeListConsistency(); // Initialize locks for parallel case. if (CollectedHeap::use_parallel_gc_threads()) { for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { _indexedFreeListParLocks[i] = new Mutex(Mutex::leaf - 1, // == ExpandHeap_lock - 1 "a freelist par lock", true); if (_indexedFreeListParLocks[i] == NULL) vm_exit_during_initialization("Could not allocate a par lock"); DEBUG_ONLY( _indexedFreeList[i].set_protecting_lock(_indexedFreeListParLocks[i]); ) } _dictionary->set_par_lock(&_parDictionaryAllocLock); } } // Like CompactibleSpace forward() but always calls cross_threshold() to // update the block offset table. Removed initialize_threshold call because // CFLS does not use a block offset array for contiguous spaces. HeapWord* CompactibleFreeListSpace::forward(oop q, size_t size, CompactPoint* cp, HeapWord* compact_top) { // q is alive // First check if we should switch compaction space assert(this == cp->space, "'this' should be current compaction space."); size_t compaction_max_size = pointer_delta(end(), compact_top); assert(adjustObjectSize(size) == cp->space->adjust_object_size_v(size), "virtual adjustObjectSize_v() method is not correct"); size_t adjusted_size = adjustObjectSize(size); assert(compaction_max_size >= MinChunkSize || compaction_max_size == 0, "no small fragments allowed"); assert(minimum_free_block_size() == MinChunkSize, "for de-virtualized reference below"); // Can't leave a nonzero size, residual fragment smaller than MinChunkSize if (adjusted_size + MinChunkSize > compaction_max_size && adjusted_size != compaction_max_size) { do { // switch to next compaction space cp->space->set_compaction_top(compact_top); cp->space = cp->space->next_compaction_space(); if (cp->space == NULL) { cp->gen = GenCollectedHeap::heap()->prev_gen(cp->gen); assert(cp->gen != NULL, "compaction must succeed"); cp->space = cp->gen->first_compaction_space(); assert(cp->space != NULL, "generation must have a first compaction space"); } compact_top = cp->space->bottom(); cp->space->set_compaction_top(compact_top); // The correct adjusted_size may not be the same as that for this method // (i.e., cp->space may no longer be "this" so adjust the size again. // Use the virtual method which is not used above to save the virtual // dispatch. adjusted_size = cp->space->adjust_object_size_v(size); compaction_max_size = pointer_delta(cp->space->end(), compact_top); assert(cp->space->minimum_free_block_size() == 0, "just checking"); } while (adjusted_size > compaction_max_size); } // store the forwarding pointer into the mark word if ((HeapWord*)q != compact_top) { q->forward_to(oop(compact_top)); assert(q->is_gc_marked(), "encoding the pointer should preserve the mark"); } else { // if the object isn't moving we can just set the mark to the default // mark and handle it specially later on. q->init_mark(); assert(q->forwardee() == NULL, "should be forwarded to NULL"); } VALIDATE_MARK_SWEEP_ONLY(MarkSweep::register_live_oop(q, adjusted_size)); compact_top += adjusted_size; // we need to update the offset table so that the beginnings of objects can be // found during scavenge. Note that we are updating the offset table based on // where the object will be once the compaction phase finishes. // Always call cross_threshold(). A contiguous space can only call it when // the compaction_top exceeds the current threshold but not for an // non-contiguous space. cp->threshold = cp->space->cross_threshold(compact_top - adjusted_size, compact_top); return compact_top; } // A modified copy of OffsetTableContigSpace::cross_threshold() with _offsets -> _bt // and use of single_block instead of alloc_block. The name here is not really // appropriate - maybe a more general name could be invented for both the // contiguous and noncontiguous spaces. HeapWord* CompactibleFreeListSpace::cross_threshold(HeapWord* start, HeapWord* the_end) { _bt.single_block(start, the_end); return end(); } // Initialize them to NULL. void CompactibleFreeListSpace::initializeIndexedFreeListArray() { for (size_t i = 0; i < IndexSetSize; i++) { // Note that on platforms where objects are double word aligned, // the odd array elements are not used. It is convenient, however, // to map directly from the object size to the array element. _indexedFreeList[i].reset(IndexSetSize); _indexedFreeList[i].set_size(i); assert(_indexedFreeList[i].count() == 0, "reset check failed"); assert(_indexedFreeList[i].head() == NULL, "reset check failed"); assert(_indexedFreeList[i].tail() == NULL, "reset check failed"); assert(_indexedFreeList[i].hint() == IndexSetSize, "reset check failed"); } } void CompactibleFreeListSpace::resetIndexedFreeListArray() { for (int i = 1; i < IndexSetSize; i++) { assert(_indexedFreeList[i].size() == (size_t) i, "Indexed free list sizes are incorrect"); _indexedFreeList[i].reset(IndexSetSize); assert(_indexedFreeList[i].count() == 0, "reset check failed"); assert(_indexedFreeList[i].head() == NULL, "reset check failed"); assert(_indexedFreeList[i].tail() == NULL, "reset check failed"); assert(_indexedFreeList[i].hint() == IndexSetSize, "reset check failed"); } } void CompactibleFreeListSpace::reset(MemRegion mr) { resetIndexedFreeListArray(); dictionary()->reset(); if (BlockOffsetArrayUseUnallocatedBlock) { assert(end() == mr.end(), "We are compacting to the bottom of CMS gen"); // Everything's allocated until proven otherwise. _bt.set_unallocated_block(end()); } if (!mr.is_empty()) { assert(mr.word_size() >= MinChunkSize, "Chunk size is too small"); _bt.single_block(mr.start(), mr.word_size()); FreeChunk* fc = (FreeChunk*) mr.start(); fc->setSize(mr.word_size()); if (mr.word_size() >= IndexSetSize ) { returnChunkToDictionary(fc); } else { _bt.verify_not_unallocated((HeapWord*)fc, fc->size()); _indexedFreeList[mr.word_size()].returnChunkAtHead(fc); } } _promoInfo.reset(); _smallLinearAllocBlock._ptr = NULL; _smallLinearAllocBlock._word_size = 0; } void CompactibleFreeListSpace::reset_after_compaction() { // Reset the space to the new reality - one free chunk. MemRegion mr(compaction_top(), end()); reset(mr); // Now refill the linear allocation block(s) if possible. if (_adaptive_freelists) { refillLinearAllocBlocksIfNeeded(); } else { // Place as much of mr in the linAB as we can get, // provided it was big enough to go into the dictionary. FreeChunk* fc = dictionary()->findLargestDict(); if (fc != NULL) { assert(fc->size() == mr.word_size(), "Why was the chunk broken up?"); removeChunkFromDictionary(fc); HeapWord* addr = (HeapWord*) fc; _smallLinearAllocBlock.set(addr, fc->size() , 1024*SmallForLinearAlloc, fc->size()); // Note that _unallocated_block is not updated here. } } } // Walks the entire dictionary, returning a coterminal // chunk, if it exists. Use with caution since it involves // a potentially complete walk of a potentially large tree. FreeChunk* CompactibleFreeListSpace::find_chunk_at_end() { assert_lock_strong(&_freelistLock); return dictionary()->find_chunk_ends_at(end()); } #ifndef PRODUCT void CompactibleFreeListSpace::initializeIndexedFreeListArrayReturnedBytes() { for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { _indexedFreeList[i].allocation_stats()->set_returnedBytes(0); } } size_t CompactibleFreeListSpace::sumIndexedFreeListArrayReturnedBytes() { size_t sum = 0; for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { sum += _indexedFreeList[i].allocation_stats()->returnedBytes(); } return sum; } size_t CompactibleFreeListSpace::totalCountInIndexedFreeLists() const { size_t count = 0; for (int i = (int)MinChunkSize; i < IndexSetSize; i++) { debug_only( ssize_t total_list_count = 0; for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL; fc = fc->next()) { total_list_count++; } assert(total_list_count == _indexedFreeList[i].count(), "Count in list is incorrect"); ) count += _indexedFreeList[i].count(); } return count; } size_t CompactibleFreeListSpace::totalCount() { size_t num = totalCountInIndexedFreeLists(); num += dictionary()->totalCount(); if (_smallLinearAllocBlock._word_size != 0) { num++; } return num; } #endif bool CompactibleFreeListSpace::is_free_block(const HeapWord* p) const { FreeChunk* fc = (FreeChunk*) p; return fc->isFree(); } size_t CompactibleFreeListSpace::used() const { return capacity() - free(); } size_t CompactibleFreeListSpace::free() const { // "MT-safe, but not MT-precise"(TM), if you will: i.e. // if you do this while the structures are in flux you // may get an approximate answer only; for instance // because there is concurrent allocation either // directly by mutators or for promotion during a GC. // It's "MT-safe", however, in the sense that you are guaranteed // not to crash and burn, for instance, because of walking // pointers that could disappear as you were walking them. // The approximation is because the various components // that are read below are not read atomically (and // further the computation of totalSizeInIndexedFreeLists() // is itself a non-atomic computation. The normal use of // this is during a resize operation at the end of GC // and at that time you are guaranteed to get the // correct actual value. However, for instance, this is // also read completely asynchronously by the "perf-sampler" // that supports jvmstat, and you are apt to see the values // flicker in such cases. assert(_dictionary != NULL, "No _dictionary?"); return (_dictionary->totalChunkSize(DEBUG_ONLY(freelistLock())) + totalSizeInIndexedFreeLists() + _smallLinearAllocBlock._word_size) * HeapWordSize; } size_t CompactibleFreeListSpace::max_alloc_in_words() const { assert(_dictionary != NULL, "No _dictionary?"); assert_locked(); size_t res = _dictionary->maxChunkSize(); res = MAX2(res, MIN2(_smallLinearAllocBlock._word_size, (size_t) SmallForLinearAlloc - 1)); // XXX the following could potentially be pretty slow; // should one, pesimally for the rare cases when res // caclulated above is less than IndexSetSize, // just return res calculated above? My reasoning was that // those cases will be so rare that the extra time spent doesn't // really matter.... // Note: do not change the loop test i >= res + IndexSetStride // to i > res below, because i is unsigned and res may be zero. for (size_t i = IndexSetSize - 1; i >= res + IndexSetStride; i -= IndexSetStride) { if (_indexedFreeList[i].head() != NULL) { assert(_indexedFreeList[i].count() != 0, "Inconsistent FreeList"); return i; } } return res; } void LinearAllocBlock::print_on(outputStream* st) const { st->print_cr(" LinearAllocBlock: ptr = " PTR_FORMAT ", word_size = " SIZE_FORMAT ", refillsize = " SIZE_FORMAT ", allocation_size_limit = " SIZE_FORMAT, _ptr, _word_size, _refillSize, _allocation_size_limit); } void CompactibleFreeListSpace::print_on(outputStream* st) const { st->print_cr("COMPACTIBLE FREELIST SPACE"); st->print_cr(" Space:"); Space::print_on(st); st->print_cr("promoInfo:"); _promoInfo.print_on(st); st->print_cr("_smallLinearAllocBlock"); _smallLinearAllocBlock.print_on(st); // dump_memory_block(_smallLinearAllocBlock->_ptr, 128); st->print_cr(" _fitStrategy = %s, _adaptive_freelists = %s", _fitStrategy?"true":"false", _adaptive_freelists?"true":"false"); } void CompactibleFreeListSpace::print_indexed_free_lists(outputStream* st) const { reportIndexedFreeListStatistics(); gclog_or_tty->print_cr("Layout of Indexed Freelists"); gclog_or_tty->print_cr("---------------------------"); FreeList::print_labels_on(st, "size"); for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { _indexedFreeList[i].print_on(gclog_or_tty); for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL; fc = fc->next()) { gclog_or_tty->print_cr("\t[" PTR_FORMAT "," PTR_FORMAT ") %s", fc, (HeapWord*)fc + i, fc->cantCoalesce() ? "\t CC" : ""); } } } void CompactibleFreeListSpace::print_promo_info_blocks(outputStream* st) const { _promoInfo.print_on(st); } void CompactibleFreeListSpace::print_dictionary_free_lists(outputStream* st) const { _dictionary->reportStatistics(); st->print_cr("Layout of Freelists in Tree"); st->print_cr("---------------------------"); _dictionary->print_free_lists(st); } class BlkPrintingClosure: public BlkClosure { const CMSCollector* _collector; const CompactibleFreeListSpace* _sp; const CMSBitMap* _live_bit_map; const bool _post_remark; outputStream* _st; public: BlkPrintingClosure(const CMSCollector* collector, const CompactibleFreeListSpace* sp, const CMSBitMap* live_bit_map, outputStream* st): _collector(collector), _sp(sp), _live_bit_map(live_bit_map), _post_remark(collector->abstract_state() > CMSCollector::FinalMarking), _st(st) { } size_t do_blk(HeapWord* addr); }; size_t BlkPrintingClosure::do_blk(HeapWord* addr) { size_t sz = _sp->block_size_no_stall(addr, _collector); assert(sz != 0, "Should always be able to compute a size"); if (_sp->block_is_obj(addr)) { const bool dead = _post_remark && !_live_bit_map->isMarked(addr); _st->print_cr(PTR_FORMAT ": %s object of size " SIZE_FORMAT "%s", addr, dead ? "dead" : "live", sz, (!dead && CMSPrintObjectsInDump) ? ":" : "."); if (CMSPrintObjectsInDump && !dead) { oop(addr)->print_on(_st); _st->print_cr("--------------------------------------"); } } else { // free block _st->print_cr(PTR_FORMAT ": free block of size " SIZE_FORMAT "%s", addr, sz, CMSPrintChunksInDump ? ":" : "."); if (CMSPrintChunksInDump) { ((FreeChunk*)addr)->print_on(_st); _st->print_cr("--------------------------------------"); } } return sz; } void CompactibleFreeListSpace::dump_at_safepoint_with_locks(CMSCollector* c, outputStream* st) { st->print_cr("\n========================="); st->print_cr("Block layout in CMS Heap:"); st->print_cr("========================="); BlkPrintingClosure bpcl(c, this, c->markBitMap(), st); blk_iterate(&bpcl); st->print_cr("\n======================================="); st->print_cr("Order & Layout of Promotion Info Blocks"); st->print_cr("======================================="); print_promo_info_blocks(st); st->print_cr("\n==========================="); st->print_cr("Order of Indexed Free Lists"); st->print_cr("========================="); print_indexed_free_lists(st); st->print_cr("\n================================="); st->print_cr("Order of Free Lists in Dictionary"); st->print_cr("================================="); print_dictionary_free_lists(st); } void CompactibleFreeListSpace::reportFreeListStatistics() const { assert_lock_strong(&_freelistLock); assert(PrintFLSStatistics != 0, "Reporting error"); _dictionary->reportStatistics(); if (PrintFLSStatistics > 1) { reportIndexedFreeListStatistics(); size_t totalSize = totalSizeInIndexedFreeLists() + _dictionary->totalChunkSize(DEBUG_ONLY(freelistLock())); gclog_or_tty->print(" free=%ld frag=%1.4f\n", totalSize, flsFrag()); } } void CompactibleFreeListSpace::reportIndexedFreeListStatistics() const { assert_lock_strong(&_freelistLock); gclog_or_tty->print("Statistics for IndexedFreeLists:\n" "--------------------------------\n"); size_t totalSize = totalSizeInIndexedFreeLists(); size_t freeBlocks = numFreeBlocksInIndexedFreeLists(); gclog_or_tty->print("Total Free Space: %d\n", totalSize); gclog_or_tty->print("Max Chunk Size: %d\n", maxChunkSizeInIndexedFreeLists()); gclog_or_tty->print("Number of Blocks: %d\n", freeBlocks); if (freeBlocks != 0) { gclog_or_tty->print("Av. Block Size: %d\n", totalSize/freeBlocks); } } size_t CompactibleFreeListSpace::numFreeBlocksInIndexedFreeLists() const { size_t res = 0; for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { debug_only( ssize_t recount = 0; for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL; fc = fc->next()) { recount += 1; } assert(recount == _indexedFreeList[i].count(), "Incorrect count in list"); ) res += _indexedFreeList[i].count(); } return res; } size_t CompactibleFreeListSpace::maxChunkSizeInIndexedFreeLists() const { for (size_t i = IndexSetSize - 1; i != 0; i -= IndexSetStride) { if (_indexedFreeList[i].head() != NULL) { assert(_indexedFreeList[i].count() != 0, "Inconsistent FreeList"); return (size_t)i; } } return 0; } void CompactibleFreeListSpace::set_end(HeapWord* value) { HeapWord* prevEnd = end(); assert(prevEnd != value, "unnecessary set_end call"); assert(prevEnd == NULL || !BlockOffsetArrayUseUnallocatedBlock || value >= unallocated_block(), "New end is below unallocated block"); _end = value; if (prevEnd != NULL) { // Resize the underlying block offset table. _bt.resize(pointer_delta(value, bottom())); if (value <= prevEnd) { assert(!BlockOffsetArrayUseUnallocatedBlock || value >= unallocated_block(), "New end is below unallocated block"); } else { // Now, take this new chunk and add it to the free blocks. // Note that the BOT has not yet been updated for this block. size_t newFcSize = pointer_delta(value, prevEnd); // XXX This is REALLY UGLY and should be fixed up. XXX if (!_adaptive_freelists && _smallLinearAllocBlock._ptr == NULL) { // Mark the boundary of the new block in BOT _bt.mark_block(prevEnd, value); // put it all in the linAB if (ParallelGCThreads == 0) { _smallLinearAllocBlock._ptr = prevEnd; _smallLinearAllocBlock._word_size = newFcSize; repairLinearAllocBlock(&_smallLinearAllocBlock); } else { // ParallelGCThreads > 0 MutexLockerEx x(parDictionaryAllocLock(), Mutex::_no_safepoint_check_flag); _smallLinearAllocBlock._ptr = prevEnd; _smallLinearAllocBlock._word_size = newFcSize; repairLinearAllocBlock(&_smallLinearAllocBlock); } // Births of chunks put into a LinAB are not recorded. Births // of chunks as they are allocated out of a LinAB are. } else { // Add the block to the free lists, if possible coalescing it // with the last free block, and update the BOT and census data. addChunkToFreeListsAtEndRecordingStats(prevEnd, newFcSize); } } } } class FreeListSpace_DCTOC : public Filtering_DCTOC { CompactibleFreeListSpace* _cfls; CMSCollector* _collector; protected: // Override. #define walk_mem_region_with_cl_DECL(ClosureType) \ virtual void walk_mem_region_with_cl(MemRegion mr, \ HeapWord* bottom, HeapWord* top, \ ClosureType* cl); \ void walk_mem_region_with_cl_par(MemRegion mr, \ HeapWord* bottom, HeapWord* top, \ ClosureType* cl); \ void walk_mem_region_with_cl_nopar(MemRegion mr, \ HeapWord* bottom, HeapWord* top, \ ClosureType* cl) walk_mem_region_with_cl_DECL(OopClosure); walk_mem_region_with_cl_DECL(FilteringClosure); public: FreeListSpace_DCTOC(CompactibleFreeListSpace* sp, CMSCollector* collector, OopClosure* cl, CardTableModRefBS::PrecisionStyle precision, HeapWord* boundary) : Filtering_DCTOC(sp, cl, precision, boundary), _cfls(sp), _collector(collector) {} }; // We de-virtualize the block-related calls below, since we know that our // space is a CompactibleFreeListSpace. #define FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(ClosureType) \ void FreeListSpace_DCTOC::walk_mem_region_with_cl(MemRegion mr, \ HeapWord* bottom, \ HeapWord* top, \ ClosureType* cl) { \ if (SharedHeap::heap()->n_par_threads() > 0) { \ walk_mem_region_with_cl_par(mr, bottom, top, cl); \ } else { \ walk_mem_region_with_cl_nopar(mr, bottom, top, cl); \ } \ } \ void FreeListSpace_DCTOC::walk_mem_region_with_cl_par(MemRegion mr, \ HeapWord* bottom, \ HeapWord* top, \ ClosureType* cl) { \ /* Skip parts that are before "mr", in case "block_start" sent us \ back too far. */ \ HeapWord* mr_start = mr.start(); \ size_t bot_size = _cfls->CompactibleFreeListSpace::block_size(bottom); \ HeapWord* next = bottom + bot_size; \ while (next < mr_start) { \ bottom = next; \ bot_size = _cfls->CompactibleFreeListSpace::block_size(bottom); \ next = bottom + bot_size; \ } \ \ while (bottom < top) { \ if (_cfls->CompactibleFreeListSpace::block_is_obj(bottom) && \ !_cfls->CompactibleFreeListSpace::obj_allocated_since_save_marks( \ oop(bottom)) && \ !_collector->CMSCollector::is_dead_obj(oop(bottom))) { \ size_t word_sz = oop(bottom)->oop_iterate(cl, mr); \ bottom += _cfls->adjustObjectSize(word_sz); \ } else { \ bottom += _cfls->CompactibleFreeListSpace::block_size(bottom); \ } \ } \ } \ void FreeListSpace_DCTOC::walk_mem_region_with_cl_nopar(MemRegion mr, \ HeapWord* bottom, \ HeapWord* top, \ ClosureType* cl) { \ /* Skip parts that are before "mr", in case "block_start" sent us \ back too far. */ \ HeapWord* mr_start = mr.start(); \ size_t bot_size = _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \ HeapWord* next = bottom + bot_size; \ while (next < mr_start) { \ bottom = next; \ bot_size = _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \ next = bottom + bot_size; \ } \ \ while (bottom < top) { \ if (_cfls->CompactibleFreeListSpace::block_is_obj_nopar(bottom) && \ !_cfls->CompactibleFreeListSpace::obj_allocated_since_save_marks( \ oop(bottom)) && \ !_collector->CMSCollector::is_dead_obj(oop(bottom))) { \ size_t word_sz = oop(bottom)->oop_iterate(cl, mr); \ bottom += _cfls->adjustObjectSize(word_sz); \ } else { \ bottom += _cfls->CompactibleFreeListSpace::block_size_nopar(bottom); \ } \ } \ } // (There are only two of these, rather than N, because the split is due // only to the introduction of the FilteringClosure, a local part of the // impl of this abstraction.) FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(OopClosure) FreeListSpace_DCTOC__walk_mem_region_with_cl_DEFN(FilteringClosure) DirtyCardToOopClosure* CompactibleFreeListSpace::new_dcto_cl(OopClosure* cl, CardTableModRefBS::PrecisionStyle precision, HeapWord* boundary) { return new FreeListSpace_DCTOC(this, _collector, cl, precision, boundary); } // Note on locking for the space iteration functions: // since the collector's iteration activities are concurrent with // allocation activities by mutators, absent a suitable mutual exclusion // mechanism the iterators may go awry. For instace a block being iterated // may suddenly be allocated or divided up and part of it allocated and // so on. // Apply the given closure to each block in the space. void CompactibleFreeListSpace::blk_iterate_careful(BlkClosureCareful* cl) { assert_lock_strong(freelistLock()); HeapWord *cur, *limit; for (cur = bottom(), limit = end(); cur < limit; cur += cl->do_blk_careful(cur)); } // Apply the given closure to each block in the space. void CompactibleFreeListSpace::blk_iterate(BlkClosure* cl) { assert_lock_strong(freelistLock()); HeapWord *cur, *limit; for (cur = bottom(), limit = end(); cur < limit; cur += cl->do_blk(cur)); } // Apply the given closure to each oop in the space. void CompactibleFreeListSpace::oop_iterate(OopClosure* cl) { assert_lock_strong(freelistLock()); HeapWord *cur, *limit; size_t curSize; for (cur = bottom(), limit = end(); cur < limit; cur += curSize) { curSize = block_size(cur); if (block_is_obj(cur)) { oop(cur)->oop_iterate(cl); } } } // Apply the given closure to each oop in the space \intersect memory region. void CompactibleFreeListSpace::oop_iterate(MemRegion mr, OopClosure* cl) { assert_lock_strong(freelistLock()); if (is_empty()) { return; } MemRegion cur = MemRegion(bottom(), end()); mr = mr.intersection(cur); if (mr.is_empty()) { return; } if (mr.equals(cur)) { oop_iterate(cl); return; } assert(mr.end() <= end(), "just took an intersection above"); HeapWord* obj_addr = block_start(mr.start()); HeapWord* t = mr.end(); SpaceMemRegionOopsIterClosure smr_blk(cl, mr); if (block_is_obj(obj_addr)) { // Handle first object specially. oop obj = oop(obj_addr); obj_addr += adjustObjectSize(obj->oop_iterate(&smr_blk)); } else { FreeChunk* fc = (FreeChunk*)obj_addr; obj_addr += fc->size(); } while (obj_addr < t) { HeapWord* obj = obj_addr; obj_addr += block_size(obj_addr); // If "obj_addr" is not greater than top, then the // entire object "obj" is within the region. if (obj_addr <= t) { if (block_is_obj(obj)) { oop(obj)->oop_iterate(cl); } } else { // "obj" extends beyond end of region if (block_is_obj(obj)) { oop(obj)->oop_iterate(&smr_blk); } break; } } } // NOTE: In the following methods, in order to safely be able to // apply the closure to an object, we need to be sure that the // object has been initialized. We are guaranteed that an object // is initialized if we are holding the Heap_lock with the // world stopped. void CompactibleFreeListSpace::verify_objects_initialized() const { if (is_init_completed()) { assert_locked_or_safepoint(Heap_lock); if (Universe::is_fully_initialized()) { guarantee(SafepointSynchronize::is_at_safepoint(), "Required for objects to be initialized"); } } // else make a concession at vm start-up } // Apply the given closure to each object in the space void CompactibleFreeListSpace::object_iterate(ObjectClosure* blk) { assert_lock_strong(freelistLock()); NOT_PRODUCT(verify_objects_initialized()); HeapWord *cur, *limit; size_t curSize; for (cur = bottom(), limit = end(); cur < limit; cur += curSize) { curSize = block_size(cur); if (block_is_obj(cur)) { blk->do_object(oop(cur)); } } } // Apply the given closure to each live object in the space // The usage of CompactibleFreeListSpace // by the ConcurrentMarkSweepGeneration for concurrent GC's allows // objects in the space with references to objects that are no longer // valid. For example, an object may reference another object // that has already been sweep up (collected). This method uses // obj_is_alive() to determine whether it is safe to apply the closure to // an object. See obj_is_alive() for details on how liveness of an // object is decided. void CompactibleFreeListSpace::safe_object_iterate(ObjectClosure* blk) { assert_lock_strong(freelistLock()); NOT_PRODUCT(verify_objects_initialized()); HeapWord *cur, *limit; size_t curSize; for (cur = bottom(), limit = end(); cur < limit; cur += curSize) { curSize = block_size(cur); if (block_is_obj(cur) && obj_is_alive(cur)) { blk->do_object(oop(cur)); } } } void CompactibleFreeListSpace::object_iterate_mem(MemRegion mr, UpwardsObjectClosure* cl) { assert_locked(freelistLock()); NOT_PRODUCT(verify_objects_initialized()); Space::object_iterate_mem(mr, cl); } // Callers of this iterator beware: The closure application should // be robust in the face of uninitialized objects and should (always) // return a correct size so that the next addr + size below gives us a // valid block boundary. [See for instance, // ScanMarkedObjectsAgainCarefullyClosure::do_object_careful() // in ConcurrentMarkSweepGeneration.cpp.] HeapWord* CompactibleFreeListSpace::object_iterate_careful(ObjectClosureCareful* cl) { assert_lock_strong(freelistLock()); HeapWord *addr, *last; size_t size; for (addr = bottom(), last = end(); addr < last; addr += size) { FreeChunk* fc = (FreeChunk*)addr; if (fc->isFree()) { // Since we hold the free list lock, which protects direct // allocation in this generation by mutators, a free object // will remain free throughout this iteration code. size = fc->size(); } else { // Note that the object need not necessarily be initialized, // because (for instance) the free list lock does NOT protect // object initialization. The closure application below must // therefore be correct in the face of uninitialized objects. size = cl->do_object_careful(oop(addr)); if (size == 0) { // An unparsable object found. Signal early termination. return addr; } } } return NULL; } // Callers of this iterator beware: The closure application should // be robust in the face of uninitialized objects and should (always) // return a correct size so that the next addr + size below gives us a // valid block boundary. [See for instance, // ScanMarkedObjectsAgainCarefullyClosure::do_object_careful() // in ConcurrentMarkSweepGeneration.cpp.] HeapWord* CompactibleFreeListSpace::object_iterate_careful_m(MemRegion mr, ObjectClosureCareful* cl) { assert_lock_strong(freelistLock()); // Can't use used_region() below because it may not necessarily // be the same as [bottom(),end()); although we could // use [used_region().start(),round_to(used_region().end(),CardSize)), // that appears too cumbersome, so we just do the simpler check // in the assertion below. assert(!mr.is_empty() && MemRegion(bottom(),end()).contains(mr), "mr should be non-empty and within used space"); HeapWord *addr, *end; size_t size; for (addr = block_start_careful(mr.start()), end = mr.end(); addr < end; addr += size) { FreeChunk* fc = (FreeChunk*)addr; if (fc->isFree()) { // Since we hold the free list lock, which protects direct // allocation in this generation by mutators, a free object // will remain free throughout this iteration code. size = fc->size(); } else { // Note that the object need not necessarily be initialized, // because (for instance) the free list lock does NOT protect // object initialization. The closure application below must // therefore be correct in the face of uninitialized objects. size = cl->do_object_careful_m(oop(addr), mr); if (size == 0) { // An unparsable object found. Signal early termination. return addr; } } } return NULL; } HeapWord* CompactibleFreeListSpace::block_start_const(const void* p) const { NOT_PRODUCT(verify_objects_initialized()); return _bt.block_start(p); } HeapWord* CompactibleFreeListSpace::block_start_careful(const void* p) const { return _bt.block_start_careful(p); } size_t CompactibleFreeListSpace::block_size(const HeapWord* p) const { NOT_PRODUCT(verify_objects_initialized()); // This must be volatile, or else there is a danger that the compiler // will compile the code below into a sometimes-infinite loop, by keeping // the value read the first time in a register. while (true) { // We must do this until we get a consistent view of the object. if (FreeChunk::indicatesFreeChunk(p)) { volatile FreeChunk* fc = (volatile FreeChunk*)p; size_t res = fc->size(); // If the object is still a free chunk, return the size, else it // has been allocated so try again. if (FreeChunk::indicatesFreeChunk(p)) { assert(res != 0, "Block size should not be 0"); return res; } } else { // must read from what 'p' points to in each loop. klassOop k = ((volatile oopDesc*)p)->klass_or_null(); if (k != NULL) { assert(k->is_oop(true /* ignore mark word */), "Should be klass oop"); oop o = (oop)p; assert(o->is_parsable(), "Should be parsable"); assert(o->is_oop(true /* ignore mark word */), "Should be an oop."); size_t res = o->size_given_klass(k->klass_part()); res = adjustObjectSize(res); assert(res != 0, "Block size should not be 0"); return res; } } } } // A variant of the above that uses the Printezis bits for // unparsable but allocated objects. This avoids any possible // stalls waiting for mutators to initialize objects, and is // thus potentially faster than the variant above. However, // this variant may return a zero size for a block that is // under mutation and for which a consistent size cannot be // inferred without stalling; see CMSCollector::block_size_if_printezis_bits(). size_t CompactibleFreeListSpace::block_size_no_stall(HeapWord* p, const CMSCollector* c) const { assert(MemRegion(bottom(), end()).contains(p), "p not in space"); // This must be volatile, or else there is a danger that the compiler // will compile the code below into a sometimes-infinite loop, by keeping // the value read the first time in a register. DEBUG_ONLY(uint loops = 0;) while (true) { // We must do this until we get a consistent view of the object. if (FreeChunk::indicatesFreeChunk(p)) { volatile FreeChunk* fc = (volatile FreeChunk*)p; size_t res = fc->size(); if (FreeChunk::indicatesFreeChunk(p)) { assert(res != 0, "Block size should not be 0"); assert(loops == 0, "Should be 0"); return res; } } else { // must read from what 'p' points to in each loop. klassOop k = ((volatile oopDesc*)p)->klass_or_null(); // We trust the size of any object that has a non-NULL // klass and (for those in the perm gen) is parsable // -- irrespective of its conc_safe-ty. if (k != NULL && ((oopDesc*)p)->is_parsable()) { assert(k->is_oop(), "Should really be klass oop."); oop o = (oop)p; assert(o->is_oop(), "Should be an oop"); size_t res = o->size_given_klass(k->klass_part()); res = adjustObjectSize(res); assert(res != 0, "Block size should not be 0"); return res; } else { // May return 0 if P-bits not present. return c->block_size_if_printezis_bits(p); } } assert(loops == 0, "Can loop at most once"); DEBUG_ONLY(loops++;) } } size_t CompactibleFreeListSpace::block_size_nopar(const HeapWord* p) const { NOT_PRODUCT(verify_objects_initialized()); assert(MemRegion(bottom(), end()).contains(p), "p not in space"); FreeChunk* fc = (FreeChunk*)p; if (fc->isFree()) { return fc->size(); } else { // Ignore mark word because this may be a recently promoted // object whose mark word is used to chain together grey // objects (the last one would have a null value). assert(oop(p)->is_oop(true), "Should be an oop"); return adjustObjectSize(oop(p)->size()); } } // This implementation assumes that the property of "being an object" is // stable. But being a free chunk may not be (because of parallel // promotion.) bool CompactibleFreeListSpace::block_is_obj(const HeapWord* p) const { FreeChunk* fc = (FreeChunk*)p; assert(is_in_reserved(p), "Should be in space"); // When doing a mark-sweep-compact of the CMS generation, this // assertion may fail because prepare_for_compaction() uses // space that is garbage to maintain information on ranges of // live objects so that these live ranges can be moved as a whole. // Comment out this assertion until that problem can be solved // (i.e., that the block start calculation may look at objects // at address below "p" in finding the object that contains "p" // and those objects (if garbage) may have been modified to hold // live range information. // assert(CollectedHeap::use_parallel_gc_threads() || _bt.block_start(p) == p, // "Should be a block boundary"); if (FreeChunk::indicatesFreeChunk(p)) return false; klassOop k = oop(p)->klass_or_null(); if (k != NULL) { // Ignore mark word because it may have been used to // chain together promoted objects (the last one // would have a null value). assert(oop(p)->is_oop(true), "Should be an oop"); return true; } else { return false; // Was not an object at the start of collection. } } // Check if the object is alive. This fact is checked either by consulting // the main marking bitmap in the sweeping phase or, if it's a permanent // generation and we're not in the sweeping phase, by checking the // perm_gen_verify_bit_map where we store the "deadness" information if // we did not sweep the perm gen in the most recent previous GC cycle. bool CompactibleFreeListSpace::obj_is_alive(const HeapWord* p) const { assert(SafepointSynchronize::is_at_safepoint() || !is_init_completed(), "Else races are possible"); assert(block_is_obj(p), "The address should point to an object"); // If we're sweeping, we use object liveness information from the main bit map // for both perm gen and old gen. // We don't need to lock the bitmap (live_map or dead_map below), because // EITHER we are in the middle of the sweeping phase, and the // main marking bit map (live_map below) is locked, // OR we're in other phases and perm_gen_verify_bit_map (dead_map below) // is stable, because it's mutated only in the sweeping phase. // NOTE: This method is also used by jmap where, if class unloading is // off, the results can return "false" for legitimate perm objects, // when we are not in the midst of a sweeping phase, which can result // in jmap not reporting certain perm gen objects. This will be moot // if/when the perm gen goes away in the future. if (_collector->abstract_state() == CMSCollector::Sweeping) { CMSBitMap* live_map = _collector->markBitMap(); return live_map->par_isMarked((HeapWord*) p); } else { // If we're not currently sweeping and we haven't swept the perm gen in // the previous concurrent cycle then we may have dead but unswept objects // in the perm gen. In this case, we use the "deadness" information // that we had saved in perm_gen_verify_bit_map at the last sweep. if (!CMSClassUnloadingEnabled && _collector->_permGen->reserved().contains(p)) { if (_collector->verifying()) { CMSBitMap* dead_map = _collector->perm_gen_verify_bit_map(); // Object is marked in the dead_map bitmap at the previous sweep // when we know that it's dead; if the bitmap is not allocated then // the object is alive. return (dead_map->sizeInBits() == 0) // bit_map has been allocated || !dead_map->par_isMarked((HeapWord*) p); } else { return false; // We can't say for sure if it's live, so we say that it's dead. } } } return true; } bool CompactibleFreeListSpace::block_is_obj_nopar(const HeapWord* p) const { FreeChunk* fc = (FreeChunk*)p; assert(is_in_reserved(p), "Should be in space"); assert(_bt.block_start(p) == p, "Should be a block boundary"); if (!fc->isFree()) { // Ignore mark word because it may have been used to // chain together promoted objects (the last one // would have a null value). assert(oop(p)->is_oop(true), "Should be an oop"); return true; } return false; } // "MT-safe but not guaranteed MT-precise" (TM); you may get an // approximate answer if you don't hold the freelistlock when you call this. size_t CompactibleFreeListSpace::totalSizeInIndexedFreeLists() const { size_t size = 0; for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { debug_only( // We may be calling here without the lock in which case we // won't do this modest sanity check. if (freelistLock()->owned_by_self()) { size_t total_list_size = 0; for (FreeChunk* fc = _indexedFreeList[i].head(); fc != NULL; fc = fc->next()) { total_list_size += i; } assert(total_list_size == i * _indexedFreeList[i].count(), "Count in list is incorrect"); } ) size += i * _indexedFreeList[i].count(); } return size; } HeapWord* CompactibleFreeListSpace::par_allocate(size_t size) { MutexLockerEx x(freelistLock(), Mutex::_no_safepoint_check_flag); return allocate(size); } HeapWord* CompactibleFreeListSpace::getChunkFromSmallLinearAllocBlockRemainder(size_t size) { return getChunkFromLinearAllocBlockRemainder(&_smallLinearAllocBlock, size); } HeapWord* CompactibleFreeListSpace::allocate(size_t size) { assert_lock_strong(freelistLock()); HeapWord* res = NULL; assert(size == adjustObjectSize(size), "use adjustObjectSize() before calling into allocate()"); if (_adaptive_freelists) { res = allocate_adaptive_freelists(size); } else { // non-adaptive free lists res = allocate_non_adaptive_freelists(size); } if (res != NULL) { // check that res does lie in this space! assert(is_in_reserved(res), "Not in this space!"); assert(is_aligned((void*)res), "alignment check"); FreeChunk* fc = (FreeChunk*)res; fc->markNotFree(); assert(!fc->isFree(), "shouldn't be marked free"); assert(oop(fc)->klass_or_null() == NULL, "should look uninitialized"); // Verify that the block offset table shows this to // be a single block, but not one which is unallocated. _bt.verify_single_block(res, size); _bt.verify_not_unallocated(res, size); // mangle a just allocated object with a distinct pattern. debug_only(fc->mangleAllocated(size)); } return res; } HeapWord* CompactibleFreeListSpace::allocate_non_adaptive_freelists(size_t size) { HeapWord* res = NULL; // try and use linear allocation for smaller blocks if (size < _smallLinearAllocBlock._allocation_size_limit) { // if successful, the following also adjusts block offset table res = getChunkFromSmallLinearAllocBlock(size); } // Else triage to indexed lists for smaller sizes if (res == NULL) { if (size < SmallForDictionary) { res = (HeapWord*) getChunkFromIndexedFreeList(size); } else { // else get it from the big dictionary; if even this doesn't // work we are out of luck. res = (HeapWord*)getChunkFromDictionaryExact(size); } } return res; } HeapWord* CompactibleFreeListSpace::allocate_adaptive_freelists(size_t size) { assert_lock_strong(freelistLock()); HeapWord* res = NULL; assert(size == adjustObjectSize(size), "use adjustObjectSize() before calling into allocate()"); // Strategy // if small // exact size from small object indexed list if small // small or large linear allocation block (linAB) as appropriate // take from lists of greater sized chunks // else // dictionary // small or large linear allocation block if it has the space // Try allocating exact size from indexTable first if (size < IndexSetSize) { res = (HeapWord*) getChunkFromIndexedFreeList(size); if(res != NULL) { assert(res != (HeapWord*)_indexedFreeList[size].head(), "Not removed from free list"); // no block offset table adjustment is necessary on blocks in // the indexed lists. // Try allocating from the small LinAB } else if (size < _smallLinearAllocBlock._allocation_size_limit && (res = getChunkFromSmallLinearAllocBlock(size)) != NULL) { // if successful, the above also adjusts block offset table // Note that this call will refill the LinAB to // satisfy the request. This is different that // evm. // Don't record chunk off a LinAB? smallSplitBirth(size); } else { // Raid the exact free lists larger than size, even if they are not // overpopulated. res = (HeapWord*) getChunkFromGreater(size); } } else { // Big objects get allocated directly from the dictionary. res = (HeapWord*) getChunkFromDictionaryExact(size); if (res == NULL) { // Try hard not to fail since an allocation failure will likely // trigger a synchronous GC. Try to get the space from the // allocation blocks. res = getChunkFromSmallLinearAllocBlockRemainder(size); } } return res; } // A worst-case estimate of the space required (in HeapWords) to expand the heap // when promoting obj. size_t CompactibleFreeListSpace::expansionSpaceRequired(size_t obj_size) const { // Depending on the object size, expansion may require refilling either a // bigLAB or a smallLAB plus refilling a PromotionInfo object. MinChunkSize // is added because the dictionary may over-allocate to avoid fragmentation. size_t space = obj_size; if (!_adaptive_freelists) { space = MAX2(space, _smallLinearAllocBlock._refillSize); } space += _promoInfo.refillSize() + 2 * MinChunkSize; return space; } FreeChunk* CompactibleFreeListSpace::getChunkFromGreater(size_t numWords) { FreeChunk* ret; assert(numWords >= MinChunkSize, "Size is less than minimum"); assert(linearAllocationWouldFail() || bestFitFirst(), "Should not be here"); size_t i; size_t currSize = numWords + MinChunkSize; assert(currSize % MinObjAlignment == 0, "currSize should be aligned"); for (i = currSize; i < IndexSetSize; i += IndexSetStride) { FreeList* fl = &_indexedFreeList[i]; if (fl->head()) { ret = getFromListGreater(fl, numWords); assert(ret == NULL || ret->isFree(), "Should be returning a free chunk"); return ret; } } currSize = MAX2((size_t)SmallForDictionary, (size_t)(numWords + MinChunkSize)); /* Try to get a chunk that satisfies request, while avoiding fragmentation that can't be handled. */ { ret = dictionary()->getChunk(currSize); if (ret != NULL) { assert(ret->size() - numWords >= MinChunkSize, "Chunk is too small"); _bt.allocated((HeapWord*)ret, ret->size()); /* Carve returned chunk. */ (void) splitChunkAndReturnRemainder(ret, numWords); /* Label this as no longer a free chunk. */ assert(ret->isFree(), "This chunk should be free"); ret->linkPrev(NULL); } assert(ret == NULL || ret->isFree(), "Should be returning a free chunk"); return ret; } ShouldNotReachHere(); } bool CompactibleFreeListSpace::verifyChunkInIndexedFreeLists(FreeChunk* fc) const { assert(fc->size() < IndexSetSize, "Size of chunk is too large"); return _indexedFreeList[fc->size()].verifyChunkInFreeLists(fc); } bool CompactibleFreeListSpace::verifyChunkInFreeLists(FreeChunk* fc) const { if (fc->size() >= IndexSetSize) { return dictionary()->verifyChunkInFreeLists(fc); } else { return verifyChunkInIndexedFreeLists(fc); } } #ifndef PRODUCT void CompactibleFreeListSpace::assert_locked() const { CMSLockVerifier::assert_locked(freelistLock(), parDictionaryAllocLock()); } void CompactibleFreeListSpace::assert_locked(const Mutex* lock) const { CMSLockVerifier::assert_locked(lock); } #endif FreeChunk* CompactibleFreeListSpace::allocateScratch(size_t size) { // In the parallel case, the main thread holds the free list lock // on behalf the parallel threads. FreeChunk* fc; { // If GC is parallel, this might be called by several threads. // This should be rare enough that the locking overhead won't affect // the sequential code. MutexLockerEx x(parDictionaryAllocLock(), Mutex::_no_safepoint_check_flag); fc = getChunkFromDictionary(size); } if (fc != NULL) { fc->dontCoalesce(); assert(fc->isFree(), "Should be free, but not coalescable"); // Verify that the block offset table shows this to // be a single block, but not one which is unallocated. _bt.verify_single_block((HeapWord*)fc, fc->size()); _bt.verify_not_unallocated((HeapWord*)fc, fc->size()); } return fc; } oop CompactibleFreeListSpace::promote(oop obj, size_t obj_size) { assert(obj_size == (size_t)obj->size(), "bad obj_size passed in"); assert_locked(); // if we are tracking promotions, then first ensure space for // promotion (including spooling space for saving header if necessary). // then allocate and copy, then track promoted info if needed. // When tracking (see PromotionInfo::track()), the mark word may // be displaced and in this case restoration of the mark word // occurs in the (oop_since_save_marks_)iterate phase. if (_promoInfo.tracking() && !_promoInfo.ensure_spooling_space()) { return NULL; } // Call the allocate(size_t, bool) form directly to avoid the // additional call through the allocate(size_t) form. Having // the compile inline the call is problematic because allocate(size_t) // is a virtual method. HeapWord* res = allocate(adjustObjectSize(obj_size)); if (res != NULL) { Copy::aligned_disjoint_words((HeapWord*)obj, res, obj_size); // if we should be tracking promotions, do so. if (_promoInfo.tracking()) { _promoInfo.track((PromotedObject*)res); } } return oop(res); } HeapWord* CompactibleFreeListSpace::getChunkFromSmallLinearAllocBlock(size_t size) { assert_locked(); assert(size >= MinChunkSize, "minimum chunk size"); assert(size < _smallLinearAllocBlock._allocation_size_limit, "maximum from smallLinearAllocBlock"); return getChunkFromLinearAllocBlock(&_smallLinearAllocBlock, size); } HeapWord* CompactibleFreeListSpace::getChunkFromLinearAllocBlock(LinearAllocBlock *blk, size_t size) { assert_locked(); assert(size >= MinChunkSize, "too small"); HeapWord* res = NULL; // Try to do linear allocation from blk, making sure that if (blk->_word_size == 0) { // We have probably been unable to fill this either in the prologue or // when it was exhausted at the last linear allocation. Bail out until // next time. assert(blk->_ptr == NULL, "consistency check"); return NULL; } assert(blk->_word_size != 0 && blk->_ptr != NULL, "consistency check"); res = getChunkFromLinearAllocBlockRemainder(blk, size); if (res != NULL) return res; // about to exhaust this linear allocation block if (blk->_word_size == size) { // exactly satisfied res = blk->_ptr; _bt.allocated(res, blk->_word_size); } else if (size + MinChunkSize <= blk->_refillSize) { size_t sz = blk->_word_size; // Update _unallocated_block if the size is such that chunk would be // returned to the indexed free list. All other chunks in the indexed // free lists are allocated from the dictionary so that _unallocated_block // has already been adjusted for them. Do it here so that the cost // for all chunks added back to the indexed free lists. if (sz < SmallForDictionary) { _bt.allocated(blk->_ptr, sz); } // Return the chunk that isn't big enough, and then refill below. addChunkToFreeLists(blk->_ptr, sz); splitBirth(sz); // Don't keep statistics on adding back chunk from a LinAB. } else { // A refilled block would not satisfy the request. return NULL; } blk->_ptr = NULL; blk->_word_size = 0; refillLinearAllocBlock(blk); assert(blk->_ptr == NULL || blk->_word_size >= size + MinChunkSize, "block was replenished"); if (res != NULL) { splitBirth(size); repairLinearAllocBlock(blk); } else if (blk->_ptr != NULL) { res = blk->_ptr; size_t blk_size = blk->_word_size; blk->_word_size -= size; blk->_ptr += size; splitBirth(size); repairLinearAllocBlock(blk); // Update BOT last so that other (parallel) GC threads see a consistent // view of the BOT and free blocks. // Above must occur before BOT is updated below. OrderAccess::storestore(); _bt.split_block(res, blk_size, size); // adjust block offset table } return res; } HeapWord* CompactibleFreeListSpace::getChunkFromLinearAllocBlockRemainder( LinearAllocBlock* blk, size_t size) { assert_locked(); assert(size >= MinChunkSize, "too small"); HeapWord* res = NULL; // This is the common case. Keep it simple. if (blk->_word_size >= size + MinChunkSize) { assert(blk->_ptr != NULL, "consistency check"); res = blk->_ptr; // Note that the BOT is up-to-date for the linAB before allocation. It // indicates the start of the linAB. The split_block() updates the // BOT for the linAB after the allocation (indicates the start of the // next chunk to be allocated). size_t blk_size = blk->_word_size; blk->_word_size -= size; blk->_ptr += size; splitBirth(size); repairLinearAllocBlock(blk); // Update BOT last so that other (parallel) GC threads see a consistent // view of the BOT and free blocks. // Above must occur before BOT is updated below. OrderAccess::storestore(); _bt.split_block(res, blk_size, size); // adjust block offset table _bt.allocated(res, size); } return res; } FreeChunk* CompactibleFreeListSpace::getChunkFromIndexedFreeList(size_t size) { assert_locked(); assert(size < SmallForDictionary, "just checking"); FreeChunk* res; res = _indexedFreeList[size].getChunkAtHead(); if (res == NULL) { res = getChunkFromIndexedFreeListHelper(size); } _bt.verify_not_unallocated((HeapWord*) res, size); assert(res == NULL || res->size() == size, "Incorrect block size"); return res; } FreeChunk* CompactibleFreeListSpace::getChunkFromIndexedFreeListHelper(size_t size, bool replenish) { assert_locked(); FreeChunk* fc = NULL; if (size < SmallForDictionary) { assert(_indexedFreeList[size].head() == NULL || _indexedFreeList[size].surplus() <= 0, "List for this size should be empty or under populated"); // Try best fit in exact lists before replenishing the list if (!bestFitFirst() || (fc = bestFitSmall(size)) == NULL) { // Replenish list. // // Things tried that failed. // Tried allocating out of the two LinAB's first before // replenishing lists. // Tried small linAB of size 256 (size in indexed list) // and replenishing indexed lists from the small linAB. // FreeChunk* newFc = NULL; const size_t replenish_size = CMSIndexedFreeListReplenish * size; if (replenish_size < SmallForDictionary) { // Do not replenish from an underpopulated size. if (_indexedFreeList[replenish_size].surplus() > 0 && _indexedFreeList[replenish_size].head() != NULL) { newFc = _indexedFreeList[replenish_size].getChunkAtHead(); } else if (bestFitFirst()) { newFc = bestFitSmall(replenish_size); } } if (newFc == NULL && replenish_size > size) { assert(CMSIndexedFreeListReplenish > 1, "ctl pt invariant"); newFc = getChunkFromIndexedFreeListHelper(replenish_size, false); } // Note: The stats update re split-death of block obtained above // will be recorded below precisely when we know we are going to // be actually splitting it into more than one pieces below. if (newFc != NULL) { if (replenish || CMSReplenishIntermediate) { // Replenish this list and return one block to caller. size_t i; FreeChunk *curFc, *nextFc; size_t num_blk = newFc->size() / size; assert(num_blk >= 1, "Smaller than requested?"); assert(newFc->size() % size == 0, "Should be integral multiple of request"); if (num_blk > 1) { // we are sure we will be splitting the block just obtained // into multiple pieces; record the split-death of the original splitDeath(replenish_size); } // carve up and link blocks 0, ..., num_blk - 2 // The last chunk is not added to the lists but is returned as the // free chunk. for (curFc = newFc, nextFc = (FreeChunk*)((HeapWord*)curFc + size), i = 0; i < (num_blk - 1); curFc = nextFc, nextFc = (FreeChunk*)((HeapWord*)nextFc + size), i++) { curFc->setSize(size); // Don't record this as a return in order to try and // determine the "returns" from a GC. _bt.verify_not_unallocated((HeapWord*) fc, size); _indexedFreeList[size].returnChunkAtTail(curFc, false); _bt.mark_block((HeapWord*)curFc, size); splitBirth(size); // Don't record the initial population of the indexed list // as a split birth. } // check that the arithmetic was OK above assert((HeapWord*)nextFc == (HeapWord*)newFc + num_blk*size, "inconsistency in carving newFc"); curFc->setSize(size); _bt.mark_block((HeapWord*)curFc, size); splitBirth(size); fc = curFc; } else { // Return entire block to caller fc = newFc; } } } } else { // Get a free chunk from the free chunk dictionary to be returned to // replenish the indexed free list. fc = getChunkFromDictionaryExact(size); } // assert(fc == NULL || fc->isFree(), "Should be returning a free chunk"); return fc; } FreeChunk* CompactibleFreeListSpace::getChunkFromDictionary(size_t size) { assert_locked(); FreeChunk* fc = _dictionary->getChunk(size); if (fc == NULL) { return NULL; } _bt.allocated((HeapWord*)fc, fc->size()); if (fc->size() >= size + MinChunkSize) { fc = splitChunkAndReturnRemainder(fc, size); } assert(fc->size() >= size, "chunk too small"); assert(fc->size() < size + MinChunkSize, "chunk too big"); _bt.verify_single_block((HeapWord*)fc, fc->size()); return fc; } FreeChunk* CompactibleFreeListSpace::getChunkFromDictionaryExact(size_t size) { assert_locked(); FreeChunk* fc = _dictionary->getChunk(size); if (fc == NULL) { return fc; } _bt.allocated((HeapWord*)fc, fc->size()); if (fc->size() == size) { _bt.verify_single_block((HeapWord*)fc, size); return fc; } assert(fc->size() > size, "getChunk() guarantee"); if (fc->size() < size + MinChunkSize) { // Return the chunk to the dictionary and go get a bigger one. returnChunkToDictionary(fc); fc = _dictionary->getChunk(size + MinChunkSize); if (fc == NULL) { return NULL; } _bt.allocated((HeapWord*)fc, fc->size()); } assert(fc->size() >= size + MinChunkSize, "tautology"); fc = splitChunkAndReturnRemainder(fc, size); assert(fc->size() == size, "chunk is wrong size"); _bt.verify_single_block((HeapWord*)fc, size); return fc; } void CompactibleFreeListSpace::returnChunkToDictionary(FreeChunk* chunk) { assert_locked(); size_t size = chunk->size(); _bt.verify_single_block((HeapWord*)chunk, size); // adjust _unallocated_block downward, as necessary _bt.freed((HeapWord*)chunk, size); _dictionary->returnChunk(chunk); #ifndef PRODUCT if (CMSCollector::abstract_state() != CMSCollector::Sweeping) { TreeChunk::as_TreeChunk(chunk)->list()->verify_stats(); } #endif // PRODUCT } void CompactibleFreeListSpace::returnChunkToFreeList(FreeChunk* fc) { assert_locked(); size_t size = fc->size(); _bt.verify_single_block((HeapWord*) fc, size); _bt.verify_not_unallocated((HeapWord*) fc, size); if (_adaptive_freelists) { _indexedFreeList[size].returnChunkAtTail(fc); } else { _indexedFreeList[size].returnChunkAtHead(fc); } #ifndef PRODUCT if (CMSCollector::abstract_state() != CMSCollector::Sweeping) { _indexedFreeList[size].verify_stats(); } #endif // PRODUCT } // Add chunk to end of last block -- if it's the largest // block -- and update BOT and census data. We would // of course have preferred to coalesce it with the // last block, but it's currently less expensive to find the // largest block than it is to find the last. void CompactibleFreeListSpace::addChunkToFreeListsAtEndRecordingStats( HeapWord* chunk, size_t size) { // check that the chunk does lie in this space! assert(chunk != NULL && is_in_reserved(chunk), "Not in this space!"); // One of the parallel gc task threads may be here // whilst others are allocating. Mutex* lock = NULL; if (ParallelGCThreads != 0) { lock = &_parDictionaryAllocLock; } FreeChunk* ec; { MutexLockerEx x(lock, Mutex::_no_safepoint_check_flag); ec = dictionary()->findLargestDict(); // get largest block if (ec != NULL && ec->end() == chunk) { // It's a coterminal block - we can coalesce. size_t old_size = ec->size(); coalDeath(old_size); removeChunkFromDictionary(ec); size += old_size; } else { ec = (FreeChunk*)chunk; } } ec->setSize(size); debug_only(ec->mangleFreed(size)); if (size < SmallForDictionary) { lock = _indexedFreeListParLocks[size]; } MutexLockerEx x(lock, Mutex::_no_safepoint_check_flag); addChunkAndRepairOffsetTable((HeapWord*)ec, size, true); // record the birth under the lock since the recording involves // manipulation of the list on which the chunk lives and // if the chunk is allocated and is the last on the list, // the list can go away. coalBirth(size); } void CompactibleFreeListSpace::addChunkToFreeLists(HeapWord* chunk, size_t size) { // check that the chunk does lie in this space! assert(chunk != NULL && is_in_reserved(chunk), "Not in this space!"); assert_locked(); _bt.verify_single_block(chunk, size); FreeChunk* fc = (FreeChunk*) chunk; fc->setSize(size); debug_only(fc->mangleFreed(size)); if (size < SmallForDictionary) { returnChunkToFreeList(fc); } else { returnChunkToDictionary(fc); } } void CompactibleFreeListSpace::addChunkAndRepairOffsetTable(HeapWord* chunk, size_t size, bool coalesced) { assert_locked(); assert(chunk != NULL, "null chunk"); if (coalesced) { // repair BOT _bt.single_block(chunk, size); } addChunkToFreeLists(chunk, size); } // We _must_ find the purported chunk on our free lists; // we assert if we don't. void CompactibleFreeListSpace::removeFreeChunkFromFreeLists(FreeChunk* fc) { size_t size = fc->size(); assert_locked(); debug_only(verifyFreeLists()); if (size < SmallForDictionary) { removeChunkFromIndexedFreeList(fc); } else { removeChunkFromDictionary(fc); } _bt.verify_single_block((HeapWord*)fc, size); debug_only(verifyFreeLists()); } void CompactibleFreeListSpace::removeChunkFromDictionary(FreeChunk* fc) { size_t size = fc->size(); assert_locked(); assert(fc != NULL, "null chunk"); _bt.verify_single_block((HeapWord*)fc, size); _dictionary->removeChunk(fc); // adjust _unallocated_block upward, as necessary _bt.allocated((HeapWord*)fc, size); } void CompactibleFreeListSpace::removeChunkFromIndexedFreeList(FreeChunk* fc) { assert_locked(); size_t size = fc->size(); _bt.verify_single_block((HeapWord*)fc, size); NOT_PRODUCT( if (FLSVerifyIndexTable) { verifyIndexedFreeList(size); } ) _indexedFreeList[size].removeChunk(fc); debug_only(fc->clearNext()); debug_only(fc->clearPrev()); NOT_PRODUCT( if (FLSVerifyIndexTable) { verifyIndexedFreeList(size); } ) } FreeChunk* CompactibleFreeListSpace::bestFitSmall(size_t numWords) { /* A hint is the next larger size that has a surplus. Start search at a size large enough to guarantee that the excess is >= MIN_CHUNK. */ size_t start = align_object_size(numWords + MinChunkSize); if (start < IndexSetSize) { FreeList* it = _indexedFreeList; size_t hint = _indexedFreeList[start].hint(); while (hint < IndexSetSize) { assert(hint % MinObjAlignment == 0, "hint should be aligned"); FreeList *fl = &_indexedFreeList[hint]; if (fl->surplus() > 0 && fl->head() != NULL) { // Found a list with surplus, reset original hint // and split out a free chunk which is returned. _indexedFreeList[start].set_hint(hint); FreeChunk* res = getFromListGreater(fl, numWords); assert(res == NULL || res->isFree(), "Should be returning a free chunk"); return res; } hint = fl->hint(); /* keep looking */ } /* None found. */ it[start].set_hint(IndexSetSize); } return NULL; } /* Requires fl->size >= numWords + MinChunkSize */ FreeChunk* CompactibleFreeListSpace::getFromListGreater(FreeList* fl, size_t numWords) { FreeChunk *curr = fl->head(); size_t oldNumWords = curr->size(); assert(numWords >= MinChunkSize, "Word size is too small"); assert(curr != NULL, "List is empty"); assert(oldNumWords >= numWords + MinChunkSize, "Size of chunks in the list is too small"); fl->removeChunk(curr); // recorded indirectly by splitChunkAndReturnRemainder - // smallSplit(oldNumWords, numWords); FreeChunk* new_chunk = splitChunkAndReturnRemainder(curr, numWords); // Does anything have to be done for the remainder in terms of // fixing the card table? assert(new_chunk == NULL || new_chunk->isFree(), "Should be returning a free chunk"); return new_chunk; } FreeChunk* CompactibleFreeListSpace::splitChunkAndReturnRemainder(FreeChunk* chunk, size_t new_size) { assert_locked(); size_t size = chunk->size(); assert(size > new_size, "Split from a smaller block?"); assert(is_aligned(chunk), "alignment problem"); assert(size == adjustObjectSize(size), "alignment problem"); size_t rem_size = size - new_size; assert(rem_size == adjustObjectSize(rem_size), "alignment problem"); assert(rem_size >= MinChunkSize, "Free chunk smaller than minimum"); FreeChunk* ffc = (FreeChunk*)((HeapWord*)chunk + new_size); assert(is_aligned(ffc), "alignment problem"); ffc->setSize(rem_size); ffc->linkNext(NULL); ffc->linkPrev(NULL); // Mark as a free block for other (parallel) GC threads. // Above must occur before BOT is updated below. // adjust block offset table OrderAccess::storestore(); assert(chunk->isFree() && ffc->isFree(), "Error"); _bt.split_block((HeapWord*)chunk, chunk->size(), new_size); if (rem_size < SmallForDictionary) { bool is_par = (SharedHeap::heap()->n_par_threads() > 0); if (is_par) _indexedFreeListParLocks[rem_size]->lock(); returnChunkToFreeList(ffc); split(size, rem_size); if (is_par) _indexedFreeListParLocks[rem_size]->unlock(); } else { returnChunkToDictionary(ffc); split(size ,rem_size); } chunk->setSize(new_size); return chunk; } void CompactibleFreeListSpace::sweep_completed() { // Now that space is probably plentiful, refill linear // allocation blocks as needed. refillLinearAllocBlocksIfNeeded(); } void CompactibleFreeListSpace::gc_prologue() { assert_locked(); if (PrintFLSStatistics != 0) { gclog_or_tty->print("Before GC:\n"); reportFreeListStatistics(); } refillLinearAllocBlocksIfNeeded(); } void CompactibleFreeListSpace::gc_epilogue() { assert_locked(); if (PrintGCDetails && Verbose && !_adaptive_freelists) { if (_smallLinearAllocBlock._word_size == 0) warning("CompactibleFreeListSpace(epilogue):: Linear allocation failure"); } assert(_promoInfo.noPromotions(), "_promoInfo inconsistency"); _promoInfo.stopTrackingPromotions(); repairLinearAllocationBlocks(); // Print Space's stats if (PrintFLSStatistics != 0) { gclog_or_tty->print("After GC:\n"); reportFreeListStatistics(); } } // Iteration support, mostly delegated from a CMS generation void CompactibleFreeListSpace::save_marks() { // mark the "end" of the used space at the time of this call; // note, however, that promoted objects from this point // on are tracked in the _promoInfo below. set_saved_mark_word(unallocated_block()); // inform allocator that promotions should be tracked. assert(_promoInfo.noPromotions(), "_promoInfo inconsistency"); _promoInfo.startTrackingPromotions(); } bool CompactibleFreeListSpace::no_allocs_since_save_marks() { assert(_promoInfo.tracking(), "No preceding save_marks?"); assert(SharedHeap::heap()->n_par_threads() == 0, "Shouldn't be called if using parallel gc."); return _promoInfo.noPromotions(); } #define CFLS_OOP_SINCE_SAVE_MARKS_DEFN(OopClosureType, nv_suffix) \ \ void CompactibleFreeListSpace:: \ oop_since_save_marks_iterate##nv_suffix(OopClosureType* blk) { \ assert(SharedHeap::heap()->n_par_threads() == 0, \ "Shouldn't be called (yet) during parallel part of gc."); \ _promoInfo.promoted_oops_iterate##nv_suffix(blk); \ /* \ * This also restores any displaced headers and removes the elements from \ * the iteration set as they are processed, so that we have a clean slate \ * at the end of the iteration. Note, thus, that if new objects are \ * promoted as a result of the iteration they are iterated over as well. \ */ \ assert(_promoInfo.noPromotions(), "_promoInfo inconsistency"); \ } ALL_SINCE_SAVE_MARKS_CLOSURES(CFLS_OOP_SINCE_SAVE_MARKS_DEFN) void CompactibleFreeListSpace::object_iterate_since_last_GC(ObjectClosure* cl) { // ugghh... how would one do this efficiently for a non-contiguous space? guarantee(false, "NYI"); } bool CompactibleFreeListSpace::linearAllocationWouldFail() const { return _smallLinearAllocBlock._word_size == 0; } void CompactibleFreeListSpace::repairLinearAllocationBlocks() { // Fix up linear allocation blocks to look like free blocks repairLinearAllocBlock(&_smallLinearAllocBlock); } void CompactibleFreeListSpace::repairLinearAllocBlock(LinearAllocBlock* blk) { assert_locked(); if (blk->_ptr != NULL) { assert(blk->_word_size != 0 && blk->_word_size >= MinChunkSize, "Minimum block size requirement"); FreeChunk* fc = (FreeChunk*)(blk->_ptr); fc->setSize(blk->_word_size); fc->linkPrev(NULL); // mark as free fc->dontCoalesce(); assert(fc->isFree(), "just marked it free"); assert(fc->cantCoalesce(), "just marked it uncoalescable"); } } void CompactibleFreeListSpace::refillLinearAllocBlocksIfNeeded() { assert_locked(); if (_smallLinearAllocBlock._ptr == NULL) { assert(_smallLinearAllocBlock._word_size == 0, "Size of linAB should be zero if the ptr is NULL"); // Reset the linAB refill and allocation size limit. _smallLinearAllocBlock.set(0, 0, 1024*SmallForLinearAlloc, SmallForLinearAlloc); } refillLinearAllocBlockIfNeeded(&_smallLinearAllocBlock); } void CompactibleFreeListSpace::refillLinearAllocBlockIfNeeded(LinearAllocBlock* blk) { assert_locked(); assert((blk->_ptr == NULL && blk->_word_size == 0) || (blk->_ptr != NULL && blk->_word_size >= MinChunkSize), "blk invariant"); if (blk->_ptr == NULL) { refillLinearAllocBlock(blk); } if (PrintMiscellaneous && Verbose) { if (blk->_word_size == 0) { warning("CompactibleFreeListSpace(prologue):: Linear allocation failure"); } } } void CompactibleFreeListSpace::refillLinearAllocBlock(LinearAllocBlock* blk) { assert_locked(); assert(blk->_word_size == 0 && blk->_ptr == NULL, "linear allocation block should be empty"); FreeChunk* fc; if (blk->_refillSize < SmallForDictionary && (fc = getChunkFromIndexedFreeList(blk->_refillSize)) != NULL) { // A linAB's strategy might be to use small sizes to reduce // fragmentation but still get the benefits of allocation from a // linAB. } else { fc = getChunkFromDictionary(blk->_refillSize); } if (fc != NULL) { blk->_ptr = (HeapWord*)fc; blk->_word_size = fc->size(); fc->dontCoalesce(); // to prevent sweeper from sweeping us up } } // Support for concurrent collection policy decisions. bool CompactibleFreeListSpace::should_concurrent_collect() const { // In the future we might want to add in frgamentation stats -- // including erosion of the "mountain" into this decision as well. return !adaptive_freelists() && linearAllocationWouldFail(); } // Support for compaction void CompactibleFreeListSpace::prepare_for_compaction(CompactPoint* cp) { SCAN_AND_FORWARD(cp,end,block_is_obj,block_size); // prepare_for_compaction() uses the space between live objects // so that later phase can skip dead space quickly. So verification // of the free lists doesn't work after. } #define obj_size(q) adjustObjectSize(oop(q)->size()) #define adjust_obj_size(s) adjustObjectSize(s) void CompactibleFreeListSpace::adjust_pointers() { // In other versions of adjust_pointers(), a bail out // based on the amount of live data in the generation // (i.e., if 0, bail out) may be used. // Cannot test used() == 0 here because the free lists have already // been mangled by the compaction. SCAN_AND_ADJUST_POINTERS(adjust_obj_size); // See note about verification in prepare_for_compaction(). } void CompactibleFreeListSpace::compact() { SCAN_AND_COMPACT(obj_size); } // fragmentation_metric = 1 - [sum of (fbs**2) / (sum of fbs)**2] // where fbs is free block sizes double CompactibleFreeListSpace::flsFrag() const { size_t itabFree = totalSizeInIndexedFreeLists(); double frag = 0.0; size_t i; for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { double sz = i; frag += _indexedFreeList[i].count() * (sz * sz); } double totFree = itabFree + _dictionary->totalChunkSize(DEBUG_ONLY(freelistLock())); if (totFree > 0) { frag = ((frag + _dictionary->sum_of_squared_block_sizes()) / (totFree * totFree)); frag = (double)1.0 - frag; } else { assert(frag == 0.0, "Follows from totFree == 0"); } return frag; } void CompactibleFreeListSpace::beginSweepFLCensus( float inter_sweep_current, float inter_sweep_estimate, float intra_sweep_estimate) { assert_locked(); size_t i; for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { FreeList* fl = &_indexedFreeList[i]; if (PrintFLSStatistics > 1) { gclog_or_tty->print("size[%d] : ", i); } fl->compute_desired(inter_sweep_current, inter_sweep_estimate, intra_sweep_estimate); fl->set_coalDesired((ssize_t)((double)fl->desired() * CMSSmallCoalSurplusPercent)); fl->set_beforeSweep(fl->count()); fl->set_bfrSurp(fl->surplus()); } _dictionary->beginSweepDictCensus(CMSLargeCoalSurplusPercent, inter_sweep_current, inter_sweep_estimate, intra_sweep_estimate); } void CompactibleFreeListSpace::setFLSurplus() { assert_locked(); size_t i; for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { FreeList *fl = &_indexedFreeList[i]; fl->set_surplus(fl->count() - (ssize_t)((double)fl->desired() * CMSSmallSplitSurplusPercent)); } } void CompactibleFreeListSpace::setFLHints() { assert_locked(); size_t i; size_t h = IndexSetSize; for (i = IndexSetSize - 1; i != 0; i -= IndexSetStride) { FreeList *fl = &_indexedFreeList[i]; fl->set_hint(h); if (fl->surplus() > 0) { h = i; } } } void CompactibleFreeListSpace::clearFLCensus() { assert_locked(); int i; for (i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { FreeList *fl = &_indexedFreeList[i]; fl->set_prevSweep(fl->count()); fl->set_coalBirths(0); fl->set_coalDeaths(0); fl->set_splitBirths(0); fl->set_splitDeaths(0); } } void CompactibleFreeListSpace::endSweepFLCensus(size_t sweep_count) { if (PrintFLSStatistics > 0) { HeapWord* largestAddr = (HeapWord*) dictionary()->findLargestDict(); gclog_or_tty->print_cr("CMS: Large block " PTR_FORMAT, largestAddr); } setFLSurplus(); setFLHints(); if (PrintGC && PrintFLSCensus > 0) { printFLCensus(sweep_count); } clearFLCensus(); assert_locked(); _dictionary->endSweepDictCensus(CMSLargeSplitSurplusPercent); } bool CompactibleFreeListSpace::coalOverPopulated(size_t size) { if (size < SmallForDictionary) { FreeList *fl = &_indexedFreeList[size]; return (fl->coalDesired() < 0) || ((int)fl->count() > fl->coalDesired()); } else { return dictionary()->coalDictOverPopulated(size); } } void CompactibleFreeListSpace::smallCoalBirth(size_t size) { assert(size < SmallForDictionary, "Size too large for indexed list"); FreeList *fl = &_indexedFreeList[size]; fl->increment_coalBirths(); fl->increment_surplus(); } void CompactibleFreeListSpace::smallCoalDeath(size_t size) { assert(size < SmallForDictionary, "Size too large for indexed list"); FreeList *fl = &_indexedFreeList[size]; fl->increment_coalDeaths(); fl->decrement_surplus(); } void CompactibleFreeListSpace::coalBirth(size_t size) { if (size < SmallForDictionary) { smallCoalBirth(size); } else { dictionary()->dictCensusUpdate(size, false /* split */, true /* birth */); } } void CompactibleFreeListSpace::coalDeath(size_t size) { if(size < SmallForDictionary) { smallCoalDeath(size); } else { dictionary()->dictCensusUpdate(size, false /* split */, false /* birth */); } } void CompactibleFreeListSpace::smallSplitBirth(size_t size) { assert(size < SmallForDictionary, "Size too large for indexed list"); FreeList *fl = &_indexedFreeList[size]; fl->increment_splitBirths(); fl->increment_surplus(); } void CompactibleFreeListSpace::smallSplitDeath(size_t size) { assert(size < SmallForDictionary, "Size too large for indexed list"); FreeList *fl = &_indexedFreeList[size]; fl->increment_splitDeaths(); fl->decrement_surplus(); } void CompactibleFreeListSpace::splitBirth(size_t size) { if (size < SmallForDictionary) { smallSplitBirth(size); } else { dictionary()->dictCensusUpdate(size, true /* split */, true /* birth */); } } void CompactibleFreeListSpace::splitDeath(size_t size) { if (size < SmallForDictionary) { smallSplitDeath(size); } else { dictionary()->dictCensusUpdate(size, true /* split */, false /* birth */); } } void CompactibleFreeListSpace::split(size_t from, size_t to1) { size_t to2 = from - to1; splitDeath(from); splitBirth(to1); splitBirth(to2); } void CompactibleFreeListSpace::print() const { print_on(tty); } void CompactibleFreeListSpace::prepare_for_verify() { assert_locked(); repairLinearAllocationBlocks(); // Verify that the SpoolBlocks look like free blocks of // appropriate sizes... To be done ... } class VerifyAllBlksClosure: public BlkClosure { private: const CompactibleFreeListSpace* _sp; const MemRegion _span; HeapWord* _last_addr; size_t _last_size; bool _last_was_obj; bool _last_was_live; public: VerifyAllBlksClosure(const CompactibleFreeListSpace* sp, MemRegion span) : _sp(sp), _span(span), _last_addr(NULL), _last_size(0), _last_was_obj(false), _last_was_live(false) { } virtual size_t do_blk(HeapWord* addr) { size_t res; bool was_obj = false; bool was_live = false; if (_sp->block_is_obj(addr)) { was_obj = true; oop p = oop(addr); guarantee(p->is_oop(), "Should be an oop"); res = _sp->adjustObjectSize(p->size()); if (_sp->obj_is_alive(addr)) { was_live = true; p->verify(); } } else { FreeChunk* fc = (FreeChunk*)addr; res = fc->size(); if (FLSVerifyLists && !fc->cantCoalesce()) { guarantee(_sp->verifyChunkInFreeLists(fc), "Chunk should be on a free list"); } } if (res == 0) { gclog_or_tty->print_cr("Livelock: no rank reduction!"); gclog_or_tty->print_cr( " Current: addr = " PTR_FORMAT ", size = " SIZE_FORMAT ", obj = %s, live = %s \n" " Previous: addr = " PTR_FORMAT ", size = " SIZE_FORMAT ", obj = %s, live = %s \n", addr, res, was_obj ?"true":"false", was_live ?"true":"false", _last_addr, _last_size, _last_was_obj?"true":"false", _last_was_live?"true":"false"); _sp->print_on(gclog_or_tty); guarantee(false, "Seppuku!"); } _last_addr = addr; _last_size = res; _last_was_obj = was_obj; _last_was_live = was_live; return res; } }; class VerifyAllOopsClosure: public OopClosure { private: const CMSCollector* _collector; const CompactibleFreeListSpace* _sp; const MemRegion _span; const bool _past_remark; const CMSBitMap* _bit_map; protected: void do_oop(void* p, oop obj) { if (_span.contains(obj)) { // the interior oop points into CMS heap if (!_span.contains(p)) { // reference from outside CMS heap // Should be a valid object; the first disjunct below allows // us to sidestep an assertion in block_is_obj() that insists // that p be in _sp. Note that several generations (and spaces) // are spanned by _span (CMS heap) above. guarantee(!_sp->is_in_reserved(obj) || _sp->block_is_obj((HeapWord*)obj), "Should be an object"); guarantee(obj->is_oop(), "Should be an oop"); obj->verify(); if (_past_remark) { // Remark has been completed, the object should be marked _bit_map->isMarked((HeapWord*)obj); } } else { // reference within CMS heap if (_past_remark) { // Remark has been completed -- so the referent should have // been marked, if referring object is. if (_bit_map->isMarked(_collector->block_start(p))) { guarantee(_bit_map->isMarked((HeapWord*)obj), "Marking error?"); } } } } else if (_sp->is_in_reserved(p)) { // the reference is from FLS, and points out of FLS guarantee(obj->is_oop(), "Should be an oop"); obj->verify(); } } template <class T> void do_oop_work(T* p) { T heap_oop = oopDesc::load_heap_oop(p); if (!oopDesc::is_null(heap_oop)) { oop obj = oopDesc::decode_heap_oop_not_null(heap_oop); do_oop(p, obj); } } public: VerifyAllOopsClosure(const CMSCollector* collector, const CompactibleFreeListSpace* sp, MemRegion span, bool past_remark, CMSBitMap* bit_map) : OopClosure(), _collector(collector), _sp(sp), _span(span), _past_remark(past_remark), _bit_map(bit_map) { } virtual void do_oop(oop* p) { VerifyAllOopsClosure::do_oop_work(p); } virtual void do_oop(narrowOop* p) { VerifyAllOopsClosure::do_oop_work(p); } }; void CompactibleFreeListSpace::verify(bool ignored) const { assert_lock_strong(&_freelistLock); verify_objects_initialized(); MemRegion span = _collector->_span; bool past_remark = (_collector->abstract_state() == CMSCollector::Sweeping); ResourceMark rm; HandleMark hm; // Check integrity of CFL data structures _promoInfo.verify(); _dictionary->verify(); if (FLSVerifyIndexTable) { verifyIndexedFreeLists(); } // Check integrity of all objects and free blocks in space { VerifyAllBlksClosure cl(this, span); ((CompactibleFreeListSpace*)this)->blk_iterate(&cl); // cast off const } // Check that all references in the heap to FLS // are to valid objects in FLS or that references in // FLS are to valid objects elsewhere in the heap if (FLSVerifyAllHeapReferences) { VerifyAllOopsClosure cl(_collector, this, span, past_remark, _collector->markBitMap()); CollectedHeap* ch = Universe::heap(); ch->oop_iterate(&cl); // all oops in generations ch->permanent_oop_iterate(&cl); // all oops in perm gen } if (VerifyObjectStartArray) { // Verify the block offset table _bt.verify(); } } #ifndef PRODUCT void CompactibleFreeListSpace::verifyFreeLists() const { if (FLSVerifyLists) { _dictionary->verify(); verifyIndexedFreeLists(); } else { if (FLSVerifyDictionary) { _dictionary->verify(); } if (FLSVerifyIndexTable) { verifyIndexedFreeLists(); } } } #endif void CompactibleFreeListSpace::verifyIndexedFreeLists() const { size_t i = 0; for (; i < MinChunkSize; i++) { guarantee(_indexedFreeList[i].head() == NULL, "should be NULL"); } for (; i < IndexSetSize; i++) { verifyIndexedFreeList(i); } } void CompactibleFreeListSpace::verifyIndexedFreeList(size_t size) const { FreeChunk* fc = _indexedFreeList[size].head(); FreeChunk* tail = _indexedFreeList[size].tail(); size_t num = _indexedFreeList[size].count(); size_t n = 0; guarantee((size % 2 == 0) || fc == NULL, "Odd slots should be empty"); for (; fc != NULL; fc = fc->next(), n++) { guarantee(fc->size() == size, "Size inconsistency"); guarantee(fc->isFree(), "!free?"); guarantee(fc->next() == NULL || fc->next()->prev() == fc, "Broken list"); guarantee((fc->next() == NULL) == (fc == tail), "Incorrect tail"); } guarantee(n == num, "Incorrect count"); } #ifndef PRODUCT void CompactibleFreeListSpace::checkFreeListConsistency() const { assert(_dictionary->minSize() <= IndexSetSize, "Some sizes can't be allocated without recourse to" " linear allocation buffers"); assert(MIN_TREE_CHUNK_SIZE*HeapWordSize == sizeof(TreeChunk), "else MIN_TREE_CHUNK_SIZE is wrong"); assert((IndexSetStride == 2 && IndexSetStart == 2) || (IndexSetStride == 1 && IndexSetStart == 1), "just checking"); assert((IndexSetStride != 2) || (MinChunkSize % 2 == 0), "Some for-loops may be incorrectly initialized"); assert((IndexSetStride != 2) || (IndexSetSize % 2 == 1), "For-loops that iterate over IndexSet with stride 2 may be wrong"); } #endif void CompactibleFreeListSpace::printFLCensus(size_t sweep_count) const { assert_lock_strong(&_freelistLock); FreeList total; gclog_or_tty->print("end sweep# " SIZE_FORMAT "\n", sweep_count); FreeList::print_labels_on(gclog_or_tty, "size"); size_t totalFree = 0; for (size_t i = IndexSetStart; i < IndexSetSize; i += IndexSetStride) { const FreeList *fl = &_indexedFreeList[i]; totalFree += fl->count() * fl->size(); if (i % (40*IndexSetStride) == 0) { FreeList::print_labels_on(gclog_or_tty, "size"); } fl->print_on(gclog_or_tty); total.set_bfrSurp( total.bfrSurp() + fl->bfrSurp() ); total.set_surplus( total.surplus() + fl->surplus() ); total.set_desired( total.desired() + fl->desired() ); total.set_prevSweep( total.prevSweep() + fl->prevSweep() ); total.set_beforeSweep(total.beforeSweep() + fl->beforeSweep()); total.set_count( total.count() + fl->count() ); total.set_coalBirths( total.coalBirths() + fl->coalBirths() ); total.set_coalDeaths( total.coalDeaths() + fl->coalDeaths() ); total.set_splitBirths(total.splitBirths() + fl->splitBirths()); total.set_splitDeaths(total.splitDeaths() + fl->splitDeaths()); } total.print_on(gclog_or_tty, "TOTAL"); gclog_or_tty->print_cr("Total free in indexed lists " SIZE_FORMAT " words", totalFree); gclog_or_tty->print("growth: %8.5f deficit: %8.5f\n", (double)(total.splitBirths()+total.coalBirths()-total.splitDeaths()-total.coalDeaths())/ (total.prevSweep() != 0 ? (double)total.prevSweep() : 1.0), (double)(total.desired() - total.count())/(total.desired() != 0 ? (double)total.desired() : 1.0)); _dictionary->printDictCensus(); } /////////////////////////////////////////////////////////////////////////// // CFLS_LAB /////////////////////////////////////////////////////////////////////////// #define VECTOR_257(x) \ /* 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 */ \ { 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, \ 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, \ 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, \ 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, \ 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, \ 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, \ 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, \ 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, \ x } // Initialize with default setting of CMSParPromoteBlocksToClaim, _not_ // OldPLABSize, whose static default is different; if overridden at the // command-line, this will get reinitialized via a call to // modify_initialization() below. AdaptiveWeightedAverage CFLS_LAB::_blocks_to_claim[] = VECTOR_257(AdaptiveWeightedAverage(OldPLABWeight, (float)CMSParPromoteBlocksToClaim)); size_t CFLS_LAB::_global_num_blocks[] = VECTOR_257(0); int CFLS_LAB::_global_num_workers[] = VECTOR_257(0); CFLS_LAB::CFLS_LAB(CompactibleFreeListSpace* cfls) : _cfls(cfls) { assert(CompactibleFreeListSpace::IndexSetSize == 257, "Modify VECTOR_257() macro above"); for (size_t i = CompactibleFreeListSpace::IndexSetStart; i < CompactibleFreeListSpace::IndexSetSize; i += CompactibleFreeListSpace::IndexSetStride) { _indexedFreeList[i].set_size(i); _num_blocks[i] = 0; } } static bool _CFLS_LAB_modified = false; void CFLS_LAB::modify_initialization(size_t n, unsigned wt) { assert(!_CFLS_LAB_modified, "Call only once"); _CFLS_LAB_modified = true; for (size_t i = CompactibleFreeListSpace::IndexSetStart; i < CompactibleFreeListSpace::IndexSetSize; i += CompactibleFreeListSpace::IndexSetStride) { _blocks_to_claim[i].modify(n, wt, true /* force */); } } HeapWord* CFLS_LAB::alloc(size_t word_sz) { FreeChunk* res; assert(word_sz == _cfls->adjustObjectSize(word_sz), "Error"); if (word_sz >= CompactibleFreeListSpace::IndexSetSize) { // This locking manages sync with other large object allocations. MutexLockerEx x(_cfls->parDictionaryAllocLock(), Mutex::_no_safepoint_check_flag); res = _cfls->getChunkFromDictionaryExact(word_sz); if (res == NULL) return NULL; } else { FreeList* fl = &_indexedFreeList[word_sz]; if (fl->count() == 0) { // Attempt to refill this local free list. get_from_global_pool(word_sz, fl); // If it didn't work, give up. if (fl->count() == 0) return NULL; } res = fl->getChunkAtHead(); assert(res != NULL, "Why was count non-zero?"); } res->markNotFree(); assert(!res->isFree(), "shouldn't be marked free"); assert(oop(res)->klass_or_null() == NULL, "should look uninitialized"); // mangle a just allocated object with a distinct pattern. debug_only(res->mangleAllocated(word_sz)); return (HeapWord*)res; } // Get a chunk of blocks of the right size and update related // book-keeping stats void CFLS_LAB::get_from_global_pool(size_t word_sz, FreeList* fl) { // Get the #blocks we want to claim size_t n_blks = (size_t)_blocks_to_claim[word_sz].average(); assert(n_blks > 0, "Error"); assert(ResizePLAB || n_blks == OldPLABSize, "Error"); // In some cases, when the application has a phase change, // there may be a sudden and sharp shift in the object survival // profile, and updating the counts at the end of a scavenge // may not be quick enough, giving rise to large scavenge pauses // during these phase changes. It is beneficial to detect such // changes on-the-fly during a scavenge and avoid such a phase-change // pothole. The following code is a heuristic attempt to do that. // It is protected by a product flag until we have gained // enough experience with this heuristic and fine-tuned its behaviour. // WARNING: This might increase fragmentation if we overreact to // small spikes, so some kind of historical smoothing based on // previous experience with the greater reactivity might be useful. // Lacking sufficient experience, CMSOldPLABResizeQuicker is disabled by // default. if (ResizeOldPLAB && CMSOldPLABResizeQuicker) { size_t multiple = _num_blocks[word_sz]/(CMSOldPLABToleranceFactor*CMSOldPLABNumRefills*n_blks); n_blks += CMSOldPLABReactivityFactor*multiple*n_blks; n_blks = MIN2(n_blks, CMSOldPLABMax); } assert(n_blks > 0, "Error"); _cfls->par_get_chunk_of_blocks(word_sz, n_blks, fl); // Update stats table entry for this block size _num_blocks[word_sz] += fl->count(); } void CFLS_LAB::compute_desired_plab_size() { for (size_t i = CompactibleFreeListSpace::IndexSetStart; i < CompactibleFreeListSpace::IndexSetSize; i += CompactibleFreeListSpace::IndexSetStride) { assert((_global_num_workers[i] == 0) == (_global_num_blocks[i] == 0), "Counter inconsistency"); if (_global_num_workers[i] > 0) { // Need to smooth wrt historical average if (ResizeOldPLAB) { _blocks_to_claim[i].sample( MAX2((size_t)CMSOldPLABMin, MIN2((size_t)CMSOldPLABMax, _global_num_blocks[i]/(_global_num_workers[i]*CMSOldPLABNumRefills)))); } // Reset counters for next round _global_num_workers[i] = 0; _global_num_blocks[i] = 0; if (PrintOldPLAB) { gclog_or_tty->print_cr("[%d]: %d", i, (size_t)_blocks_to_claim[i].average()); } } } } void CFLS_LAB::retire(int tid) { // We run this single threaded with the world stopped; // so no need for locks and such. #define CFLS_LAB_PARALLEL_ACCESS 0 NOT_PRODUCT(Thread* t = Thread::current();) assert(Thread::current()->is_VM_thread(), "Error"); assert(CompactibleFreeListSpace::IndexSetStart == CompactibleFreeListSpace::IndexSetStride, "Will access to uninitialized slot below"); #if CFLS_LAB_PARALLEL_ACCESS for (size_t i = CompactibleFreeListSpace::IndexSetSize - 1; i > 0; i -= CompactibleFreeListSpace::IndexSetStride) { #else // CFLS_LAB_PARALLEL_ACCESS for (size_t i = CompactibleFreeListSpace::IndexSetStart; i < CompactibleFreeListSpace::IndexSetSize; i += CompactibleFreeListSpace::IndexSetStride) { #endif // !CFLS_LAB_PARALLEL_ACCESS assert(_num_blocks[i] >= (size_t)_indexedFreeList[i].count(), "Can't retire more than what we obtained"); if (_num_blocks[i] > 0) { size_t num_retire = _indexedFreeList[i].count(); assert(_num_blocks[i] > num_retire, "Should have used at least one"); { #if CFLS_LAB_PARALLEL_ACCESS MutexLockerEx x(_cfls->_indexedFreeListParLocks[i], Mutex::_no_safepoint_check_flag); #endif // CFLS_LAB_PARALLEL_ACCESS // Update globals stats for num_blocks used _global_num_blocks[i] += (_num_blocks[i] - num_retire); _global_num_workers[i]++; assert(_global_num_workers[i] <= (ssize_t)ParallelGCThreads, "Too big"); if (num_retire > 0) { _cfls->_indexedFreeList[i].prepend(&_indexedFreeList[i]); // Reset this list. _indexedFreeList[i] = FreeList(); _indexedFreeList[i].set_size(i); } } if (PrintOldPLAB) { gclog_or_tty->print_cr("%d[%d]: %d/%d/%d", tid, i, num_retire, _num_blocks[i], (size_t)_blocks_to_claim[i].average()); } // Reset stats for next round _num_blocks[i] = 0; } } } void CompactibleFreeListSpace:: par_get_chunk_of_blocks(size_t word_sz, size_t n, FreeList* fl) { assert(fl->count() == 0, "Precondition."); assert(word_sz < CompactibleFreeListSpace::IndexSetSize, "Precondition"); // We'll try all multiples of word_sz in the indexed set, starting with // word_sz itself and, if CMSSplitIndexedFreeListBlocks, try larger multiples, // then try getting a big chunk and splitting it. { bool found; int k; size_t cur_sz; for (k = 1, cur_sz = k * word_sz, found = false; (cur_sz < CompactibleFreeListSpace::IndexSetSize) && (CMSSplitIndexedFreeListBlocks || k <= 1); k++, cur_sz = k * word_sz) { FreeList fl_for_cur_sz; // Empty. fl_for_cur_sz.set_size(cur_sz); { MutexLockerEx x(_indexedFreeListParLocks[cur_sz], Mutex::_no_safepoint_check_flag); FreeList* gfl = &_indexedFreeList[cur_sz]; if (gfl->count() != 0) { // nn is the number of chunks of size cur_sz that // we'd need to split k-ways each, in order to create // "n" chunks of size word_sz each. const size_t nn = MAX2(n/k, (size_t)1); gfl->getFirstNChunksFromList(nn, &fl_for_cur_sz); found = true; if (k > 1) { // Update split death stats for the cur_sz-size blocks list: // we increment the split death count by the number of blocks // we just took from the cur_sz-size blocks list and which // we will be splitting below. ssize_t deaths = gfl->splitDeaths() + fl_for_cur_sz.count(); gfl->set_splitDeaths(deaths); } } } // Now transfer fl_for_cur_sz to fl. Common case, we hope, is k = 1. if (found) { if (k == 1) { fl->prepend(&fl_for_cur_sz); } else { // Divide each block on fl_for_cur_sz up k ways. FreeChunk* fc; while ((fc = fl_for_cur_sz.getChunkAtHead()) != NULL) { // Must do this in reverse order, so that anybody attempting to // access the main chunk sees it as a single free block until we // change it. size_t fc_size = fc->size(); assert(fc->isFree(), "Error"); for (int i = k-1; i >= 0; i--) { FreeChunk* ffc = (FreeChunk*)((HeapWord*)fc + i * word_sz); assert((i != 0) || ((fc == ffc) && ffc->isFree() && (ffc->size() == k*word_sz) && (fc_size == word_sz)), "Counting error"); ffc->setSize(word_sz); ffc->linkPrev(NULL); // Mark as a free block for other (parallel) GC threads. ffc->linkNext(NULL); // Above must occur before BOT is updated below. OrderAccess::storestore(); // splitting from the right, fc_size == i * word_sz _bt.mark_block((HeapWord*)ffc, word_sz, true /* reducing */); fc_size -= word_sz; assert(fc_size == i*word_sz, "Error"); _bt.verify_not_unallocated((HeapWord*)ffc, word_sz); _bt.verify_single_block((HeapWord*)fc, fc_size); _bt.verify_single_block((HeapWord*)ffc, word_sz); // Push this on "fl". fl->returnChunkAtHead(ffc); } // TRAP assert(fl->tail()->next() == NULL, "List invariant."); } } // Update birth stats for this block size. size_t num = fl->count(); MutexLockerEx x(_indexedFreeListParLocks[word_sz], Mutex::_no_safepoint_check_flag); ssize_t births = _indexedFreeList[word_sz].splitBirths() + num; _indexedFreeList[word_sz].set_splitBirths(births); return; } } } // Otherwise, we'll split a block from the dictionary. FreeChunk* fc = NULL; FreeChunk* rem_fc = NULL; size_t rem; { MutexLockerEx x(parDictionaryAllocLock(), Mutex::_no_safepoint_check_flag); while (n > 0) { fc = dictionary()->getChunk(MAX2(n * word_sz, _dictionary->minSize()), FreeBlockDictionary::atLeast); if (fc != NULL) { _bt.allocated((HeapWord*)fc, fc->size(), true /* reducing */); // update _unallocated_blk dictionary()->dictCensusUpdate(fc->size(), true /*split*/, false /*birth*/); break; } else { n--; } } if (fc == NULL) return; // Otherwise, split up that block. assert((ssize_t)n >= 1, "Control point invariant"); assert(fc->isFree(), "Error: should be a free block"); _bt.verify_single_block((HeapWord*)fc, fc->size()); const size_t nn = fc->size() / word_sz; n = MIN2(nn, n); assert((ssize_t)n >= 1, "Control point invariant"); rem = fc->size() - n * word_sz; // If there is a remainder, and it's too small, allocate one fewer. if (rem > 0 && rem < MinChunkSize) { n--; rem += word_sz; } // Note that at this point we may have n == 0. assert((ssize_t)n >= 0, "Control point invariant"); // If n is 0, the chunk fc that was found is not large // enough to leave a viable remainder. We are unable to // allocate even one block. Return fc to the // dictionary and return, leaving "fl" empty. if (n == 0) { returnChunkToDictionary(fc); assert(fl->count() == 0, "We never allocated any blocks"); return; } // First return the remainder, if any. // Note that we hold the lock until we decide if we're going to give // back the remainder to the dictionary, since a concurrent allocation // may otherwise see the heap as empty. (We're willing to take that // hit if the block is a small block.) if (rem > 0) { size_t prefix_size = n * word_sz; rem_fc = (FreeChunk*)((HeapWord*)fc + prefix_size); rem_fc->setSize(rem); rem_fc->linkPrev(NULL); // Mark as a free block for other (parallel) GC threads. rem_fc->linkNext(NULL); // Above must occur before BOT is updated below. assert((ssize_t)n > 0 && prefix_size > 0 && rem_fc > fc, "Error"); OrderAccess::storestore(); _bt.split_block((HeapWord*)fc, fc->size(), prefix_size); assert(fc->isFree(), "Error"); fc->setSize(prefix_size); if (rem >= IndexSetSize) { returnChunkToDictionary(rem_fc); dictionary()->dictCensusUpdate(rem, true /*split*/, true /*birth*/); rem_fc = NULL; } // Otherwise, return it to the small list below. } } if (rem_fc != NULL) { MutexLockerEx x(_indexedFreeListParLocks[rem], Mutex::_no_safepoint_check_flag); _bt.verify_not_unallocated((HeapWord*)rem_fc, rem_fc->size()); _indexedFreeList[rem].returnChunkAtHead(rem_fc); smallSplitBirth(rem); } assert((ssize_t)n > 0 && fc != NULL, "Consistency"); // Now do the splitting up. // Must do this in reverse order, so that anybody attempting to // access the main chunk sees it as a single free block until we // change it. size_t fc_size = n * word_sz; // All but first chunk in this loop for (ssize_t i = n-1; i > 0; i--) { FreeChunk* ffc = (FreeChunk*)((HeapWord*)fc + i * word_sz); ffc->setSize(word_sz); ffc->linkPrev(NULL); // Mark as a free block for other (parallel) GC threads. ffc->linkNext(NULL); // Above must occur before BOT is updated below. OrderAccess::storestore(); // splitting from the right, fc_size == (n - i + 1) * wordsize _bt.mark_block((HeapWord*)ffc, word_sz, true /* reducing */); fc_size -= word_sz; _bt.verify_not_unallocated((HeapWord*)ffc, ffc->size()); _bt.verify_single_block((HeapWord*)ffc, ffc->size()); _bt.verify_single_block((HeapWord*)fc, fc_size); // Push this on "fl". fl->returnChunkAtHead(ffc); } // First chunk assert(fc->isFree() && fc->size() == n*word_sz, "Error: should still be a free block"); // The blocks above should show their new sizes before the first block below fc->setSize(word_sz); fc->linkPrev(NULL); // idempotent wrt free-ness, see assert above fc->linkNext(NULL); _bt.verify_not_unallocated((HeapWord*)fc, fc->size()); _bt.verify_single_block((HeapWord*)fc, fc->size()); fl->returnChunkAtHead(fc); assert((ssize_t)n > 0 && (ssize_t)n == fl->count(), "Incorrect number of blocks"); { // Update the stats for this block size. MutexLockerEx x(_indexedFreeListParLocks[word_sz], Mutex::_no_safepoint_check_flag); const ssize_t births = _indexedFreeList[word_sz].splitBirths() + n; _indexedFreeList[word_sz].set_splitBirths(births); // ssize_t new_surplus = _indexedFreeList[word_sz].surplus() + n; // _indexedFreeList[word_sz].set_surplus(new_surplus); } // TRAP assert(fl->tail()->next() == NULL, "List invariant."); } // Set up the space's par_seq_tasks structure for work claiming // for parallel rescan. See CMSParRemarkTask where this is currently used. // XXX Need to suitably abstract and generalize this and the next // method into one. void CompactibleFreeListSpace:: initialize_sequential_subtasks_for_rescan(int n_threads) { // The "size" of each task is fixed according to rescan_task_size. assert(n_threads > 0, "Unexpected n_threads argument"); const size_t task_size = rescan_task_size(); size_t n_tasks = (used_region().word_size() + task_size - 1)/task_size; assert((n_tasks == 0) == used_region().is_empty(), "n_tasks incorrect"); assert(n_tasks == 0 || ((used_region().start() + (n_tasks - 1)*task_size < used_region().end()) && (used_region().start() + n_tasks*task_size >= used_region().end())), "n_tasks calculation incorrect"); SequentialSubTasksDone* pst = conc_par_seq_tasks(); assert(!pst->valid(), "Clobbering existing data?"); // Sets the condition for completion of the subtask (how many threads // need to finish in order to be done). pst->set_n_threads(n_threads); pst->set_n_tasks((int)n_tasks); } // Set up the space's par_seq_tasks structure for work claiming // for parallel concurrent marking. See CMSConcMarkTask where this is currently used. void CompactibleFreeListSpace:: initialize_sequential_subtasks_for_marking(int n_threads, HeapWord* low) { // The "size" of each task is fixed according to rescan_task_size. assert(n_threads > 0, "Unexpected n_threads argument"); const size_t task_size = marking_task_size(); assert(task_size > CardTableModRefBS::card_size_in_words && (task_size % CardTableModRefBS::card_size_in_words == 0), "Otherwise arithmetic below would be incorrect"); MemRegion span = _gen->reserved(); if (low != NULL) { if (span.contains(low)) { // Align low down to a card boundary so that // we can use block_offset_careful() on span boundaries. HeapWord* aligned_low = (HeapWord*)align_size_down((uintptr_t)low, CardTableModRefBS::card_size); // Clip span prefix at aligned_low span = span.intersection(MemRegion(aligned_low, span.end())); } else if (low > span.end()) { span = MemRegion(low, low); // Null region } // else use entire span } assert(span.is_empty() || ((uintptr_t)span.start() % CardTableModRefBS::card_size == 0), "span should start at a card boundary"); size_t n_tasks = (span.word_size() + task_size - 1)/task_size; assert((n_tasks == 0) == span.is_empty(), "Inconsistency"); assert(n_tasks == 0 || ((span.start() + (n_tasks - 1)*task_size < span.end()) && (span.start() + n_tasks*task_size >= span.end())), "n_tasks calculation incorrect"); SequentialSubTasksDone* pst = conc_par_seq_tasks(); assert(!pst->valid(), "Clobbering existing data?"); // Sets the condition for completion of the subtask (how many threads // need to finish in order to be done). pst->set_n_threads(n_threads); pst->set_n_tasks((int)n_tasks); }