Mercurial > hg > graal-compiler
view src/share/vm/services/g1MemoryPool.hpp @ 2149:7e37af9d69ef
7011379: G1: overly long concurrent marking cycles
Summary: This changeset introduces filtering of SATB buffers at the point when they are about to be enqueued. If this filtering clears enough entries on each buffer, the buffer can then be re-used and not enqueued. This cuts down the number of SATB buffers that need to be processed by the concurrent marking threads.
Reviewed-by: johnc, ysr
author | tonyp |
---|---|
date | Wed, 19 Jan 2011 09:35:17 -0500 |
parents | f95d63e2154a |
children | b52782ae3880 |
line wrap: on
line source
/* * Copyright (c) 2007, 2010, 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. * */ #ifndef SHARE_VM_SERVICES_G1MEMORYPOOL_HPP #define SHARE_VM_SERVICES_G1MEMORYPOOL_HPP #ifndef SERIALGC #include "services/memoryPool.hpp" #include "services/memoryUsage.hpp" #endif class G1CollectedHeap; // This file contains the three classes that represent the memory // pools of the G1 spaces: G1EdenPool, G1SurvivorPool, and // G1OldGenPool. In G1, unlike our other GCs, we do not have a // physical space for each of those spaces. Instead, we allocate // regions for all three spaces out of a single pool of regions (that // pool basically covers the entire heap). As a result, the eden, // survivor, and old gen are considered logical spaces in G1, as each // is a set of non-contiguous regions. This is also reflected in the // way we map them to memory pools here. The easiest way to have done // this would have been to map the entire G1 heap to a single memory // pool. However, it's helpful to show how large the eden and survivor // get, as this does affect the performance and behavior of G1. Which // is why we introduce the three memory pools implemented here. // // The above approach inroduces a couple of challenging issues in the // implementation of the three memory pools: // // 1) The used space calculation for a pool is not necessarily // independent of the others. We can easily get from G1 the overall // used space in the entire heap, the number of regions in the young // generation (includes both eden and survivors), and the number of // survivor regions. So, from that we calculate: // // survivor_used = survivor_num * region_size // eden_used = young_region_num * region_size - survivor_used // old_gen_used = overall_used - eden_used - survivor_used // // Note that survivor_used and eden_used are upper bounds. To get the // actual value we would have to iterate over the regions and add up // ->used(). But that'd be expensive. So, we'll accept some lack of // accuracy for those two. But, we have to be careful when calculating // old_gen_used, in case we subtract from overall_used more then the // actual number and our result goes negative. // // 2) Calculating the used space is straightforward, as described // above. However, how do we calculate the committed space, given that // we allocate space for the eden, survivor, and old gen out of the // same pool of regions? One way to do this is to use the used value // as also the committed value for the eden and survivor spaces and // then calculate the old gen committed space as follows: // // old_gen_committed = overall_committed - eden_committed - survivor_committed // // Maybe a better way to do that would be to calculate used for eden // and survivor as a sum of ->used() over their regions and then // calculate committed as region_num * region_size (i.e., what we use // to calculate the used space now). This is something to consider // in the future. // // 3) Another decision that is again not straightforward is what is // the max size that each memory pool can grow to. One way to do this // would be to use the committed size for the max for the eden and // survivors and calculate the old gen max as follows (basically, it's // a similar pattern to what we use for the committed space, as // described above): // // old_gen_max = overall_max - eden_max - survivor_max // // Unfortunately, the above makes the max of each pool fluctuate over // time and, even though this is allowed according to the spec, it // broke several assumptions in the M&M framework (there were cases // where used would reach a value greater than max). So, for max we // use -1, which means "undefined" according to the spec. // // 4) Now, there is a very subtle issue with all the above. The // framework will call get_memory_usage() on the three pools // asynchronously. As a result, each call might get a different value // for, say, survivor_num which will yield inconsistent values for // eden_used, survivor_used, and old_gen_used (as survivor_num is used // in the calculation of all three). This would normally be // ok. However, it's possible that this might cause the sum of // eden_used, survivor_used, and old_gen_used to go over the max heap // size and this seems to sometimes cause JConsole (and maybe other // clients) to get confused. There's not a really an easy / clean // solution to this problem, due to the asynchrounous nature of the // framework. // This class is shared by the three G1 memory pool classes // (G1EdenPool, G1SurvivorPool, G1OldGenPool). Given that the way we // calculate used / committed bytes for these three pools is related // (see comment above), we put the calculations in this class so that // we can easily share them among the subclasses. class G1MemoryPoolSuper : public CollectedMemoryPool { private: // It returns x - y if x > y, 0 otherwise. // As described in the comment above, some of the inputs to the // calculations we have to do are obtained concurrently and hence // may be inconsistent with each other. So, this provides a // defensive way of performing the subtraction and avoids the value // going negative (which would mean a very large result, given that // the parameter are size_t). static size_t subtract_up_to_zero(size_t x, size_t y) { if (x > y) { return x - y; } else { return 0; } } protected: G1CollectedHeap* _g1h; // Would only be called from subclasses. G1MemoryPoolSuper(G1CollectedHeap* g1h, const char* name, size_t init_size, bool support_usage_threshold); // The reason why all the code is in static methods is so that it // can be safely called from the constructors of the subclasses. static size_t undefined_max() { return (size_t) -1; } static size_t overall_committed(G1CollectedHeap* g1h) { return g1h->capacity(); } static size_t overall_used(G1CollectedHeap* g1h) { return g1h->used_unlocked(); } static size_t eden_space_committed(G1CollectedHeap* g1h); static size_t eden_space_used(G1CollectedHeap* g1h); static size_t survivor_space_committed(G1CollectedHeap* g1h); static size_t survivor_space_used(G1CollectedHeap* g1h); static size_t old_space_committed(G1CollectedHeap* g1h); static size_t old_space_used(G1CollectedHeap* g1h); }; // Memory pool that represents the G1 eden. class G1EdenPool : public G1MemoryPoolSuper { public: G1EdenPool(G1CollectedHeap* g1h); size_t used_in_bytes() { return eden_space_used(_g1h); } size_t max_size() const { return undefined_max(); } MemoryUsage get_memory_usage(); }; // Memory pool that represents the G1 survivor. class G1SurvivorPool : public G1MemoryPoolSuper { public: G1SurvivorPool(G1CollectedHeap* g1h); size_t used_in_bytes() { return survivor_space_used(_g1h); } size_t max_size() const { return undefined_max(); } MemoryUsage get_memory_usage(); }; // Memory pool that represents the G1 old gen. class G1OldGenPool : public G1MemoryPoolSuper { public: G1OldGenPool(G1CollectedHeap* g1h); size_t used_in_bytes() { return old_space_used(_g1h); } size_t max_size() const { return undefined_max(); } MemoryUsage get_memory_usage(); }; #endif // SHARE_VM_SERVICES_G1MEMORYPOOL_HPP