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_HEAP_SPACES_H_ 6 #define V8_HEAP_SPACES_H_ 7 8 #include <list> 9 #include <memory> 10 #include <unordered_set> 11 12 #include "src/allocation.h" 13 #include "src/base/atomic-utils.h" 14 #include "src/base/atomicops.h" 15 #include "src/base/bits.h" 16 #include "src/base/hashmap.h" 17 #include "src/base/platform/mutex.h" 18 #include "src/flags.h" 19 #include "src/globals.h" 20 #include "src/heap/heap.h" 21 #include "src/heap/marking.h" 22 #include "src/list.h" 23 #include "src/objects.h" 24 #include "src/utils.h" 25 26 namespace v8 { 27 namespace internal { 28 29 class AllocationInfo; 30 class AllocationObserver; 31 class CompactionSpace; 32 class CompactionSpaceCollection; 33 class FreeList; 34 class Isolate; 35 class LocalArrayBufferTracker; 36 class MemoryAllocator; 37 class MemoryChunk; 38 class Page; 39 class PagedSpace; 40 class SemiSpace; 41 class SkipList; 42 class SlotsBuffer; 43 class SlotSet; 44 class TypedSlotSet; 45 class Space; 46 47 // ----------------------------------------------------------------------------- 48 // Heap structures: 49 // 50 // A JS heap consists of a young generation, an old generation, and a large 51 // object space. The young generation is divided into two semispaces. A 52 // scavenger implements Cheney's copying algorithm. The old generation is 53 // separated into a map space and an old object space. The map space contains 54 // all (and only) map objects, the rest of old objects go into the old space. 55 // The old generation is collected by a mark-sweep-compact collector. 56 // 57 // The semispaces of the young generation are contiguous. The old and map 58 // spaces consists of a list of pages. A page has a page header and an object 59 // area. 60 // 61 // There is a separate large object space for objects larger than 62 // kMaxRegularHeapObjectSize, so that they do not have to move during 63 // collection. The large object space is paged. Pages in large object space 64 // may be larger than the page size. 65 // 66 // A store-buffer based write barrier is used to keep track of intergenerational 67 // references. See heap/store-buffer.h. 68 // 69 // During scavenges and mark-sweep collections we sometimes (after a store 70 // buffer overflow) iterate intergenerational pointers without decoding heap 71 // object maps so if the page belongs to old space or large object space 72 // it is essential to guarantee that the page does not contain any 73 // garbage pointers to new space: every pointer aligned word which satisfies 74 // the Heap::InNewSpace() predicate must be a pointer to a live heap object in 75 // new space. Thus objects in old space and large object spaces should have a 76 // special layout (e.g. no bare integer fields). This requirement does not 77 // apply to map space which is iterated in a special fashion. However we still 78 // require pointer fields of dead maps to be cleaned. 79 // 80 // To enable lazy cleaning of old space pages we can mark chunks of the page 81 // as being garbage. Garbage sections are marked with a special map. These 82 // sections are skipped when scanning the page, even if we are otherwise 83 // scanning without regard for object boundaries. Garbage sections are chained 84 // together to form a free list after a GC. Garbage sections created outside 85 // of GCs by object trunctation etc. may not be in the free list chain. Very 86 // small free spaces are ignored, they need only be cleaned of bogus pointers 87 // into new space. 88 // 89 // Each page may have up to one special garbage section. The start of this 90 // section is denoted by the top field in the space. The end of the section 91 // is denoted by the limit field in the space. This special garbage section 92 // is not marked with a free space map in the data. The point of this section 93 // is to enable linear allocation without having to constantly update the byte 94 // array every time the top field is updated and a new object is created. The 95 // special garbage section is not in the chain of garbage sections. 96 // 97 // Since the top and limit fields are in the space, not the page, only one page 98 // has a special garbage section, and if the top and limit are equal then there 99 // is no special garbage section. 100 101 // Some assertion macros used in the debugging mode. 102 103 #define DCHECK_PAGE_ALIGNED(address) \ 104 DCHECK((OffsetFrom(address) & Page::kPageAlignmentMask) == 0) 105 106 #define DCHECK_OBJECT_ALIGNED(address) \ 107 DCHECK((OffsetFrom(address) & kObjectAlignmentMask) == 0) 108 109 #define DCHECK_OBJECT_SIZE(size) \ 110 DCHECK((0 < size) && (size <= kMaxRegularHeapObjectSize)) 111 112 #define DCHECK_CODEOBJECT_SIZE(size, code_space) \ 113 DCHECK((0 < size) && (size <= code_space->AreaSize())) 114 115 #define DCHECK_PAGE_OFFSET(offset) \ 116 DCHECK((Page::kObjectStartOffset <= offset) && (offset <= Page::kPageSize)) 117 118 enum FreeListCategoryType { 119 kTiniest, 120 kTiny, 121 kSmall, 122 kMedium, 123 kLarge, 124 kHuge, 125 126 kFirstCategory = kTiniest, 127 kLastCategory = kHuge, 128 kNumberOfCategories = kLastCategory + 1, 129 kInvalidCategory 130 }; 131 132 enum FreeMode { kLinkCategory, kDoNotLinkCategory }; 133 134 // A free list category maintains a linked list of free memory blocks. 135 class FreeListCategory { 136 public: 137 static const int kSize = kIntSize + // FreeListCategoryType type_ 138 kIntSize + // padding for type_ 139 kSizetSize + // size_t available_ 140 kPointerSize + // FreeSpace* top_ 141 kPointerSize + // FreeListCategory* prev_ 142 kPointerSize; // FreeListCategory* next_ 143 FreeListCategory()144 FreeListCategory() 145 : type_(kInvalidCategory), 146 available_(0), 147 top_(nullptr), 148 prev_(nullptr), 149 next_(nullptr) {} 150 Initialize(FreeListCategoryType type)151 void Initialize(FreeListCategoryType type) { 152 type_ = type; 153 available_ = 0; 154 top_ = nullptr; 155 prev_ = nullptr; 156 next_ = nullptr; 157 } 158 159 void Invalidate(); 160 161 void Reset(); 162 ResetStats()163 void ResetStats() { Reset(); } 164 165 void RepairFreeList(Heap* heap); 166 167 // Relinks the category into the currently owning free list. Requires that the 168 // category is currently unlinked. 169 void Relink(); 170 171 bool Free(FreeSpace* node, size_t size_in_bytes, FreeMode mode); 172 173 // Picks a node from the list and stores its size in |node_size|. Returns 174 // nullptr if the category is empty. 175 FreeSpace* PickNodeFromList(size_t* node_size); 176 177 // Performs a single try to pick a node of at least |minimum_size| from the 178 // category. Stores the actual size in |node_size|. Returns nullptr if no 179 // node is found. 180 FreeSpace* TryPickNodeFromList(size_t minimum_size, size_t* node_size); 181 182 // Picks a node of at least |minimum_size| from the category. Stores the 183 // actual size in |node_size|. Returns nullptr if no node is found. 184 FreeSpace* SearchForNodeInList(size_t minimum_size, size_t* node_size); 185 186 inline FreeList* owner(); 187 inline bool is_linked(); is_empty()188 bool is_empty() { return top() == nullptr; } available()189 size_t available() const { return available_; } 190 191 #ifdef DEBUG 192 size_t SumFreeList(); 193 int FreeListLength(); 194 #endif 195 196 private: 197 // For debug builds we accurately compute free lists lengths up until 198 // {kVeryLongFreeList} by manually walking the list. 199 static const int kVeryLongFreeList = 500; 200 201 inline Page* page(); 202 top()203 FreeSpace* top() { return top_; } set_top(FreeSpace * top)204 void set_top(FreeSpace* top) { top_ = top; } prev()205 FreeListCategory* prev() { return prev_; } set_prev(FreeListCategory * prev)206 void set_prev(FreeListCategory* prev) { prev_ = prev; } next()207 FreeListCategory* next() { return next_; } set_next(FreeListCategory * next)208 void set_next(FreeListCategory* next) { next_ = next; } 209 210 // |type_|: The type of this free list category. 211 FreeListCategoryType type_; 212 213 // |available_|: Total available bytes in all blocks of this free list 214 // category. 215 size_t available_; 216 217 // |top_|: Points to the top FreeSpace* in the free list category. 218 FreeSpace* top_; 219 220 FreeListCategory* prev_; 221 FreeListCategory* next_; 222 223 friend class FreeList; 224 friend class PagedSpace; 225 }; 226 227 // MemoryChunk represents a memory region owned by a specific space. 228 // It is divided into the header and the body. Chunk start is always 229 // 1MB aligned. Start of the body is aligned so it can accommodate 230 // any heap object. 231 class MemoryChunk { 232 public: 233 enum Flag { 234 NO_FLAGS = 0u, 235 IS_EXECUTABLE = 1u << 0, 236 POINTERS_TO_HERE_ARE_INTERESTING = 1u << 1, 237 POINTERS_FROM_HERE_ARE_INTERESTING = 1u << 2, 238 // A page in new space has one of the next to flags set. 239 IN_FROM_SPACE = 1u << 3, 240 IN_TO_SPACE = 1u << 4, 241 NEW_SPACE_BELOW_AGE_MARK = 1u << 5, 242 EVACUATION_CANDIDATE = 1u << 6, 243 NEVER_EVACUATE = 1u << 7, 244 245 // Large objects can have a progress bar in their page header. These object 246 // are scanned in increments and will be kept black while being scanned. 247 // Even if the mutator writes to them they will be kept black and a white 248 // to grey transition is performed in the value. 249 HAS_PROGRESS_BAR = 1u << 8, 250 251 // |PAGE_NEW_OLD_PROMOTION|: A page tagged with this flag has been promoted 252 // from new to old space during evacuation. 253 PAGE_NEW_OLD_PROMOTION = 1u << 9, 254 255 // |PAGE_NEW_NEW_PROMOTION|: A page tagged with this flag has been moved 256 // within the new space during evacuation. 257 PAGE_NEW_NEW_PROMOTION = 1u << 10, 258 259 // This flag is intended to be used for testing. Works only when both 260 // FLAG_stress_compaction and FLAG_manual_evacuation_candidates_selection 261 // are set. It forces the page to become an evacuation candidate at next 262 // candidates selection cycle. 263 FORCE_EVACUATION_CANDIDATE_FOR_TESTING = 1u << 11, 264 265 // This flag is intended to be used for testing. 266 NEVER_ALLOCATE_ON_PAGE = 1u << 12, 267 268 // The memory chunk is already logically freed, however the actual freeing 269 // still has to be performed. 270 PRE_FREED = 1u << 13, 271 272 // |POOLED|: When actually freeing this chunk, only uncommit and do not 273 // give up the reservation as we still reuse the chunk at some point. 274 POOLED = 1u << 14, 275 276 // |COMPACTION_WAS_ABORTED|: Indicates that the compaction in this page 277 // has been aborted and needs special handling by the sweeper. 278 COMPACTION_WAS_ABORTED = 1u << 15, 279 280 // |COMPACTION_WAS_ABORTED_FOR_TESTING|: During stress testing evacuation 281 // on pages is sometimes aborted. The flag is used to avoid repeatedly 282 // triggering on the same page. 283 COMPACTION_WAS_ABORTED_FOR_TESTING = 1u << 16, 284 285 // |ANCHOR|: Flag is set if page is an anchor. 286 ANCHOR = 1u << 17, 287 }; 288 typedef base::Flags<Flag, uintptr_t> Flags; 289 290 static const int kPointersToHereAreInterestingMask = 291 POINTERS_TO_HERE_ARE_INTERESTING; 292 293 static const int kPointersFromHereAreInterestingMask = 294 POINTERS_FROM_HERE_ARE_INTERESTING; 295 296 static const int kEvacuationCandidateMask = EVACUATION_CANDIDATE; 297 298 static const int kIsInNewSpaceMask = IN_FROM_SPACE | IN_TO_SPACE; 299 300 static const int kSkipEvacuationSlotsRecordingMask = 301 kEvacuationCandidateMask | kIsInNewSpaceMask; 302 303 // |kSweepingDone|: The page state when sweeping is complete or sweeping must 304 // not be performed on that page. Sweeper threads that are done with their 305 // work will set this value and not touch the page anymore. 306 // |kSweepingPending|: This page is ready for parallel sweeping. 307 // |kSweepingInProgress|: This page is currently swept by a sweeper thread. 308 enum ConcurrentSweepingState { 309 kSweepingDone, 310 kSweepingPending, 311 kSweepingInProgress, 312 }; 313 314 static const intptr_t kAlignment = 315 (static_cast<uintptr_t>(1) << kPageSizeBits); 316 317 static const intptr_t kAlignmentMask = kAlignment - 1; 318 319 static const intptr_t kSizeOffset = 0; 320 321 static const intptr_t kFlagsOffset = kSizeOffset + kPointerSize; 322 323 static const size_t kMinHeaderSize = 324 kSizeOffset + kSizetSize // size_t size 325 + kIntptrSize // Flags flags_ 326 + kPointerSize // Address area_start_ 327 + kPointerSize // Address area_end_ 328 + 2 * kPointerSize // base::VirtualMemory reservation_ 329 + kPointerSize // Address owner_ 330 + kPointerSize // Heap* heap_ 331 + kIntSize // int progress_bar_ 332 + kIntSize // int live_bytes_count_ 333 + kPointerSize // SlotSet* old_to_new_slots_ 334 + kPointerSize // SlotSet* old_to_old_slots_ 335 + kPointerSize // TypedSlotSet* typed_old_to_new_slots_ 336 + kPointerSize // TypedSlotSet* typed_old_to_old_slots_ 337 + kPointerSize // SkipList* skip_list_ 338 + kPointerSize // AtomicValue high_water_mark_ 339 + kPointerSize // base::Mutex* mutex_ 340 + kPointerSize // base::AtomicWord concurrent_sweeping_ 341 + 2 * kSizetSize // AtomicNumber free-list statistics 342 + kPointerSize // AtomicValue next_chunk_ 343 + kPointerSize // AtomicValue prev_chunk_ 344 // FreeListCategory categories_[kNumberOfCategories] 345 + FreeListCategory::kSize * kNumberOfCategories + 346 kPointerSize; // LocalArrayBufferTracker* local_tracker_ 347 348 // We add some more space to the computed header size to amount for missing 349 // alignment requirements in our computation. 350 // Try to get kHeaderSize properly aligned on 32-bit and 64-bit machines. 351 static const size_t kHeaderSize = kMinHeaderSize; 352 353 static const int kBodyOffset = 354 CODE_POINTER_ALIGN(kHeaderSize + Bitmap::kSize); 355 356 // The start offset of the object area in a page. Aligned to both maps and 357 // code alignment to be suitable for both. Also aligned to 32 words because 358 // the marking bitmap is arranged in 32 bit chunks. 359 static const int kObjectStartAlignment = 32 * kPointerSize; 360 static const int kObjectStartOffset = 361 kBodyOffset - 1 + 362 (kObjectStartAlignment - (kBodyOffset - 1) % kObjectStartAlignment); 363 364 // Page size in bytes. This must be a multiple of the OS page size. 365 static const int kPageSize = 1 << kPageSizeBits; 366 static const intptr_t kPageAlignmentMask = (1 << kPageSizeBits) - 1; 367 368 static const int kAllocatableMemory = kPageSize - kObjectStartOffset; 369 370 static inline void IncrementLiveBytesFromMutator(HeapObject* object, int by); 371 static inline void IncrementLiveBytesFromGC(HeapObject* object, int by); 372 373 // Only works if the pointer is in the first kPageSize of the MemoryChunk. FromAddress(Address a)374 static MemoryChunk* FromAddress(Address a) { 375 return reinterpret_cast<MemoryChunk*>(OffsetFrom(a) & ~kAlignmentMask); 376 } 377 378 static inline MemoryChunk* FromAnyPointerAddress(Heap* heap, Address addr); 379 UpdateHighWaterMark(Address mark)380 static inline void UpdateHighWaterMark(Address mark) { 381 if (mark == nullptr) return; 382 // Need to subtract one from the mark because when a chunk is full the 383 // top points to the next address after the chunk, which effectively belongs 384 // to another chunk. See the comment to Page::FromTopOrLimit. 385 MemoryChunk* chunk = MemoryChunk::FromAddress(mark - 1); 386 intptr_t new_mark = static_cast<intptr_t>(mark - chunk->address()); 387 intptr_t old_mark = 0; 388 do { 389 old_mark = chunk->high_water_mark_.Value(); 390 } while ((new_mark > old_mark) && 391 !chunk->high_water_mark_.TrySetValue(old_mark, new_mark)); 392 } 393 IsValid(MemoryChunk * chunk)394 static bool IsValid(MemoryChunk* chunk) { return chunk != nullptr; } 395 address()396 Address address() { return reinterpret_cast<Address>(this); } 397 mutex()398 base::Mutex* mutex() { return mutex_; } 399 Contains(Address addr)400 bool Contains(Address addr) { 401 return addr >= area_start() && addr < area_end(); 402 } 403 404 // Checks whether |addr| can be a limit of addresses in this page. It's a 405 // limit if it's in the page, or if it's just after the last byte of the page. ContainsLimit(Address addr)406 bool ContainsLimit(Address addr) { 407 return addr >= area_start() && addr <= area_end(); 408 } 409 concurrent_sweeping_state()410 base::AtomicValue<ConcurrentSweepingState>& concurrent_sweeping_state() { 411 return concurrent_sweeping_; 412 } 413 SweepingDone()414 bool SweepingDone() { 415 return concurrent_sweeping_state().Value() == kSweepingDone; 416 } 417 418 // Manage live byte count, i.e., count of bytes in black objects. 419 inline void ResetLiveBytes(); 420 inline void IncrementLiveBytes(int by); 421 LiveBytes()422 int LiveBytes() { 423 DCHECK_LE(static_cast<unsigned>(live_byte_count_), size_); 424 return live_byte_count_; 425 } 426 SetLiveBytes(int live_bytes)427 void SetLiveBytes(int live_bytes) { 428 DCHECK_GE(live_bytes, 0); 429 DCHECK_LE(static_cast<size_t>(live_bytes), size_); 430 live_byte_count_ = live_bytes; 431 } 432 size()433 size_t size() const { return size_; } set_size(size_t size)434 void set_size(size_t size) { size_ = size; } 435 heap()436 inline Heap* heap() const { return heap_; } 437 skip_list()438 inline SkipList* skip_list() { return skip_list_; } 439 set_skip_list(SkipList * skip_list)440 inline void set_skip_list(SkipList* skip_list) { skip_list_ = skip_list; } 441 old_to_new_slots()442 inline SlotSet* old_to_new_slots() { return old_to_new_slots_.Value(); } old_to_old_slots()443 inline SlotSet* old_to_old_slots() { return old_to_old_slots_; } typed_old_to_new_slots()444 inline TypedSlotSet* typed_old_to_new_slots() { 445 return typed_old_to_new_slots_.Value(); 446 } typed_old_to_old_slots()447 inline TypedSlotSet* typed_old_to_old_slots() { 448 return typed_old_to_old_slots_; 449 } local_tracker()450 inline LocalArrayBufferTracker* local_tracker() { return local_tracker_; } 451 452 V8_EXPORT_PRIVATE void AllocateOldToNewSlots(); 453 void ReleaseOldToNewSlots(); 454 V8_EXPORT_PRIVATE void AllocateOldToOldSlots(); 455 void ReleaseOldToOldSlots(); 456 void AllocateTypedOldToNewSlots(); 457 void ReleaseTypedOldToNewSlots(); 458 void AllocateTypedOldToOldSlots(); 459 void ReleaseTypedOldToOldSlots(); 460 void AllocateLocalTracker(); 461 void ReleaseLocalTracker(); 462 area_start()463 Address area_start() { return area_start_; } area_end()464 Address area_end() { return area_end_; } area_size()465 size_t area_size() { return static_cast<size_t>(area_end() - area_start()); } 466 467 bool CommitArea(size_t requested); 468 469 // Approximate amount of physical memory committed for this chunk. 470 size_t CommittedPhysicalMemory(); 471 HighWaterMark()472 Address HighWaterMark() { return address() + high_water_mark_.Value(); } 473 progress_bar()474 int progress_bar() { 475 DCHECK(IsFlagSet(HAS_PROGRESS_BAR)); 476 return progress_bar_; 477 } 478 set_progress_bar(int progress_bar)479 void set_progress_bar(int progress_bar) { 480 DCHECK(IsFlagSet(HAS_PROGRESS_BAR)); 481 progress_bar_ = progress_bar; 482 } 483 ResetProgressBar()484 void ResetProgressBar() { 485 if (IsFlagSet(MemoryChunk::HAS_PROGRESS_BAR)) { 486 set_progress_bar(0); 487 } 488 } 489 markbits()490 inline Bitmap* markbits() { 491 return Bitmap::FromAddress(address() + kHeaderSize); 492 } 493 AddressToMarkbitIndex(Address addr)494 inline uint32_t AddressToMarkbitIndex(Address addr) { 495 return static_cast<uint32_t>(addr - this->address()) >> kPointerSizeLog2; 496 } 497 MarkbitIndexToAddress(uint32_t index)498 inline Address MarkbitIndexToAddress(uint32_t index) { 499 return this->address() + (index << kPointerSizeLog2); 500 } 501 502 void ClearLiveness(); 503 PrintMarkbits()504 void PrintMarkbits() { markbits()->Print(); } 505 SetFlag(Flag flag)506 void SetFlag(Flag flag) { flags_ |= flag; } ClearFlag(Flag flag)507 void ClearFlag(Flag flag) { flags_ &= ~Flags(flag); } IsFlagSet(Flag flag)508 bool IsFlagSet(Flag flag) { return flags_ & flag; } 509 510 // Set or clear multiple flags at a time. The flags in the mask are set to 511 // the value in "flags", the rest retain the current value in |flags_|. SetFlags(uintptr_t flags,uintptr_t mask)512 void SetFlags(uintptr_t flags, uintptr_t mask) { 513 flags_ = (flags_ & ~Flags(mask)) | (Flags(flags) & Flags(mask)); 514 } 515 516 // Return all current flags. GetFlags()517 uintptr_t GetFlags() { return flags_; } 518 NeverEvacuate()519 bool NeverEvacuate() { return IsFlagSet(NEVER_EVACUATE); } 520 MarkNeverEvacuate()521 void MarkNeverEvacuate() { SetFlag(NEVER_EVACUATE); } 522 IsEvacuationCandidate()523 bool IsEvacuationCandidate() { 524 DCHECK(!(IsFlagSet(NEVER_EVACUATE) && IsFlagSet(EVACUATION_CANDIDATE))); 525 return IsFlagSet(EVACUATION_CANDIDATE); 526 } 527 CanAllocate()528 bool CanAllocate() { 529 return !IsEvacuationCandidate() && !IsFlagSet(NEVER_ALLOCATE_ON_PAGE); 530 } 531 ShouldSkipEvacuationSlotRecording()532 bool ShouldSkipEvacuationSlotRecording() { 533 return ((flags_ & kSkipEvacuationSlotsRecordingMask) != 0) && 534 !IsFlagSet(COMPACTION_WAS_ABORTED); 535 } 536 executable()537 Executability executable() { 538 return IsFlagSet(IS_EXECUTABLE) ? EXECUTABLE : NOT_EXECUTABLE; 539 } 540 InNewSpace()541 bool InNewSpace() { return (flags_ & kIsInNewSpaceMask) != 0; } 542 InToSpace()543 bool InToSpace() { return IsFlagSet(IN_TO_SPACE); } 544 InFromSpace()545 bool InFromSpace() { return IsFlagSet(IN_FROM_SPACE); } 546 next_chunk()547 MemoryChunk* next_chunk() { return next_chunk_.Value(); } 548 prev_chunk()549 MemoryChunk* prev_chunk() { return prev_chunk_.Value(); } 550 set_next_chunk(MemoryChunk * next)551 void set_next_chunk(MemoryChunk* next) { next_chunk_.SetValue(next); } 552 set_prev_chunk(MemoryChunk * prev)553 void set_prev_chunk(MemoryChunk* prev) { prev_chunk_.SetValue(prev); } 554 owner()555 Space* owner() const { 556 if ((reinterpret_cast<intptr_t>(owner_) & kPageHeaderTagMask) == 557 kPageHeaderTag) { 558 return reinterpret_cast<Space*>(reinterpret_cast<intptr_t>(owner_) - 559 kPageHeaderTag); 560 } else { 561 return nullptr; 562 } 563 } 564 set_owner(Space * space)565 void set_owner(Space* space) { 566 DCHECK((reinterpret_cast<intptr_t>(space) & kPageHeaderTagMask) == 0); 567 owner_ = reinterpret_cast<Address>(space) + kPageHeaderTag; 568 DCHECK((reinterpret_cast<intptr_t>(owner_) & kPageHeaderTagMask) == 569 kPageHeaderTag); 570 } 571 HasPageHeader()572 bool HasPageHeader() { return owner() != nullptr; } 573 574 void InsertAfter(MemoryChunk* other); 575 void Unlink(); 576 577 protected: 578 static MemoryChunk* Initialize(Heap* heap, Address base, size_t size, 579 Address area_start, Address area_end, 580 Executability executable, Space* owner, 581 base::VirtualMemory* reservation); 582 583 // Should be called when memory chunk is about to be freed. 584 void ReleaseAllocatedMemory(); 585 reserved_memory()586 base::VirtualMemory* reserved_memory() { return &reservation_; } 587 588 size_t size_; 589 Flags flags_; 590 591 // Start and end of allocatable memory on this chunk. 592 Address area_start_; 593 Address area_end_; 594 595 // If the chunk needs to remember its memory reservation, it is stored here. 596 base::VirtualMemory reservation_; 597 598 // The identity of the owning space. This is tagged as a failure pointer, but 599 // no failure can be in an object, so this can be distinguished from any entry 600 // in a fixed array. 601 Address owner_; 602 603 Heap* heap_; 604 605 // Used by the incremental marker to keep track of the scanning progress in 606 // large objects that have a progress bar and are scanned in increments. 607 int progress_bar_; 608 609 // Count of bytes marked black on page. 610 int live_byte_count_; 611 612 // A single slot set for small pages (of size kPageSize) or an array of slot 613 // set for large pages. In the latter case the number of entries in the array 614 // is ceil(size() / kPageSize). 615 base::AtomicValue<SlotSet*> old_to_new_slots_; 616 SlotSet* old_to_old_slots_; 617 base::AtomicValue<TypedSlotSet*> typed_old_to_new_slots_; 618 TypedSlotSet* typed_old_to_old_slots_; 619 620 SkipList* skip_list_; 621 622 // Assuming the initial allocation on a page is sequential, 623 // count highest number of bytes ever allocated on the page. 624 base::AtomicValue<intptr_t> high_water_mark_; 625 626 base::Mutex* mutex_; 627 628 base::AtomicValue<ConcurrentSweepingState> concurrent_sweeping_; 629 630 // PagedSpace free-list statistics. 631 base::AtomicNumber<intptr_t> available_in_free_list_; 632 base::AtomicNumber<intptr_t> wasted_memory_; 633 634 // next_chunk_ holds a pointer of type MemoryChunk 635 base::AtomicValue<MemoryChunk*> next_chunk_; 636 // prev_chunk_ holds a pointer of type MemoryChunk 637 base::AtomicValue<MemoryChunk*> prev_chunk_; 638 639 FreeListCategory categories_[kNumberOfCategories]; 640 641 LocalArrayBufferTracker* local_tracker_; 642 643 private: InitializeReservedMemory()644 void InitializeReservedMemory() { reservation_.Reset(); } 645 646 friend class MemoryAllocator; 647 friend class MemoryChunkValidator; 648 }; 649 650 DEFINE_OPERATORS_FOR_FLAGS(MemoryChunk::Flags) 651 652 static_assert(kMaxRegularHeapObjectSize <= MemoryChunk::kAllocatableMemory, 653 "kMaxRegularHeapObjectSize <= MemoryChunk::kAllocatableMemory"); 654 655 // ----------------------------------------------------------------------------- 656 // A page is a memory chunk of a size 1MB. Large object pages may be larger. 657 // 658 // The only way to get a page pointer is by calling factory methods: 659 // Page* p = Page::FromAddress(addr); or 660 // Page* p = Page::FromTopOrLimit(top); 661 class Page : public MemoryChunk { 662 public: 663 static const intptr_t kCopyAllFlags = ~0; 664 665 // Page flags copied from from-space to to-space when flipping semispaces. 666 static const intptr_t kCopyOnFlipFlagsMask = 667 static_cast<intptr_t>(MemoryChunk::POINTERS_TO_HERE_ARE_INTERESTING) | 668 static_cast<intptr_t>(MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING); 669 670 static inline Page* ConvertNewToOld(Page* old_page); 671 672 // Returns the page containing a given address. The address ranges 673 // from [page_addr .. page_addr + kPageSize[. This only works if the object 674 // is in fact in a page. FromAddress(Address addr)675 static Page* FromAddress(Address addr) { 676 return reinterpret_cast<Page*>(OffsetFrom(addr) & ~kPageAlignmentMask); 677 } 678 679 // Returns the page containing the address provided. The address can 680 // potentially point righter after the page. To be also safe for tagged values 681 // we subtract a hole word. The valid address ranges from 682 // [page_addr + kObjectStartOffset .. page_addr + kPageSize + kPointerSize]. FromAllocationAreaAddress(Address address)683 static Page* FromAllocationAreaAddress(Address address) { 684 return Page::FromAddress(address - kPointerSize); 685 } 686 687 // Checks if address1 and address2 are on the same new space page. OnSamePage(Address address1,Address address2)688 static bool OnSamePage(Address address1, Address address2) { 689 return Page::FromAddress(address1) == Page::FromAddress(address2); 690 } 691 692 // Checks whether an address is page aligned. IsAlignedToPageSize(Address addr)693 static bool IsAlignedToPageSize(Address addr) { 694 return (OffsetFrom(addr) & kPageAlignmentMask) == 0; 695 } 696 IsAtObjectStart(Address addr)697 static bool IsAtObjectStart(Address addr) { 698 return (reinterpret_cast<intptr_t>(addr) & kPageAlignmentMask) == 699 kObjectStartOffset; 700 } 701 702 inline static Page* FromAnyPointerAddress(Heap* heap, Address addr); 703 704 // Create a Page object that is only used as anchor for the doubly-linked 705 // list of real pages. Page(Space * owner)706 explicit Page(Space* owner) { InitializeAsAnchor(owner); } 707 708 inline void MarkNeverAllocateForTesting(); 709 inline void MarkEvacuationCandidate(); 710 inline void ClearEvacuationCandidate(); 711 next_page()712 Page* next_page() { return static_cast<Page*>(next_chunk()); } prev_page()713 Page* prev_page() { return static_cast<Page*>(prev_chunk()); } set_next_page(Page * page)714 void set_next_page(Page* page) { set_next_chunk(page); } set_prev_page(Page * page)715 void set_prev_page(Page* page) { set_prev_chunk(page); } 716 717 template <typename Callback> ForAllFreeListCategories(Callback callback)718 inline void ForAllFreeListCategories(Callback callback) { 719 for (int i = kFirstCategory; i < kNumberOfCategories; i++) { 720 callback(&categories_[i]); 721 } 722 } 723 724 // Returns the offset of a given address to this page. Offset(Address a)725 inline size_t Offset(Address a) { return static_cast<size_t>(a - address()); } 726 727 // Returns the address for a given offset to the this page. OffsetToAddress(size_t offset)728 Address OffsetToAddress(size_t offset) { 729 DCHECK_PAGE_OFFSET(offset); 730 return address() + offset; 731 } 732 733 // WaitUntilSweepingCompleted only works when concurrent sweeping is in 734 // progress. In particular, when we know that right before this call a 735 // sweeper thread was sweeping this page. WaitUntilSweepingCompleted()736 void WaitUntilSweepingCompleted() { 737 mutex_->Lock(); 738 mutex_->Unlock(); 739 DCHECK(SweepingDone()); 740 } 741 742 void ResetFreeListStatistics(); 743 744 size_t AvailableInFreeList(); 745 LiveBytesFromFreeList()746 size_t LiveBytesFromFreeList() { 747 DCHECK_GE(area_size(), wasted_memory() + available_in_free_list()); 748 return area_size() - wasted_memory() - available_in_free_list(); 749 } 750 free_list_category(FreeListCategoryType type)751 FreeListCategory* free_list_category(FreeListCategoryType type) { 752 return &categories_[type]; 753 } 754 is_anchor()755 bool is_anchor() { return IsFlagSet(Page::ANCHOR); } 756 wasted_memory()757 size_t wasted_memory() { return wasted_memory_.Value(); } add_wasted_memory(size_t waste)758 void add_wasted_memory(size_t waste) { wasted_memory_.Increment(waste); } available_in_free_list()759 size_t available_in_free_list() { return available_in_free_list_.Value(); } add_available_in_free_list(size_t available)760 void add_available_in_free_list(size_t available) { 761 DCHECK_LE(available, area_size()); 762 available_in_free_list_.Increment(available); 763 } remove_available_in_free_list(size_t available)764 void remove_available_in_free_list(size_t available) { 765 DCHECK_LE(available, area_size()); 766 DCHECK_GE(available_in_free_list(), available); 767 available_in_free_list_.Decrement(available); 768 } 769 770 size_t ShrinkToHighWaterMark(); 771 772 #ifdef DEBUG 773 void Print(); 774 #endif // DEBUG 775 776 private: 777 enum InitializationMode { kFreeMemory, kDoNotFreeMemory }; 778 779 template <InitializationMode mode = kFreeMemory> 780 static inline Page* Initialize(Heap* heap, MemoryChunk* chunk, 781 Executability executable, PagedSpace* owner); 782 static inline Page* Initialize(Heap* heap, MemoryChunk* chunk, 783 Executability executable, SemiSpace* owner); 784 785 inline void InitializeFreeListCategories(); 786 787 void InitializeAsAnchor(Space* owner); 788 789 friend class MemoryAllocator; 790 }; 791 792 class LargePage : public MemoryChunk { 793 public: GetObject()794 HeapObject* GetObject() { return HeapObject::FromAddress(area_start()); } 795 next_page()796 inline LargePage* next_page() { 797 return static_cast<LargePage*>(next_chunk()); 798 } 799 set_next_page(LargePage * page)800 inline void set_next_page(LargePage* page) { set_next_chunk(page); } 801 802 // Uncommit memory that is not in use anymore by the object. If the object 803 // cannot be shrunk 0 is returned. 804 Address GetAddressToShrink(); 805 806 void ClearOutOfLiveRangeSlots(Address free_start); 807 808 // A limit to guarantee that we do not overflow typed slot offset in 809 // the old to old remembered set. 810 // Note that this limit is higher than what assembler already imposes on 811 // x64 and ia32 architectures. 812 static const int kMaxCodePageSize = 512 * MB; 813 814 private: 815 static inline LargePage* Initialize(Heap* heap, MemoryChunk* chunk, 816 Executability executable, Space* owner); 817 818 friend class MemoryAllocator; 819 }; 820 821 822 // ---------------------------------------------------------------------------- 823 // Space is the abstract superclass for all allocation spaces. 824 class Space : public Malloced { 825 public: Space(Heap * heap,AllocationSpace id,Executability executable)826 Space(Heap* heap, AllocationSpace id, Executability executable) 827 : allocation_observers_(new List<AllocationObserver*>()), 828 allocation_observers_paused_(false), 829 heap_(heap), 830 id_(id), 831 executable_(executable), 832 committed_(0), 833 max_committed_(0) {} 834 ~Space()835 virtual ~Space() {} 836 heap()837 Heap* heap() const { return heap_; } 838 839 // Does the space need executable memory? executable()840 Executability executable() { return executable_; } 841 842 // Identity used in error reporting. identity()843 AllocationSpace identity() { return id_; } 844 AddAllocationObserver(AllocationObserver * observer)845 virtual void AddAllocationObserver(AllocationObserver* observer) { 846 allocation_observers_->Add(observer); 847 } 848 RemoveAllocationObserver(AllocationObserver * observer)849 virtual void RemoveAllocationObserver(AllocationObserver* observer) { 850 bool removed = allocation_observers_->RemoveElement(observer); 851 USE(removed); 852 DCHECK(removed); 853 } 854 PauseAllocationObservers()855 virtual void PauseAllocationObservers() { 856 allocation_observers_paused_ = true; 857 } 858 ResumeAllocationObservers()859 virtual void ResumeAllocationObservers() { 860 allocation_observers_paused_ = false; 861 } 862 863 void AllocationStep(Address soon_object, int size); 864 865 // Return the total amount committed memory for this space, i.e., allocatable 866 // memory and page headers. CommittedMemory()867 virtual size_t CommittedMemory() { return committed_; } 868 MaximumCommittedMemory()869 virtual size_t MaximumCommittedMemory() { return max_committed_; } 870 871 // Returns allocated size. 872 virtual size_t Size() = 0; 873 874 // Returns size of objects. Can differ from the allocated size 875 // (e.g. see LargeObjectSpace). SizeOfObjects()876 virtual size_t SizeOfObjects() { return Size(); } 877 878 // Approximate amount of physical memory committed for this space. 879 virtual size_t CommittedPhysicalMemory() = 0; 880 881 // Return the available bytes without growing. 882 virtual size_t Available() = 0; 883 RoundSizeDownToObjectAlignment(int size)884 virtual int RoundSizeDownToObjectAlignment(int size) { 885 if (id_ == CODE_SPACE) { 886 return RoundDown(size, kCodeAlignment); 887 } else { 888 return RoundDown(size, kPointerSize); 889 } 890 } 891 892 virtual std::unique_ptr<ObjectIterator> GetObjectIterator() = 0; 893 AccountCommitted(size_t bytes)894 void AccountCommitted(size_t bytes) { 895 DCHECK_GE(committed_ + bytes, committed_); 896 committed_ += bytes; 897 if (committed_ > max_committed_) { 898 max_committed_ = committed_; 899 } 900 } 901 AccountUncommitted(size_t bytes)902 void AccountUncommitted(size_t bytes) { 903 DCHECK_GE(committed_, committed_ - bytes); 904 committed_ -= bytes; 905 } 906 907 #ifdef DEBUG 908 virtual void Print() = 0; 909 #endif 910 911 protected: 912 std::unique_ptr<List<AllocationObserver*>> allocation_observers_; 913 bool allocation_observers_paused_; 914 915 private: 916 Heap* heap_; 917 AllocationSpace id_; 918 Executability executable_; 919 920 // Keeps track of committed memory in a space. 921 size_t committed_; 922 size_t max_committed_; 923 924 DISALLOW_COPY_AND_ASSIGN(Space); 925 }; 926 927 928 class MemoryChunkValidator { 929 // Computed offsets should match the compiler generated ones. 930 STATIC_ASSERT(MemoryChunk::kSizeOffset == offsetof(MemoryChunk, size_)); 931 932 // Validate our estimates on the header size. 933 STATIC_ASSERT(sizeof(MemoryChunk) <= MemoryChunk::kHeaderSize); 934 STATIC_ASSERT(sizeof(LargePage) <= MemoryChunk::kHeaderSize); 935 STATIC_ASSERT(sizeof(Page) <= MemoryChunk::kHeaderSize); 936 }; 937 938 939 // ---------------------------------------------------------------------------- 940 // All heap objects containing executable code (code objects) must be allocated 941 // from a 2 GB range of memory, so that they can call each other using 32-bit 942 // displacements. This happens automatically on 32-bit platforms, where 32-bit 943 // displacements cover the entire 4GB virtual address space. On 64-bit 944 // platforms, we support this using the CodeRange object, which reserves and 945 // manages a range of virtual memory. 946 class CodeRange { 947 public: 948 explicit CodeRange(Isolate* isolate); ~CodeRange()949 ~CodeRange() { TearDown(); } 950 951 // Reserves a range of virtual memory, but does not commit any of it. 952 // Can only be called once, at heap initialization time. 953 // Returns false on failure. 954 bool SetUp(size_t requested_size); 955 valid()956 bool valid() { return code_range_ != NULL; } start()957 Address start() { 958 DCHECK(valid()); 959 return static_cast<Address>(code_range_->address()); 960 } size()961 size_t size() { 962 DCHECK(valid()); 963 return code_range_->size(); 964 } contains(Address address)965 bool contains(Address address) { 966 if (!valid()) return false; 967 Address start = static_cast<Address>(code_range_->address()); 968 return start <= address && address < start + code_range_->size(); 969 } 970 971 // Allocates a chunk of memory from the large-object portion of 972 // the code range. On platforms with no separate code range, should 973 // not be called. 974 MUST_USE_RESULT Address AllocateRawMemory(const size_t requested_size, 975 const size_t commit_size, 976 size_t* allocated); 977 bool CommitRawMemory(Address start, size_t length); 978 bool UncommitRawMemory(Address start, size_t length); 979 void FreeRawMemory(Address buf, size_t length); 980 981 private: 982 class FreeBlock { 983 public: FreeBlock()984 FreeBlock() : start(0), size(0) {} FreeBlock(Address start_arg,size_t size_arg)985 FreeBlock(Address start_arg, size_t size_arg) 986 : start(start_arg), size(size_arg) { 987 DCHECK(IsAddressAligned(start, MemoryChunk::kAlignment)); 988 DCHECK(size >= static_cast<size_t>(Page::kPageSize)); 989 } FreeBlock(void * start_arg,size_t size_arg)990 FreeBlock(void* start_arg, size_t size_arg) 991 : start(static_cast<Address>(start_arg)), size(size_arg) { 992 DCHECK(IsAddressAligned(start, MemoryChunk::kAlignment)); 993 DCHECK(size >= static_cast<size_t>(Page::kPageSize)); 994 } 995 996 Address start; 997 size_t size; 998 }; 999 1000 // Frees the range of virtual memory, and frees the data structures used to 1001 // manage it. 1002 void TearDown(); 1003 1004 // Finds a block on the allocation list that contains at least the 1005 // requested amount of memory. If none is found, sorts and merges 1006 // the existing free memory blocks, and searches again. 1007 // If none can be found, returns false. 1008 bool GetNextAllocationBlock(size_t requested); 1009 // Compares the start addresses of two free blocks. 1010 static int CompareFreeBlockAddress(const FreeBlock* left, 1011 const FreeBlock* right); 1012 bool ReserveBlock(const size_t requested_size, FreeBlock* block); 1013 void ReleaseBlock(const FreeBlock* block); 1014 1015 Isolate* isolate_; 1016 1017 // The reserved range of virtual memory that all code objects are put in. 1018 base::VirtualMemory* code_range_; 1019 1020 // The global mutex guards free_list_ and allocation_list_ as GC threads may 1021 // access both lists concurrently to the main thread. 1022 base::Mutex code_range_mutex_; 1023 1024 // Freed blocks of memory are added to the free list. When the allocation 1025 // list is exhausted, the free list is sorted and merged to make the new 1026 // allocation list. 1027 List<FreeBlock> free_list_; 1028 1029 // Memory is allocated from the free blocks on the allocation list. 1030 // The block at current_allocation_block_index_ is the current block. 1031 List<FreeBlock> allocation_list_; 1032 int current_allocation_block_index_; 1033 1034 DISALLOW_COPY_AND_ASSIGN(CodeRange); 1035 }; 1036 1037 1038 class SkipList { 1039 public: SkipList()1040 SkipList() { Clear(); } 1041 Clear()1042 void Clear() { 1043 for (int idx = 0; idx < kSize; idx++) { 1044 starts_[idx] = reinterpret_cast<Address>(-1); 1045 } 1046 } 1047 StartFor(Address addr)1048 Address StartFor(Address addr) { return starts_[RegionNumber(addr)]; } 1049 AddObject(Address addr,int size)1050 void AddObject(Address addr, int size) { 1051 int start_region = RegionNumber(addr); 1052 int end_region = RegionNumber(addr + size - kPointerSize); 1053 for (int idx = start_region; idx <= end_region; idx++) { 1054 if (starts_[idx] > addr) { 1055 starts_[idx] = addr; 1056 } else { 1057 // In the first region, there may already be an object closer to the 1058 // start of the region. Do not change the start in that case. If this 1059 // is not the first region, you probably added overlapping objects. 1060 DCHECK_EQ(start_region, idx); 1061 } 1062 } 1063 } 1064 RegionNumber(Address addr)1065 static inline int RegionNumber(Address addr) { 1066 return (OffsetFrom(addr) & Page::kPageAlignmentMask) >> kRegionSizeLog2; 1067 } 1068 Update(Address addr,int size)1069 static void Update(Address addr, int size) { 1070 Page* page = Page::FromAddress(addr); 1071 SkipList* list = page->skip_list(); 1072 if (list == NULL) { 1073 list = new SkipList(); 1074 page->set_skip_list(list); 1075 } 1076 1077 list->AddObject(addr, size); 1078 } 1079 1080 private: 1081 static const int kRegionSizeLog2 = 13; 1082 static const int kRegionSize = 1 << kRegionSizeLog2; 1083 static const int kSize = Page::kPageSize / kRegionSize; 1084 1085 STATIC_ASSERT(Page::kPageSize % kRegionSize == 0); 1086 1087 Address starts_[kSize]; 1088 }; 1089 1090 1091 // ---------------------------------------------------------------------------- 1092 // A space acquires chunks of memory from the operating system. The memory 1093 // allocator allocates and deallocates pages for the paged heap spaces and large 1094 // pages for large object space. 1095 class MemoryAllocator { 1096 public: 1097 // Unmapper takes care of concurrently unmapping and uncommitting memory 1098 // chunks. 1099 class Unmapper { 1100 public: 1101 class UnmapFreeMemoryTask; 1102 Unmapper(MemoryAllocator * allocator)1103 explicit Unmapper(MemoryAllocator* allocator) 1104 : allocator_(allocator), 1105 pending_unmapping_tasks_semaphore_(0), 1106 concurrent_unmapping_tasks_active_(0) {} 1107 AddMemoryChunkSafe(MemoryChunk * chunk)1108 void AddMemoryChunkSafe(MemoryChunk* chunk) { 1109 if ((chunk->size() == Page::kPageSize) && 1110 (chunk->executable() != EXECUTABLE)) { 1111 AddMemoryChunkSafe<kRegular>(chunk); 1112 } else { 1113 AddMemoryChunkSafe<kNonRegular>(chunk); 1114 } 1115 } 1116 TryGetPooledMemoryChunkSafe()1117 MemoryChunk* TryGetPooledMemoryChunkSafe() { 1118 // Procedure: 1119 // (1) Try to get a chunk that was declared as pooled and already has 1120 // been uncommitted. 1121 // (2) Try to steal any memory chunk of kPageSize that would've been 1122 // unmapped. 1123 MemoryChunk* chunk = GetMemoryChunkSafe<kPooled>(); 1124 if (chunk == nullptr) { 1125 chunk = GetMemoryChunkSafe<kRegular>(); 1126 if (chunk != nullptr) { 1127 // For stolen chunks we need to manually free any allocated memory. 1128 chunk->ReleaseAllocatedMemory(); 1129 } 1130 } 1131 return chunk; 1132 } 1133 1134 void FreeQueuedChunks(); 1135 bool WaitUntilCompleted(); 1136 void TearDown(); 1137 1138 private: 1139 enum ChunkQueueType { 1140 kRegular, // Pages of kPageSize that do not live in a CodeRange and 1141 // can thus be used for stealing. 1142 kNonRegular, // Large chunks and executable chunks. 1143 kPooled, // Pooled chunks, already uncommited and ready for reuse. 1144 kNumberOfChunkQueues, 1145 }; 1146 1147 template <ChunkQueueType type> AddMemoryChunkSafe(MemoryChunk * chunk)1148 void AddMemoryChunkSafe(MemoryChunk* chunk) { 1149 base::LockGuard<base::Mutex> guard(&mutex_); 1150 if (type != kRegular || allocator_->CanFreeMemoryChunk(chunk)) { 1151 chunks_[type].push_back(chunk); 1152 } else { 1153 DCHECK_EQ(type, kRegular); 1154 delayed_regular_chunks_.push_back(chunk); 1155 } 1156 } 1157 1158 template <ChunkQueueType type> GetMemoryChunkSafe()1159 MemoryChunk* GetMemoryChunkSafe() { 1160 base::LockGuard<base::Mutex> guard(&mutex_); 1161 if (chunks_[type].empty()) return nullptr; 1162 MemoryChunk* chunk = chunks_[type].front(); 1163 chunks_[type].pop_front(); 1164 return chunk; 1165 } 1166 1167 void ReconsiderDelayedChunks(); 1168 void PerformFreeMemoryOnQueuedChunks(); 1169 1170 base::Mutex mutex_; 1171 MemoryAllocator* allocator_; 1172 std::list<MemoryChunk*> chunks_[kNumberOfChunkQueues]; 1173 // Delayed chunks cannot be processed in the current unmapping cycle because 1174 // of dependencies such as an active sweeper. 1175 // See MemoryAllocator::CanFreeMemoryChunk. 1176 std::list<MemoryChunk*> delayed_regular_chunks_; 1177 base::Semaphore pending_unmapping_tasks_semaphore_; 1178 intptr_t concurrent_unmapping_tasks_active_; 1179 1180 friend class MemoryAllocator; 1181 }; 1182 1183 enum AllocationMode { 1184 kRegular, 1185 kPooled, 1186 }; 1187 1188 enum FreeMode { 1189 kFull, 1190 kPreFreeAndQueue, 1191 kPooledAndQueue, 1192 }; 1193 1194 static size_t CodePageGuardStartOffset(); 1195 1196 static size_t CodePageGuardSize(); 1197 1198 static size_t CodePageAreaStartOffset(); 1199 1200 static size_t CodePageAreaEndOffset(); 1201 CodePageAreaSize()1202 static size_t CodePageAreaSize() { 1203 return CodePageAreaEndOffset() - CodePageAreaStartOffset(); 1204 } 1205 PageAreaSize(AllocationSpace space)1206 static size_t PageAreaSize(AllocationSpace space) { 1207 DCHECK_NE(LO_SPACE, space); 1208 return (space == CODE_SPACE) ? CodePageAreaSize() 1209 : Page::kAllocatableMemory; 1210 } 1211 1212 static intptr_t GetCommitPageSize(); 1213 1214 explicit MemoryAllocator(Isolate* isolate); 1215 1216 // Initializes its internal bookkeeping structures. 1217 // Max capacity of the total space and executable memory limit. 1218 bool SetUp(size_t max_capacity, size_t capacity_executable, 1219 size_t code_range_size); 1220 1221 void TearDown(); 1222 1223 // Allocates a Page from the allocator. AllocationMode is used to indicate 1224 // whether pooled allocation, which only works for MemoryChunk::kPageSize, 1225 // should be tried first. 1226 template <MemoryAllocator::AllocationMode alloc_mode = kRegular, 1227 typename SpaceType> 1228 Page* AllocatePage(size_t size, SpaceType* owner, Executability executable); 1229 1230 LargePage* AllocateLargePage(size_t size, LargeObjectSpace* owner, 1231 Executability executable); 1232 1233 template <MemoryAllocator::FreeMode mode = kFull> 1234 void Free(MemoryChunk* chunk); 1235 1236 bool CanFreeMemoryChunk(MemoryChunk* chunk); 1237 1238 // Returns allocated spaces in bytes. Size()1239 size_t Size() { return size_.Value(); } 1240 1241 // Returns allocated executable spaces in bytes. SizeExecutable()1242 size_t SizeExecutable() { return size_executable_.Value(); } 1243 1244 // Returns the maximum available bytes of heaps. Available()1245 size_t Available() { 1246 const size_t size = Size(); 1247 return capacity_ < size ? 0 : capacity_ - size; 1248 } 1249 1250 // Returns the maximum available executable bytes of heaps. AvailableExecutable()1251 size_t AvailableExecutable() { 1252 const size_t executable_size = SizeExecutable(); 1253 if (capacity_executable_ < executable_size) return 0; 1254 return capacity_executable_ - executable_size; 1255 } 1256 1257 // Returns maximum available bytes that the old space can have. MaxAvailable()1258 size_t MaxAvailable() { 1259 return (Available() / Page::kPageSize) * Page::kAllocatableMemory; 1260 } 1261 1262 // Returns an indication of whether a pointer is in a space that has 1263 // been allocated by this MemoryAllocator. IsOutsideAllocatedSpace(const void * address)1264 V8_INLINE bool IsOutsideAllocatedSpace(const void* address) { 1265 return address < lowest_ever_allocated_.Value() || 1266 address >= highest_ever_allocated_.Value(); 1267 } 1268 1269 // Returns a MemoryChunk in which the memory region from commit_area_size to 1270 // reserve_area_size of the chunk area is reserved but not committed, it 1271 // could be committed later by calling MemoryChunk::CommitArea. 1272 MemoryChunk* AllocateChunk(size_t reserve_area_size, size_t commit_area_size, 1273 Executability executable, Space* space); 1274 1275 void ShrinkChunk(MemoryChunk* chunk, size_t bytes_to_shrink); 1276 1277 Address ReserveAlignedMemory(size_t requested, size_t alignment, 1278 base::VirtualMemory* controller); 1279 Address AllocateAlignedMemory(size_t reserve_size, size_t commit_size, 1280 size_t alignment, Executability executable, 1281 base::VirtualMemory* controller); 1282 1283 bool CommitMemory(Address addr, size_t size, Executability executable); 1284 1285 void FreeMemory(base::VirtualMemory* reservation, Executability executable); 1286 void PartialFreeMemory(MemoryChunk* chunk, Address start_free); 1287 void FreeMemory(Address addr, size_t size, Executability executable); 1288 1289 // Commit a contiguous block of memory from the initial chunk. Assumes that 1290 // the address is not NULL, the size is greater than zero, and that the 1291 // block is contained in the initial chunk. Returns true if it succeeded 1292 // and false otherwise. 1293 bool CommitBlock(Address start, size_t size, Executability executable); 1294 1295 // Uncommit a contiguous block of memory [start..(start+size)[. 1296 // start is not NULL, the size is greater than zero, and the 1297 // block is contained in the initial chunk. Returns true if it succeeded 1298 // and false otherwise. 1299 bool UncommitBlock(Address start, size_t size); 1300 1301 // Zaps a contiguous block of memory [start..(start+size)[ thus 1302 // filling it up with a recognizable non-NULL bit pattern. 1303 void ZapBlock(Address start, size_t size); 1304 1305 MUST_USE_RESULT bool CommitExecutableMemory(base::VirtualMemory* vm, 1306 Address start, size_t commit_size, 1307 size_t reserved_size); 1308 code_range()1309 CodeRange* code_range() { return code_range_; } unmapper()1310 Unmapper* unmapper() { return &unmapper_; } 1311 1312 #ifdef DEBUG 1313 // Reports statistic info of the space. 1314 void ReportStatistics(); 1315 #endif 1316 1317 private: 1318 // PreFree logically frees the object, i.e., it takes care of the size 1319 // bookkeeping and calls the allocation callback. 1320 void PreFreeMemory(MemoryChunk* chunk); 1321 1322 // FreeMemory can be called concurrently when PreFree was executed before. 1323 void PerformFreeMemory(MemoryChunk* chunk); 1324 1325 // See AllocatePage for public interface. Note that currently we only support 1326 // pools for NOT_EXECUTABLE pages of size MemoryChunk::kPageSize. 1327 template <typename SpaceType> 1328 MemoryChunk* AllocatePagePooled(SpaceType* owner); 1329 1330 // Initializes pages in a chunk. Returns the first page address. 1331 // This function and GetChunkId() are provided for the mark-compact 1332 // collector to rebuild page headers in the from space, which is 1333 // used as a marking stack and its page headers are destroyed. 1334 Page* InitializePagesInChunk(int chunk_id, int pages_in_chunk, 1335 PagedSpace* owner); 1336 UpdateAllocatedSpaceLimits(void * low,void * high)1337 void UpdateAllocatedSpaceLimits(void* low, void* high) { 1338 // The use of atomic primitives does not guarantee correctness (wrt. 1339 // desired semantics) by default. The loop here ensures that we update the 1340 // values only if they did not change in between. 1341 void* ptr = nullptr; 1342 do { 1343 ptr = lowest_ever_allocated_.Value(); 1344 } while ((low < ptr) && !lowest_ever_allocated_.TrySetValue(ptr, low)); 1345 do { 1346 ptr = highest_ever_allocated_.Value(); 1347 } while ((high > ptr) && !highest_ever_allocated_.TrySetValue(ptr, high)); 1348 } 1349 1350 Isolate* isolate_; 1351 CodeRange* code_range_; 1352 1353 // Maximum space size in bytes. 1354 size_t capacity_; 1355 // Maximum subset of capacity_ that can be executable 1356 size_t capacity_executable_; 1357 1358 // Allocated space size in bytes. 1359 base::AtomicNumber<size_t> size_; 1360 // Allocated executable space size in bytes. 1361 base::AtomicNumber<size_t> size_executable_; 1362 1363 // We keep the lowest and highest addresses allocated as a quick way 1364 // of determining that pointers are outside the heap. The estimate is 1365 // conservative, i.e. not all addresses in 'allocated' space are allocated 1366 // to our heap. The range is [lowest, highest[, inclusive on the low end 1367 // and exclusive on the high end. 1368 base::AtomicValue<void*> lowest_ever_allocated_; 1369 base::AtomicValue<void*> highest_ever_allocated_; 1370 1371 base::VirtualMemory last_chunk_; 1372 Unmapper unmapper_; 1373 1374 friend class TestCodeRangeScope; 1375 1376 DISALLOW_IMPLICIT_CONSTRUCTORS(MemoryAllocator); 1377 }; 1378 1379 1380 // ----------------------------------------------------------------------------- 1381 // Interface for heap object iterator to be implemented by all object space 1382 // object iterators. 1383 // 1384 // NOTE: The space specific object iterators also implements the own next() 1385 // method which is used to avoid using virtual functions 1386 // iterating a specific space. 1387 1388 class V8_EXPORT_PRIVATE ObjectIterator : public Malloced { 1389 public: ~ObjectIterator()1390 virtual ~ObjectIterator() {} 1391 virtual HeapObject* Next() = 0; 1392 }; 1393 1394 template <class PAGE_TYPE> 1395 class PageIteratorImpl 1396 : public std::iterator<std::forward_iterator_tag, PAGE_TYPE> { 1397 public: PageIteratorImpl(PAGE_TYPE * p)1398 explicit PageIteratorImpl(PAGE_TYPE* p) : p_(p) {} PageIteratorImpl(const PageIteratorImpl<PAGE_TYPE> & other)1399 PageIteratorImpl(const PageIteratorImpl<PAGE_TYPE>& other) : p_(other.p_) {} 1400 PAGE_TYPE* operator*() { return p_; } 1401 bool operator==(const PageIteratorImpl<PAGE_TYPE>& rhs) { 1402 return rhs.p_ == p_; 1403 } 1404 bool operator!=(const PageIteratorImpl<PAGE_TYPE>& rhs) { 1405 return rhs.p_ != p_; 1406 } 1407 inline PageIteratorImpl<PAGE_TYPE>& operator++(); 1408 inline PageIteratorImpl<PAGE_TYPE> operator++(int); 1409 1410 private: 1411 PAGE_TYPE* p_; 1412 }; 1413 1414 typedef PageIteratorImpl<Page> PageIterator; 1415 typedef PageIteratorImpl<LargePage> LargePageIterator; 1416 1417 class PageRange { 1418 public: 1419 typedef PageIterator iterator; PageRange(Page * begin,Page * end)1420 PageRange(Page* begin, Page* end) : begin_(begin), end_(end) {} PageRange(Page * page)1421 explicit PageRange(Page* page) : PageRange(page, page->next_page()) {} begin()1422 iterator begin() { return iterator(begin_); } end()1423 iterator end() { return iterator(end_); } 1424 1425 private: 1426 Page* begin_; 1427 Page* end_; 1428 }; 1429 1430 // ----------------------------------------------------------------------------- 1431 // Heap object iterator in new/old/map spaces. 1432 // 1433 // A HeapObjectIterator iterates objects from the bottom of the given space 1434 // to its top or from the bottom of the given page to its top. 1435 // 1436 // If objects are allocated in the page during iteration the iterator may 1437 // or may not iterate over those objects. The caller must create a new 1438 // iterator in order to be sure to visit these new objects. 1439 class V8_EXPORT_PRIVATE HeapObjectIterator : public ObjectIterator { 1440 public: 1441 // Creates a new object iterator in a given space. 1442 explicit HeapObjectIterator(PagedSpace* space); 1443 explicit HeapObjectIterator(Page* page); 1444 1445 // Advance to the next object, skipping free spaces and other fillers and 1446 // skipping the special garbage section of which there is one per space. 1447 // Returns nullptr when the iteration has ended. 1448 inline HeapObject* Next() override; 1449 1450 private: 1451 // Fast (inlined) path of next(). 1452 inline HeapObject* FromCurrentPage(); 1453 1454 // Slow path of next(), goes into the next page. Returns false if the 1455 // iteration has ended. 1456 bool AdvanceToNextPage(); 1457 1458 Address cur_addr_; // Current iteration point. 1459 Address cur_end_; // End iteration point. 1460 PagedSpace* space_; 1461 PageRange page_range_; 1462 PageRange::iterator current_page_; 1463 }; 1464 1465 1466 // ----------------------------------------------------------------------------- 1467 // A space has a circular list of pages. The next page can be accessed via 1468 // Page::next_page() call. 1469 1470 // An abstraction of allocation and relocation pointers in a page-structured 1471 // space. 1472 class AllocationInfo { 1473 public: AllocationInfo()1474 AllocationInfo() : original_top_(nullptr), top_(nullptr), limit_(nullptr) {} AllocationInfo(Address top,Address limit)1475 AllocationInfo(Address top, Address limit) 1476 : original_top_(top), top_(top), limit_(limit) {} 1477 Reset(Address top,Address limit)1478 void Reset(Address top, Address limit) { 1479 original_top_ = top; 1480 set_top(top); 1481 set_limit(limit); 1482 } 1483 original_top()1484 Address original_top() { 1485 SLOW_DCHECK(top_ == NULL || 1486 (reinterpret_cast<intptr_t>(top_) & kHeapObjectTagMask) == 0); 1487 return original_top_; 1488 } 1489 INLINE(void set_top (Address top))1490 INLINE(void set_top(Address top)) { 1491 SLOW_DCHECK(top == NULL || 1492 (reinterpret_cast<intptr_t>(top) & kHeapObjectTagMask) == 0); 1493 top_ = top; 1494 } 1495 INLINE(Address top ())1496 INLINE(Address top()) const { 1497 SLOW_DCHECK(top_ == NULL || 1498 (reinterpret_cast<intptr_t>(top_) & kHeapObjectTagMask) == 0); 1499 return top_; 1500 } 1501 top_address()1502 Address* top_address() { return &top_; } 1503 INLINE(void set_limit (Address limit))1504 INLINE(void set_limit(Address limit)) { 1505 limit_ = limit; 1506 } 1507 INLINE(Address limit ())1508 INLINE(Address limit()) const { 1509 return limit_; 1510 } 1511 limit_address()1512 Address* limit_address() { return &limit_; } 1513 1514 #ifdef DEBUG VerifyPagedAllocation()1515 bool VerifyPagedAllocation() { 1516 return (Page::FromAllocationAreaAddress(top_) == 1517 Page::FromAllocationAreaAddress(limit_)) && 1518 (top_ <= limit_); 1519 } 1520 #endif 1521 1522 private: 1523 // The original top address when the allocation info was initialized. 1524 Address original_top_; 1525 // Current allocation top. 1526 Address top_; 1527 // Current allocation limit. 1528 Address limit_; 1529 }; 1530 1531 1532 // An abstraction of the accounting statistics of a page-structured space. 1533 // 1534 // The stats are only set by functions that ensure they stay balanced. These 1535 // functions increase or decrease one of the non-capacity stats in conjunction 1536 // with capacity, or else they always balance increases and decreases to the 1537 // non-capacity stats. 1538 class AllocationStats BASE_EMBEDDED { 1539 public: AllocationStats()1540 AllocationStats() { Clear(); } 1541 1542 // Zero out all the allocation statistics (i.e., no capacity). Clear()1543 void Clear() { 1544 capacity_ = 0; 1545 max_capacity_ = 0; 1546 size_ = 0; 1547 } 1548 ClearSize()1549 void ClearSize() { size_ = capacity_; } 1550 1551 // Accessors for the allocation statistics. Capacity()1552 size_t Capacity() { return capacity_; } MaxCapacity()1553 size_t MaxCapacity() { return max_capacity_; } Size()1554 size_t Size() { return size_; } 1555 1556 // Grow the space by adding available bytes. They are initially marked as 1557 // being in use (part of the size), but will normally be immediately freed, 1558 // putting them on the free list and removing them from size_. ExpandSpace(size_t bytes)1559 void ExpandSpace(size_t bytes) { 1560 DCHECK_GE(size_ + bytes, size_); 1561 DCHECK_GE(capacity_ + bytes, capacity_); 1562 capacity_ += bytes; 1563 size_ += bytes; 1564 if (capacity_ > max_capacity_) { 1565 max_capacity_ = capacity_; 1566 } 1567 } 1568 1569 // Shrink the space by removing available bytes. Since shrinking is done 1570 // during sweeping, bytes have been marked as being in use (part of the size) 1571 // and are hereby freed. ShrinkSpace(size_t bytes)1572 void ShrinkSpace(size_t bytes) { 1573 DCHECK_GE(capacity_, bytes); 1574 DCHECK_GE(size_, bytes); 1575 capacity_ -= bytes; 1576 size_ -= bytes; 1577 } 1578 AllocateBytes(size_t bytes)1579 void AllocateBytes(size_t bytes) { 1580 DCHECK_GE(size_ + bytes, size_); 1581 size_ += bytes; 1582 } 1583 DeallocateBytes(size_t bytes)1584 void DeallocateBytes(size_t bytes) { 1585 DCHECK_GE(size_, bytes); 1586 size_ -= bytes; 1587 } 1588 DecreaseCapacity(size_t bytes)1589 void DecreaseCapacity(size_t bytes) { 1590 DCHECK_GE(capacity_, bytes); 1591 DCHECK_GE(capacity_ - bytes, size_); 1592 capacity_ -= bytes; 1593 } 1594 IncreaseCapacity(size_t bytes)1595 void IncreaseCapacity(size_t bytes) { 1596 DCHECK_GE(capacity_ + bytes, capacity_); 1597 capacity_ += bytes; 1598 } 1599 1600 // Merge |other| into |this|. Merge(const AllocationStats & other)1601 void Merge(const AllocationStats& other) { 1602 DCHECK_GE(capacity_ + other.capacity_, capacity_); 1603 DCHECK_GE(size_ + other.size_, size_); 1604 capacity_ += other.capacity_; 1605 size_ += other.size_; 1606 if (other.max_capacity_ > max_capacity_) { 1607 max_capacity_ = other.max_capacity_; 1608 } 1609 } 1610 1611 private: 1612 // |capacity_|: The number of object-area bytes (i.e., not including page 1613 // bookkeeping structures) currently in the space. 1614 size_t capacity_; 1615 1616 // |max_capacity_|: The maximum capacity ever observed. 1617 size_t max_capacity_; 1618 1619 // |size_|: The number of allocated bytes. 1620 size_t size_; 1621 }; 1622 1623 // A free list maintaining free blocks of memory. The free list is organized in 1624 // a way to encourage objects allocated around the same time to be near each 1625 // other. The normal way to allocate is intended to be by bumping a 'top' 1626 // pointer until it hits a 'limit' pointer. When the limit is hit we need to 1627 // find a new space to allocate from. This is done with the free list, which is 1628 // divided up into rough categories to cut down on waste. Having finer 1629 // categories would scatter allocation more. 1630 1631 // The free list is organized in categories as follows: 1632 // kMinBlockSize-10 words (tiniest): The tiniest blocks are only used for 1633 // allocation, when categories >= small do not have entries anymore. 1634 // 11-31 words (tiny): The tiny blocks are only used for allocation, when 1635 // categories >= small do not have entries anymore. 1636 // 32-255 words (small): Used for allocating free space between 1-31 words in 1637 // size. 1638 // 256-2047 words (medium): Used for allocating free space between 32-255 words 1639 // in size. 1640 // 1048-16383 words (large): Used for allocating free space between 256-2047 1641 // words in size. 1642 // At least 16384 words (huge): This list is for objects of 2048 words or 1643 // larger. Empty pages are also added to this list. 1644 class FreeList { 1645 public: 1646 // This method returns how much memory can be allocated after freeing 1647 // maximum_freed memory. GuaranteedAllocatable(size_t maximum_freed)1648 static inline size_t GuaranteedAllocatable(size_t maximum_freed) { 1649 if (maximum_freed <= kTiniestListMax) { 1650 // Since we are not iterating over all list entries, we cannot guarantee 1651 // that we can find the maximum freed block in that free list. 1652 return 0; 1653 } else if (maximum_freed <= kTinyListMax) { 1654 return kTinyAllocationMax; 1655 } else if (maximum_freed <= kSmallListMax) { 1656 return kSmallAllocationMax; 1657 } else if (maximum_freed <= kMediumListMax) { 1658 return kMediumAllocationMax; 1659 } else if (maximum_freed <= kLargeListMax) { 1660 return kLargeAllocationMax; 1661 } 1662 return maximum_freed; 1663 } 1664 1665 explicit FreeList(PagedSpace* owner); 1666 1667 // Adds a node on the free list. The block of size {size_in_bytes} starting 1668 // at {start} is placed on the free list. The return value is the number of 1669 // bytes that were not added to the free list, because they freed memory block 1670 // was too small. Bookkeeping information will be written to the block, i.e., 1671 // its contents will be destroyed. The start address should be word aligned, 1672 // and the size should be a non-zero multiple of the word size. 1673 size_t Free(Address start, size_t size_in_bytes, FreeMode mode); 1674 1675 // Allocate a block of size {size_in_bytes} from the free list. The block is 1676 // unitialized. A failure is returned if no block is available. The size 1677 // should be a non-zero multiple of the word size. 1678 MUST_USE_RESULT HeapObject* Allocate(size_t size_in_bytes); 1679 1680 // Clear the free list. 1681 void Reset(); 1682 ResetStats()1683 void ResetStats() { 1684 wasted_bytes_.SetValue(0); 1685 ForAllFreeListCategories( 1686 [](FreeListCategory* category) { category->ResetStats(); }); 1687 } 1688 1689 // Return the number of bytes available on the free list. Available()1690 size_t Available() { 1691 size_t available = 0; 1692 ForAllFreeListCategories([&available](FreeListCategory* category) { 1693 available += category->available(); 1694 }); 1695 return available; 1696 } 1697 IsEmpty()1698 bool IsEmpty() { 1699 bool empty = true; 1700 ForAllFreeListCategories([&empty](FreeListCategory* category) { 1701 if (!category->is_empty()) empty = false; 1702 }); 1703 return empty; 1704 } 1705 1706 // Used after booting the VM. 1707 void RepairLists(Heap* heap); 1708 1709 size_t EvictFreeListItems(Page* page); 1710 bool ContainsPageFreeListItems(Page* page); 1711 owner()1712 PagedSpace* owner() { return owner_; } wasted_bytes()1713 size_t wasted_bytes() { return wasted_bytes_.Value(); } 1714 1715 template <typename Callback> ForAllFreeListCategories(FreeListCategoryType type,Callback callback)1716 void ForAllFreeListCategories(FreeListCategoryType type, Callback callback) { 1717 FreeListCategory* current = categories_[type]; 1718 while (current != nullptr) { 1719 FreeListCategory* next = current->next(); 1720 callback(current); 1721 current = next; 1722 } 1723 } 1724 1725 template <typename Callback> ForAllFreeListCategories(Callback callback)1726 void ForAllFreeListCategories(Callback callback) { 1727 for (int i = kFirstCategory; i < kNumberOfCategories; i++) { 1728 ForAllFreeListCategories(static_cast<FreeListCategoryType>(i), callback); 1729 } 1730 } 1731 1732 bool AddCategory(FreeListCategory* category); 1733 void RemoveCategory(FreeListCategory* category); 1734 void PrintCategories(FreeListCategoryType type); 1735 1736 #ifdef DEBUG 1737 size_t SumFreeLists(); 1738 bool IsVeryLong(); 1739 #endif 1740 1741 private: 1742 class FreeListCategoryIterator { 1743 public: FreeListCategoryIterator(FreeList * free_list,FreeListCategoryType type)1744 FreeListCategoryIterator(FreeList* free_list, FreeListCategoryType type) 1745 : current_(free_list->categories_[type]) {} 1746 HasNext()1747 bool HasNext() { return current_ != nullptr; } 1748 Next()1749 FreeListCategory* Next() { 1750 DCHECK(HasNext()); 1751 FreeListCategory* tmp = current_; 1752 current_ = current_->next(); 1753 return tmp; 1754 } 1755 1756 private: 1757 FreeListCategory* current_; 1758 }; 1759 1760 // The size range of blocks, in bytes. 1761 static const size_t kMinBlockSize = 3 * kPointerSize; 1762 static const size_t kMaxBlockSize = Page::kAllocatableMemory; 1763 1764 static const size_t kTiniestListMax = 0xa * kPointerSize; 1765 static const size_t kTinyListMax = 0x1f * kPointerSize; 1766 static const size_t kSmallListMax = 0xff * kPointerSize; 1767 static const size_t kMediumListMax = 0x7ff * kPointerSize; 1768 static const size_t kLargeListMax = 0x3fff * kPointerSize; 1769 static const size_t kTinyAllocationMax = kTiniestListMax; 1770 static const size_t kSmallAllocationMax = kTinyListMax; 1771 static const size_t kMediumAllocationMax = kSmallListMax; 1772 static const size_t kLargeAllocationMax = kMediumListMax; 1773 1774 FreeSpace* FindNodeFor(size_t size_in_bytes, size_t* node_size); 1775 1776 // Walks all available categories for a given |type| and tries to retrieve 1777 // a node. Returns nullptr if the category is empty. 1778 FreeSpace* FindNodeIn(FreeListCategoryType type, size_t* node_size); 1779 1780 // Tries to retrieve a node from the first category in a given |type|. 1781 // Returns nullptr if the category is empty. 1782 FreeSpace* TryFindNodeIn(FreeListCategoryType type, size_t* node_size, 1783 size_t minimum_size); 1784 1785 // Searches a given |type| for a node of at least |minimum_size|. 1786 FreeSpace* SearchForNodeInList(FreeListCategoryType type, size_t* node_size, 1787 size_t minimum_size); 1788 SelectFreeListCategoryType(size_t size_in_bytes)1789 FreeListCategoryType SelectFreeListCategoryType(size_t size_in_bytes) { 1790 if (size_in_bytes <= kTiniestListMax) { 1791 return kTiniest; 1792 } else if (size_in_bytes <= kTinyListMax) { 1793 return kTiny; 1794 } else if (size_in_bytes <= kSmallListMax) { 1795 return kSmall; 1796 } else if (size_in_bytes <= kMediumListMax) { 1797 return kMedium; 1798 } else if (size_in_bytes <= kLargeListMax) { 1799 return kLarge; 1800 } 1801 return kHuge; 1802 } 1803 1804 // The tiny categories are not used for fast allocation. SelectFastAllocationFreeListCategoryType(size_t size_in_bytes)1805 FreeListCategoryType SelectFastAllocationFreeListCategoryType( 1806 size_t size_in_bytes) { 1807 if (size_in_bytes <= kSmallAllocationMax) { 1808 return kSmall; 1809 } else if (size_in_bytes <= kMediumAllocationMax) { 1810 return kMedium; 1811 } else if (size_in_bytes <= kLargeAllocationMax) { 1812 return kLarge; 1813 } 1814 return kHuge; 1815 } 1816 top(FreeListCategoryType type)1817 FreeListCategory* top(FreeListCategoryType type) { return categories_[type]; } 1818 1819 PagedSpace* owner_; 1820 base::AtomicNumber<size_t> wasted_bytes_; 1821 FreeListCategory* categories_[kNumberOfCategories]; 1822 1823 friend class FreeListCategory; 1824 1825 DISALLOW_IMPLICIT_CONSTRUCTORS(FreeList); 1826 }; 1827 1828 // LocalAllocationBuffer represents a linear allocation area that is created 1829 // from a given {AllocationResult} and can be used to allocate memory without 1830 // synchronization. 1831 // 1832 // The buffer is properly closed upon destruction and reassignment. 1833 // Example: 1834 // { 1835 // AllocationResult result = ...; 1836 // LocalAllocationBuffer a(heap, result, size); 1837 // LocalAllocationBuffer b = a; 1838 // CHECK(!a.IsValid()); 1839 // CHECK(b.IsValid()); 1840 // // {a} is invalid now and cannot be used for further allocations. 1841 // } 1842 // // Since {b} went out of scope, the LAB is closed, resulting in creating a 1843 // // filler object for the remaining area. 1844 class LocalAllocationBuffer { 1845 public: 1846 // Indicates that a buffer cannot be used for allocations anymore. Can result 1847 // from either reassigning a buffer, or trying to construct it from an 1848 // invalid {AllocationResult}. 1849 static inline LocalAllocationBuffer InvalidBuffer(); 1850 1851 // Creates a new LAB from a given {AllocationResult}. Results in 1852 // InvalidBuffer if the result indicates a retry. 1853 static inline LocalAllocationBuffer FromResult(Heap* heap, 1854 AllocationResult result, 1855 intptr_t size); 1856 ~LocalAllocationBuffer()1857 ~LocalAllocationBuffer() { Close(); } 1858 1859 // Convert to C++11 move-semantics once allowed by the style guide. 1860 LocalAllocationBuffer(const LocalAllocationBuffer& other); 1861 LocalAllocationBuffer& operator=(const LocalAllocationBuffer& other); 1862 1863 MUST_USE_RESULT inline AllocationResult AllocateRawAligned( 1864 int size_in_bytes, AllocationAlignment alignment); 1865 IsValid()1866 inline bool IsValid() { return allocation_info_.top() != nullptr; } 1867 1868 // Try to merge LABs, which is only possible when they are adjacent in memory. 1869 // Returns true if the merge was successful, false otherwise. 1870 inline bool TryMerge(LocalAllocationBuffer* other); 1871 1872 private: 1873 LocalAllocationBuffer(Heap* heap, AllocationInfo allocation_info); 1874 1875 void Close(); 1876 1877 Heap* heap_; 1878 AllocationInfo allocation_info_; 1879 }; 1880 1881 class NewSpacePageRange { 1882 public: 1883 typedef PageRange::iterator iterator; 1884 inline NewSpacePageRange(Address start, Address limit); begin()1885 iterator begin() { return range_.begin(); } end()1886 iterator end() { return range_.end(); } 1887 1888 private: 1889 PageRange range_; 1890 }; 1891 1892 class PagedSpace : public Space { 1893 public: 1894 typedef PageIterator iterator; 1895 1896 static const intptr_t kCompactionMemoryWanted = 500 * KB; 1897 1898 // Creates a space with an id. 1899 PagedSpace(Heap* heap, AllocationSpace id, Executability executable); 1900 ~PagedSpace()1901 ~PagedSpace() override { TearDown(); } 1902 1903 // Set up the space using the given address range of virtual memory (from 1904 // the memory allocator's initial chunk) if possible. If the block of 1905 // addresses is not big enough to contain a single page-aligned page, a 1906 // fresh chunk will be allocated. 1907 bool SetUp(); 1908 1909 // Returns true if the space has been successfully set up and not 1910 // subsequently torn down. 1911 bool HasBeenSetUp(); 1912 1913 // Checks whether an object/address is in this space. 1914 inline bool Contains(Address a); 1915 inline bool Contains(Object* o); 1916 bool ContainsSlow(Address addr); 1917 1918 // Given an address occupied by a live object, return that object if it is 1919 // in this space, or a Smi if it is not. The implementation iterates over 1920 // objects in the page containing the address, the cost is linear in the 1921 // number of objects in the page. It may be slow. 1922 Object* FindObject(Address addr); 1923 1924 // During boot the free_space_map is created, and afterwards we may need 1925 // to write it into the free list nodes that were already created. 1926 void RepairFreeListsAfterDeserialization(); 1927 1928 // Prepares for a mark-compact GC. 1929 void PrepareForMarkCompact(); 1930 1931 // Current capacity without growing (Size() + Available()). Capacity()1932 size_t Capacity() { return accounting_stats_.Capacity(); } 1933 1934 // Approximate amount of physical memory committed for this space. 1935 size_t CommittedPhysicalMemory() override; 1936 1937 void ResetFreeListStatistics(); 1938 1939 // Sets the capacity, the available space and the wasted space to zero. 1940 // The stats are rebuilt during sweeping by adding each page to the 1941 // capacity and the size when it is encountered. As free spaces are 1942 // discovered during the sweeping they are subtracted from the size and added 1943 // to the available and wasted totals. ClearStats()1944 void ClearStats() { 1945 accounting_stats_.ClearSize(); 1946 free_list_.ResetStats(); 1947 ResetFreeListStatistics(); 1948 } 1949 1950 // Available bytes without growing. These are the bytes on the free list. 1951 // The bytes in the linear allocation area are not included in this total 1952 // because updating the stats would slow down allocation. New pages are 1953 // immediately added to the free list so they show up here. Available()1954 size_t Available() override { return free_list_.Available(); } 1955 1956 // Allocated bytes in this space. Garbage bytes that were not found due to 1957 // concurrent sweeping are counted as being allocated! The bytes in the 1958 // current linear allocation area (between top and limit) are also counted 1959 // here. Size()1960 size_t Size() override { return accounting_stats_.Size(); } 1961 1962 // As size, but the bytes in lazily swept pages are estimated and the bytes 1963 // in the current linear allocation area are not included. 1964 size_t SizeOfObjects() override; 1965 1966 // Wasted bytes in this space. These are just the bytes that were thrown away 1967 // due to being too small to use for allocation. Waste()1968 virtual size_t Waste() { return free_list_.wasted_bytes(); } 1969 1970 // Returns the allocation pointer in this space. top()1971 Address top() { return allocation_info_.top(); } limit()1972 Address limit() { return allocation_info_.limit(); } 1973 1974 // The allocation top address. allocation_top_address()1975 Address* allocation_top_address() { return allocation_info_.top_address(); } 1976 1977 // The allocation limit address. allocation_limit_address()1978 Address* allocation_limit_address() { 1979 return allocation_info_.limit_address(); 1980 } 1981 1982 enum UpdateSkipList { UPDATE_SKIP_LIST, IGNORE_SKIP_LIST }; 1983 1984 // Allocate the requested number of bytes in the space if possible, return a 1985 // failure object if not. Only use IGNORE_SKIP_LIST if the skip list is going 1986 // to be manually updated later. 1987 MUST_USE_RESULT inline AllocationResult AllocateRawUnaligned( 1988 int size_in_bytes, UpdateSkipList update_skip_list = UPDATE_SKIP_LIST); 1989 1990 MUST_USE_RESULT inline AllocationResult AllocateRawUnalignedSynchronized( 1991 int size_in_bytes); 1992 1993 // Allocate the requested number of bytes in the space double aligned if 1994 // possible, return a failure object if not. 1995 MUST_USE_RESULT inline AllocationResult AllocateRawAligned( 1996 int size_in_bytes, AllocationAlignment alignment); 1997 1998 // Allocate the requested number of bytes in the space and consider allocation 1999 // alignment if needed. 2000 MUST_USE_RESULT inline AllocationResult AllocateRaw( 2001 int size_in_bytes, AllocationAlignment alignment); 2002 2003 // Give a block of memory to the space's free list. It might be added to 2004 // the free list or accounted as waste. 2005 // If add_to_freelist is false then just accounting stats are updated and 2006 // no attempt to add area to free list is made. Free(Address start,size_t size_in_bytes)2007 size_t Free(Address start, size_t size_in_bytes) { 2008 size_t wasted = free_list_.Free(start, size_in_bytes, kLinkCategory); 2009 accounting_stats_.DeallocateBytes(size_in_bytes); 2010 DCHECK_GE(size_in_bytes, wasted); 2011 return size_in_bytes - wasted; 2012 } 2013 UnaccountedFree(Address start,size_t size_in_bytes)2014 size_t UnaccountedFree(Address start, size_t size_in_bytes) { 2015 size_t wasted = free_list_.Free(start, size_in_bytes, kDoNotLinkCategory); 2016 DCHECK_GE(size_in_bytes, wasted); 2017 return size_in_bytes - wasted; 2018 } 2019 ResetFreeList()2020 void ResetFreeList() { free_list_.Reset(); } 2021 2022 // Set space allocation info. SetTopAndLimit(Address top,Address limit)2023 void SetTopAndLimit(Address top, Address limit) { 2024 DCHECK(top == limit || 2025 Page::FromAddress(top) == Page::FromAddress(limit - 1)); 2026 MemoryChunk::UpdateHighWaterMark(allocation_info_.top()); 2027 allocation_info_.Reset(top, limit); 2028 } 2029 2030 void SetAllocationInfo(Address top, Address limit); 2031 2032 // Empty space allocation info, returning unused area to free list. 2033 void EmptyAllocationInfo(); 2034 2035 void MarkAllocationInfoBlack(); 2036 Allocate(int bytes)2037 void Allocate(int bytes) { accounting_stats_.AllocateBytes(bytes); } 2038 2039 void IncreaseCapacity(size_t bytes); 2040 2041 // Releases an unused page and shrinks the space. 2042 void ReleasePage(Page* page); 2043 2044 // The dummy page that anchors the linked list of pages. anchor()2045 Page* anchor() { return &anchor_; } 2046 2047 2048 #ifdef VERIFY_HEAP 2049 // Verify integrity of this space. 2050 virtual void Verify(ObjectVisitor* visitor); 2051 2052 // Overridden by subclasses to verify space-specific object 2053 // properties (e.g., only maps or free-list nodes are in map space). VerifyObject(HeapObject * obj)2054 virtual void VerifyObject(HeapObject* obj) {} 2055 #endif 2056 2057 #ifdef DEBUG 2058 // Print meta info and objects in this space. 2059 void Print() override; 2060 2061 // Reports statistics for the space 2062 void ReportStatistics(); 2063 2064 // Report code object related statistics 2065 static void ReportCodeStatistics(Isolate* isolate); 2066 static void ResetCodeStatistics(Isolate* isolate); 2067 #endif 2068 FirstPage()2069 Page* FirstPage() { return anchor_.next_page(); } LastPage()2070 Page* LastPage() { return anchor_.prev_page(); } 2071 2072 bool CanExpand(size_t size); 2073 2074 // Returns the number of total pages in this space. 2075 int CountTotalPages(); 2076 2077 // Return size of allocatable area on a page in this space. AreaSize()2078 inline int AreaSize() { return static_cast<int>(area_size_); } 2079 is_local()2080 virtual bool is_local() { return false; } 2081 2082 // Merges {other} into the current space. Note that this modifies {other}, 2083 // e.g., removes its bump pointer area and resets statistics. 2084 void MergeCompactionSpace(CompactionSpace* other); 2085 2086 // Refills the free list from the corresponding free list filled by the 2087 // sweeper. 2088 virtual void RefillFreeList(); 2089 free_list()2090 FreeList* free_list() { return &free_list_; } 2091 mutex()2092 base::Mutex* mutex() { return &space_mutex_; } 2093 2094 inline void UnlinkFreeListCategories(Page* page); 2095 inline intptr_t RelinkFreeListCategories(Page* page); 2096 begin()2097 iterator begin() { return iterator(anchor_.next_page()); } end()2098 iterator end() { return iterator(&anchor_); } 2099 2100 // Shrink immortal immovable pages of the space to be exactly the size needed 2101 // using the high water mark. 2102 void ShrinkImmortalImmovablePages(); 2103 2104 std::unique_ptr<ObjectIterator> GetObjectIterator() override; 2105 2106 protected: 2107 // PagedSpaces that should be included in snapshots have different, i.e., 2108 // smaller, initial pages. snapshotable()2109 virtual bool snapshotable() { return true; } 2110 HasPages()2111 bool HasPages() { return anchor_.next_page() != &anchor_; } 2112 2113 // Cleans up the space, frees all pages in this space except those belonging 2114 // to the initial chunk, uncommits addresses in the initial chunk. 2115 void TearDown(); 2116 2117 // Expands the space by allocating a fixed number of pages. Returns false if 2118 // it cannot allocate requested number of pages from OS, or if the hard heap 2119 // size limit has been hit. 2120 bool Expand(); 2121 2122 // Generic fast case allocation function that tries linear allocation at the 2123 // address denoted by top in allocation_info_. 2124 inline HeapObject* AllocateLinearly(int size_in_bytes); 2125 2126 // Generic fast case allocation function that tries aligned linear allocation 2127 // at the address denoted by top in allocation_info_. Writes the aligned 2128 // allocation size, which includes the filler size, to size_in_bytes. 2129 inline HeapObject* AllocateLinearlyAligned(int* size_in_bytes, 2130 AllocationAlignment alignment); 2131 2132 // If sweeping is still in progress try to sweep unswept pages. If that is 2133 // not successful, wait for the sweeper threads and re-try free-list 2134 // allocation. 2135 MUST_USE_RESULT virtual HeapObject* SweepAndRetryAllocation( 2136 int size_in_bytes); 2137 2138 // Slow path of AllocateRaw. This function is space-dependent. 2139 MUST_USE_RESULT HeapObject* SlowAllocateRaw(int size_in_bytes); 2140 2141 size_t area_size_; 2142 2143 // Accounting information for this space. 2144 AllocationStats accounting_stats_; 2145 2146 // The dummy page that anchors the double linked list of pages. 2147 Page anchor_; 2148 2149 // The space's free list. 2150 FreeList free_list_; 2151 2152 // Normal allocation information. 2153 AllocationInfo allocation_info_; 2154 2155 // Mutex guarding any concurrent access to the space. 2156 base::Mutex space_mutex_; 2157 2158 friend class IncrementalMarking; 2159 friend class MarkCompactCollector; 2160 2161 // Used in cctest. 2162 friend class HeapTester; 2163 }; 2164 2165 enum SemiSpaceId { kFromSpace = 0, kToSpace = 1 }; 2166 2167 // ----------------------------------------------------------------------------- 2168 // SemiSpace in young generation 2169 // 2170 // A SemiSpace is a contiguous chunk of memory holding page-like memory chunks. 2171 // The mark-compact collector uses the memory of the first page in the from 2172 // space as a marking stack when tracing live objects. 2173 class SemiSpace : public Space { 2174 public: 2175 typedef PageIterator iterator; 2176 2177 static void Swap(SemiSpace* from, SemiSpace* to); 2178 SemiSpace(Heap * heap,SemiSpaceId semispace)2179 SemiSpace(Heap* heap, SemiSpaceId semispace) 2180 : Space(heap, NEW_SPACE, NOT_EXECUTABLE), 2181 current_capacity_(0), 2182 maximum_capacity_(0), 2183 minimum_capacity_(0), 2184 age_mark_(nullptr), 2185 committed_(false), 2186 id_(semispace), 2187 anchor_(this), 2188 current_page_(nullptr), 2189 pages_used_(0) {} 2190 2191 inline bool Contains(HeapObject* o); 2192 inline bool Contains(Object* o); 2193 inline bool ContainsSlow(Address a); 2194 2195 void SetUp(size_t initial_capacity, size_t maximum_capacity); 2196 void TearDown(); HasBeenSetUp()2197 bool HasBeenSetUp() { return maximum_capacity_ != 0; } 2198 2199 bool Commit(); 2200 bool Uncommit(); is_committed()2201 bool is_committed() { return committed_; } 2202 2203 // Grow the semispace to the new capacity. The new capacity requested must 2204 // be larger than the current capacity and less than the maximum capacity. 2205 bool GrowTo(size_t new_capacity); 2206 2207 // Shrinks the semispace to the new capacity. The new capacity requested 2208 // must be more than the amount of used memory in the semispace and less 2209 // than the current capacity. 2210 bool ShrinkTo(size_t new_capacity); 2211 2212 bool EnsureCurrentCapacity(); 2213 2214 // Returns the start address of the first page of the space. space_start()2215 Address space_start() { 2216 DCHECK_NE(anchor_.next_page(), anchor()); 2217 return anchor_.next_page()->area_start(); 2218 } 2219 first_page()2220 Page* first_page() { return anchor_.next_page(); } current_page()2221 Page* current_page() { return current_page_; } pages_used()2222 int pages_used() { return pages_used_; } 2223 2224 // Returns one past the end address of the space. space_end()2225 Address space_end() { return anchor_.prev_page()->area_end(); } 2226 2227 // Returns the start address of the current page of the space. page_low()2228 Address page_low() { return current_page_->area_start(); } 2229 2230 // Returns one past the end address of the current page of the space. page_high()2231 Address page_high() { return current_page_->area_end(); } 2232 AdvancePage()2233 bool AdvancePage() { 2234 Page* next_page = current_page_->next_page(); 2235 // We cannot expand if we reached the maximum number of pages already. Note 2236 // that we need to account for the next page already for this check as we 2237 // could potentially fill the whole page after advancing. 2238 const bool reached_max_pages = (pages_used_ + 1) == max_pages(); 2239 if (next_page == anchor() || reached_max_pages) { 2240 return false; 2241 } 2242 current_page_ = next_page; 2243 pages_used_++; 2244 return true; 2245 } 2246 2247 // Resets the space to using the first page. 2248 void Reset(); 2249 2250 void RemovePage(Page* page); 2251 void PrependPage(Page* page); 2252 2253 // Age mark accessors. age_mark()2254 Address age_mark() { return age_mark_; } 2255 void set_age_mark(Address mark); 2256 2257 // Returns the current capacity of the semispace. current_capacity()2258 size_t current_capacity() { return current_capacity_; } 2259 2260 // Returns the maximum capacity of the semispace. maximum_capacity()2261 size_t maximum_capacity() { return maximum_capacity_; } 2262 2263 // Returns the initial capacity of the semispace. minimum_capacity()2264 size_t minimum_capacity() { return minimum_capacity_; } 2265 id()2266 SemiSpaceId id() { return id_; } 2267 2268 // Approximate amount of physical memory committed for this space. 2269 size_t CommittedPhysicalMemory() override; 2270 2271 // If we don't have these here then SemiSpace will be abstract. However 2272 // they should never be called: 2273 Size()2274 size_t Size() override { 2275 UNREACHABLE(); 2276 return 0; 2277 } 2278 SizeOfObjects()2279 size_t SizeOfObjects() override { return Size(); } 2280 Available()2281 size_t Available() override { 2282 UNREACHABLE(); 2283 return 0; 2284 } 2285 begin()2286 iterator begin() { return iterator(anchor_.next_page()); } end()2287 iterator end() { return iterator(anchor()); } 2288 2289 std::unique_ptr<ObjectIterator> GetObjectIterator() override; 2290 2291 #ifdef DEBUG 2292 void Print() override; 2293 // Validate a range of of addresses in a SemiSpace. 2294 // The "from" address must be on a page prior to the "to" address, 2295 // in the linked page order, or it must be earlier on the same page. 2296 static void AssertValidRange(Address from, Address to); 2297 #else 2298 // Do nothing. AssertValidRange(Address from,Address to)2299 inline static void AssertValidRange(Address from, Address to) {} 2300 #endif 2301 2302 #ifdef VERIFY_HEAP 2303 virtual void Verify(); 2304 #endif 2305 2306 private: 2307 void RewindPages(Page* start, int num_pages); 2308 anchor()2309 inline Page* anchor() { return &anchor_; } max_pages()2310 inline int max_pages() { 2311 return static_cast<int>(current_capacity_ / Page::kPageSize); 2312 } 2313 2314 // Copies the flags into the masked positions on all pages in the space. 2315 void FixPagesFlags(intptr_t flags, intptr_t flag_mask); 2316 2317 // The currently committed space capacity. 2318 size_t current_capacity_; 2319 2320 // The maximum capacity that can be used by this space. A space cannot grow 2321 // beyond that size. 2322 size_t maximum_capacity_; 2323 2324 // The minimum capacity for the space. A space cannot shrink below this size. 2325 size_t minimum_capacity_; 2326 2327 // Used to govern object promotion during mark-compact collection. 2328 Address age_mark_; 2329 2330 bool committed_; 2331 SemiSpaceId id_; 2332 2333 Page anchor_; 2334 Page* current_page_; 2335 int pages_used_; 2336 2337 friend class NewSpace; 2338 friend class SemiSpaceIterator; 2339 }; 2340 2341 2342 // A SemiSpaceIterator is an ObjectIterator that iterates over the active 2343 // semispace of the heap's new space. It iterates over the objects in the 2344 // semispace from a given start address (defaulting to the bottom of the 2345 // semispace) to the top of the semispace. New objects allocated after the 2346 // iterator is created are not iterated. 2347 class SemiSpaceIterator : public ObjectIterator { 2348 public: 2349 // Create an iterator over the allocated objects in the given to-space. 2350 explicit SemiSpaceIterator(NewSpace* space); 2351 2352 inline HeapObject* Next() override; 2353 2354 private: 2355 void Initialize(Address start, Address end); 2356 2357 // The current iteration point. 2358 Address current_; 2359 // The end of iteration. 2360 Address limit_; 2361 }; 2362 2363 // ----------------------------------------------------------------------------- 2364 // The young generation space. 2365 // 2366 // The new space consists of a contiguous pair of semispaces. It simply 2367 // forwards most functions to the appropriate semispace. 2368 2369 class NewSpace : public Space { 2370 public: 2371 typedef PageIterator iterator; 2372 NewSpace(Heap * heap)2373 explicit NewSpace(Heap* heap) 2374 : Space(heap, NEW_SPACE, NOT_EXECUTABLE), 2375 to_space_(heap, kToSpace), 2376 from_space_(heap, kFromSpace), 2377 reservation_(), 2378 top_on_previous_step_(0), 2379 allocated_histogram_(nullptr), 2380 promoted_histogram_(nullptr) {} 2381 2382 inline bool Contains(HeapObject* o); 2383 inline bool ContainsSlow(Address a); 2384 inline bool Contains(Object* o); 2385 2386 bool SetUp(size_t initial_semispace_capacity, size_t max_semispace_capacity); 2387 2388 // Tears down the space. Heap memory was not allocated by the space, so it 2389 // is not deallocated here. 2390 void TearDown(); 2391 2392 // True if the space has been set up but not torn down. HasBeenSetUp()2393 bool HasBeenSetUp() { 2394 return to_space_.HasBeenSetUp() && from_space_.HasBeenSetUp(); 2395 } 2396 2397 // Flip the pair of spaces. 2398 void Flip(); 2399 2400 // Grow the capacity of the semispaces. Assumes that they are not at 2401 // their maximum capacity. 2402 void Grow(); 2403 2404 // Shrink the capacity of the semispaces. 2405 void Shrink(); 2406 2407 // Return the allocated bytes in the active semispace. Size()2408 size_t Size() override { 2409 DCHECK_GE(top(), to_space_.page_low()); 2410 return to_space_.pages_used() * Page::kAllocatableMemory + 2411 static_cast<size_t>(top() - to_space_.page_low()); 2412 } 2413 SizeOfObjects()2414 size_t SizeOfObjects() override { return Size(); } 2415 2416 // Return the allocatable capacity of a semispace. Capacity()2417 size_t Capacity() { 2418 SLOW_DCHECK(to_space_.current_capacity() == from_space_.current_capacity()); 2419 return (to_space_.current_capacity() / Page::kPageSize) * 2420 Page::kAllocatableMemory; 2421 } 2422 2423 // Return the current size of a semispace, allocatable and non-allocatable 2424 // memory. TotalCapacity()2425 size_t TotalCapacity() { 2426 DCHECK(to_space_.current_capacity() == from_space_.current_capacity()); 2427 return to_space_.current_capacity(); 2428 } 2429 2430 // Committed memory for NewSpace is the committed memory of both semi-spaces 2431 // combined. CommittedMemory()2432 size_t CommittedMemory() override { 2433 return from_space_.CommittedMemory() + to_space_.CommittedMemory(); 2434 } 2435 MaximumCommittedMemory()2436 size_t MaximumCommittedMemory() override { 2437 return from_space_.MaximumCommittedMemory() + 2438 to_space_.MaximumCommittedMemory(); 2439 } 2440 2441 // Approximate amount of physical memory committed for this space. 2442 size_t CommittedPhysicalMemory() override; 2443 2444 // Return the available bytes without growing. Available()2445 size_t Available() override { 2446 DCHECK_GE(Capacity(), Size()); 2447 return Capacity() - Size(); 2448 } 2449 AllocatedSinceLastGC()2450 size_t AllocatedSinceLastGC() { 2451 bool seen_age_mark = false; 2452 Address age_mark = to_space_.age_mark(); 2453 Page* current_page = to_space_.first_page(); 2454 Page* age_mark_page = Page::FromAddress(age_mark); 2455 Page* last_page = Page::FromAddress(top() - kPointerSize); 2456 if (age_mark_page == last_page) { 2457 if (top() - age_mark >= 0) { 2458 return top() - age_mark; 2459 } 2460 // Top was reset at some point, invalidating this metric. 2461 return 0; 2462 } 2463 while (current_page != last_page) { 2464 if (current_page == age_mark_page) { 2465 seen_age_mark = true; 2466 break; 2467 } 2468 current_page = current_page->next_page(); 2469 } 2470 if (!seen_age_mark) { 2471 // Top was reset at some point, invalidating this metric. 2472 return 0; 2473 } 2474 DCHECK_GE(age_mark_page->area_end(), age_mark); 2475 size_t allocated = age_mark_page->area_end() - age_mark; 2476 DCHECK_EQ(current_page, age_mark_page); 2477 current_page = age_mark_page->next_page(); 2478 while (current_page != last_page) { 2479 allocated += Page::kAllocatableMemory; 2480 current_page = current_page->next_page(); 2481 } 2482 DCHECK_GE(top(), current_page->area_start()); 2483 allocated += top() - current_page->area_start(); 2484 DCHECK_LE(allocated, Size()); 2485 return allocated; 2486 } 2487 MovePageFromSpaceToSpace(Page * page)2488 void MovePageFromSpaceToSpace(Page* page) { 2489 DCHECK(page->InFromSpace()); 2490 from_space_.RemovePage(page); 2491 to_space_.PrependPage(page); 2492 } 2493 2494 bool Rebalance(); 2495 2496 // Return the maximum capacity of a semispace. MaximumCapacity()2497 size_t MaximumCapacity() { 2498 DCHECK(to_space_.maximum_capacity() == from_space_.maximum_capacity()); 2499 return to_space_.maximum_capacity(); 2500 } 2501 IsAtMaximumCapacity()2502 bool IsAtMaximumCapacity() { return TotalCapacity() == MaximumCapacity(); } 2503 2504 // Returns the initial capacity of a semispace. InitialTotalCapacity()2505 size_t InitialTotalCapacity() { 2506 DCHECK(to_space_.minimum_capacity() == from_space_.minimum_capacity()); 2507 return to_space_.minimum_capacity(); 2508 } 2509 2510 // Return the address of the allocation pointer in the active semispace. top()2511 Address top() { 2512 DCHECK(to_space_.current_page()->ContainsLimit(allocation_info_.top())); 2513 return allocation_info_.top(); 2514 } 2515 2516 // Return the address of the allocation pointer limit in the active semispace. limit()2517 Address limit() { 2518 DCHECK(to_space_.current_page()->ContainsLimit(allocation_info_.limit())); 2519 return allocation_info_.limit(); 2520 } 2521 2522 // Return the address of the first object in the active semispace. bottom()2523 Address bottom() { return to_space_.space_start(); } 2524 2525 // Get the age mark of the inactive semispace. age_mark()2526 Address age_mark() { return from_space_.age_mark(); } 2527 // Set the age mark in the active semispace. set_age_mark(Address mark)2528 void set_age_mark(Address mark) { to_space_.set_age_mark(mark); } 2529 2530 // The allocation top and limit address. allocation_top_address()2531 Address* allocation_top_address() { return allocation_info_.top_address(); } 2532 2533 // The allocation limit address. allocation_limit_address()2534 Address* allocation_limit_address() { 2535 return allocation_info_.limit_address(); 2536 } 2537 2538 MUST_USE_RESULT INLINE(AllocationResult AllocateRawAligned( 2539 int size_in_bytes, AllocationAlignment alignment)); 2540 2541 MUST_USE_RESULT INLINE( 2542 AllocationResult AllocateRawUnaligned(int size_in_bytes)); 2543 2544 MUST_USE_RESULT INLINE(AllocationResult AllocateRaw( 2545 int size_in_bytes, AllocationAlignment alignment)); 2546 2547 MUST_USE_RESULT inline AllocationResult AllocateRawSynchronized( 2548 int size_in_bytes, AllocationAlignment alignment); 2549 2550 // Reset the allocation pointer to the beginning of the active semispace. 2551 void ResetAllocationInfo(); 2552 2553 // When inline allocation stepping is active, either because of incremental 2554 // marking, idle scavenge, or allocation statistics gathering, we 'interrupt' 2555 // inline allocation every once in a while. This is done by setting 2556 // allocation_info_.limit to be lower than the actual limit and and increasing 2557 // it in steps to guarantee that the observers are notified periodically. 2558 void UpdateInlineAllocationLimit(int size_in_bytes); 2559 DisableInlineAllocationSteps()2560 void DisableInlineAllocationSteps() { 2561 top_on_previous_step_ = 0; 2562 UpdateInlineAllocationLimit(0); 2563 } 2564 2565 // Allows observation of inline allocation. The observer->Step() method gets 2566 // called after every step_size bytes have been allocated (approximately). 2567 // This works by adjusting the allocation limit to a lower value and adjusting 2568 // it after each step. 2569 void AddAllocationObserver(AllocationObserver* observer) override; 2570 2571 void RemoveAllocationObserver(AllocationObserver* observer) override; 2572 2573 // Get the extent of the inactive semispace (for use as a marking stack, 2574 // or to zap it). Notice: space-addresses are not necessarily on the 2575 // same page, so FromSpaceStart() might be above FromSpaceEnd(). FromSpacePageLow()2576 Address FromSpacePageLow() { return from_space_.page_low(); } FromSpacePageHigh()2577 Address FromSpacePageHigh() { return from_space_.page_high(); } FromSpaceStart()2578 Address FromSpaceStart() { return from_space_.space_start(); } FromSpaceEnd()2579 Address FromSpaceEnd() { return from_space_.space_end(); } 2580 2581 // Get the extent of the active semispace's pages' memory. ToSpaceStart()2582 Address ToSpaceStart() { return to_space_.space_start(); } ToSpaceEnd()2583 Address ToSpaceEnd() { return to_space_.space_end(); } 2584 2585 inline bool ToSpaceContainsSlow(Address a); 2586 inline bool FromSpaceContainsSlow(Address a); 2587 inline bool ToSpaceContains(Object* o); 2588 inline bool FromSpaceContains(Object* o); 2589 2590 // Try to switch the active semispace to a new, empty, page. 2591 // Returns false if this isn't possible or reasonable (i.e., there 2592 // are no pages, or the current page is already empty), or true 2593 // if successful. 2594 bool AddFreshPage(); 2595 bool AddFreshPageSynchronized(); 2596 2597 #ifdef VERIFY_HEAP 2598 // Verify the active semispace. 2599 virtual void Verify(); 2600 #endif 2601 2602 #ifdef DEBUG 2603 // Print the active semispace. Print()2604 void Print() override { to_space_.Print(); } 2605 #endif 2606 2607 // Iterates the active semispace to collect statistics. 2608 void CollectStatistics(); 2609 // Reports previously collected statistics of the active semispace. 2610 void ReportStatistics(); 2611 // Clears previously collected statistics. 2612 void ClearHistograms(); 2613 2614 // Record the allocation or promotion of a heap object. Note that we don't 2615 // record every single allocation, but only those that happen in the 2616 // to space during a scavenge GC. 2617 void RecordAllocation(HeapObject* obj); 2618 void RecordPromotion(HeapObject* obj); 2619 2620 // Return whether the operation succeded. CommitFromSpaceIfNeeded()2621 bool CommitFromSpaceIfNeeded() { 2622 if (from_space_.is_committed()) return true; 2623 return from_space_.Commit(); 2624 } 2625 UncommitFromSpace()2626 bool UncommitFromSpace() { 2627 if (!from_space_.is_committed()) return true; 2628 return from_space_.Uncommit(); 2629 } 2630 IsFromSpaceCommitted()2631 bool IsFromSpaceCommitted() { return from_space_.is_committed(); } 2632 active_space()2633 SemiSpace* active_space() { return &to_space_; } 2634 2635 void PauseAllocationObservers() override; 2636 void ResumeAllocationObservers() override; 2637 begin()2638 iterator begin() { return to_space_.begin(); } end()2639 iterator end() { return to_space_.end(); } 2640 2641 std::unique_ptr<ObjectIterator> GetObjectIterator() override; 2642 2643 private: 2644 // Update allocation info to match the current to-space page. 2645 void UpdateAllocationInfo(); 2646 2647 base::Mutex mutex_; 2648 2649 // The semispaces. 2650 SemiSpace to_space_; 2651 SemiSpace from_space_; 2652 base::VirtualMemory reservation_; 2653 2654 // Allocation pointer and limit for normal allocation and allocation during 2655 // mark-compact collection. 2656 AllocationInfo allocation_info_; 2657 2658 Address top_on_previous_step_; 2659 2660 HistogramInfo* allocated_histogram_; 2661 HistogramInfo* promoted_histogram_; 2662 2663 bool EnsureAllocation(int size_in_bytes, AllocationAlignment alignment); 2664 2665 // If we are doing inline allocation in steps, this method performs the 'step' 2666 // operation. top is the memory address of the bump pointer at the last 2667 // inline allocation (i.e. it determines the numbers of bytes actually 2668 // allocated since the last step.) new_top is the address of the bump pointer 2669 // where the next byte is going to be allocated from. top and new_top may be 2670 // different when we cross a page boundary or reset the space. 2671 void InlineAllocationStep(Address top, Address new_top, Address soon_object, 2672 size_t size); 2673 intptr_t GetNextInlineAllocationStepSize(); 2674 void StartNextInlineAllocationStep(); 2675 2676 friend class SemiSpaceIterator; 2677 }; 2678 2679 class PauseAllocationObserversScope { 2680 public: 2681 explicit PauseAllocationObserversScope(Heap* heap); 2682 ~PauseAllocationObserversScope(); 2683 2684 private: 2685 Heap* heap_; 2686 DISALLOW_COPY_AND_ASSIGN(PauseAllocationObserversScope); 2687 }; 2688 2689 // ----------------------------------------------------------------------------- 2690 // Compaction space that is used temporarily during compaction. 2691 2692 class CompactionSpace : public PagedSpace { 2693 public: CompactionSpace(Heap * heap,AllocationSpace id,Executability executable)2694 CompactionSpace(Heap* heap, AllocationSpace id, Executability executable) 2695 : PagedSpace(heap, id, executable) {} 2696 is_local()2697 bool is_local() override { return true; } 2698 2699 protected: 2700 // The space is temporary and not included in any snapshots. snapshotable()2701 bool snapshotable() override { return false; } 2702 2703 MUST_USE_RESULT HeapObject* SweepAndRetryAllocation( 2704 int size_in_bytes) override; 2705 }; 2706 2707 2708 // A collection of |CompactionSpace|s used by a single compaction task. 2709 class CompactionSpaceCollection : public Malloced { 2710 public: CompactionSpaceCollection(Heap * heap)2711 explicit CompactionSpaceCollection(Heap* heap) 2712 : old_space_(heap, OLD_SPACE, Executability::NOT_EXECUTABLE), 2713 code_space_(heap, CODE_SPACE, Executability::EXECUTABLE) {} 2714 Get(AllocationSpace space)2715 CompactionSpace* Get(AllocationSpace space) { 2716 switch (space) { 2717 case OLD_SPACE: 2718 return &old_space_; 2719 case CODE_SPACE: 2720 return &code_space_; 2721 default: 2722 UNREACHABLE(); 2723 } 2724 UNREACHABLE(); 2725 return nullptr; 2726 } 2727 2728 private: 2729 CompactionSpace old_space_; 2730 CompactionSpace code_space_; 2731 }; 2732 2733 2734 // ----------------------------------------------------------------------------- 2735 // Old object space (includes the old space of objects and code space) 2736 2737 class OldSpace : public PagedSpace { 2738 public: 2739 // Creates an old space object. The constructor does not allocate pages 2740 // from OS. OldSpace(Heap * heap,AllocationSpace id,Executability executable)2741 OldSpace(Heap* heap, AllocationSpace id, Executability executable) 2742 : PagedSpace(heap, id, executable) {} 2743 }; 2744 2745 2746 // For contiguous spaces, top should be in the space (or at the end) and limit 2747 // should be the end of the space. 2748 #define DCHECK_SEMISPACE_ALLOCATION_INFO(info, space) \ 2749 SLOW_DCHECK((space).page_low() <= (info).top() && \ 2750 (info).top() <= (space).page_high() && \ 2751 (info).limit() <= (space).page_high()) 2752 2753 2754 // ----------------------------------------------------------------------------- 2755 // Old space for all map objects 2756 2757 class MapSpace : public PagedSpace { 2758 public: 2759 // Creates a map space object. MapSpace(Heap * heap,AllocationSpace id)2760 MapSpace(Heap* heap, AllocationSpace id) 2761 : PagedSpace(heap, id, NOT_EXECUTABLE) {} 2762 RoundSizeDownToObjectAlignment(int size)2763 int RoundSizeDownToObjectAlignment(int size) override { 2764 if (base::bits::IsPowerOfTwo32(Map::kSize)) { 2765 return RoundDown(size, Map::kSize); 2766 } else { 2767 return (size / Map::kSize) * Map::kSize; 2768 } 2769 } 2770 2771 #ifdef VERIFY_HEAP 2772 void VerifyObject(HeapObject* obj) override; 2773 #endif 2774 }; 2775 2776 2777 // ----------------------------------------------------------------------------- 2778 // Large objects ( > kMaxRegularHeapObjectSize ) are allocated and 2779 // managed by the large object space. A large object is allocated from OS 2780 // heap with extra padding bytes (Page::kPageSize + Page::kObjectStartOffset). 2781 // A large object always starts at Page::kObjectStartOffset to a page. 2782 // Large objects do not move during garbage collections. 2783 2784 class LargeObjectSpace : public Space { 2785 public: 2786 typedef LargePageIterator iterator; 2787 2788 LargeObjectSpace(Heap* heap, AllocationSpace id); 2789 virtual ~LargeObjectSpace(); 2790 2791 // Initializes internal data structures. 2792 bool SetUp(); 2793 2794 // Releases internal resources, frees objects in this space. 2795 void TearDown(); 2796 ObjectSizeFor(size_t chunk_size)2797 static size_t ObjectSizeFor(size_t chunk_size) { 2798 if (chunk_size <= (Page::kPageSize + Page::kObjectStartOffset)) return 0; 2799 return chunk_size - Page::kPageSize - Page::kObjectStartOffset; 2800 } 2801 2802 // Shared implementation of AllocateRaw, AllocateRawCode and 2803 // AllocateRawFixedArray. 2804 MUST_USE_RESULT AllocationResult 2805 AllocateRaw(int object_size, Executability executable); 2806 2807 // Available bytes for objects in this space. 2808 inline size_t Available() override; 2809 Size()2810 size_t Size() override { return size_; } 2811 SizeOfObjects()2812 size_t SizeOfObjects() override { return objects_size_; } 2813 2814 // Approximate amount of physical memory committed for this space. 2815 size_t CommittedPhysicalMemory() override; 2816 PageCount()2817 int PageCount() { return page_count_; } 2818 2819 // Finds an object for a given address, returns a Smi if it is not found. 2820 // The function iterates through all objects in this space, may be slow. 2821 Object* FindObject(Address a); 2822 2823 // Finds a large object page containing the given address, returns NULL 2824 // if such a page doesn't exist. 2825 LargePage* FindPage(Address a); 2826 2827 // Clears the marking state of live objects. 2828 void ClearMarkingStateOfLiveObjects(); 2829 2830 // Frees unmarked objects. 2831 void FreeUnmarkedObjects(); 2832 2833 void InsertChunkMapEntries(LargePage* page); 2834 void RemoveChunkMapEntries(LargePage* page); 2835 void RemoveChunkMapEntries(LargePage* page, Address free_start); 2836 2837 // Checks whether a heap object is in this space; O(1). 2838 bool Contains(HeapObject* obj); 2839 // Checks whether an address is in the object area in this space. Iterates 2840 // all objects in the space. May be slow. ContainsSlow(Address addr)2841 bool ContainsSlow(Address addr) { return FindObject(addr)->IsHeapObject(); } 2842 2843 // Checks whether the space is empty. IsEmpty()2844 bool IsEmpty() { return first_page_ == NULL; } 2845 AdjustLiveBytes(int by)2846 void AdjustLiveBytes(int by) { objects_size_ += by; } 2847 first_page()2848 LargePage* first_page() { return first_page_; } 2849 2850 // Collect code statistics. 2851 void CollectCodeStatistics(); 2852 begin()2853 iterator begin() { return iterator(first_page_); } end()2854 iterator end() { return iterator(nullptr); } 2855 2856 std::unique_ptr<ObjectIterator> GetObjectIterator() override; 2857 2858 #ifdef VERIFY_HEAP 2859 virtual void Verify(); 2860 #endif 2861 2862 #ifdef DEBUG 2863 void Print() override; 2864 void ReportStatistics(); 2865 #endif 2866 2867 private: 2868 // The head of the linked list of large object chunks. 2869 LargePage* first_page_; 2870 size_t size_; // allocated bytes 2871 int page_count_; // number of chunks 2872 size_t objects_size_; // size of objects 2873 // Map MemoryChunk::kAlignment-aligned chunks to large pages covering them 2874 base::HashMap chunk_map_; 2875 2876 friend class LargeObjectIterator; 2877 }; 2878 2879 2880 class LargeObjectIterator : public ObjectIterator { 2881 public: 2882 explicit LargeObjectIterator(LargeObjectSpace* space); 2883 2884 HeapObject* Next() override; 2885 2886 private: 2887 LargePage* current_; 2888 }; 2889 2890 // Iterates over the chunks (pages and large object pages) that can contain 2891 // pointers to new space or to evacuation candidates. 2892 class MemoryChunkIterator BASE_EMBEDDED { 2893 public: 2894 inline explicit MemoryChunkIterator(Heap* heap); 2895 2896 // Return NULL when the iterator is done. 2897 inline MemoryChunk* next(); 2898 2899 private: 2900 enum State { 2901 kOldSpaceState, 2902 kMapState, 2903 kCodeState, 2904 kLargeObjectState, 2905 kFinishedState 2906 }; 2907 Heap* heap_; 2908 State state_; 2909 PageIterator old_iterator_; 2910 PageIterator code_iterator_; 2911 PageIterator map_iterator_; 2912 LargePageIterator lo_iterator_; 2913 }; 2914 2915 } // namespace internal 2916 } // namespace v8 2917 2918 #endif // V8_HEAP_SPACES_H_ 2919