1 /* 2 * Copyright (c) 2001, 2011, 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 #include "precompiled.hpp" 26 #include "gc_implementation/g1/concurrentG1Refine.hpp" 27 #include "gc_implementation/g1/concurrentMark.hpp" 28 #include "gc_implementation/g1/concurrentMarkThread.inline.hpp" 29 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp" 30 #include "gc_implementation/g1/g1CollectorPolicy.hpp" 31 #include "gc_implementation/g1/heapRegionRemSet.hpp" 32 #include "gc_implementation/shared/gcPolicyCounters.hpp" 33 #include "runtime/arguments.hpp" 34 #include "runtime/java.hpp" 35 #include "runtime/mutexLocker.hpp" 36 #include "utilities/debug.hpp" 37 38 #define PREDICTIONS_VERBOSE 0 39 40 // <NEW PREDICTION> 41 42 // Different defaults for different number of GC threads 43 // They were chosen by running GCOld and SPECjbb on debris with different 44 // numbers of GC threads and choosing them based on the results 45 46 // all the same 47 static double rs_length_diff_defaults[] = { 48 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 49 }; 50 51 static double cost_per_card_ms_defaults[] = { 52 0.01, 0.005, 0.005, 0.003, 0.003, 0.002, 0.002, 0.0015 53 }; 54 55 // all the same 56 static double fully_young_cards_per_entry_ratio_defaults[] = { 57 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 58 }; 59 60 static double cost_per_entry_ms_defaults[] = { 61 0.015, 0.01, 0.01, 0.008, 0.008, 0.0055, 0.0055, 0.005 62 }; 63 64 static double cost_per_byte_ms_defaults[] = { 65 0.00006, 0.00003, 0.00003, 0.000015, 0.000015, 0.00001, 0.00001, 0.000009 66 }; 67 68 // these should be pretty consistent 69 static double constant_other_time_ms_defaults[] = { 70 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0, 5.0 71 }; 72 73 74 static double young_other_cost_per_region_ms_defaults[] = { 75 0.3, 0.2, 0.2, 0.15, 0.15, 0.12, 0.12, 0.1 76 }; 77 78 static double non_young_other_cost_per_region_ms_defaults[] = { 79 1.0, 0.7, 0.7, 0.5, 0.5, 0.42, 0.42, 0.30 80 }; 81 82 // </NEW PREDICTION> 83 84 // Help class for avoiding interleaved logging 85 class LineBuffer: public StackObj { 86 87 private: 88 static const int BUFFER_LEN = 1024; 89 static const int INDENT_CHARS = 3; 90 char _buffer[BUFFER_LEN]; 91 int _indent_level; 92 int _cur; 93 94 void vappend(const char* format, va_list ap) { 95 int res = vsnprintf(&_buffer[_cur], BUFFER_LEN - _cur, format, ap); 96 if (res != -1) { 97 _cur += res; 98 } else { 99 DEBUG_ONLY(warning("buffer too small in LineBuffer");) 100 _buffer[BUFFER_LEN -1] = 0; 101 _cur = BUFFER_LEN; // vsnprintf above should not add to _buffer if we are called again 102 } 103 } 104 105 public: 106 explicit LineBuffer(int indent_level): _indent_level(indent_level), _cur(0) { 107 for (; (_cur < BUFFER_LEN && _cur < (_indent_level * INDENT_CHARS)); _cur++) { 108 _buffer[_cur] = ' '; 109 } 110 } 111 112 #ifndef PRODUCT 113 ~LineBuffer() { 114 assert(_cur == _indent_level * INDENT_CHARS, "pending data in buffer - append_and_print_cr() not called?"); 115 } 116 #endif 117 118 void append(const char* format, ...) { 119 va_list ap; 120 va_start(ap, format); 121 vappend(format, ap); 122 va_end(ap); 123 } 124 125 void append_and_print_cr(const char* format, ...) { 126 va_list ap; 127 va_start(ap, format); 128 vappend(format, ap); 129 va_end(ap); 130 gclog_or_tty->print_cr("%s", _buffer); 131 _cur = _indent_level * INDENT_CHARS; 132 } 133 }; 134 135 G1CollectorPolicy::G1CollectorPolicy() : 136 _parallel_gc_threads(G1CollectedHeap::use_parallel_gc_threads() 137 ? ParallelGCThreads : 1), 138 139 _n_pauses(0), 140 _recent_rs_scan_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 141 _recent_pause_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 142 _recent_rs_sizes(new TruncatedSeq(NumPrevPausesForHeuristics)), 143 _recent_gc_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 144 _all_pause_times_ms(new NumberSeq()), 145 _stop_world_start(0.0), 146 _all_stop_world_times_ms(new NumberSeq()), 147 _all_yield_times_ms(new NumberSeq()), 148 149 _all_mod_union_times_ms(new NumberSeq()), 150 151 _summary(new Summary()), 152 153 #ifndef PRODUCT 154 _cur_clear_ct_time_ms(0.0), 155 _min_clear_cc_time_ms(-1.0), 156 _max_clear_cc_time_ms(-1.0), 157 _cur_clear_cc_time_ms(0.0), 158 _cum_clear_cc_time_ms(0.0), 159 _num_cc_clears(0L), 160 #endif 161 162 _region_num_young(0), 163 _region_num_tenured(0), 164 _prev_region_num_young(0), 165 _prev_region_num_tenured(0), 166 167 _aux_num(10), 168 _all_aux_times_ms(new NumberSeq[_aux_num]), 169 _cur_aux_start_times_ms(new double[_aux_num]), 170 _cur_aux_times_ms(new double[_aux_num]), 171 _cur_aux_times_set(new bool[_aux_num]), 172 173 _concurrent_mark_init_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 174 _concurrent_mark_remark_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 175 _concurrent_mark_cleanup_times_ms(new TruncatedSeq(NumPrevPausesForHeuristics)), 176 177 // <NEW PREDICTION> 178 179 _alloc_rate_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 180 _prev_collection_pause_end_ms(0.0), 181 _pending_card_diff_seq(new TruncatedSeq(TruncatedSeqLength)), 182 _rs_length_diff_seq(new TruncatedSeq(TruncatedSeqLength)), 183 _cost_per_card_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 184 _fully_young_cards_per_entry_ratio_seq(new TruncatedSeq(TruncatedSeqLength)), 185 _partially_young_cards_per_entry_ratio_seq( 186 new TruncatedSeq(TruncatedSeqLength)), 187 _cost_per_entry_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 188 _partially_young_cost_per_entry_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 189 _cost_per_byte_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 190 _cost_per_byte_ms_during_cm_seq(new TruncatedSeq(TruncatedSeqLength)), 191 _constant_other_time_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 192 _young_other_cost_per_region_ms_seq(new TruncatedSeq(TruncatedSeqLength)), 193 _non_young_other_cost_per_region_ms_seq( 194 new TruncatedSeq(TruncatedSeqLength)), 195 196 _pending_cards_seq(new TruncatedSeq(TruncatedSeqLength)), 197 _scanned_cards_seq(new TruncatedSeq(TruncatedSeqLength)), 198 _rs_lengths_seq(new TruncatedSeq(TruncatedSeqLength)), 199 200 _pause_time_target_ms((double) MaxGCPauseMillis), 201 202 // </NEW PREDICTION> 203 204 _full_young_gcs(true), 205 _full_young_pause_num(0), 206 _partial_young_pause_num(0), 207 208 _during_marking(false), 209 _in_marking_window(false), 210 _in_marking_window_im(false), 211 212 _known_garbage_ratio(0.0), 213 _known_garbage_bytes(0), 214 215 _young_gc_eff_seq(new TruncatedSeq(TruncatedSeqLength)), 216 217 _recent_prev_end_times_for_all_gcs_sec(new TruncatedSeq(NumPrevPausesForHeuristics)), 218 219 _recent_CS_bytes_used_before(new TruncatedSeq(NumPrevPausesForHeuristics)), 220 _recent_CS_bytes_surviving(new TruncatedSeq(NumPrevPausesForHeuristics)), 221 222 _recent_avg_pause_time_ratio(0.0), 223 _num_markings(0), 224 _n_marks(0), 225 _n_pauses_at_mark_end(0), 226 227 _all_full_gc_times_ms(new NumberSeq()), 228 229 // G1PausesBtwnConcMark defaults to -1 230 // so the hack is to do the cast QQQ FIXME 231 _pauses_btwn_concurrent_mark((size_t)G1PausesBtwnConcMark), 232 _n_marks_since_last_pause(0), 233 _initiate_conc_mark_if_possible(false), 234 _during_initial_mark_pause(false), 235 _should_revert_to_full_young_gcs(false), 236 _last_full_young_gc(false), 237 238 _eden_bytes_before_gc(0), 239 _survivor_bytes_before_gc(0), 240 _capacity_before_gc(0), 241 242 _prev_collection_pause_used_at_end_bytes(0), 243 244 _collection_set(NULL), 245 _collection_set_size(0), 246 _collection_set_bytes_used_before(0), 247 248 // Incremental CSet attributes 249 _inc_cset_build_state(Inactive), 250 _inc_cset_head(NULL), 251 _inc_cset_tail(NULL), 252 _inc_cset_size(0), 253 _inc_cset_young_index(0), 254 _inc_cset_bytes_used_before(0), 255 _inc_cset_max_finger(NULL), 256 _inc_cset_recorded_young_bytes(0), 257 _inc_cset_recorded_rs_lengths(0), 258 _inc_cset_predicted_elapsed_time_ms(0.0), 259 _inc_cset_predicted_bytes_to_copy(0), 260 261 #ifdef _MSC_VER // the use of 'this' below gets a warning, make it go away 262 #pragma warning( disable:4355 ) // 'this' : used in base member initializer list 263 #endif // _MSC_VER 264 265 _short_lived_surv_rate_group(new SurvRateGroup(this, "Short Lived", 266 G1YoungSurvRateNumRegionsSummary)), 267 _survivor_surv_rate_group(new SurvRateGroup(this, "Survivor", 268 G1YoungSurvRateNumRegionsSummary)), 269 // add here any more surv rate groups 270 _recorded_survivor_regions(0), 271 _recorded_survivor_head(NULL), 272 _recorded_survivor_tail(NULL), 273 _survivors_age_table(true), 274 275 _gc_overhead_perc(0.0) 276 277 { 278 // Set up the region size and associated fields. Given that the 279 // policy is created before the heap, we have to set this up here, 280 // so it's done as soon as possible. 281 HeapRegion::setup_heap_region_size(Arguments::min_heap_size()); 282 HeapRegionRemSet::setup_remset_size(); 283 284 // Verify PLAB sizes 285 const uint region_size = HeapRegion::GrainWords; 286 if (YoungPLABSize > region_size || OldPLABSize > region_size) { 287 char buffer[128]; 288 jio_snprintf(buffer, sizeof(buffer), "%sPLABSize should be at most %u", 289 OldPLABSize > region_size ? "Old" : "Young", region_size); 290 vm_exit_during_initialization(buffer); 291 } 292 293 _recent_prev_end_times_for_all_gcs_sec->add(os::elapsedTime()); 294 _prev_collection_pause_end_ms = os::elapsedTime() * 1000.0; 295 296 _par_last_gc_worker_start_times_ms = new double[_parallel_gc_threads]; 297 _par_last_ext_root_scan_times_ms = new double[_parallel_gc_threads]; 298 _par_last_mark_stack_scan_times_ms = new double[_parallel_gc_threads]; 299 300 _par_last_update_rs_times_ms = new double[_parallel_gc_threads]; 301 _par_last_update_rs_processed_buffers = new double[_parallel_gc_threads]; 302 303 _par_last_scan_rs_times_ms = new double[_parallel_gc_threads]; 304 305 _par_last_obj_copy_times_ms = new double[_parallel_gc_threads]; 306 307 _par_last_termination_times_ms = new double[_parallel_gc_threads]; 308 _par_last_termination_attempts = new double[_parallel_gc_threads]; 309 _par_last_gc_worker_end_times_ms = new double[_parallel_gc_threads]; 310 _par_last_gc_worker_times_ms = new double[_parallel_gc_threads]; 311 312 // start conservatively 313 _expensive_region_limit_ms = 0.5 * (double) MaxGCPauseMillis; 314 315 // <NEW PREDICTION> 316 317 int index; 318 if (ParallelGCThreads == 0) 319 index = 0; 320 else if (ParallelGCThreads > 8) 321 index = 7; 322 else 323 index = ParallelGCThreads - 1; 324 325 _pending_card_diff_seq->add(0.0); 326 _rs_length_diff_seq->add(rs_length_diff_defaults[index]); 327 _cost_per_card_ms_seq->add(cost_per_card_ms_defaults[index]); 328 _fully_young_cards_per_entry_ratio_seq->add( 329 fully_young_cards_per_entry_ratio_defaults[index]); 330 _cost_per_entry_ms_seq->add(cost_per_entry_ms_defaults[index]); 331 _cost_per_byte_ms_seq->add(cost_per_byte_ms_defaults[index]); 332 _constant_other_time_ms_seq->add(constant_other_time_ms_defaults[index]); 333 _young_other_cost_per_region_ms_seq->add( 334 young_other_cost_per_region_ms_defaults[index]); 335 _non_young_other_cost_per_region_ms_seq->add( 336 non_young_other_cost_per_region_ms_defaults[index]); 337 338 // </NEW PREDICTION> 339 340 // Below, we might need to calculate the pause time target based on 341 // the pause interval. When we do so we are going to give G1 maximum 342 // flexibility and allow it to do pauses when it needs to. So, we'll 343 // arrange that the pause interval to be pause time target + 1 to 344 // ensure that a) the pause time target is maximized with respect to 345 // the pause interval and b) we maintain the invariant that pause 346 // time target < pause interval. If the user does not want this 347 // maximum flexibility, they will have to set the pause interval 348 // explicitly. 349 350 // First make sure that, if either parameter is set, its value is 351 // reasonable. 352 if (!FLAG_IS_DEFAULT(MaxGCPauseMillis)) { 353 if (MaxGCPauseMillis < 1) { 354 vm_exit_during_initialization("MaxGCPauseMillis should be " 355 "greater than 0"); 356 } 357 } 358 if (!FLAG_IS_DEFAULT(GCPauseIntervalMillis)) { 359 if (GCPauseIntervalMillis < 1) { 360 vm_exit_during_initialization("GCPauseIntervalMillis should be " 361 "greater than 0"); 362 } 363 } 364 365 // Then, if the pause time target parameter was not set, set it to 366 // the default value. 367 if (FLAG_IS_DEFAULT(MaxGCPauseMillis)) { 368 if (FLAG_IS_DEFAULT(GCPauseIntervalMillis)) { 369 // The default pause time target in G1 is 200ms 370 FLAG_SET_DEFAULT(MaxGCPauseMillis, 200); 371 } else { 372 // We do not allow the pause interval to be set without the 373 // pause time target 374 vm_exit_during_initialization("GCPauseIntervalMillis cannot be set " 375 "without setting MaxGCPauseMillis"); 376 } 377 } 378 379 // Then, if the interval parameter was not set, set it according to 380 // the pause time target (this will also deal with the case when the 381 // pause time target is the default value). 382 if (FLAG_IS_DEFAULT(GCPauseIntervalMillis)) { 383 FLAG_SET_DEFAULT(GCPauseIntervalMillis, MaxGCPauseMillis + 1); 384 } 385 386 // Finally, make sure that the two parameters are consistent. 387 if (MaxGCPauseMillis >= GCPauseIntervalMillis) { 388 char buffer[256]; 389 jio_snprintf(buffer, 256, 390 "MaxGCPauseMillis (%u) should be less than " 391 "GCPauseIntervalMillis (%u)", 392 MaxGCPauseMillis, GCPauseIntervalMillis); 393 vm_exit_during_initialization(buffer); 394 } 395 396 double max_gc_time = (double) MaxGCPauseMillis / 1000.0; 397 double time_slice = (double) GCPauseIntervalMillis / 1000.0; 398 _mmu_tracker = new G1MMUTrackerQueue(time_slice, max_gc_time); 399 _sigma = (double) G1ConfidencePercent / 100.0; 400 401 // start conservatively (around 50ms is about right) 402 _concurrent_mark_init_times_ms->add(0.05); 403 _concurrent_mark_remark_times_ms->add(0.05); 404 _concurrent_mark_cleanup_times_ms->add(0.20); 405 _tenuring_threshold = MaxTenuringThreshold; 406 407 // if G1FixedSurvivorSpaceSize is 0 which means the size is not 408 // fixed, then _max_survivor_regions will be calculated at 409 // calculate_young_list_target_length during initialization 410 _max_survivor_regions = G1FixedSurvivorSpaceSize / HeapRegion::GrainBytes; 411 412 assert(GCTimeRatio > 0, 413 "we should have set it to a default value set_g1_gc_flags() " 414 "if a user set it to 0"); 415 _gc_overhead_perc = 100.0 * (1.0 / (1.0 + GCTimeRatio)); 416 417 initialize_all(); 418 } 419 420 // Increment "i", mod "len" 421 static void inc_mod(int& i, int len) { 422 i++; if (i == len) i = 0; 423 } 424 425 void G1CollectorPolicy::initialize_flags() { 426 set_min_alignment(HeapRegion::GrainBytes); 427 set_max_alignment(GenRemSet::max_alignment_constraint(rem_set_name())); 428 if (SurvivorRatio < 1) { 429 vm_exit_during_initialization("Invalid survivor ratio specified"); 430 } 431 CollectorPolicy::initialize_flags(); 432 } 433 434 // The easiest way to deal with the parsing of the NewSize / 435 // MaxNewSize / etc. parameteres is to re-use the code in the 436 // TwoGenerationCollectorPolicy class. This is similar to what 437 // ParallelScavenge does with its GenerationSizer class (see 438 // ParallelScavengeHeap::initialize()). We might change this in the 439 // future, but it's a good start. 440 class G1YoungGenSizer : public TwoGenerationCollectorPolicy { 441 size_t size_to_region_num(size_t byte_size) { 442 return MAX2((size_t) 1, byte_size / HeapRegion::GrainBytes); 443 } 444 445 public: 446 G1YoungGenSizer() { 447 initialize_flags(); 448 initialize_size_info(); 449 } 450 451 size_t min_young_region_num() { 452 return size_to_region_num(_min_gen0_size); 453 } 454 size_t initial_young_region_num() { 455 return size_to_region_num(_initial_gen0_size); 456 } 457 size_t max_young_region_num() { 458 return size_to_region_num(_max_gen0_size); 459 } 460 }; 461 462 void G1CollectorPolicy::init() { 463 // Set aside an initial future to_space. 464 _g1 = G1CollectedHeap::heap(); 465 466 assert(Heap_lock->owned_by_self(), "Locking discipline."); 467 468 initialize_gc_policy_counters(); 469 470 G1YoungGenSizer sizer; 471 size_t initial_region_num = sizer.initial_young_region_num(); 472 473 if (UseAdaptiveSizePolicy) { 474 set_adaptive_young_list_length(true); 475 _young_list_fixed_length = 0; 476 } else { 477 set_adaptive_young_list_length(false); 478 _young_list_fixed_length = initial_region_num; 479 } 480 _free_regions_at_end_of_collection = _g1->free_regions(); 481 calculate_young_list_min_length(); 482 guarantee( _young_list_min_length == 0, "invariant, not enough info" ); 483 calculate_young_list_target_length(); 484 485 // We may immediately start allocating regions and placing them on the 486 // collection set list. Initialize the per-collection set info 487 start_incremental_cset_building(); 488 } 489 490 // Create the jstat counters for the policy. 491 void G1CollectorPolicy::initialize_gc_policy_counters() 492 { 493 _gc_policy_counters = new GCPolicyCounters("GarbageFirst", 1, 3); 494 } 495 496 void G1CollectorPolicy::calculate_young_list_min_length() { 497 _young_list_min_length = 0; 498 499 if (!adaptive_young_list_length()) 500 return; 501 502 if (_alloc_rate_ms_seq->num() > 3) { 503 double now_sec = os::elapsedTime(); 504 double when_ms = _mmu_tracker->when_max_gc_sec(now_sec) * 1000.0; 505 double alloc_rate_ms = predict_alloc_rate_ms(); 506 size_t min_regions = (size_t) ceil(alloc_rate_ms * when_ms); 507 size_t current_region_num = _g1->young_list()->length(); 508 _young_list_min_length = min_regions + current_region_num; 509 } 510 } 511 512 void G1CollectorPolicy::calculate_young_list_target_length() { 513 if (adaptive_young_list_length()) { 514 size_t rs_lengths = (size_t) get_new_prediction(_rs_lengths_seq); 515 calculate_young_list_target_length(rs_lengths); 516 } else { 517 if (full_young_gcs()) 518 _young_list_target_length = _young_list_fixed_length; 519 else 520 _young_list_target_length = _young_list_fixed_length / 2; 521 } 522 523 // Make sure we allow the application to allocate at least one 524 // region before we need to do a collection again. 525 size_t min_length = _g1->young_list()->length() + 1; 526 _young_list_target_length = MAX2(_young_list_target_length, min_length); 527 calculate_max_gc_locker_expansion(); 528 calculate_survivors_policy(); 529 } 530 531 void G1CollectorPolicy::calculate_young_list_target_length(size_t rs_lengths) { 532 guarantee( adaptive_young_list_length(), "pre-condition" ); 533 guarantee( !_in_marking_window || !_last_full_young_gc, "invariant" ); 534 535 double start_time_sec = os::elapsedTime(); 536 size_t min_reserve_perc = MAX2((size_t)2, (size_t)G1ReservePercent); 537 min_reserve_perc = MIN2((size_t) 50, min_reserve_perc); 538 size_t reserve_regions = 539 (size_t) ((double) min_reserve_perc * (double) _g1->n_regions() / 100.0); 540 541 if (full_young_gcs() && _free_regions_at_end_of_collection > 0) { 542 // we are in fully-young mode and there are free regions in the heap 543 544 double survivor_regions_evac_time = 545 predict_survivor_regions_evac_time(); 546 547 double target_pause_time_ms = _mmu_tracker->max_gc_time() * 1000.0; 548 size_t pending_cards = (size_t) get_new_prediction(_pending_cards_seq); 549 size_t adj_rs_lengths = rs_lengths + predict_rs_length_diff(); 550 size_t scanned_cards = predict_young_card_num(adj_rs_lengths); 551 double base_time_ms = predict_base_elapsed_time_ms(pending_cards, scanned_cards) 552 + survivor_regions_evac_time; 553 554 // the result 555 size_t final_young_length = 0; 556 557 size_t init_free_regions = 558 MAX2((size_t)0, _free_regions_at_end_of_collection - reserve_regions); 559 560 // if we're still under the pause target... 561 if (base_time_ms <= target_pause_time_ms) { 562 // We make sure that the shortest young length that makes sense 563 // fits within the target pause time. 564 size_t min_young_length = 1; 565 566 if (predict_will_fit(min_young_length, base_time_ms, 567 init_free_regions, target_pause_time_ms)) { 568 // The shortest young length will fit within the target pause time; 569 // we'll now check whether the absolute maximum number of young 570 // regions will fit in the target pause time. If not, we'll do 571 // a binary search between min_young_length and max_young_length 572 size_t abs_max_young_length = _free_regions_at_end_of_collection - 1; 573 size_t max_young_length = abs_max_young_length; 574 575 if (max_young_length > min_young_length) { 576 // Let's check if the initial max young length will fit within the 577 // target pause. If so then there is no need to search for a maximal 578 // young length - we'll return the initial maximum 579 580 if (predict_will_fit(max_young_length, base_time_ms, 581 init_free_regions, target_pause_time_ms)) { 582 // The maximum young length will satisfy the target pause time. 583 // We are done so set min young length to this maximum length. 584 // The code after the loop will then set final_young_length using 585 // the value cached in the minimum length. 586 min_young_length = max_young_length; 587 } else { 588 // The maximum possible number of young regions will not fit within 589 // the target pause time so let's search.... 590 591 size_t diff = (max_young_length - min_young_length) / 2; 592 max_young_length = min_young_length + diff; 593 594 while (max_young_length > min_young_length) { 595 if (predict_will_fit(max_young_length, base_time_ms, 596 init_free_regions, target_pause_time_ms)) { 597 598 // The current max young length will fit within the target 599 // pause time. Note we do not exit the loop here. By setting 600 // min = max, and then increasing the max below means that 601 // we will continue searching for an upper bound in the 602 // range [max..max+diff] 603 min_young_length = max_young_length; 604 } 605 diff = (max_young_length - min_young_length) / 2; 606 max_young_length = min_young_length + diff; 607 } 608 // the above loop found a maximal young length that will fit 609 // within the target pause time. 610 } 611 assert(min_young_length <= abs_max_young_length, "just checking"); 612 } 613 final_young_length = min_young_length; 614 } 615 } 616 // and we're done! 617 618 // we should have at least one region in the target young length 619 _young_list_target_length = 620 final_young_length + _recorded_survivor_regions; 621 622 // let's keep an eye of how long we spend on this calculation 623 // right now, I assume that we'll print it when we need it; we 624 // should really adde it to the breakdown of a pause 625 double end_time_sec = os::elapsedTime(); 626 double elapsed_time_ms = (end_time_sec - start_time_sec) * 1000.0; 627 628 #ifdef TRACE_CALC_YOUNG_LENGTH 629 // leave this in for debugging, just in case 630 gclog_or_tty->print_cr("target = %1.1lf ms, young = " SIZE_FORMAT ", " 631 "elapsed %1.2lf ms, (%s%s) " SIZE_FORMAT SIZE_FORMAT, 632 target_pause_time_ms, 633 _young_list_target_length 634 elapsed_time_ms, 635 full_young_gcs() ? "full" : "partial", 636 during_initial_mark_pause() ? " i-m" : "", 637 _in_marking_window, 638 _in_marking_window_im); 639 #endif // TRACE_CALC_YOUNG_LENGTH 640 641 if (_young_list_target_length < _young_list_min_length) { 642 // bummer; this means that, if we do a pause when the maximal 643 // length dictates, we'll violate the pause spacing target (the 644 // min length was calculate based on the application's current 645 // alloc rate); 646 647 // so, we have to bite the bullet, and allocate the minimum 648 // number. We'll violate our target, but we just can't meet it. 649 650 #ifdef TRACE_CALC_YOUNG_LENGTH 651 // leave this in for debugging, just in case 652 gclog_or_tty->print_cr("adjusted target length from " 653 SIZE_FORMAT " to " SIZE_FORMAT, 654 _young_list_target_length, _young_list_min_length); 655 #endif // TRACE_CALC_YOUNG_LENGTH 656 657 _young_list_target_length = _young_list_min_length; 658 } 659 } else { 660 // we are in a partially-young mode or we've run out of regions (due 661 // to evacuation failure) 662 663 #ifdef TRACE_CALC_YOUNG_LENGTH 664 // leave this in for debugging, just in case 665 gclog_or_tty->print_cr("(partial) setting target to " SIZE_FORMAT 666 _young_list_min_length); 667 #endif // TRACE_CALC_YOUNG_LENGTH 668 // we'll do the pause as soon as possible by choosing the minimum 669 _young_list_target_length = _young_list_min_length; 670 } 671 672 _rs_lengths_prediction = rs_lengths; 673 } 674 675 // This is used by: calculate_young_list_target_length(rs_length). It 676 // returns true iff: 677 // the predicted pause time for the given young list will not overflow 678 // the target pause time 679 // and: 680 // the predicted amount of surviving data will not overflow the 681 // the amount of free space available for survivor regions. 682 // 683 bool 684 G1CollectorPolicy::predict_will_fit(size_t young_length, 685 double base_time_ms, 686 size_t init_free_regions, 687 double target_pause_time_ms) { 688 689 if (young_length >= init_free_regions) 690 // end condition 1: not enough space for the young regions 691 return false; 692 693 double accum_surv_rate_adj = 0.0; 694 double accum_surv_rate = 695 accum_yg_surv_rate_pred((int)(young_length - 1)) - accum_surv_rate_adj; 696 697 size_t bytes_to_copy = 698 (size_t) (accum_surv_rate * (double) HeapRegion::GrainBytes); 699 700 double copy_time_ms = predict_object_copy_time_ms(bytes_to_copy); 701 702 double young_other_time_ms = 703 predict_young_other_time_ms(young_length); 704 705 double pause_time_ms = 706 base_time_ms + copy_time_ms + young_other_time_ms; 707 708 if (pause_time_ms > target_pause_time_ms) 709 // end condition 2: over the target pause time 710 return false; 711 712 size_t free_bytes = 713 (init_free_regions - young_length) * HeapRegion::GrainBytes; 714 715 if ((2.0 + sigma()) * (double) bytes_to_copy > (double) free_bytes) 716 // end condition 3: out of to-space (conservatively) 717 return false; 718 719 // success! 720 return true; 721 } 722 723 double G1CollectorPolicy::predict_survivor_regions_evac_time() { 724 double survivor_regions_evac_time = 0.0; 725 for (HeapRegion * r = _recorded_survivor_head; 726 r != NULL && r != _recorded_survivor_tail->get_next_young_region(); 727 r = r->get_next_young_region()) { 728 survivor_regions_evac_time += predict_region_elapsed_time_ms(r, true); 729 } 730 return survivor_regions_evac_time; 731 } 732 733 void G1CollectorPolicy::check_prediction_validity() { 734 guarantee( adaptive_young_list_length(), "should not call this otherwise" ); 735 736 size_t rs_lengths = _g1->young_list()->sampled_rs_lengths(); 737 if (rs_lengths > _rs_lengths_prediction) { 738 // add 10% to avoid having to recalculate often 739 size_t rs_lengths_prediction = rs_lengths * 1100 / 1000; 740 calculate_young_list_target_length(rs_lengths_prediction); 741 } 742 } 743 744 HeapWord* G1CollectorPolicy::mem_allocate_work(size_t size, 745 bool is_tlab, 746 bool* gc_overhead_limit_was_exceeded) { 747 guarantee(false, "Not using this policy feature yet."); 748 return NULL; 749 } 750 751 // This method controls how a collector handles one or more 752 // of its generations being fully allocated. 753 HeapWord* G1CollectorPolicy::satisfy_failed_allocation(size_t size, 754 bool is_tlab) { 755 guarantee(false, "Not using this policy feature yet."); 756 return NULL; 757 } 758 759 760 #ifndef PRODUCT 761 bool G1CollectorPolicy::verify_young_ages() { 762 HeapRegion* head = _g1->young_list()->first_region(); 763 return 764 verify_young_ages(head, _short_lived_surv_rate_group); 765 // also call verify_young_ages on any additional surv rate groups 766 } 767 768 bool 769 G1CollectorPolicy::verify_young_ages(HeapRegion* head, 770 SurvRateGroup *surv_rate_group) { 771 guarantee( surv_rate_group != NULL, "pre-condition" ); 772 773 const char* name = surv_rate_group->name(); 774 bool ret = true; 775 int prev_age = -1; 776 777 for (HeapRegion* curr = head; 778 curr != NULL; 779 curr = curr->get_next_young_region()) { 780 SurvRateGroup* group = curr->surv_rate_group(); 781 if (group == NULL && !curr->is_survivor()) { 782 gclog_or_tty->print_cr("## %s: encountered NULL surv_rate_group", name); 783 ret = false; 784 } 785 786 if (surv_rate_group == group) { 787 int age = curr->age_in_surv_rate_group(); 788 789 if (age < 0) { 790 gclog_or_tty->print_cr("## %s: encountered negative age", name); 791 ret = false; 792 } 793 794 if (age <= prev_age) { 795 gclog_or_tty->print_cr("## %s: region ages are not strictly increasing " 796 "(%d, %d)", name, age, prev_age); 797 ret = false; 798 } 799 prev_age = age; 800 } 801 } 802 803 return ret; 804 } 805 #endif // PRODUCT 806 807 void G1CollectorPolicy::record_full_collection_start() { 808 _cur_collection_start_sec = os::elapsedTime(); 809 // Release the future to-space so that it is available for compaction into. 810 _g1->set_full_collection(); 811 } 812 813 void G1CollectorPolicy::record_full_collection_end() { 814 // Consider this like a collection pause for the purposes of allocation 815 // since last pause. 816 double end_sec = os::elapsedTime(); 817 double full_gc_time_sec = end_sec - _cur_collection_start_sec; 818 double full_gc_time_ms = full_gc_time_sec * 1000.0; 819 820 _all_full_gc_times_ms->add(full_gc_time_ms); 821 822 update_recent_gc_times(end_sec, full_gc_time_ms); 823 824 _g1->clear_full_collection(); 825 826 // "Nuke" the heuristics that control the fully/partially young GC 827 // transitions and make sure we start with fully young GCs after the 828 // Full GC. 829 set_full_young_gcs(true); 830 _last_full_young_gc = false; 831 _should_revert_to_full_young_gcs = false; 832 clear_initiate_conc_mark_if_possible(); 833 clear_during_initial_mark_pause(); 834 _known_garbage_bytes = 0; 835 _known_garbage_ratio = 0.0; 836 _in_marking_window = false; 837 _in_marking_window_im = false; 838 839 _short_lived_surv_rate_group->start_adding_regions(); 840 // also call this on any additional surv rate groups 841 842 record_survivor_regions(0, NULL, NULL); 843 844 _prev_region_num_young = _region_num_young; 845 _prev_region_num_tenured = _region_num_tenured; 846 847 _free_regions_at_end_of_collection = _g1->free_regions(); 848 // Reset survivors SurvRateGroup. 849 _survivor_surv_rate_group->reset(); 850 calculate_young_list_min_length(); 851 calculate_young_list_target_length(); 852 } 853 854 void G1CollectorPolicy::record_stop_world_start() { 855 _stop_world_start = os::elapsedTime(); 856 } 857 858 void G1CollectorPolicy::record_collection_pause_start(double start_time_sec, 859 size_t start_used) { 860 if (PrintGCDetails) { 861 gclog_or_tty->stamp(PrintGCTimeStamps); 862 gclog_or_tty->print("[GC pause"); 863 gclog_or_tty->print(" (%s)", full_young_gcs() ? "young" : "partial"); 864 } 865 866 assert(_g1->used() == _g1->recalculate_used(), 867 err_msg("sanity, used: "SIZE_FORMAT" recalculate_used: "SIZE_FORMAT, 868 _g1->used(), _g1->recalculate_used())); 869 870 double s_w_t_ms = (start_time_sec - _stop_world_start) * 1000.0; 871 _all_stop_world_times_ms->add(s_w_t_ms); 872 _stop_world_start = 0.0; 873 874 _cur_collection_start_sec = start_time_sec; 875 _cur_collection_pause_used_at_start_bytes = start_used; 876 _cur_collection_pause_used_regions_at_start = _g1->used_regions(); 877 _pending_cards = _g1->pending_card_num(); 878 _max_pending_cards = _g1->max_pending_card_num(); 879 880 _bytes_in_collection_set_before_gc = 0; 881 _bytes_copied_during_gc = 0; 882 883 YoungList* young_list = _g1->young_list(); 884 _eden_bytes_before_gc = young_list->eden_used_bytes(); 885 _survivor_bytes_before_gc = young_list->survivor_used_bytes(); 886 _capacity_before_gc = _g1->capacity(); 887 888 #ifdef DEBUG 889 // initialise these to something well known so that we can spot 890 // if they are not set properly 891 892 for (int i = 0; i < _parallel_gc_threads; ++i) { 893 _par_last_gc_worker_start_times_ms[i] = -1234.0; 894 _par_last_ext_root_scan_times_ms[i] = -1234.0; 895 _par_last_mark_stack_scan_times_ms[i] = -1234.0; 896 _par_last_update_rs_times_ms[i] = -1234.0; 897 _par_last_update_rs_processed_buffers[i] = -1234.0; 898 _par_last_scan_rs_times_ms[i] = -1234.0; 899 _par_last_obj_copy_times_ms[i] = -1234.0; 900 _par_last_termination_times_ms[i] = -1234.0; 901 _par_last_termination_attempts[i] = -1234.0; 902 _par_last_gc_worker_end_times_ms[i] = -1234.0; 903 _par_last_gc_worker_times_ms[i] = -1234.0; 904 } 905 #endif 906 907 for (int i = 0; i < _aux_num; ++i) { 908 _cur_aux_times_ms[i] = 0.0; 909 _cur_aux_times_set[i] = false; 910 } 911 912 _satb_drain_time_set = false; 913 _last_satb_drain_processed_buffers = -1; 914 915 _last_young_gc_full = false; 916 917 // do that for any other surv rate groups 918 _short_lived_surv_rate_group->stop_adding_regions(); 919 _survivors_age_table.clear(); 920 921 assert( verify_young_ages(), "region age verification" ); 922 } 923 924 void G1CollectorPolicy::record_mark_closure_time(double mark_closure_time_ms) { 925 _mark_closure_time_ms = mark_closure_time_ms; 926 } 927 928 void G1CollectorPolicy::record_concurrent_mark_init_end_pre(double 929 mark_init_elapsed_time_ms) { 930 _during_marking = true; 931 assert(!initiate_conc_mark_if_possible(), "we should have cleared it by now"); 932 clear_during_initial_mark_pause(); 933 _cur_mark_stop_world_time_ms = mark_init_elapsed_time_ms; 934 } 935 936 void G1CollectorPolicy::record_concurrent_mark_remark_start() { 937 _mark_remark_start_sec = os::elapsedTime(); 938 _during_marking = false; 939 } 940 941 void G1CollectorPolicy::record_concurrent_mark_remark_end() { 942 double end_time_sec = os::elapsedTime(); 943 double elapsed_time_ms = (end_time_sec - _mark_remark_start_sec)*1000.0; 944 _concurrent_mark_remark_times_ms->add(elapsed_time_ms); 945 _cur_mark_stop_world_time_ms += elapsed_time_ms; 946 _prev_collection_pause_end_ms += elapsed_time_ms; 947 948 _mmu_tracker->add_pause(_mark_remark_start_sec, end_time_sec, true); 949 } 950 951 void G1CollectorPolicy::record_concurrent_mark_cleanup_start() { 952 _mark_cleanup_start_sec = os::elapsedTime(); 953 } 954 955 void 956 G1CollectorPolicy::record_concurrent_mark_cleanup_end(size_t freed_bytes, 957 size_t max_live_bytes) { 958 record_concurrent_mark_cleanup_end_work1(freed_bytes, max_live_bytes); 959 record_concurrent_mark_cleanup_end_work2(); 960 } 961 962 void 963 G1CollectorPolicy:: 964 record_concurrent_mark_cleanup_end_work1(size_t freed_bytes, 965 size_t max_live_bytes) { 966 if (_n_marks < 2) _n_marks++; 967 if (G1PolicyVerbose > 0) 968 gclog_or_tty->print_cr("At end of marking, max_live is " SIZE_FORMAT " MB " 969 " (of " SIZE_FORMAT " MB heap).", 970 max_live_bytes/M, _g1->capacity()/M); 971 } 972 973 // The important thing about this is that it includes "os::elapsedTime". 974 void G1CollectorPolicy::record_concurrent_mark_cleanup_end_work2() { 975 double end_time_sec = os::elapsedTime(); 976 double elapsed_time_ms = (end_time_sec - _mark_cleanup_start_sec)*1000.0; 977 _concurrent_mark_cleanup_times_ms->add(elapsed_time_ms); 978 _cur_mark_stop_world_time_ms += elapsed_time_ms; 979 _prev_collection_pause_end_ms += elapsed_time_ms; 980 981 _mmu_tracker->add_pause(_mark_cleanup_start_sec, end_time_sec, true); 982 983 _num_markings++; 984 985 // We did a marking, so reset the "since_last_mark" variables. 986 double considerConcMarkCost = 1.0; 987 // If there are available processors, concurrent activity is free... 988 if (Threads::number_of_non_daemon_threads() * 2 < 989 os::active_processor_count()) { 990 considerConcMarkCost = 0.0; 991 } 992 _n_pauses_at_mark_end = _n_pauses; 993 _n_marks_since_last_pause++; 994 } 995 996 void 997 G1CollectorPolicy::record_concurrent_mark_cleanup_completed() { 998 _should_revert_to_full_young_gcs = false; 999 _last_full_young_gc = true; 1000 _in_marking_window = false; 1001 if (adaptive_young_list_length()) 1002 calculate_young_list_target_length(); 1003 } 1004 1005 void G1CollectorPolicy::record_concurrent_pause() { 1006 if (_stop_world_start > 0.0) { 1007 double yield_ms = (os::elapsedTime() - _stop_world_start) * 1000.0; 1008 _all_yield_times_ms->add(yield_ms); 1009 } 1010 } 1011 1012 void G1CollectorPolicy::record_concurrent_pause_end() { 1013 } 1014 1015 template<class T> 1016 T sum_of(T* sum_arr, int start, int n, int N) { 1017 T sum = (T)0; 1018 for (int i = 0; i < n; i++) { 1019 int j = (start + i) % N; 1020 sum += sum_arr[j]; 1021 } 1022 return sum; 1023 } 1024 1025 void G1CollectorPolicy::print_par_stats(int level, 1026 const char* str, 1027 double* data) { 1028 double min = data[0], max = data[0]; 1029 double total = 0.0; 1030 LineBuffer buf(level); 1031 buf.append("[%s (ms):", str); 1032 for (uint i = 0; i < ParallelGCThreads; ++i) { 1033 double val = data[i]; 1034 if (val < min) 1035 min = val; 1036 if (val > max) 1037 max = val; 1038 total += val; 1039 buf.append(" %3.1lf", val); 1040 } 1041 buf.append_and_print_cr(""); 1042 double avg = total / (double) ParallelGCThreads; 1043 buf.append_and_print_cr(" Avg: %5.1lf, Min: %5.1lf, Max: %5.1lf, Diff: %5.1lf]", 1044 avg, min, max, max - min); 1045 } 1046 1047 void G1CollectorPolicy::print_par_sizes(int level, 1048 const char* str, 1049 double* data) { 1050 double min = data[0], max = data[0]; 1051 double total = 0.0; 1052 LineBuffer buf(level); 1053 buf.append("[%s :", str); 1054 for (uint i = 0; i < ParallelGCThreads; ++i) { 1055 double val = data[i]; 1056 if (val < min) 1057 min = val; 1058 if (val > max) 1059 max = val; 1060 total += val; 1061 buf.append(" %d", (int) val); 1062 } 1063 buf.append_and_print_cr(""); 1064 double avg = total / (double) ParallelGCThreads; 1065 buf.append_and_print_cr(" Sum: %d, Avg: %d, Min: %d, Max: %d, Diff: %d]", 1066 (int)total, (int)avg, (int)min, (int)max, (int)max - (int)min); 1067 } 1068 1069 void G1CollectorPolicy::print_stats (int level, 1070 const char* str, 1071 double value) { 1072 LineBuffer(level).append_and_print_cr("[%s: %5.1lf ms]", str, value); 1073 } 1074 1075 void G1CollectorPolicy::print_stats (int level, 1076 const char* str, 1077 int value) { 1078 LineBuffer(level).append_and_print_cr("[%s: %d]", str, value); 1079 } 1080 1081 double G1CollectorPolicy::avg_value (double* data) { 1082 if (G1CollectedHeap::use_parallel_gc_threads()) { 1083 double ret = 0.0; 1084 for (uint i = 0; i < ParallelGCThreads; ++i) 1085 ret += data[i]; 1086 return ret / (double) ParallelGCThreads; 1087 } else { 1088 return data[0]; 1089 } 1090 } 1091 1092 double G1CollectorPolicy::max_value (double* data) { 1093 if (G1CollectedHeap::use_parallel_gc_threads()) { 1094 double ret = data[0]; 1095 for (uint i = 1; i < ParallelGCThreads; ++i) 1096 if (data[i] > ret) 1097 ret = data[i]; 1098 return ret; 1099 } else { 1100 return data[0]; 1101 } 1102 } 1103 1104 double G1CollectorPolicy::sum_of_values (double* data) { 1105 if (G1CollectedHeap::use_parallel_gc_threads()) { 1106 double sum = 0.0; 1107 for (uint i = 0; i < ParallelGCThreads; i++) 1108 sum += data[i]; 1109 return sum; 1110 } else { 1111 return data[0]; 1112 } 1113 } 1114 1115 double G1CollectorPolicy::max_sum (double* data1, 1116 double* data2) { 1117 double ret = data1[0] + data2[0]; 1118 1119 if (G1CollectedHeap::use_parallel_gc_threads()) { 1120 for (uint i = 1; i < ParallelGCThreads; ++i) { 1121 double data = data1[i] + data2[i]; 1122 if (data > ret) 1123 ret = data; 1124 } 1125 } 1126 return ret; 1127 } 1128 1129 // Anything below that is considered to be zero 1130 #define MIN_TIMER_GRANULARITY 0.0000001 1131 1132 void G1CollectorPolicy::record_collection_pause_end() { 1133 double end_time_sec = os::elapsedTime(); 1134 double elapsed_ms = _last_pause_time_ms; 1135 bool parallel = G1CollectedHeap::use_parallel_gc_threads(); 1136 size_t rs_size = 1137 _cur_collection_pause_used_regions_at_start - collection_set_size(); 1138 size_t cur_used_bytes = _g1->used(); 1139 assert(cur_used_bytes == _g1->recalculate_used(), "It should!"); 1140 bool last_pause_included_initial_mark = false; 1141 bool update_stats = !_g1->evacuation_failed(); 1142 1143 #ifndef PRODUCT 1144 if (G1YoungSurvRateVerbose) { 1145 gclog_or_tty->print_cr(""); 1146 _short_lived_surv_rate_group->print(); 1147 // do that for any other surv rate groups too 1148 } 1149 #endif // PRODUCT 1150 1151 last_pause_included_initial_mark = during_initial_mark_pause(); 1152 if (last_pause_included_initial_mark) 1153 record_concurrent_mark_init_end_pre(0.0); 1154 1155 size_t min_used_targ = 1156 (_g1->capacity() / 100) * InitiatingHeapOccupancyPercent; 1157 1158 1159 if (!_g1->mark_in_progress() && !_last_full_young_gc) { 1160 assert(!last_pause_included_initial_mark, "invariant"); 1161 if (cur_used_bytes > min_used_targ && 1162 cur_used_bytes > _prev_collection_pause_used_at_end_bytes) { 1163 assert(!during_initial_mark_pause(), "we should not see this here"); 1164 1165 // Note: this might have already been set, if during the last 1166 // pause we decided to start a cycle but at the beginning of 1167 // this pause we decided to postpone it. That's OK. 1168 set_initiate_conc_mark_if_possible(); 1169 } 1170 } 1171 1172 _prev_collection_pause_used_at_end_bytes = cur_used_bytes; 1173 1174 _mmu_tracker->add_pause(end_time_sec - elapsed_ms/1000.0, 1175 end_time_sec, false); 1176 1177 guarantee(_cur_collection_pause_used_regions_at_start >= 1178 collection_set_size(), 1179 "Negative RS size?"); 1180 1181 // This assert is exempted when we're doing parallel collection pauses, 1182 // because the fragmentation caused by the parallel GC allocation buffers 1183 // can lead to more memory being used during collection than was used 1184 // before. Best leave this out until the fragmentation problem is fixed. 1185 // Pauses in which evacuation failed can also lead to negative 1186 // collections, since no space is reclaimed from a region containing an 1187 // object whose evacuation failed. 1188 // Further, we're now always doing parallel collection. But I'm still 1189 // leaving this here as a placeholder for a more precise assertion later. 1190 // (DLD, 10/05.) 1191 assert((true || parallel) // Always using GC LABs now. 1192 || _g1->evacuation_failed() 1193 || _cur_collection_pause_used_at_start_bytes >= cur_used_bytes, 1194 "Negative collection"); 1195 1196 size_t freed_bytes = 1197 _cur_collection_pause_used_at_start_bytes - cur_used_bytes; 1198 size_t surviving_bytes = _collection_set_bytes_used_before - freed_bytes; 1199 1200 double survival_fraction = 1201 (double)surviving_bytes/ 1202 (double)_collection_set_bytes_used_before; 1203 1204 _n_pauses++; 1205 1206 double ext_root_scan_time = avg_value(_par_last_ext_root_scan_times_ms); 1207 double mark_stack_scan_time = avg_value(_par_last_mark_stack_scan_times_ms); 1208 double update_rs_time = avg_value(_par_last_update_rs_times_ms); 1209 double update_rs_processed_buffers = 1210 sum_of_values(_par_last_update_rs_processed_buffers); 1211 double scan_rs_time = avg_value(_par_last_scan_rs_times_ms); 1212 double obj_copy_time = avg_value(_par_last_obj_copy_times_ms); 1213 double termination_time = avg_value(_par_last_termination_times_ms); 1214 1215 double parallel_known_time = update_rs_time + 1216 ext_root_scan_time + 1217 mark_stack_scan_time + 1218 scan_rs_time + 1219 obj_copy_time + 1220 termination_time; 1221 1222 double parallel_other_time = _cur_collection_par_time_ms - parallel_known_time; 1223 1224 PauseSummary* summary = _summary; 1225 1226 if (update_stats) { 1227 _recent_rs_scan_times_ms->add(scan_rs_time); 1228 _recent_pause_times_ms->add(elapsed_ms); 1229 _recent_rs_sizes->add(rs_size); 1230 1231 MainBodySummary* body_summary = summary->main_body_summary(); 1232 guarantee(body_summary != NULL, "should not be null!"); 1233 1234 if (_satb_drain_time_set) 1235 body_summary->record_satb_drain_time_ms(_cur_satb_drain_time_ms); 1236 else 1237 body_summary->record_satb_drain_time_ms(0.0); 1238 1239 body_summary->record_ext_root_scan_time_ms(ext_root_scan_time); 1240 body_summary->record_mark_stack_scan_time_ms(mark_stack_scan_time); 1241 body_summary->record_update_rs_time_ms(update_rs_time); 1242 body_summary->record_scan_rs_time_ms(scan_rs_time); 1243 body_summary->record_obj_copy_time_ms(obj_copy_time); 1244 if (parallel) { 1245 body_summary->record_parallel_time_ms(_cur_collection_par_time_ms); 1246 body_summary->record_clear_ct_time_ms(_cur_clear_ct_time_ms); 1247 body_summary->record_termination_time_ms(termination_time); 1248 body_summary->record_parallel_other_time_ms(parallel_other_time); 1249 } 1250 body_summary->record_mark_closure_time_ms(_mark_closure_time_ms); 1251 1252 // We exempt parallel collection from this check because Alloc Buffer 1253 // fragmentation can produce negative collections. Same with evac 1254 // failure. 1255 // Further, we're now always doing parallel collection. But I'm still 1256 // leaving this here as a placeholder for a more precise assertion later. 1257 // (DLD, 10/05. 1258 assert((true || parallel) 1259 || _g1->evacuation_failed() 1260 || surviving_bytes <= _collection_set_bytes_used_before, 1261 "Or else negative collection!"); 1262 _recent_CS_bytes_used_before->add(_collection_set_bytes_used_before); 1263 _recent_CS_bytes_surviving->add(surviving_bytes); 1264 1265 // this is where we update the allocation rate of the application 1266 double app_time_ms = 1267 (_cur_collection_start_sec * 1000.0 - _prev_collection_pause_end_ms); 1268 if (app_time_ms < MIN_TIMER_GRANULARITY) { 1269 // This usually happens due to the timer not having the required 1270 // granularity. Some Linuxes are the usual culprits. 1271 // We'll just set it to something (arbitrarily) small. 1272 app_time_ms = 1.0; 1273 } 1274 size_t regions_allocated = 1275 (_region_num_young - _prev_region_num_young) + 1276 (_region_num_tenured - _prev_region_num_tenured); 1277 double alloc_rate_ms = (double) regions_allocated / app_time_ms; 1278 _alloc_rate_ms_seq->add(alloc_rate_ms); 1279 _prev_region_num_young = _region_num_young; 1280 _prev_region_num_tenured = _region_num_tenured; 1281 1282 double interval_ms = 1283 (end_time_sec - _recent_prev_end_times_for_all_gcs_sec->oldest()) * 1000.0; 1284 update_recent_gc_times(end_time_sec, elapsed_ms); 1285 _recent_avg_pause_time_ratio = _recent_gc_times_ms->sum()/interval_ms; 1286 if (recent_avg_pause_time_ratio() < 0.0 || 1287 (recent_avg_pause_time_ratio() - 1.0 > 0.0)) { 1288 #ifndef PRODUCT 1289 // Dump info to allow post-facto debugging 1290 gclog_or_tty->print_cr("recent_avg_pause_time_ratio() out of bounds"); 1291 gclog_or_tty->print_cr("-------------------------------------------"); 1292 gclog_or_tty->print_cr("Recent GC Times (ms):"); 1293 _recent_gc_times_ms->dump(); 1294 gclog_or_tty->print_cr("(End Time=%3.3f) Recent GC End Times (s):", end_time_sec); 1295 _recent_prev_end_times_for_all_gcs_sec->dump(); 1296 gclog_or_tty->print_cr("GC = %3.3f, Interval = %3.3f, Ratio = %3.3f", 1297 _recent_gc_times_ms->sum(), interval_ms, recent_avg_pause_time_ratio()); 1298 // In debug mode, terminate the JVM if the user wants to debug at this point. 1299 assert(!G1FailOnFPError, "Debugging data for CR 6898948 has been dumped above"); 1300 #endif // !PRODUCT 1301 // Clip ratio between 0.0 and 1.0, and continue. This will be fixed in 1302 // CR 6902692 by redoing the manner in which the ratio is incrementally computed. 1303 if (_recent_avg_pause_time_ratio < 0.0) { 1304 _recent_avg_pause_time_ratio = 0.0; 1305 } else { 1306 assert(_recent_avg_pause_time_ratio - 1.0 > 0.0, "Ctl-point invariant"); 1307 _recent_avg_pause_time_ratio = 1.0; 1308 } 1309 } 1310 } 1311 1312 if (G1PolicyVerbose > 1) { 1313 gclog_or_tty->print_cr(" Recording collection pause(%d)", _n_pauses); 1314 } 1315 1316 if (G1PolicyVerbose > 1) { 1317 gclog_or_tty->print_cr(" ET: %10.6f ms (avg: %10.6f ms)\n" 1318 " ET-RS: %10.6f ms (avg: %10.6f ms)\n" 1319 " |RS|: " SIZE_FORMAT, 1320 elapsed_ms, recent_avg_time_for_pauses_ms(), 1321 scan_rs_time, recent_avg_time_for_rs_scan_ms(), 1322 rs_size); 1323 1324 gclog_or_tty->print_cr(" Used at start: " SIZE_FORMAT"K" 1325 " At end " SIZE_FORMAT "K\n" 1326 " garbage : " SIZE_FORMAT "K" 1327 " of " SIZE_FORMAT "K\n" 1328 " survival : %6.2f%% (%6.2f%% avg)", 1329 _cur_collection_pause_used_at_start_bytes/K, 1330 _g1->used()/K, freed_bytes/K, 1331 _collection_set_bytes_used_before/K, 1332 survival_fraction*100.0, 1333 recent_avg_survival_fraction()*100.0); 1334 gclog_or_tty->print_cr(" Recent %% gc pause time: %6.2f", 1335 recent_avg_pause_time_ratio() * 100.0); 1336 } 1337 1338 double other_time_ms = elapsed_ms; 1339 1340 if (_satb_drain_time_set) { 1341 other_time_ms -= _cur_satb_drain_time_ms; 1342 } 1343 1344 if (parallel) { 1345 other_time_ms -= _cur_collection_par_time_ms + _cur_clear_ct_time_ms; 1346 } else { 1347 other_time_ms -= 1348 update_rs_time + 1349 ext_root_scan_time + mark_stack_scan_time + 1350 scan_rs_time + obj_copy_time; 1351 } 1352 1353 if (PrintGCDetails) { 1354 gclog_or_tty->print_cr("%s, %1.8lf secs]", 1355 (last_pause_included_initial_mark) ? " (initial-mark)" : "", 1356 elapsed_ms / 1000.0); 1357 1358 if (_satb_drain_time_set) { 1359 print_stats(1, "SATB Drain Time", _cur_satb_drain_time_ms); 1360 } 1361 if (_last_satb_drain_processed_buffers >= 0) { 1362 print_stats(2, "Processed Buffers", _last_satb_drain_processed_buffers); 1363 } 1364 if (parallel) { 1365 print_stats(1, "Parallel Time", _cur_collection_par_time_ms); 1366 print_par_stats(2, "GC Worker Start Time", _par_last_gc_worker_start_times_ms); 1367 print_par_stats(2, "Update RS", _par_last_update_rs_times_ms); 1368 print_par_sizes(3, "Processed Buffers", _par_last_update_rs_processed_buffers); 1369 print_par_stats(2, "Ext Root Scanning", _par_last_ext_root_scan_times_ms); 1370 print_par_stats(2, "Mark Stack Scanning", _par_last_mark_stack_scan_times_ms); 1371 print_par_stats(2, "Scan RS", _par_last_scan_rs_times_ms); 1372 print_par_stats(2, "Object Copy", _par_last_obj_copy_times_ms); 1373 print_par_stats(2, "Termination", _par_last_termination_times_ms); 1374 print_par_sizes(3, "Termination Attempts", _par_last_termination_attempts); 1375 print_par_stats(2, "GC Worker End Time", _par_last_gc_worker_end_times_ms); 1376 1377 for (int i = 0; i < _parallel_gc_threads; i++) { 1378 _par_last_gc_worker_times_ms[i] = _par_last_gc_worker_end_times_ms[i] - _par_last_gc_worker_start_times_ms[i]; 1379 } 1380 print_par_stats(2, "GC Worker Times", _par_last_gc_worker_times_ms); 1381 1382 print_stats(2, "Parallel Other", parallel_other_time); 1383 print_stats(1, "Clear CT", _cur_clear_ct_time_ms); 1384 } else { 1385 print_stats(1, "Update RS", update_rs_time); 1386 print_stats(2, "Processed Buffers", 1387 (int)update_rs_processed_buffers); 1388 print_stats(1, "Ext Root Scanning", ext_root_scan_time); 1389 print_stats(1, "Mark Stack Scanning", mark_stack_scan_time); 1390 print_stats(1, "Scan RS", scan_rs_time); 1391 print_stats(1, "Object Copying", obj_copy_time); 1392 } 1393 #ifndef PRODUCT 1394 print_stats(1, "Cur Clear CC", _cur_clear_cc_time_ms); 1395 print_stats(1, "Cum Clear CC", _cum_clear_cc_time_ms); 1396 print_stats(1, "Min Clear CC", _min_clear_cc_time_ms); 1397 print_stats(1, "Max Clear CC", _max_clear_cc_time_ms); 1398 if (_num_cc_clears > 0) { 1399 print_stats(1, "Avg Clear CC", _cum_clear_cc_time_ms / ((double)_num_cc_clears)); 1400 } 1401 #endif 1402 print_stats(1, "Other", other_time_ms); 1403 print_stats(2, "Choose CSet", _recorded_young_cset_choice_time_ms); 1404 1405 for (int i = 0; i < _aux_num; ++i) { 1406 if (_cur_aux_times_set[i]) { 1407 char buffer[96]; 1408 sprintf(buffer, "Aux%d", i); 1409 print_stats(1, buffer, _cur_aux_times_ms[i]); 1410 } 1411 } 1412 } 1413 1414 _all_pause_times_ms->add(elapsed_ms); 1415 if (update_stats) { 1416 summary->record_total_time_ms(elapsed_ms); 1417 summary->record_other_time_ms(other_time_ms); 1418 } 1419 for (int i = 0; i < _aux_num; ++i) 1420 if (_cur_aux_times_set[i]) 1421 _all_aux_times_ms[i].add(_cur_aux_times_ms[i]); 1422 1423 // Reset marks-between-pauses counter. 1424 _n_marks_since_last_pause = 0; 1425 1426 // Update the efficiency-since-mark vars. 1427 double proc_ms = elapsed_ms * (double) _parallel_gc_threads; 1428 if (elapsed_ms < MIN_TIMER_GRANULARITY) { 1429 // This usually happens due to the timer not having the required 1430 // granularity. Some Linuxes are the usual culprits. 1431 // We'll just set it to something (arbitrarily) small. 1432 proc_ms = 1.0; 1433 } 1434 double cur_efficiency = (double) freed_bytes / proc_ms; 1435 1436 bool new_in_marking_window = _in_marking_window; 1437 bool new_in_marking_window_im = false; 1438 if (during_initial_mark_pause()) { 1439 new_in_marking_window = true; 1440 new_in_marking_window_im = true; 1441 } 1442 1443 if (_last_full_young_gc) { 1444 set_full_young_gcs(false); 1445 _last_full_young_gc = false; 1446 } 1447 1448 if ( !_last_young_gc_full ) { 1449 if ( _should_revert_to_full_young_gcs || 1450 _known_garbage_ratio < 0.05 || 1451 (adaptive_young_list_length() && 1452 (get_gc_eff_factor() * cur_efficiency < predict_young_gc_eff())) ) { 1453 set_full_young_gcs(true); 1454 } 1455 } 1456 _should_revert_to_full_young_gcs = false; 1457 1458 if (_last_young_gc_full && !_during_marking) { 1459 _young_gc_eff_seq->add(cur_efficiency); 1460 } 1461 1462 _short_lived_surv_rate_group->start_adding_regions(); 1463 // do that for any other surv rate groupsx 1464 1465 // <NEW PREDICTION> 1466 1467 if (update_stats) { 1468 double pause_time_ms = elapsed_ms; 1469 1470 size_t diff = 0; 1471 if (_max_pending_cards >= _pending_cards) 1472 diff = _max_pending_cards - _pending_cards; 1473 _pending_card_diff_seq->add((double) diff); 1474 1475 double cost_per_card_ms = 0.0; 1476 if (_pending_cards > 0) { 1477 cost_per_card_ms = update_rs_time / (double) _pending_cards; 1478 _cost_per_card_ms_seq->add(cost_per_card_ms); 1479 } 1480 1481 size_t cards_scanned = _g1->cards_scanned(); 1482 1483 double cost_per_entry_ms = 0.0; 1484 if (cards_scanned > 10) { 1485 cost_per_entry_ms = scan_rs_time / (double) cards_scanned; 1486 if (_last_young_gc_full) 1487 _cost_per_entry_ms_seq->add(cost_per_entry_ms); 1488 else 1489 _partially_young_cost_per_entry_ms_seq->add(cost_per_entry_ms); 1490 } 1491 1492 if (_max_rs_lengths > 0) { 1493 double cards_per_entry_ratio = 1494 (double) cards_scanned / (double) _max_rs_lengths; 1495 if (_last_young_gc_full) 1496 _fully_young_cards_per_entry_ratio_seq->add(cards_per_entry_ratio); 1497 else 1498 _partially_young_cards_per_entry_ratio_seq->add(cards_per_entry_ratio); 1499 } 1500 1501 size_t rs_length_diff = _max_rs_lengths - _recorded_rs_lengths; 1502 if (rs_length_diff >= 0) 1503 _rs_length_diff_seq->add((double) rs_length_diff); 1504 1505 size_t copied_bytes = surviving_bytes; 1506 double cost_per_byte_ms = 0.0; 1507 if (copied_bytes > 0) { 1508 cost_per_byte_ms = obj_copy_time / (double) copied_bytes; 1509 if (_in_marking_window) 1510 _cost_per_byte_ms_during_cm_seq->add(cost_per_byte_ms); 1511 else 1512 _cost_per_byte_ms_seq->add(cost_per_byte_ms); 1513 } 1514 1515 double all_other_time_ms = pause_time_ms - 1516 (update_rs_time + scan_rs_time + obj_copy_time + 1517 _mark_closure_time_ms + termination_time); 1518 1519 double young_other_time_ms = 0.0; 1520 if (_recorded_young_regions > 0) { 1521 young_other_time_ms = 1522 _recorded_young_cset_choice_time_ms + 1523 _recorded_young_free_cset_time_ms; 1524 _young_other_cost_per_region_ms_seq->add(young_other_time_ms / 1525 (double) _recorded_young_regions); 1526 } 1527 double non_young_other_time_ms = 0.0; 1528 if (_recorded_non_young_regions > 0) { 1529 non_young_other_time_ms = 1530 _recorded_non_young_cset_choice_time_ms + 1531 _recorded_non_young_free_cset_time_ms; 1532 1533 _non_young_other_cost_per_region_ms_seq->add(non_young_other_time_ms / 1534 (double) _recorded_non_young_regions); 1535 } 1536 1537 double constant_other_time_ms = all_other_time_ms - 1538 (young_other_time_ms + non_young_other_time_ms); 1539 _constant_other_time_ms_seq->add(constant_other_time_ms); 1540 1541 double survival_ratio = 0.0; 1542 if (_bytes_in_collection_set_before_gc > 0) { 1543 survival_ratio = (double) _bytes_copied_during_gc / 1544 (double) _bytes_in_collection_set_before_gc; 1545 } 1546 1547 _pending_cards_seq->add((double) _pending_cards); 1548 _scanned_cards_seq->add((double) cards_scanned); 1549 _rs_lengths_seq->add((double) _max_rs_lengths); 1550 1551 double expensive_region_limit_ms = 1552 (double) MaxGCPauseMillis - predict_constant_other_time_ms(); 1553 if (expensive_region_limit_ms < 0.0) { 1554 // this means that the other time was predicted to be longer than 1555 // than the max pause time 1556 expensive_region_limit_ms = (double) MaxGCPauseMillis; 1557 } 1558 _expensive_region_limit_ms = expensive_region_limit_ms; 1559 1560 if (PREDICTIONS_VERBOSE) { 1561 gclog_or_tty->print_cr(""); 1562 gclog_or_tty->print_cr("PREDICTIONS %1.4lf %d " 1563 "REGIONS %d %d %d " 1564 "PENDING_CARDS %d %d " 1565 "CARDS_SCANNED %d %d " 1566 "RS_LENGTHS %d %d " 1567 "RS_UPDATE %1.6lf %1.6lf RS_SCAN %1.6lf %1.6lf " 1568 "SURVIVAL_RATIO %1.6lf %1.6lf " 1569 "OBJECT_COPY %1.6lf %1.6lf OTHER_CONSTANT %1.6lf %1.6lf " 1570 "OTHER_YOUNG %1.6lf %1.6lf " 1571 "OTHER_NON_YOUNG %1.6lf %1.6lf " 1572 "VTIME_DIFF %1.6lf TERMINATION %1.6lf " 1573 "ELAPSED %1.6lf %1.6lf ", 1574 _cur_collection_start_sec, 1575 (!_last_young_gc_full) ? 2 : 1576 (last_pause_included_initial_mark) ? 1 : 0, 1577 _recorded_region_num, 1578 _recorded_young_regions, 1579 _recorded_non_young_regions, 1580 _predicted_pending_cards, _pending_cards, 1581 _predicted_cards_scanned, cards_scanned, 1582 _predicted_rs_lengths, _max_rs_lengths, 1583 _predicted_rs_update_time_ms, update_rs_time, 1584 _predicted_rs_scan_time_ms, scan_rs_time, 1585 _predicted_survival_ratio, survival_ratio, 1586 _predicted_object_copy_time_ms, obj_copy_time, 1587 _predicted_constant_other_time_ms, constant_other_time_ms, 1588 _predicted_young_other_time_ms, young_other_time_ms, 1589 _predicted_non_young_other_time_ms, 1590 non_young_other_time_ms, 1591 _vtime_diff_ms, termination_time, 1592 _predicted_pause_time_ms, elapsed_ms); 1593 } 1594 1595 if (G1PolicyVerbose > 0) { 1596 gclog_or_tty->print_cr("Pause Time, predicted: %1.4lfms (predicted %s), actual: %1.4lfms", 1597 _predicted_pause_time_ms, 1598 (_within_target) ? "within" : "outside", 1599 elapsed_ms); 1600 } 1601 1602 } 1603 1604 _in_marking_window = new_in_marking_window; 1605 _in_marking_window_im = new_in_marking_window_im; 1606 _free_regions_at_end_of_collection = _g1->free_regions(); 1607 calculate_young_list_min_length(); 1608 calculate_young_list_target_length(); 1609 1610 // Note that _mmu_tracker->max_gc_time() returns the time in seconds. 1611 double update_rs_time_goal_ms = _mmu_tracker->max_gc_time() * MILLIUNITS * G1RSetUpdatingPauseTimePercent / 100.0; 1612 adjust_concurrent_refinement(update_rs_time, update_rs_processed_buffers, update_rs_time_goal_ms); 1613 // </NEW PREDICTION> 1614 } 1615 1616 #define EXT_SIZE_FORMAT "%d%s" 1617 #define EXT_SIZE_PARAMS(bytes) \ 1618 byte_size_in_proper_unit((bytes)), \ 1619 proper_unit_for_byte_size((bytes)) 1620 1621 void G1CollectorPolicy::print_heap_transition() { 1622 if (PrintGCDetails) { 1623 YoungList* young_list = _g1->young_list(); 1624 size_t eden_bytes = young_list->eden_used_bytes(); 1625 size_t survivor_bytes = young_list->survivor_used_bytes(); 1626 size_t used_before_gc = _cur_collection_pause_used_at_start_bytes; 1627 size_t used = _g1->used(); 1628 size_t capacity = _g1->capacity(); 1629 1630 gclog_or_tty->print_cr( 1631 " [Eden: "EXT_SIZE_FORMAT"->"EXT_SIZE_FORMAT" " 1632 "Survivors: "EXT_SIZE_FORMAT"->"EXT_SIZE_FORMAT" " 1633 "Heap: "EXT_SIZE_FORMAT"("EXT_SIZE_FORMAT")->" 1634 EXT_SIZE_FORMAT"("EXT_SIZE_FORMAT")]", 1635 EXT_SIZE_PARAMS(_eden_bytes_before_gc), 1636 EXT_SIZE_PARAMS(eden_bytes), 1637 EXT_SIZE_PARAMS(_survivor_bytes_before_gc), 1638 EXT_SIZE_PARAMS(survivor_bytes), 1639 EXT_SIZE_PARAMS(used_before_gc), 1640 EXT_SIZE_PARAMS(_capacity_before_gc), 1641 EXT_SIZE_PARAMS(used), 1642 EXT_SIZE_PARAMS(capacity)); 1643 } else if (PrintGC) { 1644 _g1->print_size_transition(gclog_or_tty, 1645 _cur_collection_pause_used_at_start_bytes, 1646 _g1->used(), _g1->capacity()); 1647 } 1648 } 1649 1650 // <NEW PREDICTION> 1651 1652 void G1CollectorPolicy::adjust_concurrent_refinement(double update_rs_time, 1653 double update_rs_processed_buffers, 1654 double goal_ms) { 1655 DirtyCardQueueSet& dcqs = JavaThread::dirty_card_queue_set(); 1656 ConcurrentG1Refine *cg1r = G1CollectedHeap::heap()->concurrent_g1_refine(); 1657 1658 if (G1UseAdaptiveConcRefinement) { 1659 const int k_gy = 3, k_gr = 6; 1660 const double inc_k = 1.1, dec_k = 0.9; 1661 1662 int g = cg1r->green_zone(); 1663 if (update_rs_time > goal_ms) { 1664 g = (int)(g * dec_k); // Can become 0, that's OK. That would mean a mutator-only processing. 1665 } else { 1666 if (update_rs_time < goal_ms && update_rs_processed_buffers > g) { 1667 g = (int)MAX2(g * inc_k, g + 1.0); 1668 } 1669 } 1670 // Change the refinement threads params 1671 cg1r->set_green_zone(g); 1672 cg1r->set_yellow_zone(g * k_gy); 1673 cg1r->set_red_zone(g * k_gr); 1674 cg1r->reinitialize_threads(); 1675 1676 int processing_threshold_delta = MAX2((int)(cg1r->green_zone() * sigma()), 1); 1677 int processing_threshold = MIN2(cg1r->green_zone() + processing_threshold_delta, 1678 cg1r->yellow_zone()); 1679 // Change the barrier params 1680 dcqs.set_process_completed_threshold(processing_threshold); 1681 dcqs.set_max_completed_queue(cg1r->red_zone()); 1682 } 1683 1684 int curr_queue_size = dcqs.completed_buffers_num(); 1685 if (curr_queue_size >= cg1r->yellow_zone()) { 1686 dcqs.set_completed_queue_padding(curr_queue_size); 1687 } else { 1688 dcqs.set_completed_queue_padding(0); 1689 } 1690 dcqs.notify_if_necessary(); 1691 } 1692 1693 double 1694 G1CollectorPolicy:: 1695 predict_young_collection_elapsed_time_ms(size_t adjustment) { 1696 guarantee( adjustment == 0 || adjustment == 1, "invariant" ); 1697 1698 G1CollectedHeap* g1h = G1CollectedHeap::heap(); 1699 size_t young_num = g1h->young_list()->length(); 1700 if (young_num == 0) 1701 return 0.0; 1702 1703 young_num += adjustment; 1704 size_t pending_cards = predict_pending_cards(); 1705 size_t rs_lengths = g1h->young_list()->sampled_rs_lengths() + 1706 predict_rs_length_diff(); 1707 size_t card_num; 1708 if (full_young_gcs()) 1709 card_num = predict_young_card_num(rs_lengths); 1710 else 1711 card_num = predict_non_young_card_num(rs_lengths); 1712 size_t young_byte_size = young_num * HeapRegion::GrainBytes; 1713 double accum_yg_surv_rate = 1714 _short_lived_surv_rate_group->accum_surv_rate(adjustment); 1715 1716 size_t bytes_to_copy = 1717 (size_t) (accum_yg_surv_rate * (double) HeapRegion::GrainBytes); 1718 1719 return 1720 predict_rs_update_time_ms(pending_cards) + 1721 predict_rs_scan_time_ms(card_num) + 1722 predict_object_copy_time_ms(bytes_to_copy) + 1723 predict_young_other_time_ms(young_num) + 1724 predict_constant_other_time_ms(); 1725 } 1726 1727 double 1728 G1CollectorPolicy::predict_base_elapsed_time_ms(size_t pending_cards) { 1729 size_t rs_length = predict_rs_length_diff(); 1730 size_t card_num; 1731 if (full_young_gcs()) 1732 card_num = predict_young_card_num(rs_length); 1733 else 1734 card_num = predict_non_young_card_num(rs_length); 1735 return predict_base_elapsed_time_ms(pending_cards, card_num); 1736 } 1737 1738 double 1739 G1CollectorPolicy::predict_base_elapsed_time_ms(size_t pending_cards, 1740 size_t scanned_cards) { 1741 return 1742 predict_rs_update_time_ms(pending_cards) + 1743 predict_rs_scan_time_ms(scanned_cards) + 1744 predict_constant_other_time_ms(); 1745 } 1746 1747 double 1748 G1CollectorPolicy::predict_region_elapsed_time_ms(HeapRegion* hr, 1749 bool young) { 1750 size_t rs_length = hr->rem_set()->occupied(); 1751 size_t card_num; 1752 if (full_young_gcs()) 1753 card_num = predict_young_card_num(rs_length); 1754 else 1755 card_num = predict_non_young_card_num(rs_length); 1756 size_t bytes_to_copy = predict_bytes_to_copy(hr); 1757 1758 double region_elapsed_time_ms = 1759 predict_rs_scan_time_ms(card_num) + 1760 predict_object_copy_time_ms(bytes_to_copy); 1761 1762 if (young) 1763 region_elapsed_time_ms += predict_young_other_time_ms(1); 1764 else 1765 region_elapsed_time_ms += predict_non_young_other_time_ms(1); 1766 1767 return region_elapsed_time_ms; 1768 } 1769 1770 size_t 1771 G1CollectorPolicy::predict_bytes_to_copy(HeapRegion* hr) { 1772 size_t bytes_to_copy; 1773 if (hr->is_marked()) 1774 bytes_to_copy = hr->max_live_bytes(); 1775 else { 1776 guarantee( hr->is_young() && hr->age_in_surv_rate_group() != -1, 1777 "invariant" ); 1778 int age = hr->age_in_surv_rate_group(); 1779 double yg_surv_rate = predict_yg_surv_rate(age, hr->surv_rate_group()); 1780 bytes_to_copy = (size_t) ((double) hr->used() * yg_surv_rate); 1781 } 1782 1783 return bytes_to_copy; 1784 } 1785 1786 void 1787 G1CollectorPolicy::start_recording_regions() { 1788 _recorded_rs_lengths = 0; 1789 _recorded_young_regions = 0; 1790 _recorded_non_young_regions = 0; 1791 1792 #if PREDICTIONS_VERBOSE 1793 _recorded_marked_bytes = 0; 1794 _recorded_young_bytes = 0; 1795 _predicted_bytes_to_copy = 0; 1796 _predicted_rs_lengths = 0; 1797 _predicted_cards_scanned = 0; 1798 #endif // PREDICTIONS_VERBOSE 1799 } 1800 1801 void 1802 G1CollectorPolicy::record_cset_region_info(HeapRegion* hr, bool young) { 1803 #if PREDICTIONS_VERBOSE 1804 if (!young) { 1805 _recorded_marked_bytes += hr->max_live_bytes(); 1806 } 1807 _predicted_bytes_to_copy += predict_bytes_to_copy(hr); 1808 #endif // PREDICTIONS_VERBOSE 1809 1810 size_t rs_length = hr->rem_set()->occupied(); 1811 _recorded_rs_lengths += rs_length; 1812 } 1813 1814 void 1815 G1CollectorPolicy::record_non_young_cset_region(HeapRegion* hr) { 1816 assert(!hr->is_young(), "should not call this"); 1817 ++_recorded_non_young_regions; 1818 record_cset_region_info(hr, false); 1819 } 1820 1821 void 1822 G1CollectorPolicy::set_recorded_young_regions(size_t n_regions) { 1823 _recorded_young_regions = n_regions; 1824 } 1825 1826 void G1CollectorPolicy::set_recorded_young_bytes(size_t bytes) { 1827 #if PREDICTIONS_VERBOSE 1828 _recorded_young_bytes = bytes; 1829 #endif // PREDICTIONS_VERBOSE 1830 } 1831 1832 void G1CollectorPolicy::set_recorded_rs_lengths(size_t rs_lengths) { 1833 _recorded_rs_lengths = rs_lengths; 1834 } 1835 1836 void G1CollectorPolicy::set_predicted_bytes_to_copy(size_t bytes) { 1837 _predicted_bytes_to_copy = bytes; 1838 } 1839 1840 void 1841 G1CollectorPolicy::end_recording_regions() { 1842 // The _predicted_pause_time_ms field is referenced in code 1843 // not under PREDICTIONS_VERBOSE. Let's initialize it. 1844 _predicted_pause_time_ms = -1.0; 1845 1846 #if PREDICTIONS_VERBOSE 1847 _predicted_pending_cards = predict_pending_cards(); 1848 _predicted_rs_lengths = _recorded_rs_lengths + predict_rs_length_diff(); 1849 if (full_young_gcs()) 1850 _predicted_cards_scanned += predict_young_card_num(_predicted_rs_lengths); 1851 else 1852 _predicted_cards_scanned += 1853 predict_non_young_card_num(_predicted_rs_lengths); 1854 _recorded_region_num = _recorded_young_regions + _recorded_non_young_regions; 1855 1856 _predicted_rs_update_time_ms = 1857 predict_rs_update_time_ms(_g1->pending_card_num()); 1858 _predicted_rs_scan_time_ms = 1859 predict_rs_scan_time_ms(_predicted_cards_scanned); 1860 _predicted_object_copy_time_ms = 1861 predict_object_copy_time_ms(_predicted_bytes_to_copy); 1862 _predicted_constant_other_time_ms = 1863 predict_constant_other_time_ms(); 1864 _predicted_young_other_time_ms = 1865 predict_young_other_time_ms(_recorded_young_regions); 1866 _predicted_non_young_other_time_ms = 1867 predict_non_young_other_time_ms(_recorded_non_young_regions); 1868 1869 _predicted_pause_time_ms = 1870 _predicted_rs_update_time_ms + 1871 _predicted_rs_scan_time_ms + 1872 _predicted_object_copy_time_ms + 1873 _predicted_constant_other_time_ms + 1874 _predicted_young_other_time_ms + 1875 _predicted_non_young_other_time_ms; 1876 #endif // PREDICTIONS_VERBOSE 1877 } 1878 1879 void G1CollectorPolicy::check_if_region_is_too_expensive(double 1880 predicted_time_ms) { 1881 // I don't think we need to do this when in young GC mode since 1882 // marking will be initiated next time we hit the soft limit anyway... 1883 if (predicted_time_ms > _expensive_region_limit_ms) { 1884 // no point in doing another partial one 1885 _should_revert_to_full_young_gcs = true; 1886 } 1887 } 1888 1889 // </NEW PREDICTION> 1890 1891 1892 void G1CollectorPolicy::update_recent_gc_times(double end_time_sec, 1893 double elapsed_ms) { 1894 _recent_gc_times_ms->add(elapsed_ms); 1895 _recent_prev_end_times_for_all_gcs_sec->add(end_time_sec); 1896 _prev_collection_pause_end_ms = end_time_sec * 1000.0; 1897 } 1898 1899 double G1CollectorPolicy::recent_avg_time_for_pauses_ms() { 1900 if (_recent_pause_times_ms->num() == 0) { 1901 return (double) MaxGCPauseMillis; 1902 } 1903 return _recent_pause_times_ms->avg(); 1904 } 1905 1906 double G1CollectorPolicy::recent_avg_time_for_rs_scan_ms() { 1907 if (_recent_rs_scan_times_ms->num() == 0) { 1908 return (double)MaxGCPauseMillis/3.0; 1909 } 1910 return _recent_rs_scan_times_ms->avg(); 1911 } 1912 1913 int G1CollectorPolicy::number_of_recent_gcs() { 1914 assert(_recent_rs_scan_times_ms->num() == 1915 _recent_pause_times_ms->num(), "Sequence out of sync"); 1916 assert(_recent_pause_times_ms->num() == 1917 _recent_CS_bytes_used_before->num(), "Sequence out of sync"); 1918 assert(_recent_CS_bytes_used_before->num() == 1919 _recent_CS_bytes_surviving->num(), "Sequence out of sync"); 1920 1921 return _recent_pause_times_ms->num(); 1922 } 1923 1924 double G1CollectorPolicy::recent_avg_survival_fraction() { 1925 return recent_avg_survival_fraction_work(_recent_CS_bytes_surviving, 1926 _recent_CS_bytes_used_before); 1927 } 1928 1929 double G1CollectorPolicy::last_survival_fraction() { 1930 return last_survival_fraction_work(_recent_CS_bytes_surviving, 1931 _recent_CS_bytes_used_before); 1932 } 1933 1934 double 1935 G1CollectorPolicy::recent_avg_survival_fraction_work(TruncatedSeq* surviving, 1936 TruncatedSeq* before) { 1937 assert(surviving->num() == before->num(), "Sequence out of sync"); 1938 if (before->sum() > 0.0) { 1939 double recent_survival_rate = surviving->sum() / before->sum(); 1940 // We exempt parallel collection from this check because Alloc Buffer 1941 // fragmentation can produce negative collections. 1942 // Further, we're now always doing parallel collection. But I'm still 1943 // leaving this here as a placeholder for a more precise assertion later. 1944 // (DLD, 10/05.) 1945 assert((true || G1CollectedHeap::use_parallel_gc_threads()) || 1946 _g1->evacuation_failed() || 1947 recent_survival_rate <= 1.0, "Or bad frac"); 1948 return recent_survival_rate; 1949 } else { 1950 return 1.0; // Be conservative. 1951 } 1952 } 1953 1954 double 1955 G1CollectorPolicy::last_survival_fraction_work(TruncatedSeq* surviving, 1956 TruncatedSeq* before) { 1957 assert(surviving->num() == before->num(), "Sequence out of sync"); 1958 if (surviving->num() > 0 && before->last() > 0.0) { 1959 double last_survival_rate = surviving->last() / before->last(); 1960 // We exempt parallel collection from this check because Alloc Buffer 1961 // fragmentation can produce negative collections. 1962 // Further, we're now always doing parallel collection. But I'm still 1963 // leaving this here as a placeholder for a more precise assertion later. 1964 // (DLD, 10/05.) 1965 assert((true || G1CollectedHeap::use_parallel_gc_threads()) || 1966 last_survival_rate <= 1.0, "Or bad frac"); 1967 return last_survival_rate; 1968 } else { 1969 return 1.0; 1970 } 1971 } 1972 1973 static const int survival_min_obs = 5; 1974 static double survival_min_obs_limits[] = { 0.9, 0.7, 0.5, 0.3, 0.1 }; 1975 static const double min_survival_rate = 0.1; 1976 1977 double 1978 G1CollectorPolicy::conservative_avg_survival_fraction_work(double avg, 1979 double latest) { 1980 double res = avg; 1981 if (number_of_recent_gcs() < survival_min_obs) { 1982 res = MAX2(res, survival_min_obs_limits[number_of_recent_gcs()]); 1983 } 1984 res = MAX2(res, latest); 1985 res = MAX2(res, min_survival_rate); 1986 // In the parallel case, LAB fragmentation can produce "negative 1987 // collections"; so can evac failure. Cap at 1.0 1988 res = MIN2(res, 1.0); 1989 return res; 1990 } 1991 1992 size_t G1CollectorPolicy::expansion_amount() { 1993 if ((recent_avg_pause_time_ratio() * 100.0) > _gc_overhead_perc) { 1994 // We will double the existing space, or take 1995 // G1ExpandByPercentOfAvailable % of the available expansion 1996 // space, whichever is smaller, bounded below by a minimum 1997 // expansion (unless that's all that's left.) 1998 const size_t min_expand_bytes = 1*M; 1999 size_t reserved_bytes = _g1->max_capacity(); 2000 size_t committed_bytes = _g1->capacity(); 2001 size_t uncommitted_bytes = reserved_bytes - committed_bytes; 2002 size_t expand_bytes; 2003 size_t expand_bytes_via_pct = 2004 uncommitted_bytes * G1ExpandByPercentOfAvailable / 100; 2005 expand_bytes = MIN2(expand_bytes_via_pct, committed_bytes); 2006 expand_bytes = MAX2(expand_bytes, min_expand_bytes); 2007 expand_bytes = MIN2(expand_bytes, uncommitted_bytes); 2008 if (G1PolicyVerbose > 1) { 2009 gclog_or_tty->print("Decided to expand: ratio = %5.2f, " 2010 "committed = %d%s, uncommited = %d%s, via pct = %d%s.\n" 2011 " Answer = %d.\n", 2012 recent_avg_pause_time_ratio(), 2013 byte_size_in_proper_unit(committed_bytes), 2014 proper_unit_for_byte_size(committed_bytes), 2015 byte_size_in_proper_unit(uncommitted_bytes), 2016 proper_unit_for_byte_size(uncommitted_bytes), 2017 byte_size_in_proper_unit(expand_bytes_via_pct), 2018 proper_unit_for_byte_size(expand_bytes_via_pct), 2019 byte_size_in_proper_unit(expand_bytes), 2020 proper_unit_for_byte_size(expand_bytes)); 2021 } 2022 return expand_bytes; 2023 } else { 2024 return 0; 2025 } 2026 } 2027 2028 void G1CollectorPolicy::note_start_of_mark_thread() { 2029 _mark_thread_startup_sec = os::elapsedTime(); 2030 } 2031 2032 class CountCSClosure: public HeapRegionClosure { 2033 G1CollectorPolicy* _g1_policy; 2034 public: 2035 CountCSClosure(G1CollectorPolicy* g1_policy) : 2036 _g1_policy(g1_policy) {} 2037 bool doHeapRegion(HeapRegion* r) { 2038 _g1_policy->_bytes_in_collection_set_before_gc += r->used(); 2039 return false; 2040 } 2041 }; 2042 2043 void G1CollectorPolicy::count_CS_bytes_used() { 2044 CountCSClosure cs_closure(this); 2045 _g1->collection_set_iterate(&cs_closure); 2046 } 2047 2048 void G1CollectorPolicy::print_summary (int level, 2049 const char* str, 2050 NumberSeq* seq) const { 2051 double sum = seq->sum(); 2052 LineBuffer(level + 1).append_and_print_cr("%-24s = %8.2lf s (avg = %8.2lf ms)", 2053 str, sum / 1000.0, seq->avg()); 2054 } 2055 2056 void G1CollectorPolicy::print_summary_sd (int level, 2057 const char* str, 2058 NumberSeq* seq) const { 2059 print_summary(level, str, seq); 2060 LineBuffer(level + 6).append_and_print_cr("(num = %5d, std dev = %8.2lf ms, max = %8.2lf ms)", 2061 seq->num(), seq->sd(), seq->maximum()); 2062 } 2063 2064 void G1CollectorPolicy::check_other_times(int level, 2065 NumberSeq* other_times_ms, 2066 NumberSeq* calc_other_times_ms) const { 2067 bool should_print = false; 2068 LineBuffer buf(level + 2); 2069 2070 double max_sum = MAX2(fabs(other_times_ms->sum()), 2071 fabs(calc_other_times_ms->sum())); 2072 double min_sum = MIN2(fabs(other_times_ms->sum()), 2073 fabs(calc_other_times_ms->sum())); 2074 double sum_ratio = max_sum / min_sum; 2075 if (sum_ratio > 1.1) { 2076 should_print = true; 2077 buf.append_and_print_cr("## CALCULATED OTHER SUM DOESN'T MATCH RECORDED ###"); 2078 } 2079 2080 double max_avg = MAX2(fabs(other_times_ms->avg()), 2081 fabs(calc_other_times_ms->avg())); 2082 double min_avg = MIN2(fabs(other_times_ms->avg()), 2083 fabs(calc_other_times_ms->avg())); 2084 double avg_ratio = max_avg / min_avg; 2085 if (avg_ratio > 1.1) { 2086 should_print = true; 2087 buf.append_and_print_cr("## CALCULATED OTHER AVG DOESN'T MATCH RECORDED ###"); 2088 } 2089 2090 if (other_times_ms->sum() < -0.01) { 2091 buf.append_and_print_cr("## RECORDED OTHER SUM IS NEGATIVE ###"); 2092 } 2093 2094 if (other_times_ms->avg() < -0.01) { 2095 buf.append_and_print_cr("## RECORDED OTHER AVG IS NEGATIVE ###"); 2096 } 2097 2098 if (calc_other_times_ms->sum() < -0.01) { 2099 should_print = true; 2100 buf.append_and_print_cr("## CALCULATED OTHER SUM IS NEGATIVE ###"); 2101 } 2102 2103 if (calc_other_times_ms->avg() < -0.01) { 2104 should_print = true; 2105 buf.append_and_print_cr("## CALCULATED OTHER AVG IS NEGATIVE ###"); 2106 } 2107 2108 if (should_print) 2109 print_summary(level, "Other(Calc)", calc_other_times_ms); 2110 } 2111 2112 void G1CollectorPolicy::print_summary(PauseSummary* summary) const { 2113 bool parallel = G1CollectedHeap::use_parallel_gc_threads(); 2114 MainBodySummary* body_summary = summary->main_body_summary(); 2115 if (summary->get_total_seq()->num() > 0) { 2116 print_summary_sd(0, "Evacuation Pauses", summary->get_total_seq()); 2117 if (body_summary != NULL) { 2118 print_summary(1, "SATB Drain", body_summary->get_satb_drain_seq()); 2119 if (parallel) { 2120 print_summary(1, "Parallel Time", body_summary->get_parallel_seq()); 2121 print_summary(2, "Update RS", body_summary->get_update_rs_seq()); 2122 print_summary(2, "Ext Root Scanning", 2123 body_summary->get_ext_root_scan_seq()); 2124 print_summary(2, "Mark Stack Scanning", 2125 body_summary->get_mark_stack_scan_seq()); 2126 print_summary(2, "Scan RS", body_summary->get_scan_rs_seq()); 2127 print_summary(2, "Object Copy", body_summary->get_obj_copy_seq()); 2128 print_summary(2, "Termination", body_summary->get_termination_seq()); 2129 print_summary(2, "Other", body_summary->get_parallel_other_seq()); 2130 { 2131 NumberSeq* other_parts[] = { 2132 body_summary->get_update_rs_seq(), 2133 body_summary->get_ext_root_scan_seq(), 2134 body_summary->get_mark_stack_scan_seq(), 2135 body_summary->get_scan_rs_seq(), 2136 body_summary->get_obj_copy_seq(), 2137 body_summary->get_termination_seq() 2138 }; 2139 NumberSeq calc_other_times_ms(body_summary->get_parallel_seq(), 2140 6, other_parts); 2141 check_other_times(2, body_summary->get_parallel_other_seq(), 2142 &calc_other_times_ms); 2143 } 2144 print_summary(1, "Mark Closure", body_summary->get_mark_closure_seq()); 2145 print_summary(1, "Clear CT", body_summary->get_clear_ct_seq()); 2146 } else { 2147 print_summary(1, "Update RS", body_summary->get_update_rs_seq()); 2148 print_summary(1, "Ext Root Scanning", 2149 body_summary->get_ext_root_scan_seq()); 2150 print_summary(1, "Mark Stack Scanning", 2151 body_summary->get_mark_stack_scan_seq()); 2152 print_summary(1, "Scan RS", body_summary->get_scan_rs_seq()); 2153 print_summary(1, "Object Copy", body_summary->get_obj_copy_seq()); 2154 } 2155 } 2156 print_summary(1, "Other", summary->get_other_seq()); 2157 { 2158 if (body_summary != NULL) { 2159 NumberSeq calc_other_times_ms; 2160 if (parallel) { 2161 // parallel 2162 NumberSeq* other_parts[] = { 2163 body_summary->get_satb_drain_seq(), 2164 body_summary->get_parallel_seq(), 2165 body_summary->get_clear_ct_seq() 2166 }; 2167 calc_other_times_ms = NumberSeq(summary->get_total_seq(), 2168 3, other_parts); 2169 } else { 2170 // serial 2171 NumberSeq* other_parts[] = { 2172 body_summary->get_satb_drain_seq(), 2173 body_summary->get_update_rs_seq(), 2174 body_summary->get_ext_root_scan_seq(), 2175 body_summary->get_mark_stack_scan_seq(), 2176 body_summary->get_scan_rs_seq(), 2177 body_summary->get_obj_copy_seq() 2178 }; 2179 calc_other_times_ms = NumberSeq(summary->get_total_seq(), 2180 6, other_parts); 2181 } 2182 check_other_times(1, summary->get_other_seq(), &calc_other_times_ms); 2183 } 2184 } 2185 } else { 2186 LineBuffer(1).append_and_print_cr("none"); 2187 } 2188 LineBuffer(0).append_and_print_cr(""); 2189 } 2190 2191 void G1CollectorPolicy::print_tracing_info() const { 2192 if (TraceGen0Time) { 2193 gclog_or_tty->print_cr("ALL PAUSES"); 2194 print_summary_sd(0, "Total", _all_pause_times_ms); 2195 gclog_or_tty->print_cr(""); 2196 gclog_or_tty->print_cr(""); 2197 gclog_or_tty->print_cr(" Full Young GC Pauses: %8d", _full_young_pause_num); 2198 gclog_or_tty->print_cr(" Partial Young GC Pauses: %8d", _partial_young_pause_num); 2199 gclog_or_tty->print_cr(""); 2200 2201 gclog_or_tty->print_cr("EVACUATION PAUSES"); 2202 print_summary(_summary); 2203 2204 gclog_or_tty->print_cr("MISC"); 2205 print_summary_sd(0, "Stop World", _all_stop_world_times_ms); 2206 print_summary_sd(0, "Yields", _all_yield_times_ms); 2207 for (int i = 0; i < _aux_num; ++i) { 2208 if (_all_aux_times_ms[i].num() > 0) { 2209 char buffer[96]; 2210 sprintf(buffer, "Aux%d", i); 2211 print_summary_sd(0, buffer, &_all_aux_times_ms[i]); 2212 } 2213 } 2214 2215 size_t all_region_num = _region_num_young + _region_num_tenured; 2216 gclog_or_tty->print_cr(" New Regions %8d, Young %8d (%6.2lf%%), " 2217 "Tenured %8d (%6.2lf%%)", 2218 all_region_num, 2219 _region_num_young, 2220 (double) _region_num_young / (double) all_region_num * 100.0, 2221 _region_num_tenured, 2222 (double) _region_num_tenured / (double) all_region_num * 100.0); 2223 } 2224 if (TraceGen1Time) { 2225 if (_all_full_gc_times_ms->num() > 0) { 2226 gclog_or_tty->print("\n%4d full_gcs: total time = %8.2f s", 2227 _all_full_gc_times_ms->num(), 2228 _all_full_gc_times_ms->sum() / 1000.0); 2229 gclog_or_tty->print_cr(" (avg = %8.2fms).", _all_full_gc_times_ms->avg()); 2230 gclog_or_tty->print_cr(" [std. dev = %8.2f ms, max = %8.2f ms]", 2231 _all_full_gc_times_ms->sd(), 2232 _all_full_gc_times_ms->maximum()); 2233 } 2234 } 2235 } 2236 2237 void G1CollectorPolicy::print_yg_surv_rate_info() const { 2238 #ifndef PRODUCT 2239 _short_lived_surv_rate_group->print_surv_rate_summary(); 2240 // add this call for any other surv rate groups 2241 #endif // PRODUCT 2242 } 2243 2244 void 2245 G1CollectorPolicy::update_region_num(bool young) { 2246 if (young) { 2247 ++_region_num_young; 2248 } else { 2249 ++_region_num_tenured; 2250 } 2251 } 2252 2253 #ifndef PRODUCT 2254 // for debugging, bit of a hack... 2255 static char* 2256 region_num_to_mbs(int length) { 2257 static char buffer[64]; 2258 double bytes = (double) (length * HeapRegion::GrainBytes); 2259 double mbs = bytes / (double) (1024 * 1024); 2260 sprintf(buffer, "%7.2lfMB", mbs); 2261 return buffer; 2262 } 2263 #endif // PRODUCT 2264 2265 size_t G1CollectorPolicy::max_regions(int purpose) { 2266 switch (purpose) { 2267 case GCAllocForSurvived: 2268 return _max_survivor_regions; 2269 case GCAllocForTenured: 2270 return REGIONS_UNLIMITED; 2271 default: 2272 ShouldNotReachHere(); 2273 return REGIONS_UNLIMITED; 2274 }; 2275 } 2276 2277 void G1CollectorPolicy::calculate_max_gc_locker_expansion() { 2278 size_t expansion_region_num = 0; 2279 if (GCLockerEdenExpansionPercent > 0) { 2280 double perc = (double) GCLockerEdenExpansionPercent / 100.0; 2281 double expansion_region_num_d = perc * (double) _young_list_target_length; 2282 // We use ceiling so that if expansion_region_num_d is > 0.0 (but 2283 // less than 1.0) we'll get 1. 2284 expansion_region_num = (size_t) ceil(expansion_region_num_d); 2285 } else { 2286 assert(expansion_region_num == 0, "sanity"); 2287 } 2288 _young_list_max_length = _young_list_target_length + expansion_region_num; 2289 assert(_young_list_target_length <= _young_list_max_length, "post-condition"); 2290 } 2291 2292 // Calculates survivor space parameters. 2293 void G1CollectorPolicy::calculate_survivors_policy() 2294 { 2295 if (G1FixedSurvivorSpaceSize == 0) { 2296 _max_survivor_regions = _young_list_target_length / SurvivorRatio; 2297 } else { 2298 _max_survivor_regions = G1FixedSurvivorSpaceSize / HeapRegion::GrainBytes; 2299 } 2300 2301 if (G1FixedTenuringThreshold) { 2302 _tenuring_threshold = MaxTenuringThreshold; 2303 } else { 2304 _tenuring_threshold = _survivors_age_table.compute_tenuring_threshold( 2305 HeapRegion::GrainWords * _max_survivor_regions); 2306 } 2307 } 2308 2309 #ifndef PRODUCT 2310 class HRSortIndexIsOKClosure: public HeapRegionClosure { 2311 CollectionSetChooser* _chooser; 2312 public: 2313 HRSortIndexIsOKClosure(CollectionSetChooser* chooser) : 2314 _chooser(chooser) {} 2315 2316 bool doHeapRegion(HeapRegion* r) { 2317 if (!r->continuesHumongous()) { 2318 assert(_chooser->regionProperlyOrdered(r), "Ought to be."); 2319 } 2320 return false; 2321 } 2322 }; 2323 2324 bool G1CollectorPolicy_BestRegionsFirst::assertMarkedBytesDataOK() { 2325 HRSortIndexIsOKClosure cl(_collectionSetChooser); 2326 _g1->heap_region_iterate(&cl); 2327 return true; 2328 } 2329 #endif 2330 2331 bool 2332 G1CollectorPolicy::force_initial_mark_if_outside_cycle() { 2333 bool during_cycle = _g1->concurrent_mark()->cmThread()->during_cycle(); 2334 if (!during_cycle) { 2335 set_initiate_conc_mark_if_possible(); 2336 return true; 2337 } else { 2338 return false; 2339 } 2340 } 2341 2342 void 2343 G1CollectorPolicy::decide_on_conc_mark_initiation() { 2344 // We are about to decide on whether this pause will be an 2345 // initial-mark pause. 2346 2347 // First, during_initial_mark_pause() should not be already set. We 2348 // will set it here if we have to. However, it should be cleared by 2349 // the end of the pause (it's only set for the duration of an 2350 // initial-mark pause). 2351 assert(!during_initial_mark_pause(), "pre-condition"); 2352 2353 if (initiate_conc_mark_if_possible()) { 2354 // We had noticed on a previous pause that the heap occupancy has 2355 // gone over the initiating threshold and we should start a 2356 // concurrent marking cycle. So we might initiate one. 2357 2358 bool during_cycle = _g1->concurrent_mark()->cmThread()->during_cycle(); 2359 if (!during_cycle) { 2360 // The concurrent marking thread is not "during a cycle", i.e., 2361 // it has completed the last one. So we can go ahead and 2362 // initiate a new cycle. 2363 2364 set_during_initial_mark_pause(); 2365 2366 // And we can now clear initiate_conc_mark_if_possible() as 2367 // we've already acted on it. 2368 clear_initiate_conc_mark_if_possible(); 2369 } else { 2370 // The concurrent marking thread is still finishing up the 2371 // previous cycle. If we start one right now the two cycles 2372 // overlap. In particular, the concurrent marking thread might 2373 // be in the process of clearing the next marking bitmap (which 2374 // we will use for the next cycle if we start one). Starting a 2375 // cycle now will be bad given that parts of the marking 2376 // information might get cleared by the marking thread. And we 2377 // cannot wait for the marking thread to finish the cycle as it 2378 // periodically yields while clearing the next marking bitmap 2379 // and, if it's in a yield point, it's waiting for us to 2380 // finish. So, at this point we will not start a cycle and we'll 2381 // let the concurrent marking thread complete the last one. 2382 } 2383 } 2384 } 2385 2386 void 2387 G1CollectorPolicy_BestRegionsFirst:: 2388 record_collection_pause_start(double start_time_sec, size_t start_used) { 2389 G1CollectorPolicy::record_collection_pause_start(start_time_sec, start_used); 2390 } 2391 2392 class KnownGarbageClosure: public HeapRegionClosure { 2393 CollectionSetChooser* _hrSorted; 2394 2395 public: 2396 KnownGarbageClosure(CollectionSetChooser* hrSorted) : 2397 _hrSorted(hrSorted) 2398 {} 2399 2400 bool doHeapRegion(HeapRegion* r) { 2401 // We only include humongous regions in collection 2402 // sets when concurrent mark shows that their contained object is 2403 // unreachable. 2404 2405 // Do we have any marking information for this region? 2406 if (r->is_marked()) { 2407 // We don't include humongous regions in collection 2408 // sets because we collect them immediately at the end of a marking 2409 // cycle. We also don't include young regions because we *must* 2410 // include them in the next collection pause. 2411 if (!r->isHumongous() && !r->is_young()) { 2412 _hrSorted->addMarkedHeapRegion(r); 2413 } 2414 } 2415 return false; 2416 } 2417 }; 2418 2419 class ParKnownGarbageHRClosure: public HeapRegionClosure { 2420 CollectionSetChooser* _hrSorted; 2421 jint _marked_regions_added; 2422 jint _chunk_size; 2423 jint _cur_chunk_idx; 2424 jint _cur_chunk_end; // Cur chunk [_cur_chunk_idx, _cur_chunk_end) 2425 int _worker; 2426 int _invokes; 2427 2428 void get_new_chunk() { 2429 _cur_chunk_idx = _hrSorted->getParMarkedHeapRegionChunk(_chunk_size); 2430 _cur_chunk_end = _cur_chunk_idx + _chunk_size; 2431 } 2432 void add_region(HeapRegion* r) { 2433 if (_cur_chunk_idx == _cur_chunk_end) { 2434 get_new_chunk(); 2435 } 2436 assert(_cur_chunk_idx < _cur_chunk_end, "postcondition"); 2437 _hrSorted->setMarkedHeapRegion(_cur_chunk_idx, r); 2438 _marked_regions_added++; 2439 _cur_chunk_idx++; 2440 } 2441 2442 public: 2443 ParKnownGarbageHRClosure(CollectionSetChooser* hrSorted, 2444 jint chunk_size, 2445 int worker) : 2446 _hrSorted(hrSorted), _chunk_size(chunk_size), _worker(worker), 2447 _marked_regions_added(0), _cur_chunk_idx(0), _cur_chunk_end(0), 2448 _invokes(0) 2449 {} 2450 2451 bool doHeapRegion(HeapRegion* r) { 2452 // We only include humongous regions in collection 2453 // sets when concurrent mark shows that their contained object is 2454 // unreachable. 2455 _invokes++; 2456 2457 // Do we have any marking information for this region? 2458 if (r->is_marked()) { 2459 // We don't include humongous regions in collection 2460 // sets because we collect them immediately at the end of a marking 2461 // cycle. 2462 // We also do not include young regions in collection sets 2463 if (!r->isHumongous() && !r->is_young()) { 2464 add_region(r); 2465 } 2466 } 2467 return false; 2468 } 2469 jint marked_regions_added() { return _marked_regions_added; } 2470 int invokes() { return _invokes; } 2471 }; 2472 2473 class ParKnownGarbageTask: public AbstractGangTask { 2474 CollectionSetChooser* _hrSorted; 2475 jint _chunk_size; 2476 G1CollectedHeap* _g1; 2477 public: 2478 ParKnownGarbageTask(CollectionSetChooser* hrSorted, jint chunk_size) : 2479 AbstractGangTask("ParKnownGarbageTask"), 2480 _hrSorted(hrSorted), _chunk_size(chunk_size), 2481 _g1(G1CollectedHeap::heap()) 2482 {} 2483 2484 void work(int i) { 2485 ParKnownGarbageHRClosure parKnownGarbageCl(_hrSorted, _chunk_size, i); 2486 // Back to zero for the claim value. 2487 _g1->heap_region_par_iterate_chunked(&parKnownGarbageCl, i, 2488 HeapRegion::InitialClaimValue); 2489 jint regions_added = parKnownGarbageCl.marked_regions_added(); 2490 _hrSorted->incNumMarkedHeapRegions(regions_added); 2491 if (G1PrintParCleanupStats) { 2492 gclog_or_tty->print_cr(" Thread %d called %d times, added %d regions to list.", 2493 i, parKnownGarbageCl.invokes(), regions_added); 2494 } 2495 } 2496 }; 2497 2498 void 2499 G1CollectorPolicy_BestRegionsFirst:: 2500 record_concurrent_mark_cleanup_end(size_t freed_bytes, 2501 size_t max_live_bytes) { 2502 double start; 2503 if (G1PrintParCleanupStats) start = os::elapsedTime(); 2504 record_concurrent_mark_cleanup_end_work1(freed_bytes, max_live_bytes); 2505 2506 _collectionSetChooser->clearMarkedHeapRegions(); 2507 double clear_marked_end; 2508 if (G1PrintParCleanupStats) { 2509 clear_marked_end = os::elapsedTime(); 2510 gclog_or_tty->print_cr(" clear marked regions + work1: %8.3f ms.", 2511 (clear_marked_end - start)*1000.0); 2512 } 2513 if (G1CollectedHeap::use_parallel_gc_threads()) { 2514 const size_t OverpartitionFactor = 4; 2515 const size_t MinWorkUnit = 8; 2516 const size_t WorkUnit = 2517 MAX2(_g1->n_regions() / (ParallelGCThreads * OverpartitionFactor), 2518 MinWorkUnit); 2519 _collectionSetChooser->prepareForAddMarkedHeapRegionsPar(_g1->n_regions(), 2520 WorkUnit); 2521 ParKnownGarbageTask parKnownGarbageTask(_collectionSetChooser, 2522 (int) WorkUnit); 2523 _g1->workers()->run_task(&parKnownGarbageTask); 2524 2525 assert(_g1->check_heap_region_claim_values(HeapRegion::InitialClaimValue), 2526 "sanity check"); 2527 } else { 2528 KnownGarbageClosure knownGarbagecl(_collectionSetChooser); 2529 _g1->heap_region_iterate(&knownGarbagecl); 2530 } 2531 double known_garbage_end; 2532 if (G1PrintParCleanupStats) { 2533 known_garbage_end = os::elapsedTime(); 2534 gclog_or_tty->print_cr(" compute known garbage: %8.3f ms.", 2535 (known_garbage_end - clear_marked_end)*1000.0); 2536 } 2537 _collectionSetChooser->sortMarkedHeapRegions(); 2538 double sort_end; 2539 if (G1PrintParCleanupStats) { 2540 sort_end = os::elapsedTime(); 2541 gclog_or_tty->print_cr(" sorting: %8.3f ms.", 2542 (sort_end - known_garbage_end)*1000.0); 2543 } 2544 2545 record_concurrent_mark_cleanup_end_work2(); 2546 double work2_end; 2547 if (G1PrintParCleanupStats) { 2548 work2_end = os::elapsedTime(); 2549 gclog_or_tty->print_cr(" work2: %8.3f ms.", 2550 (work2_end - sort_end)*1000.0); 2551 } 2552 } 2553 2554 // Add the heap region at the head of the non-incremental collection set 2555 void G1CollectorPolicy:: 2556 add_to_collection_set(HeapRegion* hr) { 2557 assert(_inc_cset_build_state == Active, "Precondition"); 2558 assert(!hr->is_young(), "non-incremental add of young region"); 2559 2560 if (_g1->mark_in_progress()) 2561 _g1->concurrent_mark()->registerCSetRegion(hr); 2562 2563 assert(!hr->in_collection_set(), "should not already be in the CSet"); 2564 hr->set_in_collection_set(true); 2565 hr->set_next_in_collection_set(_collection_set); 2566 _collection_set = hr; 2567 _collection_set_size++; 2568 _collection_set_bytes_used_before += hr->used(); 2569 _g1->register_region_with_in_cset_fast_test(hr); 2570 } 2571 2572 // Initialize the per-collection-set information 2573 void G1CollectorPolicy::start_incremental_cset_building() { 2574 assert(_inc_cset_build_state == Inactive, "Precondition"); 2575 2576 _inc_cset_head = NULL; 2577 _inc_cset_tail = NULL; 2578 _inc_cset_size = 0; 2579 _inc_cset_bytes_used_before = 0; 2580 2581 _inc_cset_young_index = 0; 2582 2583 _inc_cset_max_finger = 0; 2584 _inc_cset_recorded_young_bytes = 0; 2585 _inc_cset_recorded_rs_lengths = 0; 2586 _inc_cset_predicted_elapsed_time_ms = 0; 2587 _inc_cset_predicted_bytes_to_copy = 0; 2588 _inc_cset_build_state = Active; 2589 } 2590 2591 void G1CollectorPolicy::add_to_incremental_cset_info(HeapRegion* hr, size_t rs_length) { 2592 // This routine is used when: 2593 // * adding survivor regions to the incremental cset at the end of an 2594 // evacuation pause, 2595 // * adding the current allocation region to the incremental cset 2596 // when it is retired, and 2597 // * updating existing policy information for a region in the 2598 // incremental cset via young list RSet sampling. 2599 // Therefore this routine may be called at a safepoint by the 2600 // VM thread, or in-between safepoints by mutator threads (when 2601 // retiring the current allocation region) or a concurrent 2602 // refine thread (RSet sampling). 2603 2604 double region_elapsed_time_ms = predict_region_elapsed_time_ms(hr, true); 2605 size_t used_bytes = hr->used(); 2606 2607 _inc_cset_recorded_rs_lengths += rs_length; 2608 _inc_cset_predicted_elapsed_time_ms += region_elapsed_time_ms; 2609 2610 _inc_cset_bytes_used_before += used_bytes; 2611 2612 // Cache the values we have added to the aggregated informtion 2613 // in the heap region in case we have to remove this region from 2614 // the incremental collection set, or it is updated by the 2615 // rset sampling code 2616 hr->set_recorded_rs_length(rs_length); 2617 hr->set_predicted_elapsed_time_ms(region_elapsed_time_ms); 2618 2619 #if PREDICTIONS_VERBOSE 2620 size_t bytes_to_copy = predict_bytes_to_copy(hr); 2621 _inc_cset_predicted_bytes_to_copy += bytes_to_copy; 2622 2623 // Record the number of bytes used in this region 2624 _inc_cset_recorded_young_bytes += used_bytes; 2625 2626 // Cache the values we have added to the aggregated informtion 2627 // in the heap region in case we have to remove this region from 2628 // the incremental collection set, or it is updated by the 2629 // rset sampling code 2630 hr->set_predicted_bytes_to_copy(bytes_to_copy); 2631 #endif // PREDICTIONS_VERBOSE 2632 } 2633 2634 void G1CollectorPolicy::remove_from_incremental_cset_info(HeapRegion* hr) { 2635 // This routine is currently only called as part of the updating of 2636 // existing policy information for regions in the incremental cset that 2637 // is performed by the concurrent refine thread(s) as part of young list 2638 // RSet sampling. Therefore we should not be at a safepoint. 2639 2640 assert(!SafepointSynchronize::is_at_safepoint(), "should not be at safepoint"); 2641 assert(hr->is_young(), "it should be"); 2642 2643 size_t used_bytes = hr->used(); 2644 size_t old_rs_length = hr->recorded_rs_length(); 2645 double old_elapsed_time_ms = hr->predicted_elapsed_time_ms(); 2646 2647 // Subtract the old recorded/predicted policy information for 2648 // the given heap region from the collection set info. 2649 _inc_cset_recorded_rs_lengths -= old_rs_length; 2650 _inc_cset_predicted_elapsed_time_ms -= old_elapsed_time_ms; 2651 2652 _inc_cset_bytes_used_before -= used_bytes; 2653 2654 // Clear the values cached in the heap region 2655 hr->set_recorded_rs_length(0); 2656 hr->set_predicted_elapsed_time_ms(0); 2657 2658 #if PREDICTIONS_VERBOSE 2659 size_t old_predicted_bytes_to_copy = hr->predicted_bytes_to_copy(); 2660 _inc_cset_predicted_bytes_to_copy -= old_predicted_bytes_to_copy; 2661 2662 // Subtract the number of bytes used in this region 2663 _inc_cset_recorded_young_bytes -= used_bytes; 2664 2665 // Clear the values cached in the heap region 2666 hr->set_predicted_bytes_to_copy(0); 2667 #endif // PREDICTIONS_VERBOSE 2668 } 2669 2670 void G1CollectorPolicy::update_incremental_cset_info(HeapRegion* hr, size_t new_rs_length) { 2671 // Update the collection set information that is dependent on the new RS length 2672 assert(hr->is_young(), "Precondition"); 2673 2674 remove_from_incremental_cset_info(hr); 2675 add_to_incremental_cset_info(hr, new_rs_length); 2676 } 2677 2678 void G1CollectorPolicy::add_region_to_incremental_cset_common(HeapRegion* hr) { 2679 assert( hr->is_young(), "invariant"); 2680 assert( hr->young_index_in_cset() == -1, "invariant" ); 2681 assert(_inc_cset_build_state == Active, "Precondition"); 2682 2683 // We need to clear and set the cached recorded/cached collection set 2684 // information in the heap region here (before the region gets added 2685 // to the collection set). An individual heap region's cached values 2686 // are calculated, aggregated with the policy collection set info, 2687 // and cached in the heap region here (initially) and (subsequently) 2688 // by the Young List sampling code. 2689 2690 size_t rs_length = hr->rem_set()->occupied(); 2691 add_to_incremental_cset_info(hr, rs_length); 2692 2693 HeapWord* hr_end = hr->end(); 2694 _inc_cset_max_finger = MAX2(_inc_cset_max_finger, hr_end); 2695 2696 assert(!hr->in_collection_set(), "invariant"); 2697 hr->set_in_collection_set(true); 2698 assert( hr->next_in_collection_set() == NULL, "invariant"); 2699 2700 _inc_cset_size++; 2701 _g1->register_region_with_in_cset_fast_test(hr); 2702 2703 hr->set_young_index_in_cset((int) _inc_cset_young_index); 2704 ++_inc_cset_young_index; 2705 } 2706 2707 // Add the region at the RHS of the incremental cset 2708 void G1CollectorPolicy::add_region_to_incremental_cset_rhs(HeapRegion* hr) { 2709 // We should only ever be appending survivors at the end of a pause 2710 assert( hr->is_survivor(), "Logic"); 2711 2712 // Do the 'common' stuff 2713 add_region_to_incremental_cset_common(hr); 2714 2715 // Now add the region at the right hand side 2716 if (_inc_cset_tail == NULL) { 2717 assert(_inc_cset_head == NULL, "invariant"); 2718 _inc_cset_head = hr; 2719 } else { 2720 _inc_cset_tail->set_next_in_collection_set(hr); 2721 } 2722 _inc_cset_tail = hr; 2723 } 2724 2725 // Add the region to the LHS of the incremental cset 2726 void G1CollectorPolicy::add_region_to_incremental_cset_lhs(HeapRegion* hr) { 2727 // Survivors should be added to the RHS at the end of a pause 2728 assert(!hr->is_survivor(), "Logic"); 2729 2730 // Do the 'common' stuff 2731 add_region_to_incremental_cset_common(hr); 2732 2733 // Add the region at the left hand side 2734 hr->set_next_in_collection_set(_inc_cset_head); 2735 if (_inc_cset_head == NULL) { 2736 assert(_inc_cset_tail == NULL, "Invariant"); 2737 _inc_cset_tail = hr; 2738 } 2739 _inc_cset_head = hr; 2740 } 2741 2742 #ifndef PRODUCT 2743 void G1CollectorPolicy::print_collection_set(HeapRegion* list_head, outputStream* st) { 2744 assert(list_head == inc_cset_head() || list_head == collection_set(), "must be"); 2745 2746 st->print_cr("\nCollection_set:"); 2747 HeapRegion* csr = list_head; 2748 while (csr != NULL) { 2749 HeapRegion* next = csr->next_in_collection_set(); 2750 assert(csr->in_collection_set(), "bad CS"); 2751 st->print_cr(" [%08x-%08x], t: %08x, P: %08x, N: %08x, C: %08x, " 2752 "age: %4d, y: %d, surv: %d", 2753 csr->bottom(), csr->end(), 2754 csr->top(), 2755 csr->prev_top_at_mark_start(), 2756 csr->next_top_at_mark_start(), 2757 csr->top_at_conc_mark_count(), 2758 csr->age_in_surv_rate_group_cond(), 2759 csr->is_young(), 2760 csr->is_survivor()); 2761 csr = next; 2762 } 2763 } 2764 #endif // !PRODUCT 2765 2766 void 2767 G1CollectorPolicy_BestRegionsFirst::choose_collection_set( 2768 double target_pause_time_ms) { 2769 // Set this here - in case we're not doing young collections. 2770 double non_young_start_time_sec = os::elapsedTime(); 2771 2772 start_recording_regions(); 2773 2774 guarantee(target_pause_time_ms > 0.0, 2775 err_msg("target_pause_time_ms = %1.6lf should be positive", 2776 target_pause_time_ms)); 2777 guarantee(_collection_set == NULL, "Precondition"); 2778 2779 double base_time_ms = predict_base_elapsed_time_ms(_pending_cards); 2780 double predicted_pause_time_ms = base_time_ms; 2781 2782 double time_remaining_ms = target_pause_time_ms - base_time_ms; 2783 2784 // the 10% and 50% values are arbitrary... 2785 if (time_remaining_ms < 0.10 * target_pause_time_ms) { 2786 time_remaining_ms = 0.50 * target_pause_time_ms; 2787 _within_target = false; 2788 } else { 2789 _within_target = true; 2790 } 2791 2792 // We figure out the number of bytes available for future to-space. 2793 // For new regions without marking information, we must assume the 2794 // worst-case of complete survival. If we have marking information for a 2795 // region, we can bound the amount of live data. We can add a number of 2796 // such regions, as long as the sum of the live data bounds does not 2797 // exceed the available evacuation space. 2798 size_t max_live_bytes = _g1->free_regions() * HeapRegion::GrainBytes; 2799 2800 size_t expansion_bytes = 2801 _g1->expansion_regions() * HeapRegion::GrainBytes; 2802 2803 _collection_set_bytes_used_before = 0; 2804 _collection_set_size = 0; 2805 2806 // Adjust for expansion and slop. 2807 max_live_bytes = max_live_bytes + expansion_bytes; 2808 2809 HeapRegion* hr; 2810 double young_start_time_sec = os::elapsedTime(); 2811 2812 if (G1PolicyVerbose > 0) { 2813 gclog_or_tty->print_cr("Adding %d young regions to the CSet", 2814 _g1->young_list()->length()); 2815 } 2816 2817 _young_cset_length = 0; 2818 _last_young_gc_full = full_young_gcs() ? true : false; 2819 2820 if (_last_young_gc_full) 2821 ++_full_young_pause_num; 2822 else 2823 ++_partial_young_pause_num; 2824 2825 // The young list is laid with the survivor regions from the previous 2826 // pause are appended to the RHS of the young list, i.e. 2827 // [Newly Young Regions ++ Survivors from last pause]. 2828 2829 hr = _g1->young_list()->first_survivor_region(); 2830 while (hr != NULL) { 2831 assert(hr->is_survivor(), "badly formed young list"); 2832 hr->set_young(); 2833 hr = hr->get_next_young_region(); 2834 } 2835 2836 // Clear the fields that point to the survivor list - they are 2837 // all young now. 2838 _g1->young_list()->clear_survivors(); 2839 2840 if (_g1->mark_in_progress()) 2841 _g1->concurrent_mark()->register_collection_set_finger(_inc_cset_max_finger); 2842 2843 _young_cset_length = _inc_cset_young_index; 2844 _collection_set = _inc_cset_head; 2845 _collection_set_size = _inc_cset_size; 2846 _collection_set_bytes_used_before = _inc_cset_bytes_used_before; 2847 2848 // For young regions in the collection set, we assume the worst 2849 // case of complete survival 2850 max_live_bytes -= _inc_cset_size * HeapRegion::GrainBytes; 2851 2852 time_remaining_ms -= _inc_cset_predicted_elapsed_time_ms; 2853 predicted_pause_time_ms += _inc_cset_predicted_elapsed_time_ms; 2854 2855 // The number of recorded young regions is the incremental 2856 // collection set's current size 2857 set_recorded_young_regions(_inc_cset_size); 2858 set_recorded_rs_lengths(_inc_cset_recorded_rs_lengths); 2859 set_recorded_young_bytes(_inc_cset_recorded_young_bytes); 2860 #if PREDICTIONS_VERBOSE 2861 set_predicted_bytes_to_copy(_inc_cset_predicted_bytes_to_copy); 2862 #endif // PREDICTIONS_VERBOSE 2863 2864 if (G1PolicyVerbose > 0) { 2865 gclog_or_tty->print_cr(" Added " PTR_FORMAT " Young Regions to CS.", 2866 _inc_cset_size); 2867 gclog_or_tty->print_cr(" (" SIZE_FORMAT " KB left in heap.)", 2868 max_live_bytes/K); 2869 } 2870 2871 assert(_inc_cset_size == _g1->young_list()->length(), "Invariant"); 2872 2873 double young_end_time_sec = os::elapsedTime(); 2874 _recorded_young_cset_choice_time_ms = 2875 (young_end_time_sec - young_start_time_sec) * 1000.0; 2876 2877 // We are doing young collections so reset this. 2878 non_young_start_time_sec = young_end_time_sec; 2879 2880 // Note we can use either _collection_set_size or 2881 // _young_cset_length here 2882 if (_collection_set_size > 0 && _last_young_gc_full) { 2883 // don't bother adding more regions... 2884 goto choose_collection_set_end; 2885 } 2886 2887 if (!full_young_gcs()) { 2888 bool should_continue = true; 2889 NumberSeq seq; 2890 double avg_prediction = 100000000000000000.0; // something very large 2891 2892 do { 2893 hr = _collectionSetChooser->getNextMarkedRegion(time_remaining_ms, 2894 avg_prediction); 2895 if (hr != NULL) { 2896 double predicted_time_ms = predict_region_elapsed_time_ms(hr, false); 2897 time_remaining_ms -= predicted_time_ms; 2898 predicted_pause_time_ms += predicted_time_ms; 2899 add_to_collection_set(hr); 2900 record_non_young_cset_region(hr); 2901 max_live_bytes -= MIN2(hr->max_live_bytes(), max_live_bytes); 2902 if (G1PolicyVerbose > 0) { 2903 gclog_or_tty->print_cr(" (" SIZE_FORMAT " KB left in heap.)", 2904 max_live_bytes/K); 2905 } 2906 seq.add(predicted_time_ms); 2907 avg_prediction = seq.avg() + seq.sd(); 2908 } 2909 should_continue = 2910 ( hr != NULL) && 2911 ( (adaptive_young_list_length()) ? time_remaining_ms > 0.0 2912 : _collection_set_size < _young_list_fixed_length ); 2913 } while (should_continue); 2914 2915 if (!adaptive_young_list_length() && 2916 _collection_set_size < _young_list_fixed_length) 2917 _should_revert_to_full_young_gcs = true; 2918 } 2919 2920 choose_collection_set_end: 2921 stop_incremental_cset_building(); 2922 2923 count_CS_bytes_used(); 2924 2925 end_recording_regions(); 2926 2927 double non_young_end_time_sec = os::elapsedTime(); 2928 _recorded_non_young_cset_choice_time_ms = 2929 (non_young_end_time_sec - non_young_start_time_sec) * 1000.0; 2930 } 2931 2932 void G1CollectorPolicy_BestRegionsFirst::record_full_collection_end() { 2933 G1CollectorPolicy::record_full_collection_end(); 2934 _collectionSetChooser->updateAfterFullCollection(); 2935 } 2936 2937 void G1CollectorPolicy_BestRegionsFirst:: 2938 expand_if_possible(size_t numRegions) { 2939 size_t expansion_bytes = numRegions * HeapRegion::GrainBytes; 2940 _g1->expand(expansion_bytes); 2941 } 2942 2943 void G1CollectorPolicy_BestRegionsFirst:: 2944 record_collection_pause_end() { 2945 G1CollectorPolicy::record_collection_pause_end(); 2946 assert(assertMarkedBytesDataOK(), "Marked regions not OK at pause end."); 2947 }