Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(446)

Side by Side Diff: src/spaces.h

Issue 437993003: Move a bunch of GC related files to heap/ subdirectory (Closed) Base URL: https://v8.googlecode.com/svn/branches/bleeding_edge
Patch Set: make presubmit happy Created 6 years, 4 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch | Annotate | Revision Log
« no previous file with comments | « src/safepoint-table.h ('k') | src/spaces.cc » ('j') | no next file with comments »
Toggle Intra-line Diffs ('i') | Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
OLDNEW
(Empty)
1 // Copyright 2011 the V8 project authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 #ifndef V8_SPACES_H_
6 #define V8_SPACES_H_
7
8 #include "src/allocation.h"
9 #include "src/base/atomicops.h"
10 #include "src/base/platform/mutex.h"
11 #include "src/hashmap.h"
12 #include "src/list.h"
13 #include "src/log.h"
14 #include "src/utils.h"
15
16 namespace v8 {
17 namespace internal {
18
19 class Isolate;
20
21 // -----------------------------------------------------------------------------
22 // Heap structures:
23 //
24 // A JS heap consists of a young generation, an old generation, and a large
25 // object space. The young generation is divided into two semispaces. A
26 // scavenger implements Cheney's copying algorithm. The old generation is
27 // separated into a map space and an old object space. The map space contains
28 // all (and only) map objects, the rest of old objects go into the old space.
29 // The old generation is collected by a mark-sweep-compact collector.
30 //
31 // The semispaces of the young generation are contiguous. The old and map
32 // spaces consists of a list of pages. A page has a page header and an object
33 // area.
34 //
35 // There is a separate large object space for objects larger than
36 // Page::kMaxHeapObjectSize, so that they do not have to move during
37 // collection. The large object space is paged. Pages in large object space
38 // may be larger than the page size.
39 //
40 // A store-buffer based write barrier is used to keep track of intergenerational
41 // references. See store-buffer.h.
42 //
43 // During scavenges and mark-sweep collections we sometimes (after a store
44 // buffer overflow) iterate intergenerational pointers without decoding heap
45 // object maps so if the page belongs to old pointer space or large object
46 // space it is essential to guarantee that the page does not contain any
47 // garbage pointers to new space: every pointer aligned word which satisfies
48 // the Heap::InNewSpace() predicate must be a pointer to a live heap object in
49 // new space. Thus objects in old pointer and large object spaces should have a
50 // special layout (e.g. no bare integer fields). This requirement does not
51 // apply to map space which is iterated in a special fashion. However we still
52 // require pointer fields of dead maps to be cleaned.
53 //
54 // To enable lazy cleaning of old space pages we can mark chunks of the page
55 // as being garbage. Garbage sections are marked with a special map. These
56 // sections are skipped when scanning the page, even if we are otherwise
57 // scanning without regard for object boundaries. Garbage sections are chained
58 // together to form a free list after a GC. Garbage sections created outside
59 // of GCs by object trunctation etc. may not be in the free list chain. Very
60 // small free spaces are ignored, they need only be cleaned of bogus pointers
61 // into new space.
62 //
63 // Each page may have up to one special garbage section. The start of this
64 // section is denoted by the top field in the space. The end of the section
65 // is denoted by the limit field in the space. This special garbage section
66 // is not marked with a free space map in the data. The point of this section
67 // is to enable linear allocation without having to constantly update the byte
68 // array every time the top field is updated and a new object is created. The
69 // special garbage section is not in the chain of garbage sections.
70 //
71 // Since the top and limit fields are in the space, not the page, only one page
72 // has a special garbage section, and if the top and limit are equal then there
73 // is no special garbage section.
74
75 // Some assertion macros used in the debugging mode.
76
77 #define DCHECK_PAGE_ALIGNED(address) \
78 DCHECK((OffsetFrom(address) & Page::kPageAlignmentMask) == 0)
79
80 #define DCHECK_OBJECT_ALIGNED(address) \
81 DCHECK((OffsetFrom(address) & kObjectAlignmentMask) == 0)
82
83 #define DCHECK_OBJECT_SIZE(size) \
84 DCHECK((0 < size) && (size <= Page::kMaxRegularHeapObjectSize))
85
86 #define DCHECK_PAGE_OFFSET(offset) \
87 DCHECK((Page::kObjectStartOffset <= offset) \
88 && (offset <= Page::kPageSize))
89
90 #define DCHECK_MAP_PAGE_INDEX(index) \
91 DCHECK((0 <= index) && (index <= MapSpace::kMaxMapPageIndex))
92
93
94 class PagedSpace;
95 class MemoryAllocator;
96 class AllocationInfo;
97 class Space;
98 class FreeList;
99 class MemoryChunk;
100
101 class MarkBit {
102 public:
103 typedef uint32_t CellType;
104
105 inline MarkBit(CellType* cell, CellType mask, bool data_only)
106 : cell_(cell), mask_(mask), data_only_(data_only) { }
107
108 inline CellType* cell() { return cell_; }
109 inline CellType mask() { return mask_; }
110
111 #ifdef DEBUG
112 bool operator==(const MarkBit& other) {
113 return cell_ == other.cell_ && mask_ == other.mask_;
114 }
115 #endif
116
117 inline void Set() { *cell_ |= mask_; }
118 inline bool Get() { return (*cell_ & mask_) != 0; }
119 inline void Clear() { *cell_ &= ~mask_; }
120
121 inline bool data_only() { return data_only_; }
122
123 inline MarkBit Next() {
124 CellType new_mask = mask_ << 1;
125 if (new_mask == 0) {
126 return MarkBit(cell_ + 1, 1, data_only_);
127 } else {
128 return MarkBit(cell_, new_mask, data_only_);
129 }
130 }
131
132 private:
133 CellType* cell_;
134 CellType mask_;
135 // This boolean indicates that the object is in a data-only space with no
136 // pointers. This enables some optimizations when marking.
137 // It is expected that this field is inlined and turned into control flow
138 // at the place where the MarkBit object is created.
139 bool data_only_;
140 };
141
142
143 // Bitmap is a sequence of cells each containing fixed number of bits.
144 class Bitmap {
145 public:
146 static const uint32_t kBitsPerCell = 32;
147 static const uint32_t kBitsPerCellLog2 = 5;
148 static const uint32_t kBitIndexMask = kBitsPerCell - 1;
149 static const uint32_t kBytesPerCell = kBitsPerCell / kBitsPerByte;
150 static const uint32_t kBytesPerCellLog2 = kBitsPerCellLog2 - kBitsPerByteLog2;
151
152 static const size_t kLength =
153 (1 << kPageSizeBits) >> (kPointerSizeLog2);
154
155 static const size_t kSize =
156 (1 << kPageSizeBits) >> (kPointerSizeLog2 + kBitsPerByteLog2);
157
158
159 static int CellsForLength(int length) {
160 return (length + kBitsPerCell - 1) >> kBitsPerCellLog2;
161 }
162
163 int CellsCount() {
164 return CellsForLength(kLength);
165 }
166
167 static int SizeFor(int cells_count) {
168 return sizeof(MarkBit::CellType) * cells_count;
169 }
170
171 INLINE(static uint32_t IndexToCell(uint32_t index)) {
172 return index >> kBitsPerCellLog2;
173 }
174
175 INLINE(static uint32_t CellToIndex(uint32_t index)) {
176 return index << kBitsPerCellLog2;
177 }
178
179 INLINE(static uint32_t CellAlignIndex(uint32_t index)) {
180 return (index + kBitIndexMask) & ~kBitIndexMask;
181 }
182
183 INLINE(MarkBit::CellType* cells()) {
184 return reinterpret_cast<MarkBit::CellType*>(this);
185 }
186
187 INLINE(Address address()) {
188 return reinterpret_cast<Address>(this);
189 }
190
191 INLINE(static Bitmap* FromAddress(Address addr)) {
192 return reinterpret_cast<Bitmap*>(addr);
193 }
194
195 inline MarkBit MarkBitFromIndex(uint32_t index, bool data_only = false) {
196 MarkBit::CellType mask = 1 << (index & kBitIndexMask);
197 MarkBit::CellType* cell = this->cells() + (index >> kBitsPerCellLog2);
198 return MarkBit(cell, mask, data_only);
199 }
200
201 static inline void Clear(MemoryChunk* chunk);
202
203 static void PrintWord(uint32_t word, uint32_t himask = 0) {
204 for (uint32_t mask = 1; mask != 0; mask <<= 1) {
205 if ((mask & himask) != 0) PrintF("[");
206 PrintF((mask & word) ? "1" : "0");
207 if ((mask & himask) != 0) PrintF("]");
208 }
209 }
210
211 class CellPrinter {
212 public:
213 CellPrinter() : seq_start(0), seq_type(0), seq_length(0) { }
214
215 void Print(uint32_t pos, uint32_t cell) {
216 if (cell == seq_type) {
217 seq_length++;
218 return;
219 }
220
221 Flush();
222
223 if (IsSeq(cell)) {
224 seq_start = pos;
225 seq_length = 0;
226 seq_type = cell;
227 return;
228 }
229
230 PrintF("%d: ", pos);
231 PrintWord(cell);
232 PrintF("\n");
233 }
234
235 void Flush() {
236 if (seq_length > 0) {
237 PrintF("%d: %dx%d\n",
238 seq_start,
239 seq_type == 0 ? 0 : 1,
240 seq_length * kBitsPerCell);
241 seq_length = 0;
242 }
243 }
244
245 static bool IsSeq(uint32_t cell) { return cell == 0 || cell == 0xFFFFFFFF; }
246
247 private:
248 uint32_t seq_start;
249 uint32_t seq_type;
250 uint32_t seq_length;
251 };
252
253 void Print() {
254 CellPrinter printer;
255 for (int i = 0; i < CellsCount(); i++) {
256 printer.Print(i, cells()[i]);
257 }
258 printer.Flush();
259 PrintF("\n");
260 }
261
262 bool IsClean() {
263 for (int i = 0; i < CellsCount(); i++) {
264 if (cells()[i] != 0) {
265 return false;
266 }
267 }
268 return true;
269 }
270 };
271
272
273 class SkipList;
274 class SlotsBuffer;
275
276 // MemoryChunk represents a memory region owned by a specific space.
277 // It is divided into the header and the body. Chunk start is always
278 // 1MB aligned. Start of the body is aligned so it can accommodate
279 // any heap object.
280 class MemoryChunk {
281 public:
282 // Only works if the pointer is in the first kPageSize of the MemoryChunk.
283 static MemoryChunk* FromAddress(Address a) {
284 return reinterpret_cast<MemoryChunk*>(OffsetFrom(a) & ~kAlignmentMask);
285 }
286 static const MemoryChunk* FromAddress(const byte* a) {
287 return reinterpret_cast<const MemoryChunk*>(
288 OffsetFrom(a) & ~kAlignmentMask);
289 }
290
291 // Only works for addresses in pointer spaces, not data or code spaces.
292 static inline MemoryChunk* FromAnyPointerAddress(Heap* heap, Address addr);
293
294 Address address() { return reinterpret_cast<Address>(this); }
295
296 bool is_valid() { return address() != NULL; }
297
298 MemoryChunk* next_chunk() const {
299 return reinterpret_cast<MemoryChunk*>(base::Acquire_Load(&next_chunk_));
300 }
301
302 MemoryChunk* prev_chunk() const {
303 return reinterpret_cast<MemoryChunk*>(base::Acquire_Load(&prev_chunk_));
304 }
305
306 void set_next_chunk(MemoryChunk* next) {
307 base::Release_Store(&next_chunk_, reinterpret_cast<base::AtomicWord>(next));
308 }
309
310 void set_prev_chunk(MemoryChunk* prev) {
311 base::Release_Store(&prev_chunk_, reinterpret_cast<base::AtomicWord>(prev));
312 }
313
314 Space* owner() const {
315 if ((reinterpret_cast<intptr_t>(owner_) & kPageHeaderTagMask) ==
316 kPageHeaderTag) {
317 return reinterpret_cast<Space*>(reinterpret_cast<intptr_t>(owner_) -
318 kPageHeaderTag);
319 } else {
320 return NULL;
321 }
322 }
323
324 void set_owner(Space* space) {
325 DCHECK((reinterpret_cast<intptr_t>(space) & kPageHeaderTagMask) == 0);
326 owner_ = reinterpret_cast<Address>(space) + kPageHeaderTag;
327 DCHECK((reinterpret_cast<intptr_t>(owner_) & kPageHeaderTagMask) ==
328 kPageHeaderTag);
329 }
330
331 base::VirtualMemory* reserved_memory() {
332 return &reservation_;
333 }
334
335 void InitializeReservedMemory() {
336 reservation_.Reset();
337 }
338
339 void set_reserved_memory(base::VirtualMemory* reservation) {
340 DCHECK_NOT_NULL(reservation);
341 reservation_.TakeControl(reservation);
342 }
343
344 bool scan_on_scavenge() { return IsFlagSet(SCAN_ON_SCAVENGE); }
345 void initialize_scan_on_scavenge(bool scan) {
346 if (scan) {
347 SetFlag(SCAN_ON_SCAVENGE);
348 } else {
349 ClearFlag(SCAN_ON_SCAVENGE);
350 }
351 }
352 inline void set_scan_on_scavenge(bool scan);
353
354 int store_buffer_counter() { return store_buffer_counter_; }
355 void set_store_buffer_counter(int counter) {
356 store_buffer_counter_ = counter;
357 }
358
359 bool Contains(Address addr) {
360 return addr >= area_start() && addr < area_end();
361 }
362
363 // Checks whether addr can be a limit of addresses in this page.
364 // It's a limit if it's in the page, or if it's just after the
365 // last byte of the page.
366 bool ContainsLimit(Address addr) {
367 return addr >= area_start() && addr <= area_end();
368 }
369
370 // Every n write barrier invocations we go to runtime even though
371 // we could have handled it in generated code. This lets us check
372 // whether we have hit the limit and should do some more marking.
373 static const int kWriteBarrierCounterGranularity = 500;
374
375 enum MemoryChunkFlags {
376 IS_EXECUTABLE,
377 ABOUT_TO_BE_FREED,
378 POINTERS_TO_HERE_ARE_INTERESTING,
379 POINTERS_FROM_HERE_ARE_INTERESTING,
380 SCAN_ON_SCAVENGE,
381 IN_FROM_SPACE, // Mutually exclusive with IN_TO_SPACE.
382 IN_TO_SPACE, // All pages in new space has one of these two set.
383 NEW_SPACE_BELOW_AGE_MARK,
384 CONTAINS_ONLY_DATA,
385 EVACUATION_CANDIDATE,
386 RESCAN_ON_EVACUATION,
387
388 // Pages swept precisely can be iterated, hitting only the live objects.
389 // Whereas those swept conservatively cannot be iterated over. Both flags
390 // indicate that marking bits have been cleared by the sweeper, otherwise
391 // marking bits are still intact.
392 WAS_SWEPT_PRECISELY,
393 WAS_SWEPT_CONSERVATIVELY,
394
395 // Large objects can have a progress bar in their page header. These object
396 // are scanned in increments and will be kept black while being scanned.
397 // Even if the mutator writes to them they will be kept black and a white
398 // to grey transition is performed in the value.
399 HAS_PROGRESS_BAR,
400
401 // Last flag, keep at bottom.
402 NUM_MEMORY_CHUNK_FLAGS
403 };
404
405
406 static const int kPointersToHereAreInterestingMask =
407 1 << POINTERS_TO_HERE_ARE_INTERESTING;
408
409 static const int kPointersFromHereAreInterestingMask =
410 1 << POINTERS_FROM_HERE_ARE_INTERESTING;
411
412 static const int kEvacuationCandidateMask =
413 1 << EVACUATION_CANDIDATE;
414
415 static const int kSkipEvacuationSlotsRecordingMask =
416 (1 << EVACUATION_CANDIDATE) |
417 (1 << RESCAN_ON_EVACUATION) |
418 (1 << IN_FROM_SPACE) |
419 (1 << IN_TO_SPACE);
420
421
422 void SetFlag(int flag) {
423 flags_ |= static_cast<uintptr_t>(1) << flag;
424 }
425
426 void ClearFlag(int flag) {
427 flags_ &= ~(static_cast<uintptr_t>(1) << flag);
428 }
429
430 void SetFlagTo(int flag, bool value) {
431 if (value) {
432 SetFlag(flag);
433 } else {
434 ClearFlag(flag);
435 }
436 }
437
438 bool IsFlagSet(int flag) {
439 return (flags_ & (static_cast<uintptr_t>(1) << flag)) != 0;
440 }
441
442 // Set or clear multiple flags at a time. The flags in the mask
443 // are set to the value in "flags", the rest retain the current value
444 // in flags_.
445 void SetFlags(intptr_t flags, intptr_t mask) {
446 flags_ = (flags_ & ~mask) | (flags & mask);
447 }
448
449 // Return all current flags.
450 intptr_t GetFlags() { return flags_; }
451
452
453 // SWEEPING_DONE - The page state when sweeping is complete or sweeping must
454 // not be performed on that page.
455 // SWEEPING_FINALIZE - A sweeper thread is done sweeping this page and will
456 // not touch the page memory anymore.
457 // SWEEPING_IN_PROGRESS - This page is currently swept by a sweeper thread.
458 // SWEEPING_PENDING - This page is ready for parallel sweeping.
459 enum ParallelSweepingState {
460 SWEEPING_DONE,
461 SWEEPING_FINALIZE,
462 SWEEPING_IN_PROGRESS,
463 SWEEPING_PENDING
464 };
465
466 ParallelSweepingState parallel_sweeping() {
467 return static_cast<ParallelSweepingState>(
468 base::Acquire_Load(&parallel_sweeping_));
469 }
470
471 void set_parallel_sweeping(ParallelSweepingState state) {
472 base::Release_Store(&parallel_sweeping_, state);
473 }
474
475 bool TryParallelSweeping() {
476 return base::Acquire_CompareAndSwap(
477 &parallel_sweeping_, SWEEPING_PENDING, SWEEPING_IN_PROGRESS) ==
478 SWEEPING_PENDING;
479 }
480
481 bool SweepingCompleted() { return parallel_sweeping() <= SWEEPING_FINALIZE; }
482
483 // Manage live byte count (count of bytes known to be live,
484 // because they are marked black).
485 void ResetLiveBytes() {
486 if (FLAG_gc_verbose) {
487 PrintF("ResetLiveBytes:%p:%x->0\n",
488 static_cast<void*>(this), live_byte_count_);
489 }
490 live_byte_count_ = 0;
491 }
492 void IncrementLiveBytes(int by) {
493 if (FLAG_gc_verbose) {
494 printf("UpdateLiveBytes:%p:%x%c=%x->%x\n",
495 static_cast<void*>(this), live_byte_count_,
496 ((by < 0) ? '-' : '+'), ((by < 0) ? -by : by),
497 live_byte_count_ + by);
498 }
499 live_byte_count_ += by;
500 DCHECK_LE(static_cast<unsigned>(live_byte_count_), size_);
501 }
502 int LiveBytes() {
503 DCHECK(static_cast<unsigned>(live_byte_count_) <= size_);
504 return live_byte_count_;
505 }
506
507 int write_barrier_counter() {
508 return static_cast<int>(write_barrier_counter_);
509 }
510
511 void set_write_barrier_counter(int counter) {
512 write_barrier_counter_ = counter;
513 }
514
515 int progress_bar() {
516 DCHECK(IsFlagSet(HAS_PROGRESS_BAR));
517 return progress_bar_;
518 }
519
520 void set_progress_bar(int progress_bar) {
521 DCHECK(IsFlagSet(HAS_PROGRESS_BAR));
522 progress_bar_ = progress_bar;
523 }
524
525 void ResetProgressBar() {
526 if (IsFlagSet(MemoryChunk::HAS_PROGRESS_BAR)) {
527 set_progress_bar(0);
528 ClearFlag(MemoryChunk::HAS_PROGRESS_BAR);
529 }
530 }
531
532 bool IsLeftOfProgressBar(Object** slot) {
533 Address slot_address = reinterpret_cast<Address>(slot);
534 DCHECK(slot_address > this->address());
535 return (slot_address - (this->address() + kObjectStartOffset)) <
536 progress_bar();
537 }
538
539 static void IncrementLiveBytesFromGC(Address address, int by) {
540 MemoryChunk::FromAddress(address)->IncrementLiveBytes(by);
541 }
542
543 static void IncrementLiveBytesFromMutator(Address address, int by);
544
545 static const intptr_t kAlignment =
546 (static_cast<uintptr_t>(1) << kPageSizeBits);
547
548 static const intptr_t kAlignmentMask = kAlignment - 1;
549
550 static const intptr_t kSizeOffset = 0;
551
552 static const intptr_t kLiveBytesOffset =
553 kSizeOffset + kPointerSize + kPointerSize + kPointerSize +
554 kPointerSize + kPointerSize +
555 kPointerSize + kPointerSize + kPointerSize + kIntSize;
556
557 static const size_t kSlotsBufferOffset = kLiveBytesOffset + kIntSize;
558
559 static const size_t kWriteBarrierCounterOffset =
560 kSlotsBufferOffset + kPointerSize + kPointerSize;
561
562 static const size_t kHeaderSize = kWriteBarrierCounterOffset + kPointerSize +
563 kIntSize + kIntSize + kPointerSize +
564 5 * kPointerSize +
565 kPointerSize + kPointerSize;
566
567 static const int kBodyOffset =
568 CODE_POINTER_ALIGN(kHeaderSize + Bitmap::kSize);
569
570 // The start offset of the object area in a page. Aligned to both maps and
571 // code alignment to be suitable for both. Also aligned to 32 words because
572 // the marking bitmap is arranged in 32 bit chunks.
573 static const int kObjectStartAlignment = 32 * kPointerSize;
574 static const int kObjectStartOffset = kBodyOffset - 1 +
575 (kObjectStartAlignment - (kBodyOffset - 1) % kObjectStartAlignment);
576
577 size_t size() const { return size_; }
578
579 void set_size(size_t size) {
580 size_ = size;
581 }
582
583 void SetArea(Address area_start, Address area_end) {
584 area_start_ = area_start;
585 area_end_ = area_end;
586 }
587
588 Executability executable() {
589 return IsFlagSet(IS_EXECUTABLE) ? EXECUTABLE : NOT_EXECUTABLE;
590 }
591
592 bool ContainsOnlyData() {
593 return IsFlagSet(CONTAINS_ONLY_DATA);
594 }
595
596 bool InNewSpace() {
597 return (flags_ & ((1 << IN_FROM_SPACE) | (1 << IN_TO_SPACE))) != 0;
598 }
599
600 bool InToSpace() {
601 return IsFlagSet(IN_TO_SPACE);
602 }
603
604 bool InFromSpace() {
605 return IsFlagSet(IN_FROM_SPACE);
606 }
607
608 // ---------------------------------------------------------------------
609 // Markbits support
610
611 inline Bitmap* markbits() {
612 return Bitmap::FromAddress(address() + kHeaderSize);
613 }
614
615 void PrintMarkbits() { markbits()->Print(); }
616
617 inline uint32_t AddressToMarkbitIndex(Address addr) {
618 return static_cast<uint32_t>(addr - this->address()) >> kPointerSizeLog2;
619 }
620
621 inline static uint32_t FastAddressToMarkbitIndex(Address addr) {
622 const intptr_t offset =
623 reinterpret_cast<intptr_t>(addr) & kAlignmentMask;
624
625 return static_cast<uint32_t>(offset) >> kPointerSizeLog2;
626 }
627
628 inline Address MarkbitIndexToAddress(uint32_t index) {
629 return this->address() + (index << kPointerSizeLog2);
630 }
631
632 void InsertAfter(MemoryChunk* other);
633 void Unlink();
634
635 inline Heap* heap() const { return heap_; }
636
637 static const int kFlagsOffset = kPointerSize;
638
639 bool IsEvacuationCandidate() { return IsFlagSet(EVACUATION_CANDIDATE); }
640
641 bool ShouldSkipEvacuationSlotRecording() {
642 return (flags_ & kSkipEvacuationSlotsRecordingMask) != 0;
643 }
644
645 inline SkipList* skip_list() {
646 return skip_list_;
647 }
648
649 inline void set_skip_list(SkipList* skip_list) {
650 skip_list_ = skip_list;
651 }
652
653 inline SlotsBuffer* slots_buffer() {
654 return slots_buffer_;
655 }
656
657 inline SlotsBuffer** slots_buffer_address() {
658 return &slots_buffer_;
659 }
660
661 void MarkEvacuationCandidate() {
662 DCHECK(slots_buffer_ == NULL);
663 SetFlag(EVACUATION_CANDIDATE);
664 }
665
666 void ClearEvacuationCandidate() {
667 DCHECK(slots_buffer_ == NULL);
668 ClearFlag(EVACUATION_CANDIDATE);
669 }
670
671 Address area_start() { return area_start_; }
672 Address area_end() { return area_end_; }
673 int area_size() {
674 return static_cast<int>(area_end() - area_start());
675 }
676 bool CommitArea(size_t requested);
677
678 // Approximate amount of physical memory committed for this chunk.
679 size_t CommittedPhysicalMemory() {
680 return high_water_mark_;
681 }
682
683 static inline void UpdateHighWaterMark(Address mark);
684
685 protected:
686 size_t size_;
687 intptr_t flags_;
688
689 // Start and end of allocatable memory on this chunk.
690 Address area_start_;
691 Address area_end_;
692
693 // If the chunk needs to remember its memory reservation, it is stored here.
694 base::VirtualMemory reservation_;
695 // The identity of the owning space. This is tagged as a failure pointer, but
696 // no failure can be in an object, so this can be distinguished from any entry
697 // in a fixed array.
698 Address owner_;
699 Heap* heap_;
700 // Used by the store buffer to keep track of which pages to mark scan-on-
701 // scavenge.
702 int store_buffer_counter_;
703 // Count of bytes marked black on page.
704 int live_byte_count_;
705 SlotsBuffer* slots_buffer_;
706 SkipList* skip_list_;
707 intptr_t write_barrier_counter_;
708 // Used by the incremental marker to keep track of the scanning progress in
709 // large objects that have a progress bar and are scanned in increments.
710 int progress_bar_;
711 // Assuming the initial allocation on a page is sequential,
712 // count highest number of bytes ever allocated on the page.
713 int high_water_mark_;
714
715 base::AtomicWord parallel_sweeping_;
716
717 // PagedSpace free-list statistics.
718 intptr_t available_in_small_free_list_;
719 intptr_t available_in_medium_free_list_;
720 intptr_t available_in_large_free_list_;
721 intptr_t available_in_huge_free_list_;
722 intptr_t non_available_small_blocks_;
723
724 static MemoryChunk* Initialize(Heap* heap,
725 Address base,
726 size_t size,
727 Address area_start,
728 Address area_end,
729 Executability executable,
730 Space* owner);
731
732 private:
733 // next_chunk_ holds a pointer of type MemoryChunk
734 base::AtomicWord next_chunk_;
735 // prev_chunk_ holds a pointer of type MemoryChunk
736 base::AtomicWord prev_chunk_;
737
738 friend class MemoryAllocator;
739 };
740
741
742 STATIC_ASSERT(sizeof(MemoryChunk) <= MemoryChunk::kHeaderSize);
743
744
745 // -----------------------------------------------------------------------------
746 // A page is a memory chunk of a size 1MB. Large object pages may be larger.
747 //
748 // The only way to get a page pointer is by calling factory methods:
749 // Page* p = Page::FromAddress(addr); or
750 // Page* p = Page::FromAllocationTop(top);
751 class Page : public MemoryChunk {
752 public:
753 // Returns the page containing a given address. The address ranges
754 // from [page_addr .. page_addr + kPageSize[
755 // This only works if the object is in fact in a page. See also MemoryChunk::
756 // FromAddress() and FromAnyAddress().
757 INLINE(static Page* FromAddress(Address a)) {
758 return reinterpret_cast<Page*>(OffsetFrom(a) & ~kPageAlignmentMask);
759 }
760
761 // Returns the page containing an allocation top. Because an allocation
762 // top address can be the upper bound of the page, we need to subtract
763 // it with kPointerSize first. The address ranges from
764 // [page_addr + kObjectStartOffset .. page_addr + kPageSize].
765 INLINE(static Page* FromAllocationTop(Address top)) {
766 Page* p = FromAddress(top - kPointerSize);
767 return p;
768 }
769
770 // Returns the next page in the chain of pages owned by a space.
771 inline Page* next_page();
772 inline Page* prev_page();
773 inline void set_next_page(Page* page);
774 inline void set_prev_page(Page* page);
775
776 // Checks whether an address is page aligned.
777 static bool IsAlignedToPageSize(Address a) {
778 return 0 == (OffsetFrom(a) & kPageAlignmentMask);
779 }
780
781 // Returns the offset of a given address to this page.
782 INLINE(int Offset(Address a)) {
783 int offset = static_cast<int>(a - address());
784 return offset;
785 }
786
787 // Returns the address for a given offset to the this page.
788 Address OffsetToAddress(int offset) {
789 DCHECK_PAGE_OFFSET(offset);
790 return address() + offset;
791 }
792
793 // ---------------------------------------------------------------------
794
795 // Page size in bytes. This must be a multiple of the OS page size.
796 static const int kPageSize = 1 << kPageSizeBits;
797
798 // Maximum object size that fits in a page. Objects larger than that size
799 // are allocated in large object space and are never moved in memory. This
800 // also applies to new space allocation, since objects are never migrated
801 // from new space to large object space. Takes double alignment into account.
802 static const int kMaxRegularHeapObjectSize = kPageSize - kObjectStartOffset;
803
804 // Page size mask.
805 static const intptr_t kPageAlignmentMask = (1 << kPageSizeBits) - 1;
806
807 inline void ClearGCFields();
808
809 static inline Page* Initialize(Heap* heap,
810 MemoryChunk* chunk,
811 Executability executable,
812 PagedSpace* owner);
813
814 void InitializeAsAnchor(PagedSpace* owner);
815
816 bool WasSweptPrecisely() { return IsFlagSet(WAS_SWEPT_PRECISELY); }
817 bool WasSweptConservatively() { return IsFlagSet(WAS_SWEPT_CONSERVATIVELY); }
818 bool WasSwept() { return WasSweptPrecisely() || WasSweptConservatively(); }
819
820 void MarkSweptPrecisely() { SetFlag(WAS_SWEPT_PRECISELY); }
821 void MarkSweptConservatively() { SetFlag(WAS_SWEPT_CONSERVATIVELY); }
822
823 void ClearSweptPrecisely() { ClearFlag(WAS_SWEPT_PRECISELY); }
824 void ClearSweptConservatively() { ClearFlag(WAS_SWEPT_CONSERVATIVELY); }
825
826 void ResetFreeListStatistics();
827
828 #define FRAGMENTATION_STATS_ACCESSORS(type, name) \
829 type name() { return name##_; } \
830 void set_##name(type name) { name##_ = name; } \
831 void add_##name(type name) { name##_ += name; }
832
833 FRAGMENTATION_STATS_ACCESSORS(intptr_t, non_available_small_blocks)
834 FRAGMENTATION_STATS_ACCESSORS(intptr_t, available_in_small_free_list)
835 FRAGMENTATION_STATS_ACCESSORS(intptr_t, available_in_medium_free_list)
836 FRAGMENTATION_STATS_ACCESSORS(intptr_t, available_in_large_free_list)
837 FRAGMENTATION_STATS_ACCESSORS(intptr_t, available_in_huge_free_list)
838
839 #undef FRAGMENTATION_STATS_ACCESSORS
840
841 #ifdef DEBUG
842 void Print();
843 #endif // DEBUG
844
845 friend class MemoryAllocator;
846 };
847
848
849 STATIC_ASSERT(sizeof(Page) <= MemoryChunk::kHeaderSize);
850
851
852 class LargePage : public MemoryChunk {
853 public:
854 HeapObject* GetObject() {
855 return HeapObject::FromAddress(area_start());
856 }
857
858 inline LargePage* next_page() const {
859 return static_cast<LargePage*>(next_chunk());
860 }
861
862 inline void set_next_page(LargePage* page) {
863 set_next_chunk(page);
864 }
865 private:
866 static inline LargePage* Initialize(Heap* heap, MemoryChunk* chunk);
867
868 friend class MemoryAllocator;
869 };
870
871 STATIC_ASSERT(sizeof(LargePage) <= MemoryChunk::kHeaderSize);
872
873 // ----------------------------------------------------------------------------
874 // Space is the abstract superclass for all allocation spaces.
875 class Space : public Malloced {
876 public:
877 Space(Heap* heap, AllocationSpace id, Executability executable)
878 : heap_(heap), id_(id), executable_(executable) {}
879
880 virtual ~Space() {}
881
882 Heap* heap() const { return heap_; }
883
884 // Does the space need executable memory?
885 Executability executable() { return executable_; }
886
887 // Identity used in error reporting.
888 AllocationSpace identity() { return id_; }
889
890 // Returns allocated size.
891 virtual intptr_t Size() = 0;
892
893 // Returns size of objects. Can differ from the allocated size
894 // (e.g. see LargeObjectSpace).
895 virtual intptr_t SizeOfObjects() { return Size(); }
896
897 virtual int RoundSizeDownToObjectAlignment(int size) {
898 if (id_ == CODE_SPACE) {
899 return RoundDown(size, kCodeAlignment);
900 } else {
901 return RoundDown(size, kPointerSize);
902 }
903 }
904
905 #ifdef DEBUG
906 virtual void Print() = 0;
907 #endif
908
909 private:
910 Heap* heap_;
911 AllocationSpace id_;
912 Executability executable_;
913 };
914
915
916 // ----------------------------------------------------------------------------
917 // All heap objects containing executable code (code objects) must be allocated
918 // from a 2 GB range of memory, so that they can call each other using 32-bit
919 // displacements. This happens automatically on 32-bit platforms, where 32-bit
920 // displacements cover the entire 4GB virtual address space. On 64-bit
921 // platforms, we support this using the CodeRange object, which reserves and
922 // manages a range of virtual memory.
923 class CodeRange {
924 public:
925 explicit CodeRange(Isolate* isolate);
926 ~CodeRange() { TearDown(); }
927
928 // Reserves a range of virtual memory, but does not commit any of it.
929 // Can only be called once, at heap initialization time.
930 // Returns false on failure.
931 bool SetUp(size_t requested_size);
932
933 // Frees the range of virtual memory, and frees the data structures used to
934 // manage it.
935 void TearDown();
936
937 bool valid() { return code_range_ != NULL; }
938 Address start() {
939 DCHECK(valid());
940 return static_cast<Address>(code_range_->address());
941 }
942 bool contains(Address address) {
943 if (!valid()) return false;
944 Address start = static_cast<Address>(code_range_->address());
945 return start <= address && address < start + code_range_->size();
946 }
947
948 // Allocates a chunk of memory from the large-object portion of
949 // the code range. On platforms with no separate code range, should
950 // not be called.
951 MUST_USE_RESULT Address AllocateRawMemory(const size_t requested_size,
952 const size_t commit_size,
953 size_t* allocated);
954 bool CommitRawMemory(Address start, size_t length);
955 bool UncommitRawMemory(Address start, size_t length);
956 void FreeRawMemory(Address buf, size_t length);
957
958 private:
959 Isolate* isolate_;
960
961 // The reserved range of virtual memory that all code objects are put in.
962 base::VirtualMemory* code_range_;
963 // Plain old data class, just a struct plus a constructor.
964 class FreeBlock {
965 public:
966 FreeBlock(Address start_arg, size_t size_arg)
967 : start(start_arg), size(size_arg) {
968 DCHECK(IsAddressAligned(start, MemoryChunk::kAlignment));
969 DCHECK(size >= static_cast<size_t>(Page::kPageSize));
970 }
971 FreeBlock(void* start_arg, size_t size_arg)
972 : start(static_cast<Address>(start_arg)), size(size_arg) {
973 DCHECK(IsAddressAligned(start, MemoryChunk::kAlignment));
974 DCHECK(size >= static_cast<size_t>(Page::kPageSize));
975 }
976
977 Address start;
978 size_t size;
979 };
980
981 // Freed blocks of memory are added to the free list. When the allocation
982 // list is exhausted, the free list is sorted and merged to make the new
983 // allocation list.
984 List<FreeBlock> free_list_;
985 // Memory is allocated from the free blocks on the allocation list.
986 // The block at current_allocation_block_index_ is the current block.
987 List<FreeBlock> allocation_list_;
988 int current_allocation_block_index_;
989
990 // Finds a block on the allocation list that contains at least the
991 // requested amount of memory. If none is found, sorts and merges
992 // the existing free memory blocks, and searches again.
993 // If none can be found, returns false.
994 bool GetNextAllocationBlock(size_t requested);
995 // Compares the start addresses of two free blocks.
996 static int CompareFreeBlockAddress(const FreeBlock* left,
997 const FreeBlock* right);
998
999 DISALLOW_COPY_AND_ASSIGN(CodeRange);
1000 };
1001
1002
1003 class SkipList {
1004 public:
1005 SkipList() {
1006 Clear();
1007 }
1008
1009 void Clear() {
1010 for (int idx = 0; idx < kSize; idx++) {
1011 starts_[idx] = reinterpret_cast<Address>(-1);
1012 }
1013 }
1014
1015 Address StartFor(Address addr) {
1016 return starts_[RegionNumber(addr)];
1017 }
1018
1019 void AddObject(Address addr, int size) {
1020 int start_region = RegionNumber(addr);
1021 int end_region = RegionNumber(addr + size - kPointerSize);
1022 for (int idx = start_region; idx <= end_region; idx++) {
1023 if (starts_[idx] > addr) starts_[idx] = addr;
1024 }
1025 }
1026
1027 static inline int RegionNumber(Address addr) {
1028 return (OffsetFrom(addr) & Page::kPageAlignmentMask) >> kRegionSizeLog2;
1029 }
1030
1031 static void Update(Address addr, int size) {
1032 Page* page = Page::FromAddress(addr);
1033 SkipList* list = page->skip_list();
1034 if (list == NULL) {
1035 list = new SkipList();
1036 page->set_skip_list(list);
1037 }
1038
1039 list->AddObject(addr, size);
1040 }
1041
1042 private:
1043 static const int kRegionSizeLog2 = 13;
1044 static const int kRegionSize = 1 << kRegionSizeLog2;
1045 static const int kSize = Page::kPageSize / kRegionSize;
1046
1047 STATIC_ASSERT(Page::kPageSize % kRegionSize == 0);
1048
1049 Address starts_[kSize];
1050 };
1051
1052
1053 // ----------------------------------------------------------------------------
1054 // A space acquires chunks of memory from the operating system. The memory
1055 // allocator allocated and deallocates pages for the paged heap spaces and large
1056 // pages for large object space.
1057 //
1058 // Each space has to manage it's own pages.
1059 //
1060 class MemoryAllocator {
1061 public:
1062 explicit MemoryAllocator(Isolate* isolate);
1063
1064 // Initializes its internal bookkeeping structures.
1065 // Max capacity of the total space and executable memory limit.
1066 bool SetUp(intptr_t max_capacity, intptr_t capacity_executable);
1067
1068 void TearDown();
1069
1070 Page* AllocatePage(
1071 intptr_t size, PagedSpace* owner, Executability executable);
1072
1073 LargePage* AllocateLargePage(
1074 intptr_t object_size, Space* owner, Executability executable);
1075
1076 void Free(MemoryChunk* chunk);
1077
1078 // Returns the maximum available bytes of heaps.
1079 intptr_t Available() { return capacity_ < size_ ? 0 : capacity_ - size_; }
1080
1081 // Returns allocated spaces in bytes.
1082 intptr_t Size() { return size_; }
1083
1084 // Returns the maximum available executable bytes of heaps.
1085 intptr_t AvailableExecutable() {
1086 if (capacity_executable_ < size_executable_) return 0;
1087 return capacity_executable_ - size_executable_;
1088 }
1089
1090 // Returns allocated executable spaces in bytes.
1091 intptr_t SizeExecutable() { return size_executable_; }
1092
1093 // Returns maximum available bytes that the old space can have.
1094 intptr_t MaxAvailable() {
1095 return (Available() / Page::kPageSize) * Page::kMaxRegularHeapObjectSize;
1096 }
1097
1098 // Returns an indication of whether a pointer is in a space that has
1099 // been allocated by this MemoryAllocator.
1100 V8_INLINE bool IsOutsideAllocatedSpace(const void* address) const {
1101 return address < lowest_ever_allocated_ ||
1102 address >= highest_ever_allocated_;
1103 }
1104
1105 #ifdef DEBUG
1106 // Reports statistic info of the space.
1107 void ReportStatistics();
1108 #endif
1109
1110 // Returns a MemoryChunk in which the memory region from commit_area_size to
1111 // reserve_area_size of the chunk area is reserved but not committed, it
1112 // could be committed later by calling MemoryChunk::CommitArea.
1113 MemoryChunk* AllocateChunk(intptr_t reserve_area_size,
1114 intptr_t commit_area_size,
1115 Executability executable,
1116 Space* space);
1117
1118 Address ReserveAlignedMemory(size_t requested,
1119 size_t alignment,
1120 base::VirtualMemory* controller);
1121 Address AllocateAlignedMemory(size_t reserve_size,
1122 size_t commit_size,
1123 size_t alignment,
1124 Executability executable,
1125 base::VirtualMemory* controller);
1126
1127 bool CommitMemory(Address addr, size_t size, Executability executable);
1128
1129 void FreeMemory(base::VirtualMemory* reservation, Executability executable);
1130 void FreeMemory(Address addr, size_t size, Executability executable);
1131
1132 // Commit a contiguous block of memory from the initial chunk. Assumes that
1133 // the address is not NULL, the size is greater than zero, and that the
1134 // block is contained in the initial chunk. Returns true if it succeeded
1135 // and false otherwise.
1136 bool CommitBlock(Address start, size_t size, Executability executable);
1137
1138 // Uncommit a contiguous block of memory [start..(start+size)[.
1139 // start is not NULL, the size is greater than zero, and the
1140 // block is contained in the initial chunk. Returns true if it succeeded
1141 // and false otherwise.
1142 bool UncommitBlock(Address start, size_t size);
1143
1144 // Zaps a contiguous block of memory [start..(start+size)[ thus
1145 // filling it up with a recognizable non-NULL bit pattern.
1146 void ZapBlock(Address start, size_t size);
1147
1148 void PerformAllocationCallback(ObjectSpace space,
1149 AllocationAction action,
1150 size_t size);
1151
1152 void AddMemoryAllocationCallback(MemoryAllocationCallback callback,
1153 ObjectSpace space,
1154 AllocationAction action);
1155
1156 void RemoveMemoryAllocationCallback(
1157 MemoryAllocationCallback callback);
1158
1159 bool MemoryAllocationCallbackRegistered(
1160 MemoryAllocationCallback callback);
1161
1162 static int CodePageGuardStartOffset();
1163
1164 static int CodePageGuardSize();
1165
1166 static int CodePageAreaStartOffset();
1167
1168 static int CodePageAreaEndOffset();
1169
1170 static int CodePageAreaSize() {
1171 return CodePageAreaEndOffset() - CodePageAreaStartOffset();
1172 }
1173
1174 MUST_USE_RESULT bool CommitExecutableMemory(base::VirtualMemory* vm,
1175 Address start,
1176 size_t commit_size,
1177 size_t reserved_size);
1178
1179 private:
1180 Isolate* isolate_;
1181
1182 // Maximum space size in bytes.
1183 size_t capacity_;
1184 // Maximum subset of capacity_ that can be executable
1185 size_t capacity_executable_;
1186
1187 // Allocated space size in bytes.
1188 size_t size_;
1189 // Allocated executable space size in bytes.
1190 size_t size_executable_;
1191
1192 // We keep the lowest and highest addresses allocated as a quick way
1193 // of determining that pointers are outside the heap. The estimate is
1194 // conservative, i.e. not all addrsses in 'allocated' space are allocated
1195 // to our heap. The range is [lowest, highest[, inclusive on the low end
1196 // and exclusive on the high end.
1197 void* lowest_ever_allocated_;
1198 void* highest_ever_allocated_;
1199
1200 struct MemoryAllocationCallbackRegistration {
1201 MemoryAllocationCallbackRegistration(MemoryAllocationCallback callback,
1202 ObjectSpace space,
1203 AllocationAction action)
1204 : callback(callback), space(space), action(action) {
1205 }
1206 MemoryAllocationCallback callback;
1207 ObjectSpace space;
1208 AllocationAction action;
1209 };
1210
1211 // A List of callback that are triggered when memory is allocated or free'd
1212 List<MemoryAllocationCallbackRegistration>
1213 memory_allocation_callbacks_;
1214
1215 // Initializes pages in a chunk. Returns the first page address.
1216 // This function and GetChunkId() are provided for the mark-compact
1217 // collector to rebuild page headers in the from space, which is
1218 // used as a marking stack and its page headers are destroyed.
1219 Page* InitializePagesInChunk(int chunk_id, int pages_in_chunk,
1220 PagedSpace* owner);
1221
1222 void UpdateAllocatedSpaceLimits(void* low, void* high) {
1223 lowest_ever_allocated_ = Min(lowest_ever_allocated_, low);
1224 highest_ever_allocated_ = Max(highest_ever_allocated_, high);
1225 }
1226
1227 DISALLOW_IMPLICIT_CONSTRUCTORS(MemoryAllocator);
1228 };
1229
1230
1231 // -----------------------------------------------------------------------------
1232 // Interface for heap object iterator to be implemented by all object space
1233 // object iterators.
1234 //
1235 // NOTE: The space specific object iterators also implements the own next()
1236 // method which is used to avoid using virtual functions
1237 // iterating a specific space.
1238
1239 class ObjectIterator : public Malloced {
1240 public:
1241 virtual ~ObjectIterator() { }
1242
1243 virtual HeapObject* next_object() = 0;
1244 };
1245
1246
1247 // -----------------------------------------------------------------------------
1248 // Heap object iterator in new/old/map spaces.
1249 //
1250 // A HeapObjectIterator iterates objects from the bottom of the given space
1251 // to its top or from the bottom of the given page to its top.
1252 //
1253 // If objects are allocated in the page during iteration the iterator may
1254 // or may not iterate over those objects. The caller must create a new
1255 // iterator in order to be sure to visit these new objects.
1256 class HeapObjectIterator: public ObjectIterator {
1257 public:
1258 // Creates a new object iterator in a given space.
1259 // If the size function is not given, the iterator calls the default
1260 // Object::Size().
1261 explicit HeapObjectIterator(PagedSpace* space);
1262 HeapObjectIterator(PagedSpace* space, HeapObjectCallback size_func);
1263 HeapObjectIterator(Page* page, HeapObjectCallback size_func);
1264
1265 // Advance to the next object, skipping free spaces and other fillers and
1266 // skipping the special garbage section of which there is one per space.
1267 // Returns NULL when the iteration has ended.
1268 inline HeapObject* Next() {
1269 do {
1270 HeapObject* next_obj = FromCurrentPage();
1271 if (next_obj != NULL) return next_obj;
1272 } while (AdvanceToNextPage());
1273 return NULL;
1274 }
1275
1276 virtual HeapObject* next_object() {
1277 return Next();
1278 }
1279
1280 private:
1281 enum PageMode { kOnePageOnly, kAllPagesInSpace };
1282
1283 Address cur_addr_; // Current iteration point.
1284 Address cur_end_; // End iteration point.
1285 HeapObjectCallback size_func_; // Size function or NULL.
1286 PagedSpace* space_;
1287 PageMode page_mode_;
1288
1289 // Fast (inlined) path of next().
1290 inline HeapObject* FromCurrentPage();
1291
1292 // Slow path of next(), goes into the next page. Returns false if the
1293 // iteration has ended.
1294 bool AdvanceToNextPage();
1295
1296 // Initializes fields.
1297 inline void Initialize(PagedSpace* owner,
1298 Address start,
1299 Address end,
1300 PageMode mode,
1301 HeapObjectCallback size_func);
1302 };
1303
1304
1305 // -----------------------------------------------------------------------------
1306 // A PageIterator iterates the pages in a paged space.
1307
1308 class PageIterator BASE_EMBEDDED {
1309 public:
1310 explicit inline PageIterator(PagedSpace* space);
1311
1312 inline bool has_next();
1313 inline Page* next();
1314
1315 private:
1316 PagedSpace* space_;
1317 Page* prev_page_; // Previous page returned.
1318 // Next page that will be returned. Cached here so that we can use this
1319 // iterator for operations that deallocate pages.
1320 Page* next_page_;
1321 };
1322
1323
1324 // -----------------------------------------------------------------------------
1325 // A space has a circular list of pages. The next page can be accessed via
1326 // Page::next_page() call.
1327
1328 // An abstraction of allocation and relocation pointers in a page-structured
1329 // space.
1330 class AllocationInfo {
1331 public:
1332 AllocationInfo() : top_(NULL), limit_(NULL) {
1333 }
1334
1335 INLINE(void set_top(Address top)) {
1336 SLOW_DCHECK(top == NULL ||
1337 (reinterpret_cast<intptr_t>(top) & HeapObjectTagMask()) == 0);
1338 top_ = top;
1339 }
1340
1341 INLINE(Address top()) const {
1342 SLOW_DCHECK(top_ == NULL ||
1343 (reinterpret_cast<intptr_t>(top_) & HeapObjectTagMask()) == 0);
1344 return top_;
1345 }
1346
1347 Address* top_address() {
1348 return &top_;
1349 }
1350
1351 INLINE(void set_limit(Address limit)) {
1352 SLOW_DCHECK(limit == NULL ||
1353 (reinterpret_cast<intptr_t>(limit) & HeapObjectTagMask()) == 0);
1354 limit_ = limit;
1355 }
1356
1357 INLINE(Address limit()) const {
1358 SLOW_DCHECK(limit_ == NULL ||
1359 (reinterpret_cast<intptr_t>(limit_) & HeapObjectTagMask()) == 0);
1360 return limit_;
1361 }
1362
1363 Address* limit_address() {
1364 return &limit_;
1365 }
1366
1367 #ifdef DEBUG
1368 bool VerifyPagedAllocation() {
1369 return (Page::FromAllocationTop(top_) == Page::FromAllocationTop(limit_))
1370 && (top_ <= limit_);
1371 }
1372 #endif
1373
1374 private:
1375 // Current allocation top.
1376 Address top_;
1377 // Current allocation limit.
1378 Address limit_;
1379 };
1380
1381
1382 // An abstraction of the accounting statistics of a page-structured space.
1383 // The 'capacity' of a space is the number of object-area bytes (i.e., not
1384 // including page bookkeeping structures) currently in the space. The 'size'
1385 // of a space is the number of allocated bytes, the 'waste' in the space is
1386 // the number of bytes that are not allocated and not available to
1387 // allocation without reorganizing the space via a GC (e.g. small blocks due
1388 // to internal fragmentation, top of page areas in map space), and the bytes
1389 // 'available' is the number of unallocated bytes that are not waste. The
1390 // capacity is the sum of size, waste, and available.
1391 //
1392 // The stats are only set by functions that ensure they stay balanced. These
1393 // functions increase or decrease one of the non-capacity stats in
1394 // conjunction with capacity, or else they always balance increases and
1395 // decreases to the non-capacity stats.
1396 class AllocationStats BASE_EMBEDDED {
1397 public:
1398 AllocationStats() { Clear(); }
1399
1400 // Zero out all the allocation statistics (i.e., no capacity).
1401 void Clear() {
1402 capacity_ = 0;
1403 max_capacity_ = 0;
1404 size_ = 0;
1405 waste_ = 0;
1406 }
1407
1408 void ClearSizeWaste() {
1409 size_ = capacity_;
1410 waste_ = 0;
1411 }
1412
1413 // Reset the allocation statistics (i.e., available = capacity with no
1414 // wasted or allocated bytes).
1415 void Reset() {
1416 size_ = 0;
1417 waste_ = 0;
1418 }
1419
1420 // Accessors for the allocation statistics.
1421 intptr_t Capacity() { return capacity_; }
1422 intptr_t MaxCapacity() { return max_capacity_; }
1423 intptr_t Size() { return size_; }
1424 intptr_t Waste() { return waste_; }
1425
1426 // Grow the space by adding available bytes. They are initially marked as
1427 // being in use (part of the size), but will normally be immediately freed,
1428 // putting them on the free list and removing them from size_.
1429 void ExpandSpace(int size_in_bytes) {
1430 capacity_ += size_in_bytes;
1431 size_ += size_in_bytes;
1432 if (capacity_ > max_capacity_) {
1433 max_capacity_ = capacity_;
1434 }
1435 DCHECK(size_ >= 0);
1436 }
1437
1438 // Shrink the space by removing available bytes. Since shrinking is done
1439 // during sweeping, bytes have been marked as being in use (part of the size)
1440 // and are hereby freed.
1441 void ShrinkSpace(int size_in_bytes) {
1442 capacity_ -= size_in_bytes;
1443 size_ -= size_in_bytes;
1444 DCHECK(size_ >= 0);
1445 }
1446
1447 // Allocate from available bytes (available -> size).
1448 void AllocateBytes(intptr_t size_in_bytes) {
1449 size_ += size_in_bytes;
1450 DCHECK(size_ >= 0);
1451 }
1452
1453 // Free allocated bytes, making them available (size -> available).
1454 void DeallocateBytes(intptr_t size_in_bytes) {
1455 size_ -= size_in_bytes;
1456 DCHECK(size_ >= 0);
1457 }
1458
1459 // Waste free bytes (available -> waste).
1460 void WasteBytes(int size_in_bytes) {
1461 DCHECK(size_in_bytes >= 0);
1462 waste_ += size_in_bytes;
1463 }
1464
1465 private:
1466 intptr_t capacity_;
1467 intptr_t max_capacity_;
1468 intptr_t size_;
1469 intptr_t waste_;
1470 };
1471
1472
1473 // -----------------------------------------------------------------------------
1474 // Free lists for old object spaces
1475 //
1476 // Free-list nodes are free blocks in the heap. They look like heap objects
1477 // (free-list node pointers have the heap object tag, and they have a map like
1478 // a heap object). They have a size and a next pointer. The next pointer is
1479 // the raw address of the next free list node (or NULL).
1480 class FreeListNode: public HeapObject {
1481 public:
1482 // Obtain a free-list node from a raw address. This is not a cast because
1483 // it does not check nor require that the first word at the address is a map
1484 // pointer.
1485 static FreeListNode* FromAddress(Address address) {
1486 return reinterpret_cast<FreeListNode*>(HeapObject::FromAddress(address));
1487 }
1488
1489 static inline bool IsFreeListNode(HeapObject* object);
1490
1491 // Set the size in bytes, which can be read with HeapObject::Size(). This
1492 // function also writes a map to the first word of the block so that it
1493 // looks like a heap object to the garbage collector and heap iteration
1494 // functions.
1495 void set_size(Heap* heap, int size_in_bytes);
1496
1497 // Accessors for the next field.
1498 inline FreeListNode* next();
1499 inline FreeListNode** next_address();
1500 inline void set_next(FreeListNode* next);
1501
1502 inline void Zap();
1503
1504 static inline FreeListNode* cast(Object* object) {
1505 return reinterpret_cast<FreeListNode*>(object);
1506 }
1507
1508 private:
1509 static const int kNextOffset = POINTER_SIZE_ALIGN(FreeSpace::kHeaderSize);
1510
1511 DISALLOW_IMPLICIT_CONSTRUCTORS(FreeListNode);
1512 };
1513
1514
1515 // The free list category holds a pointer to the top element and a pointer to
1516 // the end element of the linked list of free memory blocks.
1517 class FreeListCategory {
1518 public:
1519 FreeListCategory() :
1520 top_(0),
1521 end_(NULL),
1522 available_(0) {}
1523
1524 intptr_t Concatenate(FreeListCategory* category);
1525
1526 void Reset();
1527
1528 void Free(FreeListNode* node, int size_in_bytes);
1529
1530 FreeListNode* PickNodeFromList(int *node_size);
1531 FreeListNode* PickNodeFromList(int size_in_bytes, int *node_size);
1532
1533 intptr_t EvictFreeListItemsInList(Page* p);
1534 bool ContainsPageFreeListItemsInList(Page* p);
1535
1536 void RepairFreeList(Heap* heap);
1537
1538 FreeListNode* top() const {
1539 return reinterpret_cast<FreeListNode*>(base::NoBarrier_Load(&top_));
1540 }
1541
1542 void set_top(FreeListNode* top) {
1543 base::NoBarrier_Store(&top_, reinterpret_cast<base::AtomicWord>(top));
1544 }
1545
1546 FreeListNode** GetEndAddress() { return &end_; }
1547 FreeListNode* end() const { return end_; }
1548 void set_end(FreeListNode* end) { end_ = end; }
1549
1550 int* GetAvailableAddress() { return &available_; }
1551 int available() const { return available_; }
1552 void set_available(int available) { available_ = available; }
1553
1554 base::Mutex* mutex() { return &mutex_; }
1555
1556 bool IsEmpty() {
1557 return top() == 0;
1558 }
1559
1560 #ifdef DEBUG
1561 intptr_t SumFreeList();
1562 int FreeListLength();
1563 #endif
1564
1565 private:
1566 // top_ points to the top FreeListNode* in the free list category.
1567 base::AtomicWord top_;
1568 FreeListNode* end_;
1569 base::Mutex mutex_;
1570
1571 // Total available bytes in all blocks of this free list category.
1572 int available_;
1573 };
1574
1575
1576 // The free list for the old space. The free list is organized in such a way
1577 // as to encourage objects allocated around the same time to be near each
1578 // other. The normal way to allocate is intended to be by bumping a 'top'
1579 // pointer until it hits a 'limit' pointer. When the limit is hit we need to
1580 // find a new space to allocate from. This is done with the free list, which
1581 // is divided up into rough categories to cut down on waste. Having finer
1582 // categories would scatter allocation more.
1583
1584 // The old space free list is organized in categories.
1585 // 1-31 words: Such small free areas are discarded for efficiency reasons.
1586 // They can be reclaimed by the compactor. However the distance between top
1587 // and limit may be this small.
1588 // 32-255 words: There is a list of spaces this large. It is used for top and
1589 // limit when the object we need to allocate is 1-31 words in size. These
1590 // spaces are called small.
1591 // 256-2047 words: There is a list of spaces this large. It is used for top and
1592 // limit when the object we need to allocate is 32-255 words in size. These
1593 // spaces are called medium.
1594 // 1048-16383 words: There is a list of spaces this large. It is used for top
1595 // and limit when the object we need to allocate is 256-2047 words in size.
1596 // These spaces are call large.
1597 // At least 16384 words. This list is for objects of 2048 words or larger.
1598 // Empty pages are added to this list. These spaces are called huge.
1599 class FreeList {
1600 public:
1601 explicit FreeList(PagedSpace* owner);
1602
1603 intptr_t Concatenate(FreeList* free_list);
1604
1605 // Clear the free list.
1606 void Reset();
1607
1608 // Return the number of bytes available on the free list.
1609 intptr_t available() {
1610 return small_list_.available() + medium_list_.available() +
1611 large_list_.available() + huge_list_.available();
1612 }
1613
1614 // Place a node on the free list. The block of size 'size_in_bytes'
1615 // starting at 'start' is placed on the free list. The return value is the
1616 // number of bytes that have been lost due to internal fragmentation by
1617 // freeing the block. Bookkeeping information will be written to the block,
1618 // i.e., its contents will be destroyed. The start address should be word
1619 // aligned, and the size should be a non-zero multiple of the word size.
1620 int Free(Address start, int size_in_bytes);
1621
1622 // This method returns how much memory can be allocated after freeing
1623 // maximum_freed memory.
1624 static inline int GuaranteedAllocatable(int maximum_freed) {
1625 if (maximum_freed < kSmallListMin) {
1626 return 0;
1627 } else if (maximum_freed <= kSmallListMax) {
1628 return kSmallAllocationMax;
1629 } else if (maximum_freed <= kMediumListMax) {
1630 return kMediumAllocationMax;
1631 } else if (maximum_freed <= kLargeListMax) {
1632 return kLargeAllocationMax;
1633 }
1634 return maximum_freed;
1635 }
1636
1637 // Allocate a block of size 'size_in_bytes' from the free list. The block
1638 // is unitialized. A failure is returned if no block is available. The
1639 // number of bytes lost to fragmentation is returned in the output parameter
1640 // 'wasted_bytes'. The size should be a non-zero multiple of the word size.
1641 MUST_USE_RESULT HeapObject* Allocate(int size_in_bytes);
1642
1643 bool IsEmpty() {
1644 return small_list_.IsEmpty() && medium_list_.IsEmpty() &&
1645 large_list_.IsEmpty() && huge_list_.IsEmpty();
1646 }
1647
1648 #ifdef DEBUG
1649 void Zap();
1650 intptr_t SumFreeLists();
1651 bool IsVeryLong();
1652 #endif
1653
1654 // Used after booting the VM.
1655 void RepairLists(Heap* heap);
1656
1657 intptr_t EvictFreeListItems(Page* p);
1658 bool ContainsPageFreeListItems(Page* p);
1659
1660 FreeListCategory* small_list() { return &small_list_; }
1661 FreeListCategory* medium_list() { return &medium_list_; }
1662 FreeListCategory* large_list() { return &large_list_; }
1663 FreeListCategory* huge_list() { return &huge_list_; }
1664
1665 private:
1666 // The size range of blocks, in bytes.
1667 static const int kMinBlockSize = 3 * kPointerSize;
1668 static const int kMaxBlockSize = Page::kMaxRegularHeapObjectSize;
1669
1670 FreeListNode* FindNodeFor(int size_in_bytes, int* node_size);
1671
1672 PagedSpace* owner_;
1673 Heap* heap_;
1674
1675 static const int kSmallListMin = 0x20 * kPointerSize;
1676 static const int kSmallListMax = 0xff * kPointerSize;
1677 static const int kMediumListMax = 0x7ff * kPointerSize;
1678 static const int kLargeListMax = 0x3fff * kPointerSize;
1679 static const int kSmallAllocationMax = kSmallListMin - kPointerSize;
1680 static const int kMediumAllocationMax = kSmallListMax;
1681 static const int kLargeAllocationMax = kMediumListMax;
1682 FreeListCategory small_list_;
1683 FreeListCategory medium_list_;
1684 FreeListCategory large_list_;
1685 FreeListCategory huge_list_;
1686
1687 DISALLOW_IMPLICIT_CONSTRUCTORS(FreeList);
1688 };
1689
1690
1691 class AllocationResult {
1692 public:
1693 // Implicit constructor from Object*.
1694 AllocationResult(Object* object) : object_(object), // NOLINT
1695 retry_space_(INVALID_SPACE) { }
1696
1697 AllocationResult() : object_(NULL),
1698 retry_space_(INVALID_SPACE) { }
1699
1700 static inline AllocationResult Retry(AllocationSpace space = NEW_SPACE) {
1701 return AllocationResult(space);
1702 }
1703
1704 inline bool IsRetry() { return retry_space_ != INVALID_SPACE; }
1705
1706 template <typename T>
1707 bool To(T** obj) {
1708 if (IsRetry()) return false;
1709 *obj = T::cast(object_);
1710 return true;
1711 }
1712
1713 Object* ToObjectChecked() {
1714 CHECK(!IsRetry());
1715 return object_;
1716 }
1717
1718 AllocationSpace RetrySpace() {
1719 DCHECK(IsRetry());
1720 return retry_space_;
1721 }
1722
1723 private:
1724 explicit AllocationResult(AllocationSpace space) : object_(NULL),
1725 retry_space_(space) { }
1726
1727 Object* object_;
1728 AllocationSpace retry_space_;
1729 };
1730
1731
1732 class PagedSpace : public Space {
1733 public:
1734 // Creates a space with a maximum capacity, and an id.
1735 PagedSpace(Heap* heap,
1736 intptr_t max_capacity,
1737 AllocationSpace id,
1738 Executability executable);
1739
1740 virtual ~PagedSpace() {}
1741
1742 // Set up the space using the given address range of virtual memory (from
1743 // the memory allocator's initial chunk) if possible. If the block of
1744 // addresses is not big enough to contain a single page-aligned page, a
1745 // fresh chunk will be allocated.
1746 bool SetUp();
1747
1748 // Returns true if the space has been successfully set up and not
1749 // subsequently torn down.
1750 bool HasBeenSetUp();
1751
1752 // Cleans up the space, frees all pages in this space except those belonging
1753 // to the initial chunk, uncommits addresses in the initial chunk.
1754 void TearDown();
1755
1756 // Checks whether an object/address is in this space.
1757 inline bool Contains(Address a);
1758 bool Contains(HeapObject* o) { return Contains(o->address()); }
1759
1760 // Given an address occupied by a live object, return that object if it is
1761 // in this space, or a Smi if it is not. The implementation iterates over
1762 // objects in the page containing the address, the cost is linear in the
1763 // number of objects in the page. It may be slow.
1764 Object* FindObject(Address addr);
1765
1766 // During boot the free_space_map is created, and afterwards we may need
1767 // to write it into the free list nodes that were already created.
1768 void RepairFreeListsAfterBoot();
1769
1770 // Prepares for a mark-compact GC.
1771 void PrepareForMarkCompact();
1772
1773 // Current capacity without growing (Size() + Available()).
1774 intptr_t Capacity() { return accounting_stats_.Capacity(); }
1775
1776 // Total amount of memory committed for this space. For paged
1777 // spaces this equals the capacity.
1778 intptr_t CommittedMemory() { return Capacity(); }
1779
1780 // The maximum amount of memory ever committed for this space.
1781 intptr_t MaximumCommittedMemory() { return accounting_stats_.MaxCapacity(); }
1782
1783 // Approximate amount of physical memory committed for this space.
1784 size_t CommittedPhysicalMemory();
1785
1786 struct SizeStats {
1787 intptr_t Total() {
1788 return small_size_ + medium_size_ + large_size_ + huge_size_;
1789 }
1790
1791 intptr_t small_size_;
1792 intptr_t medium_size_;
1793 intptr_t large_size_;
1794 intptr_t huge_size_;
1795 };
1796
1797 void ObtainFreeListStatistics(Page* p, SizeStats* sizes);
1798 void ResetFreeListStatistics();
1799
1800 // Sets the capacity, the available space and the wasted space to zero.
1801 // The stats are rebuilt during sweeping by adding each page to the
1802 // capacity and the size when it is encountered. As free spaces are
1803 // discovered during the sweeping they are subtracted from the size and added
1804 // to the available and wasted totals.
1805 void ClearStats() {
1806 accounting_stats_.ClearSizeWaste();
1807 ResetFreeListStatistics();
1808 }
1809
1810 // Increases the number of available bytes of that space.
1811 void AddToAccountingStats(intptr_t bytes) {
1812 accounting_stats_.DeallocateBytes(bytes);
1813 }
1814
1815 // Available bytes without growing. These are the bytes on the free list.
1816 // The bytes in the linear allocation area are not included in this total
1817 // because updating the stats would slow down allocation. New pages are
1818 // immediately added to the free list so they show up here.
1819 intptr_t Available() { return free_list_.available(); }
1820
1821 // Allocated bytes in this space. Garbage bytes that were not found due to
1822 // concurrent sweeping are counted as being allocated! The bytes in the
1823 // current linear allocation area (between top and limit) are also counted
1824 // here.
1825 virtual intptr_t Size() { return accounting_stats_.Size(); }
1826
1827 // As size, but the bytes in lazily swept pages are estimated and the bytes
1828 // in the current linear allocation area are not included.
1829 virtual intptr_t SizeOfObjects();
1830
1831 // Wasted bytes in this space. These are just the bytes that were thrown away
1832 // due to being too small to use for allocation. They do not include the
1833 // free bytes that were not found at all due to lazy sweeping.
1834 virtual intptr_t Waste() { return accounting_stats_.Waste(); }
1835
1836 // Returns the allocation pointer in this space.
1837 Address top() { return allocation_info_.top(); }
1838 Address limit() { return allocation_info_.limit(); }
1839
1840 // The allocation top address.
1841 Address* allocation_top_address() {
1842 return allocation_info_.top_address();
1843 }
1844
1845 // The allocation limit address.
1846 Address* allocation_limit_address() {
1847 return allocation_info_.limit_address();
1848 }
1849
1850 // Allocate the requested number of bytes in the space if possible, return a
1851 // failure object if not.
1852 MUST_USE_RESULT inline AllocationResult AllocateRaw(int size_in_bytes);
1853
1854 // Give a block of memory to the space's free list. It might be added to
1855 // the free list or accounted as waste.
1856 // If add_to_freelist is false then just accounting stats are updated and
1857 // no attempt to add area to free list is made.
1858 int Free(Address start, int size_in_bytes) {
1859 int wasted = free_list_.Free(start, size_in_bytes);
1860 accounting_stats_.DeallocateBytes(size_in_bytes);
1861 accounting_stats_.WasteBytes(wasted);
1862 return size_in_bytes - wasted;
1863 }
1864
1865 void ResetFreeList() {
1866 free_list_.Reset();
1867 }
1868
1869 // Set space allocation info.
1870 void SetTopAndLimit(Address top, Address limit) {
1871 DCHECK(top == limit ||
1872 Page::FromAddress(top) == Page::FromAddress(limit - 1));
1873 MemoryChunk::UpdateHighWaterMark(allocation_info_.top());
1874 allocation_info_.set_top(top);
1875 allocation_info_.set_limit(limit);
1876 }
1877
1878 // Empty space allocation info, returning unused area to free list.
1879 void EmptyAllocationInfo() {
1880 // Mark the old linear allocation area with a free space map so it can be
1881 // skipped when scanning the heap.
1882 int old_linear_size = static_cast<int>(limit() - top());
1883 Free(top(), old_linear_size);
1884 SetTopAndLimit(NULL, NULL);
1885 }
1886
1887 void Allocate(int bytes) {
1888 accounting_stats_.AllocateBytes(bytes);
1889 }
1890
1891 void IncreaseCapacity(int size);
1892
1893 // Releases an unused page and shrinks the space.
1894 void ReleasePage(Page* page);
1895
1896 // The dummy page that anchors the linked list of pages.
1897 Page* anchor() { return &anchor_; }
1898
1899 #ifdef VERIFY_HEAP
1900 // Verify integrity of this space.
1901 virtual void Verify(ObjectVisitor* visitor);
1902
1903 // Overridden by subclasses to verify space-specific object
1904 // properties (e.g., only maps or free-list nodes are in map space).
1905 virtual void VerifyObject(HeapObject* obj) {}
1906 #endif
1907
1908 #ifdef DEBUG
1909 // Print meta info and objects in this space.
1910 virtual void Print();
1911
1912 // Reports statistics for the space
1913 void ReportStatistics();
1914
1915 // Report code object related statistics
1916 void CollectCodeStatistics();
1917 static void ReportCodeStatistics(Isolate* isolate);
1918 static void ResetCodeStatistics(Isolate* isolate);
1919 #endif
1920
1921 bool swept_precisely() { return swept_precisely_; }
1922 void set_swept_precisely(bool b) { swept_precisely_ = b; }
1923
1924 // Evacuation candidates are swept by evacuator. Needs to return a valid
1925 // result before _and_ after evacuation has finished.
1926 static bool ShouldBeSweptBySweeperThreads(Page* p) {
1927 return !p->IsEvacuationCandidate() &&
1928 !p->IsFlagSet(Page::RESCAN_ON_EVACUATION) &&
1929 !p->WasSweptPrecisely();
1930 }
1931
1932 void IncrementUnsweptFreeBytes(intptr_t by) {
1933 unswept_free_bytes_ += by;
1934 }
1935
1936 void IncreaseUnsweptFreeBytes(Page* p) {
1937 DCHECK(ShouldBeSweptBySweeperThreads(p));
1938 unswept_free_bytes_ += (p->area_size() - p->LiveBytes());
1939 }
1940
1941 void DecrementUnsweptFreeBytes(intptr_t by) {
1942 unswept_free_bytes_ -= by;
1943 }
1944
1945 void DecreaseUnsweptFreeBytes(Page* p) {
1946 DCHECK(ShouldBeSweptBySweeperThreads(p));
1947 unswept_free_bytes_ -= (p->area_size() - p->LiveBytes());
1948 }
1949
1950 void ResetUnsweptFreeBytes() {
1951 unswept_free_bytes_ = 0;
1952 }
1953
1954 // This function tries to steal size_in_bytes memory from the sweeper threads
1955 // free-lists. If it does not succeed stealing enough memory, it will wait
1956 // for the sweeper threads to finish sweeping.
1957 // It returns true when sweeping is completed and false otherwise.
1958 bool EnsureSweeperProgress(intptr_t size_in_bytes);
1959
1960 void set_end_of_unswept_pages(Page* page) {
1961 end_of_unswept_pages_ = page;
1962 }
1963
1964 Page* end_of_unswept_pages() {
1965 return end_of_unswept_pages_;
1966 }
1967
1968 Page* FirstPage() { return anchor_.next_page(); }
1969 Page* LastPage() { return anchor_.prev_page(); }
1970
1971 void EvictEvacuationCandidatesFromFreeLists();
1972
1973 bool CanExpand();
1974
1975 // Returns the number of total pages in this space.
1976 int CountTotalPages();
1977
1978 // Return size of allocatable area on a page in this space.
1979 inline int AreaSize() {
1980 return area_size_;
1981 }
1982
1983 void CreateEmergencyMemory();
1984 void FreeEmergencyMemory();
1985 void UseEmergencyMemory();
1986
1987 bool HasEmergencyMemory() { return emergency_memory_ != NULL; }
1988
1989 protected:
1990 FreeList* free_list() { return &free_list_; }
1991
1992 int area_size_;
1993
1994 // Maximum capacity of this space.
1995 intptr_t max_capacity_;
1996
1997 intptr_t SizeOfFirstPage();
1998
1999 // Accounting information for this space.
2000 AllocationStats accounting_stats_;
2001
2002 // The dummy page that anchors the double linked list of pages.
2003 Page anchor_;
2004
2005 // The space's free list.
2006 FreeList free_list_;
2007
2008 // Normal allocation information.
2009 AllocationInfo allocation_info_;
2010
2011 // This space was swept precisely, hence it is iterable.
2012 bool swept_precisely_;
2013
2014 // The number of free bytes which could be reclaimed by advancing the
2015 // concurrent sweeper threads. This is only an estimation because concurrent
2016 // sweeping is done conservatively.
2017 intptr_t unswept_free_bytes_;
2018
2019 // The sweeper threads iterate over the list of pointer and data space pages
2020 // and sweep these pages concurrently. They will stop sweeping after the
2021 // end_of_unswept_pages_ page.
2022 Page* end_of_unswept_pages_;
2023
2024 // Emergency memory is the memory of a full page for a given space, allocated
2025 // conservatively before evacuating a page. If compaction fails due to out
2026 // of memory error the emergency memory can be used to complete compaction.
2027 // If not used, the emergency memory is released after compaction.
2028 MemoryChunk* emergency_memory_;
2029
2030 // Expands the space by allocating a fixed number of pages. Returns false if
2031 // it cannot allocate requested number of pages from OS, or if the hard heap
2032 // size limit has been hit.
2033 bool Expand();
2034
2035 // Generic fast case allocation function that tries linear allocation at the
2036 // address denoted by top in allocation_info_.
2037 inline HeapObject* AllocateLinearly(int size_in_bytes);
2038
2039 // If sweeping is still in progress try to sweep unswept pages. If that is
2040 // not successful, wait for the sweeper threads and re-try free-list
2041 // allocation.
2042 MUST_USE_RESULT HeapObject* WaitForSweeperThreadsAndRetryAllocation(
2043 int size_in_bytes);
2044
2045 // Slow path of AllocateRaw. This function is space-dependent.
2046 MUST_USE_RESULT HeapObject* SlowAllocateRaw(int size_in_bytes);
2047
2048 friend class PageIterator;
2049 friend class MarkCompactCollector;
2050 };
2051
2052
2053 class NumberAndSizeInfo BASE_EMBEDDED {
2054 public:
2055 NumberAndSizeInfo() : number_(0), bytes_(0) {}
2056
2057 int number() const { return number_; }
2058 void increment_number(int num) { number_ += num; }
2059
2060 int bytes() const { return bytes_; }
2061 void increment_bytes(int size) { bytes_ += size; }
2062
2063 void clear() {
2064 number_ = 0;
2065 bytes_ = 0;
2066 }
2067
2068 private:
2069 int number_;
2070 int bytes_;
2071 };
2072
2073
2074 // HistogramInfo class for recording a single "bar" of a histogram. This
2075 // class is used for collecting statistics to print to the log file.
2076 class HistogramInfo: public NumberAndSizeInfo {
2077 public:
2078 HistogramInfo() : NumberAndSizeInfo() {}
2079
2080 const char* name() { return name_; }
2081 void set_name(const char* name) { name_ = name; }
2082
2083 private:
2084 const char* name_;
2085 };
2086
2087
2088 enum SemiSpaceId {
2089 kFromSpace = 0,
2090 kToSpace = 1
2091 };
2092
2093
2094 class SemiSpace;
2095
2096
2097 class NewSpacePage : public MemoryChunk {
2098 public:
2099 // GC related flags copied from from-space to to-space when
2100 // flipping semispaces.
2101 static const intptr_t kCopyOnFlipFlagsMask =
2102 (1 << MemoryChunk::POINTERS_TO_HERE_ARE_INTERESTING) |
2103 (1 << MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING) |
2104 (1 << MemoryChunk::SCAN_ON_SCAVENGE);
2105
2106 static const int kAreaSize = Page::kMaxRegularHeapObjectSize;
2107
2108 inline NewSpacePage* next_page() const {
2109 return static_cast<NewSpacePage*>(next_chunk());
2110 }
2111
2112 inline void set_next_page(NewSpacePage* page) {
2113 set_next_chunk(page);
2114 }
2115
2116 inline NewSpacePage* prev_page() const {
2117 return static_cast<NewSpacePage*>(prev_chunk());
2118 }
2119
2120 inline void set_prev_page(NewSpacePage* page) {
2121 set_prev_chunk(page);
2122 }
2123
2124 SemiSpace* semi_space() {
2125 return reinterpret_cast<SemiSpace*>(owner());
2126 }
2127
2128 bool is_anchor() { return !this->InNewSpace(); }
2129
2130 static bool IsAtStart(Address addr) {
2131 return (reinterpret_cast<intptr_t>(addr) & Page::kPageAlignmentMask)
2132 == kObjectStartOffset;
2133 }
2134
2135 static bool IsAtEnd(Address addr) {
2136 return (reinterpret_cast<intptr_t>(addr) & Page::kPageAlignmentMask) == 0;
2137 }
2138
2139 Address address() {
2140 return reinterpret_cast<Address>(this);
2141 }
2142
2143 // Finds the NewSpacePage containg the given address.
2144 static inline NewSpacePage* FromAddress(Address address_in_page) {
2145 Address page_start =
2146 reinterpret_cast<Address>(reinterpret_cast<uintptr_t>(address_in_page) &
2147 ~Page::kPageAlignmentMask);
2148 NewSpacePage* page = reinterpret_cast<NewSpacePage*>(page_start);
2149 return page;
2150 }
2151
2152 // Find the page for a limit address. A limit address is either an address
2153 // inside a page, or the address right after the last byte of a page.
2154 static inline NewSpacePage* FromLimit(Address address_limit) {
2155 return NewSpacePage::FromAddress(address_limit - 1);
2156 }
2157
2158 // Checks if address1 and address2 are on the same new space page.
2159 static inline bool OnSamePage(Address address1, Address address2) {
2160 return NewSpacePage::FromAddress(address1) ==
2161 NewSpacePage::FromAddress(address2);
2162 }
2163
2164 private:
2165 // Create a NewSpacePage object that is only used as anchor
2166 // for the doubly-linked list of real pages.
2167 explicit NewSpacePage(SemiSpace* owner) {
2168 InitializeAsAnchor(owner);
2169 }
2170
2171 static NewSpacePage* Initialize(Heap* heap,
2172 Address start,
2173 SemiSpace* semi_space);
2174
2175 // Intialize a fake NewSpacePage used as sentinel at the ends
2176 // of a doubly-linked list of real NewSpacePages.
2177 // Only uses the prev/next links, and sets flags to not be in new-space.
2178 void InitializeAsAnchor(SemiSpace* owner);
2179
2180 friend class SemiSpace;
2181 friend class SemiSpaceIterator;
2182 };
2183
2184
2185 // -----------------------------------------------------------------------------
2186 // SemiSpace in young generation
2187 //
2188 // A semispace is a contiguous chunk of memory holding page-like memory
2189 // chunks. The mark-compact collector uses the memory of the first page in
2190 // the from space as a marking stack when tracing live objects.
2191
2192 class SemiSpace : public Space {
2193 public:
2194 // Constructor.
2195 SemiSpace(Heap* heap, SemiSpaceId semispace)
2196 : Space(heap, NEW_SPACE, NOT_EXECUTABLE),
2197 start_(NULL),
2198 age_mark_(NULL),
2199 id_(semispace),
2200 anchor_(this),
2201 current_page_(NULL) { }
2202
2203 // Sets up the semispace using the given chunk.
2204 void SetUp(Address start, int initial_capacity, int maximum_capacity);
2205
2206 // Tear down the space. Heap memory was not allocated by the space, so it
2207 // is not deallocated here.
2208 void TearDown();
2209
2210 // True if the space has been set up but not torn down.
2211 bool HasBeenSetUp() { return start_ != NULL; }
2212
2213 // Grow the semispace to the new capacity. The new capacity
2214 // requested must be larger than the current capacity and less than
2215 // the maximum capacity.
2216 bool GrowTo(int new_capacity);
2217
2218 // Shrinks the semispace to the new capacity. The new capacity
2219 // requested must be more than the amount of used memory in the
2220 // semispace and less than the current capacity.
2221 bool ShrinkTo(int new_capacity);
2222
2223 // Returns the start address of the first page of the space.
2224 Address space_start() {
2225 DCHECK(anchor_.next_page() != &anchor_);
2226 return anchor_.next_page()->area_start();
2227 }
2228
2229 // Returns the start address of the current page of the space.
2230 Address page_low() {
2231 return current_page_->area_start();
2232 }
2233
2234 // Returns one past the end address of the space.
2235 Address space_end() {
2236 return anchor_.prev_page()->area_end();
2237 }
2238
2239 // Returns one past the end address of the current page of the space.
2240 Address page_high() {
2241 return current_page_->area_end();
2242 }
2243
2244 bool AdvancePage() {
2245 NewSpacePage* next_page = current_page_->next_page();
2246 if (next_page == anchor()) return false;
2247 current_page_ = next_page;
2248 return true;
2249 }
2250
2251 // Resets the space to using the first page.
2252 void Reset();
2253
2254 // Age mark accessors.
2255 Address age_mark() { return age_mark_; }
2256 void set_age_mark(Address mark);
2257
2258 // True if the address is in the address range of this semispace (not
2259 // necessarily below the allocation pointer).
2260 bool Contains(Address a) {
2261 return (reinterpret_cast<uintptr_t>(a) & address_mask_)
2262 == reinterpret_cast<uintptr_t>(start_);
2263 }
2264
2265 // True if the object is a heap object in the address range of this
2266 // semispace (not necessarily below the allocation pointer).
2267 bool Contains(Object* o) {
2268 return (reinterpret_cast<uintptr_t>(o) & object_mask_) == object_expected_;
2269 }
2270
2271 // If we don't have these here then SemiSpace will be abstract. However
2272 // they should never be called.
2273 virtual intptr_t Size() {
2274 UNREACHABLE();
2275 return 0;
2276 }
2277
2278 bool is_committed() { return committed_; }
2279 bool Commit();
2280 bool Uncommit();
2281
2282 NewSpacePage* first_page() { return anchor_.next_page(); }
2283 NewSpacePage* current_page() { return current_page_; }
2284
2285 #ifdef VERIFY_HEAP
2286 virtual void Verify();
2287 #endif
2288
2289 #ifdef DEBUG
2290 virtual void Print();
2291 // Validate a range of of addresses in a SemiSpace.
2292 // The "from" address must be on a page prior to the "to" address,
2293 // in the linked page order, or it must be earlier on the same page.
2294 static void AssertValidRange(Address from, Address to);
2295 #else
2296 // Do nothing.
2297 inline static void AssertValidRange(Address from, Address to) {}
2298 #endif
2299
2300 // Returns the current capacity of the semi space.
2301 int Capacity() { return capacity_; }
2302
2303 // Returns the maximum capacity of the semi space.
2304 int MaximumCapacity() { return maximum_capacity_; }
2305
2306 // Returns the initial capacity of the semi space.
2307 int InitialCapacity() { return initial_capacity_; }
2308
2309 SemiSpaceId id() { return id_; }
2310
2311 static void Swap(SemiSpace* from, SemiSpace* to);
2312
2313 // Returns the maximum amount of memory ever committed by the semi space.
2314 size_t MaximumCommittedMemory() { return maximum_committed_; }
2315
2316 // Approximate amount of physical memory committed for this space.
2317 size_t CommittedPhysicalMemory();
2318
2319 private:
2320 // Flips the semispace between being from-space and to-space.
2321 // Copies the flags into the masked positions on all pages in the space.
2322 void FlipPages(intptr_t flags, intptr_t flag_mask);
2323
2324 // Updates Capacity and MaximumCommitted based on new capacity.
2325 void SetCapacity(int new_capacity);
2326
2327 NewSpacePage* anchor() { return &anchor_; }
2328
2329 // The current and maximum capacity of the space.
2330 int capacity_;
2331 int maximum_capacity_;
2332 int initial_capacity_;
2333
2334 intptr_t maximum_committed_;
2335
2336 // The start address of the space.
2337 Address start_;
2338 // Used to govern object promotion during mark-compact collection.
2339 Address age_mark_;
2340
2341 // Masks and comparison values to test for containment in this semispace.
2342 uintptr_t address_mask_;
2343 uintptr_t object_mask_;
2344 uintptr_t object_expected_;
2345
2346 bool committed_;
2347 SemiSpaceId id_;
2348
2349 NewSpacePage anchor_;
2350 NewSpacePage* current_page_;
2351
2352 friend class SemiSpaceIterator;
2353 friend class NewSpacePageIterator;
2354 public:
2355 TRACK_MEMORY("SemiSpace")
2356 };
2357
2358
2359 // A SemiSpaceIterator is an ObjectIterator that iterates over the active
2360 // semispace of the heap's new space. It iterates over the objects in the
2361 // semispace from a given start address (defaulting to the bottom of the
2362 // semispace) to the top of the semispace. New objects allocated after the
2363 // iterator is created are not iterated.
2364 class SemiSpaceIterator : public ObjectIterator {
2365 public:
2366 // Create an iterator over the objects in the given space. If no start
2367 // address is given, the iterator starts from the bottom of the space. If
2368 // no size function is given, the iterator calls Object::Size().
2369
2370 // Iterate over all of allocated to-space.
2371 explicit SemiSpaceIterator(NewSpace* space);
2372 // Iterate over all of allocated to-space, with a custome size function.
2373 SemiSpaceIterator(NewSpace* space, HeapObjectCallback size_func);
2374 // Iterate over part of allocated to-space, from start to the end
2375 // of allocation.
2376 SemiSpaceIterator(NewSpace* space, Address start);
2377 // Iterate from one address to another in the same semi-space.
2378 SemiSpaceIterator(Address from, Address to);
2379
2380 HeapObject* Next() {
2381 if (current_ == limit_) return NULL;
2382 if (NewSpacePage::IsAtEnd(current_)) {
2383 NewSpacePage* page = NewSpacePage::FromLimit(current_);
2384 page = page->next_page();
2385 DCHECK(!page->is_anchor());
2386 current_ = page->area_start();
2387 if (current_ == limit_) return NULL;
2388 }
2389
2390 HeapObject* object = HeapObject::FromAddress(current_);
2391 int size = (size_func_ == NULL) ? object->Size() : size_func_(object);
2392
2393 current_ += size;
2394 return object;
2395 }
2396
2397 // Implementation of the ObjectIterator functions.
2398 virtual HeapObject* next_object() { return Next(); }
2399
2400 private:
2401 void Initialize(Address start,
2402 Address end,
2403 HeapObjectCallback size_func);
2404
2405 // The current iteration point.
2406 Address current_;
2407 // The end of iteration.
2408 Address limit_;
2409 // The callback function.
2410 HeapObjectCallback size_func_;
2411 };
2412
2413
2414 // -----------------------------------------------------------------------------
2415 // A PageIterator iterates the pages in a semi-space.
2416 class NewSpacePageIterator BASE_EMBEDDED {
2417 public:
2418 // Make an iterator that runs over all pages in to-space.
2419 explicit inline NewSpacePageIterator(NewSpace* space);
2420
2421 // Make an iterator that runs over all pages in the given semispace,
2422 // even those not used in allocation.
2423 explicit inline NewSpacePageIterator(SemiSpace* space);
2424
2425 // Make iterator that iterates from the page containing start
2426 // to the page that contains limit in the same semispace.
2427 inline NewSpacePageIterator(Address start, Address limit);
2428
2429 inline bool has_next();
2430 inline NewSpacePage* next();
2431
2432 private:
2433 NewSpacePage* prev_page_; // Previous page returned.
2434 // Next page that will be returned. Cached here so that we can use this
2435 // iterator for operations that deallocate pages.
2436 NewSpacePage* next_page_;
2437 // Last page returned.
2438 NewSpacePage* last_page_;
2439 };
2440
2441
2442 // -----------------------------------------------------------------------------
2443 // The young generation space.
2444 //
2445 // The new space consists of a contiguous pair of semispaces. It simply
2446 // forwards most functions to the appropriate semispace.
2447
2448 class NewSpace : public Space {
2449 public:
2450 // Constructor.
2451 explicit NewSpace(Heap* heap)
2452 : Space(heap, NEW_SPACE, NOT_EXECUTABLE),
2453 to_space_(heap, kToSpace),
2454 from_space_(heap, kFromSpace),
2455 reservation_(),
2456 inline_allocation_limit_step_(0) {}
2457
2458 // Sets up the new space using the given chunk.
2459 bool SetUp(int reserved_semispace_size_, int max_semi_space_size);
2460
2461 // Tears down the space. Heap memory was not allocated by the space, so it
2462 // is not deallocated here.
2463 void TearDown();
2464
2465 // True if the space has been set up but not torn down.
2466 bool HasBeenSetUp() {
2467 return to_space_.HasBeenSetUp() && from_space_.HasBeenSetUp();
2468 }
2469
2470 // Flip the pair of spaces.
2471 void Flip();
2472
2473 // Grow the capacity of the semispaces. Assumes that they are not at
2474 // their maximum capacity.
2475 void Grow();
2476
2477 // Shrink the capacity of the semispaces.
2478 void Shrink();
2479
2480 // True if the address or object lies in the address range of either
2481 // semispace (not necessarily below the allocation pointer).
2482 bool Contains(Address a) {
2483 return (reinterpret_cast<uintptr_t>(a) & address_mask_)
2484 == reinterpret_cast<uintptr_t>(start_);
2485 }
2486
2487 bool Contains(Object* o) {
2488 Address a = reinterpret_cast<Address>(o);
2489 return (reinterpret_cast<uintptr_t>(a) & object_mask_) == object_expected_;
2490 }
2491
2492 // Return the allocated bytes in the active semispace.
2493 virtual intptr_t Size() {
2494 return pages_used_ * NewSpacePage::kAreaSize +
2495 static_cast<int>(top() - to_space_.page_low());
2496 }
2497
2498 // The same, but returning an int. We have to have the one that returns
2499 // intptr_t because it is inherited, but if we know we are dealing with the
2500 // new space, which can't get as big as the other spaces then this is useful:
2501 int SizeAsInt() { return static_cast<int>(Size()); }
2502
2503 // Return the current capacity of a semispace.
2504 intptr_t EffectiveCapacity() {
2505 SLOW_DCHECK(to_space_.Capacity() == from_space_.Capacity());
2506 return (to_space_.Capacity() / Page::kPageSize) * NewSpacePage::kAreaSize;
2507 }
2508
2509 // Return the current capacity of a semispace.
2510 intptr_t Capacity() {
2511 DCHECK(to_space_.Capacity() == from_space_.Capacity());
2512 return to_space_.Capacity();
2513 }
2514
2515 // Return the total amount of memory committed for new space.
2516 intptr_t CommittedMemory() {
2517 if (from_space_.is_committed()) return 2 * Capacity();
2518 return Capacity();
2519 }
2520
2521 // Return the total amount of memory committed for new space.
2522 intptr_t MaximumCommittedMemory() {
2523 return to_space_.MaximumCommittedMemory() +
2524 from_space_.MaximumCommittedMemory();
2525 }
2526
2527 // Approximate amount of physical memory committed for this space.
2528 size_t CommittedPhysicalMemory();
2529
2530 // Return the available bytes without growing.
2531 intptr_t Available() {
2532 return Capacity() - Size();
2533 }
2534
2535 // Return the maximum capacity of a semispace.
2536 int MaximumCapacity() {
2537 DCHECK(to_space_.MaximumCapacity() == from_space_.MaximumCapacity());
2538 return to_space_.MaximumCapacity();
2539 }
2540
2541 bool IsAtMaximumCapacity() {
2542 return Capacity() == MaximumCapacity();
2543 }
2544
2545 // Returns the initial capacity of a semispace.
2546 int InitialCapacity() {
2547 DCHECK(to_space_.InitialCapacity() == from_space_.InitialCapacity());
2548 return to_space_.InitialCapacity();
2549 }
2550
2551 // Return the address of the allocation pointer in the active semispace.
2552 Address top() {
2553 DCHECK(to_space_.current_page()->ContainsLimit(allocation_info_.top()));
2554 return allocation_info_.top();
2555 }
2556
2557 void set_top(Address top) {
2558 DCHECK(to_space_.current_page()->ContainsLimit(top));
2559 allocation_info_.set_top(top);
2560 }
2561
2562 // Return the address of the allocation pointer limit in the active semispace.
2563 Address limit() {
2564 DCHECK(to_space_.current_page()->ContainsLimit(allocation_info_.limit()));
2565 return allocation_info_.limit();
2566 }
2567
2568 // Return the address of the first object in the active semispace.
2569 Address bottom() { return to_space_.space_start(); }
2570
2571 // Get the age mark of the inactive semispace.
2572 Address age_mark() { return from_space_.age_mark(); }
2573 // Set the age mark in the active semispace.
2574 void set_age_mark(Address mark) { to_space_.set_age_mark(mark); }
2575
2576 // The start address of the space and a bit mask. Anding an address in the
2577 // new space with the mask will result in the start address.
2578 Address start() { return start_; }
2579 uintptr_t mask() { return address_mask_; }
2580
2581 INLINE(uint32_t AddressToMarkbitIndex(Address addr)) {
2582 DCHECK(Contains(addr));
2583 DCHECK(IsAligned(OffsetFrom(addr), kPointerSize) ||
2584 IsAligned(OffsetFrom(addr) - 1, kPointerSize));
2585 return static_cast<uint32_t>(addr - start_) >> kPointerSizeLog2;
2586 }
2587
2588 INLINE(Address MarkbitIndexToAddress(uint32_t index)) {
2589 return reinterpret_cast<Address>(index << kPointerSizeLog2);
2590 }
2591
2592 // The allocation top and limit address.
2593 Address* allocation_top_address() {
2594 return allocation_info_.top_address();
2595 }
2596
2597 // The allocation limit address.
2598 Address* allocation_limit_address() {
2599 return allocation_info_.limit_address();
2600 }
2601
2602 MUST_USE_RESULT INLINE(AllocationResult AllocateRaw(int size_in_bytes));
2603
2604 // Reset the allocation pointer to the beginning of the active semispace.
2605 void ResetAllocationInfo();
2606
2607 void UpdateInlineAllocationLimit(int size_in_bytes);
2608 void LowerInlineAllocationLimit(intptr_t step) {
2609 inline_allocation_limit_step_ = step;
2610 UpdateInlineAllocationLimit(0);
2611 top_on_previous_step_ = allocation_info_.top();
2612 }
2613
2614 // Get the extent of the inactive semispace (for use as a marking stack,
2615 // or to zap it). Notice: space-addresses are not necessarily on the
2616 // same page, so FromSpaceStart() might be above FromSpaceEnd().
2617 Address FromSpacePageLow() { return from_space_.page_low(); }
2618 Address FromSpacePageHigh() { return from_space_.page_high(); }
2619 Address FromSpaceStart() { return from_space_.space_start(); }
2620 Address FromSpaceEnd() { return from_space_.space_end(); }
2621
2622 // Get the extent of the active semispace's pages' memory.
2623 Address ToSpaceStart() { return to_space_.space_start(); }
2624 Address ToSpaceEnd() { return to_space_.space_end(); }
2625
2626 inline bool ToSpaceContains(Address address) {
2627 return to_space_.Contains(address);
2628 }
2629 inline bool FromSpaceContains(Address address) {
2630 return from_space_.Contains(address);
2631 }
2632
2633 // True if the object is a heap object in the address range of the
2634 // respective semispace (not necessarily below the allocation pointer of the
2635 // semispace).
2636 inline bool ToSpaceContains(Object* o) { return to_space_.Contains(o); }
2637 inline bool FromSpaceContains(Object* o) { return from_space_.Contains(o); }
2638
2639 // Try to switch the active semispace to a new, empty, page.
2640 // Returns false if this isn't possible or reasonable (i.e., there
2641 // are no pages, or the current page is already empty), or true
2642 // if successful.
2643 bool AddFreshPage();
2644
2645 #ifdef VERIFY_HEAP
2646 // Verify the active semispace.
2647 virtual void Verify();
2648 #endif
2649
2650 #ifdef DEBUG
2651 // Print the active semispace.
2652 virtual void Print() { to_space_.Print(); }
2653 #endif
2654
2655 // Iterates the active semispace to collect statistics.
2656 void CollectStatistics();
2657 // Reports previously collected statistics of the active semispace.
2658 void ReportStatistics();
2659 // Clears previously collected statistics.
2660 void ClearHistograms();
2661
2662 // Record the allocation or promotion of a heap object. Note that we don't
2663 // record every single allocation, but only those that happen in the
2664 // to space during a scavenge GC.
2665 void RecordAllocation(HeapObject* obj);
2666 void RecordPromotion(HeapObject* obj);
2667
2668 // Return whether the operation succeded.
2669 bool CommitFromSpaceIfNeeded() {
2670 if (from_space_.is_committed()) return true;
2671 return from_space_.Commit();
2672 }
2673
2674 bool UncommitFromSpace() {
2675 if (!from_space_.is_committed()) return true;
2676 return from_space_.Uncommit();
2677 }
2678
2679 inline intptr_t inline_allocation_limit_step() {
2680 return inline_allocation_limit_step_;
2681 }
2682
2683 SemiSpace* active_space() { return &to_space_; }
2684
2685 private:
2686 // Update allocation info to match the current to-space page.
2687 void UpdateAllocationInfo();
2688
2689 Address chunk_base_;
2690 uintptr_t chunk_size_;
2691
2692 // The semispaces.
2693 SemiSpace to_space_;
2694 SemiSpace from_space_;
2695 base::VirtualMemory reservation_;
2696 int pages_used_;
2697
2698 // Start address and bit mask for containment testing.
2699 Address start_;
2700 uintptr_t address_mask_;
2701 uintptr_t object_mask_;
2702 uintptr_t object_expected_;
2703
2704 // Allocation pointer and limit for normal allocation and allocation during
2705 // mark-compact collection.
2706 AllocationInfo allocation_info_;
2707
2708 // When incremental marking is active we will set allocation_info_.limit
2709 // to be lower than actual limit and then will gradually increase it
2710 // in steps to guarantee that we do incremental marking steps even
2711 // when all allocation is performed from inlined generated code.
2712 intptr_t inline_allocation_limit_step_;
2713
2714 Address top_on_previous_step_;
2715
2716 HistogramInfo* allocated_histogram_;
2717 HistogramInfo* promoted_histogram_;
2718
2719 MUST_USE_RESULT AllocationResult SlowAllocateRaw(int size_in_bytes);
2720
2721 friend class SemiSpaceIterator;
2722
2723 public:
2724 TRACK_MEMORY("NewSpace")
2725 };
2726
2727
2728 // -----------------------------------------------------------------------------
2729 // Old object space (excluding map objects)
2730
2731 class OldSpace : public PagedSpace {
2732 public:
2733 // Creates an old space object with a given maximum capacity.
2734 // The constructor does not allocate pages from OS.
2735 OldSpace(Heap* heap,
2736 intptr_t max_capacity,
2737 AllocationSpace id,
2738 Executability executable)
2739 : PagedSpace(heap, max_capacity, id, executable) {
2740 }
2741
2742 public:
2743 TRACK_MEMORY("OldSpace")
2744 };
2745
2746
2747 // For contiguous spaces, top should be in the space (or at the end) and limit
2748 // should be the end of the space.
2749 #define DCHECK_SEMISPACE_ALLOCATION_INFO(info, space) \
2750 SLOW_DCHECK((space).page_low() <= (info).top() \
2751 && (info).top() <= (space).page_high() \
2752 && (info).limit() <= (space).page_high())
2753
2754
2755 // -----------------------------------------------------------------------------
2756 // Old space for all map objects
2757
2758 class MapSpace : public PagedSpace {
2759 public:
2760 // Creates a map space object with a maximum capacity.
2761 MapSpace(Heap* heap, intptr_t max_capacity, AllocationSpace id)
2762 : PagedSpace(heap, max_capacity, id, NOT_EXECUTABLE),
2763 max_map_space_pages_(kMaxMapPageIndex - 1) {
2764 }
2765
2766 // Given an index, returns the page address.
2767 // TODO(1600): this limit is artifical just to keep code compilable
2768 static const int kMaxMapPageIndex = 1 << 16;
2769
2770 virtual int RoundSizeDownToObjectAlignment(int size) {
2771 if (IsPowerOf2(Map::kSize)) {
2772 return RoundDown(size, Map::kSize);
2773 } else {
2774 return (size / Map::kSize) * Map::kSize;
2775 }
2776 }
2777
2778 protected:
2779 virtual void VerifyObject(HeapObject* obj);
2780
2781 private:
2782 static const int kMapsPerPage = Page::kMaxRegularHeapObjectSize / Map::kSize;
2783
2784 // Do map space compaction if there is a page gap.
2785 int CompactionThreshold() {
2786 return kMapsPerPage * (max_map_space_pages_ - 1);
2787 }
2788
2789 const int max_map_space_pages_;
2790
2791 public:
2792 TRACK_MEMORY("MapSpace")
2793 };
2794
2795
2796 // -----------------------------------------------------------------------------
2797 // Old space for simple property cell objects
2798
2799 class CellSpace : public PagedSpace {
2800 public:
2801 // Creates a property cell space object with a maximum capacity.
2802 CellSpace(Heap* heap, intptr_t max_capacity, AllocationSpace id)
2803 : PagedSpace(heap, max_capacity, id, NOT_EXECUTABLE) {
2804 }
2805
2806 virtual int RoundSizeDownToObjectAlignment(int size) {
2807 if (IsPowerOf2(Cell::kSize)) {
2808 return RoundDown(size, Cell::kSize);
2809 } else {
2810 return (size / Cell::kSize) * Cell::kSize;
2811 }
2812 }
2813
2814 protected:
2815 virtual void VerifyObject(HeapObject* obj);
2816
2817 public:
2818 TRACK_MEMORY("CellSpace")
2819 };
2820
2821
2822 // -----------------------------------------------------------------------------
2823 // Old space for all global object property cell objects
2824
2825 class PropertyCellSpace : public PagedSpace {
2826 public:
2827 // Creates a property cell space object with a maximum capacity.
2828 PropertyCellSpace(Heap* heap, intptr_t max_capacity,
2829 AllocationSpace id)
2830 : PagedSpace(heap, max_capacity, id, NOT_EXECUTABLE) {
2831 }
2832
2833 virtual int RoundSizeDownToObjectAlignment(int size) {
2834 if (IsPowerOf2(PropertyCell::kSize)) {
2835 return RoundDown(size, PropertyCell::kSize);
2836 } else {
2837 return (size / PropertyCell::kSize) * PropertyCell::kSize;
2838 }
2839 }
2840
2841 protected:
2842 virtual void VerifyObject(HeapObject* obj);
2843
2844 public:
2845 TRACK_MEMORY("PropertyCellSpace")
2846 };
2847
2848
2849 // -----------------------------------------------------------------------------
2850 // Large objects ( > Page::kMaxHeapObjectSize ) are allocated and managed by
2851 // the large object space. A large object is allocated from OS heap with
2852 // extra padding bytes (Page::kPageSize + Page::kObjectStartOffset).
2853 // A large object always starts at Page::kObjectStartOffset to a page.
2854 // Large objects do not move during garbage collections.
2855
2856 class LargeObjectSpace : public Space {
2857 public:
2858 LargeObjectSpace(Heap* heap, intptr_t max_capacity, AllocationSpace id);
2859 virtual ~LargeObjectSpace() {}
2860
2861 // Initializes internal data structures.
2862 bool SetUp();
2863
2864 // Releases internal resources, frees objects in this space.
2865 void TearDown();
2866
2867 static intptr_t ObjectSizeFor(intptr_t chunk_size) {
2868 if (chunk_size <= (Page::kPageSize + Page::kObjectStartOffset)) return 0;
2869 return chunk_size - Page::kPageSize - Page::kObjectStartOffset;
2870 }
2871
2872 // Shared implementation of AllocateRaw, AllocateRawCode and
2873 // AllocateRawFixedArray.
2874 MUST_USE_RESULT AllocationResult AllocateRaw(int object_size,
2875 Executability executable);
2876
2877 // Available bytes for objects in this space.
2878 inline intptr_t Available();
2879
2880 virtual intptr_t Size() {
2881 return size_;
2882 }
2883
2884 virtual intptr_t SizeOfObjects() {
2885 return objects_size_;
2886 }
2887
2888 intptr_t MaximumCommittedMemory() {
2889 return maximum_committed_;
2890 }
2891
2892 intptr_t CommittedMemory() {
2893 return Size();
2894 }
2895
2896 // Approximate amount of physical memory committed for this space.
2897 size_t CommittedPhysicalMemory();
2898
2899 int PageCount() {
2900 return page_count_;
2901 }
2902
2903 // Finds an object for a given address, returns a Smi if it is not found.
2904 // The function iterates through all objects in this space, may be slow.
2905 Object* FindObject(Address a);
2906
2907 // Finds a large object page containing the given address, returns NULL
2908 // if such a page doesn't exist.
2909 LargePage* FindPage(Address a);
2910
2911 // Frees unmarked objects.
2912 void FreeUnmarkedObjects();
2913
2914 // Checks whether a heap object is in this space; O(1).
2915 bool Contains(HeapObject* obj);
2916
2917 // Checks whether the space is empty.
2918 bool IsEmpty() { return first_page_ == NULL; }
2919
2920 LargePage* first_page() { return first_page_; }
2921
2922 #ifdef VERIFY_HEAP
2923 virtual void Verify();
2924 #endif
2925
2926 #ifdef DEBUG
2927 virtual void Print();
2928 void ReportStatistics();
2929 void CollectCodeStatistics();
2930 #endif
2931 // Checks whether an address is in the object area in this space. It
2932 // iterates all objects in the space. May be slow.
2933 bool SlowContains(Address addr) { return FindObject(addr)->IsHeapObject(); }
2934
2935 private:
2936 intptr_t max_capacity_;
2937 intptr_t maximum_committed_;
2938 // The head of the linked list of large object chunks.
2939 LargePage* first_page_;
2940 intptr_t size_; // allocated bytes
2941 int page_count_; // number of chunks
2942 intptr_t objects_size_; // size of objects
2943 // Map MemoryChunk::kAlignment-aligned chunks to large pages covering them
2944 HashMap chunk_map_;
2945
2946 friend class LargeObjectIterator;
2947
2948 public:
2949 TRACK_MEMORY("LargeObjectSpace")
2950 };
2951
2952
2953 class LargeObjectIterator: public ObjectIterator {
2954 public:
2955 explicit LargeObjectIterator(LargeObjectSpace* space);
2956 LargeObjectIterator(LargeObjectSpace* space, HeapObjectCallback size_func);
2957
2958 HeapObject* Next();
2959
2960 // implementation of ObjectIterator.
2961 virtual HeapObject* next_object() { return Next(); }
2962
2963 private:
2964 LargePage* current_;
2965 HeapObjectCallback size_func_;
2966 };
2967
2968
2969 // Iterates over the chunks (pages and large object pages) that can contain
2970 // pointers to new space.
2971 class PointerChunkIterator BASE_EMBEDDED {
2972 public:
2973 inline explicit PointerChunkIterator(Heap* heap);
2974
2975 // Return NULL when the iterator is done.
2976 MemoryChunk* next() {
2977 switch (state_) {
2978 case kOldPointerState: {
2979 if (old_pointer_iterator_.has_next()) {
2980 return old_pointer_iterator_.next();
2981 }
2982 state_ = kMapState;
2983 // Fall through.
2984 }
2985 case kMapState: {
2986 if (map_iterator_.has_next()) {
2987 return map_iterator_.next();
2988 }
2989 state_ = kLargeObjectState;
2990 // Fall through.
2991 }
2992 case kLargeObjectState: {
2993 HeapObject* heap_object;
2994 do {
2995 heap_object = lo_iterator_.Next();
2996 if (heap_object == NULL) {
2997 state_ = kFinishedState;
2998 return NULL;
2999 }
3000 // Fixed arrays are the only pointer-containing objects in large
3001 // object space.
3002 } while (!heap_object->IsFixedArray());
3003 MemoryChunk* answer = MemoryChunk::FromAddress(heap_object->address());
3004 return answer;
3005 }
3006 case kFinishedState:
3007 return NULL;
3008 default:
3009 break;
3010 }
3011 UNREACHABLE();
3012 return NULL;
3013 }
3014
3015
3016 private:
3017 enum State {
3018 kOldPointerState,
3019 kMapState,
3020 kLargeObjectState,
3021 kFinishedState
3022 };
3023 State state_;
3024 PageIterator old_pointer_iterator_;
3025 PageIterator map_iterator_;
3026 LargeObjectIterator lo_iterator_;
3027 };
3028
3029
3030 #ifdef DEBUG
3031 struct CommentStatistic {
3032 const char* comment;
3033 int size;
3034 int count;
3035 void Clear() {
3036 comment = NULL;
3037 size = 0;
3038 count = 0;
3039 }
3040 // Must be small, since an iteration is used for lookup.
3041 static const int kMaxComments = 64;
3042 };
3043 #endif
3044
3045
3046 } } // namespace v8::internal
3047
3048 #endif // V8_SPACES_H_
OLDNEW
« no previous file with comments | « src/safepoint-table.h ('k') | src/spaces.cc » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698