Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
PojavLauncherTeam
GitHub Repository: PojavLauncherTeam/mobile
Path: blob/master/src/hotspot/share/gc/shared/adaptiveSizePolicy.hpp
40957 views
1
/*
2
* Copyright (c) 2004, 2019, Oracle and/or its affiliates. All rights reserved.
3
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4
*
5
* This code is free software; you can redistribute it and/or modify it
6
* under the terms of the GNU General Public License version 2 only, as
7
* published by the Free Software Foundation.
8
*
9
* This code is distributed in the hope that it will be useful, but WITHOUT
10
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12
* version 2 for more details (a copy is included in the LICENSE file that
13
* accompanied this code).
14
*
15
* You should have received a copy of the GNU General Public License version
16
* 2 along with this work; if not, write to the Free Software Foundation,
17
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18
*
19
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20
* or visit www.oracle.com if you need additional information or have any
21
* questions.
22
*
23
*/
24
25
#ifndef SHARE_GC_SHARED_ADAPTIVESIZEPOLICY_HPP
26
#define SHARE_GC_SHARED_ADAPTIVESIZEPOLICY_HPP
27
28
#include "gc/shared/gcCause.hpp"
29
#include "gc/shared/gcOverheadChecker.hpp"
30
#include "gc/shared/gcUtil.hpp"
31
#include "memory/allocation.hpp"
32
33
// This class keeps statistical information and computes the
34
// size of the heap.
35
36
// Forward decls
37
class elapsedTimer;
38
39
class AdaptiveSizePolicy : public CHeapObj<mtGC> {
40
friend class GCAdaptivePolicyCounters;
41
friend class PSGCAdaptivePolicyCounters;
42
protected:
43
44
enum GCPolicyKind {
45
_gc_adaptive_size_policy,
46
_gc_ps_adaptive_size_policy
47
};
48
virtual GCPolicyKind kind() const { return _gc_adaptive_size_policy; }
49
50
enum SizePolicyTrueValues {
51
decrease_old_gen_for_throughput_true = -7,
52
decrease_young_gen_for_througput_true = -6,
53
54
increase_old_gen_for_min_pauses_true = -5,
55
decrease_old_gen_for_min_pauses_true = -4,
56
decrease_young_gen_for_maj_pauses_true = -3,
57
increase_young_gen_for_min_pauses_true = -2,
58
increase_old_gen_for_maj_pauses_true = -1,
59
60
decrease_young_gen_for_min_pauses_true = 1,
61
decrease_old_gen_for_maj_pauses_true = 2,
62
increase_young_gen_for_maj_pauses_true = 3,
63
64
increase_old_gen_for_throughput_true = 4,
65
increase_young_gen_for_througput_true = 5,
66
67
decrease_young_gen_for_footprint_true = 6,
68
decrease_old_gen_for_footprint_true = 7,
69
decide_at_full_gc_true = 8
70
};
71
72
// Goal for the fraction of the total time during which application
73
// threads run
74
const double _throughput_goal;
75
76
// Last calculated sizes, in bytes, and aligned
77
size_t _eden_size; // calculated eden free space in bytes
78
size_t _promo_size; // calculated promoted free space in bytes
79
80
size_t _survivor_size; // calculated survivor size in bytes
81
82
// Support for UseGCOverheadLimit
83
GCOverheadChecker _overhead_checker;
84
85
// Minor collection timers used to determine both
86
// pause and interval times for collections
87
static elapsedTimer _minor_timer;
88
89
// Major collection timers, used to determine both
90
// pause and interval times for collections
91
static elapsedTimer _major_timer;
92
93
// Time statistics
94
AdaptivePaddedAverage* _avg_minor_pause;
95
AdaptiveWeightedAverage* _avg_minor_interval;
96
AdaptiveWeightedAverage* _avg_minor_gc_cost;
97
98
AdaptiveWeightedAverage* _avg_major_interval;
99
AdaptiveWeightedAverage* _avg_major_gc_cost;
100
101
// Footprint statistics
102
AdaptiveWeightedAverage* _avg_young_live;
103
AdaptiveWeightedAverage* _avg_eden_live;
104
AdaptiveWeightedAverage* _avg_old_live;
105
106
// Statistics for survivor space calculation for young generation
107
AdaptivePaddedAverage* _avg_survived;
108
109
// Objects that have been directly allocated in the old generation
110
AdaptivePaddedNoZeroDevAverage* _avg_pretenured;
111
112
// Variable for estimating the major and minor pause times.
113
// These variables represent linear least-squares fits of
114
// the data.
115
// minor pause time vs. old gen size
116
LinearLeastSquareFit* _minor_pause_old_estimator;
117
// minor pause time vs. young gen size
118
LinearLeastSquareFit* _minor_pause_young_estimator;
119
120
// Variables for estimating the major and minor collection costs
121
// minor collection time vs. young gen size
122
LinearLeastSquareFit* _minor_collection_estimator;
123
// major collection time vs. old gen size
124
LinearLeastSquareFit* _major_collection_estimator;
125
126
// These record the most recent collection times. They
127
// are available as an alternative to using the averages
128
// for making ergonomic decisions.
129
double _latest_minor_mutator_interval_seconds;
130
131
// Allowed difference between major and minor GC times, used
132
// for computing tenuring_threshold
133
const double _threshold_tolerance_percent;
134
135
const double _gc_pause_goal_sec; // Goal for maximum GC pause
136
137
// Flag indicating that the adaptive policy is ready to use
138
bool _young_gen_policy_is_ready;
139
140
// Decrease/increase the young generation for minor pause time
141
int _change_young_gen_for_min_pauses;
142
143
// Decrease/increase the old generation for major pause time
144
int _change_old_gen_for_maj_pauses;
145
146
// change old generation for throughput
147
int _change_old_gen_for_throughput;
148
149
// change young generation for throughput
150
int _change_young_gen_for_throughput;
151
152
// Flag indicating that the policy would
153
// increase the tenuring threshold because of the total major GC cost
154
// is greater than the total minor GC cost
155
bool _increment_tenuring_threshold_for_gc_cost;
156
// decrease the tenuring threshold because of the the total minor GC
157
// cost is greater than the total major GC cost
158
bool _decrement_tenuring_threshold_for_gc_cost;
159
// decrease due to survivor size limit
160
bool _decrement_tenuring_threshold_for_survivor_limit;
161
162
// decrease generation sizes for footprint
163
int _decrease_for_footprint;
164
165
// Set if the ergonomic decisions were made at a full GC.
166
int _decide_at_full_gc;
167
168
// Changing the generation sizing depends on the data that is
169
// gathered about the effects of changes on the pause times and
170
// throughput. These variable count the number of data points
171
// gathered. The policy may use these counters as a threshold
172
// for reliable data.
173
julong _young_gen_change_for_minor_throughput;
174
julong _old_gen_change_for_major_throughput;
175
176
// Accessors
177
178
double gc_pause_goal_sec() const { return _gc_pause_goal_sec; }
179
// The value returned is unitless: it's the proportion of time
180
// spent in a particular collection type.
181
// An interval time will be 0.0 if a collection type hasn't occurred yet.
182
// The 1.4.2 implementation put a floor on the values of major_gc_cost
183
// and minor_gc_cost. This was useful because of the way major_gc_cost
184
// and minor_gc_cost was used in calculating the sizes of the generations.
185
// Do not use a floor in this implementation because any finite value
186
// will put a limit on the throughput that can be achieved and any
187
// throughput goal above that limit will drive the generations sizes
188
// to extremes.
189
double major_gc_cost() const {
190
return MAX2(0.0F, _avg_major_gc_cost->average());
191
}
192
193
// The value returned is unitless: it's the proportion of time
194
// spent in a particular collection type.
195
// An interval time will be 0.0 if a collection type hasn't occurred yet.
196
// The 1.4.2 implementation put a floor on the values of major_gc_cost
197
// and minor_gc_cost. This was useful because of the way major_gc_cost
198
// and minor_gc_cost was used in calculating the sizes of the generations.
199
// Do not use a floor in this implementation because any finite value
200
// will put a limit on the throughput that can be achieved and any
201
// throughput goal above that limit will drive the generations sizes
202
// to extremes.
203
204
double minor_gc_cost() const {
205
return MAX2(0.0F, _avg_minor_gc_cost->average());
206
}
207
208
// Because we're dealing with averages, gc_cost() can be
209
// larger than 1.0 if just the sum of the minor cost the
210
// the major cost is used. Worse than that is the
211
// fact that the minor cost and the major cost each
212
// tend toward 1.0 in the extreme of high GC costs.
213
// Limit the value of gc_cost to 1.0 so that the mutator
214
// cost stays non-negative.
215
virtual double gc_cost() const {
216
double result = MIN2(1.0, minor_gc_cost() + major_gc_cost());
217
assert(result >= 0.0, "Both minor and major costs are non-negative");
218
return result;
219
}
220
221
// Elapsed time since the last major collection.
222
virtual double time_since_major_gc() const;
223
224
// Average interval between major collections to be used
225
// in calculating the decaying major GC cost. An overestimate
226
// of this time would be a conservative estimate because
227
// this time is used to decide if the major GC cost
228
// should be decayed (i.e., if the time since the last
229
// major GC is long compared to the time returned here,
230
// then the major GC cost will be decayed). See the
231
// implementations for the specifics.
232
virtual double major_gc_interval_average_for_decay() const {
233
return _avg_major_interval->average();
234
}
235
236
// Return the cost of the GC where the major GC cost
237
// has been decayed based on the time since the last
238
// major collection.
239
double decaying_gc_cost() const;
240
241
// Decay the major GC cost. Use this only for decisions on
242
// whether to adjust, not to determine by how much to adjust.
243
// This approximation is crude and may not be good enough for the
244
// latter.
245
double decaying_major_gc_cost() const;
246
247
// Return the mutator cost using the decayed
248
// GC cost.
249
double adjusted_mutator_cost() const {
250
double result = 1.0 - decaying_gc_cost();
251
assert(result >= 0.0, "adjusted mutator cost calculation is incorrect");
252
return result;
253
}
254
255
virtual double mutator_cost() const {
256
double result = 1.0 - gc_cost();
257
assert(result >= 0.0, "mutator cost calculation is incorrect");
258
return result;
259
}
260
261
262
bool young_gen_policy_is_ready() { return _young_gen_policy_is_ready; }
263
264
void update_minor_pause_young_estimator(double minor_pause_in_ms);
265
virtual void update_minor_pause_old_estimator(double minor_pause_in_ms) {
266
// This is not meaningful for all policies but needs to be present
267
// to use minor_collection_end() in its current form.
268
}
269
270
virtual size_t eden_increment(size_t cur_eden);
271
virtual size_t eden_increment(size_t cur_eden, uint percent_change);
272
virtual size_t eden_decrement(size_t cur_eden);
273
virtual size_t promo_increment(size_t cur_eden);
274
virtual size_t promo_increment(size_t cur_eden, uint percent_change);
275
virtual size_t promo_decrement(size_t cur_eden);
276
277
virtual void clear_generation_free_space_flags();
278
279
int change_old_gen_for_throughput() const {
280
return _change_old_gen_for_throughput;
281
}
282
void set_change_old_gen_for_throughput(int v) {
283
_change_old_gen_for_throughput = v;
284
}
285
int change_young_gen_for_throughput() const {
286
return _change_young_gen_for_throughput;
287
}
288
void set_change_young_gen_for_throughput(int v) {
289
_change_young_gen_for_throughput = v;
290
}
291
292
int change_old_gen_for_maj_pauses() const {
293
return _change_old_gen_for_maj_pauses;
294
}
295
void set_change_old_gen_for_maj_pauses(int v) {
296
_change_old_gen_for_maj_pauses = v;
297
}
298
299
bool decrement_tenuring_threshold_for_gc_cost() const {
300
return _decrement_tenuring_threshold_for_gc_cost;
301
}
302
void set_decrement_tenuring_threshold_for_gc_cost(bool v) {
303
_decrement_tenuring_threshold_for_gc_cost = v;
304
}
305
bool increment_tenuring_threshold_for_gc_cost() const {
306
return _increment_tenuring_threshold_for_gc_cost;
307
}
308
void set_increment_tenuring_threshold_for_gc_cost(bool v) {
309
_increment_tenuring_threshold_for_gc_cost = v;
310
}
311
bool decrement_tenuring_threshold_for_survivor_limit() const {
312
return _decrement_tenuring_threshold_for_survivor_limit;
313
}
314
void set_decrement_tenuring_threshold_for_survivor_limit(bool v) {
315
_decrement_tenuring_threshold_for_survivor_limit = v;
316
}
317
// Return true if the policy suggested a change.
318
bool tenuring_threshold_change() const;
319
320
public:
321
AdaptiveSizePolicy(size_t init_eden_size,
322
size_t init_promo_size,
323
size_t init_survivor_size,
324
double gc_pause_goal_sec,
325
uint gc_cost_ratio);
326
327
bool is_gc_ps_adaptive_size_policy() {
328
return kind() == _gc_ps_adaptive_size_policy;
329
}
330
331
AdaptivePaddedAverage* avg_minor_pause() const { return _avg_minor_pause; }
332
AdaptiveWeightedAverage* avg_minor_interval() const {
333
return _avg_minor_interval;
334
}
335
AdaptiveWeightedAverage* avg_minor_gc_cost() const {
336
return _avg_minor_gc_cost;
337
}
338
339
AdaptiveWeightedAverage* avg_major_gc_cost() const {
340
return _avg_major_gc_cost;
341
}
342
343
AdaptiveWeightedAverage* avg_young_live() const { return _avg_young_live; }
344
AdaptiveWeightedAverage* avg_eden_live() const { return _avg_eden_live; }
345
AdaptiveWeightedAverage* avg_old_live() const { return _avg_old_live; }
346
347
AdaptivePaddedAverage* avg_survived() const { return _avg_survived; }
348
AdaptivePaddedNoZeroDevAverage* avg_pretenured() { return _avg_pretenured; }
349
350
// Methods indicating events of interest to the adaptive size policy,
351
// called by GC algorithms. It is the responsibility of users of this
352
// policy to call these methods at the correct times!
353
virtual void minor_collection_begin();
354
virtual void minor_collection_end(GCCause::Cause gc_cause);
355
virtual LinearLeastSquareFit* minor_pause_old_estimator() const {
356
return _minor_pause_old_estimator;
357
}
358
359
LinearLeastSquareFit* minor_pause_young_estimator() {
360
return _minor_pause_young_estimator;
361
}
362
LinearLeastSquareFit* minor_collection_estimator() {
363
return _minor_collection_estimator;
364
}
365
366
LinearLeastSquareFit* major_collection_estimator() {
367
return _major_collection_estimator;
368
}
369
370
float minor_pause_young_slope() {
371
return _minor_pause_young_estimator->slope();
372
}
373
374
float minor_collection_slope() { return _minor_collection_estimator->slope();}
375
float major_collection_slope() { return _major_collection_estimator->slope();}
376
377
float minor_pause_old_slope() {
378
return _minor_pause_old_estimator->slope();
379
}
380
381
void set_eden_size(size_t new_size) {
382
_eden_size = new_size;
383
}
384
void set_survivor_size(size_t new_size) {
385
_survivor_size = new_size;
386
}
387
388
size_t calculated_eden_size_in_bytes() const {
389
return _eden_size;
390
}
391
392
size_t calculated_promo_size_in_bytes() const {
393
return _promo_size;
394
}
395
396
size_t calculated_survivor_size_in_bytes() const {
397
return _survivor_size;
398
}
399
400
bool gc_overhead_limit_exceeded() {
401
return _overhead_checker.gc_overhead_limit_exceeded();
402
}
403
void set_gc_overhead_limit_exceeded(bool v) {
404
_overhead_checker.set_gc_overhead_limit_exceeded(v);
405
}
406
407
bool gc_overhead_limit_near() {
408
return _overhead_checker.gc_overhead_limit_near();
409
}
410
411
void reset_gc_overhead_limit_count() {
412
_overhead_checker.reset_gc_overhead_limit_count();
413
}
414
// accessors for flags recording the decisions to resize the
415
// generations to meet the pause goal.
416
417
int change_young_gen_for_min_pauses() const {
418
return _change_young_gen_for_min_pauses;
419
}
420
void set_change_young_gen_for_min_pauses(int v) {
421
_change_young_gen_for_min_pauses = v;
422
}
423
void set_decrease_for_footprint(int v) { _decrease_for_footprint = v; }
424
int decrease_for_footprint() const { return _decrease_for_footprint; }
425
int decide_at_full_gc() { return _decide_at_full_gc; }
426
void set_decide_at_full_gc(int v) { _decide_at_full_gc = v; }
427
428
// Check the conditions for an out-of-memory due to excessive GC time.
429
// Set _gc_overhead_limit_exceeded if all the conditions have been met.
430
void check_gc_overhead_limit(size_t eden_live,
431
size_t max_old_gen_size,
432
size_t max_eden_size,
433
bool is_full_gc,
434
GCCause::Cause gc_cause,
435
SoftRefPolicy* soft_ref_policy);
436
437
static bool should_update_promo_stats(GCCause::Cause cause) {
438
return ((GCCause::is_user_requested_gc(cause) &&
439
UseAdaptiveSizePolicyWithSystemGC) ||
440
GCCause::is_tenured_allocation_failure_gc(cause));
441
}
442
443
static bool should_update_eden_stats(GCCause::Cause cause) {
444
return ((GCCause::is_user_requested_gc(cause) &&
445
UseAdaptiveSizePolicyWithSystemGC) ||
446
GCCause::is_allocation_failure_gc(cause));
447
}
448
449
// Printing support
450
virtual bool print() const;
451
void print_tenuring_threshold(uint new_tenuring_threshold) const;
452
};
453
454
#endif // SHARE_GC_SHARED_ADAPTIVESIZEPOLICY_HPP
455
456