blob: 2c8cbdfc32c11b02dbc32b904d42f3cef8cb85d8 [file] [log] [blame]
// Copyright 2011 the V8 project authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "src/heap/spaces.h"
#include <cinttypes>
#include <utility>
#include "src/base/bits.h"
#include "src/base/macros.h"
#include "src/base/platform/semaphore.h"
#include "src/base/template-utils.h"
#include "src/execution/vm-state-inl.h"
#include "src/heap/array-buffer-tracker.h"
#include "src/heap/combined-heap.h"
#include "src/heap/concurrent-marking.h"
#include "src/heap/gc-tracer.h"
#include "src/heap/heap-controller.h"
#include "src/heap/incremental-marking-inl.h"
#include "src/heap/mark-compact.h"
#include "src/heap/read-only-heap.h"
#include "src/heap/remembered-set.h"
#include "src/heap/slot-set.h"
#include "src/heap/sweeper.h"
#include "src/init/v8.h"
#include "src/logging/counters.h"
#include "src/objects/free-space-inl.h"
#include "src/objects/js-array-buffer-inl.h"
#include "src/objects/objects-inl.h"
#include "src/sanitizer/msan.h"
#include "src/snapshot/snapshot.h"
#include "src/utils/ostreams.h"
namespace v8 {
namespace internal {
// These checks are here to ensure that the lower 32 bits of any real heap
// object can't overlap with the lower 32 bits of cleared weak reference value
// and therefore it's enough to compare only the lower 32 bits of a MaybeObject
// in order to figure out if it's a cleared weak reference or not.
STATIC_ASSERT(kClearedWeakHeapObjectLower32 > 0);
STATIC_ASSERT(kClearedWeakHeapObjectLower32 < Page::kHeaderSize);
STATIC_ASSERT(kClearedWeakHeapObjectLower32 < LargePage::kHeaderSize);
// ----------------------------------------------------------------------------
// HeapObjectIterator
HeapObjectIterator::HeapObjectIterator(PagedSpace* space)
: cur_addr_(kNullAddress),
cur_end_(kNullAddress),
space_(space),
page_range_(space->first_page(), nullptr),
current_page_(page_range_.begin()) {
#ifdef V8_SHARED_RO_HEAP
DCHECK_NE(space->identity(), RO_SPACE);
#endif
}
HeapObjectIterator::HeapObjectIterator(Page* page)
: cur_addr_(kNullAddress),
cur_end_(kNullAddress),
space_(reinterpret_cast<PagedSpace*>(page->owner())),
page_range_(page),
current_page_(page_range_.begin()) {
#ifdef DEBUG
Space* owner = page->owner();
// TODO(v8:7464): Always enforce this once PagedSpace::Verify is no longer
// used to verify read-only space for non-shared builds.
#ifdef V8_SHARED_RO_HEAP
DCHECK_NE(owner->identity(), RO_SPACE);
#endif
// Do not access the heap of the read-only space.
DCHECK(owner->identity() == RO_SPACE || owner->identity() == OLD_SPACE ||
owner->identity() == MAP_SPACE || owner->identity() == CODE_SPACE);
#endif // DEBUG
}
// We have hit the end of the page and should advance to the next block of
// objects. This happens at the end of the page.
bool HeapObjectIterator::AdvanceToNextPage() {
DCHECK_EQ(cur_addr_, cur_end_);
if (current_page_ == page_range_.end()) return false;
Page* cur_page = *(current_page_++);
#ifdef ENABLE_MINOR_MC
Heap* heap = space_->heap();
heap->mark_compact_collector()->sweeper()->EnsurePageIsIterable(cur_page);
if (cur_page->IsFlagSet(Page::SWEEP_TO_ITERATE)) {
heap->minor_mark_compact_collector()->MakeIterable(
cur_page, MarkingTreatmentMode::CLEAR,
FreeSpaceTreatmentMode::IGNORE_FREE_SPACE);
}
#else
DCHECK(!cur_page->IsFlagSet(Page::SWEEP_TO_ITERATE));
#endif // ENABLE_MINOR_MC
cur_addr_ = cur_page->area_start();
cur_end_ = cur_page->area_end();
DCHECK(cur_page->SweepingDone());
return true;
}
PauseAllocationObserversScope::PauseAllocationObserversScope(Heap* heap)
: heap_(heap) {
DCHECK_EQ(heap->gc_state(), Heap::NOT_IN_GC);
for (SpaceIterator it(heap_); it.has_next();) {
it.next()->PauseAllocationObservers();
}
}
PauseAllocationObserversScope::~PauseAllocationObserversScope() {
for (SpaceIterator it(heap_); it.has_next();) {
it.next()->ResumeAllocationObservers();
}
}
static base::LazyInstance<CodeRangeAddressHint>::type code_range_address_hint =
LAZY_INSTANCE_INITIALIZER;
Address CodeRangeAddressHint::GetAddressHint(size_t code_range_size) {
base::MutexGuard guard(&mutex_);
auto it = recently_freed_.find(code_range_size);
if (it == recently_freed_.end() || it->second.empty()) {
return reinterpret_cast<Address>(GetRandomMmapAddr());
}
Address result = it->second.back();
it->second.pop_back();
return result;
}
void CodeRangeAddressHint::NotifyFreedCodeRange(Address code_range_start,
size_t code_range_size) {
base::MutexGuard guard(&mutex_);
recently_freed_[code_range_size].push_back(code_range_start);
}
// -----------------------------------------------------------------------------
// MemoryAllocator
//
MemoryAllocator::MemoryAllocator(Isolate* isolate, size_t capacity,
size_t code_range_size)
: isolate_(isolate),
data_page_allocator_(isolate->page_allocator()),
code_page_allocator_(nullptr),
capacity_(RoundUp(capacity, Page::kPageSize)),
size_(0),
size_executable_(0),
lowest_ever_allocated_(static_cast<Address>(-1ll)),
highest_ever_allocated_(kNullAddress),
unmapper_(isolate->heap(), this) {
InitializeCodePageAllocator(data_page_allocator_, code_range_size);
}
void MemoryAllocator::InitializeCodePageAllocator(
v8::PageAllocator* page_allocator, size_t requested) {
DCHECK_NULL(code_page_allocator_instance_.get());
code_page_allocator_ = page_allocator;
if (requested == 0) {
if (!kRequiresCodeRange) return;
// When a target requires the code range feature, we put all code objects
// in a kMaximalCodeRangeSize range of virtual address space, so that
// they can call each other with near calls.
requested = kMaximalCodeRangeSize;
} else if (requested <= kMinimumCodeRangeSize) {
requested = kMinimumCodeRangeSize;
}
const size_t reserved_area =
kReservedCodeRangePages * MemoryAllocator::GetCommitPageSize();
if (requested < (kMaximalCodeRangeSize - reserved_area)) {
requested += RoundUp(reserved_area, MemoryChunk::kPageSize);
// Fullfilling both reserved pages requirement and huge code area
// alignments is not supported (requires re-implementation).
DCHECK_LE(kMinExpectedOSPageSize, page_allocator->AllocatePageSize());
}
DCHECK(!kRequiresCodeRange || requested <= kMaximalCodeRangeSize);
Address hint =
RoundDown(code_range_address_hint.Pointer()->GetAddressHint(requested),
page_allocator->AllocatePageSize());
VirtualMemory reservation(
page_allocator, requested, reinterpret_cast<void*>(hint),
Max(kMinExpectedOSPageSize, page_allocator->AllocatePageSize()));
if (!reservation.IsReserved()) {
V8::FatalProcessOutOfMemory(isolate_,
"CodeRange setup: allocate virtual memory");
}
code_range_ = reservation.region();
// We are sure that we have mapped a block of requested addresses.
DCHECK_GE(reservation.size(), requested);
Address base = reservation.address();
// On some platforms, specifically Win64, we need to reserve some pages at
// the beginning of an executable space. See
// https://cs.chromium.org/chromium/src/components/crash/content/
// app/crashpad_win.cc?rcl=fd680447881449fba2edcf0589320e7253719212&l=204
// for details.
if (reserved_area > 0) {
if (!reservation.SetPermissions(base, reserved_area,
PageAllocator::kReadWrite))
V8::FatalProcessOutOfMemory(isolate_, "CodeRange setup: set permissions");
base += reserved_area;
}
Address aligned_base = RoundUp(base, MemoryChunk::kAlignment);
size_t size =
RoundDown(reservation.size() - (aligned_base - base) - reserved_area,
MemoryChunk::kPageSize);
DCHECK(IsAligned(aligned_base, kMinExpectedOSPageSize));
LOG(isolate_,
NewEvent("CodeRange", reinterpret_cast<void*>(reservation.address()),
requested));
heap_reservation_ = std::move(reservation);
code_page_allocator_instance_ = base::make_unique<base::BoundedPageAllocator>(
page_allocator, aligned_base, size,
static_cast<size_t>(MemoryChunk::kAlignment));
code_page_allocator_ = code_page_allocator_instance_.get();
}
void MemoryAllocator::TearDown() {
unmapper()->TearDown();
// Check that spaces were torn down before MemoryAllocator.
DCHECK_EQ(size_, 0u);
// TODO(gc) this will be true again when we fix FreeMemory.
// DCHECK_EQ(0, size_executable_);
capacity_ = 0;
if (last_chunk_.IsReserved()) {
last_chunk_.Free();
}
if (code_page_allocator_instance_.get()) {
DCHECK(!code_range_.is_empty());
code_range_address_hint.Pointer()->NotifyFreedCodeRange(code_range_.begin(),
code_range_.size());
code_range_ = base::AddressRegion();
code_page_allocator_instance_.reset();
}
code_page_allocator_ = nullptr;
data_page_allocator_ = nullptr;
}
class MemoryAllocator::Unmapper::UnmapFreeMemoryTask : public CancelableTask {
public:
explicit UnmapFreeMemoryTask(Isolate* isolate, Unmapper* unmapper)
: CancelableTask(isolate),
unmapper_(unmapper),
tracer_(isolate->heap()->tracer()) {}
private:
void RunInternal() override {
TRACE_BACKGROUND_GC(tracer_,
GCTracer::BackgroundScope::BACKGROUND_UNMAPPER);
unmapper_->PerformFreeMemoryOnQueuedChunks<FreeMode::kUncommitPooled>();
unmapper_->active_unmapping_tasks_--;
unmapper_->pending_unmapping_tasks_semaphore_.Signal();
if (FLAG_trace_unmapper) {
PrintIsolate(unmapper_->heap_->isolate(),
"UnmapFreeMemoryTask Done: id=%" PRIu64 "\n", id());
}
}
Unmapper* const unmapper_;
GCTracer* const tracer_;
DISALLOW_COPY_AND_ASSIGN(UnmapFreeMemoryTask);
};
void MemoryAllocator::Unmapper::FreeQueuedChunks() {
if (!heap_->IsTearingDown() && FLAG_concurrent_sweeping) {
if (!MakeRoomForNewTasks()) {
// kMaxUnmapperTasks are already running. Avoid creating any more.
if (FLAG_trace_unmapper) {
PrintIsolate(heap_->isolate(),
"Unmapper::FreeQueuedChunks: reached task limit (%d)\n",
kMaxUnmapperTasks);
}
return;
}
auto task = base::make_unique<UnmapFreeMemoryTask>(heap_->isolate(), this);
if (FLAG_trace_unmapper) {
PrintIsolate(heap_->isolate(),
"Unmapper::FreeQueuedChunks: new task id=%" PRIu64 "\n",
task->id());
}
DCHECK_LT(pending_unmapping_tasks_, kMaxUnmapperTasks);
DCHECK_LE(active_unmapping_tasks_, pending_unmapping_tasks_);
DCHECK_GE(active_unmapping_tasks_, 0);
active_unmapping_tasks_++;
task_ids_[pending_unmapping_tasks_++] = task->id();
V8::GetCurrentPlatform()->CallOnWorkerThread(std::move(task));
} else {
PerformFreeMemoryOnQueuedChunks<FreeMode::kUncommitPooled>();
}
}
void MemoryAllocator::Unmapper::CancelAndWaitForPendingTasks() {
for (int i = 0; i < pending_unmapping_tasks_; i++) {
if (heap_->isolate()->cancelable_task_manager()->TryAbort(task_ids_[i]) !=
TryAbortResult::kTaskAborted) {
pending_unmapping_tasks_semaphore_.Wait();
}
}
pending_unmapping_tasks_ = 0;
active_unmapping_tasks_ = 0;
if (FLAG_trace_unmapper) {
PrintIsolate(
heap_->isolate(),
"Unmapper::CancelAndWaitForPendingTasks: no tasks remaining\n");
}
}
void MemoryAllocator::Unmapper::PrepareForGC() {
// Free non-regular chunks because they cannot be re-used.
PerformFreeMemoryOnQueuedNonRegularChunks();
}
void MemoryAllocator::Unmapper::EnsureUnmappingCompleted() {
CancelAndWaitForPendingTasks();
PerformFreeMemoryOnQueuedChunks<FreeMode::kReleasePooled>();
}
bool MemoryAllocator::Unmapper::MakeRoomForNewTasks() {
DCHECK_LE(pending_unmapping_tasks_, kMaxUnmapperTasks);
if (active_unmapping_tasks_ == 0 && pending_unmapping_tasks_ > 0) {
// All previous unmapping tasks have been run to completion.
// Finalize those tasks to make room for new ones.
CancelAndWaitForPendingTasks();
}
return pending_unmapping_tasks_ != kMaxUnmapperTasks;
}
void MemoryAllocator::Unmapper::PerformFreeMemoryOnQueuedNonRegularChunks() {
MemoryChunk* chunk = nullptr;
while ((chunk = GetMemoryChunkSafe<kNonRegular>()) != nullptr) {
allocator_->PerformFreeMemory(chunk);
}
}
template <MemoryAllocator::Unmapper::FreeMode mode>
void MemoryAllocator::Unmapper::PerformFreeMemoryOnQueuedChunks() {
MemoryChunk* chunk = nullptr;
if (FLAG_trace_unmapper) {
PrintIsolate(
heap_->isolate(),
"Unmapper::PerformFreeMemoryOnQueuedChunks: %d queued chunks\n",
NumberOfChunks());
}
// Regular chunks.
while ((chunk = GetMemoryChunkSafe<kRegular>()) != nullptr) {
bool pooled = chunk->IsFlagSet(MemoryChunk::POOLED);
allocator_->PerformFreeMemory(chunk);
if (pooled) AddMemoryChunkSafe<kPooled>(chunk);
}
if (mode == MemoryAllocator::Unmapper::FreeMode::kReleasePooled) {
// The previous loop uncommitted any pages marked as pooled and added them
// to the pooled list. In case of kReleasePooled we need to free them
// though.
while ((chunk = GetMemoryChunkSafe<kPooled>()) != nullptr) {
allocator_->Free<MemoryAllocator::kAlreadyPooled>(chunk);
}
}
PerformFreeMemoryOnQueuedNonRegularChunks();
}
void MemoryAllocator::Unmapper::TearDown() {
CHECK_EQ(0, pending_unmapping_tasks_);
PerformFreeMemoryOnQueuedChunks<FreeMode::kReleasePooled>();
for (int i = 0; i < kNumberOfChunkQueues; i++) {
DCHECK(chunks_[i].empty());
}
}
size_t MemoryAllocator::Unmapper::NumberOfCommittedChunks() {
base::MutexGuard guard(&mutex_);
return chunks_[kRegular].size() + chunks_[kNonRegular].size();
}
int MemoryAllocator::Unmapper::NumberOfChunks() {
base::MutexGuard guard(&mutex_);
size_t result = 0;
for (int i = 0; i < kNumberOfChunkQueues; i++) {
result += chunks_[i].size();
}
return static_cast<int>(result);
}
size_t MemoryAllocator::Unmapper::CommittedBufferedMemory() {
base::MutexGuard guard(&mutex_);
size_t sum = 0;
// kPooled chunks are already uncommited. We only have to account for
// kRegular and kNonRegular chunks.
for (auto& chunk : chunks_[kRegular]) {
sum += chunk->size();
}
for (auto& chunk : chunks_[kNonRegular]) {
sum += chunk->size();
}
return sum;
}
bool MemoryAllocator::CommitMemory(VirtualMemory* reservation) {
Address base = reservation->address();
size_t size = reservation->size();
if (!reservation->SetPermissions(base, size, PageAllocator::kReadWrite)) {
return false;
}
UpdateAllocatedSpaceLimits(base, base + size);
isolate_->counters()->memory_allocated()->Increment(static_cast<int>(size));
return true;
}
bool MemoryAllocator::UncommitMemory(VirtualMemory* reservation) {
size_t size = reservation->size();
if (!reservation->SetPermissions(reservation->address(), size,
PageAllocator::kNoAccess)) {
return false;
}
isolate_->counters()->memory_allocated()->Decrement(static_cast<int>(size));
return true;
}
void MemoryAllocator::FreeMemory(v8::PageAllocator* page_allocator,
Address base, size_t size) {
CHECK(FreePages(page_allocator, reinterpret_cast<void*>(base), size));
}
Address MemoryAllocator::AllocateAlignedMemory(
size_t reserve_size, size_t commit_size, size_t alignment,
Executability executable, void* hint, VirtualMemory* controller) {
v8::PageAllocator* page_allocator = this->page_allocator(executable);
DCHECK(commit_size <= reserve_size);
VirtualMemory reservation(page_allocator, reserve_size, hint, alignment);
if (!reservation.IsReserved()) return kNullAddress;
Address base = reservation.address();
size_ += reservation.size();
if (executable == EXECUTABLE) {
if (!CommitExecutableMemory(&reservation, base, commit_size,
reserve_size)) {
base = kNullAddress;
}
} else {
if (reservation.SetPermissions(base, commit_size,
PageAllocator::kReadWrite)) {
UpdateAllocatedSpaceLimits(base, base + commit_size);
} else {
base = kNullAddress;
}
}
if (base == kNullAddress) {
// Failed to commit the body. Free the mapping and any partially committed
// regions inside it.
reservation.Free();
size_ -= reserve_size;
return kNullAddress;
}
*controller = std::move(reservation);
return base;
}
void MemoryChunk::DiscardUnusedMemory(Address addr, size_t size) {
base::AddressRegion memory_area =
MemoryAllocator::ComputeDiscardMemoryArea(addr, size);
if (memory_area.size() != 0) {
MemoryAllocator* memory_allocator = heap_->memory_allocator();
v8::PageAllocator* page_allocator =
memory_allocator->page_allocator(executable());
CHECK(page_allocator->DiscardSystemPages(
reinterpret_cast<void*>(memory_area.begin()), memory_area.size()));
}
}
size_t MemoryChunkLayout::CodePageGuardStartOffset() {
// We are guarding code pages: the first OS page after the header
// will be protected as non-writable.
return ::RoundUp(Page::kHeaderSize, MemoryAllocator::GetCommitPageSize());
}
size_t MemoryChunkLayout::CodePageGuardSize() {
return MemoryAllocator::GetCommitPageSize();
}
intptr_t MemoryChunkLayout::ObjectStartOffsetInCodePage() {
// We are guarding code pages: the first OS page after the header
// will be protected as non-writable.
return CodePageGuardStartOffset() + CodePageGuardSize();
}
intptr_t MemoryChunkLayout::ObjectEndOffsetInCodePage() {
// We are guarding code pages: the last OS page will be protected as
// non-writable.
return Page::kPageSize -
static_cast<int>(MemoryAllocator::GetCommitPageSize());
}
size_t MemoryChunkLayout::AllocatableMemoryInCodePage() {
size_t memory = ObjectEndOffsetInCodePage() - ObjectStartOffsetInCodePage();
DCHECK_LE(kMaxRegularHeapObjectSize, memory);
return memory;
}
intptr_t MemoryChunkLayout::ObjectStartOffsetInDataPage() {
return RoundUp(MemoryChunk::kHeaderSize, kTaggedSize);
}
size_t MemoryChunkLayout::ObjectStartOffsetInMemoryChunk(
AllocationSpace space) {
if (space == CODE_SPACE) {
return ObjectStartOffsetInCodePage();
}
return ObjectStartOffsetInDataPage();
}
size_t MemoryChunkLayout::AllocatableMemoryInDataPage() {
size_t memory = MemoryChunk::kPageSize - ObjectStartOffsetInDataPage();
DCHECK_LE(kMaxRegularHeapObjectSize, memory);
return memory;
}
size_t MemoryChunkLayout::AllocatableMemoryInMemoryChunk(
AllocationSpace space) {
if (space == CODE_SPACE) {
return AllocatableMemoryInCodePage();
}
return AllocatableMemoryInDataPage();
}
Heap* MemoryChunk::synchronized_heap() {
return reinterpret_cast<Heap*>(
base::Acquire_Load(reinterpret_cast<base::AtomicWord*>(&heap_)));
}
void MemoryChunk::InitializationMemoryFence() {
base::SeqCst_MemoryFence();
#ifdef THREAD_SANITIZER
// Since TSAN does not process memory fences, we use the following annotation
// to tell TSAN that there is no data race when emitting a
// InitializationMemoryFence. Note that the other thread still needs to
// perform MemoryChunk::synchronized_heap().
base::Release_Store(reinterpret_cast<base::AtomicWord*>(&heap_),
reinterpret_cast<base::AtomicWord>(heap_));
#endif
}
void MemoryChunk::DecrementWriteUnprotectCounterAndMaybeSetPermissions(
PageAllocator::Permission permission) {
DCHECK(permission == PageAllocator::kRead ||
permission == PageAllocator::kReadExecute);
DCHECK(IsFlagSet(MemoryChunk::IS_EXECUTABLE));
DCHECK(owner()->identity() == CODE_SPACE ||
owner()->identity() == CODE_LO_SPACE);
// Decrementing the write_unprotect_counter_ and changing the page
// protection mode has to be atomic.
base::MutexGuard guard(page_protection_change_mutex_);
if (write_unprotect_counter_ == 0) {
// This is a corner case that may happen when we have a
// CodeSpaceMemoryModificationScope open and this page was newly
// added.
return;
}
write_unprotect_counter_--;
DCHECK_LT(write_unprotect_counter_, kMaxWriteUnprotectCounter);
if (write_unprotect_counter_ == 0) {
Address protect_start =
address() + MemoryChunkLayout::ObjectStartOffsetInCodePage();
size_t page_size = MemoryAllocator::GetCommitPageSize();
DCHECK(IsAligned(protect_start, page_size));
size_t protect_size = RoundUp(area_size(), page_size);
CHECK(reservation_.SetPermissions(protect_start, protect_size, permission));
}
}
void MemoryChunk::SetReadable() {
DecrementWriteUnprotectCounterAndMaybeSetPermissions(PageAllocator::kRead);
}
void MemoryChunk::SetReadAndExecutable() {
DCHECK(!FLAG_jitless);
DecrementWriteUnprotectCounterAndMaybeSetPermissions(
PageAllocator::kReadExecute);
}
void MemoryChunk::SetReadAndWritable() {
DCHECK(IsFlagSet(MemoryChunk::IS_EXECUTABLE));
DCHECK(owner()->identity() == CODE_SPACE ||
owner()->identity() == CODE_LO_SPACE);
// Incrementing the write_unprotect_counter_ and changing the page
// protection mode has to be atomic.
base::MutexGuard guard(page_protection_change_mutex_);
write_unprotect_counter_++;
DCHECK_LE(write_unprotect_counter_, kMaxWriteUnprotectCounter);
if (write_unprotect_counter_ == 1) {
Address unprotect_start =
address() + MemoryChunkLayout::ObjectStartOffsetInCodePage();
size_t page_size = MemoryAllocator::GetCommitPageSize();
DCHECK(IsAligned(unprotect_start, page_size));
size_t unprotect_size = RoundUp(area_size(), page_size);
CHECK(reservation_.SetPermissions(unprotect_start, unprotect_size,
PageAllocator::kReadWrite));
}
}
void CodeObjectRegistry::RegisterNewlyAllocatedCodeObject(Address code) {
auto result = code_object_registry_newly_allocated_.insert(code);
USE(result);
DCHECK(result.second);
}
void CodeObjectRegistry::RegisterAlreadyExistingCodeObject(Address code) {
code_object_registry_already_existing_.push_back(code);
}
void CodeObjectRegistry::Clear() {
code_object_registry_already_existing_.clear();
code_object_registry_newly_allocated_.clear();
}
void CodeObjectRegistry::Finalize() {
code_object_registry_already_existing_.shrink_to_fit();
}
bool CodeObjectRegistry::Contains(Address object) const {
return (code_object_registry_newly_allocated_.find(object) !=
code_object_registry_newly_allocated_.end()) ||
(std::binary_search(code_object_registry_already_existing_.begin(),
code_object_registry_already_existing_.end(),
object));
}
Address CodeObjectRegistry::GetCodeObjectStartFromInnerAddress(
Address address) const {
// Let's first find the object which comes right before address in the vector
// of already existing code objects.
Address already_existing_set_ = 0;
Address newly_allocated_set_ = 0;
if (!code_object_registry_already_existing_.empty()) {
auto it =
std::upper_bound(code_object_registry_already_existing_.begin(),
code_object_registry_already_existing_.end(), address);
if (it != code_object_registry_already_existing_.begin()) {
already_existing_set_ = *(--it);
}
}
// Next, let's find the object which comes right before address in the set
// of newly allocated code objects.
if (!code_object_registry_newly_allocated_.empty()) {
auto it = code_object_registry_newly_allocated_.upper_bound(address);
if (it != code_object_registry_newly_allocated_.begin()) {
newly_allocated_set_ = *(--it);
}
}
// The code objects which contains address has to be in one of the two
// data structures.
DCHECK(already_existing_set_ != 0 || newly_allocated_set_ != 0);
// The address which is closest to the given address is the code object.
return already_existing_set_ > newly_allocated_set_ ? already_existing_set_
: newly_allocated_set_;
}
namespace {
PageAllocator::Permission DefaultWritableCodePermissions() {
return FLAG_jitless ? PageAllocator::kReadWrite
: PageAllocator::kReadWriteExecute;
}
} // namespace
MemoryChunk* MemoryChunk::Initialize(Heap* heap, Address base, size_t size,
Address area_start, Address area_end,
Executability executable, Space* owner,
VirtualMemory reservation) {
MemoryChunk* chunk = FromAddress(base);
DCHECK_EQ(base, chunk->address());
chunk->heap_ = heap;
chunk->size_ = size;
chunk->header_sentinel_ = HeapObject::FromAddress(base).ptr();
DCHECK(HasHeaderSentinel(area_start));
chunk->area_start_ = area_start;
chunk->area_end_ = area_end;
chunk->flags_ = Flags(NO_FLAGS);
chunk->set_owner(owner);
chunk->InitializeReservedMemory();
base::AsAtomicPointer::Release_Store(&chunk->slot_set_[OLD_TO_NEW], nullptr);
base::AsAtomicPointer::Release_Store(&chunk->slot_set_[OLD_TO_OLD], nullptr);
base::AsAtomicPointer::Release_Store(&chunk->typed_slot_set_[OLD_TO_NEW],
nullptr);
base::AsAtomicPointer::Release_Store(&chunk->typed_slot_set_[OLD_TO_OLD],
nullptr);
chunk->invalidated_slots_ = nullptr;
chunk->progress_bar_ = 0;
chunk->high_water_mark_ = static_cast<intptr_t>(area_start - base);
chunk->set_concurrent_sweeping_state(kSweepingDone);
chunk->page_protection_change_mutex_ = new base::Mutex();
chunk->write_unprotect_counter_ = 0;
chunk->mutex_ = new base::Mutex();
chunk->allocated_bytes_ = chunk->area_size();
chunk->wasted_memory_ = 0;
chunk->young_generation_bitmap_ = nullptr;
chunk->marking_bitmap_ = nullptr;
chunk->local_tracker_ = nullptr;
chunk->external_backing_store_bytes_[ExternalBackingStoreType::kArrayBuffer] =
0;
chunk->external_backing_store_bytes_
[ExternalBackingStoreType::kExternalString] = 0;
for (int i = kFirstCategory; i < kNumberOfCategories; i++) {
chunk->categories_[i] = nullptr;
}
chunk->AllocateMarkingBitmap();
if (owner->identity() == RO_SPACE) {
heap->incremental_marking()
->non_atomic_marking_state()
->bitmap(chunk)
->MarkAllBits();
} else {
heap->incremental_marking()->non_atomic_marking_state()->SetLiveBytes(chunk,
0);
}
DCHECK_EQ(kFlagsOffset, OFFSET_OF(MemoryChunk, flags_));
DCHECK_EQ(kHeapOffset, OFFSET_OF(MemoryChunk, heap_));
DCHECK_EQ(kOwnerOffset, OFFSET_OF(MemoryChunk, owner_));
if (executable == EXECUTABLE) {
chunk->SetFlag(IS_EXECUTABLE);
if (heap->write_protect_code_memory()) {
chunk->write_unprotect_counter_ =
heap->code_space_memory_modification_scope_depth();
} else {
size_t page_size = MemoryAllocator::GetCommitPageSize();
DCHECK(IsAligned(area_start, page_size));
size_t area_size = RoundUp(area_end - area_start, page_size);
CHECK(reservation.SetPermissions(area_start, area_size,
DefaultWritableCodePermissions()));
}
}
chunk->reservation_ = std::move(reservation);
if (owner->identity() == CODE_SPACE) {
chunk->code_object_registry_ = new CodeObjectRegistry();
} else {
chunk->code_object_registry_ = nullptr;
}
return chunk;
}
Page* PagedSpace::InitializePage(MemoryChunk* chunk, Executability executable) {
Page* page = static_cast<Page*>(chunk);
DCHECK_EQ(MemoryChunkLayout::AllocatableMemoryInMemoryChunk(
page->owner()->identity()),
page->area_size());
// Make sure that categories are initialized before freeing the area.
page->ResetAllocationStatistics();
page->SetOldGenerationPageFlags(heap()->incremental_marking()->IsMarking());
page->AllocateFreeListCategories();
page->InitializeFreeListCategories();
page->list_node().Initialize();
page->InitializationMemoryFence();
return page;
}
Page* SemiSpace::InitializePage(MemoryChunk* chunk, Executability executable) {
DCHECK_EQ(executable, Executability::NOT_EXECUTABLE);
bool in_to_space = (id() != kFromSpace);
chunk->SetFlag(in_to_space ? MemoryChunk::TO_PAGE : MemoryChunk::FROM_PAGE);
Page* page = static_cast<Page*>(chunk);
page->SetYoungGenerationPageFlags(heap()->incremental_marking()->IsMarking());
page->AllocateLocalTracker();
page->list_node().Initialize();
#ifdef ENABLE_MINOR_MC
if (FLAG_minor_mc) {
page->AllocateYoungGenerationBitmap();
heap()
->minor_mark_compact_collector()
->non_atomic_marking_state()
->ClearLiveness(page);
}
#endif // ENABLE_MINOR_MC
page->InitializationMemoryFence();
return page;
}
LargePage* LargePage::Initialize(Heap* heap, MemoryChunk* chunk,
Executability executable) {
if (executable && chunk->size() > LargePage::kMaxCodePageSize) {
STATIC_ASSERT(LargePage::kMaxCodePageSize <= TypedSlotSet::kMaxOffset);
FATAL("Code page is too large.");
}
MSAN_ALLOCATED_UNINITIALIZED_MEMORY(chunk->area_start(), chunk->area_size());
// Initialize the sentinel value for each page boundary since the mutator
// may initialize the object starting from its end.
Address sentinel = chunk->address() + MemoryChunk::kHeaderSentinelOffset +
MemoryChunk::kPageSize;
while (sentinel < chunk->area_end()) {
*reinterpret_cast<intptr_t*>(sentinel) = kNullAddress;
sentinel += MemoryChunk::kPageSize;
}
LargePage* page = static_cast<LargePage*>(chunk);
page->SetFlag(MemoryChunk::LARGE_PAGE);
page->list_node().Initialize();
return page;
}
void Page::AllocateFreeListCategories() {
for (int i = kFirstCategory; i < kNumberOfCategories; i++) {
categories_[i] = new FreeListCategory(
reinterpret_cast<PagedSpace*>(owner())->free_list(), this);
}
}
void Page::InitializeFreeListCategories() {
for (int i = kFirstCategory; i < kNumberOfCategories; i++) {
categories_[i]->Initialize(static_cast<FreeListCategoryType>(i));
}
}
void Page::ReleaseFreeListCategories() {
for (int i = kFirstCategory; i < kNumberOfCategories; i++) {
if (categories_[i] != nullptr) {
delete categories_[i];
categories_[i] = nullptr;
}
}
}
Page* Page::ConvertNewToOld(Page* old_page) {
DCHECK(old_page);
DCHECK(old_page->InNewSpace());
OldSpace* old_space = old_page->heap()->old_space();
old_page->set_owner(old_space);
old_page->SetFlags(0, static_cast<uintptr_t>(~0));
Page* new_page = old_space->InitializePage(old_page, NOT_EXECUTABLE);
old_space->AddPage(new_page);
return new_page;
}
size_t MemoryChunk::CommittedPhysicalMemory() {
if (!base::OS::HasLazyCommits() || owner()->identity() == LO_SPACE)
return size();
return high_water_mark_;
}
bool MemoryChunk::InOldSpace() const {
return owner()->identity() == OLD_SPACE;
}
bool MemoryChunk::InLargeObjectSpace() const {
return owner()->identity() == LO_SPACE;
}
MemoryChunk* MemoryAllocator::AllocateChunk(size_t reserve_area_size,
size_t commit_area_size,
Executability executable,
Space* owner) {
DCHECK_LE(commit_area_size, reserve_area_size);
size_t chunk_size;
Heap* heap = isolate_->heap();
Address base = kNullAddress;
VirtualMemory reservation;
Address area_start = kNullAddress;
Address area_end = kNullAddress;
void* address_hint =
AlignedAddress(heap->GetRandomMmapAddr(), MemoryChunk::kAlignment);
//
// MemoryChunk layout:
//
// Executable
// +----------------------------+<- base aligned with MemoryChunk::kAlignment
// | Header |
// +----------------------------+<- base + CodePageGuardStartOffset
// | Guard |
// +----------------------------+<- area_start_
// | Area |
// +----------------------------+<- area_end_ (area_start + commit_area_size)
// | Committed but not used |
// +----------------------------+<- aligned at OS page boundary
// | Reserved but not committed |
// +----------------------------+<- aligned at OS page boundary
// | Guard |
// +----------------------------+<- base + chunk_size
//
// Non-executable
// +----------------------------+<- base aligned with MemoryChunk::kAlignment
// | Header |
// +----------------------------+<- area_start_ (base + area_start_)
// | Area |
// +----------------------------+<- area_end_ (area_start + commit_area_size)
// | Committed but not used |
// +----------------------------+<- aligned at OS page boundary
// | Reserved but not committed |
// +----------------------------+<- base + chunk_size
//
if (executable == EXECUTABLE) {
chunk_size = ::RoundUp(MemoryChunkLayout::ObjectStartOffsetInCodePage() +
reserve_area_size +
MemoryChunkLayout::CodePageGuardSize(),
GetCommitPageSize());
// Size of header (not executable) plus area (executable).
size_t commit_size = ::RoundUp(
MemoryChunkLayout::CodePageGuardStartOffset() + commit_area_size,
GetCommitPageSize());
base =
AllocateAlignedMemory(chunk_size, commit_size, MemoryChunk::kAlignment,
executable, address_hint, &reservation);
if (base == kNullAddress) return nullptr;
// Update executable memory size.
size_executable_ += reservation.size();
if (Heap::ShouldZapGarbage()) {
ZapBlock(base, MemoryChunkLayout::CodePageGuardStartOffset(), kZapValue);
ZapBlock(base + MemoryChunkLayout::ObjectStartOffsetInCodePage(),
commit_area_size, kZapValue);
}
area_start = base + MemoryChunkLayout::ObjectStartOffsetInCodePage();
area_end = area_start + commit_area_size;
} else {
chunk_size = ::RoundUp(
MemoryChunkLayout::ObjectStartOffsetInDataPage() + reserve_area_size,
GetCommitPageSize());
size_t commit_size = ::RoundUp(
MemoryChunkLayout::ObjectStartOffsetInDataPage() + commit_area_size,
GetCommitPageSize());
base =
AllocateAlignedMemory(chunk_size, commit_size, MemoryChunk::kAlignment,
executable, address_hint, &reservation);
if (base == kNullAddress) return nullptr;
if (Heap::ShouldZapGarbage()) {
ZapBlock(
base,
MemoryChunkLayout::ObjectStartOffsetInDataPage() + commit_area_size,
kZapValue);
}
area_start = base + MemoryChunkLayout::ObjectStartOffsetInDataPage();
area_end = area_start + commit_area_size;
}
// Use chunk_size for statistics and callbacks because we assume that they
// treat reserved but not-yet committed memory regions of chunks as allocated.
isolate_->counters()->memory_allocated()->Increment(
static_cast<int>(chunk_size));
LOG(isolate_,
NewEvent("MemoryChunk", reinterpret_cast<void*>(base), chunk_size));
// We cannot use the last chunk in the address space because we would
// overflow when comparing top and limit if this chunk is used for a
// linear allocation area.
if ((base + chunk_size) == 0u) {
CHECK(!last_chunk_.IsReserved());
last_chunk_ = std::move(reservation);
UncommitMemory(&last_chunk_);
size_ -= chunk_size;
if (executable == EXECUTABLE) {
size_executable_ -= chunk_size;
}
CHECK(last_chunk_.IsReserved());
return AllocateChunk(reserve_area_size, commit_area_size, executable,
owner);
}
MemoryChunk* chunk =
MemoryChunk::Initialize(heap, base, chunk_size, area_start, area_end,
executable, owner, std::move(reservation));
if (chunk->executable()) RegisterExecutableMemoryChunk(chunk);
return chunk;
}
void MemoryChunk::SetOldGenerationPageFlags(bool is_marking) {
if (is_marking) {
SetFlag(MemoryChunk::POINTERS_TO_HERE_ARE_INTERESTING);
SetFlag(MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING);
SetFlag(MemoryChunk::INCREMENTAL_MARKING);
} else {
ClearFlag(MemoryChunk::POINTERS_TO_HERE_ARE_INTERESTING);
SetFlag(MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING);
ClearFlag(MemoryChunk::INCREMENTAL_MARKING);
}
}
void MemoryChunk::SetYoungGenerationPageFlags(bool is_marking) {
SetFlag(MemoryChunk::POINTERS_TO_HERE_ARE_INTERESTING);
if (is_marking) {
SetFlag(MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING);
SetFlag(MemoryChunk::INCREMENTAL_MARKING);
} else {
ClearFlag(MemoryChunk::POINTERS_FROM_HERE_ARE_INTERESTING);
ClearFlag(MemoryChunk::INCREMENTAL_MARKING);
}
}
void Page::ResetAllocationStatistics() {
allocated_bytes_ = area_size();
wasted_memory_ = 0;
}
void Page::AllocateLocalTracker() {
DCHECK_NULL(local_tracker_);
local_tracker_ = new LocalArrayBufferTracker(this);
}
bool Page::contains_array_buffers() {
return local_tracker_ != nullptr && !local_tracker_->IsEmpty();
}
size_t Page::AvailableInFreeList() {
size_t sum = 0;
ForAllFreeListCategories([&sum](FreeListCategory* category) {
sum += category->available();
});
return sum;
}
#ifdef DEBUG
namespace {
// Skips filler starting from the given filler until the end address.
// Returns the first address after the skipped fillers.
Address SkipFillers(HeapObject filler, Address end) {
Address addr = filler.address();
while (addr < end) {
filler = HeapObject::FromAddress(addr);
CHECK(filler.IsFiller());
addr = filler.address() + filler.Size();
}
return addr;
}
} // anonymous namespace
#endif // DEBUG
size_t Page::ShrinkToHighWaterMark() {
// Shrinking only makes sense outside of the CodeRange, where we don't care
// about address space fragmentation.
VirtualMemory* reservation = reserved_memory();
if (!reservation->IsReserved()) return 0;
// Shrink pages to high water mark. The water mark points either to a filler
// or the area_end.
HeapObject filler = HeapObject::FromAddress(HighWaterMark());
if (filler.address() == area_end()) return 0;
CHECK(filler.IsFiller());
// Ensure that no objects were allocated in [filler, area_end) region.
DCHECK_EQ(area_end(), SkipFillers(filler, area_end()));
// Ensure that no objects will be allocated on this page.
DCHECK_EQ(0u, AvailableInFreeList());
size_t unused = RoundDown(static_cast<size_t>(area_end() - filler.address()),
MemoryAllocator::GetCommitPageSize());
if (unused > 0) {
DCHECK_EQ(0u, unused % MemoryAllocator::GetCommitPageSize());
if (FLAG_trace_gc_verbose) {
PrintIsolate(heap()->isolate(), "Shrinking page %p: end %p -> %p\n",
reinterpret_cast<void*>(this),
reinterpret_cast<void*>(area_end()),
reinterpret_cast<void*>(area_end() - unused));
}
heap()->CreateFillerObjectAt(
filler.address(),
static_cast<int>(area_end() - filler.address() - unused),
ClearRecordedSlots::kNo);
heap()->memory_allocator()->PartialFreeMemory(
this, address() + size() - unused, unused, area_end() - unused);
if (filler.address() != area_end()) {
CHECK(filler.IsFiller());
CHECK_EQ(filler.address() + filler.Size(), area_end());
}
}
return unused;
}
void Page::CreateBlackArea(Address start, Address end) {
DCHECK(heap()->incremental_marking()->black_allocation());
DCHECK_EQ(Page::FromAddress(start), this);
DCHECK_NE(start, end);
DCHECK_EQ(Page::FromAddress(end - 1), this);
IncrementalMarking::MarkingState* marking_state =
heap()->incremental_marking()->marking_state();
marking_state->bitmap(this)->SetRange(AddressToMarkbitIndex(start),
AddressToMarkbitIndex(end));
marking_state->IncrementLiveBytes(this, static_cast<intptr_t>(end - start));
}
void Page::DestroyBlackArea(Address start, Address end) {
DCHECK(heap()->incremental_marking()->black_allocation());
DCHECK_EQ(Page::FromAddress(start), this);
DCHECK_NE(start, end);
DCHECK_EQ(Page::FromAddress(end - 1), this);
IncrementalMarking::MarkingState* marking_state =
heap()->incremental_marking()->marking_state();
marking_state->bitmap(this)->ClearRange(AddressToMarkbitIndex(start),
AddressToMarkbitIndex(end));
marking_state->IncrementLiveBytes(this, -static_cast<intptr_t>(end - start));
}
void MemoryAllocator::PartialFreeMemory(MemoryChunk* chunk, Address start_free,
size_t bytes_to_free,
Address new_area_end) {
VirtualMemory* reservation = chunk->reserved_memory();
DCHECK(reservation->IsReserved());
chunk->size_ -= bytes_to_free;
chunk->area_end_ = new_area_end;
if (chunk->IsFlagSet(MemoryChunk::IS_EXECUTABLE)) {
// Add guard page at the end.
size_t page_size = GetCommitPageSize();
DCHECK_EQ(0, chunk->area_end_ % static_cast<Address>(page_size));
DCHECK_EQ(chunk->address() + chunk->size(),
chunk->area_end() + MemoryChunkLayout::CodePageGuardSize());
reservation->SetPermissions(chunk->area_end_, page_size,
PageAllocator::kNoAccess);
}
// On e.g. Windows, a reservation may be larger than a page and releasing
// partially starting at |start_free| will also release the potentially
// unused part behind the current page.
const size_t released_bytes = reservation->Release(start_free);
DCHECK_GE(size_, released_bytes);
size_ -= released_bytes;
isolate_->counters()->memory_allocated()->Decrement(
static_cast<int>(released_bytes));
}
void MemoryAllocator::UnregisterMemory(MemoryChunk* chunk) {
DCHECK(!chunk->IsFlagSet(MemoryChunk::UNREGISTERED));
VirtualMemory* reservation = chunk->reserved_memory();
const size_t size =
reservation->IsReserved() ? reservation->size() : chunk->size();
DCHECK_GE(size_, static_cast<size_t>(size));
size_ -= size;
isolate_->counters()->memory_allocated()->Decrement(static_cast<int>(size));
if (chunk->executable() == EXECUTABLE) {
DCHECK_GE(size_executable_, size);
size_executable_ -= size;
}
if (chunk->executable()) UnregisterExecutableMemoryChunk(chunk);
chunk->SetFlag(MemoryChunk::UNREGISTERED);
}
void MemoryAllocator::PreFreeMemory(MemoryChunk* chunk) {
DCHECK(!chunk->IsFlagSet(MemoryChunk::PRE_FREED));
LOG(isolate_, DeleteEvent("MemoryChunk", chunk));
UnregisterMemory(chunk);
isolate_->heap()->RememberUnmappedPage(reinterpret_cast<Address>(chunk),
chunk->IsEvacuationCandidate());
chunk->SetFlag(MemoryChunk::PRE_FREED);
}
void MemoryAllocator::PerformFreeMemory(MemoryChunk* chunk) {
DCHECK(chunk->IsFlagSet(MemoryChunk::UNREGISTERED));
DCHECK(chunk->IsFlagSet(MemoryChunk::PRE_FREED));
chunk->ReleaseAllocatedMemory();
VirtualMemory* reservation = chunk->reserved_memory();
if (chunk->IsFlagSet(MemoryChunk::POOLED)) {
UncommitMemory(reservation);
} else {
if (reservation->IsReserved()) {
reservation->Free();
} else {
// Only read-only pages can have non-initialized reservation object.
DCHECK_EQ(RO_SPACE, chunk->owner()->identity());
FreeMemory(page_allocator(chunk->executable()), chunk->address(),
chunk->size());
}
}
}
template <MemoryAllocator::FreeMode mode>
void MemoryAllocator::Free(MemoryChunk* chunk) {
switch (mode) {
case kFull:
PreFreeMemory(chunk);
PerformFreeMemory(chunk);
break;
case kAlreadyPooled:
// Pooled pages cannot be touched anymore as their memory is uncommitted.
// Pooled pages are not-executable.
FreeMemory(data_page_allocator(), chunk->address(),
static_cast<size_t>(MemoryChunk::kPageSize));
break;
case kPooledAndQueue:
DCHECK_EQ(chunk->size(), static_cast<size_t>(MemoryChunk::kPageSize));
DCHECK_EQ(chunk->executable(), NOT_EXECUTABLE);
chunk->SetFlag(MemoryChunk::POOLED);
V8_FALLTHROUGH;
case kPreFreeAndQueue:
PreFreeMemory(chunk);
// The chunks added to this queue will be freed by a concurrent thread.
unmapper()->AddMemoryChunkSafe(chunk);
break;
}
}
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE) void MemoryAllocator::Free<
MemoryAllocator::kFull>(MemoryChunk* chunk);
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE) void MemoryAllocator::Free<
MemoryAllocator::kAlreadyPooled>(MemoryChunk* chunk);
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE) void MemoryAllocator::Free<
MemoryAllocator::kPreFreeAndQueue>(MemoryChunk* chunk);
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE) void MemoryAllocator::Free<
MemoryAllocator::kPooledAndQueue>(MemoryChunk* chunk);
template <MemoryAllocator::AllocationMode alloc_mode, typename SpaceType>
Page* MemoryAllocator::AllocatePage(size_t size, SpaceType* owner,
Executability executable) {
MemoryChunk* chunk = nullptr;
if (alloc_mode == kPooled) {
DCHECK_EQ(size, static_cast<size_t>(
MemoryChunkLayout::AllocatableMemoryInMemoryChunk(
owner->identity())));
DCHECK_EQ(executable, NOT_EXECUTABLE);
chunk = AllocatePagePooled(owner);
}
if (chunk == nullptr) {
chunk = AllocateChunk(size, size, executable, owner);
}
if (chunk == nullptr) return nullptr;
return owner->InitializePage(chunk, executable);
}
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE)
Page* MemoryAllocator::AllocatePage<MemoryAllocator::kRegular, PagedSpace>(
size_t size, PagedSpace* owner, Executability executable);
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE)
Page* MemoryAllocator::AllocatePage<MemoryAllocator::kRegular, SemiSpace>(
size_t size, SemiSpace* owner, Executability executable);
template EXPORT_TEMPLATE_DEFINE(V8_EXPORT_PRIVATE)
Page* MemoryAllocator::AllocatePage<MemoryAllocator::kPooled, SemiSpace>(
size_t size, SemiSpace* owner, Executability executable);
LargePage* MemoryAllocator::AllocateLargePage(size_t size,
LargeObjectSpace* owner,
Executability executable) {
MemoryChunk* chunk = AllocateChunk(size, size, executable, owner);
if (chunk == nullptr) return nullptr;
return LargePage::Initialize(isolate_->heap(), chunk, executable);
}
template <typename SpaceType>
MemoryChunk* MemoryAllocator::AllocatePagePooled(SpaceType* owner) {
MemoryChunk* chunk = unmapper()->TryGetPooledMemoryChunkSafe();
if (chunk == nullptr) return nullptr;
const int size = MemoryChunk::kPageSize;
const Address start = reinterpret_cast<Address>(chunk);
const Address area_start =
start +
MemoryChunkLayout::ObjectStartOffsetInMemoryChunk(owner->identity());
const Address area_end = start + size;
// Pooled pages are always regular data pages.
DCHECK_NE(CODE_SPACE, owner->identity());
VirtualMemory reservation(data_page_allocator(), start, size);
if (!CommitMemory(&reservation)) return nullptr;
if (Heap::ShouldZapGarbage()) {
ZapBlock(start, size, kZapValue);
}
MemoryChunk::Initialize(isolate_->heap(), start, size, area_start, area_end,
NOT_EXECUTABLE, owner, std::move(reservation));
size_ += size;
return chunk;
}
void MemoryAllocator::ZapBlock(Address start, size_t size,
uintptr_t zap_value) {
DCHECK(IsAligned(start, kTaggedSize));
DCHECK(IsAligned(size, kTaggedSize));
MemsetTagged(ObjectSlot(start), Object(static_cast<Address>(zap_value)),
size >> kTaggedSizeLog2);
}
intptr_t MemoryAllocator::GetCommitPageSize() {
if (FLAG_v8_os_page_size != 0) {
DCHECK(base::bits::IsPowerOfTwo(FLAG_v8_os_page_size));
return FLAG_v8_os_page_size * KB;
} else {
return CommitPageSize();
}
}
base::AddressRegion MemoryAllocator::ComputeDiscardMemoryArea(Address addr,
size_t size) {
size_t page_size = MemoryAllocator::GetCommitPageSize();
if (size < page_size + FreeSpace::kSize) {
return base::AddressRegion(0, 0);
}
Address discardable_start = RoundUp(addr + FreeSpace::kSize, page_size);
Address discardable_end = RoundDown(addr + size, page_size);
if (discardable_start >= discardable_end) return base::AddressRegion(0, 0);
return base::AddressRegion(discardable_start,
discardable_end - discardable_start);
}
bool MemoryAllocator::CommitExecutableMemory(VirtualMemory* vm, Address start,
size_t commit_size,
size_t reserved_size) {
const size_t page_size = GetCommitPageSize();
// All addresses and sizes must be aligned to the commit page size.
DCHECK(IsAligned(start, page_size));
DCHECK_EQ(0, commit_size % page_size);
DCHECK_EQ(0, reserved_size % page_size);
const size_t guard_size = MemoryChunkLayout::CodePageGuardSize();
const size_t pre_guard_offset = MemoryChunkLayout::CodePageGuardStartOffset();
const size_t code_area_offset =
MemoryChunkLayout::ObjectStartOffsetInCodePage();
// reserved_size includes two guard regions, commit_size does not.
DCHECK_LE(commit_size, reserved_size - 2 * guard_size);
const Address pre_guard_page = start + pre_guard_offset;
const Address code_area = start + code_area_offset;
const Address post_guard_page = start + reserved_size - guard_size;
// Commit the non-executable header, from start to pre-code guard page.
if (vm->SetPermissions(start, pre_guard_offset, PageAllocator::kReadWrite)) {
// Create the pre-code guard page, following the header.
if (vm->SetPermissions(pre_guard_page, page_size,
PageAllocator::kNoAccess)) {
// Commit the executable code body.
if (vm->SetPermissions(code_area, commit_size - pre_guard_offset,
PageAllocator::kReadWrite)) {
// Create the post-code guard page.
if (vm->SetPermissions(post_guard_page, page_size,
PageAllocator::kNoAccess)) {
UpdateAllocatedSpaceLimits(start, code_area + commit_size);
return true;
}
vm->SetPermissions(code_area, commit_size, PageAllocator::kNoAccess);
}
}
vm->SetPermissions(start, pre_guard_offset, PageAllocator::kNoAccess);
}
return false;
}
// -----------------------------------------------------------------------------
// MemoryChunk implementation
void MemoryChunk::ReleaseAllocatedMemory() {
if (mutex_ != nullptr) {
delete mutex_;
mutex_ = nullptr;
}
if (page_protection_change_mutex_ != nullptr) {
delete page_protection_change_mutex_;
page_protection_change_mutex_ = nullptr;
}
ReleaseSlotSet<OLD_TO_NEW>();
ReleaseSlotSet<OLD_TO_OLD>();
ReleaseTypedSlotSet<OLD_TO_NEW>();
ReleaseTypedSlotSet<OLD_TO_OLD>();
ReleaseInvalidatedSlots();
if (local_tracker_ != nullptr) ReleaseLocalTracker();
if (young_generation_bitmap_ != nullptr) ReleaseYoungGenerationBitmap();
if (marking_bitmap_ != nullptr) ReleaseMarkingBitmap();
if (code_object_registry_ != nullptr) delete code_object_registry_;
if (!IsLargePage()) {
Page* page = static_cast<Page*>(this);
page->ReleaseFreeListCategories();
}
}
static SlotSet* AllocateAndInitializeSlotSet(size_t size, Address page_start) {
size_t pages = (size + Page::kPageSize - 1) / Page::kPageSize;
DCHECK_LT(0, pages);
SlotSet* slot_set = new SlotSet[pages];
for (size_t i = 0; i < pages; i++) {
slot_set[i].SetPageStart(page_start + i * Page::kPageSize);
}
return slot_set;
}
template V8_EXPORT_PRIVATE SlotSet* MemoryChunk::AllocateSlotSet<OLD_TO_NEW>();
template V8_EXPORT_PRIVATE SlotSet* MemoryChunk::AllocateSlotSet<OLD_TO_OLD>();
template <RememberedSetType type>
SlotSet* MemoryChunk::AllocateSlotSet() {
SlotSet* slot_set = AllocateAndInitializeSlotSet(size_, address());
SlotSet* old_slot_set = base::AsAtomicPointer::Release_CompareAndSwap(
&slot_set_[type], nullptr, slot_set);
if (old_slot_set != nullptr) {
delete[] slot_set;
slot_set = old_slot_set;
}
DCHECK(slot_set);
return slot_set;
}
template void MemoryChunk::ReleaseSlotSet<OLD_TO_NEW>();
template void MemoryChunk::ReleaseSlotSet<OLD_TO_OLD>();
template <RememberedSetType type>
void MemoryChunk::ReleaseSlotSet() {
SlotSet* slot_set = slot_set_[type];
if (slot_set) {
slot_set_[type] = nullptr;
delete[] slot_set;
}
}
template TypedSlotSet* MemoryChunk::AllocateTypedSlotSet<OLD_TO_NEW>();
template TypedSlotSet* MemoryChunk::AllocateTypedSlotSet<OLD_TO_OLD>();
template <RememberedSetType type>
TypedSlotSet* MemoryChunk::AllocateTypedSlotSet() {
TypedSlotSet* typed_slot_set = new TypedSlotSet(address());
TypedSlotSet* old_value = base::AsAtomicPointer::Release_CompareAndSwap(
&typed_slot_set_[type], nullptr, typed_slot_set);
if (old_value != nullptr) {
delete typed_slot_set;
typed_slot_set = old_value;
}
DCHECK(typed_slot_set);
return typed_slot_set;
}
template void MemoryChunk::ReleaseTypedSlotSet<OLD_TO_NEW>();
template void MemoryChunk::ReleaseTypedSlotSet<OLD_TO_OLD>();
template <RememberedSetType type>
void MemoryChunk::ReleaseTypedSlotSet() {
TypedSlotSet* typed_slot_set = typed_slot_set_[type];
if (typed_slot_set) {
typed_slot_set_[type] = nullptr;
delete typed_slot_set;
}
}
InvalidatedSlots* MemoryChunk::AllocateInvalidatedSlots() {
DCHECK_NULL(invalidated_slots_);
invalidated_slots_ = new InvalidatedSlots();
return invalidated_slots_;
}
void MemoryChunk::ReleaseInvalidatedSlots() {
if (invalidated_slots_) {
delete invalidated_slots_;
invalidated_slots_ = nullptr;
}
}
void MemoryChunk::RegisterObjectWithInvalidatedSlots(HeapObject object,
int size) {
if (!ShouldSkipEvacuationSlotRecording()) {
if (invalidated_slots() == nullptr) {
AllocateInvalidatedSlots();
}
int old_size = (*invalidated_slots())[object];
(*invalidated_slots())[object] = std::max(old_size, size);
}
}
bool MemoryChunk::RegisteredObjectWithInvalidatedSlots(HeapObject object) {
if (ShouldSkipEvacuationSlotRecording()) {
// Invalidated slots do not matter if we are not recording slots.
return true;
}
if (invalidated_slots() == nullptr) {
return false;
}
return invalidated_slots()->find(object) != invalidated_slots()->end();
}
void MemoryChunk::MoveObjectWithInvalidatedSlots(HeapObject old_start,
HeapObject new_start) {
DCHECK_LT(old_start, new_start);
DCHECK_EQ(MemoryChunk::FromHeapObject(old_start),
MemoryChunk::FromHeapObject(new_start));
if (!ShouldSkipEvacuationSlotRecording() && invalidated_slots()) {
auto it = invalidated_slots()->find(old_start);
if (it != invalidated_slots()->end()) {
int old_size = it->second;
int delta = static_cast<int>(new_start.address() - old_start.address());
invalidated_slots()->erase(it);
(*invalidated_slots())[new_start] = old_size - delta;
}
}
}
void MemoryChunk::ReleaseLocalTracker() {
DCHECK_NOT_NULL(local_tracker_);
delete local_tracker_;
local_tracker_ = nullptr;
}
void MemoryChunk::AllocateYoungGenerationBitmap() {
DCHECK_NULL(young_generation_bitmap_);
young_generation_bitmap_ = static_cast<Bitmap*>(calloc(1, Bitmap::kSize));
}
void MemoryChunk::ReleaseYoungGenerationBitmap() {
DCHECK_NOT_NULL(young_generation_bitmap_);
free(young_generation_bitmap_);
young_generation_bitmap_ = nullptr;
}
void MemoryChunk::AllocateMarkingBitmap() {
DCHECK_NULL(marking_bitmap_);
marking_bitmap_ = static_cast<Bitmap*>(calloc(1, Bitmap::kSize));
}
void MemoryChunk::ReleaseMarkingBitmap() {
DCHECK_NOT_NULL(marking_bitmap_);
free(marking_bitmap_);
marking_bitmap_ = nullptr;
}
// -----------------------------------------------------------------------------
// PagedSpace implementation
void Space::CheckOffsetsAreConsistent() const {
static_assert(Space::kIdOffset == heap_internals::Space::kIdOffset,
"ID offset inconsistent");
DCHECK_EQ(Space::kIdOffset, OFFSET_OF(Space, id_));
}
void Space::AddAllocationObserver(AllocationObserver* observer) {
allocation_observers_.push_back(observer);
StartNextInlineAllocationStep();
}
void Space::RemoveAllocationObserver(AllocationObserver* observer) {
auto it = std::find(allocation_observers_.begin(),
allocation_observers_.end(), observer);
DCHECK(allocation_observers_.end() != it);
allocation_observers_.erase(it);
StartNextInlineAllocationStep();
}
void Space::PauseAllocationObservers() { allocation_observers_paused_ = true; }
void Space::ResumeAllocationObservers() {
allocation_observers_paused_ = false;
}
void Space::AllocationStep(int bytes_since_last, Address soon_object,
int size) {
if (!AllocationObserversActive()) {
return;
}
DCHECK(!heap()->allocation_step_in_progress());
heap()->set_allocation_step_in_progress(true);
heap()->CreateFillerObjectAt(soon_object, size, ClearRecordedSlots::kNo);
for (AllocationObserver* observer : allocation_observers_) {
observer->AllocationStep(bytes_since_last, soon_object, size);
}
heap()->set_allocation_step_in_progress(false);
}
intptr_t Space::GetNextInlineAllocationStepSize() {
intptr_t next_step = 0;
for (AllocationObserver* observer : allocation_observers_) {
next_step = next_step ? Min(next_step, observer->bytes_to_next_step())
: observer->bytes_to_next_step();
}
DCHECK(allocation_observers_.size() == 0 || next_step > 0);
return next_step;
}
PagedSpace::PagedSpace(Heap* heap, AllocationSpace space,
Executability executable)
: SpaceWithLinearArea(heap, space), executable_(executable) {
area_size_ = MemoryChunkLayout::AllocatableMemoryInMemoryChunk(space);
accounting_stats_.Clear();
}
void PagedSpace::TearDown() {
while (!memory_chunk_list_.Empty()) {
MemoryChunk* chunk = memory_chunk_list_.front();
memory_chunk_list_.Remove(chunk);
heap()->memory_allocator()->Free<MemoryAllocator::kFull>(chunk);
}
accounting_stats_.Clear();
}
void PagedSpace::RefillFreeList() {
// Any PagedSpace might invoke RefillFreeList. We filter all but our old
// generation spaces out.
if (identity() != OLD_SPACE && identity() != CODE_SPACE &&
identity() != MAP_SPACE && identity() != RO_SPACE) {
return;
}
MarkCompactCollector* collector = heap()->mark_compact_collector();
size_t added = 0;
{
Page* p = nullptr;
while ((p = collector->sweeper()->GetSweptPageSafe(this)) != nullptr) {
// We regularly sweep NEVER_ALLOCATE_ON_PAGE pages. We drop the freelist
// entries here to make them unavailable for allocations.
if (p->IsFlagSet(Page::NEVER_ALLOCATE_ON_PAGE)) {
p->ForAllFreeListCategories(
[](FreeListCategory* category) { category->Reset(); });
}
// Only during compaction pages can actually change ownership. This is
// safe because there exists no other competing action on the page links
// during compaction.
if (is_local()) {
DCHECK_NE(this, p->owner());
PagedSpace* owner = reinterpret_cast<PagedSpace*>(p->owner());
base::MutexGuard guard(owner->mutex());
owner->RefineAllocatedBytesAfterSweeping(p);
owner->RemovePage(p);
added += AddPage(p);
} else {
base::MutexGuard guard(mutex());
DCHECK_EQ(this, p->owner());
RefineAllocatedBytesAfterSweeping(p);
added += RelinkFreeListCategories(p);
}
added += p->wasted_memory();
if (is_local() && (added > kCompactionMemoryWanted)) break;
}
}
}
void PagedSpace::MergeCompactionSpace(CompactionSpace* other) {
base::MutexGuard guard(mutex());
DCHECK(identity() == other->identity());
// Unmerged fields:
// area_size_
other->FreeLinearAllocationArea();
// The linear allocation area of {other} should be destroyed now.
DCHECK_EQ(kNullAddress, other->top());
DCHECK_EQ(kNullAddress, other->limit());
// Move over pages.
for (auto it = other->begin(); it != other->end();) {
Page* p = *(it++);
// Relinking requires the category to be unlinked.
other->RemovePage(p);
AddPage(p);
DCHECK_IMPLIES(
!p->IsFlagSet(Page::NEVER_ALLOCATE_ON_PAGE),
p->AvailableInFreeList() == p->AvailableInFreeListFromAllocatedBytes());
}
DCHECK_EQ(0u, other->Size());
DCHECK_EQ(0u, other->Capacity());
}
size_t PagedSpace::CommittedPhysicalMemory() {
if (!base::OS::HasLazyCommits()) return CommittedMemory();
MemoryChunk::UpdateHighWaterMark(allocation_info_.top());
size_t size = 0;
for (Page* page : *this) {
size += page->CommittedPhysicalMemory();
}
return size;
}
bool PagedSpace::ContainsSlow(Address addr) {
Page* p = Page::FromAddress(addr);
for (Page* page : *this) {
if (page == p) return true;
}
return false;
}
void PagedSpace::RefineAllocatedBytesAfterSweeping(Page* page) {
CHECK(page->SweepingDone());
auto marking_state =
heap()->incremental_marking()->non_atomic_marking_state();
// The live_byte on the page was accounted in the space allocated
// bytes counter. After sweeping allocated_bytes() contains the
// accurate live byte count on the page.
size_t old_counter = marking_state->live_bytes(page);
size_t new_counter = page->allocated_bytes();
DCHECK_GE(old_counter, new_counter);
if (old_counter > new_counter) {
DecreaseAllocatedBytes(old_counter - new_counter, page);
// Give the heap a chance to adjust counters in response to the
// more precise and smaller old generation size.
heap()->NotifyRefinedOldGenerationSize(old_counter - new_counter);
}
marking_state->SetLiveBytes(page, 0);
}
Page* PagedSpace::RemovePageSafe(int size_in_bytes) {
base::MutexGuard guard(mutex());
// Check for pages that still contain free list entries. Bail out for smaller
// categories.
const int minimum_category =
static_cast<int>(FreeList::SelectFreeListCategoryType(size_in_bytes));
Page* page = free_list()->GetPageForCategoryType(kHuge);
if (!page && static_cast<int>(kLarge) >= minimum_category)
page = free_list()->GetPageForCategoryType(kLarge);
if (!page && static_cast<int>(kMedium) >= minimum_category)
page = free_list()->GetPageForCategoryType(kMedium);
if (!page && static_cast<int>(kSmall) >= minimum_category)
page = free_list()->GetPageForCategoryType(kSmall);
if (!page && static_cast<int>(kTiny) >= minimum_category)
page = free_list()->GetPageForCategoryType(kTiny);
if (!page && static_cast<int>(kTiniest) >= minimum_category)
page = free_list()->GetPageForCategoryType(kTiniest);
if (!page) return nullptr;
RemovePage(page);
return page;
}
size_t PagedSpace::AddPage(Page* page) {
CHECK(page->SweepingDone());
page->set_owner(this);
memory_chunk_list_.PushBack(page);
AccountCommitted(page->size());
IncreaseCapacity(page->area_size());
IncreaseAllocatedBytes(page->allocated_bytes(), page);
for (size_t i = 0; i < ExternalBackingStoreType::kNumTypes; i++) {
ExternalBackingStoreType t = static_cast<ExternalBackingStoreType>(i);
IncrementExternalBackingStoreBytes(t, page->ExternalBackingStoreBytes(t));
}
return RelinkFreeListCategories(page);
}
void PagedSpace::RemovePage(Page* page) {
CHECK(page->SweepingDone());
memory_chunk_list_.Remove(page);
UnlinkFreeListCategories(page);
DecreaseAllocatedBytes(page->allocated_bytes(), page);
DecreaseCapacity(page->area_size());
AccountUncommitted(page->size());
for (size_t i = 0; i < ExternalBackingStoreType::kNumTypes; i++) {
ExternalBackingStoreType t = static_cast<ExternalBackingStoreType>(i);
DecrementExternalBackingStoreBytes(t, page->ExternalBackingStoreBytes(t));
}
}
size_t PagedSpace::ShrinkPageToHighWaterMark(Page* page) {
size_t unused = page->ShrinkToHighWaterMark();
accounting_stats_.DecreaseCapacity(static_cast<intptr_t>(unused));
AccountUncommitted(unused);
return unused;
}
void PagedSpace::ResetFreeList() {
for (Page* page : *this) {
free_list_.EvictFreeListItems(page);
}
DCHECK(free_list_.IsEmpty());
}
void PagedSpace::ShrinkImmortalImmovablePages() {
DCHECK(!heap()->deserialization_complete());
MemoryChunk::UpdateHighWaterMark(allocation_info_.top());
FreeLinearAllocationArea();
ResetFreeList();
for (Page* page : *this) {
DCHECK(page->IsFlagSet(Page::NEVER_EVACUATE));
ShrinkPageToHighWaterMark(page);
}
}
bool PagedSpace::Expand() {
// Always lock against the main space as we can only adjust capacity and
// pages concurrently for the main paged space.
base::MutexGuard guard(heap()->paged_space(identity())->mutex());
const int size = AreaSize();
if (!heap()->CanExpandOldGeneration(size)) return false;
Page* page =
heap()->memory_allocator()->AllocatePage(size, this, executable());
if (page == nullptr) return false;
// Pages created during bootstrapping may contain immortal immovable objects.
if (!heap()->deserialization_complete()) page->MarkNeverEvacuate();
AddPage(page);
Free(page->area_start(), page->area_size(),
SpaceAccountingMode::kSpaceAccounted);
heap()->NotifyOldGenerationExpansion();
return true;
}
int PagedSpace::CountTotalPages() {
int count = 0;
for (Page* page : *this) {
count++;
USE(page);
}
return count;
}
void PagedSpace::SetLinearAllocationArea(Address top, Address limit) {
SetTopAndLimit(top, limit);
if (top != kNullAddress && top != limit &&
heap()->incremental_marking()->black_allocation()) {
Page::FromAllocationAreaAddress(top)->CreateBlackArea(top, limit);
}
}
void PagedSpace::DecreaseLimit(Address new_limit) {
Address old_limit = limit();
DCHECK_LE(top(), new_limit);
DCHECK_GE(old_limit, new_limit);
if (new_limit != old_limit) {
SetTopAndLimit(top(), new_limit);
Free(new_limit, old_limit - new_limit,
SpaceAccountingMode::kSpaceAccounted);
if (heap()->incremental_marking()->black_allocation()) {
Page::FromAllocationAreaAddress(new_limit)->DestroyBlackArea(new_limit,
old_limit);
}
}
}
Address SpaceWithLinearArea::ComputeLimit(Address start, Address end,
size_t min_size) {
DCHECK_GE(end - start, min_size);
if (heap()->inline_allocation_disabled()) {
// Fit the requested area exactly.
return start + min_size;
} else if (SupportsInlineAllocation() && AllocationObserversActive()) {
// Generated code may allocate inline from the linear allocation area for.
// To make sure we can observe these allocations, we use a lower limit.
size_t step = GetNextInlineAllocationStepSize();
// TODO(ofrobots): there is subtle difference between old space and new
// space here. Any way to avoid it? `step - 1` makes more sense as we would
// like to sample the object that straddles the `start + step` boundary.
// Rounding down further would introduce a small statistical error in
// sampling. However, presently PagedSpace requires limit to be aligned.
size_t rounded_step;
if (identity() == NEW_SPACE) {
DCHECK_GE(step, 1);
rounded_step = step - 1;
} else {
rounded_step = RoundSizeDownToObjectAlignment(static_cast<int>(step));
}
return Min(static_cast<Address>(start + min_size + rounded_step), end);
} else {
// The entire node can be used as the linear allocation area.
return end;
}
}
void PagedSpace::MarkLinearAllocationAreaBlack() {
DCHECK(heap()->incremental_marking()->black_allocation());
Address current_top = top();
Address current_limit = limit();
if (current_top != kNullAddress && current_top != current_limit) {
Page::FromAllocationAreaAddress(current_top)
->CreateBlackArea(current_top, current_limit);
}
}
void PagedSpace::UnmarkLinearAllocationArea() {
Address current_top = top();
Address current_limit = limit();
if (current_top != kNullAddress && current_top != current_limit) {
Page::FromAllocationAreaAddress(current_top)
->DestroyBlackArea(current_top, current_limit);
}
}
void PagedSpace::FreeLinearAllocationArea() {
// Mark the old linear allocation area with a free space map so it can be
// skipped when scanning the heap.
Address current_top = top();
Address current_limit = limit();
if (current_top == kNullAddress) {
DCHECK_EQ(kNullAddress, current_limit);
return;
}
if (heap()->incremental_marking()->black_allocation()) {
Page* page = Page::FromAllocationAreaAddress(current_top);
// Clear the bits in the unused black area.
if (current_top != current_limit) {
IncrementalMarking::MarkingState* marking_state =
heap()->incremental_marking()->marking_state();
marking_state->bitmap(page)->ClearRange(
page->AddressToMarkbitIndex(current_top),
page->AddressToMarkbitIndex(current_limit));
marking_state->IncrementLiveBytes(
page, -static_cast<int>(current_limit - current_top));
}
}
InlineAllocationStep(current_top, kNullAddress, kNullAddress, 0);
SetTopAndLimit(kNullAddress, kNullAddress);
DCHECK_GE(current_limit, current_top);
// The code page of the linear allocation area needs to be unprotected
// because we are going to write a filler into that memory area below.
if (identity() == CODE_SPACE) {
heap()->UnprotectAndRegisterMemoryChunk(
MemoryChunk::FromAddress(current_top));
}
Free(current_top, current_limit - current_top,
SpaceAccountingMode::kSpaceAccounted);
}
void PagedSpace::ReleasePage(Page* page) {
DCHECK_EQ(
0, heap()->incremental_marking()->non_atomic_marking_state()->live_bytes(
page));
DCHECK_EQ(page->owner(), this);
free_list_.EvictFreeListItems(page);
DCHECK(!free_list_.ContainsPageFreeListItems(page));
if (Page::FromAllocationAreaAddress(allocation_info_.top()) == page) {
DCHECK(!top_on_previous_step_);
allocation_info_.Reset(kNullAddress, kNullAddress);
}
AccountUncommitted(page->size());
accounting_stats_.DecreaseCapacity(page->area_size());
heap()->memory_allocator()->Free<MemoryAllocator::kPreFreeAndQueue>(page);
}
void PagedSpace::SetReadable() {
DCHECK(identity() == CODE_SPACE);
for (Page* page : *this) {
CHECK(heap()->memory_allocator()->IsMemoryChunkExecutable(page));
page->SetReadable();
}
}
void PagedSpace::SetReadAndExecutable() {
DCHECK(identity() == CODE_SPACE);
for (Page* page : *this) {
CHECK(heap()->memory_allocator()->IsMemoryChunkExecutable(page));
page->SetReadAndExecutable();
}
}
void PagedSpace::SetReadAndWritable() {
DCHECK(identity() == CODE_SPACE);
for (Page* page : *this) {
CHECK(heap()->memory_allocator()->IsMemoryChunkExecutable(page));
page->SetReadAndWritable();
}
}
std::unique_ptr<ObjectIterator> PagedSpace::GetObjectIterator() {
return std::unique_ptr<ObjectIterator>(new HeapObjectIterator(this));
}
bool PagedSpace::RefillLinearAllocationAreaFromFreeList(size_t size_in_bytes) {
DCHECK(IsAligned(size_in_bytes, kTaggedSize));
DCHECK_LE(top(), limit());
#ifdef DEBUG
if (top() != limit()) {
DCHECK_EQ(Page::FromAddress(top()), Page::FromAddress(limit() - 1));
}
#endif
// Don't free list allocate if there is linear space available.
DCHECK_LT(static_cast<size_t>(limit() - top()), size_in_bytes);
// Mark the old linear allocation area with a free space map so it can be
// skipped when scanning the heap. This also puts it back in the free list
// if it is big enough.
FreeLinearAllocationArea();
if (!is_local()) {
heap()->StartIncrementalMarkingIfAllocationLimitIsReached(
heap()->GCFlagsForIncrementalMarking(),
kGCCallbackScheduleIdleGarbageCollection);
}
size_t new_node_size = 0;
FreeSpace new_node = free_list_.Allocate(size_in_bytes, &new_node_size);
if (new_node.is_null()) return false;
DCHECK_GE(new_node_size, size_in_bytes);
// The old-space-step might have finished sweeping and restarted marking.
// Verify that it did not turn the page of the new node into an evacuation
// candidate.
DCHECK(!MarkCompactCollector::IsOnEvacuationCandidate(new_node));
// Memory in the linear allocation area is counted as allocated. We may free
// a little of this again immediately - see below.
Page* page = Page::FromHeapObject(new_node);
IncreaseAllocatedBytes(new_node_size, page);
Address start = new_node.address();
Address end = new_node.address() + new_node_size;
Address limit = ComputeLimit(start, end, size_in_bytes);
DCHECK_LE(limit, end);
DCHECK_LE(size_in_bytes, limit - start);
if (limit != end) {
if (identity() == CODE_SPACE) {
heap()->UnprotectAndRegisterMemoryChunk(page);
}
Free(limit, end - limit, SpaceAccountingMode::kSpaceAccounted);
}
SetLinearAllocationArea(start, limit);
return true;
}
#ifdef DEBUG
void PagedSpace::Print() {}
#endif
#ifdef VERIFY_HEAP
void PagedSpace::Verify(Isolate* isolate, ObjectVisitor* visitor) {
bool allocation_pointer_found_in_space =
(allocation_info_.top() == allocation_info_.limit());
size_t external_space_bytes[kNumTypes];
size_t external_page_bytes[kNumTypes];
for (int i = 0; i < kNumTypes; i++) {
external_space_bytes[static_cast<ExternalBackingStoreType>(i)] = 0;
}
for (Page* page : *this) {
CHECK(page->owner() == this);
for (int i = 0; i < kNumTypes; i++) {
external_page_bytes[static_cast<ExternalBackingStoreType>(i)] = 0;
}
if (page == Page::FromAllocationAreaAddress(allocation_info_.top())) {
allocation_pointer_found_in_space = true;
}
CHECK(page->SweepingDone());
HeapObjectIterator it(page);
Address end_of_previous_object = page->area_start();
Address top = page->area_end();
for (HeapObject object = it.Next(); !object.is_null(); object = it.Next()) {
CHECK(end_of_previous_object <= object.address());
// The first word should be a map, and we expect all map pointers to
// be in map space.
Map map = object.map();
CHECK(map.IsMap());
CHECK(isolate->heap()->map_space()->Contains(map) ||
ReadOnlyHeap::Contains(map));
// Perform space-specific object verification.
VerifyObject(object);
// The object itself should look OK.
object.ObjectVerify(isolate);
if (!FLAG_verify_heap_skip_remembered_set) {
isolate->heap()->VerifyRememberedSetFor(object);
}
// All the interior pointers should be contained in the heap.
int size = object.Size();
object.IterateBody(map, size, visitor);
CHECK(object.address() + size <= top);
end_of_previous_object = object.address() + size;
if (object.IsExternalString()) {
ExternalString external_string = ExternalString::cast(object);
size_t size = external_string.ExternalPayloadSize();
external_page_bytes[ExternalBackingStoreType::kExternalString] += size;
} else if (object.IsJSArrayBuffer()) {
JSArrayBuffer array_buffer = JSArrayBuffer::cast(object);
if (ArrayBufferTracker::IsTracked(array_buffer)) {
size_t size = array_buffer.byte_length();
external_page_bytes[ExternalBackingStoreType::kArrayBuffer] += size;
}
}
}
for (int i = 0; i < kNumTypes; i++) {
ExternalBackingStoreType t = static_cast<ExternalBackingStoreType>(i);
CHECK_EQ(external_page_bytes[t], page->ExternalBackingStoreBytes(t));
external_space_bytes[t] += external_page_bytes[t];
}
}
for (int i = 0; i < kNumTypes; i++) {
ExternalBackingStoreType t = static_cast<ExternalBackingStoreType>(i);
CHECK_EQ(external_space_bytes[t], ExternalBackingStoreBytes(t));
}
CHECK(allocation_pointer_found_in_space);
#ifdef DEBUG
VerifyCountersAfterSweeping();
#endif
}
void PagedSpace::VerifyLiveBytes() {
IncrementalMarking::MarkingState* marking_state =
heap()->incremental_marking()->marking_state();
for (Page* page : *this) {
CHECK(page->SweepingDone());
HeapObjectIterator it(page);
int black_size = 0;
for (HeapObject object = it.Next(); !object.is_null(); object = it.Next()) {
// All the interior pointers should be contained in the heap.
if (marking_state->IsBlack(object)) {
black_size += object.Size();
}
}
CHECK_LE(black_size, marking_state->live_bytes(page));
}
}
#endif // VERIFY_HEAP
#ifdef DEBUG
void PagedSpace::VerifyCountersAfterSweeping() {
size_t total_capacity = 0;
size_t total_allocated = 0;
for (Page* page : *this) {
DCHECK(page->SweepingDone());
total_capacity += page->area_size();
HeapObjectIterator it(page);
size_t real_allocated = 0;
for (HeapObject object = it.Next(); !object.is_null(); object = it.Next()) {
if (!object.IsFiller()) {
real_allocated += object.Size();
}
}
total_allocated += page->allocated_bytes();
// The real size can be smaller than the accounted size if array trimming,
// object slack tracking happened after sweeping.
DCHECK_LE(real_allocated, accounting_stats_.AllocatedOnPage(page));
DCHECK_EQ(page->allocated_bytes(), accounting_stats_.AllocatedOnPage(page));
}
DCHECK_EQ(total_capacity, accounting_stats_.Capacity());
DCHECK_EQ(total_allocated, accounting_stats_.Size());
}
void PagedSpace::VerifyCountersBeforeConcurrentSweeping() {
// We need to refine the counters on pages that are already swept and have
// not been moved over to the actual space. Otherwise, the AccountingStats
// are just an over approximation.
RefillFreeList();
size_t total_capacity = 0;
size_t total_allocated = 0;
auto marking_state =
heap()->incremental_marking()->non_atomic_marking_state();
for (Page* page : *this) {
size_t page_allocated =
page->SweepingDone()
? page->allocated_bytes()
: static_cast<size_t>(marking_state->live_bytes(page));
total_capacity += page->area_size();
total_allocated += page_allocated;
DCHECK_EQ(page_allocated, accounting_stats_.AllocatedOnPage(page));
}
DCHECK_EQ(total_capacity, accounting_stats_.Capacity());
DCHECK_EQ(total_allocated, accounting_stats_.Size());
}
#endif
// -----------------------------------------------------------------------------
// NewSpace implementation
NewSpace::NewSpace(Heap* heap, v8::PageAllocator* page_allocator,
size_t initial_semispace_capacity,
size_t max_semispace_capacity)
: SpaceWithLinearArea(heap, NEW_SPACE),
to_space_(heap, kToSpace),
from_space_(heap, kFromSpace) {
DCHECK(initial_semispace_capacity <= max_semispace_capacity);
DCHECK(
base::bits::IsPowerOfTwo(static_cast<uint32_t>(max_semispace_capacity)));
to_space_.SetUp(initial_semispace_capacity, max_semispace_capacity);
from_space_.SetUp(initial_semispace_capacity, max_semispace_capacity);
if (!to_space_.Commit()) {
V8::FatalProcessOutOfMemory(heap->isolate(), "New space setup");
}
DCHECK(!from_space_.is_committed()); // No need to use memory yet.
ResetLinearAllocationArea();
}
void NewSpace::TearDown() {
allocation_info_.Reset(kNullAddress, kNullAddress);
to_space_.TearDown();
from_space_.TearDown();
}
void NewSpace::Flip() { SemiSpace::Swap(&from_space_, &to_space_); }
void NewSpace::Grow() {
// Double the semispace size but only up to maximum capacity.
DCHECK(TotalCapacity() < MaximumCapacity());
size_t new_capacity =
Min(MaximumCapacity(),
static_cast<size_t>(FLAG_semi_space_growth_factor) * TotalCapacity());
if (to_space_.GrowTo(new_capacity)) {
// Only grow from space if we managed to grow to-space.
if (!from_space_.GrowTo(new_capacity)) {
// If we managed to grow to-space but couldn't grow from-space,
// attempt to shrink to-space.
if (!to_space_.ShrinkTo(from_space_.current_capacity())) {
// We are in an inconsistent state because we could not
// commit/uncommit memory from new space.
FATAL("inconsistent state");
}
}
}
DCHECK_SEMISPACE_ALLOCATION_INFO(allocation_info_, to_space_);
}
void NewSpace::Shrink() {
size_t new_capacity = Max(InitialTotalCapacity(), 2 * Size());
size_t rounded_new_capacity = ::RoundUp(new_capacity, Page::kPageSize);
if (rounded_new_capacity < TotalCapacity() &&
to_space_.ShrinkTo(rounded_new_capacity)) {
// Only shrink from-space if we managed to shrink to-space.
from_space_.Reset();
if (!from_space_.ShrinkTo(rounded_new_capacity)) {
// If we managed to shrink to-space but couldn't shrink from
// space, attempt to grow to-space again.
if (!to_space_.GrowTo(from_space_.current_capacity())) {
// We are in an inconsistent state because we could not
// commit/uncommit memory from new space.
FATAL("inconsistent state");
}
}
}
DCHECK_SEMISPACE_ALLOCATION_INFO(allocation_info_, to_space_);
}
bool NewSpace::Rebalance() {
// Order here is important to make use of the page pool.
return to_space_.EnsureCurrentCapacity() &&
from_space_.EnsureCurrentCapacity();
}
bool SemiSpace::EnsureCurrentCapacity() {
if (is_committed()) {
const int expected_pages =
static_cast<int>(current_capacity_ / Page::kPageSize);
MemoryChunk* current_page = first_page();
int actual_pages = 0;
// First iterate through the pages list until expected pages if so many
// pages exist.
while (current_page != nullptr && actual_pages < expected_pages) {
actual_pages++;
current_page = current_page->list_node().next();
}
// Free all overallocated pages which are behind current_page.
while (current_page) {
MemoryChunk* next_current = current_page->list_node().next();
memory_chunk_list_.Remove(current_page);
// Clear new space flags to avoid this page being treated as a new
// space page that is potentially being swept.
current_page->SetFlags(0, Page::kIsInYoungGenerationMask);
heap()->memory_allocator()->Free<MemoryAllocator::kPooledAndQueue>(
current_page);
current_page = next_current;
}
// Add more pages if we have less than expected_pages.
IncrementalMarking::NonAtomicMarkingState* marking_state =
heap()->incremental_marking()->non_atomic_marking_state();
while (actual_pages < expected_pages) {
actual_pages++;
current_page =
heap()->memory_allocator()->AllocatePage<MemoryAllocator::kPooled>(
MemoryChunkLayout::AllocatableMemoryInDataPage(), this,
NOT_EXECUTABLE);
if (current_page == nullptr) return false;
DCHECK_NOT_NULL(current_page);
memory_chunk_list_.PushBack(current_page);
marking_state->ClearLiveness(current_page);
current_page->SetFlags(first_page()->GetFlags(),
static_cast<uintptr_t>(Page::kCopyAllFlags));
heap()->CreateFillerObjectAt(current_page->area_start(),
static_cast<int>(current_page->area_size()),
ClearRecordedSlots::kNo);
}
}
return true;
}
LinearAllocationArea LocalAllocationBuffer::Close() {
if (IsValid()) {
heap_->CreateFillerObjectAt(
allocation_info_.top(),
static_cast<int>(allocation_info_.limit() - allocation_info_.top()),
ClearRecordedSlots::kNo);
const LinearAllocationArea old_info = allocation_info_;
allocation_info_ = LinearAllocationArea(kNullAddress, kNullAddress);
return old_info;
}
return LinearAllocationArea(kNullAddress, kNullAddress);
}
LocalAllocationBuffer::LocalAllocationBuffer(
Heap* heap, LinearAllocationArea allocation_info) V8_NOEXCEPT
: heap_(heap),
allocation_info_(allocation_info) {
if (IsValid()) {
heap_->CreateFillerObjectAt(
allocation_info_.top(),
static_cast<int>(allocation_info_.limit() - allocation_info_.top()),
ClearRecordedSlots::kNo);
}
}
LocalAllocationBuffer::LocalAllocationBuffer(const LocalAllocationBuffer& other)
V8_NOEXCEPT {
*this = other;
}
LocalAllocationBuffer& LocalAllocationBuffer::operator=(
const LocalAllocationBuffer& other) V8_NOEXCEPT {
Close();
heap_ = other.heap_;
allocation_info_ = other.allocation_info_;
// This is needed since we (a) cannot yet use move-semantics, and (b) want
// to make the use of the class easy by it as value and (c) implicitly call
// {Close} upon copy.
const_cast<LocalAllocationBuffer&>(other).allocation_info_.Reset(
kNullAddress, kNullAddress);
return *this;
}
void NewSpace::UpdateLinearAllocationArea() {
// Make sure there is no unaccounted allocations.
DCHECK(!AllocationObserversActive() || top_on_previous_step_ == top());
Address new_top = to_space_.page_low();
MemoryChunk::UpdateHighWaterMark(allocation_info_.top());
allocation_info_.Reset(new_top, to_space_.page_high());
// The order of the following two stores is important.
// See the corresponding loads in ConcurrentMarking::Run.
original_limit_.store(limit(), std::memory_order_relaxed);
original_top_.store(top(), std::memory_order_release);
StartNextInlineAllocationStep();
DCHECK_SEMISPACE_ALLOCATION_INFO(allocation_info_, to_space_);
}
void NewSpace::ResetLinearAllocationArea() {
// Do a step to account for memory allocated so far before resetting.
InlineAllocationStep(top(), top(), kNullAddress, 0);
to_space_.Reset();
UpdateLinearAllocationArea();
// Clear all mark-bits in the to-space.
IncrementalMarking::NonAtomicMarkingState* marking_state =
heap()->incremental_marking()->non_atomic_marking_state();
for (Page* p : to_space_) {
marking_state->ClearLiveness(p);
// Concurrent marking may have local live bytes for this page.
heap()->concurrent_marking()->ClearMemoryChunkData(p);
}
}
void NewSpace::UpdateInlineAllocationLimit(size_t min_size) {
Address new_limit = ComputeLimit(top(), to_space_.page_high(), min_size);
allocation_info_.set_limit(new_limit);
DCHECK_SEMISPACE_ALLOCATION_INFO(allocation_info_, to_space_);
}
void PagedSpace::UpdateInlineAllocationLimit(size_t min_size) {
Address new_limit = ComputeLimit(top(), limit(), min_size);
DCHECK_LE(new_limit, limit());
DecreaseLimit(new_limit);
}
bool NewSpace::AddFreshPage() {
Address top = allocation_info_.top();
DCHECK(!OldSpace::IsAtPageStart(top));
// Do a step to account for memory allocated on previous page.
InlineAllocationStep(top, top, kNullAddress, 0);
if (!to_space_.AdvancePage()) {
// No more pages left to advance.
return false;
}
// Clear remainder of current page.
Address limit = Page::FromAllocationAreaAddress(top)->area_end();
int remaining_in_page = static_cast<int>(limit - top);
heap()->CreateFillerObjectAt(top, remaining_in_page, ClearRecordedSlots::kNo);
UpdateLinearAllocationArea();
return true;
}
bool NewSpace::AddFreshPageSynchronized() {
base::MutexGuard guard(&mutex_);
return AddFreshPage();
}
bool NewSpace::EnsureAllocation(int size_in_bytes,
AllocationAlignment alignment) {
Address old_top = allocation_info_.top();
Address high = to_space_.page_high();
int filler_size = Heap::GetFillToAlign(old_top, alignment);
int aligned_size_in_bytes = size_in_bytes + filler_size;
if (old_top + aligned_size_in_bytes > high) {
// Not enough room in the page, try to allocate a new one.
if (!AddFreshPage()) {
return false;
}
old_top = allocation_info_.top();
high = to_space_.page_high();
filler_size = Heap::GetFillToAlign(old_top, alignment);
}
DCHECK(old_top + aligned_size_in_bytes <= high);
if (allocation_info_.limit() < high) {
// Either the limit has been lowered because linear allocation was disabled
// or because incremental marking wants to get a chance to do a step,
// or because idle scavenge job wants to get a chance to post a task.
// Set the new limit accordingly.
Address new_top = old_top + aligned_size_in_bytes;
Address soon_object = old_top + filler_size;
InlineAllocationStep(new_top, new_top, soon_object, size_in_bytes);
UpdateInlineAllocationLimit(aligned_size_in_bytes);
}
return true;
}
size_t LargeObjectSpace::Available() {
// We return zero here since we cannot take advantage of already allocated
// large object memory.
return 0;
}
void SpaceWithLinearArea::StartNextInlineAllocationStep() {
if (heap()->allocation_step_in_progress()) {
// If we are mid-way through an existing step, don't start a new one.
return;
}
if (AllocationObserversActive()) {
top_on_previous_step_ = top();
UpdateInlineAllocationLimit(0);
} else {
DCHECK_EQ(kNullAddress, top_on_previous_step_);
}
}
void SpaceWithLinearArea::AddAllocationObserver(AllocationObserver* observer) {
InlineAllocationStep(top(), top(), kNullAddress, 0);
Space::AddAllocationObserver(observer);
DCHECK_IMPLIES(top_on_previous_step_, AllocationObserversActive());
}
void SpaceWithLinearArea::RemoveAllocationObserver(
AllocationObserver* observer) {
Address top_for_next_step =
allocation_observers_.size() == 1 ? kNullAddress : top();
InlineAllocationStep(top(), top_for_next_step, kNullAddress, 0);
Space::RemoveAllocationObserver(observer);
DCHECK_IMPLIES(top_on_previous_step_, AllocationObserversActive());
}
void SpaceWithLinearArea::PauseAllocationObservers() {
// Do a step to account for memory allocated so far.
InlineAllocationStep(top(), kNullAddress, kNullAddress, 0);
Space::PauseAllocationObservers();
DCHECK_EQ(kNullAddress, top_on_previous_step_);
UpdateInlineAllocationLimit(0);
}
void SpaceWithLinearArea::ResumeAllocationObservers() {
DCHECK_EQ(kNullAddress, top_on_previous_step_);
Space::ResumeAllocationObservers();
StartNextInlineAllocationStep();
}
void SpaceWithLinearArea::InlineAllocationStep(Address top,
Address top_for_next_step,
Address soon_object,
size_t size) {
if (heap()->allocation_step_in_progress()) {
// Avoid starting a new step if we are mid-way through an existing one.
return;
}
if (top_on_previous_step_) {
if (top < top_on_previous_step_) {
// Generated code decreased the top pointer to do folded allocations.
DCHECK_NE(top, kNullAddress);
DCHECK_EQ(Page::FromAllocationAreaAddress(top),
Page::FromAllocationAreaAddress(top_on_previous_step_));
top_on_previous_step_ = top;
}
int bytes_allocated = static_cast<int>(top - top_on_previous_step_);
AllocationStep(bytes_allocated, soon_object, static_cast<int>(size));
top_on_previous_step_ = top_for_next_step;
}
}
std::unique_ptr<ObjectIterator> NewSpace::GetObjectIterator() {
return std::unique_ptr<ObjectIterator>(new SemiSpaceIterator(this));
}
#ifdef VERIFY_HEAP
// We do not use the SemiSpaceIterator because verification doesn't assume
// that it works (it depends on the invariants we are checking).
void NewSpace::Verify(Isolate* isolate) {
// The allocation pointer should be in the space or at the very end.
DCHECK_SEMISPACE_ALLOCATION_INFO(allocation_info_, to_space_);
// There should be objects packed in from the low address up to the
// allocation pointer.
Address current = to_space_.first_page()->area_start();
CHECK_EQ(current, to_space_.space_start());
size_t external_space_bytes[kNumTypes];
for (int i = 0; i < kNumTypes; i++) {
external_space_bytes[static_cast<ExternalBackingStoreType>(i)] = 0;
}
while (current != top()) {
if (!Page::IsAlignedToPageSize(current)) {
// The allocation pointer should not be in the middle of an object.
CHECK(!Page::FromAllocationAreaAddress(current)->ContainsLimit(top()) ||
current < top());
HeapObject object = HeapObject::FromAddress(current);
// The first word should be a map, and we expect all map pointers to
// be in map space or read-only space.
Map map = object.map();
CHECK(map.IsMap());
CHECK(heap()->map_space()->Contains(map) ||
heap()->read_only_space()->Contains(map));
// The object should not be code or a map.
CHECK(!object.IsMap());
CHECK(!object.IsAbstractCode());
// The object itself should look OK.
object.ObjectVerify(isolate);
// All the interior pointers should be contained in the heap.
VerifyPointersVisitor visitor(heap());