1 /* 2 * Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved. 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * This code is free software; you can redistribute it and/or modify it 6 * under the terms of the GNU General Public License version 2 only, as 7 * published by the Free Software Foundation. 8 * 9 * This code is distributed in the hope that it will be useful, but WITHOUT 10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 12 * version 2 for more details (a copy is included in the LICENSE file that 13 * accompanied this code). 14 * 15 * You should have received a copy of the GNU General Public License version 16 * 2 along with this work; if not, write to the Free Software Foundation, 17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 18 * 19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 20 * or visit www.oracle.com if you need additional information or have any 21 * questions. 22 * 23 */ 24 25 #ifndef SHARE_VM_GC_IMPLEMENTATION_G1_CONCURRENTMARK_HPP 26 #define SHARE_VM_GC_IMPLEMENTATION_G1_CONCURRENTMARK_HPP 27 28 #include "classfile/javaClasses.hpp" 29 #include "gc_implementation/g1/heapRegionSet.hpp" 30 #include "gc_implementation/g1/g1RegionToSpaceMapper.hpp" 31 #include "gc_implementation/shared/gcId.hpp" 32 #include "utilities/taskqueue.hpp" 33 34 class G1CollectedHeap; 35 class CMBitMap; 36 class CMTask; 37 class ConcurrentMark; 38 typedef GenericTaskQueue<oop, mtGC> CMTaskQueue; 39 typedef GenericTaskQueueSet<CMTaskQueue, mtGC> CMTaskQueueSet; 40 41 // Closure used by CM during concurrent reference discovery 42 // and reference processing (during remarking) to determine 43 // if a particular object is alive. It is primarily used 44 // to determine if referents of discovered reference objects 45 // are alive. An instance is also embedded into the 46 // reference processor as the _is_alive_non_header field 47 class G1CMIsAliveClosure: public BoolObjectClosure { 48 G1CollectedHeap* _g1; 49 public: 50 G1CMIsAliveClosure(G1CollectedHeap* g1) : _g1(g1) { } 51 52 bool do_object_b(oop obj); 53 }; 54 55 // A generic CM bit map. This is essentially a wrapper around the BitMap 56 // class, with one bit per (1<<_shifter) HeapWords. 57 58 class CMBitMapRO VALUE_OBJ_CLASS_SPEC { 59 protected: 60 HeapWord* _bmStartWord; // base address of range covered by map 61 size_t _bmWordSize; // map size (in #HeapWords covered) 62 const int _shifter; // map to char or bit 63 BitMap _bm; // the bit map itself 64 65 public: 66 // constructor 67 CMBitMapRO(int shifter); 68 69 enum { do_yield = true }; 70 71 // inquiries 72 HeapWord* startWord() const { return _bmStartWord; } 73 size_t sizeInWords() const { return _bmWordSize; } 74 // the following is one past the last word in space 75 HeapWord* endWord() const { return _bmStartWord + _bmWordSize; } 76 77 // read marks 78 79 bool isMarked(HeapWord* addr) const { 80 assert(_bmStartWord <= addr && addr < (_bmStartWord + _bmWordSize), 81 "outside underlying space?"); 82 return _bm.at(heapWordToOffset(addr)); 83 } 84 85 // iteration 86 inline bool iterate(BitMapClosure* cl, MemRegion mr); 87 inline bool iterate(BitMapClosure* cl); 88 89 // Return the address corresponding to the next marked bit at or after 90 // "addr", and before "limit", if "limit" is non-NULL. If there is no 91 // such bit, returns "limit" if that is non-NULL, or else "endWord()". 92 HeapWord* getNextMarkedWordAddress(const HeapWord* addr, 93 const HeapWord* limit = NULL) const; 94 // Return the address corresponding to the next unmarked bit at or after 95 // "addr", and before "limit", if "limit" is non-NULL. If there is no 96 // such bit, returns "limit" if that is non-NULL, or else "endWord()". 97 HeapWord* getNextUnmarkedWordAddress(const HeapWord* addr, 98 const HeapWord* limit = NULL) const; 99 100 // conversion utilities 101 HeapWord* offsetToHeapWord(size_t offset) const { 102 return _bmStartWord + (offset << _shifter); 103 } 104 size_t heapWordToOffset(const HeapWord* addr) const { 105 return pointer_delta(addr, _bmStartWord) >> _shifter; 106 } 107 int heapWordDiffToOffsetDiff(size_t diff) const; 108 109 // The argument addr should be the start address of a valid object 110 HeapWord* nextObject(HeapWord* addr) { 111 oop obj = (oop) addr; 112 HeapWord* res = addr + obj->size(); 113 assert(offsetToHeapWord(heapWordToOffset(res)) == res, "sanity"); 114 return res; 115 } 116 117 void print_on_error(outputStream* st, const char* prefix) const; 118 119 // debugging 120 NOT_PRODUCT(bool covers(MemRegion rs) const;) 121 }; 122 123 class CMBitMapMappingChangedListener : public G1MappingChangedListener { 124 private: 125 CMBitMap* _bm; 126 public: 127 CMBitMapMappingChangedListener() : _bm(NULL) {} 128 129 void set_bitmap(CMBitMap* bm) { _bm = bm; } 130 131 virtual void on_commit(uint start_idx, size_t num_regions, bool zero_filled); 132 }; 133 134 class CMBitMap : public CMBitMapRO { 135 private: 136 CMBitMapMappingChangedListener _listener; 137 138 public: 139 static size_t compute_size(size_t heap_size); 140 // Returns the amount of bytes on the heap between two marks in the bitmap. 141 static size_t mark_distance(); 142 143 CMBitMap() : CMBitMapRO(LogMinObjAlignment), _listener() { _listener.set_bitmap(this); } 144 145 // Initializes the underlying BitMap to cover the given area. 146 void initialize(MemRegion heap, G1RegionToSpaceMapper* storage); 147 148 // Write marks. 149 inline void mark(HeapWord* addr); 150 inline void clear(HeapWord* addr); 151 inline bool parMark(HeapWord* addr); 152 inline bool parClear(HeapWord* addr); 153 154 void markRange(MemRegion mr); 155 void clearRange(MemRegion mr); 156 157 // Starting at the bit corresponding to "addr" (inclusive), find the next 158 // "1" bit, if any. This bit starts some run of consecutive "1"'s; find 159 // the end of this run (stopping at "end_addr"). Return the MemRegion 160 // covering from the start of the region corresponding to the first bit 161 // of the run to the end of the region corresponding to the last bit of 162 // the run. If there is no "1" bit at or after "addr", return an empty 163 // MemRegion. 164 MemRegion getAndClearMarkedRegion(HeapWord* addr, HeapWord* end_addr); 165 166 // Clear the whole mark bitmap. 167 void clearAll(); 168 }; 169 170 // Represents a marking stack used by ConcurrentMarking in the G1 collector. 171 class CMMarkStack VALUE_OBJ_CLASS_SPEC { 172 VirtualSpace _virtual_space; // Underlying backing store for actual stack 173 ConcurrentMark* _cm; 174 oop* _base; // bottom of stack 175 jint _index; // one more than last occupied index 176 jint _capacity; // max #elements 177 jint _saved_index; // value of _index saved at start of GC 178 NOT_PRODUCT(jint _max_depth;) // max depth plumbed during run 179 180 bool _overflow; 181 bool _should_expand; 182 DEBUG_ONLY(bool _drain_in_progress;) 183 DEBUG_ONLY(bool _drain_in_progress_yields;) 184 185 public: 186 CMMarkStack(ConcurrentMark* cm); 187 ~CMMarkStack(); 188 189 #ifndef PRODUCT 190 jint max_depth() const { 191 return _max_depth; 192 } 193 #endif 194 195 bool allocate(size_t capacity); 196 197 oop pop() { 198 if (!isEmpty()) { 199 return _base[--_index] ; 200 } 201 return NULL; 202 } 203 204 // If overflow happens, don't do the push, and record the overflow. 205 // *Requires* that "ptr" is already marked. 206 void push(oop ptr) { 207 if (isFull()) { 208 // Record overflow. 209 _overflow = true; 210 return; 211 } else { 212 _base[_index++] = ptr; 213 NOT_PRODUCT(_max_depth = MAX2(_max_depth, _index)); 214 } 215 } 216 // Non-block impl. Note: concurrency is allowed only with other 217 // "par_push" operations, not with "pop" or "drain". We would need 218 // parallel versions of them if such concurrency was desired. 219 void par_push(oop ptr); 220 221 // Pushes the first "n" elements of "ptr_arr" on the stack. 222 // Non-block impl. Note: concurrency is allowed only with other 223 // "par_adjoin_arr" or "push" operations, not with "pop" or "drain". 224 void par_adjoin_arr(oop* ptr_arr, int n); 225 226 // Pushes the first "n" elements of "ptr_arr" on the stack. 227 // Locking impl: concurrency is allowed only with 228 // "par_push_arr" and/or "par_pop_arr" operations, which use the same 229 // locking strategy. 230 void par_push_arr(oop* ptr_arr, int n); 231 232 // If returns false, the array was empty. Otherwise, removes up to "max" 233 // elements from the stack, and transfers them to "ptr_arr" in an 234 // unspecified order. The actual number transferred is given in "n" ("n 235 // == 0" is deliberately redundant with the return value.) Locking impl: 236 // concurrency is allowed only with "par_push_arr" and/or "par_pop_arr" 237 // operations, which use the same locking strategy. 238 bool par_pop_arr(oop* ptr_arr, int max, int* n); 239 240 // Drain the mark stack, applying the given closure to all fields of 241 // objects on the stack. (That is, continue until the stack is empty, 242 // even if closure applications add entries to the stack.) The "bm" 243 // argument, if non-null, may be used to verify that only marked objects 244 // are on the mark stack. If "yield_after" is "true", then the 245 // concurrent marker performing the drain offers to yield after 246 // processing each object. If a yield occurs, stops the drain operation 247 // and returns false. Otherwise, returns true. 248 template<class OopClosureClass> 249 bool drain(OopClosureClass* cl, CMBitMap* bm, bool yield_after = false); 250 251 bool isEmpty() { return _index == 0; } 252 bool isFull() { return _index == _capacity; } 253 int maxElems() { return _capacity; } 254 255 bool overflow() { return _overflow; } 256 void clear_overflow() { _overflow = false; } 257 258 bool should_expand() const { return _should_expand; } 259 void set_should_expand(); 260 261 // Expand the stack, typically in response to an overflow condition 262 void expand(); 263 264 int size() { return _index; } 265 266 void setEmpty() { _index = 0; clear_overflow(); } 267 268 // Record the current index. 269 void note_start_of_gc(); 270 271 // Make sure that we have not added any entries to the stack during GC. 272 void note_end_of_gc(); 273 274 // iterate over the oops in the mark stack, up to the bound recorded via 275 // the call above. 276 void oops_do(OopClosure* f); 277 }; 278 279 class ForceOverflowSettings VALUE_OBJ_CLASS_SPEC { 280 private: 281 #ifndef PRODUCT 282 uintx _num_remaining; 283 bool _force; 284 #endif // !defined(PRODUCT) 285 286 public: 287 void init() PRODUCT_RETURN; 288 void update() PRODUCT_RETURN; 289 bool should_force() PRODUCT_RETURN_( return false; ); 290 }; 291 292 // this will enable a variety of different statistics per GC task 293 #define _MARKING_STATS_ 0 294 // this will enable the higher verbose levels 295 #define _MARKING_VERBOSE_ 0 296 297 #if _MARKING_STATS_ 298 #define statsOnly(statement) \ 299 do { \ 300 statement ; \ 301 } while (0) 302 #else // _MARKING_STATS_ 303 #define statsOnly(statement) \ 304 do { \ 305 } while (0) 306 #endif // _MARKING_STATS_ 307 308 typedef enum { 309 no_verbose = 0, // verbose turned off 310 stats_verbose, // only prints stats at the end of marking 311 low_verbose, // low verbose, mostly per region and per major event 312 medium_verbose, // a bit more detailed than low 313 high_verbose // per object verbose 314 } CMVerboseLevel; 315 316 class YoungList; 317 318 // Root Regions are regions that are not empty at the beginning of a 319 // marking cycle and which we might collect during an evacuation pause 320 // while the cycle is active. Given that, during evacuation pauses, we 321 // do not copy objects that are explicitly marked, what we have to do 322 // for the root regions is to scan them and mark all objects reachable 323 // from them. According to the SATB assumptions, we only need to visit 324 // each object once during marking. So, as long as we finish this scan 325 // before the next evacuation pause, we can copy the objects from the 326 // root regions without having to mark them or do anything else to them. 327 // 328 // Currently, we only support root region scanning once (at the start 329 // of the marking cycle) and the root regions are all the survivor 330 // regions populated during the initial-mark pause. 331 class CMRootRegions VALUE_OBJ_CLASS_SPEC { 332 private: 333 YoungList* _young_list; 334 ConcurrentMark* _cm; 335 336 volatile bool _scan_in_progress; 337 volatile bool _should_abort; 338 HeapRegion* volatile _next_survivor; 339 340 public: 341 CMRootRegions(); 342 // We actually do most of the initialization in this method. 343 void init(G1CollectedHeap* g1h, ConcurrentMark* cm); 344 345 // Reset the claiming / scanning of the root regions. 346 void prepare_for_scan(); 347 348 // Forces get_next() to return NULL so that the iteration aborts early. 349 void abort() { _should_abort = true; } 350 351 // Return true if the CM thread are actively scanning root regions, 352 // false otherwise. 353 bool scan_in_progress() { return _scan_in_progress; } 354 355 // Claim the next root region to scan atomically, or return NULL if 356 // all have been claimed. 357 HeapRegion* claim_next(); 358 359 // Flag that we're done with root region scanning and notify anyone 360 // who's waiting on it. If aborted is false, assume that all regions 361 // have been claimed. 362 void scan_finished(); 363 364 // If CM threads are still scanning root regions, wait until they 365 // are done. Return true if we had to wait, false otherwise. 366 bool wait_until_scan_finished(); 367 }; 368 369 class ConcurrentMarkThread; 370 371 class ConcurrentMark: public CHeapObj<mtGC> { 372 friend class CMMarkStack; 373 friend class ConcurrentMarkThread; 374 friend class CMTask; 375 friend class CMBitMapClosure; 376 friend class CMGlobalObjectClosure; 377 friend class CMRemarkTask; 378 friend class CMConcurrentMarkingTask; 379 friend class G1ParNoteEndTask; 380 friend class CalcLiveObjectsClosure; 381 friend class G1CMRefProcTaskProxy; 382 friend class G1CMRefProcTaskExecutor; 383 friend class G1CMKeepAliveAndDrainClosure; 384 friend class G1CMDrainMarkingStackClosure; 385 386 protected: 387 ConcurrentMarkThread* _cmThread; // The thread doing the work 388 G1CollectedHeap* _g1h; // The heap 389 uint _parallel_marking_threads; // The number of marking 390 // threads we're using 391 uint _max_parallel_marking_threads; // Max number of marking 392 // threads we'll ever use 393 double _sleep_factor; // How much we have to sleep, with 394 // respect to the work we just did, to 395 // meet the marking overhead goal 396 double _marking_task_overhead; // Marking target overhead for 397 // a single task 398 399 // Same as the two above, but for the cleanup task 400 double _cleanup_sleep_factor; 401 double _cleanup_task_overhead; 402 403 FreeRegionList _cleanup_list; 404 405 // Concurrent marking support structures 406 CMBitMap _markBitMap1; 407 CMBitMap _markBitMap2; 408 CMBitMapRO* _prevMarkBitMap; // Completed mark bitmap 409 CMBitMap* _nextMarkBitMap; // Under-construction mark bitmap 410 411 BitMap _region_bm; 412 BitMap _card_bm; 413 414 // Heap bounds 415 HeapWord* _heap_start; 416 HeapWord* _heap_end; 417 418 // Root region tracking and claiming 419 CMRootRegions _root_regions; 420 421 // For gray objects 422 CMMarkStack _markStack; // Grey objects behind global finger 423 HeapWord* volatile _finger; // The global finger, region aligned, 424 // always points to the end of the 425 // last claimed region 426 427 // Marking tasks 428 uint _max_worker_id;// Maximum worker id 429 uint _active_tasks; // Task num currently active 430 CMTask** _tasks; // Task queue array (max_worker_id len) 431 CMTaskQueueSet* _task_queues; // Task queue set 432 ParallelTaskTerminator _terminator; // For termination 433 434 // Two sync barriers that are used to synchronize tasks when an 435 // overflow occurs. The algorithm is the following. All tasks enter 436 // the first one to ensure that they have all stopped manipulating 437 // the global data structures. After they exit it, they re-initialize 438 // their data structures and task 0 re-initializes the global data 439 // structures. Then, they enter the second sync barrier. This 440 // ensure, that no task starts doing work before all data 441 // structures (local and global) have been re-initialized. When they 442 // exit it, they are free to start working again. 443 WorkGangBarrierSync _first_overflow_barrier_sync; 444 WorkGangBarrierSync _second_overflow_barrier_sync; 445 446 // This is set by any task, when an overflow on the global data 447 // structures is detected 448 volatile bool _has_overflown; 449 // True: marking is concurrent, false: we're in remark 450 volatile bool _concurrent; 451 // Set at the end of a Full GC so that marking aborts 452 volatile bool _has_aborted; 453 GCId _aborted_gc_id; 454 455 // Used when remark aborts due to an overflow to indicate that 456 // another concurrent marking phase should start 457 volatile bool _restart_for_overflow; 458 459 // This is true from the very start of concurrent marking until the 460 // point when all the tasks complete their work. It is really used 461 // to determine the points between the end of concurrent marking and 462 // time of remark. 463 volatile bool _concurrent_marking_in_progress; 464 465 // Verbose level 466 CMVerboseLevel _verbose_level; 467 468 // All of these times are in ms 469 NumberSeq _init_times; 470 NumberSeq _remark_times; 471 NumberSeq _remark_mark_times; 472 NumberSeq _remark_weak_ref_times; 473 NumberSeq _cleanup_times; 474 double _total_counting_time; 475 double _total_rs_scrub_time; 476 477 double* _accum_task_vtime; // Accumulated task vtime 478 479 FlexibleWorkGang* _parallel_workers; 480 481 ForceOverflowSettings _force_overflow_conc; 482 ForceOverflowSettings _force_overflow_stw; 483 484 void weakRefsWorkParallelPart(BoolObjectClosure* is_alive, bool purged_classes); 485 void weakRefsWork(bool clear_all_soft_refs); 486 487 void swapMarkBitMaps(); 488 489 // It resets the global marking data structures, as well as the 490 // task local ones; should be called during initial mark. 491 void reset(); 492 493 // Resets all the marking data structures. Called when we have to restart 494 // marking or when marking completes (via set_non_marking_state below). 495 void reset_marking_state(bool clear_overflow = true); 496 497 // We do this after we're done with marking so that the marking data 498 // structures are initialized to a sensible and predictable state. 499 void set_non_marking_state(); 500 501 // Called to indicate how many threads are currently active. 502 void set_concurrency(uint active_tasks); 503 504 // It should be called to indicate which phase we're in (concurrent 505 // mark or remark) and how many threads are currently active. 506 void set_concurrency_and_phase(uint active_tasks, bool concurrent); 507 508 // Prints all gathered CM-related statistics 509 void print_stats(); 510 511 bool cleanup_list_is_empty() { 512 return _cleanup_list.is_empty(); 513 } 514 515 // Accessor methods 516 uint parallel_marking_threads() const { return _parallel_marking_threads; } 517 uint max_parallel_marking_threads() const { return _max_parallel_marking_threads;} 518 double sleep_factor() { return _sleep_factor; } 519 double marking_task_overhead() { return _marking_task_overhead;} 520 double cleanup_sleep_factor() { return _cleanup_sleep_factor; } 521 double cleanup_task_overhead() { return _cleanup_task_overhead;} 522 523 HeapWord* finger() { return _finger; } 524 bool concurrent() { return _concurrent; } 525 uint active_tasks() { return _active_tasks; } 526 ParallelTaskTerminator* terminator() { return &_terminator; } 527 528 // It claims the next available region to be scanned by a marking 529 // task/thread. It might return NULL if the next region is empty or 530 // we have run out of regions. In the latter case, out_of_regions() 531 // determines whether we've really run out of regions or the task 532 // should call claim_region() again. This might seem a bit 533 // awkward. Originally, the code was written so that claim_region() 534 // either successfully returned with a non-empty region or there 535 // were no more regions to be claimed. The problem with this was 536 // that, in certain circumstances, it iterated over large chunks of 537 // the heap finding only empty regions and, while it was working, it 538 // was preventing the calling task to call its regular clock 539 // method. So, this way, each task will spend very little time in 540 // claim_region() and is allowed to call the regular clock method 541 // frequently. 542 HeapRegion* claim_region(uint worker_id); 543 544 // It determines whether we've run out of regions to scan. Note that 545 // the finger can point past the heap end in case the heap was expanded 546 // to satisfy an allocation without doing a GC. This is fine, because all 547 // objects in those regions will be considered live anyway because of 548 // SATB guarantees (i.e. their TAMS will be equal to bottom). 549 bool out_of_regions() { return _finger >= _heap_end; } 550 551 // Returns the task with the given id 552 CMTask* task(int id) { 553 assert(0 <= id && id < (int) _active_tasks, 554 "task id not within active bounds"); 555 return _tasks[id]; 556 } 557 558 // Returns the task queue with the given id 559 CMTaskQueue* task_queue(int id) { 560 assert(0 <= id && id < (int) _active_tasks, 561 "task queue id not within active bounds"); 562 return (CMTaskQueue*) _task_queues->queue(id); 563 } 564 565 // Returns the task queue set 566 CMTaskQueueSet* task_queues() { return _task_queues; } 567 568 // Access / manipulation of the overflow flag which is set to 569 // indicate that the global stack has overflown 570 bool has_overflown() { return _has_overflown; } 571 void set_has_overflown() { _has_overflown = true; } 572 void clear_has_overflown() { _has_overflown = false; } 573 bool restart_for_overflow() { return _restart_for_overflow; } 574 575 // Methods to enter the two overflow sync barriers 576 void enter_first_sync_barrier(uint worker_id); 577 void enter_second_sync_barrier(uint worker_id); 578 579 ForceOverflowSettings* force_overflow_conc() { 580 return &_force_overflow_conc; 581 } 582 583 ForceOverflowSettings* force_overflow_stw() { 584 return &_force_overflow_stw; 585 } 586 587 ForceOverflowSettings* force_overflow() { 588 if (concurrent()) { 589 return force_overflow_conc(); 590 } else { 591 return force_overflow_stw(); 592 } 593 } 594 595 // Live Data Counting data structures... 596 // These data structures are initialized at the start of 597 // marking. They are written to while marking is active. 598 // They are aggregated during remark; the aggregated values 599 // are then used to populate the _region_bm, _card_bm, and 600 // the total live bytes, which are then subsequently updated 601 // during cleanup. 602 603 // An array of bitmaps (one bit map per task). Each bitmap 604 // is used to record the cards spanned by the live objects 605 // marked by that task/worker. 606 BitMap* _count_card_bitmaps; 607 608 // Used to record the number of marked live bytes 609 // (for each region, by worker thread). 610 size_t** _count_marked_bytes; 611 612 // Card index of the bottom of the G1 heap. Used for biasing indices into 613 // the card bitmaps. 614 intptr_t _heap_bottom_card_num; 615 616 // Set to true when initialization is complete 617 bool _completed_initialization; 618 619 public: 620 // Manipulation of the global mark stack. 621 // Notice that the first mark_stack_push is CAS-based, whereas the 622 // two below are Mutex-based. This is OK since the first one is only 623 // called during evacuation pauses and doesn't compete with the 624 // other two (which are called by the marking tasks during 625 // concurrent marking or remark). 626 bool mark_stack_push(oop p) { 627 _markStack.par_push(p); 628 if (_markStack.overflow()) { 629 set_has_overflown(); 630 return false; 631 } 632 return true; 633 } 634 bool mark_stack_push(oop* arr, int n) { 635 _markStack.par_push_arr(arr, n); 636 if (_markStack.overflow()) { 637 set_has_overflown(); 638 return false; 639 } 640 return true; 641 } 642 void mark_stack_pop(oop* arr, int max, int* n) { 643 _markStack.par_pop_arr(arr, max, n); 644 } 645 size_t mark_stack_size() { return _markStack.size(); } 646 size_t partial_mark_stack_size_target() { return _markStack.maxElems()/3; } 647 bool mark_stack_overflow() { return _markStack.overflow(); } 648 bool mark_stack_empty() { return _markStack.isEmpty(); } 649 650 CMRootRegions* root_regions() { return &_root_regions; } 651 652 bool concurrent_marking_in_progress() { 653 return _concurrent_marking_in_progress; 654 } 655 void set_concurrent_marking_in_progress() { 656 _concurrent_marking_in_progress = true; 657 } 658 void clear_concurrent_marking_in_progress() { 659 _concurrent_marking_in_progress = false; 660 } 661 662 void update_accum_task_vtime(int i, double vtime) { 663 _accum_task_vtime[i] += vtime; 664 } 665 666 double all_task_accum_vtime() { 667 double ret = 0.0; 668 for (uint i = 0; i < _max_worker_id; ++i) 669 ret += _accum_task_vtime[i]; 670 return ret; 671 } 672 673 // Attempts to steal an object from the task queues of other tasks 674 bool try_stealing(uint worker_id, int* hash_seed, oop& obj) { 675 return _task_queues->steal(worker_id, hash_seed, obj); 676 } 677 678 ConcurrentMark(G1CollectedHeap* g1h, 679 G1RegionToSpaceMapper* prev_bitmap_storage, 680 G1RegionToSpaceMapper* next_bitmap_storage); 681 ~ConcurrentMark(); 682 683 ConcurrentMarkThread* cmThread() { return _cmThread; } 684 685 CMBitMapRO* prevMarkBitMap() const { return _prevMarkBitMap; } 686 CMBitMap* nextMarkBitMap() const { return _nextMarkBitMap; } 687 688 // Returns the number of GC threads to be used in a concurrent 689 // phase based on the number of GC threads being used in a STW 690 // phase. 691 uint scale_parallel_threads(uint n_par_threads); 692 693 // Calculates the number of GC threads to be used in a concurrent phase. 694 uint calc_parallel_marking_threads(); 695 696 // The following three are interaction between CM and 697 // G1CollectedHeap 698 699 // This notifies CM that a root during initial-mark needs to be 700 // grayed. It is MT-safe. word_size is the size of the object in 701 // words. It is passed explicitly as sometimes we cannot calculate 702 // it from the given object because it might be in an inconsistent 703 // state (e.g., in to-space and being copied). So the caller is 704 // responsible for dealing with this issue (e.g., get the size from 705 // the from-space image when the to-space image might be 706 // inconsistent) and always passing the size. hr is the region that 707 // contains the object and it's passed optionally from callers who 708 // might already have it (no point in recalculating it). 709 inline void grayRoot(oop obj, 710 size_t word_size, 711 uint worker_id, 712 HeapRegion* hr = NULL); 713 714 // It iterates over the heap and for each object it comes across it 715 // will dump the contents of its reference fields, as well as 716 // liveness information for the object and its referents. The dump 717 // will be written to a file with the following name: 718 // G1PrintReachableBaseFile + "." + str. 719 // vo decides whether the prev (vo == UsePrevMarking), the next 720 // (vo == UseNextMarking) marking information, or the mark word 721 // (vo == UseMarkWord) will be used to determine the liveness of 722 // each object / referent. 723 // If all is true, all objects in the heap will be dumped, otherwise 724 // only the live ones. In the dump the following symbols / breviations 725 // are used: 726 // M : an explicitly live object (its bitmap bit is set) 727 // > : an implicitly live object (over tams) 728 // O : an object outside the G1 heap (typically: in the perm gen) 729 // NOT : a reference field whose referent is not live 730 // AND MARKED : indicates that an object is both explicitly and 731 // implicitly live (it should be one or the other, not both) 732 void print_reachable(const char* str, 733 VerifyOption vo, 734 bool all) PRODUCT_RETURN; 735 736 // Clear the next marking bitmap (will be called concurrently). 737 void clearNextBitmap(); 738 739 // Return whether the next mark bitmap has no marks set. To be used for assertions 740 // only. Will not yield to pause requests. 741 bool nextMarkBitmapIsClear(); 742 743 // These two do the work that needs to be done before and after the 744 // initial root checkpoint. Since this checkpoint can be done at two 745 // different points (i.e. an explicit pause or piggy-backed on a 746 // young collection), then it's nice to be able to easily share the 747 // pre/post code. It might be the case that we can put everything in 748 // the post method. TP 749 void checkpointRootsInitialPre(); 750 void checkpointRootsInitialPost(); 751 752 // Scan all the root regions and mark everything reachable from 753 // them. 754 void scanRootRegions(); 755 756 // Scan a single root region and mark everything reachable from it. 757 void scanRootRegion(HeapRegion* hr, uint worker_id); 758 759 // Do concurrent phase of marking, to a tentative transitive closure. 760 void markFromRoots(); 761 762 void checkpointRootsFinal(bool clear_all_soft_refs); 763 void checkpointRootsFinalWork(); 764 void cleanup(); 765 void completeCleanup(); 766 767 // Mark in the previous bitmap. NB: this is usually read-only, so use 768 // this carefully! 769 inline void markPrev(oop p); 770 771 // Clears marks for all objects in the given range, for the prev or 772 // next bitmaps. NB: the previous bitmap is usually 773 // read-only, so use this carefully! 774 void clearRangePrevBitmap(MemRegion mr); 775 void clearRangeNextBitmap(MemRegion mr); 776 777 // Notify data structures that a GC has started. 778 void note_start_of_gc() { 779 _markStack.note_start_of_gc(); 780 } 781 782 // Notify data structures that a GC is finished. 783 void note_end_of_gc() { 784 _markStack.note_end_of_gc(); 785 } 786 787 // Verify that there are no CSet oops on the stacks (taskqueues / 788 // global mark stack) and fingers (global / per-task). The boolean 789 // parameters decide which of the above data structures to verify. 790 // If marking is not in progress, it's a no-op. 791 void verify_no_cset_oops(bool verify_stacks, 792 bool verify_fingers) PRODUCT_RETURN; 793 794 bool isPrevMarked(oop p) const { 795 assert(p != NULL && p->is_oop(), "expected an oop"); 796 HeapWord* addr = (HeapWord*)p; 797 assert(addr >= _prevMarkBitMap->startWord() || 798 addr < _prevMarkBitMap->endWord(), "in a region"); 799 800 return _prevMarkBitMap->isMarked(addr); 801 } 802 803 inline bool do_yield_check(uint worker_i = 0); 804 805 // Called to abort the marking cycle after a Full GC takes place. 806 void abort(); 807 808 bool has_aborted() { return _has_aborted; } 809 810 const GCId& concurrent_gc_id(); 811 812 // This prints the global/local fingers. It is used for debugging. 813 NOT_PRODUCT(void print_finger();) 814 815 void print_summary_info(); 816 817 void print_worker_threads_on(outputStream* st) const; 818 819 void print_on_error(outputStream* st) const; 820 821 // The following indicate whether a given verbose level has been 822 // set. Notice that anything above stats is conditional to 823 // _MARKING_VERBOSE_ having been set to 1 824 bool verbose_stats() { 825 return _verbose_level >= stats_verbose; 826 } 827 bool verbose_low() { 828 return _MARKING_VERBOSE_ && _verbose_level >= low_verbose; 829 } 830 bool verbose_medium() { 831 return _MARKING_VERBOSE_ && _verbose_level >= medium_verbose; 832 } 833 bool verbose_high() { 834 return _MARKING_VERBOSE_ && _verbose_level >= high_verbose; 835 } 836 837 // Liveness counting 838 839 // Utility routine to set an exclusive range of cards on the given 840 // card liveness bitmap 841 inline void set_card_bitmap_range(BitMap* card_bm, 842 BitMap::idx_t start_idx, 843 BitMap::idx_t end_idx, 844 bool is_par); 845 846 // Returns the card number of the bottom of the G1 heap. 847 // Used in biasing indices into accounting card bitmaps. 848 intptr_t heap_bottom_card_num() const { 849 return _heap_bottom_card_num; 850 } 851 852 // Returns the card bitmap for a given task or worker id. 853 BitMap* count_card_bitmap_for(uint worker_id) { 854 assert(worker_id < _max_worker_id, "oob"); 855 assert(_count_card_bitmaps != NULL, "uninitialized"); 856 BitMap* task_card_bm = &_count_card_bitmaps[worker_id]; 857 assert(task_card_bm->size() == _card_bm.size(), "size mismatch"); 858 return task_card_bm; 859 } 860 861 // Returns the array containing the marked bytes for each region, 862 // for the given worker or task id. 863 size_t* count_marked_bytes_array_for(uint worker_id) { 864 assert(worker_id < _max_worker_id, "oob"); 865 assert(_count_marked_bytes != NULL, "uninitialized"); 866 size_t* marked_bytes_array = _count_marked_bytes[worker_id]; 867 assert(marked_bytes_array != NULL, "uninitialized"); 868 return marked_bytes_array; 869 } 870 871 // Returns the index in the liveness accounting card table bitmap 872 // for the given address 873 inline BitMap::idx_t card_bitmap_index_for(HeapWord* addr); 874 875 // Counts the size of the given memory region in the the given 876 // marked_bytes array slot for the given HeapRegion. 877 // Sets the bits in the given card bitmap that are associated with the 878 // cards that are spanned by the memory region. 879 inline void count_region(MemRegion mr, 880 HeapRegion* hr, 881 size_t* marked_bytes_array, 882 BitMap* task_card_bm); 883 884 // Counts the given memory region in the task/worker counting 885 // data structures for the given worker id. 886 inline void count_region(MemRegion mr, HeapRegion* hr, uint worker_id); 887 888 // Counts the given object in the given task/worker counting 889 // data structures. 890 inline void count_object(oop obj, 891 HeapRegion* hr, 892 size_t* marked_bytes_array, 893 BitMap* task_card_bm); 894 895 // Attempts to mark the given object and, if successful, counts 896 // the object in the given task/worker counting structures. 897 inline bool par_mark_and_count(oop obj, 898 HeapRegion* hr, 899 size_t* marked_bytes_array, 900 BitMap* task_card_bm); 901 902 // Attempts to mark the given object and, if successful, counts 903 // the object in the task/worker counting structures for the 904 // given worker id. 905 inline bool par_mark_and_count(oop obj, 906 size_t word_size, 907 HeapRegion* hr, 908 uint worker_id); 909 910 // Returns true if initialization was successfully completed. 911 bool completed_initialization() const { 912 return _completed_initialization; 913 } 914 915 protected: 916 // Clear all the per-task bitmaps and arrays used to store the 917 // counting data. 918 void clear_all_count_data(); 919 920 // Aggregates the counting data for each worker/task 921 // that was constructed while marking. Also sets 922 // the amount of marked bytes for each region and 923 // the top at concurrent mark count. 924 void aggregate_count_data(); 925 926 // Verification routine 927 void verify_count_data(); 928 }; 929 930 // A class representing a marking task. 931 class CMTask : public TerminatorTerminator { 932 private: 933 enum PrivateConstants { 934 // the regular clock call is called once the scanned words reaches 935 // this limit 936 words_scanned_period = 12*1024, 937 // the regular clock call is called once the number of visited 938 // references reaches this limit 939 refs_reached_period = 384, 940 // initial value for the hash seed, used in the work stealing code 941 init_hash_seed = 17, 942 // how many entries will be transferred between global stack and 943 // local queues 944 global_stack_transfer_size = 16 945 }; 946 947 uint _worker_id; 948 G1CollectedHeap* _g1h; 949 ConcurrentMark* _cm; 950 CMBitMap* _nextMarkBitMap; 951 // the task queue of this task 952 CMTaskQueue* _task_queue; 953 private: 954 // the task queue set---needed for stealing 955 CMTaskQueueSet* _task_queues; 956 // indicates whether the task has been claimed---this is only for 957 // debugging purposes 958 bool _claimed; 959 960 // number of calls to this task 961 int _calls; 962 963 // when the virtual timer reaches this time, the marking step should 964 // exit 965 double _time_target_ms; 966 // the start time of the current marking step 967 double _start_time_ms; 968 969 // the oop closure used for iterations over oops 970 G1CMOopClosure* _cm_oop_closure; 971 972 // the region this task is scanning, NULL if we're not scanning any 973 HeapRegion* _curr_region; 974 // the local finger of this task, NULL if we're not scanning a region 975 HeapWord* _finger; 976 // limit of the region this task is scanning, NULL if we're not scanning one 977 HeapWord* _region_limit; 978 979 // the number of words this task has scanned 980 size_t _words_scanned; 981 // When _words_scanned reaches this limit, the regular clock is 982 // called. Notice that this might be decreased under certain 983 // circumstances (i.e. when we believe that we did an expensive 984 // operation). 985 size_t _words_scanned_limit; 986 // the initial value of _words_scanned_limit (i.e. what it was 987 // before it was decreased). 988 size_t _real_words_scanned_limit; 989 990 // the number of references this task has visited 991 size_t _refs_reached; 992 // When _refs_reached reaches this limit, the regular clock is 993 // called. Notice this this might be decreased under certain 994 // circumstances (i.e. when we believe that we did an expensive 995 // operation). 996 size_t _refs_reached_limit; 997 // the initial value of _refs_reached_limit (i.e. what it was before 998 // it was decreased). 999 size_t _real_refs_reached_limit; 1000 1001 // used by the work stealing stuff 1002 int _hash_seed; 1003 // if this is true, then the task has aborted for some reason 1004 bool _has_aborted; 1005 // set when the task aborts because it has met its time quota 1006 bool _has_timed_out; 1007 // true when we're draining SATB buffers; this avoids the task 1008 // aborting due to SATB buffers being available (as we're already 1009 // dealing with them) 1010 bool _draining_satb_buffers; 1011 1012 // number sequence of past step times 1013 NumberSeq _step_times_ms; 1014 // elapsed time of this task 1015 double _elapsed_time_ms; 1016 // termination time of this task 1017 double _termination_time_ms; 1018 // when this task got into the termination protocol 1019 double _termination_start_time_ms; 1020 1021 // true when the task is during a concurrent phase, false when it is 1022 // in the remark phase (so, in the latter case, we do not have to 1023 // check all the things that we have to check during the concurrent 1024 // phase, i.e. SATB buffer availability...) 1025 bool _concurrent; 1026 1027 TruncatedSeq _marking_step_diffs_ms; 1028 1029 // Counting data structures. Embedding the task's marked_bytes_array 1030 // and card bitmap into the actual task saves having to go through 1031 // the ConcurrentMark object. 1032 size_t* _marked_bytes_array; 1033 BitMap* _card_bm; 1034 1035 // LOTS of statistics related with this task 1036 #if _MARKING_STATS_ 1037 NumberSeq _all_clock_intervals_ms; 1038 double _interval_start_time_ms; 1039 1040 size_t _aborted; 1041 size_t _aborted_overflow; 1042 size_t _aborted_cm_aborted; 1043 size_t _aborted_yield; 1044 size_t _aborted_timed_out; 1045 size_t _aborted_satb; 1046 size_t _aborted_termination; 1047 1048 size_t _steal_attempts; 1049 size_t _steals; 1050 1051 size_t _clock_due_to_marking; 1052 size_t _clock_due_to_scanning; 1053 1054 size_t _local_pushes; 1055 size_t _local_pops; 1056 size_t _local_max_size; 1057 size_t _objs_scanned; 1058 1059 size_t _global_pushes; 1060 size_t _global_pops; 1061 size_t _global_max_size; 1062 1063 size_t _global_transfers_to; 1064 size_t _global_transfers_from; 1065 1066 size_t _regions_claimed; 1067 size_t _objs_found_on_bitmap; 1068 1069 size_t _satb_buffers_processed; 1070 #endif // _MARKING_STATS_ 1071 1072 // it updates the local fields after this task has claimed 1073 // a new region to scan 1074 void setup_for_region(HeapRegion* hr); 1075 // it brings up-to-date the limit of the region 1076 void update_region_limit(); 1077 1078 // called when either the words scanned or the refs visited limit 1079 // has been reached 1080 void reached_limit(); 1081 // recalculates the words scanned and refs visited limits 1082 void recalculate_limits(); 1083 // decreases the words scanned and refs visited limits when we reach 1084 // an expensive operation 1085 void decrease_limits(); 1086 // it checks whether the words scanned or refs visited reached their 1087 // respective limit and calls reached_limit() if they have 1088 void check_limits() { 1089 if (_words_scanned >= _words_scanned_limit || 1090 _refs_reached >= _refs_reached_limit) { 1091 reached_limit(); 1092 } 1093 } 1094 // this is supposed to be called regularly during a marking step as 1095 // it checks a bunch of conditions that might cause the marking step 1096 // to abort 1097 void regular_clock_call(); 1098 bool concurrent() { return _concurrent; } 1099 1100 // Test whether objAddr might have already been passed over by the 1101 // mark bitmap scan, and so needs to be pushed onto the mark stack. 1102 bool is_below_finger(HeapWord* objAddr, HeapWord* global_finger) const; 1103 1104 public: 1105 // It resets the task; it should be called right at the beginning of 1106 // a marking phase. 1107 void reset(CMBitMap* _nextMarkBitMap); 1108 // it clears all the fields that correspond to a claimed region. 1109 void clear_region_fields(); 1110 1111 void set_concurrent(bool concurrent) { _concurrent = concurrent; } 1112 1113 // The main method of this class which performs a marking step 1114 // trying not to exceed the given duration. However, it might exit 1115 // prematurely, according to some conditions (i.e. SATB buffers are 1116 // available for processing). 1117 void do_marking_step(double target_ms, 1118 bool do_termination, 1119 bool is_serial); 1120 1121 // These two calls start and stop the timer 1122 void record_start_time() { 1123 _elapsed_time_ms = os::elapsedTime() * 1000.0; 1124 } 1125 void record_end_time() { 1126 _elapsed_time_ms = os::elapsedTime() * 1000.0 - _elapsed_time_ms; 1127 } 1128 1129 // returns the worker ID associated with this task. 1130 uint worker_id() { return _worker_id; } 1131 1132 // From TerminatorTerminator. It determines whether this task should 1133 // exit the termination protocol after it's entered it. 1134 virtual bool should_exit_termination(); 1135 1136 // Resets the local region fields after a task has finished scanning a 1137 // region; or when they have become stale as a result of the region 1138 // being evacuated. 1139 void giveup_current_region(); 1140 1141 HeapWord* finger() { return _finger; } 1142 1143 bool has_aborted() { return _has_aborted; } 1144 void set_has_aborted() { _has_aborted = true; } 1145 void clear_has_aborted() { _has_aborted = false; } 1146 bool has_timed_out() { return _has_timed_out; } 1147 bool claimed() { return _claimed; } 1148 1149 void set_cm_oop_closure(G1CMOopClosure* cm_oop_closure); 1150 1151 // It grays the object by marking it and, if necessary, pushing it 1152 // on the local queue 1153 inline void deal_with_reference(oop obj); 1154 1155 // It scans an object and visits its children. 1156 void scan_object(oop obj); 1157 1158 // It pushes an object on the local queue. 1159 inline void push(oop obj); 1160 1161 // These two move entries to/from the global stack. 1162 void move_entries_to_global_stack(); 1163 void get_entries_from_global_stack(); 1164 1165 // It pops and scans objects from the local queue. If partially is 1166 // true, then it stops when the queue size is of a given limit. If 1167 // partially is false, then it stops when the queue is empty. 1168 void drain_local_queue(bool partially); 1169 // It moves entries from the global stack to the local queue and 1170 // drains the local queue. If partially is true, then it stops when 1171 // both the global stack and the local queue reach a given size. If 1172 // partially if false, it tries to empty them totally. 1173 void drain_global_stack(bool partially); 1174 // It keeps picking SATB buffers and processing them until no SATB 1175 // buffers are available. 1176 void drain_satb_buffers(); 1177 1178 // moves the local finger to a new location 1179 inline void move_finger_to(HeapWord* new_finger) { 1180 assert(new_finger >= _finger && new_finger < _region_limit, "invariant"); 1181 _finger = new_finger; 1182 } 1183 1184 CMTask(uint worker_id, 1185 ConcurrentMark *cm, 1186 size_t* marked_bytes, 1187 BitMap* card_bm, 1188 CMTaskQueue* task_queue, 1189 CMTaskQueueSet* task_queues); 1190 1191 // it prints statistics associated with this task 1192 void print_stats(); 1193 1194 #if _MARKING_STATS_ 1195 void increase_objs_found_on_bitmap() { ++_objs_found_on_bitmap; } 1196 #endif // _MARKING_STATS_ 1197 }; 1198 1199 // Class that's used to to print out per-region liveness 1200 // information. It's currently used at the end of marking and also 1201 // after we sort the old regions at the end of the cleanup operation. 1202 class G1PrintRegionLivenessInfoClosure: public HeapRegionClosure { 1203 private: 1204 outputStream* _out; 1205 1206 // Accumulators for these values. 1207 size_t _total_used_bytes; 1208 size_t _total_capacity_bytes; 1209 size_t _total_prev_live_bytes; 1210 size_t _total_next_live_bytes; 1211 1212 // These are set up when we come across a "stars humongous" region 1213 // (as this is where most of this information is stored, not in the 1214 // subsequent "continues humongous" regions). After that, for every 1215 // region in a given humongous region series we deduce the right 1216 // values for it by simply subtracting the appropriate amount from 1217 // these fields. All these values should reach 0 after we've visited 1218 // the last region in the series. 1219 size_t _hum_used_bytes; 1220 size_t _hum_capacity_bytes; 1221 size_t _hum_prev_live_bytes; 1222 size_t _hum_next_live_bytes; 1223 1224 // Accumulator for the remembered set size 1225 size_t _total_remset_bytes; 1226 1227 // Accumulator for strong code roots memory size 1228 size_t _total_strong_code_roots_bytes; 1229 1230 static double perc(size_t val, size_t total) { 1231 if (total == 0) { 1232 return 0.0; 1233 } else { 1234 return 100.0 * ((double) val / (double) total); 1235 } 1236 } 1237 1238 static double bytes_to_mb(size_t val) { 1239 return (double) val / (double) M; 1240 } 1241 1242 // See the .cpp file. 1243 size_t get_hum_bytes(size_t* hum_bytes); 1244 void get_hum_bytes(size_t* used_bytes, size_t* capacity_bytes, 1245 size_t* prev_live_bytes, size_t* next_live_bytes); 1246 1247 public: 1248 // The header and footer are printed in the constructor and 1249 // destructor respectively. 1250 G1PrintRegionLivenessInfoClosure(outputStream* out, const char* phase_name); 1251 virtual bool doHeapRegion(HeapRegion* r); 1252 ~G1PrintRegionLivenessInfoClosure(); 1253 }; 1254 1255 #endif // SHARE_VM_GC_IMPLEMENTATION_G1_CONCURRENTMARK_HPP