jdk8-mips64-public/hotspot: src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp@e018e6884bd8 (annotated)

src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp@e018e6884bd8 (annotated)

src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp

Wed, 23 Dec 2009 09:23:54 -0800

author: ysr
date: Wed, 23 Dec 2009 09:23:54 -0800
changeset 1580: e018e6884bd8
parent 1040: 98cb887364d3
child 1907: c18cbe5936b8
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 2004-2006 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.
  *
  */
 // This class keeps statistical information and computes the
 // size of the heap for the concurrent mark sweep collector.
 //
 // Cost for garbage collector include cost for
 //   minor collection
 //   concurrent collection
 //      stop-the-world component
 //      concurrent component
 //   major compacting collection
 //      uses decaying cost
 // Forward decls
 class elapsedTimer;
 class CMSAdaptiveSizePolicy : public AdaptiveSizePolicy {
  friend class CMSGCAdaptivePolicyCounters;
  friend class CMSCollector;
  private:
   // Total number of processors available
   int _processor_count;
   // Number of processors used by the concurrent phases of GC
   // This number is assumed to be the same for all concurrent
   // phases.
   int _concurrent_processor_count;
   // Time that the mutators run exclusive of a particular
   // phase.  For example, the time the mutators run excluding
   // the time during which the cms collector runs concurrently
   // with the mutators.
   //   Between end of most recent cms reset and start of initial mark
                 // This may be redundant
   double _latest_cms_reset_end_to_initial_mark_start_secs;
   //   Between end of the most recent initial mark and start of remark
   double _latest_cms_initial_mark_end_to_remark_start_secs;
   //   Between end of most recent collection and start of
   //   a concurrent collection
   double _latest_cms_collection_end_to_collection_start_secs;
   //   Times of the concurrent phases of the most recent
   //   concurrent collection
   double _latest_cms_concurrent_marking_time_secs;
   double _latest_cms_concurrent_precleaning_time_secs;
   double _latest_cms_concurrent_sweeping_time_secs;
   //   Between end of most recent STW MSC and start of next STW MSC
   double _latest_cms_msc_end_to_msc_start_time_secs;
   //   Between end of most recent MS and start of next MS
   //   This does not include any time spent during a concurrent
   // collection.
   double _latest_cms_ms_end_to_ms_start;
   //   Between start and end of the initial mark of the most recent
   // concurrent collection.
   double _latest_cms_initial_mark_start_to_end_time_secs;
   //   Between start and end of the remark phase of the most recent
   // concurrent collection
   double _latest_cms_remark_start_to_end_time_secs;
   //   Between start and end of the most recent MS STW marking phase
   double _latest_cms_ms_marking_start_to_end_time_secs;
   // Pause time timers
   static elapsedTimer _STW_timer;
   // Concurrent collection timer.  Used for total of all concurrent phases
   // during 1 collection cycle.
   static elapsedTimer _concurrent_timer;
   // When the size of the generation is changed, the size
   // of the change will rounded up or down (depending on the
   // type of change) by this value.
   size_t _generation_alignment;
   // If this variable is true, the size of the young generation
   // may be changed in order to reduce the pause(s) of the
   // collection of the tenured generation in order to meet the
   // pause time goal.  It is common to change the size of the
   // tenured generation in order to meet the pause time goal
   // for the tenured generation.  With the CMS collector for
   // the tenured generation, the size of the young generation
   // can have an significant affect on the pause times for collecting the
   // tenured generation.
   // This is a duplicate of a variable in PSAdaptiveSizePolicy.  It
   // is duplicated because it is not clear that it is general enough
   // to go into AdaptiveSizePolicy.
   int _change_young_gen_for_maj_pauses;
   // Variable that is set to true after a collection.
   bool _first_after_collection;
   // Fraction of collections that are of each type
   double concurrent_fraction() const;
   double STW_msc_fraction() const;
   double STW_ms_fraction() const;
   // This call cannot be put into the epilogue as long as some
   // of the counters can be set during concurrent phases.
   virtual void clear_generation_free_space_flags();
   void set_first_after_collection() { _first_after_collection = true; }
  protected:
   // Average of the sum of the concurrent times for
   // one collection in seconds.
   AdaptiveWeightedAverage* _avg_concurrent_time;
   // Average time between concurrent collections in seconds.
   AdaptiveWeightedAverage* _avg_concurrent_interval;
   // Average cost of the concurrent part of a collection
   // in seconds.
   AdaptiveWeightedAverage* _avg_concurrent_gc_cost;
   // Average of the initial pause of a concurrent collection in seconds.
   AdaptivePaddedAverage* _avg_initial_pause;
   // Average of the remark pause of a concurrent collection in seconds.
   AdaptivePaddedAverage* _avg_remark_pause;
   // Average of the stop-the-world (STW) (initial mark + remark)
   // times in seconds for concurrent collections.
   AdaptiveWeightedAverage* _avg_cms_STW_time;
   // Average of the STW collection cost for concurrent collections.
   AdaptiveWeightedAverage* _avg_cms_STW_gc_cost;
   // Average of the bytes free at the start of the sweep.
   AdaptiveWeightedAverage* _avg_cms_free_at_sweep;
   // Average of the bytes free at the end of the collection.
   AdaptiveWeightedAverage* _avg_cms_free;
   // Average of the bytes promoted between cms collections.
   AdaptiveWeightedAverage* _avg_cms_promo;
   // stop-the-world (STW) mark-sweep-compact
   // Average of the pause time in seconds for STW mark-sweep-compact
   // collections.
   AdaptiveWeightedAverage* _avg_msc_pause;
   // Average of the interval in seconds between STW mark-sweep-compact
   // collections.
   AdaptiveWeightedAverage* _avg_msc_interval;
   // Average of the collection costs for STW mark-sweep-compact
   // collections.
   AdaptiveWeightedAverage* _avg_msc_gc_cost;
   // Averages for mark-sweep collections.
   // The collection may have started as a background collection
   // that completes in a stop-the-world (STW) collection.
   // Average of the pause time in seconds for mark-sweep
   // collections.
   AdaptiveWeightedAverage* _avg_ms_pause;
   // Average of the interval in seconds between mark-sweep
   // collections.
   AdaptiveWeightedAverage* _avg_ms_interval;
   // Average of the collection costs for mark-sweep
   // collections.
   AdaptiveWeightedAverage* _avg_ms_gc_cost;
   // These variables contain a linear fit of
   // a generation size as the independent variable
   // and a pause time as the dependent variable.
   // For example _remark_pause_old_estimator
   // is a fit of the old generation size as the
   // independent variable and the remark pause
   // as the dependent variable.
   //   remark pause time vs. cms gen size
   LinearLeastSquareFit* _remark_pause_old_estimator;
   //   initial pause time vs. cms gen size
   LinearLeastSquareFit* _initial_pause_old_estimator;
   //   remark pause time vs. young gen size
   LinearLeastSquareFit* _remark_pause_young_estimator;
   //   initial pause time vs. young gen size
   LinearLeastSquareFit* _initial_pause_young_estimator;
   // Accessors
   int processor_count() const { return _processor_count; }
   int concurrent_processor_count() const { return _concurrent_processor_count; }
   AdaptiveWeightedAverage* avg_concurrent_time() const {
     return _avg_concurrent_time;
   }
   AdaptiveWeightedAverage* avg_concurrent_interval() const {
     return _avg_concurrent_interval;
   }
   AdaptiveWeightedAverage* avg_concurrent_gc_cost() const {
     return _avg_concurrent_gc_cost;
   }
   AdaptiveWeightedAverage* avg_cms_STW_time() const {
     return _avg_cms_STW_time;
   }
   AdaptiveWeightedAverage* avg_cms_STW_gc_cost() const {
     return _avg_cms_STW_gc_cost;
   }
   AdaptivePaddedAverage* avg_initial_pause() const {
     return _avg_initial_pause;
   }
   AdaptivePaddedAverage* avg_remark_pause() const {
     return _avg_remark_pause;
   }
   AdaptiveWeightedAverage* avg_cms_free() const {
     return _avg_cms_free;
   }
   AdaptiveWeightedAverage* avg_cms_free_at_sweep() const {
     return _avg_cms_free_at_sweep;
   }
   AdaptiveWeightedAverage* avg_msc_pause() const {
     return _avg_msc_pause;
   }
   AdaptiveWeightedAverage* avg_msc_interval() const {
     return _avg_msc_interval;
   }
   AdaptiveWeightedAverage* avg_msc_gc_cost() const {
     return _avg_msc_gc_cost;
   }
   AdaptiveWeightedAverage* avg_ms_pause() const {
     return _avg_ms_pause;
   }
   AdaptiveWeightedAverage* avg_ms_interval() const {
     return _avg_ms_interval;
   }
   AdaptiveWeightedAverage* avg_ms_gc_cost() const {
     return _avg_ms_gc_cost;
   }
   LinearLeastSquareFit* remark_pause_old_estimator() {
     return _remark_pause_old_estimator;
   }
   LinearLeastSquareFit* initial_pause_old_estimator() {
     return _initial_pause_old_estimator;
   }
   LinearLeastSquareFit* remark_pause_young_estimator() {
     return _remark_pause_young_estimator;
   }
   LinearLeastSquareFit* initial_pause_young_estimator() {
     return _initial_pause_young_estimator;
   }
   // These *slope() methods return the slope
   // m for the linear fit of an independent
   // variable vs. a dependent variable.  For
   // example
   //  remark_pause = m * old_generation_size + c
   // These may be used to determine if an
   // adjustment should be made to achieve a goal.
   // For example, if remark_pause_old_slope() is
   // positive, a reduction of the old generation
   // size has on average resulted in the reduction
   // of the remark pause.
   float remark_pause_old_slope() {
     return _remark_pause_old_estimator->slope();
   }
   float initial_pause_old_slope() {
     return _initial_pause_old_estimator->slope();
   }
   float remark_pause_young_slope() {
     return _remark_pause_young_estimator->slope();
   }
   float initial_pause_young_slope() {
     return _initial_pause_young_estimator->slope();
   }
   // Update estimators
   void update_minor_pause_old_estimator(double minor_pause_in_ms);
   // Fraction of processors used by the concurrent phases.
   double concurrent_processor_fraction();
   // Returns the total times for the concurrent part of the
   // latest collection in seconds.
   double concurrent_collection_time();
   // Return the total times for the concurrent part of the
   // latest collection in seconds where the times of the various
   // concurrent phases are scaled by the processor fraction used
   // during the phase.
   double scaled_concurrent_collection_time();
   // Dimensionless concurrent GC cost for all the concurrent phases.
   double concurrent_collection_cost(double interval_in_seconds);
   // Dimensionless GC cost
   double collection_cost(double pause_in_seconds, double interval_in_seconds);
   virtual GCPolicyKind kind() const { return _gc_cms_adaptive_size_policy; }
   virtual double time_since_major_gc() const;
   // This returns the maximum average for the concurrent, ms, and
   // msc collections.  This is meant to be used for the calculation
   // of the decayed major gc cost and is not in general the
   // average of all the different types of major collections.
   virtual double major_gc_interval_average_for_decay() const;
  public:
   CMSAdaptiveSizePolicy(size_t init_eden_size,
                         size_t init_promo_size,
                         size_t init_survivor_size,
                         double max_gc_minor_pause_sec,
                         double max_gc_pause_sec,
                         uint gc_cost_ratio);
   // The timers for the stop-the-world phases measure a total
   // stop-the-world time.  The timer is started and stopped
   // for each phase but is only reset after the final checkpoint.
   void checkpoint_roots_initial_begin();
   void checkpoint_roots_initial_end(GCCause::Cause gc_cause);
   void checkpoint_roots_final_begin();
   void checkpoint_roots_final_end(GCCause::Cause gc_cause);
   // Methods for gathering information about the
   // concurrent marking phase of the collection.
   // Records the mutator times and
   // resets the concurrent timer.
   void concurrent_marking_begin();
   // Resets concurrent phase timer in the begin methods and
   // saves the time for a phase in the end methods.
   void concurrent_marking_end();
   void concurrent_sweeping_begin();
   void concurrent_sweeping_end();
   // Similar to the above (e.g., concurrent_marking_end()) and
   // is used for both the precleaning an abortable precleaing
   // phases.
   void concurrent_precleaning_begin();
   void concurrent_precleaning_end();
   // Stops the concurrent phases time.  Gathers
   // information and resets the timer.
   void concurrent_phases_end(GCCause::Cause gc_cause,
                               size_t cur_eden,
                               size_t cur_promo);
   // Methods for gather information about STW Mark-Sweep-Compact
   void msc_collection_begin();
   void msc_collection_end(GCCause::Cause gc_cause);
   // Methods for gather information about Mark-Sweep done
   // in the foreground.
   void ms_collection_begin();
   void ms_collection_end(GCCause::Cause gc_cause);
   // Cost for a mark-sweep tenured gen collection done in the foreground
   double ms_gc_cost() const {
     return MAX2(0.0F, _avg_ms_gc_cost->average());
   }
   // Cost of collecting the tenured generation.  Includes
   // concurrent collection and STW collection costs
   double cms_gc_cost() const;
   // Cost of STW mark-sweep-compact tenured gen collection.
   double msc_gc_cost() const {
     return MAX2(0.0F, _avg_msc_gc_cost->average());
   }
   //
   double compacting_gc_cost() const {
     double result = MIN2(1.0, minor_gc_cost() + msc_gc_cost());
     assert(result >= 0.0, "Both minor and major costs are non-negative");
     return result;
   }
    // Restarts the concurrent phases timer.
    void concurrent_phases_resume();
    // Time beginning and end of the marking phase for
    // a synchronous MS collection.  A MS collection
    // that finishes in the foreground can have started
    // in the background.  These methods capture the
    // completion of the marking (after the initial
    // marking) that is done in the foreground.
    void ms_collection_marking_begin();
    void ms_collection_marking_end(GCCause::Cause gc_cause);
    static elapsedTimer* concurrent_timer_ptr() {
      return &_concurrent_timer;
    }
   AdaptiveWeightedAverage* avg_cms_promo() const {
     return _avg_cms_promo;
   }
   int change_young_gen_for_maj_pauses() {
     return _change_young_gen_for_maj_pauses;
   }
   void set_change_young_gen_for_maj_pauses(int v) {
     _change_young_gen_for_maj_pauses = v;
   }
   void clear_internal_time_intervals();
   // Either calculated_promo_size_in_bytes() or promo_size()
   // should be deleted.
   size_t promo_size() { return _promo_size; }
   void set_promo_size(size_t v) { _promo_size = v; }
   // Cost of GC for all types of collections.
   virtual double gc_cost() const;
   size_t generation_alignment() { return _generation_alignment; }
   virtual void compute_young_generation_free_space(size_t cur_eden,
                                                    size_t max_eden_size);
   // Calculates new survivor space size;  returns a new tenuring threshold
   // value. Stores new survivor size in _survivor_size.
   virtual int compute_survivor_space_size_and_threshold(
                                                 bool   is_survivor_overflow,
                                                 int    tenuring_threshold,
                                                 size_t survivor_limit);
   virtual void compute_tenured_generation_free_space(size_t cur_tenured_free,
                                            size_t max_tenured_available,
                                            size_t cur_eden);
   size_t eden_decrement_aligned_down(size_t cur_eden);
   size_t eden_increment_aligned_up(size_t cur_eden);
   size_t adjust_eden_for_pause_time(size_t cur_eden);
   size_t adjust_eden_for_throughput(size_t cur_eden);
   size_t adjust_eden_for_footprint(size_t cur_eden);
   size_t promo_decrement_aligned_down(size_t cur_promo);
   size_t promo_increment_aligned_up(size_t cur_promo);
   size_t adjust_promo_for_pause_time(size_t cur_promo);
   size_t adjust_promo_for_throughput(size_t cur_promo);
   size_t adjust_promo_for_footprint(size_t cur_promo, size_t cur_eden);
   // Scale down the input size by the ratio of the cost to collect the
   // generation to the total GC cost.
   size_t scale_by_gen_gc_cost(size_t base_change, double gen_gc_cost);
   // Return the value and clear it.
   bool get_and_clear_first_after_collection();
   // Printing support
   virtual bool print_adaptive_size_policy_on(outputStream* st) const;
 };

Mercurial > jdk8-mips64-public > hotspot / annotate

src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp@e018e6884bd8 (annotated)

src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp