v8/test/cctest/heap/test-heap.cc

7105 lines
242 KiB
C++
Raw Normal View History

// Copyright 2012 the V8 project authors. All rights reserved.
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following
// disclaimer in the documentation and/or other materials provided
// with the distribution.
// * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived
// from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include <stdlib.h>
#include <utility>
#include "src/compilation-cache.h"
#include "src/context-measure.h"
#include "src/deoptimizer.h"
#include "src/elements.h"
#include "src/execution.h"
#include "src/factory.h"
#include "src/field-type.h"
#include "src/global-handles.h"
#include "src/heap/gc-tracer.h"
#include "src/heap/memory-reducer.h"
#include "src/ic/ic.h"
#include "src/macro-assembler.h"
#include "src/regexp/jsregexp.h"
#include "src/snapshot/snapshot.h"
#include "test/cctest/cctest.h"
#include "test/cctest/heap/heap-tester.h"
#include "test/cctest/heap/heap-utils.h"
#include "test/cctest/test-feedback-vector.h"
namespace v8 {
namespace internal {
static void CheckMap(Map* map, int type, int instance_size) {
CHECK(map->IsHeapObject());
#ifdef DEBUG
CHECK(CcTest::heap()->Contains(map));
#endif
CHECK_EQ(CcTest::heap()->meta_map(), map->map());
CHECK_EQ(type, map->instance_type());
CHECK_EQ(instance_size, map->instance_size());
}
TEST(HeapMaps) {
CcTest::InitializeVM();
Heap* heap = CcTest::heap();
CheckMap(heap->meta_map(), MAP_TYPE, Map::kSize);
CheckMap(heap->heap_number_map(), HEAP_NUMBER_TYPE, HeapNumber::kSize);
#define SIMD128_TYPE(TYPE, Type, type, lane_count, lane_type) \
CheckMap(heap->type##_map(), SIMD128_VALUE_TYPE, Type::kSize);
SIMD128_TYPES(SIMD128_TYPE)
#undef SIMD128_TYPE
CheckMap(heap->fixed_array_map(), FIXED_ARRAY_TYPE, kVariableSizeSentinel);
CheckMap(heap->string_map(), STRING_TYPE, kVariableSizeSentinel);
}
static void CheckOddball(Isolate* isolate, Object* obj, const char* string) {
CHECK(obj->IsOddball());
Handle<Object> handle(obj, isolate);
Object* print_string = *Object::ToString(isolate, handle).ToHandleChecked();
CHECK(String::cast(print_string)->IsUtf8EqualTo(CStrVector(string)));
}
static void CheckSmi(Isolate* isolate, int value, const char* string) {
Handle<Object> handle(Smi::FromInt(value), isolate);
Object* print_string = *Object::ToString(isolate, handle).ToHandleChecked();
CHECK(String::cast(print_string)->IsUtf8EqualTo(CStrVector(string)));
}
static void CheckNumber(Isolate* isolate, double value, const char* string) {
Handle<Object> number = isolate->factory()->NewNumber(value);
CHECK(number->IsNumber());
Handle<Object> print_string =
Object::ToString(isolate, number).ToHandleChecked();
CHECK(String::cast(*print_string)->IsUtf8EqualTo(CStrVector(string)));
}
void CheckEmbeddedObjectsAreEqual(Handle<Code> lhs, Handle<Code> rhs) {
int mode_mask = RelocInfo::ModeMask(RelocInfo::EMBEDDED_OBJECT);
RelocIterator lhs_it(*lhs, mode_mask);
RelocIterator rhs_it(*rhs, mode_mask);
while (!lhs_it.done() && !rhs_it.done()) {
CHECK(lhs_it.rinfo()->target_object() == rhs_it.rinfo()->target_object());
lhs_it.next();
rhs_it.next();
}
CHECK(lhs_it.done() == rhs_it.done());
}
HEAP_TEST(TestNewSpaceRefsInCopiedCode) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
HandleScope sc(isolate);
Handle<Object> value = factory->NewNumber(1.000123);
CHECK(heap->InNewSpace(*value));
i::byte buffer[i::Assembler::kMinimalBufferSize];
MacroAssembler masm(isolate, buffer, sizeof(buffer),
v8::internal::CodeObjectRequired::kYes);
// Add a new-space reference to the code.
masm.Push(value);
CodeDesc desc;
masm.GetCode(&desc);
Handle<Code> code = isolate->factory()->NewCode(
desc, Code::ComputeFlags(Code::STUB), Handle<Code>());
Code* tmp = nullptr;
heap->CopyCode(*code).To(&tmp);
Handle<Code> copy(tmp);
CheckEmbeddedObjectsAreEqual(code, copy);
heap->CollectAllAvailableGarbage();
CheckEmbeddedObjectsAreEqual(code, copy);
}
static void CheckFindCodeObject(Isolate* isolate) {
// Test FindCodeObject
#define __ assm.
Assembler assm(isolate, NULL, 0);
__ nop(); // supported on all architectures
CodeDesc desc;
assm.GetCode(&desc);
Handle<Code> code = isolate->factory()->NewCode(
desc, Code::ComputeFlags(Code::STUB), Handle<Code>());
CHECK(code->IsCode());
HeapObject* obj = HeapObject::cast(*code);
Address obj_addr = obj->address();
for (int i = 0; i < obj->Size(); i += kPointerSize) {
Object* found = isolate->FindCodeObject(obj_addr + i);
CHECK_EQ(*code, found);
}
Handle<Code> copy = isolate->factory()->NewCode(
desc, Code::ComputeFlags(Code::STUB), Handle<Code>());
HeapObject* obj_copy = HeapObject::cast(*copy);
Object* not_right = isolate->FindCodeObject(obj_copy->address() +
obj_copy->Size() / 2);
CHECK(not_right != *code);
}
TEST(HandleNull) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
HandleScope outer_scope(isolate);
LocalContext context;
Handle<Object> n(static_cast<Object*>(nullptr), isolate);
CHECK(!n.is_null());
}
TEST(HeapObjects) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
HandleScope sc(isolate);
Handle<Object> value = factory->NewNumber(1.000123);
CHECK(value->IsHeapNumber());
CHECK(value->IsNumber());
CHECK_EQ(1.000123, value->Number());
value = factory->NewNumber(1.0);
CHECK(value->IsSmi());
CHECK(value->IsNumber());
CHECK_EQ(1.0, value->Number());
value = factory->NewNumberFromInt(1024);
CHECK(value->IsSmi());
CHECK(value->IsNumber());
CHECK_EQ(1024.0, value->Number());
value = factory->NewNumberFromInt(Smi::kMinValue);
CHECK(value->IsSmi());
CHECK(value->IsNumber());
CHECK_EQ(Smi::kMinValue, Handle<Smi>::cast(value)->value());
value = factory->NewNumberFromInt(Smi::kMaxValue);
CHECK(value->IsSmi());
CHECK(value->IsNumber());
CHECK_EQ(Smi::kMaxValue, Handle<Smi>::cast(value)->value());
#if !defined(V8_TARGET_ARCH_64_BIT)
// TODO(lrn): We need a NumberFromIntptr function in order to test this.
value = factory->NewNumberFromInt(Smi::kMinValue - 1);
CHECK(value->IsHeapNumber());
CHECK(value->IsNumber());
CHECK_EQ(static_cast<double>(Smi::kMinValue - 1), value->Number());
#endif
value = factory->NewNumberFromUint(static_cast<uint32_t>(Smi::kMaxValue) + 1);
CHECK(value->IsHeapNumber());
CHECK(value->IsNumber());
CHECK_EQ(static_cast<double>(static_cast<uint32_t>(Smi::kMaxValue) + 1),
value->Number());
value = factory->NewNumberFromUint(static_cast<uint32_t>(1) << 31);
CHECK(value->IsHeapNumber());
CHECK(value->IsNumber());
CHECK_EQ(static_cast<double>(static_cast<uint32_t>(1) << 31),
value->Number());
// nan oddball checks
CHECK(factory->nan_value()->IsNumber());
CHECK(std::isnan(factory->nan_value()->Number()));
Handle<String> s = factory->NewStringFromStaticChars("fisk hest ");
CHECK(s->IsString());
CHECK_EQ(10, s->length());
Handle<String> object_string = Handle<String>::cast(factory->Object_string());
Handle<JSGlobalObject> global(
CcTest::i_isolate()->context()->global_object());
CHECK(Just(true) == JSReceiver::HasOwnProperty(global, object_string));
// Check ToString for oddballs
CheckOddball(isolate, heap->true_value(), "true");
CheckOddball(isolate, heap->false_value(), "false");
CheckOddball(isolate, heap->null_value(), "null");
CheckOddball(isolate, heap->undefined_value(), "undefined");
// Check ToString for Smis
CheckSmi(isolate, 0, "0");
CheckSmi(isolate, 42, "42");
CheckSmi(isolate, -42, "-42");
// Check ToString for Numbers
CheckNumber(isolate, 1.1, "1.1");
CheckFindCodeObject(isolate);
}
template <typename T, typename LANE_TYPE, int LANES>
static void CheckSimdValue(T* value, LANE_TYPE lane_values[LANES],
LANE_TYPE other_value) {
// Check against lane_values, and check that all lanes can be set to
// other_value without disturbing the other lanes.
for (int i = 0; i < LANES; i++) {
CHECK_EQ(lane_values[i], value->get_lane(i));
}
for (int i = 0; i < LANES; i++) {
value->set_lane(i, other_value); // change the value
for (int j = 0; j < LANES; j++) {
if (i != j)
CHECK_EQ(lane_values[j], value->get_lane(j));
else
CHECK_EQ(other_value, value->get_lane(j));
}
value->set_lane(i, lane_values[i]); // restore the lane
}
CHECK(value->BooleanValue()); // SIMD values are 'true'.
}
TEST(SimdObjects) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
HandleScope sc(isolate);
// Float32x4
{
float lanes[4] = {1, 2, 3, 4};
float quiet_NaN = std::numeric_limits<float>::quiet_NaN();
float signaling_NaN = std::numeric_limits<float>::signaling_NaN();
Handle<Float32x4> value = factory->NewFloat32x4(lanes);
CHECK(value->IsFloat32x4());
CheckSimdValue<Float32x4, float, 4>(*value, lanes, 3.14f);
// Check special lane values.
value->set_lane(1, -0.0);
CHECK_EQ(-0.0f, value->get_lane(1));
CHECK(std::signbit(value->get_lane(1))); // Sign bit should be preserved.
value->set_lane(2, quiet_NaN);
CHECK(std::isnan(value->get_lane(2)));
value->set_lane(3, signaling_NaN);
CHECK(std::isnan(value->get_lane(3)));
#ifdef OBJECT_PRINT
// Check value printing.
{
value = factory->NewFloat32x4(lanes);
std::ostringstream os;
value->Float32x4Print(os);
CHECK_EQ("1, 2, 3, 4", os.str());
}
{
float special_lanes[4] = {0, -0.0, quiet_NaN, signaling_NaN};
value = factory->NewFloat32x4(special_lanes);
std::ostringstream os;
value->Float32x4Print(os);
// Value printing doesn't preserve signed zeroes.
CHECK_EQ("0, 0, NaN, NaN", os.str());
}
#endif // OBJECT_PRINT
}
// Int32x4
{
int32_t lanes[4] = {1, 2, 3, 4};
Handle<Int32x4> value = factory->NewInt32x4(lanes);
CHECK(value->IsInt32x4());
CheckSimdValue<Int32x4, int32_t, 4>(*value, lanes, 3);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Int32x4Print(os);
CHECK_EQ("1, 2, 3, 4", os.str());
#endif // OBJECT_PRINT
}
// Uint32x4
{
uint32_t lanes[4] = {1, 2, 3, 4};
Handle<Uint32x4> value = factory->NewUint32x4(lanes);
CHECK(value->IsUint32x4());
CheckSimdValue<Uint32x4, uint32_t, 4>(*value, lanes, 3);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Uint32x4Print(os);
CHECK_EQ("1, 2, 3, 4", os.str());
#endif // OBJECT_PRINT
}
// Bool32x4
{
bool lanes[4] = {true, false, true, false};
Handle<Bool32x4> value = factory->NewBool32x4(lanes);
CHECK(value->IsBool32x4());
CheckSimdValue<Bool32x4, bool, 4>(*value, lanes, false);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Bool32x4Print(os);
CHECK_EQ("true, false, true, false", os.str());
#endif // OBJECT_PRINT
}
// Int16x8
{
int16_t lanes[8] = {1, 2, 3, 4, 5, 6, 7, 8};
Handle<Int16x8> value = factory->NewInt16x8(lanes);
CHECK(value->IsInt16x8());
CheckSimdValue<Int16x8, int16_t, 8>(*value, lanes, 32767);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Int16x8Print(os);
CHECK_EQ("1, 2, 3, 4, 5, 6, 7, 8", os.str());
#endif // OBJECT_PRINT
}
// Uint16x8
{
uint16_t lanes[8] = {1, 2, 3, 4, 5, 6, 7, 8};
Handle<Uint16x8> value = factory->NewUint16x8(lanes);
CHECK(value->IsUint16x8());
CheckSimdValue<Uint16x8, uint16_t, 8>(*value, lanes, 32767);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Uint16x8Print(os);
CHECK_EQ("1, 2, 3, 4, 5, 6, 7, 8", os.str());
#endif // OBJECT_PRINT
}
// Bool16x8
{
bool lanes[8] = {true, false, true, false, true, false, true, false};
Handle<Bool16x8> value = factory->NewBool16x8(lanes);
CHECK(value->IsBool16x8());
CheckSimdValue<Bool16x8, bool, 8>(*value, lanes, false);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Bool16x8Print(os);
CHECK_EQ("true, false, true, false, true, false, true, false", os.str());
#endif // OBJECT_PRINT
}
// Int8x16
{
int8_t lanes[16] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
Handle<Int8x16> value = factory->NewInt8x16(lanes);
CHECK(value->IsInt8x16());
CheckSimdValue<Int8x16, int8_t, 16>(*value, lanes, 127);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Int8x16Print(os);
CHECK_EQ("1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16", os.str());
#endif // OBJECT_PRINT
}
// Uint8x16
{
uint8_t lanes[16] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
Handle<Uint8x16> value = factory->NewUint8x16(lanes);
CHECK(value->IsUint8x16());
CheckSimdValue<Uint8x16, uint8_t, 16>(*value, lanes, 127);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Uint8x16Print(os);
CHECK_EQ("1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16", os.str());
#endif // OBJECT_PRINT
}
// Bool8x16
{
bool lanes[16] = {true, false, true, false, true, false, true, false,
true, false, true, false, true, false, true, false};
Handle<Bool8x16> value = factory->NewBool8x16(lanes);
CHECK(value->IsBool8x16());
CheckSimdValue<Bool8x16, bool, 16>(*value, lanes, false);
#ifdef OBJECT_PRINT
std::ostringstream os;
value->Bool8x16Print(os);
CHECK_EQ(
"true, false, true, false, true, false, true, false, true, false, "
"true, false, true, false, true, false",
os.str());
#endif // OBJECT_PRINT
}
}
TEST(Tagging) {
CcTest::InitializeVM();
int request = 24;
CHECK_EQ(request, static_cast<int>(OBJECT_POINTER_ALIGN(request)));
CHECK(Smi::FromInt(42)->IsSmi());
CHECK(Smi::FromInt(Smi::kMinValue)->IsSmi());
CHECK(Smi::FromInt(Smi::kMaxValue)->IsSmi());
}
TEST(GarbageCollection) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope sc(isolate);
// Check GC.
heap->CollectGarbage(NEW_SPACE);
Handle<JSGlobalObject> global(
CcTest::i_isolate()->context()->global_object());
Handle<String> name = factory->InternalizeUtf8String("theFunction");
Handle<String> prop_name = factory->InternalizeUtf8String("theSlot");
Handle<String> prop_namex = factory->InternalizeUtf8String("theSlotx");
Handle<String> obj_name = factory->InternalizeUtf8String("theObject");
Handle<Smi> twenty_three(Smi::FromInt(23), isolate);
Handle<Smi> twenty_four(Smi::FromInt(24), isolate);
{
HandleScope inner_scope(isolate);
// Allocate a function and keep it in global object's property.
Handle<JSFunction> function = factory->NewFunction(name);
JSReceiver::SetProperty(global, name, function, SLOPPY).Check();
// Allocate an object. Unrooted after leaving the scope.
Handle<JSObject> obj = factory->NewJSObject(function);
JSReceiver::SetProperty(obj, prop_name, twenty_three, SLOPPY).Check();
JSReceiver::SetProperty(obj, prop_namex, twenty_four, SLOPPY).Check();
CHECK_EQ(Smi::FromInt(23),
*Object::GetProperty(obj, prop_name).ToHandleChecked());
CHECK_EQ(Smi::FromInt(24),
*Object::GetProperty(obj, prop_namex).ToHandleChecked());
}
heap->CollectGarbage(NEW_SPACE);
// Function should be alive.
CHECK(Just(true) == JSReceiver::HasOwnProperty(global, name));
// Check function is retained.
Handle<Object> func_value =
Object::GetProperty(global, name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
{
HandleScope inner_scope(isolate);
// Allocate another object, make it reachable from global.
Handle<JSObject> obj = factory->NewJSObject(function);
JSReceiver::SetProperty(global, obj_name, obj, SLOPPY).Check();
JSReceiver::SetProperty(obj, prop_name, twenty_three, SLOPPY).Check();
}
// After gc, it should survive.
heap->CollectGarbage(NEW_SPACE);
CHECK(Just(true) == JSReceiver::HasOwnProperty(global, obj_name));
Handle<Object> obj =
Object::GetProperty(global, obj_name).ToHandleChecked();
CHECK(obj->IsJSObject());
CHECK_EQ(Smi::FromInt(23),
*Object::GetProperty(obj, prop_name).ToHandleChecked());
}
static void VerifyStringAllocation(Isolate* isolate, const char* string) {
HandleScope scope(isolate);
Handle<String> s = isolate->factory()->NewStringFromUtf8(
CStrVector(string)).ToHandleChecked();
CHECK_EQ(StrLength(string), s->length());
for (int index = 0; index < s->length(); index++) {
CHECK_EQ(static_cast<uint16_t>(string[index]), s->Get(index));
}
}
TEST(String) {
CcTest::InitializeVM();
Isolate* isolate = reinterpret_cast<Isolate*>(CcTest::isolate());
VerifyStringAllocation(isolate, "a");
VerifyStringAllocation(isolate, "ab");
VerifyStringAllocation(isolate, "abc");
VerifyStringAllocation(isolate, "abcd");
VerifyStringAllocation(isolate, "fiskerdrengen er paa havet");
}
TEST(LocalHandles) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
const char* name = "Kasper the spunky";
Handle<String> string = factory->NewStringFromAsciiChecked(name);
CHECK_EQ(StrLength(name), string->length());
}
TEST(GlobalHandles) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
GlobalHandles* global_handles = isolate->global_handles();
Handle<Object> h1;
Handle<Object> h2;
Handle<Object> h3;
Handle<Object> h4;
{
HandleScope scope(isolate);
Handle<Object> i = factory->NewStringFromStaticChars("fisk");
Handle<Object> u = factory->NewNumber(1.12344);
h1 = global_handles->Create(*i);
h2 = global_handles->Create(*u);
h3 = global_handles->Create(*i);
h4 = global_handles->Create(*u);
}
// after gc, it should survive
heap->CollectGarbage(NEW_SPACE);
CHECK((*h1)->IsString());
CHECK((*h2)->IsHeapNumber());
CHECK((*h3)->IsString());
CHECK((*h4)->IsHeapNumber());
CHECK_EQ(*h3, *h1);
GlobalHandles::Destroy(h1.location());
GlobalHandles::Destroy(h3.location());
CHECK_EQ(*h4, *h2);
GlobalHandles::Destroy(h2.location());
GlobalHandles::Destroy(h4.location());
}
static bool WeakPointerCleared = false;
static void TestWeakGlobalHandleCallback(
const v8::WeakCallbackInfo<void>& data) {
std::pair<v8::Persistent<v8::Value>*, int>* p =
reinterpret_cast<std::pair<v8::Persistent<v8::Value>*, int>*>(
data.GetParameter());
if (p->second == 1234) WeakPointerCleared = true;
p->first->Reset();
}
TEST(WeakGlobalHandlesScavenge) {
i::FLAG_stress_compaction = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
GlobalHandles* global_handles = isolate->global_handles();
WeakPointerCleared = false;
Handle<Object> h1;
Handle<Object> h2;
{
HandleScope scope(isolate);
Handle<Object> i = factory->NewStringFromStaticChars("fisk");
Handle<Object> u = factory->NewNumber(1.12344);
h1 = global_handles->Create(*i);
h2 = global_handles->Create(*u);
}
std::pair<Handle<Object>*, int> handle_and_id(&h2, 1234);
GlobalHandles::MakeWeak(
h2.location(), reinterpret_cast<void*>(&handle_and_id),
&TestWeakGlobalHandleCallback, v8::WeakCallbackType::kParameter);
// Scavenge treats weak pointers as normal roots.
heap->CollectGarbage(NEW_SPACE);
CHECK((*h1)->IsString());
CHECK((*h2)->IsHeapNumber());
CHECK(!WeakPointerCleared);
CHECK(!global_handles->IsNearDeath(h2.location()));
CHECK(!global_handles->IsNearDeath(h1.location()));
GlobalHandles::Destroy(h1.location());
GlobalHandles::Destroy(h2.location());
}
TEST(WeakGlobalHandlesMark) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
GlobalHandles* global_handles = isolate->global_handles();
WeakPointerCleared = false;
Handle<Object> h1;
Handle<Object> h2;
{
HandleScope scope(isolate);
Handle<Object> i = factory->NewStringFromStaticChars("fisk");
Handle<Object> u = factory->NewNumber(1.12344);
h1 = global_handles->Create(*i);
h2 = global_handles->Create(*u);
}
// Make sure the objects are promoted.
heap->CollectGarbage(OLD_SPACE);
heap->CollectGarbage(NEW_SPACE);
CHECK(!heap->InNewSpace(*h1) && !heap->InNewSpace(*h2));
std::pair<Handle<Object>*, int> handle_and_id(&h2, 1234);
GlobalHandles::MakeWeak(
h2.location(), reinterpret_cast<void*>(&handle_and_id),
&TestWeakGlobalHandleCallback, v8::WeakCallbackType::kParameter);
CHECK(!GlobalHandles::IsNearDeath(h1.location()));
CHECK(!GlobalHandles::IsNearDeath(h2.location()));
// Incremental marking potentially marked handles before they turned weak.
heap->CollectAllGarbage();
CHECK((*h1)->IsString());
CHECK(WeakPointerCleared);
CHECK(!GlobalHandles::IsNearDeath(h1.location()));
GlobalHandles::Destroy(h1.location());
}
TEST(DeleteWeakGlobalHandle) {
i::FLAG_stress_compaction = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
GlobalHandles* global_handles = isolate->global_handles();
WeakPointerCleared = false;
Handle<Object> h;
{
HandleScope scope(isolate);
Handle<Object> i = factory->NewStringFromStaticChars("fisk");
h = global_handles->Create(*i);
}
std::pair<Handle<Object>*, int> handle_and_id(&h, 1234);
GlobalHandles::MakeWeak(h.location(), reinterpret_cast<void*>(&handle_and_id),
&TestWeakGlobalHandleCallback,
v8::WeakCallbackType::kParameter);
// Scanvenge does not recognize weak reference.
heap->CollectGarbage(NEW_SPACE);
CHECK(!WeakPointerCleared);
// Mark-compact treats weak reference properly.
heap->CollectGarbage(OLD_SPACE);
CHECK(WeakPointerCleared);
}
TEST(DoNotPromoteWhiteObjectsOnScavenge) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope scope(isolate);
Handle<Object> white = factory->NewStringFromStaticChars("white");
CHECK(Marking::IsWhite(ObjectMarking::MarkBitFrom(HeapObject::cast(*white))));
heap->CollectGarbage(NEW_SPACE);
CHECK(heap->InNewSpace(*white));
}
TEST(PromoteGreyOrBlackObjectsOnScavenge) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope scope(isolate);
Handle<Object> marked = factory->NewStringFromStaticChars("marked");
IncrementalMarking* marking = heap->incremental_marking();
marking->Stop();
heap->StartIncrementalMarking();
while (
Marking::IsWhite(ObjectMarking::MarkBitFrom(HeapObject::cast(*marked)))) {
marking->Step(MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD,
IncrementalMarking::FORCE_MARKING,
IncrementalMarking::DO_NOT_FORCE_COMPLETION);
}
heap->CollectGarbage(NEW_SPACE);
CHECK(!heap->InNewSpace(*marked));
}
TEST(BytecodeArray) {
static const uint8_t kRawBytes[] = {0xc3, 0x7e, 0xa5, 0x5a};
static const int kRawBytesSize = sizeof(kRawBytes);
static const int kFrameSize = 32;
static const int kParameterCount = 2;
i::FLAG_manual_evacuation_candidates_selection = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope scope(isolate);
heap::SimulateFullSpace(heap->old_space());
Handle<FixedArray> constant_pool = factory->NewFixedArray(5, TENURED);
for (int i = 0; i < 5; i++) {
Handle<Object> number = factory->NewHeapNumber(i);
constant_pool->set(i, *number);
}
// Allocate and initialize BytecodeArray
Handle<BytecodeArray> array = factory->NewBytecodeArray(
kRawBytesSize, kRawBytes, kFrameSize, kParameterCount, constant_pool);
CHECK(array->IsBytecodeArray());
CHECK_EQ(array->length(), (int)sizeof(kRawBytes));
CHECK_EQ(array->frame_size(), kFrameSize);
CHECK_EQ(array->parameter_count(), kParameterCount);
CHECK_EQ(array->constant_pool(), *constant_pool);
CHECK_LE(array->address(), array->GetFirstBytecodeAddress());
CHECK_GE(array->address() + array->BytecodeArraySize(),
array->GetFirstBytecodeAddress() + array->length());
for (int i = 0; i < kRawBytesSize; i++) {
CHECK_EQ(array->GetFirstBytecodeAddress()[i], kRawBytes[i]);
CHECK_EQ(array->get(i), kRawBytes[i]);
}
FixedArray* old_constant_pool_address = *constant_pool;
// Perform a full garbage collection and force the constant pool to be on an
// evacuation candidate.
Page* evac_page = Page::FromAddress(constant_pool->address());
evac_page->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
heap->CollectAllGarbage();
// BytecodeArray should survive.
CHECK_EQ(array->length(), kRawBytesSize);
CHECK_EQ(array->frame_size(), kFrameSize);
for (int i = 0; i < kRawBytesSize; i++) {
CHECK_EQ(array->get(i), kRawBytes[i]);
CHECK_EQ(array->GetFirstBytecodeAddress()[i], kRawBytes[i]);
}
// Constant pool should have been migrated.
CHECK_EQ(array->constant_pool(), *constant_pool);
CHECK_NE(array->constant_pool(), old_constant_pool_address);
}
static const char* not_so_random_string_table[] = {
"abstract",
"boolean",
"break",
"byte",
"case",
"catch",
"char",
"class",
"const",
"continue",
"debugger",
"default",
"delete",
"do",
"double",
"else",
"enum",
"export",
"extends",
"false",
"final",
"finally",
"float",
"for",
"function",
"goto",
"if",
"implements",
"import",
"in",
"instanceof",
"int",
"interface",
"long",
"native",
"new",
"null",
"package",
"private",
"protected",
"public",
"return",
"short",
"static",
"super",
"switch",
"synchronized",
"this",
"throw",
"throws",
"transient",
"true",
"try",
"typeof",
"var",
"void",
"volatile",
"while",
"with",
0
};
static void CheckInternalizedStrings(const char** strings) {
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
for (const char* string = *strings; *strings != 0; string = *strings++) {
HandleScope scope(isolate);
Handle<String> a =
isolate->factory()->InternalizeUtf8String(CStrVector(string));
// InternalizeUtf8String may return a failure if a GC is needed.
CHECK(a->IsInternalizedString());
Handle<String> b = factory->InternalizeUtf8String(string);
CHECK_EQ(*b, *a);
CHECK(b->IsUtf8EqualTo(CStrVector(string)));
b = isolate->factory()->InternalizeUtf8String(CStrVector(string));
CHECK_EQ(*b, *a);
CHECK(b->IsUtf8EqualTo(CStrVector(string)));
}
}
TEST(StringTable) {
CcTest::InitializeVM();
v8::HandleScope sc(CcTest::isolate());
CheckInternalizedStrings(not_so_random_string_table);
CheckInternalizedStrings(not_so_random_string_table);
}
TEST(FunctionAllocation) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope sc(CcTest::isolate());
Handle<String> name = factory->InternalizeUtf8String("theFunction");
Handle<JSFunction> function = factory->NewFunction(name);
Handle<Smi> twenty_three(Smi::FromInt(23), isolate);
Handle<Smi> twenty_four(Smi::FromInt(24), isolate);
Handle<String> prop_name = factory->InternalizeUtf8String("theSlot");
Handle<JSObject> obj = factory->NewJSObject(function);
JSReceiver::SetProperty(obj, prop_name, twenty_three, SLOPPY).Check();
CHECK_EQ(Smi::FromInt(23),
*Object::GetProperty(obj, prop_name).ToHandleChecked());
// Check that we can add properties to function objects.
JSReceiver::SetProperty(function, prop_name, twenty_four, SLOPPY).Check();
CHECK_EQ(Smi::FromInt(24),
*Object::GetProperty(function, prop_name).ToHandleChecked());
}
TEST(ObjectProperties) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope sc(CcTest::isolate());
Handle<String> object_string(String::cast(CcTest::heap()->Object_string()));
Handle<Object> object = Object::GetProperty(
CcTest::i_isolate()->global_object(), object_string).ToHandleChecked();
Handle<JSFunction> constructor = Handle<JSFunction>::cast(object);
Handle<JSObject> obj = factory->NewJSObject(constructor);
Handle<String> first = factory->InternalizeUtf8String("first");
Handle<String> second = factory->InternalizeUtf8String("second");
Handle<Smi> one(Smi::FromInt(1), isolate);
Handle<Smi> two(Smi::FromInt(2), isolate);
// check for empty
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, first));
// add first
JSReceiver::SetProperty(obj, first, one, SLOPPY).Check();
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, first));
// delete first
CHECK(Just(true) == JSReceiver::DeleteProperty(obj, first, SLOPPY));
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, first));
// add first and then second
JSReceiver::SetProperty(obj, first, one, SLOPPY).Check();
JSReceiver::SetProperty(obj, second, two, SLOPPY).Check();
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, first));
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, second));
// delete first and then second
CHECK(Just(true) == JSReceiver::DeleteProperty(obj, first, SLOPPY));
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, second));
CHECK(Just(true) == JSReceiver::DeleteProperty(obj, second, SLOPPY));
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, first));
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, second));
// add first and then second
JSReceiver::SetProperty(obj, first, one, SLOPPY).Check();
JSReceiver::SetProperty(obj, second, two, SLOPPY).Check();
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, first));
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, second));
// delete second and then first
CHECK(Just(true) == JSReceiver::DeleteProperty(obj, second, SLOPPY));
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, first));
CHECK(Just(true) == JSReceiver::DeleteProperty(obj, first, SLOPPY));
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, first));
CHECK(Just(false) == JSReceiver::HasOwnProperty(obj, second));
// check string and internalized string match
const char* string1 = "fisk";
Handle<String> s1 = factory->NewStringFromAsciiChecked(string1);
JSReceiver::SetProperty(obj, s1, one, SLOPPY).Check();
Handle<String> s1_string = factory->InternalizeUtf8String(string1);
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, s1_string));
// check internalized string and string match
const char* string2 = "fugl";
Handle<String> s2_string = factory->InternalizeUtf8String(string2);
JSReceiver::SetProperty(obj, s2_string, one, SLOPPY).Check();
Handle<String> s2 = factory->NewStringFromAsciiChecked(string2);
CHECK(Just(true) == JSReceiver::HasOwnProperty(obj, s2));
}
TEST(JSObjectMaps) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope sc(CcTest::isolate());
Handle<String> name = factory->InternalizeUtf8String("theFunction");
Handle<JSFunction> function = factory->NewFunction(name);
Handle<String> prop_name = factory->InternalizeUtf8String("theSlot");
Handle<JSObject> obj = factory->NewJSObject(function);
Handle<Map> initial_map(function->initial_map());
// Set a propery
Handle<Smi> twenty_three(Smi::FromInt(23), isolate);
JSReceiver::SetProperty(obj, prop_name, twenty_three, SLOPPY).Check();
CHECK_EQ(Smi::FromInt(23),
*Object::GetProperty(obj, prop_name).ToHandleChecked());
// Check the map has changed
CHECK(*initial_map != obj->map());
}
TEST(JSArray) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope sc(CcTest::isolate());
Handle<String> name = factory->InternalizeUtf8String("Array");
Handle<Object> fun_obj = Object::GetProperty(
CcTest::i_isolate()->global_object(), name).ToHandleChecked();
Handle<JSFunction> function = Handle<JSFunction>::cast(fun_obj);
// Allocate the object.
Handle<Object> element;
Handle<JSObject> object = factory->NewJSObject(function);
Handle<JSArray> array = Handle<JSArray>::cast(object);
// We just initialized the VM, no heap allocation failure yet.
JSArray::Initialize(array, 0);
// Set array length to 0.
JSArray::SetLength(array, 0);
CHECK_EQ(Smi::FromInt(0), array->length());
// Must be in fast mode.
CHECK(array->HasFastSmiOrObjectElements());
// array[length] = name.
JSReceiver::SetElement(isolate, array, 0, name, SLOPPY).Check();
CHECK_EQ(Smi::FromInt(1), array->length());
element = i::Object::GetElement(isolate, array, 0).ToHandleChecked();
CHECK_EQ(*element, *name);
// Set array length with larger than smi value.
JSArray::SetLength(array, static_cast<uint32_t>(Smi::kMaxValue) + 1);
uint32_t int_length = 0;
CHECK(array->length()->ToArrayIndex(&int_length));
CHECK_EQ(static_cast<uint32_t>(Smi::kMaxValue) + 1, int_length);
CHECK(array->HasDictionaryElements()); // Must be in slow mode.
// array[length] = name.
JSReceiver::SetElement(isolate, array, int_length, name, SLOPPY).Check();
uint32_t new_int_length = 0;
CHECK(array->length()->ToArrayIndex(&new_int_length));
CHECK_EQ(static_cast<double>(int_length), new_int_length - 1);
element = Object::GetElement(isolate, array, int_length).ToHandleChecked();
CHECK_EQ(*element, *name);
element = Object::GetElement(isolate, array, 0).ToHandleChecked();
CHECK_EQ(*element, *name);
}
TEST(JSObjectCopy) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope sc(CcTest::isolate());
Handle<String> object_string(String::cast(CcTest::heap()->Object_string()));
Handle<Object> object = Object::GetProperty(
CcTest::i_isolate()->global_object(), object_string).ToHandleChecked();
Handle<JSFunction> constructor = Handle<JSFunction>::cast(object);
Handle<JSObject> obj = factory->NewJSObject(constructor);
Handle<String> first = factory->InternalizeUtf8String("first");
Handle<String> second = factory->InternalizeUtf8String("second");
Handle<Smi> one(Smi::FromInt(1), isolate);
Handle<Smi> two(Smi::FromInt(2), isolate);
JSReceiver::SetProperty(obj, first, one, SLOPPY).Check();
JSReceiver::SetProperty(obj, second, two, SLOPPY).Check();
JSReceiver::SetElement(isolate, obj, 0, first, SLOPPY).Check();
JSReceiver::SetElement(isolate, obj, 1, second, SLOPPY).Check();
// Make the clone.
Handle<Object> value1, value2;
Handle<JSObject> clone = factory->CopyJSObject(obj);
CHECK(!clone.is_identical_to(obj));
value1 = Object::GetElement(isolate, obj, 0).ToHandleChecked();
value2 = Object::GetElement(isolate, clone, 0).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetElement(isolate, obj, 1).ToHandleChecked();
value2 = Object::GetElement(isolate, clone, 1).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetProperty(obj, first).ToHandleChecked();
value2 = Object::GetProperty(clone, first).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetProperty(obj, second).ToHandleChecked();
value2 = Object::GetProperty(clone, second).ToHandleChecked();
CHECK_EQ(*value1, *value2);
// Flip the values.
JSReceiver::SetProperty(clone, first, two, SLOPPY).Check();
JSReceiver::SetProperty(clone, second, one, SLOPPY).Check();
JSReceiver::SetElement(isolate, clone, 0, second, SLOPPY).Check();
JSReceiver::SetElement(isolate, clone, 1, first, SLOPPY).Check();
value1 = Object::GetElement(isolate, obj, 1).ToHandleChecked();
value2 = Object::GetElement(isolate, clone, 0).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetElement(isolate, obj, 0).ToHandleChecked();
value2 = Object::GetElement(isolate, clone, 1).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetProperty(obj, second).ToHandleChecked();
value2 = Object::GetProperty(clone, first).ToHandleChecked();
CHECK_EQ(*value1, *value2);
value1 = Object::GetProperty(obj, first).ToHandleChecked();
value2 = Object::GetProperty(clone, second).ToHandleChecked();
CHECK_EQ(*value1, *value2);
}
TEST(StringAllocation) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
const unsigned char chars[] = { 0xe5, 0xa4, 0xa7 };
for (int length = 0; length < 100; length++) {
v8::HandleScope scope(CcTest::isolate());
char* non_one_byte = NewArray<char>(3 * length + 1);
char* one_byte = NewArray<char>(length + 1);
non_one_byte[3 * length] = 0;
one_byte[length] = 0;
for (int i = 0; i < length; i++) {
one_byte[i] = 'a';
non_one_byte[3 * i] = chars[0];
non_one_byte[3 * i + 1] = chars[1];
non_one_byte[3 * i + 2] = chars[2];
}
Handle<String> non_one_byte_sym = factory->InternalizeUtf8String(
Vector<const char>(non_one_byte, 3 * length));
CHECK_EQ(length, non_one_byte_sym->length());
Handle<String> one_byte_sym =
factory->InternalizeOneByteString(OneByteVector(one_byte, length));
CHECK_EQ(length, one_byte_sym->length());
Handle<String> non_one_byte_str =
factory->NewStringFromUtf8(Vector<const char>(non_one_byte, 3 * length))
.ToHandleChecked();
non_one_byte_str->Hash();
CHECK_EQ(length, non_one_byte_str->length());
Handle<String> one_byte_str =
factory->NewStringFromUtf8(Vector<const char>(one_byte, length))
.ToHandleChecked();
one_byte_str->Hash();
CHECK_EQ(length, one_byte_str->length());
DeleteArray(non_one_byte);
DeleteArray(one_byte);
}
}
static int ObjectsFoundInHeap(Heap* heap, Handle<Object> objs[], int size) {
// Count the number of objects found in the heap.
int found_count = 0;
HeapIterator iterator(heap);
for (HeapObject* obj = iterator.next(); obj != NULL; obj = iterator.next()) {
for (int i = 0; i < size; i++) {
if (*objs[i] == obj) {
found_count++;
}
}
}
return found_count;
}
TEST(Iteration) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
// Array of objects to scan haep for.
const int objs_count = 6;
Handle<Object> objs[objs_count];
int next_objs_index = 0;
// Allocate a JS array to OLD_SPACE and NEW_SPACE
objs[next_objs_index++] = factory->NewJSArray(10);
objs[next_objs_index++] =
factory->NewJSArray(10, FAST_HOLEY_ELEMENTS, TENURED);
// Allocate a small string to OLD_DATA_SPACE and NEW_SPACE
objs[next_objs_index++] = factory->NewStringFromStaticChars("abcdefghij");
objs[next_objs_index++] =
factory->NewStringFromStaticChars("abcdefghij", TENURED);
// Allocate a large string (for large object space).
int large_size = Page::kMaxRegularHeapObjectSize + 1;
char* str = new char[large_size];
for (int i = 0; i < large_size - 1; ++i) str[i] = 'a';
str[large_size - 1] = '\0';
objs[next_objs_index++] = factory->NewStringFromAsciiChecked(str, TENURED);
delete[] str;
// Add a Map object to look for.
objs[next_objs_index++] = Handle<Map>(HeapObject::cast(*objs[0])->map());
CHECK_EQ(objs_count, next_objs_index);
CHECK_EQ(objs_count, ObjectsFoundInHeap(CcTest::heap(), objs, objs_count));
}
UNINITIALIZED_TEST(TestCodeFlushing) {
// If we do not flush code this test is invalid.
if (!FLAG_flush_code) return;
i::FLAG_allow_natives_syntax = true;
i::FLAG_optimize_for_size = false;
v8::Isolate::CreateParams create_params;
create_params.array_buffer_allocator = CcTest::array_buffer_allocator();
v8::Isolate* isolate = v8::Isolate::New(create_params);
i::Isolate* i_isolate = reinterpret_cast<i::Isolate*>(isolate);
isolate->Enter();
Factory* factory = i_isolate->factory();
{
v8::HandleScope scope(isolate);
v8::Context::New(isolate)->Enter();
const char* source =
"function foo() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo()";
Handle<String> foo_name = factory->InternalizeUtf8String("foo");
// This compile will add the code to the compilation cache.
{
v8::HandleScope scope(isolate);
CompileRun(source);
}
// Check function is compiled.
Handle<Object> func_value = Object::GetProperty(i_isolate->global_object(),
foo_name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
CHECK(function->shared()->is_compiled());
// The code will survive at least two GCs.
i_isolate->heap()->CollectAllGarbage();
i_isolate->heap()->CollectAllGarbage();
CHECK(function->shared()->is_compiled());
// Simulate several GCs that use full marking.
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
i_isolate->heap()->CollectAllGarbage();
}
// foo should no longer be in the compilation cache
CHECK(!function->shared()->is_compiled() || function->IsOptimized());
CHECK(!function->is_compiled() || function->IsOptimized());
// Call foo to get it recompiled.
CompileRun("foo()");
CHECK(function->shared()->is_compiled());
CHECK(function->is_compiled());
}
isolate->Exit();
isolate->Dispose();
}
TEST(TestCodeFlushingPreAged) {
// If we do not flush code this test is invalid.
if (!FLAG_flush_code) return;
i::FLAG_allow_natives_syntax = true;
i::FLAG_optimize_for_size = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
const char* source = "function foo() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo()";
Handle<String> foo_name = factory->InternalizeUtf8String("foo");
// Compile foo, but don't run it.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun(source);
}
// Check function is compiled.
Handle<Object> func_value =
Object::GetProperty(isolate->global_object(), foo_name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
CHECK(function->shared()->is_compiled());
// The code has been run so will survive at least one GC.
CcTest::heap()->CollectAllGarbage();
CHECK(function->shared()->is_compiled());
// The code was only run once, so it should be pre-aged and collected on the
// next GC.
CcTest::heap()->CollectAllGarbage();
CHECK(!function->shared()->is_compiled() || function->IsOptimized());
// Execute the function again twice, and ensure it is reset to the young age.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun("foo();"
"foo();");
}
// The code will survive at least two GC now that it is young again.
CcTest::heap()->CollectAllGarbage();
CcTest::heap()->CollectAllGarbage();
CHECK(function->shared()->is_compiled());
// Simulate several GCs that use full marking.
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
CcTest::heap()->CollectAllGarbage();
}
// foo should no longer be in the compilation cache
CHECK(!function->shared()->is_compiled() || function->IsOptimized());
CHECK(!function->is_compiled() || function->IsOptimized());
// Call foo to get it recompiled.
CompileRun("foo()");
CHECK(function->shared()->is_compiled());
CHECK(function->is_compiled());
}
TEST(TestCodeFlushingIncremental) {
// If we do not flush code this test is invalid.
if (!FLAG_flush_code) return;
i::FLAG_allow_natives_syntax = true;
i::FLAG_optimize_for_size = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
const char* source = "function foo() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo()";
Handle<String> foo_name = factory->InternalizeUtf8String("foo");
// This compile will add the code to the compilation cache.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun(source);
}
// Check function is compiled.
Handle<Object> func_value =
Object::GetProperty(isolate->global_object(), foo_name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
CHECK(function->shared()->is_compiled());
// The code will survive at least two GCs.
CcTest::heap()->CollectAllGarbage();
CcTest::heap()->CollectAllGarbage();
CHECK(function->shared()->is_compiled());
// Simulate several GCs that use incremental marking.
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
}
CHECK(!function->shared()->is_compiled() || function->IsOptimized());
CHECK(!function->is_compiled() || function->IsOptimized());
// This compile will compile the function again.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun("foo();");
}
// Simulate several GCs that use incremental marking but make sure
// the loop breaks once the function is enqueued as a candidate.
for (int i = 0; i < kAgingThreshold; i++) {
heap::SimulateIncrementalMarking(CcTest::heap());
if (!function->next_function_link()->IsUndefined(CcTest::i_isolate()))
break;
CcTest::heap()->CollectAllGarbage();
}
// Force optimization while incremental marking is active and while
// the function is enqueued as a candidate.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun("%OptimizeFunctionOnNextCall(foo); foo();");
}
// Simulate one final GC to make sure the candidate queue is sane.
CcTest::heap()->CollectAllGarbage();
CHECK(function->shared()->is_compiled() || !function->IsOptimized());
CHECK(function->is_compiled() || !function->IsOptimized());
}
TEST(TestCodeFlushingIncrementalScavenge) {
// If we do not flush code this test is invalid.
if (!FLAG_flush_code) return;
i::FLAG_allow_natives_syntax = true;
i::FLAG_optimize_for_size = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
const char* source = "var foo = function() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo();"
"var bar = function() {"
" var x = 23;"
"};"
"bar();";
Handle<String> foo_name = factory->InternalizeUtf8String("foo");
Handle<String> bar_name = factory->InternalizeUtf8String("bar");
// Perfrom one initial GC to enable code flushing.
CcTest::heap()->CollectAllGarbage();
// This compile will add the code to the compilation cache.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun(source);
}
// Check functions are compiled.
Handle<Object> func_value =
Object::GetProperty(isolate->global_object(), foo_name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
CHECK(function->shared()->is_compiled());
Handle<Object> func_value2 =
Object::GetProperty(isolate->global_object(), bar_name).ToHandleChecked();
CHECK(func_value2->IsJSFunction());
Handle<JSFunction> function2 = Handle<JSFunction>::cast(func_value2);
CHECK(function2->shared()->is_compiled());
// Clear references to functions so that one of them can die.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun("foo = 0; bar = 0;");
}
// Bump the code age so that flushing is triggered while the function
// object is still located in new-space.
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
function->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
function2->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
// Simulate incremental marking so that the functions are enqueued as
// code flushing candidates. Then kill one of the functions. Finally
// perform a scavenge while incremental marking is still running.
heap::SimulateIncrementalMarking(CcTest::heap(), false);
*function2.location() = NULL;
CcTest::heap()->CollectGarbage(NEW_SPACE, "test scavenge while marking");
// Simulate one final GC to make sure the candidate queue is sane.
CcTest::heap()->CollectAllGarbage();
CHECK(!function->shared()->is_compiled() || function->IsOptimized());
CHECK(!function->is_compiled() || function->IsOptimized());
}
TEST(TestCodeFlushingIncrementalAbort) {
// If we do not flush code this test is invalid.
if (!FLAG_flush_code) return;
i::FLAG_allow_natives_syntax = true;
i::FLAG_optimize_for_size = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
v8::HandleScope scope(CcTest::isolate());
const char* source = "function foo() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo()";
Handle<String> foo_name = factory->InternalizeUtf8String("foo");
// This compile will add the code to the compilation cache.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun(source);
}
// Check function is compiled.
Handle<Object> func_value =
Object::GetProperty(isolate->global_object(), foo_name).ToHandleChecked();
CHECK(func_value->IsJSFunction());
Handle<JSFunction> function = Handle<JSFunction>::cast(func_value);
CHECK(function->shared()->is_compiled());
// The code will survive at least two GCs.
heap->CollectAllGarbage();
heap->CollectAllGarbage();
CHECK(function->shared()->is_compiled());
// Bump the code age so that flushing is triggered.
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
function->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
// Simulate incremental marking so that the function is enqueued as
// code flushing candidate.
heap::SimulateIncrementalMarking(heap);
// Enable the debugger and add a breakpoint while incremental marking
// is running so that incremental marking aborts and code flushing is
// disabled.
int position = function->shared()->start_position();
Handle<Object> breakpoint_object(Smi::FromInt(0), isolate);
EnableDebugger(CcTest::isolate());
isolate->debug()->SetBreakPoint(function, breakpoint_object, &position);
isolate->debug()->ClearBreakPoint(breakpoint_object);
DisableDebugger(CcTest::isolate());
// Force optimization now that code flushing is disabled.
{ v8::HandleScope scope(CcTest::isolate());
CompileRun("%OptimizeFunctionOnNextCall(foo); foo();");
}
// Simulate one final GC to make sure the candidate queue is sane.
heap->CollectAllGarbage();
CHECK(function->shared()->is_compiled() || !function->IsOptimized());
CHECK(function->is_compiled() || !function->IsOptimized());
}
TEST(TestUseOfIncrementalBarrierOnCompileLazy) {
// Turn off always_opt because it interferes with running the built-in for
// the last call to g().
i::FLAG_always_opt = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function make_closure(x) {"
" return function() { return x + 3 };"
"}"
"var f = make_closure(5); f();"
"var g = make_closure(5);");
// Check f is compiled.
Handle<String> f_name = factory->InternalizeUtf8String("f");
Handle<Object> f_value =
Object::GetProperty(isolate->global_object(), f_name).ToHandleChecked();
Handle<JSFunction> f_function = Handle<JSFunction>::cast(f_value);
CHECK(f_function->is_compiled());
// Check g is not compiled.
Handle<String> g_name = factory->InternalizeUtf8String("g");
Handle<Object> g_value =
Object::GetProperty(isolate->global_object(), g_name).ToHandleChecked();
Handle<JSFunction> g_function = Handle<JSFunction>::cast(g_value);
CHECK(!g_function->is_compiled());
heap::SimulateIncrementalMarking(heap);
CompileRun("%OptimizeFunctionOnNextCall(f); f();");
// g should now have available an optimized function, unmarked by gc. The
// CompileLazy built-in will discover it and install it in the closure, and
// the incremental write barrier should be used.
CompileRun("g();");
CHECK(g_function->is_compiled());
}
TEST(CompilationCacheCachingBehavior) {
// If we do not flush code, or have the compilation cache turned off, this
// test is invalid.
if (!FLAG_flush_code || !FLAG_compilation_cache) {
return;
}
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
CompilationCache* compilation_cache = isolate->compilation_cache();
LanguageMode language_mode = construct_language_mode(FLAG_use_strict);
v8::HandleScope scope(CcTest::isolate());
const char* raw_source =
"function foo() {"
" var x = 42;"
" var y = 42;"
" var z = x + y;"
"};"
"foo()";
Handle<String> source = factory->InternalizeUtf8String(raw_source);
Handle<Context> native_context = isolate->native_context();
{
v8::HandleScope scope(CcTest::isolate());
CompileRun(raw_source);
}
// The script should be in the cache now.
MaybeHandle<SharedFunctionInfo> info = compilation_cache->LookupScript(
source, Handle<Object>(), 0, 0,
v8::ScriptOriginOptions(false, true, false), native_context,
language_mode);
CHECK(!info.is_null());
// Check that the code cache entry survives at least on GC.
// (Unless --optimize-for-size, in which case it might get collected
// immediately.)
if (!FLAG_optimize_for_size) {
heap->CollectAllGarbage();
info = compilation_cache->LookupScript(
source, Handle<Object>(), 0, 0,
v8::ScriptOriginOptions(false, true, false), native_context,
language_mode);
CHECK(!info.is_null());
}
// Progress code age until it's old and ready for GC.
while (!info.ToHandleChecked()->code()->IsOld()) {
// To guarantee progress, we have to MakeOlder with different parities.
// We can't just use NO_MARKING_PARITY, since e.g. kExecutedOnceCodeAge is
// always NO_MARKING_PARITY and the code age only progresses if the parity
// is different.
info.ToHandleChecked()->code()->MakeOlder(ODD_MARKING_PARITY);
info.ToHandleChecked()->code()->MakeOlder(EVEN_MARKING_PARITY);
}
heap->CollectAllGarbage();
// Ensure code aging cleared the entry from the cache.
info = compilation_cache->LookupScript(
source, Handle<Object>(), 0, 0,
v8::ScriptOriginOptions(false, true, false), native_context,
language_mode);
CHECK(info.is_null());
}
static void OptimizeEmptyFunction(const char* name) {
HandleScope scope(CcTest::i_isolate());
EmbeddedVector<char, 256> source;
SNPrintF(source,
"function %s() { return 0; }"
"%s(); %s();"
"%%OptimizeFunctionOnNextCall(%s);"
"%s();",
name, name, name, name, name);
CompileRun(source.start());
}
// Count the number of native contexts in the weak list of native contexts.
int CountNativeContexts() {
int count = 0;
Object* object = CcTest::heap()->native_contexts_list();
while (!object->IsUndefined(CcTest::i_isolate())) {
count++;
object = Context::cast(object)->next_context_link();
}
return count;
}
// Count the number of user functions in the weak list of optimized
// functions attached to a native context.
static int CountOptimizedUserFunctions(v8::Local<v8::Context> context) {
int count = 0;
Handle<Context> icontext = v8::Utils::OpenHandle(*context);
Object* object = icontext->get(Context::OPTIMIZED_FUNCTIONS_LIST);
while (object->IsJSFunction() &&
!JSFunction::cast(object)->shared()->IsBuiltin()) {
count++;
object = JSFunction::cast(object)->next_function_link();
}
return count;
}
TEST(TestInternalWeakLists) {
FLAG_always_opt = false;
FLAG_allow_natives_syntax = true;
v8::V8::Initialize();
// Some flags turn Scavenge collections into Mark-sweep collections
// and hence are incompatible with this test case.
if (FLAG_gc_global || FLAG_stress_compaction) return;
FLAG_retain_maps_for_n_gc = 0;
static const int kNumTestContexts = 10;
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
v8::Local<v8::Context> ctx[kNumTestContexts];
if (!isolate->use_crankshaft()) return;
CHECK_EQ(0, CountNativeContexts());
// Create a number of global contests which gets linked together.
for (int i = 0; i < kNumTestContexts; i++) {
ctx[i] = v8::Context::New(CcTest::isolate());
// Collect garbage that might have been created by one of the
// installed extensions.
isolate->compilation_cache()->Clear();
heap->CollectAllGarbage();
CHECK_EQ(i + 1, CountNativeContexts());
ctx[i]->Enter();
// Create a handle scope so no function objects get stuck in the outer
// handle scope.
HandleScope scope(isolate);
CHECK_EQ(0, CountOptimizedUserFunctions(ctx[i]));
OptimizeEmptyFunction("f1");
CHECK_EQ(1, CountOptimizedUserFunctions(ctx[i]));
OptimizeEmptyFunction("f2");
CHECK_EQ(2, CountOptimizedUserFunctions(ctx[i]));
OptimizeEmptyFunction("f3");
CHECK_EQ(3, CountOptimizedUserFunctions(ctx[i]));
OptimizeEmptyFunction("f4");
CHECK_EQ(4, CountOptimizedUserFunctions(ctx[i]));
OptimizeEmptyFunction("f5");
CHECK_EQ(5, CountOptimizedUserFunctions(ctx[i]));
// Remove function f1, and
CompileRun("f1=null");
// Scavenge treats these references as strong.
for (int j = 0; j < 10; j++) {
CcTest::heap()->CollectGarbage(NEW_SPACE);
CHECK_EQ(5, CountOptimizedUserFunctions(ctx[i]));
}
// Mark compact handles the weak references.
isolate->compilation_cache()->Clear();
heap->CollectAllGarbage();
CHECK_EQ(4, CountOptimizedUserFunctions(ctx[i]));
// Get rid of f3 and f5 in the same way.
CompileRun("f3=null");
for (int j = 0; j < 10; j++) {
CcTest::heap()->CollectGarbage(NEW_SPACE);
CHECK_EQ(4, CountOptimizedUserFunctions(ctx[i]));
}
CcTest::heap()->CollectAllGarbage();
CHECK_EQ(3, CountOptimizedUserFunctions(ctx[i]));
CompileRun("f5=null");
for (int j = 0; j < 10; j++) {
CcTest::heap()->CollectGarbage(NEW_SPACE);
CHECK_EQ(3, CountOptimizedUserFunctions(ctx[i]));
}
CcTest::heap()->CollectAllGarbage();
CHECK_EQ(2, CountOptimizedUserFunctions(ctx[i]));
ctx[i]->Exit();
}
// Force compilation cache cleanup.
CcTest::heap()->NotifyContextDisposed(true);
CcTest::heap()->CollectAllGarbage();
// Dispose the native contexts one by one.
for (int i = 0; i < kNumTestContexts; i++) {
// TODO(dcarney): is there a better way to do this?
i::Object** unsafe = reinterpret_cast<i::Object**>(*ctx[i]);
*unsafe = CcTest::heap()->undefined_value();
ctx[i].Clear();
// Scavenge treats these references as strong.
for (int j = 0; j < 10; j++) {
CcTest::heap()->CollectGarbage(i::NEW_SPACE);
CHECK_EQ(kNumTestContexts - i, CountNativeContexts());
}
// Mark compact handles the weak references.
CcTest::heap()->CollectAllGarbage();
CHECK_EQ(kNumTestContexts - i - 1, CountNativeContexts());
}
CHECK_EQ(0, CountNativeContexts());
}
// Count the number of native contexts in the weak list of native contexts
// causing a GC after the specified number of elements.
static int CountNativeContextsWithGC(Isolate* isolate, int n) {
Heap* heap = isolate->heap();
int count = 0;
Handle<Object> object(heap->native_contexts_list(), isolate);
while (!object->IsUndefined(isolate)) {
count++;
if (count == n) heap->CollectAllGarbage();
object =
Handle<Object>(Context::cast(*object)->next_context_link(), isolate);
}
return count;
}
// Count the number of user functions in the weak list of optimized
// functions attached to a native context causing a GC after the
// specified number of elements.
static int CountOptimizedUserFunctionsWithGC(v8::Local<v8::Context> context,
int n) {
int count = 0;
Handle<Context> icontext = v8::Utils::OpenHandle(*context);
Isolate* isolate = icontext->GetIsolate();
Handle<Object> object(icontext->get(Context::OPTIMIZED_FUNCTIONS_LIST),
isolate);
while (object->IsJSFunction() &&
!Handle<JSFunction>::cast(object)->shared()->IsBuiltin()) {
count++;
if (count == n) isolate->heap()->CollectAllGarbage();
object = Handle<Object>(
Object::cast(JSFunction::cast(*object)->next_function_link()),
isolate);
}
return count;
}
TEST(TestInternalWeakListsTraverseWithGC) {
FLAG_always_opt = false;
FLAG_allow_natives_syntax = true;
v8::V8::Initialize();
static const int kNumTestContexts = 10;
Isolate* isolate = CcTest::i_isolate();
HandleScope scope(isolate);
v8::Local<v8::Context> ctx[kNumTestContexts];
if (!isolate->use_crankshaft()) return;
CHECK_EQ(0, CountNativeContexts());
// Create an number of contexts and check the length of the weak list both
// with and without GCs while iterating the list.
for (int i = 0; i < kNumTestContexts; i++) {
ctx[i] = v8::Context::New(CcTest::isolate());
CHECK_EQ(i + 1, CountNativeContexts());
CHECK_EQ(i + 1, CountNativeContextsWithGC(isolate, i / 2 + 1));
}
ctx[0]->Enter();
// Compile a number of functions the length of the weak list of optimized
// functions both with and without GCs while iterating the list.
CHECK_EQ(0, CountOptimizedUserFunctions(ctx[0]));
OptimizeEmptyFunction("f1");
CHECK_EQ(1, CountOptimizedUserFunctions(ctx[0]));
CHECK_EQ(1, CountOptimizedUserFunctionsWithGC(ctx[0], 1));
OptimizeEmptyFunction("f2");
CHECK_EQ(2, CountOptimizedUserFunctions(ctx[0]));
CHECK_EQ(2, CountOptimizedUserFunctionsWithGC(ctx[0], 1));
OptimizeEmptyFunction("f3");
CHECK_EQ(3, CountOptimizedUserFunctions(ctx[0]));
CHECK_EQ(3, CountOptimizedUserFunctionsWithGC(ctx[0], 1));
OptimizeEmptyFunction("f4");
CHECK_EQ(4, CountOptimizedUserFunctions(ctx[0]));
CHECK_EQ(4, CountOptimizedUserFunctionsWithGC(ctx[0], 2));
OptimizeEmptyFunction("f5");
CHECK_EQ(5, CountOptimizedUserFunctions(ctx[0]));
CHECK_EQ(5, CountOptimizedUserFunctionsWithGC(ctx[0], 4));
ctx[0]->Exit();
}
TEST(TestSizeOfRegExpCode) {
if (!FLAG_regexp_optimization) return;
v8::V8::Initialize();
Isolate* isolate = CcTest::i_isolate();
HandleScope scope(isolate);
LocalContext context;
// Adjust source below and this check to match
// RegExpImple::kRegExpTooLargeToOptimize.
CHECK_EQ(i::RegExpImpl::kRegExpTooLargeToOptimize, 20 * KB);
// Compile a regexp that is much larger if we are using regexp optimizations.
CompileRun(
"var reg_exp_source = '(?:a|bc|def|ghij|klmno|pqrstu)';"
"var half_size_reg_exp;"
"while (reg_exp_source.length < 20 * 1024) {"
" half_size_reg_exp = reg_exp_source;"
" reg_exp_source = reg_exp_source + reg_exp_source;"
"}"
// Flatten string.
"reg_exp_source.match(/f/);");
// Get initial heap size after several full GCs, which will stabilize
// the heap size and return with sweeping finished completely.
CcTest::heap()->CollectAllAvailableGarbage("initial cleanup");
MarkCompactCollector* collector = CcTest::heap()->mark_compact_collector();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
int initial_size = static_cast<int>(CcTest::heap()->SizeOfObjects());
CompileRun("'foo'.match(reg_exp_source);");
CcTest::heap()->CollectAllGarbage();
int size_with_regexp = static_cast<int>(CcTest::heap()->SizeOfObjects());
CompileRun("'foo'.match(half_size_reg_exp);");
CcTest::heap()->CollectAllGarbage();
int size_with_optimized_regexp =
static_cast<int>(CcTest::heap()->SizeOfObjects());
int size_of_regexp_code = size_with_regexp - initial_size;
// On some platforms the debug-code flag causes huge amounts of regexp code
// to be emitted, breaking this test.
if (!FLAG_debug_code) {
CHECK_LE(size_of_regexp_code, 1 * MB);
}
// Small regexp is half the size, but compiles to more than twice the code
// due to the optimization steps.
CHECK_GE(size_with_optimized_regexp,
size_with_regexp + size_of_regexp_code * 2);
}
HEAP_TEST(TestSizeOfObjects) {
v8::V8::Initialize();
Heap* heap = CcTest::heap();
MarkCompactCollector* collector = heap->mark_compact_collector();
// Get initial heap size after several full GCs, which will stabilize
// the heap size and return with sweeping finished completely.
heap->CollectAllAvailableGarbage("initial cleanup");
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
int initial_size = static_cast<int>(heap->SizeOfObjects());
{
// Allocate objects on several different old-space pages so that
// concurrent sweeper threads will be busy sweeping the old space on
// subsequent GC runs.
AlwaysAllocateScope always_allocate(CcTest::i_isolate());
int filler_size = static_cast<int>(FixedArray::SizeFor(8192));
for (int i = 1; i <= 100; i++) {
heap->AllocateFixedArray(8192, TENURED).ToObjectChecked();
CHECK_EQ(initial_size + i * filler_size,
static_cast<int>(heap->SizeOfObjects()));
}
}
// The heap size should go back to initial size after a full GC, even
// though sweeping didn't finish yet.
heap->CollectAllGarbage();
// Normally sweeping would not be complete here, but no guarantees.
CHECK_EQ(initial_size, static_cast<int>(heap->SizeOfObjects()));
// Waiting for sweeper threads should not change heap size.
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK_EQ(initial_size, static_cast<int>(heap->SizeOfObjects()));
}
TEST(TestAlignmentCalculations) {
// Maximum fill amounts are consistent.
int maximum_double_misalignment = kDoubleSize - kPointerSize;
int maximum_simd128_misalignment = kSimd128Size - kPointerSize;
int max_word_fill = Heap::GetMaximumFillToAlign(kWordAligned);
CHECK_EQ(0, max_word_fill);
int max_double_fill = Heap::GetMaximumFillToAlign(kDoubleAligned);
CHECK_EQ(maximum_double_misalignment, max_double_fill);
int max_double_unaligned_fill = Heap::GetMaximumFillToAlign(kDoubleUnaligned);
CHECK_EQ(maximum_double_misalignment, max_double_unaligned_fill);
int max_simd128_unaligned_fill =
Heap::GetMaximumFillToAlign(kSimd128Unaligned);
CHECK_EQ(maximum_simd128_misalignment, max_simd128_unaligned_fill);
Address base = static_cast<Address>(NULL);
int fill = 0;
// Word alignment never requires fill.
fill = Heap::GetFillToAlign(base, kWordAligned);
CHECK_EQ(0, fill);
fill = Heap::GetFillToAlign(base + kPointerSize, kWordAligned);
CHECK_EQ(0, fill);
// No fill is required when address is double aligned.
fill = Heap::GetFillToAlign(base, kDoubleAligned);
CHECK_EQ(0, fill);
// Fill is required if address is not double aligned.
fill = Heap::GetFillToAlign(base + kPointerSize, kDoubleAligned);
CHECK_EQ(maximum_double_misalignment, fill);
// kDoubleUnaligned has the opposite fill amounts.
fill = Heap::GetFillToAlign(base, kDoubleUnaligned);
CHECK_EQ(maximum_double_misalignment, fill);
fill = Heap::GetFillToAlign(base + kPointerSize, kDoubleUnaligned);
CHECK_EQ(0, fill);
// 128 bit SIMD types have 2 or 4 possible alignments, depending on platform.
fill = Heap::GetFillToAlign(base, kSimd128Unaligned);
CHECK_EQ((3 * kPointerSize) & kSimd128AlignmentMask, fill);
fill = Heap::GetFillToAlign(base + kPointerSize, kSimd128Unaligned);
CHECK_EQ((2 * kPointerSize) & kSimd128AlignmentMask, fill);
fill = Heap::GetFillToAlign(base + 2 * kPointerSize, kSimd128Unaligned);
CHECK_EQ(kPointerSize, fill);
fill = Heap::GetFillToAlign(base + 3 * kPointerSize, kSimd128Unaligned);
CHECK_EQ(0, fill);
}
static HeapObject* NewSpaceAllocateAligned(int size,
AllocationAlignment alignment) {
Heap* heap = CcTest::heap();
AllocationResult allocation =
heap->new_space()->AllocateRawAligned(size, alignment);
HeapObject* obj = NULL;
allocation.To(&obj);
heap->CreateFillerObjectAt(obj->address(), size, ClearRecordedSlots::kNo);
return obj;
}
// Get new space allocation into the desired alignment.
static Address AlignNewSpace(AllocationAlignment alignment, int offset) {
Address* top_addr = CcTest::heap()->new_space()->allocation_top_address();
int fill = Heap::GetFillToAlign(*top_addr, alignment);
if (fill) {
NewSpaceAllocateAligned(fill + offset, kWordAligned);
}
return *top_addr;
}
TEST(TestAlignedAllocation) {
// Double misalignment is 4 on 32-bit platforms, 0 on 64-bit ones.
const intptr_t double_misalignment = kDoubleSize - kPointerSize;
Address* top_addr = CcTest::heap()->new_space()->allocation_top_address();
Address start;
HeapObject* obj;
HeapObject* filler;
if (double_misalignment) {
// Allocate a pointer sized object that must be double aligned at an
// aligned address.
start = AlignNewSpace(kDoubleAligned, 0);
obj = NewSpaceAllocateAligned(kPointerSize, kDoubleAligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment));
// There is no filler.
CHECK_EQ(kPointerSize, *top_addr - start);
// Allocate a second pointer sized object that must be double aligned at an
// unaligned address.
start = AlignNewSpace(kDoubleAligned, kPointerSize);
obj = NewSpaceAllocateAligned(kPointerSize, kDoubleAligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment));
// There is a filler object before the object.
filler = HeapObject::FromAddress(start);
CHECK(obj != filler && filler->IsFiller() &&
filler->Size() == kPointerSize);
CHECK_EQ(kPointerSize + double_misalignment, *top_addr - start);
// Similarly for kDoubleUnaligned.
start = AlignNewSpace(kDoubleUnaligned, 0);
obj = NewSpaceAllocateAligned(kPointerSize, kDoubleUnaligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment, kPointerSize));
CHECK_EQ(kPointerSize, *top_addr - start);
start = AlignNewSpace(kDoubleUnaligned, kPointerSize);
obj = NewSpaceAllocateAligned(kPointerSize, kDoubleUnaligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment, kPointerSize));
// There is a filler object before the object.
filler = HeapObject::FromAddress(start);
CHECK(obj != filler && filler->IsFiller() &&
filler->Size() == kPointerSize);
CHECK_EQ(kPointerSize + double_misalignment, *top_addr - start);
}
// Now test SIMD alignment. There are 2 or 4 possible alignments, depending
// on platform.
start = AlignNewSpace(kSimd128Unaligned, 0);
obj = NewSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is no filler.
CHECK_EQ(kPointerSize, *top_addr - start);
start = AlignNewSpace(kSimd128Unaligned, kPointerSize);
obj = NewSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is a filler object before the object.
filler = HeapObject::FromAddress(start);
CHECK(obj != filler && filler->IsFiller() &&
filler->Size() == kSimd128Size - kPointerSize);
CHECK_EQ(kPointerSize + kSimd128Size - kPointerSize, *top_addr - start);
if (double_misalignment) {
// Test the 2 other alignments possible on 32 bit platforms.
start = AlignNewSpace(kSimd128Unaligned, 2 * kPointerSize);
obj = NewSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is a filler object before the object.
filler = HeapObject::FromAddress(start);
CHECK(obj != filler && filler->IsFiller() &&
filler->Size() == 2 * kPointerSize);
CHECK_EQ(kPointerSize + 2 * kPointerSize, *top_addr - start);
start = AlignNewSpace(kSimd128Unaligned, 3 * kPointerSize);
obj = NewSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is a filler object before the object.
filler = HeapObject::FromAddress(start);
CHECK(obj != filler && filler->IsFiller() &&
filler->Size() == kPointerSize);
CHECK_EQ(kPointerSize + kPointerSize, *top_addr - start);
}
}
static HeapObject* OldSpaceAllocateAligned(int size,
AllocationAlignment alignment) {
Heap* heap = CcTest::heap();
AllocationResult allocation =
heap->old_space()->AllocateRawAligned(size, alignment);
HeapObject* obj = NULL;
allocation.To(&obj);
heap->CreateFillerObjectAt(obj->address(), size, ClearRecordedSlots::kNo);
return obj;
}
// Get old space allocation into the desired alignment.
static Address AlignOldSpace(AllocationAlignment alignment, int offset) {
Address* top_addr = CcTest::heap()->old_space()->allocation_top_address();
int fill = Heap::GetFillToAlign(*top_addr, alignment);
int allocation = fill + offset;
if (allocation) {
OldSpaceAllocateAligned(allocation, kWordAligned);
}
Address top = *top_addr;
// Now force the remaining allocation onto the free list.
CcTest::heap()->old_space()->EmptyAllocationInfo();
return top;
}
// Test the case where allocation must be done from the free list, so filler
// may precede or follow the object.
TEST(TestAlignedOverAllocation) {
Heap* heap = CcTest::heap();
// Test checks for fillers before and behind objects and requires a fresh
// page and empty free list.
heap::AbandonCurrentlyFreeMemory(heap->old_space());
// Allocate a dummy object to properly set up the linear allocation info.
AllocationResult dummy =
heap->old_space()->AllocateRawUnaligned(kPointerSize);
CHECK(!dummy.IsRetry());
heap->CreateFillerObjectAt(
HeapObject::cast(dummy.ToObjectChecked())->address(), kPointerSize,
ClearRecordedSlots::kNo);
// Double misalignment is 4 on 32-bit platforms, 0 on 64-bit ones.
const intptr_t double_misalignment = kDoubleSize - kPointerSize;
Address start;
HeapObject* obj;
HeapObject* filler1;
HeapObject* filler2;
if (double_misalignment) {
start = AlignOldSpace(kDoubleAligned, 0);
obj = OldSpaceAllocateAligned(kPointerSize, kDoubleAligned);
// The object is aligned, and a filler object is created after.
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment));
filler1 = HeapObject::FromAddress(start + kPointerSize);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kPointerSize);
// Try the opposite alignment case.
start = AlignOldSpace(kDoubleAligned, kPointerSize);
obj = OldSpaceAllocateAligned(kPointerSize, kDoubleAligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment));
filler1 = HeapObject::FromAddress(start);
CHECK(obj != filler1);
CHECK(filler1->IsFiller());
CHECK(filler1->Size() == kPointerSize);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kPointerSize);
// Similarly for kDoubleUnaligned.
start = AlignOldSpace(kDoubleUnaligned, 0);
obj = OldSpaceAllocateAligned(kPointerSize, kDoubleUnaligned);
// The object is aligned, and a filler object is created after.
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment, kPointerSize));
filler1 = HeapObject::FromAddress(start + kPointerSize);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kPointerSize);
// Try the opposite alignment case.
start = AlignOldSpace(kDoubleUnaligned, kPointerSize);
obj = OldSpaceAllocateAligned(kPointerSize, kDoubleUnaligned);
CHECK(IsAddressAligned(obj->address(), kDoubleAlignment, kPointerSize));
filler1 = HeapObject::FromAddress(start);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kPointerSize);
}
// Now test SIMD alignment. There are 2 or 4 possible alignments, depending
// on platform.
start = AlignOldSpace(kSimd128Unaligned, 0);
obj = OldSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is a filler object after the object.
filler1 = HeapObject::FromAddress(start + kPointerSize);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kSimd128Size - kPointerSize);
start = AlignOldSpace(kSimd128Unaligned, kPointerSize);
obj = OldSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There is a filler object before the object.
filler1 = HeapObject::FromAddress(start);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kSimd128Size - kPointerSize);
if (double_misalignment) {
// Test the 2 other alignments possible on 32 bit platforms.
start = AlignOldSpace(kSimd128Unaligned, 2 * kPointerSize);
obj = OldSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There are filler objects before and after the object.
filler1 = HeapObject::FromAddress(start);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == 2 * kPointerSize);
filler2 = HeapObject::FromAddress(start + 3 * kPointerSize);
CHECK(obj != filler2 && filler2->IsFiller() &&
filler2->Size() == kPointerSize);
start = AlignOldSpace(kSimd128Unaligned, 3 * kPointerSize);
obj = OldSpaceAllocateAligned(kPointerSize, kSimd128Unaligned);
CHECK(IsAddressAligned(obj->address(), kSimd128Alignment, kPointerSize));
// There are filler objects before and after the object.
filler1 = HeapObject::FromAddress(start);
CHECK(obj != filler1 && filler1->IsFiller() &&
filler1->Size() == kPointerSize);
filler2 = HeapObject::FromAddress(start + 2 * kPointerSize);
CHECK(obj != filler2 && filler2->IsFiller() &&
filler2->Size() == 2 * kPointerSize);
}
}
TEST(TestSizeOfObjectsVsHeapIteratorPrecision) {
CcTest::InitializeVM();
HeapIterator iterator(CcTest::heap());
intptr_t size_of_objects_1 = CcTest::heap()->SizeOfObjects();
intptr_t size_of_objects_2 = 0;
for (HeapObject* obj = iterator.next();
obj != NULL;
obj = iterator.next()) {
Refactoring of snapshots. This simplifies and improves the speed of deserializing code. The current startup time improvement for V8 is around 6%, but code deserialization is speeded up disproportionately, and we will soon have more code in the snapshot. * Removed support for deserializing into large object space. The regular pages are 1Mbyte now and that is plenty. This is a big simplification. * Instead of reserving space for the snapshot we actually allocate it now. This removes some special casing from the memory management and simplifies deserialization since we are just bumping a pointer rather than calling the normal allocation routines during deserialization. * Record in the snapshot how much we need to boot up and allocate it instead of just assuming that allocations in a new VM will always be linear. * In the snapshot we always address an object as a negative offset from the current allocation point. We used to sometimes address from the start of the deserialized data, but this is less useful now that we have good support for roots and repetitions in the deserialization data. * Code objects were previously deserialized (like other objects) by alternating raw data (deserialized with memcpy) and pointers (to external references, other objects, etc.). Now we deserialize code objects with a single memcpy, followed by a series of skips and pointers that partially overwrite the code we memcopied out of the snapshot. The skips are sometimes merged into the following instruction in the deserialization data to reduce dispatch time. * Integers in the snapshot were stored in a variable length format that gives a compact representation for small positive integers. This is still the case, but the new encoding can be decoded without branches or conditional instructions, which is faster on a modern CPU. Review URL: https://chromiumcodereview.appspot.com/10918067 git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12505 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-14 11:16:56 +00:00
if (!obj->IsFreeSpace()) {
size_of_objects_2 += obj->Size();
}
}
// Delta must be within 5% of the larger result.
// TODO(gc): Tighten this up by distinguishing between byte
// arrays that are real and those that merely mark free space
// on the heap.
if (size_of_objects_1 > size_of_objects_2) {
intptr_t delta = size_of_objects_1 - size_of_objects_2;
PrintF("Heap::SizeOfObjects: %" V8PRIdPTR
", "
"Iterator: %" V8PRIdPTR
", "
"delta: %" V8PRIdPTR "\n",
size_of_objects_1, size_of_objects_2, delta);
CHECK_GT(size_of_objects_1 / 20, delta);
} else {
intptr_t delta = size_of_objects_2 - size_of_objects_1;
PrintF("Heap::SizeOfObjects: %" V8PRIdPTR
", "
"Iterator: %" V8PRIdPTR
", "
"delta: %" V8PRIdPTR "\n",
size_of_objects_1, size_of_objects_2, delta);
CHECK_GT(size_of_objects_2 / 20, delta);
}
}
static void FillUpNewSpace(NewSpace* new_space) {
// Fill up new space to the point that it is completely full. Make sure
// that the scavenger does not undo the filling.
Heap* heap = new_space->heap();
Isolate* isolate = heap->isolate();
Factory* factory = isolate->factory();
HandleScope scope(isolate);
AlwaysAllocateScope always_allocate(isolate);
intptr_t available = new_space->Capacity() - new_space->Size();
intptr_t number_of_fillers = (available / FixedArray::SizeFor(32)) - 1;
for (intptr_t i = 0; i < number_of_fillers; i++) {
CHECK(heap->InNewSpace(*factory->NewFixedArray(32, NOT_TENURED)));
}
}
TEST(GrowAndShrinkNewSpace) {
CcTest::InitializeVM();
Heap* heap = CcTest::heap();
NewSpace* new_space = heap->new_space();
if (heap->MaxSemiSpaceSize() == heap->InitialSemiSpaceSize()) {
return;
}
// Explicitly growing should double the space capacity.
intptr_t old_capacity, new_capacity;
old_capacity = new_space->TotalCapacity();
new_space->Grow();
new_capacity = new_space->TotalCapacity();
CHECK(2 * old_capacity == new_capacity);
old_capacity = new_space->TotalCapacity();
FillUpNewSpace(new_space);
new_capacity = new_space->TotalCapacity();
CHECK(old_capacity == new_capacity);
// Explicitly shrinking should not affect space capacity.
old_capacity = new_space->TotalCapacity();
new_space->Shrink();
new_capacity = new_space->TotalCapacity();
CHECK(old_capacity == new_capacity);
// Let the scavenger empty the new space.
heap->CollectGarbage(NEW_SPACE);
CHECK_LE(new_space->Size(), old_capacity);
// Explicitly shrinking should halve the space capacity.
old_capacity = new_space->TotalCapacity();
new_space->Shrink();
new_capacity = new_space->TotalCapacity();
CHECK(old_capacity == 2 * new_capacity);
// Consecutive shrinking should not affect space capacity.
old_capacity = new_space->TotalCapacity();
new_space->Shrink();
new_space->Shrink();
new_space->Shrink();
new_capacity = new_space->TotalCapacity();
CHECK(old_capacity == new_capacity);
}
TEST(CollectingAllAvailableGarbageShrinksNewSpace) {
CcTest::InitializeVM();
Heap* heap = CcTest::heap();
if (heap->MaxSemiSpaceSize() == heap->InitialSemiSpaceSize()) {
return;
}
v8::HandleScope scope(CcTest::isolate());
NewSpace* new_space = heap->new_space();
intptr_t old_capacity, new_capacity;
old_capacity = new_space->TotalCapacity();
new_space->Grow();
new_capacity = new_space->TotalCapacity();
CHECK(2 * old_capacity == new_capacity);
FillUpNewSpace(new_space);
heap->CollectAllAvailableGarbage();
new_capacity = new_space->TotalCapacity();
CHECK(old_capacity == new_capacity);
}
static int NumberOfGlobalObjects() {
int count = 0;
HeapIterator iterator(CcTest::heap());
for (HeapObject* obj = iterator.next(); obj != NULL; obj = iterator.next()) {
if (obj->IsJSGlobalObject()) count++;
}
return count;
}
// Test that we don't embed maps from foreign contexts into
// optimized code.
TEST(LeakNativeContextViaMap) {
i::FLAG_allow_natives_syntax = true;
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope outer_scope(isolate);
v8::Persistent<v8::Context> ctx1p;
v8::Persistent<v8::Context> ctx2p;
{
v8::HandleScope scope(isolate);
ctx1p.Reset(isolate, v8::Context::New(isolate));
ctx2p.Reset(isolate, v8::Context::New(isolate));
v8::Local<v8::Context>::New(isolate, ctx1p)->Enter();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(2, NumberOfGlobalObjects());
{
v8::HandleScope inner_scope(isolate);
CompileRun("var v = {x: 42}");
v8::Local<v8::Context> ctx1 = v8::Local<v8::Context>::New(isolate, ctx1p);
v8::Local<v8::Context> ctx2 = v8::Local<v8::Context>::New(isolate, ctx2p);
v8::Local<v8::Value> v =
ctx1->Global()->Get(ctx1, v8_str("v")).ToLocalChecked();
ctx2->Enter();
CHECK(ctx2->Global()->Set(ctx2, v8_str("o"), v).FromJust());
v8::Local<v8::Value> res = CompileRun(
"function f() { return o.x; }"
"for (var i = 0; i < 10; ++i) f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
CHECK_EQ(42, res->Int32Value(ctx2).FromJust());
CHECK(ctx2->Global()
->Set(ctx2, v8_str("o"), v8::Int32::New(isolate, 0))
.FromJust());
ctx2->Exit();
v8::Local<v8::Context>::New(isolate, ctx1)->Exit();
ctx1p.Reset();
isolate->ContextDisposedNotification();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(1, NumberOfGlobalObjects());
ctx2p.Reset();
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(0, NumberOfGlobalObjects());
}
// Test that we don't embed functions from foreign contexts into
// optimized code.
TEST(LeakNativeContextViaFunction) {
i::FLAG_allow_natives_syntax = true;
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope outer_scope(isolate);
v8::Persistent<v8::Context> ctx1p;
v8::Persistent<v8::Context> ctx2p;
{
v8::HandleScope scope(isolate);
ctx1p.Reset(isolate, v8::Context::New(isolate));
ctx2p.Reset(isolate, v8::Context::New(isolate));
v8::Local<v8::Context>::New(isolate, ctx1p)->Enter();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(2, NumberOfGlobalObjects());
{
v8::HandleScope inner_scope(isolate);
CompileRun("var v = function() { return 42; }");
v8::Local<v8::Context> ctx1 = v8::Local<v8::Context>::New(isolate, ctx1p);
v8::Local<v8::Context> ctx2 = v8::Local<v8::Context>::New(isolate, ctx2p);
v8::Local<v8::Value> v =
ctx1->Global()->Get(ctx1, v8_str("v")).ToLocalChecked();
ctx2->Enter();
CHECK(ctx2->Global()->Set(ctx2, v8_str("o"), v).FromJust());
v8::Local<v8::Value> res = CompileRun(
"function f(x) { return x(); }"
"for (var i = 0; i < 10; ++i) f(o);"
"%OptimizeFunctionOnNextCall(f);"
"f(o);");
CHECK_EQ(42, res->Int32Value(ctx2).FromJust());
CHECK(ctx2->Global()
->Set(ctx2, v8_str("o"), v8::Int32::New(isolate, 0))
.FromJust());
ctx2->Exit();
ctx1->Exit();
ctx1p.Reset();
isolate->ContextDisposedNotification();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(1, NumberOfGlobalObjects());
ctx2p.Reset();
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(0, NumberOfGlobalObjects());
}
TEST(LeakNativeContextViaMapKeyed) {
i::FLAG_allow_natives_syntax = true;
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope outer_scope(isolate);
v8::Persistent<v8::Context> ctx1p;
v8::Persistent<v8::Context> ctx2p;
{
v8::HandleScope scope(isolate);
ctx1p.Reset(isolate, v8::Context::New(isolate));
ctx2p.Reset(isolate, v8::Context::New(isolate));
v8::Local<v8::Context>::New(isolate, ctx1p)->Enter();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(2, NumberOfGlobalObjects());
{
v8::HandleScope inner_scope(isolate);
CompileRun("var v = [42, 43]");
v8::Local<v8::Context> ctx1 = v8::Local<v8::Context>::New(isolate, ctx1p);
v8::Local<v8::Context> ctx2 = v8::Local<v8::Context>::New(isolate, ctx2p);
v8::Local<v8::Value> v =
ctx1->Global()->Get(ctx1, v8_str("v")).ToLocalChecked();
ctx2->Enter();
CHECK(ctx2->Global()->Set(ctx2, v8_str("o"), v).FromJust());
v8::Local<v8::Value> res = CompileRun(
"function f() { return o[0]; }"
"for (var i = 0; i < 10; ++i) f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
CHECK_EQ(42, res->Int32Value(ctx2).FromJust());
CHECK(ctx2->Global()
->Set(ctx2, v8_str("o"), v8::Int32::New(isolate, 0))
.FromJust());
ctx2->Exit();
ctx1->Exit();
ctx1p.Reset();
isolate->ContextDisposedNotification();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(1, NumberOfGlobalObjects());
ctx2p.Reset();
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(0, NumberOfGlobalObjects());
}
TEST(LeakNativeContextViaMapProto) {
i::FLAG_allow_natives_syntax = true;
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope outer_scope(isolate);
v8::Persistent<v8::Context> ctx1p;
v8::Persistent<v8::Context> ctx2p;
{
v8::HandleScope scope(isolate);
ctx1p.Reset(isolate, v8::Context::New(isolate));
ctx2p.Reset(isolate, v8::Context::New(isolate));
v8::Local<v8::Context>::New(isolate, ctx1p)->Enter();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(2, NumberOfGlobalObjects());
{
v8::HandleScope inner_scope(isolate);
CompileRun("var v = { y: 42}");
v8::Local<v8::Context> ctx1 = v8::Local<v8::Context>::New(isolate, ctx1p);
v8::Local<v8::Context> ctx2 = v8::Local<v8::Context>::New(isolate, ctx2p);
v8::Local<v8::Value> v =
ctx1->Global()->Get(ctx1, v8_str("v")).ToLocalChecked();
ctx2->Enter();
CHECK(ctx2->Global()->Set(ctx2, v8_str("o"), v).FromJust());
v8::Local<v8::Value> res = CompileRun(
"function f() {"
" var p = {x: 42};"
" p.__proto__ = o;"
" return p.x;"
"}"
"for (var i = 0; i < 10; ++i) f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
CHECK_EQ(42, res->Int32Value(ctx2).FromJust());
CHECK(ctx2->Global()
->Set(ctx2, v8_str("o"), v8::Int32::New(isolate, 0))
.FromJust());
ctx2->Exit();
ctx1->Exit();
ctx1p.Reset();
isolate->ContextDisposedNotification();
}
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(1, NumberOfGlobalObjects());
ctx2p.Reset();
CcTest::heap()->CollectAllAvailableGarbage();
CHECK_EQ(0, NumberOfGlobalObjects());
}
TEST(InstanceOfStubWriteBarrier) {
i::FLAG_allow_natives_syntax = true;
#ifdef VERIFY_HEAP
i::FLAG_verify_heap = true;
#endif
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft()) return;
if (i::FLAG_force_marking_deque_overflows) return;
v8::HandleScope outer_scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function foo () { }"
"function mkbar () { return new (new Function(\"\")) (); }"
"function f (x) { return (x instanceof foo); }"
"function g () { f(mkbar()); }"
"f(new foo()); f(new foo());"
"%OptimizeFunctionOnNextCall(f);"
"f(new foo()); g();");
}
IncrementalMarking* marking = CcTest::heap()->incremental_marking();
marking->Stop();
CcTest::heap()->StartIncrementalMarking();
i::Handle<JSFunction> f = i::Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CHECK(f->IsOptimized());
while (!Marking::IsBlack(ObjectMarking::MarkBitFrom(f->code())) &&
!marking->IsStopped()) {
// Discard any pending GC requests otherwise we will get GC when we enter
// code below.
marking->Step(MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD);
}
CHECK(marking->IsMarking());
{
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Object> global = CcTest::global();
v8::Local<v8::Function> g = v8::Local<v8::Function>::Cast(
global->Get(ctx, v8_str("g")).ToLocalChecked());
g->Call(ctx, global, 0, nullptr).ToLocalChecked();
}
CcTest::heap()->incremental_marking()->set_should_hurry(true);
CcTest::heap()->CollectGarbage(OLD_SPACE);
}
namespace {
int GetProfilerTicks(SharedFunctionInfo* shared) {
return FLAG_ignition ? shared->profiler_ticks()
: shared->code()->profiler_ticks();
}
} // namespace
TEST(ResetSharedFunctionInfoCountersDuringIncrementalMarking) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
#ifdef VERIFY_HEAP
i::FLAG_verify_heap = true;
#endif
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft()) return;
v8::HandleScope outer_scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function f () {"
" var s = 0;"
" for (var i = 0; i < 100; i++) s += i;"
" return s;"
"}"
"f(); f();"
"%BaselineFunctionOnNextCall(f);"
"f(); f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
}
i::Handle<JSFunction> f = i::Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CHECK(f->IsOptimized());
// Make sure incremental marking it not running.
CcTest::heap()->incremental_marking()->Stop();
CcTest::heap()->StartIncrementalMarking();
// The following calls will increment CcTest::heap()->global_ic_age().
CcTest::isolate()->ContextDisposedNotification();
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CHECK_EQ(CcTest::heap()->global_ic_age(), f->shared()->ic_age());
CHECK_EQ(0, f->shared()->opt_count());
CHECK_EQ(0, GetProfilerTicks(f->shared()));
}
TEST(ResetSharedFunctionInfoCountersDuringMarkSweep) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
#ifdef VERIFY_HEAP
i::FLAG_verify_heap = true;
#endif
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft()) return;
v8::HandleScope outer_scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function f () {"
" var s = 0;"
" for (var i = 0; i < 100; i++) s += i;"
" return s;"
"}"
"f(); f();"
"%BaselineFunctionOnNextCall(f);"
"f(); f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
}
i::Handle<JSFunction> f = i::Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CHECK(f->IsOptimized());
// Make sure incremental marking it not running.
CcTest::heap()->incremental_marking()->Stop();
// The following two calls will increment CcTest::heap()->global_ic_age().
CcTest::isolate()->ContextDisposedNotification();
CcTest::heap()->CollectAllGarbage();
CHECK_EQ(CcTest::heap()->global_ic_age(), f->shared()->ic_age());
CHECK_EQ(0, f->shared()->opt_count());
CHECK_EQ(0, GetProfilerTicks(f->shared()));
}
HEAP_TEST(GCFlags) {
CcTest::InitializeVM();
Heap* heap = CcTest::heap();
heap->set_current_gc_flags(Heap::kNoGCFlags);
CHECK_EQ(Heap::kNoGCFlags, heap->current_gc_flags_);
// Set the flags to check whether we appropriately resets them after the GC.
heap->set_current_gc_flags(Heap::kAbortIncrementalMarkingMask);
heap->CollectAllGarbage(Heap::kReduceMemoryFootprintMask);
CHECK_EQ(Heap::kNoGCFlags, heap->current_gc_flags_);
MarkCompactCollector* collector = heap->mark_compact_collector();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
IncrementalMarking* marking = heap->incremental_marking();
marking->Stop();
heap->StartIncrementalMarking(Heap::kReduceMemoryFootprintMask);
CHECK_NE(0, heap->current_gc_flags_ & Heap::kReduceMemoryFootprintMask);
heap->CollectGarbage(NEW_SPACE);
// NewSpace scavenges should not overwrite the flags.
CHECK_NE(0, heap->current_gc_flags_ & Heap::kReduceMemoryFootprintMask);
heap->CollectAllGarbage(Heap::kAbortIncrementalMarkingMask);
CHECK_EQ(Heap::kNoGCFlags, heap->current_gc_flags_);
}
TEST(IdleNotificationFinishMarking) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
const int initial_gc_count = CcTest::heap()->gc_count();
heap::SimulateFullSpace(CcTest::heap()->old_space());
IncrementalMarking* marking = CcTest::heap()->incremental_marking();
marking->Stop();
CcTest::heap()->StartIncrementalMarking();
CHECK_EQ(CcTest::heap()->gc_count(), initial_gc_count);
// TODO(hpayer): We cannot write proper unit test right now for heap.
// The ideal test would call kMaxIdleMarkingDelayCounter to test the
// marking delay counter.
// Perform a huge incremental marking step but don't complete marking.
intptr_t bytes_processed = 0;
do {
bytes_processed =
marking->Step(1 * MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD,
IncrementalMarking::FORCE_MARKING,
IncrementalMarking::DO_NOT_FORCE_COMPLETION);
CHECK(!marking->IsIdleMarkingDelayCounterLimitReached());
} while (bytes_processed);
// The next invocations of incremental marking are not going to complete
// marking
// since the completion threshold is not reached
for (size_t i = 0; i < IncrementalMarking::kMaxIdleMarkingDelayCounter - 2;
i++) {
marking->Step(1 * MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD,
IncrementalMarking::FORCE_MARKING,
IncrementalMarking::DO_NOT_FORCE_COMPLETION);
CHECK(!marking->IsIdleMarkingDelayCounterLimitReached());
}
marking->SetWeakClosureWasOverApproximatedForTesting(true);
// The next idle notification has to finish incremental marking.
const double kLongIdleTime = 1000.0;
CcTest::isolate()->IdleNotificationDeadline(
(v8::base::TimeTicks::HighResolutionNow().ToInternalValue() /
static_cast<double>(v8::base::Time::kMicrosecondsPerSecond)) +
kLongIdleTime);
CHECK_EQ(CcTest::heap()->gc_count(), initial_gc_count + 1);
}
// Test that HAllocateObject will always return an object in new-space.
TEST(OptimizedAllocationAlwaysInNewSpace) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
heap::SimulateFullSpace(CcTest::heap()->new_space());
AlwaysAllocateScope always_allocate(CcTest::i_isolate());
v8::Local<v8::Value> res = CompileRun(
"function c(x) {"
" this.x = x;"
" for (var i = 0; i < 32; i++) {"
" this['x' + i] = x;"
" }"
"}"
"function f(x) { return new c(x); };"
"f(1); f(2); f(3);"
"%OptimizeFunctionOnNextCall(f);"
"f(4);");
CHECK_EQ(4, res.As<v8::Object>()
->GetRealNamedProperty(ctx, v8_str("x"))
.ToLocalChecked()
->Int32Value(ctx)
.FromJust());
i::Handle<JSReceiver> o =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res));
CHECK(CcTest::heap()->InNewSpace(*o));
}
TEST(OptimizedPretenuringAllocationFolding) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array();"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [[{}], [1.1]];"
" }"
" return elements[number_elements-1]"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
v8::Local<v8::Value> int_array =
v8::Object::Cast(*res)->Get(ctx, v8_str("0")).ToLocalChecked();
i::Handle<JSObject> int_array_handle = i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(int_array)));
v8::Local<v8::Value> double_array =
v8::Object::Cast(*res)->Get(ctx, v8_str("1")).ToLocalChecked();
i::Handle<JSObject> double_array_handle = i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(double_array)));
i::Handle<JSReceiver> o =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res));
CHECK(CcTest::heap()->InOldSpace(*o));
CHECK(CcTest::heap()->InOldSpace(*int_array_handle));
CHECK(CcTest::heap()->InOldSpace(int_array_handle->elements()));
CHECK(CcTest::heap()->InOldSpace(*double_array_handle));
CHECK(CcTest::heap()->InOldSpace(double_array_handle->elements()));
}
TEST(OptimizedPretenuringObjectArrayLiterals) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [{}, {}, {}];"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(o->elements()));
CHECK(CcTest::heap()->InOldSpace(*o));
}
TEST(OptimizedPretenuringMixedInObjectProperties) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = {a: {c: 2.2, d: {}}, b: 1.1};"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(*o));
FieldIndex idx1 = FieldIndex::ForPropertyIndex(o->map(), 0);
FieldIndex idx2 = FieldIndex::ForPropertyIndex(o->map(), 1);
CHECK(CcTest::heap()->InOldSpace(o->RawFastPropertyAt(idx1)));
if (!o->IsUnboxedDoubleField(idx2)) {
CHECK(CcTest::heap()->InOldSpace(o->RawFastPropertyAt(idx2)));
} else {
CHECK_EQ(1.1, o->RawFastDoublePropertyAt(idx2));
}
JSObject* inner_object =
reinterpret_cast<JSObject*>(o->RawFastPropertyAt(idx1));
CHECK(CcTest::heap()->InOldSpace(inner_object));
if (!inner_object->IsUnboxedDoubleField(idx1)) {
CHECK(CcTest::heap()->InOldSpace(inner_object->RawFastPropertyAt(idx1)));
} else {
CHECK_EQ(2.2, inner_object->RawFastDoublePropertyAt(idx1));
}
CHECK(CcTest::heap()->InOldSpace(inner_object->RawFastPropertyAt(idx2)));
}
TEST(OptimizedPretenuringDoubleArrayProperties) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = {a: 1.1, b: 2.2};"
" }"
" return elements[i - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(*o));
CHECK(CcTest::heap()->InOldSpace(o->properties()));
}
TEST(OptimizedPretenuringdoubleArrayLiterals) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [1.1, 2.2, 3.3];"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(o->elements()));
CHECK(CcTest::heap()->InOldSpace(*o));
}
TEST(OptimizedPretenuringNestedMixedArrayLiterals) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = 100;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [[{}, {}, {}], [1.1, 2.2, 3.3]];"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();");
v8::Local<v8::Value> res = CompileRun(source.start());
v8::Local<v8::Value> int_array =
v8::Object::Cast(*res)->Get(ctx, v8_str("0")).ToLocalChecked();
i::Handle<JSObject> int_array_handle = i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(int_array)));
v8::Local<v8::Value> double_array =
v8::Object::Cast(*res)->Get(ctx, v8_str("1")).ToLocalChecked();
i::Handle<JSObject> double_array_handle = i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(double_array)));
Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(*o));
CHECK(CcTest::heap()->InOldSpace(*int_array_handle));
CHECK(CcTest::heap()->InOldSpace(int_array_handle->elements()));
CHECK(CcTest::heap()->InOldSpace(*double_array_handle));
CHECK(CcTest::heap()->InOldSpace(double_array_handle->elements()));
}
TEST(OptimizedPretenuringNestedObjectLiterals) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [[{}, {}, {}],[{}, {}, {}]];"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
v8::Local<v8::Value> int_array_1 =
v8::Object::Cast(*res)->Get(ctx, v8_str("0")).ToLocalChecked();
Handle<JSObject> int_array_handle_1 = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(int_array_1)));
v8::Local<v8::Value> int_array_2 =
v8::Object::Cast(*res)->Get(ctx, v8_str("1")).ToLocalChecked();
Handle<JSObject> int_array_handle_2 = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(int_array_2)));
Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(*o));
CHECK(CcTest::heap()->InOldSpace(*int_array_handle_1));
CHECK(CcTest::heap()->InOldSpace(int_array_handle_1->elements()));
CHECK(CcTest::heap()->InOldSpace(*int_array_handle_2));
CHECK(CcTest::heap()->InOldSpace(int_array_handle_2->elements()));
}
TEST(OptimizedPretenuringNestedDoubleLiterals) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Grow new space unitl maximum capacity reached.
while (!CcTest::heap()->new_space()->IsAtMaximumCapacity()) {
CcTest::heap()->new_space()->Grow();
}
i::ScopedVector<char> source(1024);
i::SNPrintF(
source,
"var number_elements = %d;"
"var elements = new Array(number_elements);"
"function f() {"
" for (var i = 0; i < number_elements; i++) {"
" elements[i] = [[1.1, 1.2, 1.3],[2.1, 2.2, 2.3]];"
" }"
" return elements[number_elements - 1];"
"};"
"f(); gc();"
"f(); f();"
"%%OptimizeFunctionOnNextCall(f);"
"f();",
AllocationSite::kPretenureMinimumCreated);
v8::Local<v8::Value> res = CompileRun(source.start());
v8::Local<v8::Value> double_array_1 =
v8::Object::Cast(*res)->Get(ctx, v8_str("0")).ToLocalChecked();
i::Handle<JSObject> double_array_handle_1 = i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(double_array_1)));
v8::Local<v8::Value> double_array_2 =
v8::Object::Cast(*res)->Get(ctx, v8_str("1")).ToLocalChecked();
i::Handle<JSObject> double_array_handle_2 = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(double_array_2)));
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InOldSpace(*o));
CHECK(CcTest::heap()->InOldSpace(*double_array_handle_1));
CHECK(CcTest::heap()->InOldSpace(double_array_handle_1->elements()));
CHECK(CcTest::heap()->InOldSpace(*double_array_handle_2));
CHECK(CcTest::heap()->InOldSpace(double_array_handle_2->elements()));
}
// Test regular array literals allocation.
TEST(OptimizedAllocationArrayLiterals) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
if (!CcTest::i_isolate()->use_crankshaft() || i::FLAG_always_opt) return;
if (i::FLAG_gc_global || i::FLAG_stress_compaction) return;
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
v8::Local<v8::Value> res = CompileRun(
"function f() {"
" var numbers = new Array(1, 2, 3);"
" numbers[0] = 3.14;"
" return numbers;"
"};"
"f(); f(); f();"
"%OptimizeFunctionOnNextCall(f);"
"f();");
CHECK_EQ(static_cast<int>(3.14), v8::Object::Cast(*res)
->Get(ctx, v8_str("0"))
.ToLocalChecked()
->Int32Value(ctx)
.FromJust());
i::Handle<JSObject> o = Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(res)));
CHECK(CcTest::heap()->InNewSpace(o->elements()));
}
static int CountMapTransitions(Map* map) {
return TransitionArray::NumberOfTransitions(map->raw_transitions());
}
// Test that map transitions are cleared and maps are collected with
// incremental marking as well.
TEST(Regress1465) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
i::FLAG_trace_incremental_marking = true;
i::FLAG_retain_maps_for_n_gc = 0;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
static const int transitions_count = 256;
CompileRun("function F() {}");
{
AlwaysAllocateScope always_allocate(CcTest::i_isolate());
for (int i = 0; i < transitions_count; i++) {
EmbeddedVector<char, 64> buffer;
SNPrintF(buffer, "var o = new F; o.prop%d = %d;", i, i);
CompileRun(buffer.start());
}
CompileRun("var root = new F;");
}
i::Handle<JSReceiver> root =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(
CcTest::global()->Get(ctx, v8_str("root")).ToLocalChecked()));
// Count number of live transitions before marking.
int transitions_before = CountMapTransitions(root->map());
CompileRun("%DebugPrint(root);");
CHECK_EQ(transitions_count, transitions_before);
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
// Count number of live transitions after marking. Note that one transition
// is left, because 'o' still holds an instance of one transition target.
int transitions_after = CountMapTransitions(root->map());
CompileRun("%DebugPrint(root);");
CHECK_EQ(1, transitions_after);
}
#ifdef DEBUG
static void AddTransitions(int transitions_count) {
AlwaysAllocateScope always_allocate(CcTest::i_isolate());
for (int i = 0; i < transitions_count; i++) {
EmbeddedVector<char, 64> buffer;
SNPrintF(buffer, "var o = new F; o.prop%d = %d;", i, i);
CompileRun(buffer.start());
}
}
static i::Handle<JSObject> GetByName(const char* name) {
return i::Handle<JSObject>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(
CcTest::global()
->Get(CcTest::isolate()->GetCurrentContext(), v8_str(name))
.ToLocalChecked())));
}
static void AddPropertyTo(
int gc_count, Handle<JSObject> object, const char* property_name) {
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Handle<String> prop_name = factory->InternalizeUtf8String(property_name);
Handle<Smi> twenty_three(Smi::FromInt(23), isolate);
i::FLAG_gc_interval = gc_count;
i::FLAG_gc_global = true;
i::FLAG_retain_maps_for_n_gc = 0;
CcTest::heap()->set_allocation_timeout(gc_count);
JSReceiver::SetProperty(object, prop_name, twenty_three, SLOPPY).Check();
}
TEST(TransitionArrayShrinksDuringAllocToZero) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
static const int transitions_count = 10;
CompileRun("function F() { }");
AddTransitions(transitions_count);
CompileRun("var root = new F;");
Handle<JSObject> root = GetByName("root");
// Count number of live transitions before marking.
int transitions_before = CountMapTransitions(root->map());
CHECK_EQ(transitions_count, transitions_before);
// Get rid of o
CompileRun("o = new F;"
"root = new F");
root = GetByName("root");
AddPropertyTo(2, root, "funny");
CcTest::heap()->CollectGarbage(NEW_SPACE);
// Count number of live transitions after marking. Note that one transition
// is left, because 'o' still holds an instance of one transition target.
int transitions_after = CountMapTransitions(
Map::cast(root->map()->GetBackPointer()));
CHECK_EQ(1, transitions_after);
}
TEST(TransitionArrayShrinksDuringAllocToOne) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
static const int transitions_count = 10;
CompileRun("function F() {}");
AddTransitions(transitions_count);
CompileRun("var root = new F;");
Handle<JSObject> root = GetByName("root");
// Count number of live transitions before marking.
int transitions_before = CountMapTransitions(root->map());
CHECK_EQ(transitions_count, transitions_before);
root = GetByName("root");
AddPropertyTo(2, root, "funny");
CcTest::heap()->CollectGarbage(NEW_SPACE);
// Count number of live transitions after marking. Note that one transition
// is left, because 'o' still holds an instance of one transition target.
int transitions_after = CountMapTransitions(
Map::cast(root->map()->GetBackPointer()));
CHECK_EQ(2, transitions_after);
}
TEST(TransitionArrayShrinksDuringAllocToOnePropertyFound) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
static const int transitions_count = 10;
CompileRun("function F() {}");
AddTransitions(transitions_count);
CompileRun("var root = new F;");
Handle<JSObject> root = GetByName("root");
// Count number of live transitions before marking.
int transitions_before = CountMapTransitions(root->map());
CHECK_EQ(transitions_count, transitions_before);
root = GetByName("root");
AddPropertyTo(0, root, "prop9");
CcTest::i_isolate()->heap()->CollectGarbage(OLD_SPACE);
// Count number of live transitions after marking. Note that one transition
// is left, because 'o' still holds an instance of one transition target.
int transitions_after = CountMapTransitions(
Map::cast(root->map()->GetBackPointer()));
CHECK_EQ(1, transitions_after);
}
TEST(TransitionArraySimpleToFull) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
static const int transitions_count = 1;
CompileRun("function F() {}");
AddTransitions(transitions_count);
CompileRun("var root = new F;");
Handle<JSObject> root = GetByName("root");
// Count number of live transitions before marking.
int transitions_before = CountMapTransitions(root->map());
CHECK_EQ(transitions_count, transitions_before);
CompileRun("o = new F;"
"root = new F");
root = GetByName("root");
CHECK(TransitionArray::IsSimpleTransition(root->map()->raw_transitions()));
AddPropertyTo(2, root, "happy");
// Count number of live transitions after marking. Note that one transition
// is left, because 'o' still holds an instance of one transition target.
int transitions_after = CountMapTransitions(
Map::cast(root->map()->GetBackPointer()));
CHECK_EQ(1, transitions_after);
}
#endif // DEBUG
TEST(Regress2143a) {
i::FLAG_incremental_marking = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
// Prepare a map transition from the root object together with a yet
// untransitioned root object.
CompileRun("var root = new Object;"
"root.foo = 0;"
"root = new Object;");
heap::SimulateIncrementalMarking(CcTest::heap());
// Compile a StoreIC that performs the prepared map transition. This
// will restart incremental marking and should make sure the root is
// marked grey again.
CompileRun("function f(o) {"
" o.foo = 0;"
"}"
"f(new Object);"
"f(root);");
// This bug only triggers with aggressive IC clearing.
CcTest::heap()->AgeInlineCaches();
// Explicitly request GC to perform final marking step and sweeping.
CcTest::heap()->CollectAllGarbage();
Handle<JSReceiver> root = v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(
CcTest::global()
->Get(CcTest::isolate()->GetCurrentContext(), v8_str("root"))
.ToLocalChecked()));
// The root object should be in a sane state.
CHECK(root->IsJSObject());
CHECK(root->map()->IsMap());
}
TEST(Regress2143b) {
i::FLAG_incremental_marking = true;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
// Prepare a map transition from the root object together with a yet
// untransitioned root object.
CompileRun("var root = new Object;"
"root.foo = 0;"
"root = new Object;");
heap::SimulateIncrementalMarking(CcTest::heap());
// Compile an optimized LStoreNamedField that performs the prepared
// map transition. This will restart incremental marking and should
// make sure the root is marked grey again.
CompileRun("function f(o) {"
" o.foo = 0;"
"}"
"f(new Object);"
"f(new Object);"
"%OptimizeFunctionOnNextCall(f);"
"f(root);"
"%DeoptimizeFunction(f);");
// This bug only triggers with aggressive IC clearing.
CcTest::heap()->AgeInlineCaches();
// Explicitly request GC to perform final marking step and sweeping.
CcTest::heap()->CollectAllGarbage();
Handle<JSReceiver> root = v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(
CcTest::global()
->Get(CcTest::isolate()->GetCurrentContext(), v8_str("root"))
.ToLocalChecked()));
// The root object should be in a sane state.
CHECK(root->IsJSObject());
CHECK(root->map()->IsMap());
}
TEST(ReleaseOverReservedPages) {
if (FLAG_never_compact) return;
i::FLAG_trace_gc = true;
// The optimizer can allocate stuff, messing up the test.
i::FLAG_crankshaft = false;
i::FLAG_always_opt = false;
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction \o/" This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b. All parallelism can be turned off using --predictable, or --noparallel-compaction. This patch completely parallelizes - semispace copy: from space -> to space (within newspace) - newspace evacuation: newspace -> oldspace - oldspace compaction: oldspace -> oldspace Previously newspace has been handled sequentially (semispace copy, newspace evacuation) before compacting oldspace in parallel. However, on a high level there are no dependencies between those two actions, hence we parallelize them altogether. We base the number of evacuation tasks on the overall set of to-be-processed pages (newspace + oldspace compaction pages). Some low-level details: - The hard cap on number of tasks has been lifted - We cache store buffer entries locally before merging them back into the global StoreBuffer in a finalization phase. - We cache AllocationSite operations locally before merging them back into the global pretenuring storage in a finalization phase. - AllocationSite might be compacted while they would be needed for newspace evacuation. To mitigate any problems we defer checking allocation sites for newspace till merging locally buffered data. CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel BUG=chromium:524425 LOG=N R=hpayer@chromium.org, ulan@chromium.org Review URL: https://codereview.chromium.org/1640563004 Cr-Commit-Position: refs/heads/master@{#33552}
2016-01-27 13:23:54 +00:00
// Parallel compaction increases fragmentation, depending on how existing
// memory is distributed. Since this is non-deterministic because of
// concurrent sweeping, we disable it for this test.
i::FLAG_parallel_compaction = false;
// Concurrent sweeping adds non determinism, depending on when memory is
// available for further reuse.
i::FLAG_concurrent_sweeping = false;
// Fast evacuation of pages may result in a different page count in old space.
i::FLAG_page_promotion = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
v8::HandleScope scope(CcTest::isolate());
static const int number_of_test_pages = 20;
// Prepare many pages with low live-bytes count.
PagedSpace* old_space = heap->old_space();
const int initial_page_count = old_space->CountTotalPages();
const int overall_page_count = number_of_test_pages + initial_page_count;
for (int i = 0; i < number_of_test_pages; i++) {
AlwaysAllocateScope always_allocate(isolate);
heap::SimulateFullSpace(old_space);
factory->NewFixedArray(1, TENURED);
}
CHECK_EQ(overall_page_count, old_space->CountTotalPages());
// Triggering one GC will cause a lot of garbage to be discovered but
// even spread across all allocated pages.
heap->CollectAllGarbage(Heap::kFinalizeIncrementalMarkingMask,
"triggered for preparation");
CHECK_GE(overall_page_count, old_space->CountTotalPages());
// Triggering subsequent GCs should cause at least half of the pages
// to be released to the OS after at most two cycles.
heap->CollectAllGarbage(Heap::kFinalizeIncrementalMarkingMask,
"triggered by test 1");
CHECK_GE(overall_page_count, old_space->CountTotalPages());
heap->CollectAllGarbage(Heap::kFinalizeIncrementalMarkingMask,
"triggered by test 2");
CHECK_GE(overall_page_count, old_space->CountTotalPages() * 2);
// Triggering a last-resort GC should cause all pages to be released to the
// OS so that other processes can seize the memory. If we get a failure here
// where there are 2 pages left instead of 1, then we should increase the
// size of the first page a little in SizeOfFirstPage in spaces.cc. The
// first page should be small in order to reduce memory used when the VM
// boots, but if the 20 small arrays don't fit on the first page then that's
// an indication that it is too small.
heap->CollectAllAvailableGarbage("triggered really hard");
CHECK_EQ(initial_page_count, old_space->CountTotalPages());
}
static int forced_gc_counter = 0;
void MockUseCounterCallback(v8::Isolate* isolate,
v8::Isolate::UseCounterFeature feature) {
isolate->GetCurrentContext();
if (feature == v8::Isolate::kForcedGC) {
forced_gc_counter++;
}
}
TEST(CountForcedGC) {
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::HandleScope scope(CcTest::isolate());
isolate->SetUseCounterCallback(MockUseCounterCallback);
forced_gc_counter = 0;
const char* source = "gc();";
CompileRun(source);
CHECK_GT(forced_gc_counter, 0);
}
#ifdef OBJECT_PRINT
TEST(PrintSharedFunctionInfo) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
const char* source = "f = function() { return 987654321; }\n"
"g = function() { return 123456789; }\n";
CompileRun(source);
i::Handle<JSFunction> g = i::Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("g")).ToLocalChecked())));
OFStream os(stdout);
g->shared()->Print(os);
os << std::endl;
}
#endif // OBJECT_PRINT
TEST(IncrementalMarkingPreservesMonomorphicCallIC) {
if (i::FLAG_always_opt) return;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Value> fun1, fun2;
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
CompileRun("function fun() {};");
fun1 = CcTest::global()->Get(ctx, v8_str("fun")).ToLocalChecked();
}
{
CompileRun("function fun() {};");
fun2 = CcTest::global()->Get(ctx, v8_str("fun")).ToLocalChecked();
}
// Prepare function f that contains type feedback for the two closures.
CHECK(CcTest::global()->Set(ctx, v8_str("fun1"), fun1).FromJust());
CHECK(CcTest::global()->Set(ctx, v8_str("fun2"), fun2).FromJust());
CompileRun("function f(a, b) { a(); b(); } f(fun1, fun2);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
Handle<TypeFeedbackVector> feedback_vector(f->feedback_vector());
FeedbackVectorHelper feedback_helper(feedback_vector);
int expected_slots = 2;
CHECK_EQ(expected_slots, feedback_helper.slot_count());
int slot1 = 0;
int slot2 = 1;
CHECK(feedback_vector->Get(feedback_helper.slot(slot1))->IsWeakCell());
CHECK(feedback_vector->Get(feedback_helper.slot(slot2))->IsWeakCell());
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CHECK(!WeakCell::cast(feedback_vector->Get(feedback_helper.slot(slot1)))
->cleared());
CHECK(!WeakCell::cast(feedback_vector->Get(feedback_helper.slot(slot2)))
->cleared());
}
static Code* FindFirstIC(Code* code, Code::Kind kind) {
int mask = RelocInfo::ModeMask(RelocInfo::CODE_TARGET) |
RelocInfo::ModeMask(RelocInfo::CODE_TARGET_WITH_ID);
for (RelocIterator it(code, mask); !it.done(); it.next()) {
RelocInfo* info = it.rinfo();
Code* target = Code::GetCodeFromTargetAddress(info->target_address());
if (target->is_inline_cache_stub() && target->kind() == kind) {
return target;
}
}
return NULL;
}
static void CheckVectorIC(Handle<JSFunction> f, int slot_index,
InlineCacheState desired_state) {
Handle<TypeFeedbackVector> vector =
Handle<TypeFeedbackVector>(f->feedback_vector());
FeedbackVectorHelper helper(vector);
FeedbackVectorSlot slot = helper.slot(slot_index);
if (vector->GetKind(slot) == FeedbackVectorSlotKind::LOAD_IC) {
LoadICNexus nexus(vector, slot);
CHECK(nexus.StateFromFeedback() == desired_state);
} else {
CHECK_EQ(FeedbackVectorSlotKind::KEYED_LOAD_IC, vector->GetKind(slot));
KeyedLoadICNexus nexus(vector, slot);
CHECK(nexus.StateFromFeedback() == desired_state);
}
}
TEST(IncrementalMarkingPreservesMonomorphicConstructor) {
if (i::FLAG_always_opt) return;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Prepare function f that contains a monomorphic IC for object
// originating from the same native context.
CompileRun(
"function fun() { this.x = 1; };"
"function f(o) { return new o(); } f(fun); f(fun);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
Handle<TypeFeedbackVector> vector(f->feedback_vector());
CHECK(vector->Get(FeedbackVectorSlot(0))->IsWeakCell());
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CHECK(vector->Get(FeedbackVectorSlot(0))->IsWeakCell());
}
TEST(IncrementalMarkingPreservesMonomorphicIC) {
if (i::FLAG_always_opt) return;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
// Prepare function f that contains a monomorphic IC for object
// originating from the same native context.
CompileRun("function fun() { this.x = 1; }; var obj = new fun();"
"function f(o) { return o.x; } f(obj); f(obj);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CheckVectorIC(f, 0, MONOMORPHIC);
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CheckVectorIC(f, 0, MONOMORPHIC);
}
TEST(IncrementalMarkingPreservesPolymorphicIC) {
if (i::FLAG_always_opt) return;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Value> obj1, obj2;
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
LocalContext env;
CompileRun("function fun() { this.x = 1; }; var obj = new fun();");
obj1 = env->Global()->Get(env.local(), v8_str("obj")).ToLocalChecked();
}
{
LocalContext env;
CompileRun("function fun() { this.x = 2; }; var obj = new fun();");
obj2 = env->Global()->Get(env.local(), v8_str("obj")).ToLocalChecked();
}
// Prepare function f that contains a polymorphic IC for objects
// originating from two different native contexts.
CHECK(CcTest::global()->Set(ctx, v8_str("obj1"), obj1).FromJust());
CHECK(CcTest::global()->Set(ctx, v8_str("obj2"), obj2).FromJust());
CompileRun("function f(o) { return o.x; } f(obj1); f(obj1); f(obj2);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CheckVectorIC(f, 0, POLYMORPHIC);
// Fire context dispose notification.
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CheckVectorIC(f, 0, POLYMORPHIC);
}
TEST(ContextDisposeDoesntClearPolymorphicIC) {
if (i::FLAG_always_opt) return;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Value> obj1, obj2;
v8::Local<v8::Context> ctx = CcTest::isolate()->GetCurrentContext();
{
LocalContext env;
CompileRun("function fun() { this.x = 1; }; var obj = new fun();");
obj1 = env->Global()->Get(env.local(), v8_str("obj")).ToLocalChecked();
}
{
LocalContext env;
CompileRun("function fun() { this.x = 2; }; var obj = new fun();");
obj2 = env->Global()->Get(env.local(), v8_str("obj")).ToLocalChecked();
}
// Prepare function f that contains a polymorphic IC for objects
// originating from two different native contexts.
CHECK(CcTest::global()->Set(ctx, v8_str("obj1"), obj1).FromJust());
CHECK(CcTest::global()->Set(ctx, v8_str("obj2"), obj2).FromJust());
CompileRun("function f(o) { return o.x; } f(obj1); f(obj1); f(obj2);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(ctx, v8_str("f")).ToLocalChecked())));
CheckVectorIC(f, 0, POLYMORPHIC);
// Fire context dispose notification.
CcTest::isolate()->ContextDisposedNotification();
heap::SimulateIncrementalMarking(CcTest::heap());
CcTest::heap()->CollectAllGarbage();
CheckVectorIC(f, 0, POLYMORPHIC);
}
class SourceResource : public v8::String::ExternalOneByteStringResource {
public:
explicit SourceResource(const char* data)
: data_(data), length_(strlen(data)) { }
virtual void Dispose() {
i::DeleteArray(data_);
data_ = NULL;
}
const char* data() const { return data_; }
size_t length() const { return length_; }
bool IsDisposed() { return data_ == NULL; }
private:
const char* data_;
size_t length_;
};
void ReleaseStackTraceDataTest(v8::Isolate* isolate, const char* source,
const char* accessor) {
// Test that the data retained by the Error.stack accessor is released
// after the first time the accessor is fired. We use external string
// to check whether the data is being released since the external string
// resource's callback is fired when the external string is GC'ed.
i::Isolate* i_isolate = reinterpret_cast<i::Isolate*>(isolate);
v8::HandleScope scope(isolate);
SourceResource* resource = new SourceResource(i::StrDup(source));
{
v8::HandleScope scope(isolate);
v8::Local<v8::Context> ctx = isolate->GetCurrentContext();
v8::Local<v8::String> source_string =
v8::String::NewExternalOneByte(isolate, resource).ToLocalChecked();
i_isolate->heap()->CollectAllAvailableGarbage();
v8::Script::Compile(ctx, source_string)
.ToLocalChecked()
->Run(ctx)
.ToLocalChecked();
CHECK(!resource->IsDisposed());
}
// i_isolate->heap()->CollectAllAvailableGarbage();
CHECK(!resource->IsDisposed());
CompileRun(accessor);
i_isolate->heap()->CollectAllAvailableGarbage();
// External source has been released.
CHECK(resource->IsDisposed());
delete resource;
}
UNINITIALIZED_TEST(ReleaseStackTraceData) {
if (i::FLAG_always_opt) {
// TODO(ulan): Remove this once the memory leak via code_next_link is fixed.
// See: https://codereview.chromium.org/181833004/
return;
}
FLAG_use_ic = false; // ICs retain objects.
FLAG_concurrent_recompilation = false;
v8::Isolate::CreateParams create_params;
create_params.array_buffer_allocator = CcTest::array_buffer_allocator();
v8::Isolate* isolate = v8::Isolate::New(create_params);
{
v8::Isolate::Scope isolate_scope(isolate);
v8::HandleScope handle_scope(isolate);
v8::Context::New(isolate)->Enter();
static const char* source1 = "var error = null; "
/* Normal Error */ "try { "
" throw new Error(); "
"} catch (e) { "
" error = e; "
"} ";
static const char* source2 = "var error = null; "
/* Stack overflow */ "try { "
" (function f() { f(); })(); "
"} catch (e) { "
" error = e; "
"} ";
static const char* source3 = "var error = null; "
/* Normal Error */ "try { "
/* as prototype */ " throw new Error(); "
"} catch (e) { "
" error = {}; "
" error.__proto__ = e; "
"} ";
static const char* source4 = "var error = null; "
/* Stack overflow */ "try { "
/* as prototype */ " (function f() { f(); })(); "
"} catch (e) { "
" error = {}; "
" error.__proto__ = e; "
"} ";
static const char* getter = "error.stack";
static const char* setter = "error.stack = 0";
ReleaseStackTraceDataTest(isolate, source1, setter);
ReleaseStackTraceDataTest(isolate, source2, setter);
// We do not test source3 and source4 with setter, since the setter is
// supposed to (untypically) write to the receiver, not the holder. This is
// to emulate the behavior of a data property.
ReleaseStackTraceDataTest(isolate, source1, getter);
ReleaseStackTraceDataTest(isolate, source2, getter);
ReleaseStackTraceDataTest(isolate, source3, getter);
ReleaseStackTraceDataTest(isolate, source4, getter);
}
isolate->Dispose();
}
TEST(Regress159140) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
LocalContext env;
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Perform one initial GC to enable code flushing.
heap->CollectAllGarbage();
// Prepare several closures that are all eligible for code flushing
// because all reachable ones are not optimized. Make sure that the
// optimized code object is directly reachable through a handle so
// that it is marked black during incremental marking.
Handle<Code> code;
{
HandleScope inner_scope(isolate);
CompileRun("function h(x) {}"
"function mkClosure() {"
" return function(x) { return x + 1; };"
"}"
"var f = mkClosure();"
"var g = mkClosure();"
"f(1); f(2);"
"g(1); g(2);"
"h(1); h(2);"
"%OptimizeFunctionOnNextCall(f); f(3);"
"%OptimizeFunctionOnNextCall(h); h(3);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("f")).ToLocalChecked())));
CHECK(f->is_compiled());
CompileRun("f = null;");
Handle<JSFunction> g = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("g")).ToLocalChecked())));
CHECK(g->is_compiled());
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
g->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
code = inner_scope.CloseAndEscape(Handle<Code>(f->code()));
}
// Simulate incremental marking so that the functions are enqueued as
// code flushing candidates. Then optimize one function. Finally
// finish the GC to complete code flushing.
heap::SimulateIncrementalMarking(heap);
CompileRun("%OptimizeFunctionOnNextCall(g); g(3);");
heap->CollectAllGarbage();
// Unoptimized code is missing and the deoptimizer will go ballistic.
CompileRun("g('bozo');");
}
TEST(Regress165495) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Perform one initial GC to enable code flushing.
heap->CollectAllGarbage();
// Prepare an optimized closure that the optimized code map will get
// populated. Then age the unoptimized code to trigger code flushing
// but make sure the optimized code is unreachable.
{
HandleScope inner_scope(isolate);
LocalContext env;
CompileRun("function mkClosure() {"
" return function(x) { return x + 1; };"
"}"
"var f = mkClosure();"
"f(1); f(2);"
"%OptimizeFunctionOnNextCall(f); f(3);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("f")).ToLocalChecked())));
CHECK(f->is_compiled());
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
f->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
CompileRun("f = null;");
}
// Simulate incremental marking so that unoptimized code is flushed
// even though it still is cached in the optimized code map.
heap::SimulateIncrementalMarking(heap);
heap->CollectAllGarbage();
// Make a new closure that will get code installed from the code map.
// Unoptimized code is missing and the deoptimizer will go ballistic.
CompileRun("var g = mkClosure(); g('bozo');");
}
TEST(Regress169209) {
i::FLAG_stress_compaction = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Perform one initial GC to enable code flushing.
heap->CollectAllGarbage();
// Prepare a shared function info eligible for code flushing for which
// the unoptimized code will be replaced during optimization.
Handle<SharedFunctionInfo> shared1;
{
HandleScope inner_scope(isolate);
LocalContext env;
CompileRun("function f() { return 'foobar'; }"
"function g(x) { if (x) f(); }"
"f();"
"g(false);"
"g(false);");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("f")).ToLocalChecked())));
CHECK(f->is_compiled());
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
f->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
shared1 = inner_scope.CloseAndEscape(handle(f->shared(), isolate));
}
// Prepare a shared function info eligible for code flushing that will
// represent the dangling tail of the candidate list.
Handle<SharedFunctionInfo> shared2;
{
HandleScope inner_scope(isolate);
LocalContext env;
CompileRun("function flushMe() { return 0; }"
"flushMe(1);");
Handle<JSFunction> f = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(env.local(), v8_str("flushMe"))
.ToLocalChecked())));
CHECK(f->is_compiled());
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
f->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
shared2 = inner_scope.CloseAndEscape(handle(f->shared(), isolate));
}
// Simulate incremental marking and collect code flushing candidates.
heap::SimulateIncrementalMarking(heap);
CHECK(shared1->code()->gc_metadata() != NULL);
// Optimize function and make sure the unoptimized code is replaced.
CompileRun("%OptimizeFunctionOnNextCall(g);"
"g(false);");
// Finish garbage collection cycle.
heap->CollectAllGarbage();
CHECK(shared1->code()->gc_metadata() == NULL);
}
TEST(Regress169928) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_crankshaft = false;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
LocalContext env;
Factory* factory = isolate->factory();
v8::HandleScope scope(CcTest::isolate());
// Some flags turn Scavenge collections into Mark-sweep collections
// and hence are incompatible with this test case.
if (FLAG_gc_global || FLAG_stress_compaction) return;
// Prepare the environment
CompileRun("function fastliteralcase(literal, value) {"
" literal[0] = value;"
" return literal;"
"}"
"function get_standard_literal() {"
" var literal = [1, 2, 3];"
" return literal;"
"}"
"obj = fastliteralcase(get_standard_literal(), 1);"
"obj = fastliteralcase(get_standard_literal(), 1.5);"
"obj = fastliteralcase(get_standard_literal(), 2);");
// prepare the heap
v8::Local<v8::String> mote_code_string =
v8_str("fastliteralcase(mote, 2.5);");
v8::Local<v8::String> array_name = v8_str("mote");
CHECK(CcTest::global()
->Set(env.local(), array_name, v8::Int32::New(CcTest::isolate(), 0))
.FromJust());
// First make sure we flip spaces
CcTest::heap()->CollectGarbage(NEW_SPACE);
// Allocate the object.
Handle<FixedArray> array_data = factory->NewFixedArray(2, NOT_TENURED);
array_data->set(0, Smi::FromInt(1));
array_data->set(1, Smi::FromInt(2));
heap::AllocateAllButNBytes(
CcTest::heap()->new_space(),
JSArray::kSize + AllocationMemento::kSize + kPointerSize);
Handle<JSArray> array =
factory->NewJSArrayWithElements(array_data, FAST_SMI_ELEMENTS);
CHECK_EQ(Smi::FromInt(2), array->length());
CHECK(array->HasFastSmiOrObjectElements());
// We need filler the size of AllocationMemento object, plus an extra
// fill pointer value.
HeapObject* obj = NULL;
AllocationResult allocation =
CcTest::heap()->new_space()->AllocateRawUnaligned(
AllocationMemento::kSize + kPointerSize);
CHECK(allocation.To(&obj));
Address addr_obj = obj->address();
CcTest::heap()->CreateFillerObjectAt(addr_obj,
AllocationMemento::kSize + kPointerSize,
ClearRecordedSlots::kNo);
// Give the array a name, making sure not to allocate strings.
v8::Local<v8::Object> array_obj = v8::Utils::ToLocal(array);
CHECK(CcTest::global()->Set(env.local(), array_name, array_obj).FromJust());
// This should crash with a protection violation if we are running a build
// with the bug.
AlwaysAllocateScope aa_scope(isolate);
v8::Script::Compile(env.local(), mote_code_string)
.ToLocalChecked()
->Run(env.local())
.ToLocalChecked();
}
#ifdef DEBUG
TEST(Regress513507) {
i::FLAG_flush_optimized_code_cache = false;
i::FLAG_allow_natives_syntax = true;
i::FLAG_gc_global = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
LocalContext env;
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Prepare function whose optimized code map we can use.
Handle<SharedFunctionInfo> shared;
{
HandleScope inner_scope(isolate);
CompileRun("function f() { return 1 }"
"f(); %OptimizeFunctionOnNextCall(f); f();");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("f")).ToLocalChecked())));
shared = inner_scope.CloseAndEscape(handle(f->shared(), isolate));
CompileRun("f = null");
}
// Prepare optimized code that we can use.
Handle<Code> code;
{
HandleScope inner_scope(isolate);
CompileRun("function g() { return 2 }"
"g(); %OptimizeFunctionOnNextCall(g); g();");
Handle<JSFunction> g = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("g")).ToLocalChecked())));
code = inner_scope.CloseAndEscape(handle(g->code(), isolate));
if (!code->is_optimized_code()) return;
}
Handle<TypeFeedbackVector> vector =
TypeFeedbackVector::New(isolate, handle(shared->feedback_metadata()));
Handle<LiteralsArray> lit =
LiteralsArray::New(isolate, vector, shared->num_literals());
Handle<Context> context(isolate->context());
// Add the new code several times to the optimized code map and also set an
// allocation timeout so that expanding the code map will trigger a GC.
heap->set_allocation_timeout(5);
FLAG_gc_interval = 1000;
for (int i = 0; i < 10; ++i) {
BailoutId id = BailoutId(i);
SharedFunctionInfo::AddToOptimizedCodeMap(shared, context, code, lit, id);
}
}
#endif // DEBUG
TEST(Regress514122) {
i::FLAG_flush_optimized_code_cache = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
LocalContext env;
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Perfrom one initial GC to enable code flushing.
CcTest::heap()->CollectAllGarbage();
// Prepare function whose optimized code map we can use.
Handle<SharedFunctionInfo> shared;
{
HandleScope inner_scope(isolate);
CompileRun("function f() { return 1 }"
"f(); %OptimizeFunctionOnNextCall(f); f();");
Handle<JSFunction> f = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("f")).ToLocalChecked())));
shared = inner_scope.CloseAndEscape(handle(f->shared(), isolate));
CompileRun("f = null");
}
// Prepare optimized code that we can use.
Handle<Code> code;
{
HandleScope inner_scope(isolate);
CompileRun("function g() { return 2 }"
"g(); %OptimizeFunctionOnNextCall(g); g();");
Handle<JSFunction> g = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(env.local(), v8_str("g")).ToLocalChecked())));
code = inner_scope.CloseAndEscape(handle(g->code(), isolate));
if (!code->is_optimized_code()) return;
}
Handle<TypeFeedbackVector> vector =
TypeFeedbackVector::New(isolate, handle(shared->feedback_metadata()));
Handle<LiteralsArray> lit =
LiteralsArray::New(isolate, vector, shared->num_literals(), TENURED);
Handle<Context> context(isolate->context());
// Add the code several times to the optimized code map.
for (int i = 0; i < 3; ++i) {
HandleScope inner_scope(isolate);
BailoutId id = BailoutId(i);
SharedFunctionInfo::AddToOptimizedCodeMap(shared, context, code, lit, id);
}
shared->optimized_code_map()->Print();
// Add the code with a literals array to be evacuated.
Page* evac_page;
{
HandleScope inner_scope(isolate);
AlwaysAllocateScope always_allocate(isolate);
// Make sure literal is placed on an old-space evacuation candidate.
heap::SimulateFullSpace(heap->old_space());
// Make sure there the number of literals is > 0.
Handle<LiteralsArray> lit = LiteralsArray::New(isolate, vector, 23);
evac_page = Page::FromAddress(lit->address());
BailoutId id = BailoutId(100);
SharedFunctionInfo::AddToOptimizedCodeMap(shared, context, code, lit, id);
}
// Heap is ready, force {lit_page} to become an evacuation candidate and
// simulate incremental marking to enqueue optimized code map.
FLAG_manual_evacuation_candidates_selection = true;
evac_page->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
heap::SimulateIncrementalMarking(heap);
// No matter whether reachable or not, {boomer} is doomed.
Handle<Object> boomer(shared->optimized_code_map(), isolate);
// Add the code several times to the optimized code map. This will leave old
// copies of the optimized code map unreachable but still marked.
for (int i = 3; i < 6; ++i) {
HandleScope inner_scope(isolate);
BailoutId id = BailoutId(i);
SharedFunctionInfo::AddToOptimizedCodeMap(shared, context, code, lit, id);
}
// Trigger a GC to flush out the bug.
heap->CollectGarbage(i::OLD_SPACE, "fire in the hole");
boomer->Print();
}
TEST(OptimizedCodeMapReuseEntries) {
i::FLAG_flush_optimized_code_cache = false;
i::FLAG_allow_natives_syntax = true;
// BUG(v8:4598): Since TurboFan doesn't treat maps in code weakly, we can't
// run this test.
if (i::FLAG_turbo) return;
CcTest::InitializeVM();
v8::Isolate* v8_isolate = CcTest::isolate();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Create 3 contexts, allow the 2nd one to be disposed, and verify that
// a 4th context will re-use the weak slots in the optimized code map
// to hold data, rather than expanding the map.
v8::Local<v8::Context> c1 = v8::Context::New(v8_isolate);
const char* source = "function foo(x) { var l = [1]; return x+l[0]; }";
v8::ScriptCompiler::Source script_source(
v8::String::NewFromUtf8(v8_isolate, source, v8::NewStringType::kNormal)
.ToLocalChecked());
v8::Local<v8::UnboundScript> indep =
v8::ScriptCompiler::CompileUnboundScript(v8_isolate, &script_source)
.ToLocalChecked();
const char* toplevel = "foo(3); %OptimizeFunctionOnNextCall(foo); foo(3);";
// Perfrom one initial GC to enable code flushing.
heap->CollectAllGarbage();
c1->Enter();
indep->BindToCurrentContext()->Run(c1).ToLocalChecked();
CompileRun(toplevel);
Handle<SharedFunctionInfo> shared;
Handle<JSFunction> foo = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()->Get(c1, v8_str("foo")).ToLocalChecked())));
CHECK(foo->shared()->is_compiled());
shared = handle(foo->shared());
c1->Exit();
{
HandleScope scope(isolate);
v8::Local<v8::Context> c2 = v8::Context::New(v8_isolate);
c2->Enter();
indep->BindToCurrentContext()->Run(c2).ToLocalChecked();
CompileRun(toplevel);
c2->Exit();
}
{
HandleScope scope(isolate);
v8::Local<v8::Context> c3 = v8::Context::New(v8_isolate);
c3->Enter();
indep->BindToCurrentContext()->Run(c3).ToLocalChecked();
CompileRun(toplevel);
c3->Exit();
// Now, collect garbage. Context c2 should have no roots to it, and it's
// entry in the optimized code map should be free for a new context.
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
Handle<FixedArray> optimized_code_map =
handle(shared->optimized_code_map());
// There should be 3 entries in the map.
CHECK_EQ(
3, ((optimized_code_map->length() - SharedFunctionInfo::kEntriesStart) /
SharedFunctionInfo::kEntryLength));
// But one of them (formerly for c2) should be cleared.
int cleared_count = 0;
for (int i = SharedFunctionInfo::kEntriesStart;
i < optimized_code_map->length();
i += SharedFunctionInfo::kEntryLength) {
cleared_count +=
WeakCell::cast(
optimized_code_map->get(i + SharedFunctionInfo::kContextOffset))
->cleared()
? 1
: 0;
}
CHECK_EQ(1, cleared_count);
// Verify that a new context uses the cleared entry rather than creating a
// new
// optimized code map array.
v8::Local<v8::Context> c4 = v8::Context::New(v8_isolate);
c4->Enter();
indep->BindToCurrentContext()->Run(c4).ToLocalChecked();
CompileRun(toplevel);
c4->Exit();
CHECK_EQ(*optimized_code_map, shared->optimized_code_map());
// Now each entry is in use.
cleared_count = 0;
for (int i = SharedFunctionInfo::kEntriesStart;
i < optimized_code_map->length();
i += SharedFunctionInfo::kEntryLength) {
cleared_count +=
WeakCell::cast(
optimized_code_map->get(i + SharedFunctionInfo::kContextOffset))
->cleared()
? 1
: 0;
}
CHECK_EQ(0, cleared_count);
}
}
TEST(Regress513496) {
i::FLAG_flush_optimized_code_cache = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Perfrom one initial GC to enable code flushing.
CcTest::heap()->CollectAllGarbage();
// Prepare an optimized closure with containing an inlined function. Then age
// the inlined unoptimized code to trigger code flushing but make sure the
// outer optimized code is kept in the optimized code map.
Handle<SharedFunctionInfo> shared;
{
LocalContext context;
HandleScope inner_scope(isolate);
CompileRun(
"function g(x) { return x + 1 }"
"function mkClosure() {"
" return function(x) { return g(x); };"
"}"
"var f = mkClosure();"
"f(1); f(2);"
"%OptimizeFunctionOnNextCall(f); f(3);");
Handle<JSFunction> g = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("g"))
.ToLocalChecked())));
CHECK(g->shared()->is_compiled());
const int kAgingThreshold = 6;
for (int i = 0; i < kAgingThreshold; i++) {
g->shared()->code()->MakeOlder(static_cast<MarkingParity>(i % 2));
}
Handle<JSFunction> f = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("f"))
.ToLocalChecked())));
CHECK(f->is_compiled());
shared = inner_scope.CloseAndEscape(handle(f->shared(), isolate));
CompileRun("f = null");
}
// Lookup the optimized code and keep it alive.
CodeAndLiterals result = shared->SearchOptimizedCodeMap(
isolate->context()->native_context(), BailoutId::None());
Handle<Code> optimized_code(result.code, isolate);
// Finish a full GC cycle so that the unoptimized code of 'g' is flushed even
// though the optimized code for 'f' is reachable via the optimized code map.
heap->CollectAllGarbage();
// Make a new closure that will get code installed from the code map.
// Unoptimized code is missing and the deoptimizer will go ballistic.
CompileRun("var h = mkClosure(); h('bozo');");
}
TEST(LargeObjectSlotRecording) {
FLAG_manual_evacuation_candidates_selection = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope scope(isolate);
// Create an object on an evacuation candidate.
heap::SimulateFullSpace(heap->old_space());
Handle<FixedArray> lit = isolate->factory()->NewFixedArray(4, TENURED);
Page* evac_page = Page::FromAddress(lit->address());
evac_page->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
FixedArray* old_location = *lit;
// Allocate a large object.
int size = Max(1000000, Page::kMaxRegularHeapObjectSize + KB);
CHECK(size > Page::kMaxRegularHeapObjectSize);
Handle<FixedArray> lo = isolate->factory()->NewFixedArray(size, TENURED);
CHECK(heap->lo_space()->Contains(*lo));
// Start incremental marking to active write barrier.
heap::SimulateIncrementalMarking(heap, false);
heap->incremental_marking()->AdvanceIncrementalMarking(
10000000, IncrementalMarking::IdleStepActions());
// Create references from the large object to the object on the evacuation
// candidate.
const int kStep = size / 10;
for (int i = 0; i < size; i += kStep) {
lo->set(i, *lit);
CHECK(lo->get(i) == old_location);
}
// Move the evaucation candidate object.
CcTest::heap()->CollectAllGarbage();
// Verify that the pointers in the large object got updated.
for (int i = 0; i < size; i += kStep) {
CHECK_EQ(lo->get(i), *lit);
CHECK(lo->get(i) != old_location);
}
}
class DummyVisitor : public ObjectVisitor {
public:
void VisitPointers(Object** start, Object** end) override {}
};
TEST(DeferredHandles) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
v8::HandleScope scope(reinterpret_cast<v8::Isolate*>(isolate));
HandleScopeData* data = isolate->handle_scope_data();
Handle<Object> init(heap->empty_string(), isolate);
while (data->next < data->limit) {
Handle<Object> obj(heap->empty_string(), isolate);
}
// An entire block of handles has been filled.
// Next handle would require a new block.
CHECK(data->next == data->limit);
DeferredHandleScope deferred(isolate);
DummyVisitor visitor;
isolate->handle_scope_implementer()->Iterate(&visitor);
delete deferred.Detach();
}
TEST(IncrementalMarkingStepMakesBigProgressWithLargeObjects) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
CompileRun("function f(n) {"
" var a = new Array(n);"
" for (var i = 0; i < n; i += 100) a[i] = i;"
"};"
"f(10 * 1024 * 1024);");
IncrementalMarking* marking = CcTest::heap()->incremental_marking();
if (marking->IsStopped()) {
CcTest::heap()->StartIncrementalMarking();
}
// This big step should be sufficient to mark the whole array.
marking->Step(100 * MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD);
CHECK(marking->IsComplete() ||
marking->IsReadyToOverApproximateWeakClosure());
}
TEST(DisableInlineAllocation) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
CompileRun("function test() {"
" var x = [];"
" for (var i = 0; i < 10; i++) {"
" x[i] = [ {}, [1,2,3], [1,x,3] ];"
" }"
"}"
"function run() {"
" %OptimizeFunctionOnNextCall(test);"
" test();"
" %DeoptimizeFunction(test);"
"}");
// Warm-up with inline allocation enabled.
CompileRun("test(); test(); run();");
// Run test with inline allocation disabled.
CcTest::heap()->DisableInlineAllocation();
CompileRun("run()");
// Run test with inline allocation re-enabled.
CcTest::heap()->EnableInlineAllocation();
CompileRun("run()");
}
static int AllocationSitesCount(Heap* heap) {
int count = 0;
for (Object* site = heap->allocation_sites_list();
!(site->IsUndefined(heap->isolate()));
site = AllocationSite::cast(site)->weak_next()) {
count++;
}
return count;
}
TEST(EnsureAllocationSiteDependentCodesProcessed) {
if (i::FLAG_always_opt || !i::FLAG_crankshaft) return;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
GlobalHandles* global_handles = isolate->global_handles();
if (!isolate->use_crankshaft()) return;
// The allocation site at the head of the list is ours.
Handle<AllocationSite> site;
{
LocalContext context;
v8::HandleScope scope(context->GetIsolate());
int count = AllocationSitesCount(heap);
CompileRun("var bar = function() { return (new Array()); };"
"var a = bar();"
"bar();"
"bar();");
// One allocation site should have been created.
int new_count = AllocationSitesCount(heap);
CHECK_EQ(new_count, (count + 1));
site = Handle<AllocationSite>::cast(
global_handles->Create(
AllocationSite::cast(heap->allocation_sites_list())));
CompileRun("%OptimizeFunctionOnNextCall(bar); bar();");
Handle<JSFunction> bar_handle = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()
->Get(context.local(), v8_str("bar"))
.ToLocalChecked())));
int dependency_group_count = 0;
DependentCode* dependency = site->dependent_code();
while (dependency != heap->empty_fixed_array()) {
CHECK(dependency->group() ==
DependentCode::kAllocationSiteTransitionChangedGroup ||
dependency->group() ==
DependentCode::kAllocationSiteTenuringChangedGroup);
CHECK_EQ(1, dependency->count());
CHECK(dependency->object_at(0)->IsWeakCell());
Code* function_bar =
Code::cast(WeakCell::cast(dependency->object_at(0))->value());
CHECK_EQ(bar_handle->code(), function_bar);
dependency = dependency->next_link();
dependency_group_count++;
}
// TurboFan respects pretenuring feedback from allocation sites, Crankshaft
// does not. Either is fine for the purposes of this test.
CHECK(dependency_group_count == 1 || dependency_group_count == 2);
}
// Now make sure that a gc should get rid of the function, even though we
// still have the allocation site alive.
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
// The site still exists because of our global handle, but the code is no
// longer referred to by dependent_code().
CHECK(site->dependent_code()->object_at(0)->IsWeakCell() &&
WeakCell::cast(site->dependent_code()->object_at(0))->cleared());
}
TEST(CellsInOptimizedCodeAreWeak) {
if (i::FLAG_always_opt || !i::FLAG_crankshaft) return;
i::FLAG_weak_embedded_objects_in_optimized_code = true;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
Handle<Code> code;
{
LocalContext context;
HandleScope scope(heap->isolate());
CompileRun(
"bar = (function() {"
" function bar() {"
" return foo(1);"
" };"
" var foo = function(x) { with (x) { return 1 + x; } };"
" %NeverOptimizeFunction(foo);"
" bar(foo);"
" bar(foo);"
" bar(foo);"
" %OptimizeFunctionOnNextCall(bar);"
" bar(foo);"
" return bar;})();");
Handle<JSFunction> bar = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("bar"))
.ToLocalChecked())));
code = scope.CloseAndEscape(Handle<Code>(bar->code()));
}
// Now make sure that a gc should get rid of the function
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
CHECK(code->marked_for_deoptimization());
}
TEST(ObjectsInOptimizedCodeAreWeak) {
if (i::FLAG_always_opt || !i::FLAG_crankshaft) return;
i::FLAG_weak_embedded_objects_in_optimized_code = true;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
Handle<Code> code;
{
LocalContext context;
HandleScope scope(heap->isolate());
CompileRun(
"function bar() {"
" return foo(1);"
"};"
"function foo(x) { with (x) { return 1 + x; } };"
"%NeverOptimizeFunction(foo);"
"bar();"
"bar();"
"bar();"
"%OptimizeFunctionOnNextCall(bar);"
"bar();");
Handle<JSFunction> bar = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("bar"))
.ToLocalChecked())));
code = scope.CloseAndEscape(Handle<Code>(bar->code()));
}
// Now make sure that a gc should get rid of the function
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
CHECK(code->marked_for_deoptimization());
}
TEST(NewSpaceObjectsInOptimizedCode) {
if (i::FLAG_always_opt || !i::FLAG_crankshaft || i::FLAG_turbo) return;
i::FLAG_weak_embedded_objects_in_optimized_code = true;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
Handle<Code> code;
{
LocalContext context;
HandleScope scope(heap->isolate());
CompileRun(
"var foo;"
"var bar;"
"(function() {"
" function foo_func(x) { with (x) { return 1 + x; } };"
" %NeverOptimizeFunction(foo_func);"
" function bar_func() {"
" return foo(1);"
" };"
" bar = bar_func;"
" foo = foo_func;"
" bar_func();"
" bar_func();"
" bar_func();"
" %OptimizeFunctionOnNextCall(bar_func);"
" bar_func();"
"})();");
Handle<JSFunction> bar = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("bar"))
.ToLocalChecked())));
Handle<JSFunction> foo = Handle<JSFunction>::cast(v8::Utils::OpenHandle(
*v8::Local<v8::Function>::Cast(CcTest::global()
->Get(context.local(), v8_str("foo"))
.ToLocalChecked())));
CHECK(heap->InNewSpace(*foo));
heap->CollectGarbage(NEW_SPACE);
heap->CollectGarbage(NEW_SPACE);
CHECK(!heap->InNewSpace(*foo));
#ifdef VERIFY_HEAP
heap->Verify();
#endif
CHECK(!bar->code()->marked_for_deoptimization());
code = scope.CloseAndEscape(Handle<Code>(bar->code()));
}
// Now make sure that a gc should get rid of the function
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
CHECK(code->marked_for_deoptimization());
}
TEST(NoWeakHashTableLeakWithIncrementalMarking) {
if (i::FLAG_always_opt || !i::FLAG_crankshaft) return;
if (!i::FLAG_incremental_marking) return;
i::FLAG_weak_embedded_objects_in_optimized_code = true;
i::FLAG_allow_natives_syntax = true;
i::FLAG_compilation_cache = false;
i::FLAG_retain_maps_for_n_gc = 0;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
// Do not run for no-snap builds.
if (!i::Snapshot::HasContextSnapshot(isolate, 0)) return;
v8::internal::Heap* heap = CcTest::heap();
// Get a clean slate regarding optimized functions on the heap.
i::Deoptimizer::DeoptimizeAll(isolate);
heap->CollectAllGarbage();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
for (int i = 0; i < 3; i++) {
heap::SimulateIncrementalMarking(heap);
{
LocalContext context;
HandleScope scope(heap->isolate());
EmbeddedVector<char, 256> source;
SNPrintF(source,
"function bar%d() {"
" return foo%d(1);"
"};"
"function foo%d(x) { with (x) { return 1 + x; } };"
"bar%d();"
"bar%d();"
"bar%d();"
"%%OptimizeFunctionOnNextCall(bar%d);"
"bar%d();",
i, i, i, i, i, i, i, i);
CompileRun(source.start());
}
// We have to abort incremental marking here to abandon black pages.
heap->CollectAllGarbage(Heap::kAbortIncrementalMarkingMask);
}
int elements = 0;
if (heap->weak_object_to_code_table()->IsHashTable()) {
WeakHashTable* t = WeakHashTable::cast(heap->weak_object_to_code_table());
elements = t->NumberOfElements();
}
CHECK_EQ(0, elements);
}
static Handle<JSFunction> OptimizeDummyFunction(v8::Isolate* isolate,
const char* name) {
EmbeddedVector<char, 256> source;
SNPrintF(source,
"function %s() { return 0; }"
"%s(); %s();"
"%%OptimizeFunctionOnNextCall(%s);"
"%s();", name, name, name, name, name);
CompileRun(source.start());
i::Handle<JSFunction> fun = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()
->Get(isolate->GetCurrentContext(), v8_str(name))
.ToLocalChecked())));
return fun;
}
static int GetCodeChainLength(Code* code) {
int result = 0;
while (code->next_code_link()->IsCode()) {
result++;
code = Code::cast(code->next_code_link());
}
return result;
}
TEST(NextCodeLinkIsWeak) {
i::FLAG_always_opt = false;
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
Handle<Code> code;
heap->CollectAllAvailableGarbage();
int code_chain_length_before, code_chain_length_after;
{
HandleScope scope(heap->isolate());
Handle<JSFunction> mortal =
OptimizeDummyFunction(CcTest::isolate(), "mortal");
Handle<JSFunction> immortal =
OptimizeDummyFunction(CcTest::isolate(), "immortal");
CHECK_EQ(immortal->code()->next_code_link(), mortal->code());
code_chain_length_before = GetCodeChainLength(immortal->code());
// Keep the immortal code and let the mortal code die.
code = scope.CloseAndEscape(Handle<Code>(immortal->code()));
CompileRun("mortal = null; immortal = null;");
}
heap->CollectAllAvailableGarbage();
// Now mortal code should be dead.
code_chain_length_after = GetCodeChainLength(*code);
CHECK_EQ(code_chain_length_before - 1, code_chain_length_after);
}
static Handle<Code> DummyOptimizedCode(Isolate* isolate) {
i::byte buffer[i::Assembler::kMinimalBufferSize];
MacroAssembler masm(isolate, buffer, sizeof(buffer),
v8::internal::CodeObjectRequired::kYes);
CodeDesc desc;
masm.Push(isolate->factory()->undefined_value());
masm.Drop(1);
masm.GetCode(&desc);
Handle<Object> undefined(isolate->heap()->undefined_value(), isolate);
Handle<Code> code = isolate->factory()->NewCode(
desc, Code::ComputeFlags(Code::OPTIMIZED_FUNCTION), undefined);
CHECK(code->IsCode());
return code;
}
TEST(NextCodeLinkIsWeak2) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
if (!isolate->use_crankshaft()) return;
HandleScope outer_scope(heap->isolate());
heap->CollectAllAvailableGarbage();
Handle<Context> context(Context::cast(heap->native_contexts_list()), isolate);
Handle<Code> new_head;
Handle<Object> old_head(context->get(Context::OPTIMIZED_CODE_LIST), isolate);
{
HandleScope scope(heap->isolate());
Handle<Code> immortal = DummyOptimizedCode(isolate);
Handle<Code> mortal = DummyOptimizedCode(isolate);
mortal->set_next_code_link(*old_head);
immortal->set_next_code_link(*mortal);
context->set(Context::OPTIMIZED_CODE_LIST, *immortal);
new_head = scope.CloseAndEscape(immortal);
}
heap->CollectAllAvailableGarbage();
// Now mortal code should be dead.
CHECK_EQ(*old_head, new_head->next_code_link());
}
static bool weak_ic_cleared = false;
static void ClearWeakIC(
const v8::WeakCallbackInfo<v8::Persistent<v8::Object>>& data) {
printf("clear weak is called\n");
weak_ic_cleared = true;
data.GetParameter()->Reset();
}
TEST(WeakFunctionInConstructor) {
if (i::FLAG_always_opt) return;
i::FLAG_stress_compaction = false;
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
LocalContext env;
v8::HandleScope scope(isolate);
CompileRun(
"function createObj(obj) {"
" return new obj();"
"}");
i::Handle<JSFunction> createObj = Handle<JSFunction>::cast(
v8::Utils::OpenHandle(*v8::Local<v8::Function>::Cast(
CcTest::global()
->Get(env.local(), v8_str("createObj"))
.ToLocalChecked())));
v8::Persistent<v8::Object> garbage;
{
v8::HandleScope scope(isolate);
const char* source =
" (function() {"
" function hat() { this.x = 5; }"
" createObj(hat);"
" createObj(hat);"
" return hat;"
" })();";
garbage.Reset(isolate, CompileRun(env.local(), source)
.ToLocalChecked()
->ToObject(env.local())
.ToLocalChecked());
}
weak_ic_cleared = false;
garbage.SetWeak(&garbage, &ClearWeakIC, v8::WeakCallbackType::kParameter);
Heap* heap = CcTest::i_isolate()->heap();
heap->CollectAllGarbage();
CHECK(weak_ic_cleared);
// We've determined the constructor in createObj has had it's weak cell
// cleared. Now, verify that one additional call with a new function
// allows monomorphicity.
Handle<TypeFeedbackVector> feedback_vector = Handle<TypeFeedbackVector>(
createObj->feedback_vector(), CcTest::i_isolate());
for (int i = 0; i < 20; i++) {
Object* slot_value = feedback_vector->Get(FeedbackVectorSlot(0));
CHECK(slot_value->IsWeakCell());
if (WeakCell::cast(slot_value)->cleared()) break;
heap->CollectAllGarbage();
}
Object* slot_value = feedback_vector->Get(FeedbackVectorSlot(0));
CHECK(slot_value->IsWeakCell() && WeakCell::cast(slot_value)->cleared());
CompileRun(
"function coat() { this.x = 6; }"
"createObj(coat);");
slot_value = feedback_vector->Get(FeedbackVectorSlot(0));
CHECK(slot_value->IsWeakCell() && !WeakCell::cast(slot_value)->cleared());
}
// Checks that the value returned by execution of the source is weak.
void CheckWeakness(const char* source) {
i::FLAG_stress_compaction = false;
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
LocalContext env;
v8::HandleScope scope(isolate);
v8::Persistent<v8::Object> garbage;
{
v8::HandleScope scope(isolate);
garbage.Reset(isolate, CompileRun(env.local(), source)
.ToLocalChecked()
->ToObject(env.local())
.ToLocalChecked());
}
weak_ic_cleared = false;
garbage.SetWeak(&garbage, &ClearWeakIC, v8::WeakCallbackType::kParameter);
Heap* heap = CcTest::i_isolate()->heap();
heap->CollectAllGarbage();
CHECK(weak_ic_cleared);
}
// Each of the following "weak IC" tests creates an IC that embeds a map with
// the prototype pointing to _proto_ and checks that the _proto_ dies on GC.
TEST(WeakMapInMonomorphicLoadIC) {
CheckWeakness("function loadIC(obj) {"
" return obj.name;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" loadIC(obj);"
" loadIC(obj);"
" loadIC(obj);"
" return proto;"
" })();");
}
TEST(WeakMapInPolymorphicLoadIC) {
CheckWeakness(
"function loadIC(obj) {"
" return obj.name;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" loadIC(obj);"
" loadIC(obj);"
" loadIC(obj);"
" var poly = Object.create(proto);"
" poly.x = true;"
" loadIC(poly);"
" return proto;"
" })();");
}
TEST(WeakMapInMonomorphicKeyedLoadIC) {
CheckWeakness("function keyedLoadIC(obj, field) {"
" return obj[field];"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" keyedLoadIC(obj, 'name');"
" keyedLoadIC(obj, 'name');"
" keyedLoadIC(obj, 'name');"
" return proto;"
" })();");
}
TEST(WeakMapInPolymorphicKeyedLoadIC) {
CheckWeakness(
"function keyedLoadIC(obj, field) {"
" return obj[field];"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" keyedLoadIC(obj, 'name');"
" keyedLoadIC(obj, 'name');"
" keyedLoadIC(obj, 'name');"
" var poly = Object.create(proto);"
" poly.x = true;"
" keyedLoadIC(poly, 'name');"
" return proto;"
" })();");
}
TEST(WeakMapInMonomorphicStoreIC) {
CheckWeakness("function storeIC(obj, value) {"
" obj.name = value;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" storeIC(obj, 'x');"
" storeIC(obj, 'x');"
" storeIC(obj, 'x');"
" return proto;"
" })();");
}
TEST(WeakMapInPolymorphicStoreIC) {
CheckWeakness(
"function storeIC(obj, value) {"
" obj.name = value;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" storeIC(obj, 'x');"
" storeIC(obj, 'x');"
" storeIC(obj, 'x');"
" var poly = Object.create(proto);"
" poly.x = true;"
" storeIC(poly, 'x');"
" return proto;"
" })();");
}
TEST(WeakMapInMonomorphicKeyedStoreIC) {
CheckWeakness("function keyedStoreIC(obj, field, value) {"
" obj[field] = value;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" keyedStoreIC(obj, 'x');"
" keyedStoreIC(obj, 'x');"
" keyedStoreIC(obj, 'x');"
" return proto;"
" })();");
}
TEST(WeakMapInPolymorphicKeyedStoreIC) {
CheckWeakness(
"function keyedStoreIC(obj, field, value) {"
" obj[field] = value;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" keyedStoreIC(obj, 'x');"
" keyedStoreIC(obj, 'x');"
" keyedStoreIC(obj, 'x');"
" var poly = Object.create(proto);"
" poly.x = true;"
" keyedStoreIC(poly, 'x');"
" return proto;"
" })();");
}
TEST(WeakMapInMonomorphicCompareNilIC) {
CheckWeakness("function compareNilIC(obj) {"
" return obj == null;"
"}"
" (function() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" compareNilIC(obj);"
" compareNilIC(obj);"
" compareNilIC(obj);"
" return proto;"
" })();");
}
Handle<JSFunction> GetFunctionByName(Isolate* isolate, const char* name) {
Handle<String> str = isolate->factory()->InternalizeUtf8String(name);
Handle<Object> obj =
Object::GetProperty(isolate->global_object(), str).ToHandleChecked();
return Handle<JSFunction>::cast(obj);
}
void CheckIC(Handle<JSFunction> function, Code::Kind kind, int slot_index,
InlineCacheState state) {
if (kind == Code::LOAD_IC || kind == Code::KEYED_LOAD_IC ||
kind == Code::CALL_IC) {
TypeFeedbackVector* vector = function->feedback_vector();
FeedbackVectorSlot slot(slot_index);
if (kind == Code::LOAD_IC) {
LoadICNexus nexus(vector, slot);
CHECK_EQ(nexus.StateFromFeedback(), state);
} else if (kind == Code::KEYED_LOAD_IC) {
KeyedLoadICNexus nexus(vector, slot);
CHECK_EQ(nexus.StateFromFeedback(), state);
} else if (kind == Code::CALL_IC) {
CallICNexus nexus(vector, slot);
CHECK_EQ(nexus.StateFromFeedback(), state);
}
} else {
Code* ic = FindFirstIC(function->code(), kind);
CHECK(ic->is_inline_cache_stub());
CHECK(!IC::ICUseVector(kind));
CHECK_EQ(state, IC::StateFromCode(ic));
}
}
TEST(MonomorphicStaysMonomorphicAfterGC) {
if (FLAG_always_opt) return;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function loadIC(obj) {"
" return obj.name;"
"}"
"function testIC() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" loadIC(obj);"
" loadIC(obj);"
" loadIC(obj);"
" return proto;"
"};");
Handle<JSFunction> loadIC = GetFunctionByName(isolate, "loadIC");
{
v8::HandleScope scope(CcTest::isolate());
CompileRun("(testIC())");
}
heap->CollectAllGarbage();
CheckIC(loadIC, Code::LOAD_IC, 0, MONOMORPHIC);
{
v8::HandleScope scope(CcTest::isolate());
CompileRun("(testIC())");
}
CheckIC(loadIC, Code::LOAD_IC, 0, MONOMORPHIC);
}
TEST(PolymorphicStaysPolymorphicAfterGC) {
if (FLAG_always_opt) return;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
v8::HandleScope scope(CcTest::isolate());
CompileRun(
"function loadIC(obj) {"
" return obj.name;"
"}"
"function testIC() {"
" var proto = {'name' : 'weak'};"
" var obj = Object.create(proto);"
" loadIC(obj);"
" loadIC(obj);"
" loadIC(obj);"
" var poly = Object.create(proto);"
" poly.x = true;"
" loadIC(poly);"
" return proto;"
"};");
Handle<JSFunction> loadIC = GetFunctionByName(isolate, "loadIC");
{
v8::HandleScope scope(CcTest::isolate());
CompileRun("(testIC())");
}
heap->CollectAllGarbage();
CheckIC(loadIC, Code::LOAD_IC, 0, POLYMORPHIC);
{
v8::HandleScope scope(CcTest::isolate());
CompileRun("(testIC())");
}
CheckIC(loadIC, Code::LOAD_IC, 0, POLYMORPHIC);
}
TEST(WeakCell) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
v8::internal::Factory* factory = isolate->factory();
HandleScope outer_scope(isolate);
Handle<WeakCell> weak_cell1;
{
HandleScope inner_scope(isolate);
Handle<HeapObject> value = factory->NewFixedArray(1, NOT_TENURED);
weak_cell1 = inner_scope.CloseAndEscape(factory->NewWeakCell(value));
}
Handle<FixedArray> survivor = factory->NewFixedArray(1, NOT_TENURED);
Handle<WeakCell> weak_cell2;
{
HandleScope inner_scope(isolate);
weak_cell2 = inner_scope.CloseAndEscape(factory->NewWeakCell(survivor));
}
CHECK(weak_cell1->value()->IsFixedArray());
CHECK_EQ(*survivor, weak_cell2->value());
heap->CollectGarbage(NEW_SPACE);
CHECK(weak_cell1->value()->IsFixedArray());
CHECK_EQ(*survivor, weak_cell2->value());
heap->CollectGarbage(NEW_SPACE);
CHECK(weak_cell1->value()->IsFixedArray());
CHECK_EQ(*survivor, weak_cell2->value());
heap->CollectAllAvailableGarbage();
CHECK(weak_cell1->cleared());
CHECK_EQ(*survivor, weak_cell2->value());
}
TEST(WeakCellsWithIncrementalMarking) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
v8::internal::Heap* heap = CcTest::heap();
v8::internal::Factory* factory = isolate->factory();
const int N = 16;
HandleScope outer_scope(isolate);
Handle<FixedArray> survivor = factory->NewFixedArray(1, NOT_TENURED);
Handle<WeakCell> weak_cells[N];
for (int i = 0; i < N; i++) {
HandleScope inner_scope(isolate);
Handle<HeapObject> value =
i == 0 ? survivor : factory->NewFixedArray(1, NOT_TENURED);
Handle<WeakCell> weak_cell = factory->NewWeakCell(value);
CHECK(weak_cell->value()->IsFixedArray());
IncrementalMarking* marking = heap->incremental_marking();
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
marking->Step(128, IncrementalMarking::NO_GC_VIA_STACK_GUARD);
heap->CollectGarbage(NEW_SPACE);
CHECK(weak_cell->value()->IsFixedArray());
weak_cells[i] = inner_scope.CloseAndEscape(weak_cell);
}
// Call collect all twice to make sure that we also cleared
// weak cells that were allocated on black pages.
heap->CollectAllGarbage();
heap->CollectAllGarbage();
CHECK_EQ(*survivor, weak_cells[0]->value());
for (int i = 1; i < N; i++) {
CHECK(weak_cells[i]->cleared());
}
}
#ifdef DEBUG
TEST(AddInstructionChangesNewSpacePromotion) {
i::FLAG_allow_natives_syntax = true;
i::FLAG_expose_gc = true;
i::FLAG_stress_compaction = true;
i::FLAG_gc_interval = 1000;
CcTest::InitializeVM();
if (!i::FLAG_allocation_site_pretenuring) return;
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
LocalContext env;
CompileRun(
"function add(a, b) {"
" return a + b;"
"}"
"add(1, 2);"
"add(\"a\", \"b\");"
"var oldSpaceObject;"
"gc();"
"function crash(x) {"
" var object = {a: null, b: null};"
" var result = add(1.5, x | 0);"
" object.a = result;"
" oldSpaceObject = object;"
" return object;"
"}"
"crash(1);"
"crash(1);"
"%OptimizeFunctionOnNextCall(crash);"
"crash(1);");
v8::Local<v8::Object> global = CcTest::global();
v8::Local<v8::Function> g = v8::Local<v8::Function>::Cast(
global->Get(env.local(), v8_str("crash")).ToLocalChecked());
v8::Local<v8::Value> args1[] = {v8_num(1)};
heap->DisableInlineAllocation();
heap->set_allocation_timeout(1);
g->Call(env.local(), global, 1, args1).ToLocalChecked();
heap->CollectAllGarbage();
}
void OnFatalErrorExpectOOM(const char* location, const char* message) {
// Exit with 0 if the location matches our expectation.
exit(strcmp(location, "CALL_AND_RETRY_LAST"));
}
TEST(CEntryStubOOM) {
i::FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
CcTest::isolate()->SetFatalErrorHandler(OnFatalErrorExpectOOM);
v8::Local<v8::Value> result = CompileRun(
"%SetFlags('--gc-interval=1');"
"var a = [];"
"a.__proto__ = [];"
"a.unshift(1)");
CHECK(result->IsNumber());
}
#endif // DEBUG
static void InterruptCallback357137(v8::Isolate* isolate, void* data) { }
static void RequestInterrupt(const v8::FunctionCallbackInfo<v8::Value>& args) {
CcTest::isolate()->RequestInterrupt(&InterruptCallback357137, NULL);
}
UNINITIALIZED_TEST(Regress538257) {
i::FLAG_manual_evacuation_candidates_selection = true;
v8::Isolate::CreateParams create_params;
// Set heap limits.
create_params.constraints.set_max_semi_space_size(1 * Page::kPageSize / MB);
create_params.constraints.set_max_old_space_size(6 * Page::kPageSize / MB);
create_params.array_buffer_allocator = CcTest::array_buffer_allocator();
v8::Isolate* isolate = v8::Isolate::New(create_params);
isolate->Enter();
{
i::Isolate* i_isolate = reinterpret_cast<i::Isolate*>(isolate);
Heap* heap = i_isolate->heap();
HandleScope handle_scope(i_isolate);
PagedSpace* old_space = heap->old_space();
const int kMaxObjects = 10000;
const int kFixedArrayLen = 512;
Handle<FixedArray> objects[kMaxObjects];
for (int i = 0; (i < kMaxObjects) &&
heap->CanExpandOldGeneration(old_space->AreaSize());
i++) {
objects[i] = i_isolate->factory()->NewFixedArray(kFixedArrayLen, TENURED);
Page::FromAddress(objects[i]->address())
->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
}
heap::SimulateFullSpace(old_space);
heap->CollectGarbage(OLD_SPACE);
// If we get this far, we've successfully aborted compaction. Any further
// allocations might trigger OOM.
}
isolate->Exit();
isolate->Dispose();
}
TEST(Regress357137) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope hscope(isolate);
v8::Local<v8::ObjectTemplate> global = v8::ObjectTemplate::New(isolate);
global->Set(
v8::String::NewFromUtf8(isolate, "interrupt", v8::NewStringType::kNormal)
.ToLocalChecked(),
v8::FunctionTemplate::New(isolate, RequestInterrupt));
v8::Local<v8::Context> context = v8::Context::New(isolate, NULL, global);
CHECK(!context.IsEmpty());
v8::Context::Scope cscope(context);
v8::Local<v8::Value> result = CompileRun(
"var locals = '';"
"for (var i = 0; i < 512; i++) locals += 'var v' + i + '= 42;';"
"eval('function f() {' + locals + 'return function() { return v0; }; }');"
"interrupt();" // This triggers a fake stack overflow in f.
"f()()");
CHECK_EQ(42.0, result->ToNumber(context).ToLocalChecked()->Value());
}
TEST(Regress507979) {
const int kFixedArrayLen = 10;
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
HandleScope handle_scope(isolate);
Handle<FixedArray> o1 = isolate->factory()->NewFixedArray(kFixedArrayLen);
Handle<FixedArray> o2 = isolate->factory()->NewFixedArray(kFixedArrayLen);
CHECK(heap->InNewSpace(*o1));
CHECK(heap->InNewSpace(*o2));
HeapIterator it(heap, i::HeapIterator::kFilterUnreachable);
// Replace parts of an object placed before a live object with a filler. This
// way the filler object shares the mark bits with the following live object.
o1->Shrink(kFixedArrayLen - 1);
for (HeapObject* obj = it.next(); obj != NULL; obj = it.next()) {
// Let's not optimize the loop away.
CHECK(obj->address() != nullptr);
}
}
UNINITIALIZED_TEST(PromotionQueue) {
i::FLAG_expose_gc = true;
i::FLAG_max_semi_space_size = 2 * Page::kPageSize / MB;
i::FLAG_min_semi_space_size = i::FLAG_max_semi_space_size;
v8::Isolate::CreateParams create_params;
create_params.array_buffer_allocator = CcTest::array_buffer_allocator();
v8::Isolate* isolate = v8::Isolate::New(create_params);
i::Isolate* i_isolate = reinterpret_cast<i::Isolate*>(isolate);
{
v8::Isolate::Scope isolate_scope(isolate);
v8::HandleScope handle_scope(isolate);
v8::Context::New(isolate)->Enter();
Heap* heap = i_isolate->heap();
NewSpace* new_space = heap->new_space();
// In this test we will try to overwrite the promotion queue which is at the
// end of to-space. To actually make that possible, we need at least two
// semi-space pages and take advantage of fragmentation.
// (1) Use a semi-space consisting of two pages.
// (2) Create a few small long living objects and call the scavenger to
// move them to the other semi-space.
// (3) Create a huge object, i.e., remainder of first semi-space page and
// create another huge object which should be of maximum allocatable memory
// size of the second semi-space page.
// (4) Call the scavenger again.
// What will happen is: the scavenger will promote the objects created in
// (2) and will create promotion queue entries at the end of the second
// semi-space page during the next scavenge when it promotes the objects to
// the old generation. The first allocation of (3) will fill up the first
// semi-space page. The second allocation in (3) will not fit into the
// first semi-space page, but it will overwrite the promotion queue which
// are in the second semi-space page. If the right guards are in place, the
// promotion queue will be evacuated in that case.
CHECK(new_space->IsAtMaximumCapacity());
CHECK(i::FLAG_min_semi_space_size * MB == new_space->TotalCapacity());
// Call the scavenger two times to get an empty new space
heap->CollectGarbage(NEW_SPACE);
heap->CollectGarbage(NEW_SPACE);
// First create a few objects which will survive a scavenge, and will get
// promoted to the old generation later on. These objects will create
// promotion queue entries at the end of the second semi-space page.
const int number_handles = 12;
Handle<FixedArray> handles[number_handles];
for (int i = 0; i < number_handles; i++) {
handles[i] = i_isolate->factory()->NewFixedArray(1, NOT_TENURED);
}
heap->CollectGarbage(NEW_SPACE);
CHECK(i::FLAG_min_semi_space_size * MB == new_space->TotalCapacity());
// Fill-up the first semi-space page.
heap::FillUpOnePage(new_space);
// Create a small object to initialize the bump pointer on the second
// semi-space page.
Handle<FixedArray> small =
i_isolate->factory()->NewFixedArray(1, NOT_TENURED);
CHECK(heap->InNewSpace(*small));
// Fill-up the second semi-space page.
heap::FillUpOnePage(new_space);
// This scavenge will corrupt memory if the promotion queue is not
// evacuated.
heap->CollectGarbage(NEW_SPACE);
}
isolate->Dispose();
}
TEST(Regress388880) {
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
Handle<Map> map1 = Map::Create(isolate, 1);
Handle<String> name = factory->NewStringFromStaticChars("foo");
name = factory->InternalizeString(name);
Handle<Map> map2 =
Map::CopyWithField(map1, name, FieldType::Any(isolate), NONE,
Representation::Tagged(), OMIT_TRANSITION)
.ToHandleChecked();
int desired_offset = Page::kPageSize - map1->instance_size();
// Allocate padding objects in old pointer space so, that object allocated
// afterwards would end at the end of the page.
heap::SimulateFullSpace(heap->old_space());
int padding_size = desired_offset - Page::kObjectStartOffset;
heap::CreatePadding(heap, padding_size, TENURED);
Handle<JSObject> o = factory->NewJSObjectFromMap(map1, TENURED);
o->set_properties(*factory->empty_fixed_array());
// Ensure that the object allocated where we need it.
Page* page = Page::FromAddress(o->address());
CHECK_EQ(desired_offset, page->Offset(o->address()));
// Now we have an object right at the end of the page.
// Enable incremental marking to trigger actions in Heap::AdjustLiveBytes()
// that would cause crash.
IncrementalMarking* marking = CcTest::heap()->incremental_marking();
marking->Stop();
CcTest::heap()->StartIncrementalMarking();
CHECK(marking->IsMarking());
// Now everything is set up for crashing in JSObject::MigrateFastToFast()
// when it calls heap->AdjustLiveBytes(...).
JSObject::MigrateToMap(o, map2);
}
TEST(Regress3631) {
i::FLAG_expose_gc = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
IncrementalMarking* marking = CcTest::heap()->incremental_marking();
v8::Local<v8::Value> result = CompileRun(
"var weak_map = new WeakMap();"
"var future_keys = [];"
"for (var i = 0; i < 50; i++) {"
" var key = {'k' : i + 0.1};"
" weak_map.set(key, 1);"
" future_keys.push({'x' : i + 0.2});"
"}"
"weak_map");
if (marking->IsStopped()) {
CcTest::heap()->StartIncrementalMarking();
}
// Incrementally mark the backing store.
Handle<JSReceiver> obj =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(result));
Handle<JSWeakCollection> weak_map(reinterpret_cast<JSWeakCollection*>(*obj));
while (!Marking::IsBlack(
ObjectMarking::MarkBitFrom(HeapObject::cast(weak_map->table()))) &&
!marking->IsStopped()) {
marking->Step(MB, IncrementalMarking::NO_GC_VIA_STACK_GUARD);
}
// Stash the backing store in a handle.
Handle<Object> save(weak_map->table(), isolate);
// The following line will update the backing store.
CompileRun(
"for (var i = 0; i < 50; i++) {"
" weak_map.set(future_keys[i], i);"
"}");
heap->incremental_marking()->set_should_hurry(true);
heap->CollectGarbage(OLD_SPACE);
}
TEST(Regress442710) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope sc(isolate);
Handle<JSGlobalObject> global(
CcTest::i_isolate()->context()->global_object());
Handle<JSArray> array = factory->NewJSArray(2);
Handle<String> name = factory->InternalizeUtf8String("testArray");
JSReceiver::SetProperty(global, name, array, SLOPPY).Check();
CompileRun("testArray[0] = 1; testArray[1] = 2; testArray.shift();");
heap->CollectGarbage(OLD_SPACE);
}
HEAP_TEST(NumberStringCacheSize) {
// Test that the number-string cache has not been resized in the snapshot.
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
if (!isolate->snapshot_available()) return;
Heap* heap = isolate->heap();
CHECK_EQ(Heap::kInitialNumberStringCacheSize * 2,
heap->number_string_cache()->length());
}
TEST(Regress3877) {
CcTest::InitializeVM();
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Factory* factory = isolate->factory();
HandleScope scope(isolate);
CompileRun("function cls() { this.x = 10; }");
Handle<WeakCell> weak_prototype;
{
HandleScope inner_scope(isolate);
v8::Local<v8::Value> result = CompileRun("cls.prototype");
Handle<JSReceiver> proto =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(result));
weak_prototype = inner_scope.CloseAndEscape(factory->NewWeakCell(proto));
}
CHECK(!weak_prototype->cleared());
CompileRun(
"var a = { };"
"a.x = new cls();"
"cls.prototype = null;");
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
// The map of a.x keeps prototype alive
CHECK(!weak_prototype->cleared());
// Change the map of a.x and make the previous map garbage collectable.
CompileRun("a.x.__proto__ = {};");
for (int i = 0; i < 4; i++) {
heap->CollectAllGarbage();
}
CHECK(weak_prototype->cleared());
}
Handle<WeakCell> AddRetainedMap(Isolate* isolate, Heap* heap) {
HandleScope inner_scope(isolate);
Handle<Map> map = Map::Create(isolate, 1);
v8::Local<v8::Value> result =
CompileRun("(function () { return {x : 10}; })();");
Handle<JSReceiver> proto =
v8::Utils::OpenHandle(*v8::Local<v8::Object>::Cast(result));
Map::SetPrototype(map, proto);
heap->AddRetainedMap(map);
return inner_scope.CloseAndEscape(Map::WeakCellForMap(map));
}
void CheckMapRetainingFor(int n) {
FLAG_retain_maps_for_n_gc = n;
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
Handle<WeakCell> weak_cell = AddRetainedMap(isolate, heap);
CHECK(!weak_cell->cleared());
for (int i = 0; i < n; i++) {
heap::SimulateIncrementalMarking(heap);
heap->CollectGarbage(OLD_SPACE);
}
CHECK(!weak_cell->cleared());
heap::SimulateIncrementalMarking(heap);
heap->CollectGarbage(OLD_SPACE);
CHECK(weak_cell->cleared());
}
TEST(MapRetaining) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
CheckMapRetainingFor(FLAG_retain_maps_for_n_gc);
CheckMapRetainingFor(0);
CheckMapRetainingFor(1);
CheckMapRetainingFor(7);
}
TEST(RegressArrayListGC) {
FLAG_retain_maps_for_n_gc = 1;
FLAG_incremental_marking = 0;
FLAG_gc_global = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
AddRetainedMap(isolate, heap);
Handle<Map> map = Map::Create(isolate, 1);
heap->CollectGarbage(OLD_SPACE);
// Force GC in old space on next addition of retained map.
Map::WeakCellForMap(map);
heap::SimulateFullSpace(CcTest::heap()->new_space());
for (int i = 0; i < 10; i++) {
heap->AddRetainedMap(map);
}
heap->CollectGarbage(OLD_SPACE);
}
#ifdef DEBUG
TEST(PathTracer) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::Local<v8::Value> result = CompileRun("'abc'");
Handle<Object> o = v8::Utils::OpenHandle(*result);
CcTest::i_isolate()->heap()->TracePathToObject(*o);
}
#endif // DEBUG
TEST(WritableVsImmortalRoots) {
for (int i = 0; i < Heap::kStrongRootListLength; ++i) {
Heap::RootListIndex root_index = static_cast<Heap::RootListIndex>(i);
bool writable = Heap::RootCanBeWrittenAfterInitialization(root_index);
bool immortal = Heap::RootIsImmortalImmovable(root_index);
// A root value can be writable, immortal, or neither, but not both.
CHECK(!immortal || !writable);
}
}
static void TestRightTrimFixedTypedArray(i::ExternalArrayType type,
int initial_length,
int elements_to_trim) {
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
Heap* heap = isolate->heap();
Handle<FixedTypedArrayBase> array =
factory->NewFixedTypedArray(initial_length, type, true);
int old_size = array->size();
heap->RightTrimFixedArray<Heap::CONCURRENT_TO_SWEEPER>(*array,
elements_to_trim);
// Check that free space filler is at the right place and did not smash the
// array header.
CHECK(array->IsFixedArrayBase());
CHECK_EQ(initial_length - elements_to_trim, array->length());
int new_size = array->size();
if (new_size != old_size) {
// Free space filler should be created in this case.
Address next_obj_address = array->address() + array->size();
CHECK(HeapObject::FromAddress(next_obj_address)->IsFiller());
}
heap->CollectAllAvailableGarbage();
}
TEST(Regress472513) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
// The combination of type/initial_length/elements_to_trim triggered
// typed array header smashing with free space filler (crbug/472513).
// 64-bit cases.
TestRightTrimFixedTypedArray(i::kExternalUint8Array, 32, 6);
TestRightTrimFixedTypedArray(i::kExternalUint8Array, 32 - 7, 6);
TestRightTrimFixedTypedArray(i::kExternalUint16Array, 16, 6);
TestRightTrimFixedTypedArray(i::kExternalUint16Array, 16 - 3, 6);
TestRightTrimFixedTypedArray(i::kExternalUint32Array, 8, 6);
TestRightTrimFixedTypedArray(i::kExternalUint32Array, 8 - 1, 6);
// 32-bit cases.
TestRightTrimFixedTypedArray(i::kExternalUint8Array, 16, 3);
TestRightTrimFixedTypedArray(i::kExternalUint8Array, 16 - 3, 3);
TestRightTrimFixedTypedArray(i::kExternalUint16Array, 8, 3);
TestRightTrimFixedTypedArray(i::kExternalUint16Array, 8 - 1, 3);
TestRightTrimFixedTypedArray(i::kExternalUint32Array, 4, 3);
}
TEST(WeakFixedArray) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Handle<HeapNumber> number = CcTest::i_isolate()->factory()->NewHeapNumber(1);
Handle<WeakFixedArray> array = WeakFixedArray::Add(Handle<Object>(), number);
array->Remove(number);
array->Compact<WeakFixedArray::NullCallback>();
WeakFixedArray::Add(array, number);
}
TEST(PreprocessStackTrace) {
// Do not automatically trigger early GC.
FLAG_gc_interval = -1;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
v8::TryCatch try_catch(CcTest::isolate());
CompileRun("throw new Error();");
CHECK(try_catch.HasCaught());
Isolate* isolate = CcTest::i_isolate();
Handle<Object> exception = v8::Utils::OpenHandle(*try_catch.Exception());
Handle<Name> key = isolate->factory()->stack_trace_symbol();
Handle<Object> stack_trace =
Object::GetProperty(exception, key).ToHandleChecked();
Handle<Object> code =
Object::GetElement(isolate, stack_trace, 3).ToHandleChecked();
CHECK(code->IsCode());
isolate->heap()->CollectAllAvailableGarbage("stack trace preprocessing");
Handle<Object> pos =
Object::GetElement(isolate, stack_trace, 3).ToHandleChecked();
CHECK(pos->IsSmi());
Handle<JSArray> stack_trace_array = Handle<JSArray>::cast(stack_trace);
int array_length = Smi::cast(stack_trace_array->length())->value();
for (int i = 0; i < array_length; i++) {
Handle<Object> element =
Object::GetElement(isolate, stack_trace, i).ToHandleChecked();
CHECK(!element->IsCode());
}
}
static bool utils_has_been_collected = false;
static void UtilsHasBeenCollected(
const v8::WeakCallbackInfo<v8::Persistent<v8::Object>>& data) {
utils_has_been_collected = true;
data.GetParameter()->Reset();
}
TEST(BootstrappingExports) {
// Expose utils object and delete it to observe that it is indeed
// being garbage-collected.
FLAG_expose_natives_as = "utils";
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
LocalContext env;
if (Snapshot::HasContextSnapshot(CcTest::i_isolate(), 0)) return;
utils_has_been_collected = false;
v8::Persistent<v8::Object> utils;
{
v8::HandleScope scope(isolate);
v8::Local<v8::String> name = v8_str("utils");
utils.Reset(isolate, CcTest::global()
->Get(env.local(), name)
.ToLocalChecked()
->ToObject(env.local())
.ToLocalChecked());
CHECK(CcTest::global()->Delete(env.local(), name).FromJust());
}
utils.SetWeak(&utils, UtilsHasBeenCollected,
v8::WeakCallbackType::kParameter);
CcTest::heap()->CollectAllAvailableGarbage("fire weak callbacks");
CHECK(utils_has_been_collected);
}
TEST(Regress1878) {
FLAG_allow_natives_syntax = true;
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope scope(isolate);
v8::Local<v8::Function> constructor = v8::Utils::CallableToLocal(
CcTest::i_isolate()->internal_array_function());
LocalContext env;
CHECK(CcTest::global()
->Set(env.local(), v8_str("InternalArray"), constructor)
.FromJust());
v8::TryCatch try_catch(isolate);
CompileRun(
"var a = Array();"
"for (var i = 0; i < 1000; i++) {"
" var ai = new InternalArray(10000);"
" if (%HaveSameMap(ai, a)) throw Error();"
" if (!%HasFastObjectElements(ai)) throw Error();"
"}"
"for (var i = 0; i < 1000; i++) {"
" var ai = new InternalArray(10000);"
" if (%HaveSameMap(ai, a)) throw Error();"
" if (!%HasFastObjectElements(ai)) throw Error();"
"}");
CHECK(!try_catch.HasCaught());
}
void AllocateInSpace(Isolate* isolate, size_t bytes, AllocationSpace space) {
CHECK(bytes >= FixedArray::kHeaderSize);
CHECK(bytes % kPointerSize == 0);
Factory* factory = isolate->factory();
HandleScope scope(isolate);
AlwaysAllocateScope always_allocate(isolate);
int elements =
static_cast<int>((bytes - FixedArray::kHeaderSize) / kPointerSize);
Handle<FixedArray> array = factory->NewFixedArray(
elements, space == NEW_SPACE ? NOT_TENURED : TENURED);
CHECK((space == NEW_SPACE) == isolate->heap()->InNewSpace(*array));
CHECK_EQ(bytes, static_cast<size_t>(array->Size()));
}
TEST(NewSpaceAllocationCounter) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
size_t counter1 = heap->NewSpaceAllocationCounter();
heap->CollectGarbage(NEW_SPACE);
const size_t kSize = 1024;
AllocateInSpace(isolate, kSize, NEW_SPACE);
size_t counter2 = heap->NewSpaceAllocationCounter();
CHECK_EQ(kSize, counter2 - counter1);
heap->CollectGarbage(NEW_SPACE);
size_t counter3 = heap->NewSpaceAllocationCounter();
CHECK_EQ(0U, counter3 - counter2);
// Test counter overflow.
size_t max_counter = -1;
heap->set_new_space_allocation_counter(max_counter - 10 * kSize);
size_t start = heap->NewSpaceAllocationCounter();
for (int i = 0; i < 20; i++) {
AllocateInSpace(isolate, kSize, NEW_SPACE);
size_t counter = heap->NewSpaceAllocationCounter();
CHECK_EQ(kSize, counter - start);
start = counter;
}
}
TEST(OldSpaceAllocationCounter) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = isolate->heap();
size_t counter1 = heap->OldGenerationAllocationCounter();
heap->CollectGarbage(NEW_SPACE);
heap->CollectGarbage(NEW_SPACE);
const size_t kSize = 1024;
AllocateInSpace(isolate, kSize, OLD_SPACE);
size_t counter2 = heap->OldGenerationAllocationCounter();
// TODO(ulan): replace all CHECK_LE with CHECK_EQ after v8:4148 is fixed.
CHECK_LE(kSize, counter2 - counter1);
heap->CollectGarbage(NEW_SPACE);
size_t counter3 = heap->OldGenerationAllocationCounter();
CHECK_EQ(0u, counter3 - counter2);
AllocateInSpace(isolate, kSize, OLD_SPACE);
heap->CollectGarbage(OLD_SPACE);
size_t counter4 = heap->OldGenerationAllocationCounter();
CHECK_LE(kSize, counter4 - counter3);
// Test counter overflow.
size_t max_counter = -1;
heap->set_old_generation_allocation_counter(max_counter - 10 * kSize);
size_t start = heap->OldGenerationAllocationCounter();
for (int i = 0; i < 20; i++) {
AllocateInSpace(isolate, kSize, OLD_SPACE);
size_t counter = heap->OldGenerationAllocationCounter();
CHECK_LE(kSize, counter - start);
start = counter;
}
}
static void CheckLeak(const v8::FunctionCallbackInfo<v8::Value>& args) {
Isolate* isolate = CcTest::i_isolate();
Object* message =
*reinterpret_cast<Object**>(isolate->pending_message_obj_address());
CHECK(message->IsTheHole(isolate));
}
TEST(MessageObjectLeak) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope scope(isolate);
v8::Local<v8::ObjectTemplate> global = v8::ObjectTemplate::New(isolate);
global->Set(
v8::String::NewFromUtf8(isolate, "check", v8::NewStringType::kNormal)
.ToLocalChecked(),
v8::FunctionTemplate::New(isolate, CheckLeak));
v8::Local<v8::Context> context = v8::Context::New(isolate, NULL, global);
v8::Context::Scope cscope(context);
const char* test =
"try {"
" throw 'message 1';"
"} catch (e) {"
"}"
"check();"
"L: try {"
" throw 'message 2';"
"} finally {"
" break L;"
"}"
"check();";
CompileRun(test);
const char* flag = "--turbo-filter=*";
FlagList::SetFlagsFromString(flag, StrLength(flag));
FLAG_always_opt = true;
CompileRun(test);
}
static void CheckEqualSharedFunctionInfos(
const v8::FunctionCallbackInfo<v8::Value>& args) {
Handle<Object> obj1 = v8::Utils::OpenHandle(*args[0]);
Handle<Object> obj2 = v8::Utils::OpenHandle(*args[1]);
Handle<JSFunction> fun1 = Handle<JSFunction>::cast(obj1);
Handle<JSFunction> fun2 = Handle<JSFunction>::cast(obj2);
CHECK(fun1->shared() == fun2->shared());
}
static void RemoveCodeAndGC(const v8::FunctionCallbackInfo<v8::Value>& args) {
Isolate* isolate = CcTest::i_isolate();
Handle<Object> obj = v8::Utils::OpenHandle(*args[0]);
Handle<JSFunction> fun = Handle<JSFunction>::cast(obj);
fun->ReplaceCode(*isolate->builtins()->CompileLazy());
fun->shared()->ReplaceCode(*isolate->builtins()->CompileLazy());
fun->shared()->ClearBytecodeArray(); // Bytecode is code too.
isolate->heap()->CollectAllAvailableGarbage("remove code and gc");
}
TEST(CanonicalSharedFunctionInfo) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope scope(isolate);
v8::Local<v8::ObjectTemplate> global = v8::ObjectTemplate::New(isolate);
global->Set(isolate, "check", v8::FunctionTemplate::New(
isolate, CheckEqualSharedFunctionInfos));
global->Set(isolate, "remove",
v8::FunctionTemplate::New(isolate, RemoveCodeAndGC));
v8::Local<v8::Context> context = v8::Context::New(isolate, NULL, global);
v8::Context::Scope cscope(context);
CompileRun(
"function f() { return function g() {}; }"
"var g1 = f();"
"remove(f);"
"var g2 = f();"
"check(g1, g2);");
CompileRun(
"function f() { return (function() { return function g() {}; })(); }"
"var g1 = f();"
"remove(f);"
"var g2 = f();"
"check(g1, g2);");
}
TEST(RemoveCodeFromSharedFunctionInfoButNotFromClosure) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope scope(isolate);
v8::Local<v8::ObjectTemplate> global = v8::ObjectTemplate::New(isolate);
global->Set(isolate, "check", v8::FunctionTemplate::New(
isolate, CheckEqualSharedFunctionInfos));
global->Set(isolate, "remove",
v8::FunctionTemplate::New(isolate, RemoveCodeAndGC));
v8::Local<v8::Context> context = v8::Context::New(isolate, NULL, global);
v8::Context::Scope cscope(context);
CompileRun(
"function f() { return function g() {}; }"
"var g1 = f();"
"var g2 = f();"
"check(g1, g2);"
"g1();"
"g2();"
"remove(g1);"
"g2();"
"check(g1, g2);");
}
TEST(ContextMeasure) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
LocalContext context;
int size_upper_limit = 0;
int count_upper_limit = 0;
HeapIterator it(CcTest::heap());
for (HeapObject* obj = it.next(); obj != NULL; obj = it.next()) {
size_upper_limit += obj->Size();
count_upper_limit++;
}
ContextMeasure measure(*isolate->native_context());
PrintF("Context size : %d bytes\n", measure.Size());
PrintF("Context object count: %d\n", measure.Count());
CHECK_LE(1000, measure.Count());
CHECK_LE(50000, measure.Size());
CHECK_LE(measure.Count(), count_upper_limit);
CHECK_LE(measure.Size(), size_upper_limit);
}
TEST(ScriptIterator) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = CcTest::heap();
LocalContext context;
heap->CollectAllGarbage();
int script_count = 0;
{
HeapIterator it(heap);
for (HeapObject* obj = it.next(); obj != NULL; obj = it.next()) {
if (obj->IsScript()) script_count++;
}
}
{
Script::Iterator iterator(isolate);
while (iterator.Next()) script_count--;
}
CHECK_EQ(0, script_count);
}
TEST(SharedFunctionInfoIterator) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Isolate* isolate = CcTest::i_isolate();
Heap* heap = CcTest::heap();
LocalContext context;
heap->CollectAllGarbage();
heap->CollectAllGarbage();
int sfi_count = 0;
{
HeapIterator it(heap);
for (HeapObject* obj = it.next(); obj != NULL; obj = it.next()) {
if (!obj->IsSharedFunctionInfo()) continue;
sfi_count++;
}
}
{
SharedFunctionInfo::Iterator iterator(isolate);
while (iterator.Next()) sfi_count--;
}
CHECK_EQ(0, sfi_count);
}
template <typename T>
static UniqueId MakeUniqueId(const Persistent<T>& p) {
return UniqueId(reinterpret_cast<uintptr_t>(*v8::Utils::OpenPersistent(p)));
}
TEST(Regress519319) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
v8::HandleScope scope(isolate);
Heap* heap = CcTest::heap();
LocalContext context;
v8::Persistent<Value> parent;
v8::Persistent<Value> child;
parent.Reset(isolate, v8::Object::New(isolate));
child.Reset(isolate, v8::Object::New(isolate));
heap::SimulateFullSpace(heap->old_space());
heap->CollectGarbage(OLD_SPACE);
{
UniqueId id = MakeUniqueId(parent);
isolate->SetObjectGroupId(parent, id);
isolate->SetReferenceFromGroup(id, child);
}
// The CollectGarbage call above starts sweeper threads.
// The crash will happen if the following two functions
// are called before sweeping finishes.
heap->StartIncrementalMarking();
heap->FinalizeIncrementalMarkingIfComplete("test");
}
HEAP_TEST(TestMemoryReducerSampleJsCalls) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = CcTest::i_isolate();
MemoryReducer* memory_reducer = heap->memory_reducer_;
memory_reducer->SampleAndGetJsCallsPerMs(0);
isolate->IncrementJsCallsFromApiCounter();
isolate->IncrementJsCallsFromApiCounter();
isolate->IncrementJsCallsFromApiCounter();
double calls_per_ms = memory_reducer->SampleAndGetJsCallsPerMs(1);
CheckDoubleEquals(3, calls_per_ms);
calls_per_ms = memory_reducer->SampleAndGetJsCallsPerMs(2);
CheckDoubleEquals(0, calls_per_ms);
isolate->IncrementJsCallsFromApiCounter();
isolate->IncrementJsCallsFromApiCounter();
isolate->IncrementJsCallsFromApiCounter();
isolate->IncrementJsCallsFromApiCounter();
calls_per_ms = memory_reducer->SampleAndGetJsCallsPerMs(4);
CheckDoubleEquals(2, calls_per_ms);
}
HEAP_TEST(Regress587004) {
FLAG_concurrent_sweeping = false;
#ifdef VERIFY_HEAP
FLAG_verify_heap = false;
#endif
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
const int N = (Page::kMaxRegularHeapObjectSize - FixedArray::kHeaderSize) /
kPointerSize;
Handle<FixedArray> array = factory->NewFixedArray(N, TENURED);
CHECK(heap->old_space()->Contains(*array));
Handle<Object> number = factory->NewHeapNumber(1.0);
CHECK(heap->InNewSpace(*number));
for (int i = 0; i < N; i++) {
array->set(i, *number);
}
heap->CollectGarbage(OLD_SPACE);
heap::SimulateFullSpace(heap->old_space());
heap->RightTrimFixedArray<Heap::CONCURRENT_TO_SWEEPER>(*array, N - 1);
heap->mark_compact_collector()->EnsureSweepingCompleted();
ByteArray* byte_array;
const int M = 256;
// Don't allow old space expansion. The test works without this flag too,
// but becomes very slow.
heap->set_force_oom(true);
while (heap->AllocateByteArray(M, TENURED).To(&byte_array)) {
for (int j = 0; j < M; j++) {
byte_array->set(j, 0x31);
}
}
// Re-enable old space expansion to avoid OOM crash.
heap->set_force_oom(false);
heap->CollectGarbage(NEW_SPACE);
}
HEAP_TEST(Regress589413) {
FLAG_stress_compaction = true;
FLAG_manual_evacuation_candidates_selection = true;
FLAG_parallel_compaction = false;
FLAG_concurrent_sweeping = false;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
// Get the heap in clean state.
heap->CollectGarbage(OLD_SPACE);
heap->CollectGarbage(OLD_SPACE);
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
// Fill the new space with byte arrays with elements looking like pointers.
const int M = 256;
ByteArray* byte_array;
while (heap->AllocateByteArray(M).To(&byte_array)) {
for (int j = 0; j < M; j++) {
byte_array->set(j, 0x31);
}
// Add the array in root set.
handle(byte_array);
}
// Make sure the byte arrays will be promoted on the next GC.
heap->CollectGarbage(NEW_SPACE);
// This number is close to large free list category threshold.
const int N = 0x3eee;
{
std::vector<FixedArray*> arrays;
std::set<Page*> pages;
FixedArray* array;
// Fill all pages with fixed arrays.
heap->set_force_oom(true);
while (heap->AllocateFixedArray(N, TENURED).To(&array)) {
arrays.push_back(array);
pages.insert(Page::FromAddress(array->address()));
// Add the array in root set.
handle(array);
}
// Expand and full one complete page with fixed arrays.
heap->set_force_oom(false);
while (heap->AllocateFixedArray(N, TENURED).To(&array)) {
arrays.push_back(array);
pages.insert(Page::FromAddress(array->address()));
// Add the array in root set.
handle(array);
// Do not expand anymore.
heap->set_force_oom(true);
}
// Expand and mark the new page as evacuation candidate.
heap->set_force_oom(false);
{
AlwaysAllocateScope always_allocate(isolate);
Handle<HeapObject> ec_obj = factory->NewFixedArray(5000, TENURED);
Page* ec_page = Page::FromAddress(ec_obj->address());
ec_page->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
// Make all arrays point to evacuation candidate so that
// slots are recorded for them.
for (size_t j = 0; j < arrays.size(); j++) {
array = arrays[j];
for (int i = 0; i < N; i++) {
array->set(i, *ec_obj);
}
}
}
heap::SimulateIncrementalMarking(heap);
for (size_t j = 0; j < arrays.size(); j++) {
heap->RightTrimFixedArray<Heap::CONCURRENT_TO_SWEEPER>(arrays[j], N - 1);
}
}
// Force allocation from the free list.
heap->set_force_oom(true);
heap->CollectGarbage(OLD_SPACE);
}
TEST(Regress598319) {
// This test ensures that no white objects can cross the progress bar of large
// objects during incremental marking. It checks this by using Shift() during
// incremental marking.
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
const int kNumberOfObjects = Page::kMaxRegularHeapObjectSize / kPointerSize;
struct Arr {
Arr(Isolate* isolate, int number_of_objects) {
root = isolate->factory()->NewFixedArray(1, TENURED);
{
// Temporary scope to avoid getting any other objects into the root set.
v8::HandleScope scope(CcTest::isolate());
Handle<FixedArray> tmp =
isolate->factory()->NewFixedArray(number_of_objects);
root->set(0, *tmp);
for (int i = 0; i < get()->length(); i++) {
tmp = isolate->factory()->NewFixedArray(100, TENURED);
get()->set(i, *tmp);
}
}
}
FixedArray* get() { return FixedArray::cast(root->get(0)); }
Handle<FixedArray> root;
} arr(isolate, kNumberOfObjects);
CHECK_EQ(arr.get()->length(), kNumberOfObjects);
CHECK(heap->lo_space()->Contains(arr.get()));
LargePage* page = heap->lo_space()->FindPage(arr.get()->address());
CHECK_NOT_NULL(page);
// GC to cleanup state
heap->CollectGarbage(OLD_SPACE);
MarkCompactCollector* collector = heap->mark_compact_collector();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(heap->lo_space()->Contains(arr.get()));
CHECK(Marking::IsWhite(ObjectMarking::MarkBitFrom(arr.get())));
for (int i = 0; i < arr.get()->length(); i++) {
CHECK(Marking::IsWhite(
ObjectMarking::MarkBitFrom(HeapObject::cast(arr.get()->get(i)))));
}
// Start incremental marking.
IncrementalMarking* marking = heap->incremental_marking();
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
// Check that we have not marked the interesting array during root scanning.
for (int i = 0; i < arr.get()->length(); i++) {
CHECK(Marking::IsWhite(
ObjectMarking::MarkBitFrom(HeapObject::cast(arr.get()->get(i)))));
}
// Now we search for a state where we are in incremental marking and have
// only partially marked the large object.
while (!marking->IsComplete()) {
marking->Step(i::KB, i::IncrementalMarking::NO_GC_VIA_STACK_GUARD);
if (page->IsFlagSet(Page::HAS_PROGRESS_BAR) && page->progress_bar() > 0) {
CHECK_NE(page->progress_bar(), arr.get()->Size());
{
// Shift by 1, effectively moving one white object across the progress
// bar, meaning that we will miss marking it.
v8::HandleScope scope(CcTest::isolate());
Handle<JSArray> js_array = isolate->factory()->NewJSArrayWithElements(
Handle<FixedArray>(arr.get()));
js_array->GetElementsAccessor()->Shift(js_array);
}
break;
}
}
// Finish marking with bigger steps to speed up test.
while (!marking->IsComplete()) {
marking->Step(10 * i::MB, i::IncrementalMarking::NO_GC_VIA_STACK_GUARD);
if (marking->IsReadyToOverApproximateWeakClosure()) {
marking->FinalizeIncrementally();
}
}
CHECK(marking->IsComplete());
// All objects need to be black after marking. If a white object crossed the
// progress bar, we would fail here.
for (int i = 0; i < arr.get()->length(); i++) {
CHECK(Marking::IsBlack(
ObjectMarking::MarkBitFrom(HeapObject::cast(arr.get()->get(i)))));
}
}
TEST(Regress609761) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
intptr_t size_before = heap->SizeOfObjects();
Handle<FixedArray> array = isolate->factory()->NewFixedArray(200000);
array->Shrink(1);
intptr_t size_after = heap->SizeOfObjects();
CHECK_EQ(size_after, size_before + array->Size());
}
TEST(Regress615489) {
FLAG_black_allocation = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
heap->CollectAllGarbage();
i::MarkCompactCollector* collector = heap->mark_compact_collector();
i::IncrementalMarking* marking = heap->incremental_marking();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
marking->StartBlackAllocationForTesting();
{
AlwaysAllocateScope always_allocate(CcTest::i_isolate());
v8::HandleScope inner(CcTest::isolate());
isolate->factory()->NewFixedArray(500, TENURED)->Size();
}
while (!marking->IsComplete()) {
marking->Step(i::MB, i::IncrementalMarking::NO_GC_VIA_STACK_GUARD);
if (marking->IsReadyToOverApproximateWeakClosure()) {
marking->FinalizeIncrementally();
}
}
CHECK(marking->IsComplete());
intptr_t size_before = heap->SizeOfObjects();
CcTest::heap()->CollectAllGarbage();
intptr_t size_after = heap->SizeOfObjects();
// Live size does not increase after garbage collection.
CHECK_LE(size_after, size_before);
}
class StaticOneByteResource : public v8::String::ExternalOneByteStringResource {
public:
explicit StaticOneByteResource(const char* data) : data_(data) {}
~StaticOneByteResource() {}
const char* data() const { return data_; }
size_t length() const { return strlen(data_); }
private:
const char* data_;
};
TEST(Regress631969) {
FLAG_manual_evacuation_candidates_selection = true;
FLAG_parallel_compaction = false;
FLAG_concurrent_sweeping = false;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
// Get the heap in clean state.
heap->CollectGarbage(OLD_SPACE);
heap->CollectGarbage(OLD_SPACE);
Isolate* isolate = CcTest::i_isolate();
Factory* factory = isolate->factory();
// Allocate two strings in a fresh page and mark the page as evacuation
// candidate.
heap::SimulateFullSpace(heap->old_space());
Handle<String> s1 = factory->NewStringFromStaticChars("123456789", TENURED);
Handle<String> s2 = factory->NewStringFromStaticChars("01234", TENURED);
Page::FromAddress(s1->address())
->SetFlag(MemoryChunk::FORCE_EVACUATION_CANDIDATE_FOR_TESTING);
heap::SimulateIncrementalMarking(heap, false);
// Allocate a cons string and promote it to a fresh page in the old space.
heap::SimulateFullSpace(heap->old_space());
Handle<String> s3;
factory->NewConsString(s1, s2).ToHandle(&s3);
heap->CollectGarbage(NEW_SPACE);
heap->CollectGarbage(NEW_SPACE);
// Finish incremental marking.
IncrementalMarking* marking = heap->incremental_marking();
while (!marking->IsComplete()) {
marking->Step(MB, i::IncrementalMarking::NO_GC_VIA_STACK_GUARD);
if (marking->IsReadyToOverApproximateWeakClosure()) {
marking->FinalizeIncrementally();
}
}
{
StaticOneByteResource external_string("12345678901234");
s3->MakeExternal(&external_string);
heap->CollectGarbage(OLD_SPACE);
}
}
TEST(LeftTrimFixedArrayInBlackArea) {
FLAG_black_allocation = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
heap->CollectAllGarbage();
i::MarkCompactCollector* collector = heap->mark_compact_collector();
i::IncrementalMarking* marking = heap->incremental_marking();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
marking->StartBlackAllocationForTesting();
// Ensure that we allocate a new page, set up a bump pointer area, and
// perform the allocation in a black area.
heap::SimulateFullSpace(heap->old_space());
isolate->factory()->NewFixedArray(4, TENURED);
Handle<FixedArray> array = isolate->factory()->NewFixedArray(50, TENURED);
CHECK(heap->old_space()->Contains(*array));
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(*array)));
// Now left trim the allocated black area. A filler has to be installed
// for the trimmed area and all mark bits of the trimmed area have to be
// cleared.
FixedArrayBase* trimmed = heap->LeftTrimFixedArray(*array, 10);
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(trimmed)));
heap::GcAndSweep(heap, OLD_SPACE);
}
TEST(ContinuousLeftTrimFixedArrayInBlackArea) {
FLAG_black_allocation = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
heap->CollectAllGarbage();
i::MarkCompactCollector* collector = heap->mark_compact_collector();
i::IncrementalMarking* marking = heap->incremental_marking();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
marking->StartBlackAllocationForTesting();
// Ensure that we allocate a new page, set up a bump pointer area, and
// perform the allocation in a black area.
heap::SimulateFullSpace(heap->old_space());
isolate->factory()->NewFixedArray(10, TENURED);
// Allocate the fixed array that will be trimmed later.
Handle<FixedArray> array = isolate->factory()->NewFixedArray(100, TENURED);
Address start_address = array->address();
Address end_address = start_address + array->Size();
Page* page = Page::FromAddress(start_address);
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(*array)));
CHECK(page->markbits()->AllBitsSetInRange(
page->AddressToMarkbitIndex(start_address),
page->AddressToMarkbitIndex(end_address)));
CHECK(heap->old_space()->Contains(*array));
FixedArrayBase* previous = *array;
FixedArrayBase* trimmed;
// First trim in one word steps.
for (int i = 0; i < 10; i++) {
trimmed = heap->LeftTrimFixedArray(previous, 1);
HeapObject* filler = HeapObject::FromAddress(previous->address());
CHECK(filler->IsFiller());
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(trimmed)));
CHECK(Marking::IsImpossible(ObjectMarking::MarkBitFrom(previous)));
previous = trimmed;
}
// Then trim in two and three word steps.
for (int i = 2; i <= 3; i++) {
for (int j = 0; j < 10; j++) {
trimmed = heap->LeftTrimFixedArray(previous, i);
HeapObject* filler = HeapObject::FromAddress(previous->address());
CHECK(filler->IsFiller());
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(trimmed)));
CHECK(Marking::IsWhite(ObjectMarking::MarkBitFrom(previous)));
previous = trimmed;
}
}
heap::GcAndSweep(heap, OLD_SPACE);
}
TEST(ContinuousRightTrimFixedArrayInBlackArea) {
FLAG_black_allocation = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
heap->CollectAllGarbage();
i::MarkCompactCollector* collector = heap->mark_compact_collector();
i::IncrementalMarking* marking = heap->incremental_marking();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
marking->StartBlackAllocationForTesting();
// Ensure that we allocate a new page, set up a bump pointer area, and
// perform the allocation in a black area.
heap::SimulateFullSpace(heap->old_space());
isolate->factory()->NewFixedArray(10, TENURED);
// Allocate the fixed array that will be trimmed later.
Handle<FixedArray> array = isolate->factory()->NewFixedArray(100, TENURED);
Address start_address = array->address();
Address end_address = start_address + array->Size();
Page* page = Page::FromAddress(start_address);
CHECK(Marking::IsBlack(ObjectMarking::MarkBitFrom(*array)));
CHECK(page->markbits()->AllBitsSetInRange(
page->AddressToMarkbitIndex(start_address),
page->AddressToMarkbitIndex(end_address)));
CHECK(heap->old_space()->Contains(*array));
// Trim it once by one word to make checking for white marking color uniform.
Address previous = end_address - kPointerSize;
heap->RightTrimFixedArray<Heap::SEQUENTIAL_TO_SWEEPER>(*array, 1);
HeapObject* filler = HeapObject::FromAddress(previous);
CHECK(filler->IsFiller());
CHECK(Marking::IsImpossible(ObjectMarking::MarkBitFrom(previous)));
// Trim 10 times by one, two, and three word.
for (int i = 1; i <= 3; i++) {
for (int j = 0; j < 10; j++) {
previous -= kPointerSize * i;
heap->RightTrimFixedArray<Heap::SEQUENTIAL_TO_SWEEPER>(*array, i);
HeapObject* filler = HeapObject::FromAddress(previous);
CHECK(filler->IsFiller());
CHECK(Marking::IsWhite(ObjectMarking::MarkBitFrom(previous)));
}
}
heap::GcAndSweep(heap, OLD_SPACE);
}
TEST(SlotFilteringAfterBlackAreas) {
FLAG_black_allocation = true;
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
MarkCompactCollector* mark_compact_collector = heap->mark_compact_collector();
heap->CollectAllGarbage();
i::MarkCompactCollector* collector = heap->mark_compact_collector();
i::IncrementalMarking* marking = heap->incremental_marking();
if (collector->sweeping_in_progress()) {
collector->EnsureSweepingCompleted();
}
CHECK(marking->IsMarking() || marking->IsStopped());
if (marking->IsStopped()) {
heap->StartIncrementalMarking();
}
CHECK(marking->IsMarking());
marking->StartBlackAllocationForTesting();
// Ensure that we allocate a new page, set up a bump pointer area, and
// perform the allocation in a black area.
heap::SimulateFullSpace(heap->old_space());
Handle<FixedArray> array = isolate->factory()->NewFixedArray(10, TENURED);
Page* page = Page::FromAddress(array->address());
// After allocation we empty the allocation info to limit the black area
// only on the allocated array.
heap->old_space()->EmptyAllocationInfo();
// Slots in the black area are part of the black object.
CHECK(mark_compact_collector->IsSlotInBlackObject(page, array->address()));
CHECK(mark_compact_collector->IsSlotInBlackObject(
page, array->address() + array->Size() - kPointerSize));
// Slots after the black area are not part of the black object and have to
// be filtered out.
CHECK(!mark_compact_collector->IsSlotInBlackObject(
page, array->address() + array->Size()));
}
TEST(Regress618958) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
bool isolate_is_locked = true;
heap->update_external_memory(100 * MB);
int mark_sweep_count_before = heap->ms_count();
heap->MemoryPressureNotification(MemoryPressureLevel::kCritical,
isolate_is_locked);
int mark_sweep_count_after = heap->ms_count();
int mark_sweeps_performed = mark_sweep_count_after - mark_sweep_count_before;
// The memory pressuer handler either performed two GCs or performed one and
// started incremental marking.
CHECK(mark_sweeps_performed == 2 ||
(mark_sweeps_performed == 1 &&
!heap->incremental_marking()->IsStopped()));
}
TEST(UncommitUnusedLargeObjectMemory) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
Handle<FixedArray> array = isolate->factory()->NewFixedArray(200000);
MemoryChunk* chunk = MemoryChunk::FromAddress(array->address());
CHECK(chunk->owner()->identity() == LO_SPACE);
intptr_t size_before = array->Size();
size_t committed_memory_before = chunk->CommittedPhysicalMemory();
array->Shrink(1);
CHECK(array->Size() < size_before);
CcTest::heap()->CollectAllGarbage();
CHECK(chunk->CommittedPhysicalMemory() < committed_memory_before);
size_t shrinked_size =
RoundUp((array->address() - chunk->address()) + array->Size(),
base::OS::CommitPageSize());
CHECK_EQ(shrinked_size, chunk->CommittedPhysicalMemory());
}
TEST(RememberedSetRemoveRange) {
CcTest::InitializeVM();
v8::HandleScope scope(CcTest::isolate());
Heap* heap = CcTest::heap();
Isolate* isolate = heap->isolate();
Handle<FixedArray> array = isolate->factory()->NewFixedArray(Page::kPageSize /
kPointerSize);
MemoryChunk* chunk = MemoryChunk::FromAddress(array->address());
CHECK(chunk->owner()->identity() == LO_SPACE);
Address start = array->address();
// Maps slot to boolean indicator of whether the slot should be in the set.
std::map<Address, bool> slots;
slots[start + 0] = true;
slots[start + kPointerSize] = true;
slots[start + Page::kPageSize - kPointerSize] = true;
slots[start + Page::kPageSize] = true;
slots[start + Page::kPageSize + kPointerSize] = true;
slots[chunk->area_end() - kPointerSize] = true;
for (auto x : slots) {
RememberedSet<OLD_TO_NEW>::Insert(chunk, x.first);
}
RememberedSet<OLD_TO_NEW>::Iterate(chunk, [&slots](Address addr) {
CHECK(slots[addr]);
return KEEP_SLOT;
});
RememberedSet<OLD_TO_NEW>::RemoveRange(chunk, start, start + kPointerSize);
slots[start] = false;
RememberedSet<OLD_TO_NEW>::Iterate(chunk, [&slots](Address addr) {
CHECK(slots[addr]);
return KEEP_SLOT;
});
RememberedSet<OLD_TO_NEW>::RemoveRange(chunk, start + kPointerSize,
start + Page::kPageSize);
slots[start + kPointerSize] = false;
slots[start + Page::kPageSize - kPointerSize] = false;
RememberedSet<OLD_TO_NEW>::Iterate(chunk, [&slots](Address addr) {
CHECK(slots[addr]);
return KEEP_SLOT;
});
RememberedSet<OLD_TO_NEW>::RemoveRange(
chunk, start, start + Page::kPageSize + kPointerSize);
slots[start + Page::kPageSize] = false;
RememberedSet<OLD_TO_NEW>::Iterate(chunk, [&slots](Address addr) {
CHECK(slots[addr]);
return KEEP_SLOT;
});
RememberedSet<OLD_TO_NEW>::RemoveRange(
chunk, chunk->area_end() - kPointerSize, chunk->area_end());
slots[chunk->area_end() - kPointerSize] = false;
RememberedSet<OLD_TO_NEW>::Iterate(chunk, [&slots](Address addr) {
CHECK(slots[addr]);
return KEEP_SLOT;
});
}
} // namespace internal
} // namespace v8