Mercurial > hg > graal-compiler
view src/share/vm/gc_implementation/g1/heapRegion.inline.hpp @ 20543:e7d0505c8a30
8059758: Footprint regressions with JDK-8038423
Summary: Changes in JDK-8038423 always initialize (zero out) virtual memory used for auxiliary data structures. This causes a footprint regression for G1 in startup benchmarks. This is because they do not touch that memory at all, so the operating system does not actually commit these pages. The fix is to, if the initialization value of the data structures matches the default value of just committed memory (=0), do not do anything.
Reviewed-by: jwilhelm, brutisso
author | tschatzl |
---|---|
date | Fri, 10 Oct 2014 15:51:58 +0200 |
parents | f3aeae1f9fc5 |
children |
line wrap: on
line source
/* * Copyright (c) 2001, 2012, 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_GC_IMPLEMENTATION_G1_HEAPREGION_INLINE_HPP #define SHARE_VM_GC_IMPLEMENTATION_G1_HEAPREGION_INLINE_HPP #include "gc_implementation/g1/g1BlockOffsetTable.inline.hpp" #include "gc_implementation/g1/g1CollectedHeap.hpp" #include "gc_implementation/g1/heapRegion.hpp" #include "memory/space.hpp" #include "runtime/atomic.inline.hpp" // This version requires locking. inline HeapWord* G1OffsetTableContigSpace::allocate_impl(size_t size, HeapWord* const end_value) { HeapWord* obj = top(); if (pointer_delta(end_value, obj) >= size) { HeapWord* new_top = obj + size; set_top(new_top); assert(is_aligned(obj) && is_aligned(new_top), "checking alignment"); return obj; } else { return NULL; } } // This version is lock-free. inline HeapWord* G1OffsetTableContigSpace::par_allocate_impl(size_t size, HeapWord* const end_value) { do { HeapWord* obj = top(); if (pointer_delta(end_value, obj) >= size) { HeapWord* new_top = obj + size; HeapWord* result = (HeapWord*)Atomic::cmpxchg_ptr(new_top, top_addr(), obj); // result can be one of two: // the old top value: the exchange succeeded // otherwise: the new value of the top is returned. if (result == obj) { assert(is_aligned(obj) && is_aligned(new_top), "checking alignment"); return obj; } } else { return NULL; } } while (true); } inline HeapWord* G1OffsetTableContigSpace::allocate(size_t size) { HeapWord* res = allocate_impl(size, end()); if (res != NULL) { _offsets.alloc_block(res, size); } return res; } // Because of the requirement of keeping "_offsets" up to date with the // allocations, we sequentialize these with a lock. Therefore, best if // this is used for larger LAB allocations only. inline HeapWord* G1OffsetTableContigSpace::par_allocate(size_t size) { MutexLocker x(&_par_alloc_lock); return allocate(size); } inline HeapWord* G1OffsetTableContigSpace::block_start(const void* p) { return _offsets.block_start(p); } inline HeapWord* G1OffsetTableContigSpace::block_start_const(const void* p) const { return _offsets.block_start_const(p); } inline bool HeapRegion::block_is_obj(const HeapWord* p) const { G1CollectedHeap* g1h = G1CollectedHeap::heap(); if (ClassUnloadingWithConcurrentMark) { return !g1h->is_obj_dead(oop(p), this); } return p < top(); } inline size_t HeapRegion::block_size(const HeapWord *addr) const { if (addr == top()) { return pointer_delta(end(), addr); } if (block_is_obj(addr)) { return oop(addr)->size(); } assert(ClassUnloadingWithConcurrentMark, err_msg("All blocks should be objects if G1 Class Unloading isn't used. " "HR: ["PTR_FORMAT", "PTR_FORMAT", "PTR_FORMAT") " "addr: " PTR_FORMAT, p2i(bottom()), p2i(top()), p2i(end()), p2i(addr))); // Old regions' dead objects may have dead classes // We need to find the next live object in some other // manner than getting the oop size G1CollectedHeap* g1h = G1CollectedHeap::heap(); HeapWord* next = g1h->concurrent_mark()->prevMarkBitMap()-> getNextMarkedWordAddress(addr, prev_top_at_mark_start()); assert(next > addr, "must get the next live object"); return pointer_delta(next, addr); } inline HeapWord* HeapRegion::par_allocate_no_bot_updates(size_t word_size) { assert(is_young(), "we can only skip BOT updates on young regions"); return par_allocate_impl(word_size, end()); } inline HeapWord* HeapRegion::allocate_no_bot_updates(size_t word_size) { assert(is_young(), "we can only skip BOT updates on young regions"); return allocate_impl(word_size, end()); } inline void HeapRegion::note_start_of_marking() { _next_marked_bytes = 0; _next_top_at_mark_start = top(); } inline void HeapRegion::note_end_of_marking() { _prev_top_at_mark_start = _next_top_at_mark_start; _prev_marked_bytes = _next_marked_bytes; _next_marked_bytes = 0; assert(_prev_marked_bytes <= (size_t) pointer_delta(prev_top_at_mark_start(), bottom()) * HeapWordSize, "invariant"); } inline void HeapRegion::note_start_of_copying(bool during_initial_mark) { if (is_survivor()) { // This is how we always allocate survivors. assert(_next_top_at_mark_start == bottom(), "invariant"); } else { if (during_initial_mark) { // During initial-mark we'll explicitly mark any objects on old // regions that are pointed to by roots. Given that explicit // marks only make sense under NTAMS it'd be nice if we could // check that condition if we wanted to. Given that we don't // know where the top of this region will end up, we simply set // NTAMS to the end of the region so all marks will be below // NTAMS. We'll set it to the actual top when we retire this region. _next_top_at_mark_start = end(); } else { // We could have re-used this old region as to-space over a // couple of GCs since the start of the concurrent marking // cycle. This means that [bottom,NTAMS) will contain objects // copied up to and including initial-mark and [NTAMS, top) // will contain objects copied during the concurrent marking cycle. assert(top() >= _next_top_at_mark_start, "invariant"); } } } inline void HeapRegion::note_end_of_copying(bool during_initial_mark) { if (is_survivor()) { // This is how we always allocate survivors. assert(_next_top_at_mark_start == bottom(), "invariant"); } else { if (during_initial_mark) { // See the comment for note_start_of_copying() for the details // on this. assert(_next_top_at_mark_start == end(), "pre-condition"); _next_top_at_mark_start = top(); } else { // See the comment for note_start_of_copying() for the details // on this. assert(top() >= _next_top_at_mark_start, "invariant"); } } } #endif // SHARE_VM_GC_IMPLEMENTATION_G1_HEAPREGION_INLINE_HPP