v8/test/unittests/heap/heap-utils.cc
Omer Katz 4d95ff1a21 [heap] Prevent new space from allocation small LABs
Update the free list implementation for new space to set a larger
minimum size and skip redundant step in the allocation logic.

Bug: v8:12612
Change-Id: I480fe99cf4cfad7c25d687540b7841cd56d41d47
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3976508
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83920}
2022-10-26 09:12:04 +00:00

287 lines
11 KiB
C++

// Copyright 2020 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 "test/unittests/heap/heap-utils.h"
#include <algorithm>
#include "src/common/globals.h"
#include "src/flags/flags.h"
#include "src/heap/gc-tracer-inl.h"
#include "src/heap/incremental-marking.h"
#include "src/heap/mark-compact.h"
#include "src/heap/new-spaces.h"
#include "src/heap/safepoint.h"
#include "src/objects/free-space-inl.h"
namespace v8 {
namespace internal {
void HeapInternalsBase::SimulateIncrementalMarking(Heap* heap,
bool force_completion) {
constexpr double kStepSizeInMs = 100;
CHECK(v8_flags.incremental_marking);
i::IncrementalMarking* marking = heap->incremental_marking();
if (heap->sweeping_in_progress()) {
IsolateSafepointScope scope(heap);
heap->EnsureSweepingCompleted(
Heap::SweepingForcedFinalizationMode::kV8Only);
}
if (marking->IsStopped()) {
heap->StartIncrementalMarking(i::Heap::kNoGCFlags,
i::GarbageCollectionReason::kTesting);
}
CHECK(marking->IsMajorMarking());
if (!force_completion) return;
while (!marking->IsMajorMarkingComplete()) {
marking->AdvanceForTesting(kStepSizeInMs);
}
}
namespace {
int FixedArrayLenFromSize(int size) {
return std::min({(size - FixedArray::kHeaderSize) / kTaggedSize,
FixedArray::kMaxRegularLength});
}
void FillPageInPagedSpace(Page* page,
std::vector<Handle<FixedArray>>* out_handles) {
Heap* heap = page->heap();
DCHECK(page->SweepingDone());
PagedSpaceBase* paged_space = static_cast<PagedSpaceBase*>(page->owner());
// Make sure the LAB is empty to guarantee that all free space is accounted
// for in the freelist.
DCHECK_EQ(paged_space->limit(), paged_space->top());
PauseAllocationObserversScope no_observers_scope(heap);
CollectionEpoch full_epoch =
heap->tracer()->CurrentEpoch(GCTracer::Scope::ScopeId::MARK_COMPACTOR);
CollectionEpoch young_epoch = heap->tracer()->CurrentEpoch(
GCTracer::Scope::ScopeId::MINOR_MARK_COMPACTOR);
for (Page* p : *paged_space) {
if (p != page) paged_space->UnlinkFreeListCategories(p);
}
// If min_block_size is larger than FixedArray::kHeaderSize, all blocks in the
// free list can be used to allocate a fixed array. This guarantees that we
// can fill the whole page.
DCHECK_LT(FixedArray::kHeaderSize,
paged_space->free_list()->min_block_size());
std::vector<int> available_sizes;
// Collect all free list block sizes
page->ForAllFreeListCategories(
[&available_sizes](FreeListCategory* category) {
category->IterateNodesForTesting([&available_sizes](FreeSpace node) {
int node_size = node.Size();
if (node_size >= kMaxRegularHeapObjectSize) {
available_sizes.push_back(node_size);
}
});
});
Isolate* isolate = heap->isolate();
// Allocate as many max size arrays as possible, while making sure not to
// leave behind a block too small to fit a FixedArray.
const int max_array_length = FixedArrayLenFromSize(kMaxRegularHeapObjectSize);
for (size_t i = 0; i < available_sizes.size(); ++i) {
int available_size = available_sizes[i];
while (available_size > kMaxRegularHeapObjectSize) {
Handle<FixedArray> fixed_array = isolate->factory()->NewFixedArray(
max_array_length, AllocationType::kYoung);
if (out_handles) out_handles->push_back(fixed_array);
available_size -= kMaxRegularHeapObjectSize;
}
}
paged_space->FreeLinearAllocationArea();
// Allocate FixedArrays in remaining free list blocks, from largest
// category to smallest.
std::vector<std::vector<int>> remaining_sizes;
page->ForAllFreeListCategories(
[&remaining_sizes](FreeListCategory* category) {
remaining_sizes.push_back({});
std::vector<int>& sizes_in_category =
remaining_sizes[remaining_sizes.size() - 1];
category->IterateNodesForTesting([&sizes_in_category](FreeSpace node) {
int node_size = node.Size();
DCHECK_LT(0, FixedArrayLenFromSize(node_size));
sizes_in_category.push_back(node_size);
});
});
for (auto it = remaining_sizes.rbegin(); it != remaining_sizes.rend(); ++it) {
std::vector<int> sizes_in_category = *it;
for (int size : sizes_in_category) {
DCHECK_LE(size, kMaxRegularHeapObjectSize);
int array_length = FixedArrayLenFromSize(size);
DCHECK_LT(0, array_length);
Handle<FixedArray> fixed_array = isolate->factory()->NewFixedArray(
array_length, AllocationType::kYoung);
if (out_handles) out_handles->push_back(fixed_array);
}
}
DCHECK_EQ(0, page->AvailableInFreeList());
DCHECK_EQ(0, page->AvailableInFreeListFromAllocatedBytes());
for (Page* p : *paged_space) {
if (p != page) paged_space->RelinkFreeListCategories(p);
}
// Allocations in this method should not require a GC.
CHECK_EQ(full_epoch, heap->tracer()->CurrentEpoch(
GCTracer::Scope::ScopeId::MARK_COMPACTOR));
CHECK_EQ(young_epoch, heap->tracer()->CurrentEpoch(
GCTracer::Scope::ScopeId::MINOR_MARK_COMPACTOR));
}
} // namespace
void HeapInternalsBase::SimulateFullSpace(
v8::internal::NewSpace* space,
std::vector<Handle<FixedArray>>* out_handles) {
// If you see this check failing, disable the flag at the start of your test:
// v8_flags.stress_concurrent_allocation = false;
// Background thread allocating concurrently interferes with this function.
CHECK(!v8_flags.stress_concurrent_allocation);
space->heap()->EnsureSweepingCompleted(
Heap::SweepingForcedFinalizationMode::kV8Only);
space->FreeLinearAllocationArea();
if (v8_flags.minor_mc) {
for (Page* page : *space) {
FillPageInPagedSpace(page, out_handles);
}
DCHECK_EQ(0, space->free_list()->Available());
} else {
do {
FillCurrentPage(space, out_handles);
} while (space->AddFreshPage());
}
}
void HeapInternalsBase::SimulateFullSpace(v8::internal::PagedSpace* space) {
// If you see this check failing, disable the flag at the start of your test:
// v8_flags.stress_concurrent_allocation = false;
// Background thread allocating concurrently interferes with this function.
CHECK(!v8_flags.stress_concurrent_allocation);
Heap* heap = space->heap();
CodePageCollectionMemoryModificationScopeForTesting code_scope(heap);
if (heap->sweeping_in_progress()) {
heap->EnsureSweepingCompleted(
Heap::SweepingForcedFinalizationMode::kV8Only);
}
space->FreeLinearAllocationArea();
space->ResetFreeList();
}
namespace {
int GetSpaceRemainingOnCurrentSemiSpacePage(v8::internal::NewSpace* space) {
Address top = space->top();
if ((top & kPageAlignmentMask) == 0) {
// `top` points to the start of a page signifies that there is not room in
// the current page.
return 0;
}
return static_cast<int>(Page::FromAddress(space->top())->area_end() - top);
}
std::vector<Handle<FixedArray>> CreatePadding(Heap* heap, int padding_size,
AllocationType allocation) {
std::vector<Handle<FixedArray>> handles;
Isolate* isolate = heap->isolate();
int allocate_memory;
int length;
int free_memory = padding_size;
if (allocation == i::AllocationType::kOld) {
heap->old_space()->FreeLinearAllocationArea();
int overall_free_memory = static_cast<int>(heap->old_space()->Available());
CHECK(padding_size <= overall_free_memory || overall_free_memory == 0);
} else {
int overall_free_memory = static_cast<int>(heap->new_space()->Available());
CHECK(padding_size <= overall_free_memory || overall_free_memory == 0);
}
while (free_memory > 0) {
if (free_memory > kMaxRegularHeapObjectSize) {
allocate_memory = kMaxRegularHeapObjectSize;
length = FixedArrayLenFromSize(allocate_memory);
} else {
allocate_memory = free_memory;
length = FixedArrayLenFromSize(allocate_memory);
if (length <= 0) {
// Not enough room to create another FixedArray, so create a filler.
if (allocation == i::AllocationType::kOld) {
heap->CreateFillerObjectAt(
*heap->old_space()->allocation_top_address(), free_memory);
} else {
heap->CreateFillerObjectAt(
*heap->new_space()->allocation_top_address(), free_memory);
}
break;
}
}
handles.push_back(isolate->factory()->NewFixedArray(length, allocation));
CHECK((allocation == AllocationType::kYoung &&
heap->new_space()->Contains(*handles.back())) ||
(allocation == AllocationType::kOld &&
heap->InOldSpace(*handles.back())) ||
v8_flags.single_generation);
free_memory -= handles.back()->Size();
}
return handles;
}
void FillCurrentSemiSpacePage(v8::internal::NewSpace* space,
std::vector<Handle<FixedArray>>* out_handles) {
// We cannot rely on `space->limit()` to point to the end of the current page
// in the case where inline allocations are disabled, it actually points to
// the current allocation pointer.
DCHECK_IMPLIES(!space->IsInlineAllocationEnabled(),
space->limit() == space->top());
int space_remaining = GetSpaceRemainingOnCurrentSemiSpacePage(space);
if (space_remaining == 0) return;
std::vector<Handle<FixedArray>> handles =
CreatePadding(space->heap(), space_remaining, i::AllocationType::kYoung);
if (out_handles != nullptr) {
out_handles->insert(out_handles->end(), handles.begin(), handles.end());
}
}
void FillCurrenPagedSpacePage(v8::internal::NewSpace* space,
std::vector<Handle<FixedArray>>* out_handles) {
if (space->top() == kNullAddress) return;
Page* page = Page::FromAllocationAreaAddress(space->top());
space->heap()->EnsureSweepingCompleted(
Heap::SweepingForcedFinalizationMode::kV8Only);
space->FreeLinearAllocationArea();
FillPageInPagedSpace(page, out_handles);
}
} // namespace
void HeapInternalsBase::FillCurrentPage(
v8::internal::NewSpace* space,
std::vector<Handle<FixedArray>>* out_handles) {
PauseAllocationObserversScope pause_observers(space->heap());
if (v8_flags.minor_mc)
FillCurrenPagedSpacePage(space, out_handles);
else
FillCurrentSemiSpacePage(space, out_handles);
}
bool IsNewObjectInCorrectGeneration(HeapObject object) {
return v8_flags.single_generation ? !i::Heap::InYoungGeneration(object)
: i::Heap::InYoungGeneration(object);
}
} // namespace internal
} // namespace v8