hotspot/src/share/vm/gc_implementation/concurrentMarkSweep/freeChunk.hpp
author ysr
Wed, 23 Dec 2009 09:23:54 -0800
changeset 4574 b2d5b0975515
parent 977 b90650e2a9f7
child 5694 1e0532a6abff
child 5547 f4b087cbb361
permissions -rw-r--r--
6631166: CMS: better heuristics when combatting fragmentation Summary: Autonomic per-worker free block cache sizing, tunable coalition policies, fixes to per-size block statistics, retuned gain and bandwidth of some feedback loop filters to allow quicker reactivity to abrupt changes in ambient demand, and other heuristics to reduce fragmentation of the CMS old gen. Also tightened some assertions, including those related to locking. Reviewed-by: jmasa

/*
 * Copyright 2001-2008 Sun Microsystems, Inc.  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 Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
 * CA 95054 USA or visit www.sun.com if you need additional information or
 * have any questions.
 *
 */

//
// Free block maintenance for Concurrent Mark Sweep Generation
//
// The main data structure for free blocks are
// . an indexed array of small free blocks, and
// . a dictionary of large free blocks
//

// No virtuals in FreeChunk (don't want any vtables).

// A FreeChunk is merely a chunk that can be in a doubly linked list
// and has a size field. NOTE: FreeChunks are distinguished from allocated
// objects in two ways (by the sweeper), depending on whether the VM is 32 or
// 64 bits.
// In 32 bits or 64 bits without CompressedOops, the second word (prev) has the
// LSB set to indicate a free chunk; allocated objects' klass() pointers
// don't have their LSB set. The corresponding bit in the CMSBitMap is
// set when the chunk is allocated. There are also blocks that "look free"
// but are not part of the free list and should not be coalesced into larger
// free blocks. These free blocks have their two LSB's set.

class FreeChunk VALUE_OBJ_CLASS_SPEC {
  friend class VMStructs;
  // For 64 bit compressed oops, the markOop encodes both the size and the
  // indication that this is a FreeChunk and not an object.
  volatile size_t   _size;
  FreeChunk* _prev;
  FreeChunk* _next;

  markOop mark()     const volatile { return (markOop)_size; }
  void set_mark(markOop m)          { _size = (size_t)m; }

 public:
  NOT_PRODUCT(static const size_t header_size();)

  // Returns "true" if the address indicates that the block represents
  // a free chunk.
  static bool indicatesFreeChunk(const HeapWord* addr) {
    // Force volatile read from addr because value might change between
    // calls.  We really want the read of _mark and _prev from this pointer
    // to be volatile but making the fields volatile causes all sorts of
    // compilation errors.
    return ((volatile FreeChunk*)addr)->isFree();
  }

  bool isFree() const volatile {
    LP64_ONLY(if (UseCompressedOops) return mark()->is_cms_free_chunk(); else)
    return (((intptr_t)_prev) & 0x1) == 0x1;
  }
  bool cantCoalesce() const {
    assert(isFree(), "can't get coalesce bit on not free");
    return (((intptr_t)_prev) & 0x2) == 0x2;
  }
  void dontCoalesce() {
    // the block should be free
    assert(isFree(), "Should look like a free block");
    _prev = (FreeChunk*)(((intptr_t)_prev) | 0x2);
  }
  FreeChunk* prev() const {
    return (FreeChunk*)(((intptr_t)_prev) & ~(0x3));
  }

  debug_only(void* prev_addr() const { return (void*)&_prev; })
  debug_only(void* next_addr() const { return (void*)&_next; })
  debug_only(void* size_addr() const { return (void*)&_size; })

  size_t size() const volatile {
    LP64_ONLY(if (UseCompressedOops) return mark()->get_size(); else )
    return _size;
  }
  void setSize(size_t sz) {
    LP64_ONLY(if (UseCompressedOops) set_mark(markOopDesc::set_size_and_free(sz)); else )
    _size = sz;
  }

  FreeChunk* next()   const { return _next; }

  void linkAfter(FreeChunk* ptr) {
    linkNext(ptr);
    if (ptr != NULL) ptr->linkPrev(this);
  }
  void linkAfterNonNull(FreeChunk* ptr) {
    assert(ptr != NULL, "precondition violation");
    linkNext(ptr);
    ptr->linkPrev(this);
  }
  void linkNext(FreeChunk* ptr) { _next = ptr; }
  void linkPrev(FreeChunk* ptr) {
     LP64_ONLY(if (UseCompressedOops) _prev = ptr; else)
     _prev = (FreeChunk*)((intptr_t)ptr | 0x1);
  }
  void clearPrev()              { _prev = NULL; }
  void clearNext()              { _next = NULL; }
  void markNotFree() {
   LP64_ONLY(if (UseCompressedOops) set_mark(markOopDesc::prototype());)
   // Also set _prev to null
   _prev = NULL;
  }

  // Return the address past the end of this chunk
  HeapWord* end() const { return ((HeapWord*) this) + size(); }

  // debugging
  void verify()             const PRODUCT_RETURN;
  void verifyList()         const PRODUCT_RETURN;
  void mangleAllocated(size_t size) PRODUCT_RETURN;
  void mangleFreed(size_t size)     PRODUCT_RETURN;

  void print_on(outputStream* st);
};

// Alignment helpers etc.
#define numQuanta(x,y) ((x+y-1)/y)
enum AlignmentConstants {
  MinChunkSize = numQuanta(sizeof(FreeChunk), MinObjAlignmentInBytes) * MinObjAlignment
};