v8/src/compiler/code-assembler.cc

1960 lines
68 KiB
C++
Raw Normal View History

// Copyright 2015 the V8 project authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "src/compiler/code-assembler.h"
#include <ostream>
#include "src/codegen/code-factory.h"
#include "src/codegen/interface-descriptors.h"
#include "src/codegen/machine-type.h"
#include "src/codegen/macro-assembler.h"
#include "src/compiler/backend/instruction-selector.h"
#include "src/compiler/graph.h"
#include "src/compiler/linkage.h"
#include "src/compiler/node-matchers.h"
#include "src/compiler/pipeline.h"
#include "src/compiler/raw-machine-assembler.h"
#include "src/compiler/schedule.h"
#include "src/frames.h"
#include "src/interpreter/bytecodes.h"
#include "src/memcopy.h"
#include "src/objects-inl.h"
#include "src/objects/smi.h"
#include "src/zone/zone.h"
namespace v8 {
namespace internal {
constexpr MachineType MachineTypeOf<Smi>::value;
constexpr MachineType MachineTypeOf<Object>::value;
namespace compiler {
static_assert(std::is_convertible<TNode<Number>, TNode<Object>>::value,
"test subtyping");
static_assert(std::is_convertible<TNode<UnionT<Smi, HeapNumber>>,
TNode<UnionT<Smi, HeapObject>>>::value,
"test subtyping");
static_assert(
!std::is_convertible<TNode<UnionT<Smi, HeapObject>>, TNode<Number>>::value,
"test subtyping");
CodeAssemblerState::CodeAssemblerState(
Isolate* isolate, Zone* zone, const CallInterfaceDescriptor& descriptor,
Code::Kind kind, const char* name, PoisoningMitigationLevel poisoning_level,
int32_t builtin_index)
// TODO(rmcilroy): Should we use Linkage::GetBytecodeDispatchDescriptor for
// bytecode handlers?
: CodeAssemblerState(
isolate, zone,
Linkage::GetStubCallDescriptor(
zone, descriptor, descriptor.GetStackParameterCount(),
CallDescriptor::kNoFlags, Operator::kNoProperties),
kind, name, poisoning_level, builtin_index) {}
CodeAssemblerState::CodeAssemblerState(Isolate* isolate, Zone* zone,
int parameter_count, Code::Kind kind,
const char* name,
PoisoningMitigationLevel poisoning_level,
int32_t builtin_index)
: CodeAssemblerState(
isolate, zone,
Linkage::GetJSCallDescriptor(
zone, false, parameter_count,
(kind == Code::BUILTIN ? CallDescriptor::kPushArgumentCount
: CallDescriptor::kNoFlags) |
CallDescriptor::kCanUseRoots),
kind, name, poisoning_level, builtin_index) {}
CodeAssemblerState::CodeAssemblerState(Isolate* isolate, Zone* zone,
CallDescriptor* call_descriptor,
Code::Kind kind, const char* name,
PoisoningMitigationLevel poisoning_level,
int32_t builtin_index)
: raw_assembler_(new RawMachineAssembler(
isolate, new (zone) Graph(zone), call_descriptor,
MachineType::PointerRepresentation(),
InstructionSelector::SupportedMachineOperatorFlags(),
InstructionSelector::AlignmentRequirements(), poisoning_level)),
kind_(kind),
name_(name),
builtin_index_(builtin_index),
code_generated_(false),
variables_(zone) {}
CodeAssemblerState::~CodeAssemblerState() = default;
int CodeAssemblerState::parameter_count() const {
return static_cast<int>(raw_assembler_->call_descriptor()->ParameterCount());
}
CodeAssembler::~CodeAssembler() = default;
#if DEBUG
void CodeAssemblerState::PrintCurrentBlock(std::ostream& os) {
raw_assembler_->PrintCurrentBlock(os);
}
#endif
bool CodeAssemblerState::InsideBlock() { return raw_assembler_->InsideBlock(); }
void CodeAssemblerState::SetInitialDebugInformation(const char* msg,
const char* file,
int line) {
#if DEBUG
AssemblerDebugInfo debug_info = {msg, file, line};
raw_assembler_->SetSourcePosition(file, line);
raw_assembler_->SetInitialDebugInformation(debug_info);
#endif // DEBUG
}
class BreakOnNodeDecorator final : public GraphDecorator {
public:
explicit BreakOnNodeDecorator(NodeId node_id) : node_id_(node_id) {}
void Decorate(Node* node) final {
if (node->id() == node_id_) {
base::OS::DebugBreak();
}
}
private:
NodeId node_id_;
};
void CodeAssembler::BreakOnNode(int node_id) {
Graph* graph = raw_assembler()->graph();
Zone* zone = graph->zone();
GraphDecorator* decorator =
new (zone) BreakOnNodeDecorator(static_cast<NodeId>(node_id));
graph->AddDecorator(decorator);
}
void CodeAssembler::RegisterCallGenerationCallbacks(
const CodeAssemblerCallback& call_prologue,
const CodeAssemblerCallback& call_epilogue) {
// The callback can be registered only once.
DCHECK(!state_->call_prologue_);
DCHECK(!state_->call_epilogue_);
state_->call_prologue_ = call_prologue;
state_->call_epilogue_ = call_epilogue;
}
void CodeAssembler::UnregisterCallGenerationCallbacks() {
state_->call_prologue_ = nullptr;
state_->call_epilogue_ = nullptr;
}
void CodeAssembler::CallPrologue() {
if (state_->call_prologue_) {
state_->call_prologue_();
}
}
void CodeAssembler::CallEpilogue() {
if (state_->call_epilogue_) {
state_->call_epilogue_();
}
}
bool CodeAssembler::Word32ShiftIsSafe() const {
return raw_assembler()->machine()->Word32ShiftIsSafe();
}
PoisoningMitigationLevel CodeAssembler::poisoning_level() const {
return raw_assembler()->poisoning_level();
}
// static
Handle<Code> CodeAssembler::GenerateCode(CodeAssemblerState* state,
const AssemblerOptions& options) {
DCHECK(!state->code_generated_);
RawMachineAssembler* rasm = state->raw_assembler_.get();
Handle<Code> code;
Graph* graph = rasm->ExportForOptimization();
code = Pipeline::GenerateCodeForCodeStub(
rasm->isolate(), rasm->call_descriptor(), graph,
rasm->source_positions(), state->kind_, state->name_,
state->builtin_index_, rasm->poisoning_level(), options)
.ToHandleChecked();
state->code_generated_ = true;
return code;
}
bool CodeAssembler::Is64() const { return raw_assembler()->machine()->Is64(); }
bool CodeAssembler::IsFloat64RoundUpSupported() const {
return raw_assembler()->machine()->Float64RoundUp().IsSupported();
}
bool CodeAssembler::IsFloat64RoundDownSupported() const {
return raw_assembler()->machine()->Float64RoundDown().IsSupported();
}
bool CodeAssembler::IsFloat64RoundTiesEvenSupported() const {
return raw_assembler()->machine()->Float64RoundTiesEven().IsSupported();
}
bool CodeAssembler::IsFloat64RoundTruncateSupported() const {
return raw_assembler()->machine()->Float64RoundTruncate().IsSupported();
}
bool CodeAssembler::IsInt32AbsWithOverflowSupported() const {
return raw_assembler()->machine()->Int32AbsWithOverflow().IsSupported();
}
bool CodeAssembler::IsInt64AbsWithOverflowSupported() const {
return raw_assembler()->machine()->Int64AbsWithOverflow().IsSupported();
}
bool CodeAssembler::IsIntPtrAbsWithOverflowSupported() const {
return Is64() ? IsInt64AbsWithOverflowSupported()
: IsInt32AbsWithOverflowSupported();
}
#ifdef DEBUG
void CodeAssembler::GenerateCheckMaybeObjectIsObject(Node* node,
const char* location) {
Label ok(this);
GotoIf(WordNotEqual(WordAnd(BitcastMaybeObjectToWord(node),
IntPtrConstant(kHeapObjectTagMask)),
IntPtrConstant(kWeakHeapObjectTag)),
&ok);
Node* message_node = StringConstant(location);
DebugAbort(message_node);
Unreachable();
Bind(&ok);
}
#endif
TNode<Int32T> CodeAssembler::Int32Constant(int32_t value) {
return UncheckedCast<Int32T>(raw_assembler()->Int32Constant(value));
}
TNode<Int64T> CodeAssembler::Int64Constant(int64_t value) {
return UncheckedCast<Int64T>(raw_assembler()->Int64Constant(value));
}
TNode<IntPtrT> CodeAssembler::IntPtrConstant(intptr_t value) {
return UncheckedCast<IntPtrT>(raw_assembler()->IntPtrConstant(value));
}
TNode<Number> CodeAssembler::NumberConstant(double value) {
int smi_value;
if (DoubleToSmiInteger(value, &smi_value)) {
return UncheckedCast<Number>(SmiConstant(smi_value));
} else {
// We allocate the heap number constant eagerly at this point instead of
// deferring allocation to code generation
// (see AllocateAndInstallRequestedHeapObjects) since that makes it easier
// to generate constant lookups for embedded builtins.
return UncheckedCast<Number>(HeapConstant(
isolate()->factory()->NewHeapNumber(value, AllocationType::kOld)));
}
}
TNode<Smi> CodeAssembler::SmiConstant(Smi value) {
return UncheckedCast<Smi>(BitcastWordToTaggedSigned(
IntPtrConstant(static_cast<intptr_t>(value.ptr()))));
}
TNode<Smi> CodeAssembler::SmiConstant(int value) {
return SmiConstant(Smi::FromInt(value));
}
TNode<HeapObject> CodeAssembler::UntypedHeapConstant(
Handle<HeapObject> object) {
return UncheckedCast<HeapObject>(raw_assembler()->HeapConstant(object));
}
TNode<String> CodeAssembler::StringConstant(const char* str) {
Handle<String> internalized_string =
factory()->InternalizeOneByteString(OneByteVector(str));
return UncheckedCast<String>(HeapConstant(internalized_string));
}
TNode<Oddball> CodeAssembler::BooleanConstant(bool value) {
Handle<Object> object = isolate()->factory()->ToBoolean(value);
return UncheckedCast<Oddball>(
raw_assembler()->HeapConstant(Handle<HeapObject>::cast(object)));
}
TNode<ExternalReference> CodeAssembler::ExternalConstant(
ExternalReference address) {
return UncheckedCast<ExternalReference>(
raw_assembler()->ExternalConstant(address));
}
TNode<Float64T> CodeAssembler::Float64Constant(double value) {
return UncheckedCast<Float64T>(raw_assembler()->Float64Constant(value));
}
TNode<HeapNumber> CodeAssembler::NaNConstant() {
return UncheckedCast<HeapNumber>(LoadRoot(RootIndex::kNanValue));
}
bool CodeAssembler::ToInt32Constant(Node* node, int32_t& out_value) {
{
Int64Matcher m(node);
if (m.HasValue() && m.IsInRange(std::numeric_limits<int32_t>::min(),
std::numeric_limits<int32_t>::max())) {
out_value = static_cast<int32_t>(m.Value());
return true;
}
}
{
Int32Matcher m(node);
if (m.HasValue()) {
out_value = m.Value();
return true;
}
}
return false;
}
bool CodeAssembler::ToInt64Constant(Node* node, int64_t& out_value) {
Int64Matcher m(node);
if (m.HasValue()) out_value = m.Value();
return m.HasValue();
}
bool CodeAssembler::ToSmiConstant(Node* node, Smi* out_value) {
if (node->opcode() == IrOpcode::kBitcastWordToTaggedSigned) {
node = node->InputAt(0);
}
IntPtrMatcher m(node);
if (m.HasValue()) {
intptr_t value = m.Value();
// Make sure that the value is actually a smi
CHECK_EQ(0, value & ((static_cast<intptr_t>(1) << kSmiShiftSize) - 1));
*out_value = Smi(static_cast<Address>(value));
return true;
}
return false;
}
bool CodeAssembler::ToIntPtrConstant(Node* node, intptr_t& out_value) {
if (node->opcode() == IrOpcode::kBitcastWordToTaggedSigned ||
node->opcode() == IrOpcode::kBitcastWordToTagged) {
node = node->InputAt(0);
}
IntPtrMatcher m(node);
if (m.HasValue()) out_value = m.Value();
return m.HasValue();
}
bool CodeAssembler::IsUndefinedConstant(TNode<Object> node) {
compiler::HeapObjectMatcher m(node);
return m.Is(isolate()->factory()->undefined_value());
}
bool CodeAssembler::IsNullConstant(TNode<Object> node) {
compiler::HeapObjectMatcher m(node);
return m.Is(isolate()->factory()->null_value());
}
Node* CodeAssembler::Parameter(int index) {
if (index == kTargetParameterIndex) return raw_assembler()->TargetParameter();
return raw_assembler()->Parameter(index);
}
bool CodeAssembler::IsJSFunctionCall() const {
auto call_descriptor = raw_assembler()->call_descriptor();
return call_descriptor->IsJSFunctionCall();
}
TNode<Context> CodeAssembler::GetJSContextParameter() {
auto call_descriptor = raw_assembler()->call_descriptor();
DCHECK(call_descriptor->IsJSFunctionCall());
return CAST(Parameter(Linkage::GetJSCallContextParamIndex(
static_cast<int>(call_descriptor->JSParameterCount()))));
}
void CodeAssembler::Return(SloppyTNode<Object> value) {
return raw_assembler()->Return(value);
}
void CodeAssembler::Return(SloppyTNode<Object> value1,
SloppyTNode<Object> value2) {
return raw_assembler()->Return(value1, value2);
}
void CodeAssembler::Return(SloppyTNode<Object> value1,
SloppyTNode<Object> value2,
SloppyTNode<Object> value3) {
return raw_assembler()->Return(value1, value2, value3);
}
void CodeAssembler::PopAndReturn(Node* pop, Node* value) {
return raw_assembler()->PopAndReturn(pop, value);
}
[async-iteration] implement AsyncGenerator - Introduce new struct AsyncGeneratorRequest, which holds information pertinent to resuming execution of an AsyncGenerator, such as the Promise associated with the async generator request. It is intended to be used as a singly linked list, and holds a pointer to the next item in te queue. - Introduce JSAsyncGeneratorObject (subclass of JSGeneratorObject), which includes several new internal fields (`queue` which contains a singly linked list of AsyncGeneratorRequest objects, and `await_input` which contains the sent value from an Await expression (This is necessary to prevent function.sent (used by yield*) from having the sent value observably overwritten during execution). - Modify SuspendGenerator to accept a set of Flags, which indicate whether the suspend is for a Yield or Await, and whether it takes place on an async generator or ES6 generator. - Introduce interpreter intrinsics and TF intrinsic lowering for accessing the await input of an async generator - Modify the JSGeneratorStore operator to understand whether or not it's suspending for a normal yield, or an AsyncGenerator Await. This ensures appropriate registers are stored. - Add versions of ResumeGeneratorTrampoline which store the input value in a different field depending on wether it's an AsyncGenerator Await resume, or an ordinary resume. Also modifies whether debug code will assert that the generator object is a JSGeneratorObject or a JSAsyncGeneratorObject depending on the resume type. BUG=v8:5855 R=bmeurer@chromium.org, rmcilroy@chromium.org, jgruber@chromium.org, littledan@chromium.org, neis@chromium.org TBR=marja@chromium.org Change-Id: I9d58df1d344465fc937fe7eed322424204497187 Reviewed-on: https://chromium-review.googlesource.com/446961 Commit-Queue: Caitlin Potter <caitp@igalia.com> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org> Reviewed-by: Hannes Payer <hpayer@chromium.org> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by: Jakob Gruber <jgruber@chromium.org> Cr-Commit-Position: refs/heads/master@{#44240}
2017-03-29 13:41:45 +00:00
void CodeAssembler::ReturnIf(Node* condition, Node* value) {
Label if_return(this), if_continue(this);
Branch(condition, &if_return, &if_continue);
Bind(&if_return);
Return(value);
Bind(&if_continue);
}
void CodeAssembler::ReturnRaw(Node* value) {
return raw_assembler()->Return(value);
}
void CodeAssembler::DebugAbort(Node* message) {
raw_assembler()->DebugAbort(message);
}
void CodeAssembler::DebugBreak() { raw_assembler()->DebugBreak(); }
void CodeAssembler::Unreachable() {
DebugBreak();
raw_assembler()->Unreachable();
}
Reland "Reland "Reland "[code-comments] Put code comments into the code object""" This is a reland of 9c0a48580bc820d93a16f8914281a7359beb2a7a Original change's description: > Reland "Reland "[code-comments] Put code comments into the code object"" > > This is a reland of ed3d647284538e9d6f013ebf2c460697aa06a5df > > This reland fixes that padding at the end of Wasm instruction streams > triggered asserts in the code printer. > > Original change's description: > > Reland "[code-comments] Put code comments into the code object" > > > > This is a reland of e774cffe2bd3f00332209d4d5695221963888c96 > > > > This reland disables a test as v8:8548 is blocking it, which was > > broken by a recent CL. CQ did not catch this because the merge-base > > CQ used did not yet contain the CL that caused v8:8548. > > > > Original change's description: > > > [code-comments] Put code comments into the code object > > > > > > Code comments in the snapshot can now be enabled with gn > > > arg 'v8_enable_snapshot_code_comments' > > > > > > Bug: v8:7989 > > > Change-Id: I8bd00cafa63132d00d849394c311ba15e6b6daf3 > > > Reviewed-on: https://chromium-review.googlesource.com/c/1329173 > > > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org> > > > Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> > > > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > > > Cr-Commit-Position: refs/heads/master@{#58020} > > > > TBR=mvstanton@chromium.org,mstarzinger@chromium.org,jgruber@chromium.org,tebbi@chromium.org > > > > Bug: v8:7989, v8:8548 > > Change-Id: I464fc897205fefdf2dfc2eadc54d699c4e08a0e9 > > Reviewed-on: https://chromium-review.googlesource.com/c/1361166 > > Reviewed-by: Sigurd Schneider <sigurds@chromium.org> > > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#58028} > > Bug: v8:7989, v8:8548 > Change-Id: I254f55ff687ad049f8d92b09331ed26a2bd05d7d > Reviewed-on: https://chromium-review.googlesource.com/c/1371784 > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> > Reviewed-by: Jakob Gruber <jgruber@chromium.org> > Cr-Commit-Position: refs/heads/master@{#58221} TBR=jgruber@chromium.org,mstarzinger@chromium.org Bug: v8:7989, v8:8548, v8:8593 Change-Id: I4f7ffc98e0281c7b744eb4a04ba0763896c7b59b Reviewed-on: https://chromium-review.googlesource.com/c/1375919 Reviewed-by: Sigurd Schneider <sigurds@chromium.org> Commit-Queue: Sigurd Schneider <sigurds@chromium.org> Cr-Commit-Position: refs/heads/master@{#58232}
2018-12-13 19:30:56 +00:00
void CodeAssembler::Comment(std::string str) {
if (!FLAG_code_comments) return;
Reland "Reland "Reland "[code-comments] Put code comments into the code object""" This is a reland of 9c0a48580bc820d93a16f8914281a7359beb2a7a Original change's description: > Reland "Reland "[code-comments] Put code comments into the code object"" > > This is a reland of ed3d647284538e9d6f013ebf2c460697aa06a5df > > This reland fixes that padding at the end of Wasm instruction streams > triggered asserts in the code printer. > > Original change's description: > > Reland "[code-comments] Put code comments into the code object" > > > > This is a reland of e774cffe2bd3f00332209d4d5695221963888c96 > > > > This reland disables a test as v8:8548 is blocking it, which was > > broken by a recent CL. CQ did not catch this because the merge-base > > CQ used did not yet contain the CL that caused v8:8548. > > > > Original change's description: > > > [code-comments] Put code comments into the code object > > > > > > Code comments in the snapshot can now be enabled with gn > > > arg 'v8_enable_snapshot_code_comments' > > > > > > Bug: v8:7989 > > > Change-Id: I8bd00cafa63132d00d849394c311ba15e6b6daf3 > > > Reviewed-on: https://chromium-review.googlesource.com/c/1329173 > > > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org> > > > Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> > > > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > > > Cr-Commit-Position: refs/heads/master@{#58020} > > > > TBR=mvstanton@chromium.org,mstarzinger@chromium.org,jgruber@chromium.org,tebbi@chromium.org > > > > Bug: v8:7989, v8:8548 > > Change-Id: I464fc897205fefdf2dfc2eadc54d699c4e08a0e9 > > Reviewed-on: https://chromium-review.googlesource.com/c/1361166 > > Reviewed-by: Sigurd Schneider <sigurds@chromium.org> > > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#58028} > > Bug: v8:7989, v8:8548 > Change-Id: I254f55ff687ad049f8d92b09331ed26a2bd05d7d > Reviewed-on: https://chromium-review.googlesource.com/c/1371784 > Commit-Queue: Sigurd Schneider <sigurds@chromium.org> > Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> > Reviewed-by: Jakob Gruber <jgruber@chromium.org> > Cr-Commit-Position: refs/heads/master@{#58221} TBR=jgruber@chromium.org,mstarzinger@chromium.org Bug: v8:7989, v8:8548, v8:8593 Change-Id: I4f7ffc98e0281c7b744eb4a04ba0763896c7b59b Reviewed-on: https://chromium-review.googlesource.com/c/1375919 Reviewed-by: Sigurd Schneider <sigurds@chromium.org> Commit-Queue: Sigurd Schneider <sigurds@chromium.org> Cr-Commit-Position: refs/heads/master@{#58232}
2018-12-13 19:30:56 +00:00
raw_assembler()->Comment(str);
}
void CodeAssembler::StaticAssert(TNode<BoolT> value) {
raw_assembler()->StaticAssert(value);
}
void CodeAssembler::SetSourcePosition(const char* file, int line) {
raw_assembler()->SetSourcePosition(file, line);
}
void CodeAssembler::Bind(Label* label) { return label->Bind(); }
#if DEBUG
void CodeAssembler::Bind(Label* label, AssemblerDebugInfo debug_info) {
return label->Bind(debug_info);
}
#endif // DEBUG
Node* CodeAssembler::LoadFramePointer() {
return raw_assembler()->LoadFramePointer();
}
Node* CodeAssembler::LoadParentFramePointer() {
return raw_assembler()->LoadParentFramePointer();
}
Node* CodeAssembler::LoadStackPointer() {
return raw_assembler()->LoadStackPointer();
}
TNode<Object> CodeAssembler::TaggedPoisonOnSpeculation(
SloppyTNode<Object> value) {
return UncheckedCast<Object>(
raw_assembler()->TaggedPoisonOnSpeculation(value));
}
TNode<WordT> CodeAssembler::WordPoisonOnSpeculation(SloppyTNode<WordT> value) {
return UncheckedCast<WordT>(raw_assembler()->WordPoisonOnSpeculation(value));
}
#define DEFINE_CODE_ASSEMBLER_BINARY_OP(name, ResType, Arg1Type, Arg2Type) \
TNode<ResType> CodeAssembler::name(SloppyTNode<Arg1Type> a, \
SloppyTNode<Arg2Type> b) { \
return UncheckedCast<ResType>(raw_assembler()->name(a, b)); \
}
CODE_ASSEMBLER_BINARY_OP_LIST(DEFINE_CODE_ASSEMBLER_BINARY_OP)
#undef DEFINE_CODE_ASSEMBLER_BINARY_OP
TNode<WordT> CodeAssembler::IntPtrAdd(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant + right_constant);
}
if (left_constant == 0) {
return right;
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<WordT>(raw_assembler()->IntPtrAdd(left, right));
}
TNode<IntPtrT> CodeAssembler::IntPtrDiv(TNode<IntPtrT> left,
TNode<IntPtrT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_right_constant) {
if (is_left_constant) {
return IntPtrConstant(left_constant / right_constant);
}
if (base::bits::IsPowerOfTwo(right_constant)) {
return WordSar(left, WhichPowerOf2(right_constant));
}
}
return UncheckedCast<IntPtrT>(raw_assembler()->IntPtrDiv(left, right));
}
TNode<WordT> CodeAssembler::IntPtrSub(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant - right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<IntPtrT>(raw_assembler()->IntPtrSub(left, right));
}
TNode<WordT> CodeAssembler::IntPtrMul(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant * right_constant);
}
if (base::bits::IsPowerOfTwo(left_constant)) {
return WordShl(right, WhichPowerOf2(left_constant));
}
} else if (is_right_constant) {
if (base::bits::IsPowerOfTwo(right_constant)) {
return WordShl(left, WhichPowerOf2(right_constant));
}
}
return UncheckedCast<IntPtrT>(raw_assembler()->IntPtrMul(left, right));
}
TNode<WordT> CodeAssembler::WordShl(SloppyTNode<WordT> value, int shift) {
return (shift != 0) ? WordShl(value, IntPtrConstant(shift)) : value;
}
TNode<WordT> CodeAssembler::WordShr(SloppyTNode<WordT> value, int shift) {
return (shift != 0) ? WordShr(value, IntPtrConstant(shift)) : value;
}
TNode<WordT> CodeAssembler::WordSar(SloppyTNode<WordT> value, int shift) {
return (shift != 0) ? WordSar(value, IntPtrConstant(shift)) : value;
}
TNode<Word32T> CodeAssembler::Word32Shr(SloppyTNode<Word32T> value, int shift) {
return (shift != 0) ? Word32Shr(value, Int32Constant(shift)) : value;
}
TNode<WordT> CodeAssembler::WordOr(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant | right_constant);
}
if (left_constant == 0) {
return right;
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<WordT>(raw_assembler()->WordOr(left, right));
}
TNode<WordT> CodeAssembler::WordAnd(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant & right_constant);
}
}
return UncheckedCast<WordT>(raw_assembler()->WordAnd(left, right));
}
TNode<WordT> CodeAssembler::WordXor(SloppyTNode<WordT> left,
SloppyTNode<WordT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant ^ right_constant);
}
}
return UncheckedCast<WordT>(raw_assembler()->WordXor(left, right));
}
TNode<WordT> CodeAssembler::WordShl(SloppyTNode<WordT> left,
SloppyTNode<IntegralT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant << right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<WordT>(raw_assembler()->WordShl(left, right));
}
TNode<WordT> CodeAssembler::WordShr(SloppyTNode<WordT> left,
SloppyTNode<IntegralT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(static_cast<uintptr_t>(left_constant) >>
right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<WordT>(raw_assembler()->WordShr(left, right));
}
TNode<WordT> CodeAssembler::WordSar(SloppyTNode<WordT> left,
SloppyTNode<IntegralT> right) {
intptr_t left_constant;
bool is_left_constant = ToIntPtrConstant(left, left_constant);
intptr_t right_constant;
bool is_right_constant = ToIntPtrConstant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return IntPtrConstant(left_constant >> right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<WordT>(raw_assembler()->WordSar(left, right));
}
TNode<Word32T> CodeAssembler::Word32Or(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(left_constant | right_constant);
}
if (left_constant == 0) {
return right;
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32Or(left, right));
}
TNode<Word32T> CodeAssembler::Word32And(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(left_constant & right_constant);
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32And(left, right));
}
TNode<Word32T> CodeAssembler::Word32Xor(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(left_constant ^ right_constant);
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32Xor(left, right));
}
TNode<Word32T> CodeAssembler::Word32Shl(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(left_constant << right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32Shl(left, right));
}
TNode<Word32T> CodeAssembler::Word32Shr(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(static_cast<uint32_t>(left_constant) >>
right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32Shr(left, right));
}
TNode<Word32T> CodeAssembler::Word32Sar(SloppyTNode<Word32T> left,
SloppyTNode<Word32T> right) {
int32_t left_constant;
bool is_left_constant = ToInt32Constant(left, left_constant);
int32_t right_constant;
bool is_right_constant = ToInt32Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int32Constant(left_constant >> right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word32T>(raw_assembler()->Word32Sar(left, right));
}
TNode<Word64T> CodeAssembler::Word64Or(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(left_constant | right_constant);
}
if (left_constant == 0) {
return right;
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64Or(left, right));
}
TNode<Word64T> CodeAssembler::Word64And(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(left_constant & right_constant);
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64And(left, right));
}
TNode<Word64T> CodeAssembler::Word64Xor(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(left_constant ^ right_constant);
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64Xor(left, right));
}
TNode<Word64T> CodeAssembler::Word64Shl(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(left_constant << right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64Shl(left, right));
}
TNode<Word64T> CodeAssembler::Word64Shr(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(static_cast<uint64_t>(left_constant) >>
right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64Shr(left, right));
}
TNode<Word64T> CodeAssembler::Word64Sar(SloppyTNode<Word64T> left,
SloppyTNode<Word64T> right) {
int64_t left_constant;
bool is_left_constant = ToInt64Constant(left, left_constant);
int64_t right_constant;
bool is_right_constant = ToInt64Constant(right, right_constant);
if (is_left_constant) {
if (is_right_constant) {
return Int64Constant(left_constant >> right_constant);
}
} else if (is_right_constant) {
if (right_constant == 0) {
return left;
}
}
return UncheckedCast<Word64T>(raw_assembler()->Word64Sar(left, right));
}
#define CODE_ASSEMBLER_COMPARE(Name, ArgT, VarT, ToConstant, op) \
TNode<BoolT> CodeAssembler::Name(SloppyTNode<ArgT> left, \
SloppyTNode<ArgT> right) { \
VarT lhs, rhs; \
if (ToConstant(left, lhs) && ToConstant(right, rhs)) { \
return BoolConstant(lhs op rhs); \
} \
return UncheckedCast<BoolT>(raw_assembler()->Name(left, right)); \
}
CODE_ASSEMBLER_COMPARE(IntPtrEqual, WordT, intptr_t, ToIntPtrConstant, ==)
CODE_ASSEMBLER_COMPARE(WordEqual, WordT, intptr_t, ToIntPtrConstant, ==)
CODE_ASSEMBLER_COMPARE(WordNotEqual, WordT, intptr_t, ToIntPtrConstant, !=)
CODE_ASSEMBLER_COMPARE(Word32Equal, Word32T, int32_t, ToInt32Constant, ==)
CODE_ASSEMBLER_COMPARE(Word32NotEqual, Word32T, int32_t, ToInt32Constant, !=)
CODE_ASSEMBLER_COMPARE(Word64Equal, Word64T, int64_t, ToInt64Constant, ==)
CODE_ASSEMBLER_COMPARE(Word64NotEqual, Word64T, int64_t, ToInt64Constant, !=)
#undef CODE_ASSEMBLER_COMPARE
TNode<UintPtrT> CodeAssembler::ChangeUint32ToWord(SloppyTNode<Word32T> value) {
if (raw_assembler()->machine()->Is64()) {
return UncheckedCast<UintPtrT>(
raw_assembler()->ChangeUint32ToUint64(value));
}
return ReinterpretCast<UintPtrT>(value);
}
TNode<IntPtrT> CodeAssembler::ChangeInt32ToIntPtr(SloppyTNode<Word32T> value) {
if (raw_assembler()->machine()->Is64()) {
return ReinterpretCast<IntPtrT>(raw_assembler()->ChangeInt32ToInt64(value));
}
return ReinterpretCast<IntPtrT>(value);
}
TNode<UintPtrT> CodeAssembler::ChangeFloat64ToUintPtr(
SloppyTNode<Float64T> value) {
if (raw_assembler()->machine()->Is64()) {
return ReinterpretCast<UintPtrT>(
raw_assembler()->ChangeFloat64ToUint64(value));
}
return ReinterpretCast<UintPtrT>(
raw_assembler()->ChangeFloat64ToUint32(value));
}
TNode<Float64T> CodeAssembler::ChangeUintPtrToFloat64(TNode<UintPtrT> value) {
if (raw_assembler()->machine()->Is64()) {
// TODO(turbofan): Maybe we should introduce a ChangeUint64ToFloat64
// machine operator to TurboFan here?
return ReinterpretCast<Float64T>(
raw_assembler()->RoundUint64ToFloat64(value));
}
return ReinterpretCast<Float64T>(
raw_assembler()->ChangeUint32ToFloat64(value));
}
Node* CodeAssembler::RoundIntPtrToFloat64(Node* value) {
if (raw_assembler()->machine()->Is64()) {
return raw_assembler()->RoundInt64ToFloat64(value);
}
return raw_assembler()->ChangeInt32ToFloat64(value);
}
#define DEFINE_CODE_ASSEMBLER_UNARY_OP(name, ResType, ArgType) \
TNode<ResType> CodeAssembler::name(SloppyTNode<ArgType> a) { \
return UncheckedCast<ResType>(raw_assembler()->name(a)); \
}
CODE_ASSEMBLER_UNARY_OP_LIST(DEFINE_CODE_ASSEMBLER_UNARY_OP)
#undef DEFINE_CODE_ASSEMBLER_UNARY_OP
Node* CodeAssembler::Load(MachineType type, Node* base,
LoadSensitivity needs_poisoning) {
return raw_assembler()->Load(type, base, needs_poisoning);
}
Node* CodeAssembler::Load(MachineType type, Node* base, Node* offset,
LoadSensitivity needs_poisoning) {
return raw_assembler()->Load(type, base, offset, needs_poisoning);
}
Node* CodeAssembler::LoadFullTagged(Node* base,
LoadSensitivity needs_poisoning) {
return BitcastWordToTagged(
Load(MachineType::Pointer(), base, needs_poisoning));
}
Node* CodeAssembler::LoadFullTagged(Node* base, Node* offset,
LoadSensitivity needs_poisoning) {
return BitcastWordToTagged(
Load(MachineType::Pointer(), base, offset, needs_poisoning));
}
Node* CodeAssembler::AtomicLoad(MachineType type, Node* base, Node* offset) {
return raw_assembler()->AtomicLoad(type, base, offset);
}
TNode<Object> CodeAssembler::LoadRoot(RootIndex root_index) {
if (RootsTable::IsImmortalImmovable(root_index)) {
Handle<Object> root = isolate()->root_handle(root_index);
if (root->IsSmi()) {
return SmiConstant(Smi::cast(*root));
} else {
return HeapConstant(Handle<HeapObject>::cast(root));
}
}
// TODO(jgruber): In theory we could generate better code for this by
// letting the macro assembler decide how to load from the roots list. In most
// cases, it would boil down to loading from a fixed kRootRegister offset.
Node* isolate_root =
ExternalConstant(ExternalReference::isolate_root(isolate()));
int offset = IsolateData::root_slot_offset(root_index);
return UncheckedCast<Object>(
LoadFullTagged(isolate_root, IntPtrConstant(offset)));
}
Node* CodeAssembler::Store(Node* base, Node* value) {
return raw_assembler()->Store(MachineRepresentation::kTagged, base, value,
kFullWriteBarrier);
}
void CodeAssembler::OptimizedStoreField(MachineRepresentation rep,
TNode<HeapObject> object, int offset,
Node* value) {
raw_assembler()->OptimizedStoreField(rep, object, offset, value,
WriteBarrierKind::kFullWriteBarrier);
}
Reland: [csa] verify skipped write-barriers in MemoryOptimizer With very few exceptions, this verifies all skipped write-barriers in CSA and Torque, showing that the MemoryOptimizer together with some type information on the stored value are enough to avoid unsafe skipped write-barriers. Changes to CSA: SKIP_WRITE_BARRIER and Store*NoWriteBarrier are verified by the MemoryOptimizer by default. Type information about the stored values (TNode<Smi>) is exploited to safely skip write barriers for stored Smi values. In some cases, the code is re-structured to make it easier to consume for the MemoryOptimizer (manual branch and load elimination). Changes to the MemoryOptimizer: Improve the MemoryOptimizer to remove write barriers: - When the store happens to a CSA-generated InnerAllocate, by ignoring Bitcasts and additions. - When the stored value is the HeapConstant of an immortal immovable root. - When the stored value is a SmiConstant (recognized by BitcastToTaggedSigned). - Fast C-calls are treated as non-allocating. - Runtime calls can be white-listed as non-allocating. Remaining missing cases: - C++-style iterator loops with inner pointers. - Inner allocates that are reloaded from a field where they were just stored (for example an elements backing store). Load elimination would fix that. - Safe stored value types that cannot be expressed in CSA (e.g., Smi|Hole). We could handle that in Torque. - Double-aligned allocations, which are not lowered in the MemoryOptimizer but in CSA. Drive-by change: Avoid Smi suffix for StoreFixedArrayElement since this can be handled by overload resolution (in Torque and C++). Reland Change: Support pointer compression operands. R=jarin@chromium.org TBR=mvstanton@chromium.org Bug: v8:7793 Change-Id: I84e1831eb6bf9be14f36db3f8b485ee4fab6b22e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1612904 Auto-Submit: Tobias Tebbi <tebbi@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#61522}
2019-05-15 08:45:36 +00:00
void CodeAssembler::OptimizedStoreFieldAssertNoWriteBarrier(
MachineRepresentation rep, TNode<HeapObject> object, int offset,
Node* value) {
raw_assembler()->OptimizedStoreField(rep, object, offset, value,
WriteBarrierKind::kAssertNoWriteBarrier);
}
void CodeAssembler::OptimizedStoreFieldUnsafeNoWriteBarrier(
MachineRepresentation rep, TNode<HeapObject> object, int offset,
Node* value) {
raw_assembler()->OptimizedStoreField(rep, object, offset, value,
WriteBarrierKind::kNoWriteBarrier);
}
void CodeAssembler::OptimizedStoreMap(TNode<HeapObject> object,
TNode<Map> map) {
raw_assembler()->OptimizedStoreMap(object, map);
}
Node* CodeAssembler::Store(Node* base, Node* offset, Node* value) {
return raw_assembler()->Store(MachineRepresentation::kTagged, base, offset,
value, kFullWriteBarrier);
}
Node* CodeAssembler::StoreEphemeronKey(Node* base, Node* offset, Node* value) {
return raw_assembler()->Store(MachineRepresentation::kTagged, base, offset,
value, kEphemeronKeyWriteBarrier);
}
Node* CodeAssembler::StoreNoWriteBarrier(MachineRepresentation rep, Node* base,
Node* value) {
Reland: [csa] verify skipped write-barriers in MemoryOptimizer With very few exceptions, this verifies all skipped write-barriers in CSA and Torque, showing that the MemoryOptimizer together with some type information on the stored value are enough to avoid unsafe skipped write-barriers. Changes to CSA: SKIP_WRITE_BARRIER and Store*NoWriteBarrier are verified by the MemoryOptimizer by default. Type information about the stored values (TNode<Smi>) is exploited to safely skip write barriers for stored Smi values. In some cases, the code is re-structured to make it easier to consume for the MemoryOptimizer (manual branch and load elimination). Changes to the MemoryOptimizer: Improve the MemoryOptimizer to remove write barriers: - When the store happens to a CSA-generated InnerAllocate, by ignoring Bitcasts and additions. - When the stored value is the HeapConstant of an immortal immovable root. - When the stored value is a SmiConstant (recognized by BitcastToTaggedSigned). - Fast C-calls are treated as non-allocating. - Runtime calls can be white-listed as non-allocating. Remaining missing cases: - C++-style iterator loops with inner pointers. - Inner allocates that are reloaded from a field where they were just stored (for example an elements backing store). Load elimination would fix that. - Safe stored value types that cannot be expressed in CSA (e.g., Smi|Hole). We could handle that in Torque. - Double-aligned allocations, which are not lowered in the MemoryOptimizer but in CSA. Drive-by change: Avoid Smi suffix for StoreFixedArrayElement since this can be handled by overload resolution (in Torque and C++). Reland Change: Support pointer compression operands. R=jarin@chromium.org TBR=mvstanton@chromium.org Bug: v8:7793 Change-Id: I84e1831eb6bf9be14f36db3f8b485ee4fab6b22e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1612904 Auto-Submit: Tobias Tebbi <tebbi@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#61522}
2019-05-15 08:45:36 +00:00
return raw_assembler()->Store(
rep, base, value,
CanBeTaggedPointer(rep) ? kAssertNoWriteBarrier : kNoWriteBarrier);
}
Node* CodeAssembler::StoreNoWriteBarrier(MachineRepresentation rep, Node* base,
Node* offset, Node* value) {
Reland: [csa] verify skipped write-barriers in MemoryOptimizer With very few exceptions, this verifies all skipped write-barriers in CSA and Torque, showing that the MemoryOptimizer together with some type information on the stored value are enough to avoid unsafe skipped write-barriers. Changes to CSA: SKIP_WRITE_BARRIER and Store*NoWriteBarrier are verified by the MemoryOptimizer by default. Type information about the stored values (TNode<Smi>) is exploited to safely skip write barriers for stored Smi values. In some cases, the code is re-structured to make it easier to consume for the MemoryOptimizer (manual branch and load elimination). Changes to the MemoryOptimizer: Improve the MemoryOptimizer to remove write barriers: - When the store happens to a CSA-generated InnerAllocate, by ignoring Bitcasts and additions. - When the stored value is the HeapConstant of an immortal immovable root. - When the stored value is a SmiConstant (recognized by BitcastToTaggedSigned). - Fast C-calls are treated as non-allocating. - Runtime calls can be white-listed as non-allocating. Remaining missing cases: - C++-style iterator loops with inner pointers. - Inner allocates that are reloaded from a field where they were just stored (for example an elements backing store). Load elimination would fix that. - Safe stored value types that cannot be expressed in CSA (e.g., Smi|Hole). We could handle that in Torque. - Double-aligned allocations, which are not lowered in the MemoryOptimizer but in CSA. Drive-by change: Avoid Smi suffix for StoreFixedArrayElement since this can be handled by overload resolution (in Torque and C++). Reland Change: Support pointer compression operands. R=jarin@chromium.org TBR=mvstanton@chromium.org Bug: v8:7793 Change-Id: I84e1831eb6bf9be14f36db3f8b485ee4fab6b22e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1612904 Auto-Submit: Tobias Tebbi <tebbi@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#61522}
2019-05-15 08:45:36 +00:00
return raw_assembler()->Store(
rep, base, offset, value,
CanBeTaggedPointer(rep) ? kAssertNoWriteBarrier : kNoWriteBarrier);
}
Node* CodeAssembler::UnsafeStoreNoWriteBarrier(MachineRepresentation rep,
Node* base, Node* value) {
return raw_assembler()->Store(rep, base, value, kNoWriteBarrier);
}
Node* CodeAssembler::UnsafeStoreNoWriteBarrier(MachineRepresentation rep,
Node* base, Node* offset,
Node* value) {
return raw_assembler()->Store(rep, base, offset, value, kNoWriteBarrier);
}
Node* CodeAssembler::StoreFullTaggedNoWriteBarrier(Node* base,
Node* tagged_value) {
return StoreNoWriteBarrier(MachineType::PointerRepresentation(), base,
BitcastTaggedToWord(tagged_value));
}
Node* CodeAssembler::StoreFullTaggedNoWriteBarrier(Node* base, Node* offset,
Node* tagged_value) {
return StoreNoWriteBarrier(MachineType::PointerRepresentation(), base, offset,
BitcastTaggedToWord(tagged_value));
}
Node* CodeAssembler::AtomicStore(MachineRepresentation rep, Node* base,
Node* offset, Node* value, Node* value_high) {
return raw_assembler()->AtomicStore(rep, base, offset, value, value_high);
}
#define ATOMIC_FUNCTION(name) \
Node* CodeAssembler::Atomic##name(MachineType type, Node* base, \
Node* offset, Node* value, \
Node* value_high) { \
return raw_assembler()->Atomic##name(type, base, offset, value, \
value_high); \
}
ATOMIC_FUNCTION(Exchange)
ATOMIC_FUNCTION(Add)
ATOMIC_FUNCTION(Sub)
ATOMIC_FUNCTION(And)
ATOMIC_FUNCTION(Or)
ATOMIC_FUNCTION(Xor)
#undef ATOMIC_FUNCTION
Node* CodeAssembler::AtomicCompareExchange(MachineType type, Node* base,
Node* offset, Node* old_value,
Node* new_value,
Node* old_value_high,
Node* new_value_high) {
return raw_assembler()->AtomicCompareExchange(
type, base, offset, old_value, old_value_high, new_value, new_value_high);
}
Node* CodeAssembler::StoreRoot(RootIndex root_index, Node* value) {
DCHECK(!RootsTable::IsImmortalImmovable(root_index));
Node* isolate_root =
ExternalConstant(ExternalReference::isolate_root(isolate()));
int offset = IsolateData::root_slot_offset(root_index);
return StoreFullTaggedNoWriteBarrier(isolate_root, IntPtrConstant(offset),
value);
}
Node* CodeAssembler::Retain(Node* value) {
return raw_assembler()->Retain(value);
}
Node* CodeAssembler::ChangeTaggedToCompressed(Node* tagged) {
return raw_assembler()->ChangeTaggedToCompressed(tagged);
}
Node* CodeAssembler::ChangeCompressedToTagged(Node* compressed) {
return raw_assembler()->ChangeCompressedToTagged(compressed);
}
Node* CodeAssembler::Projection(int index, Node* value) {
DCHECK_LT(index, value->op()->ValueOutputCount());
return raw_assembler()->Projection(index, value);
}
void CodeAssembler::GotoIfException(Node* node, Label* if_exception,
Variable* exception_var) {
if (if_exception == nullptr) {
// If no handler is supplied, don't add continuations
return;
}
// No catch handlers should be active if we're using catch labels
DCHECK_EQ(state()->exception_handler_labels_.size(), 0);
DCHECK(!node->op()->HasProperty(Operator::kNoThrow));
Label success(this), exception(this, Label::kDeferred);
success.MergeVariables();
exception.MergeVariables();
raw_assembler()->Continuations(node, success.label_, exception.label_);
Bind(&exception);
const Operator* op = raw_assembler()->common()->IfException();
Node* exception_value = raw_assembler()->AddNode(op, node, node);
if (exception_var != nullptr) {
exception_var->Bind(exception_value);
}
Goto(if_exception);
Bind(&success);
raw_assembler()->AddNode(raw_assembler()->common()->IfSuccess(), node);
}
TNode<HeapObject> CodeAssembler::OptimizedAllocate(
TNode<IntPtrT> size, AllocationType allocation,
AllowLargeObjects allow_large_objects) {
return UncheckedCast<HeapObject>(raw_assembler()->OptimizedAllocate(
size, allocation, allow_large_objects));
}
void CodeAssembler::HandleException(Node* node) {
if (state_->exception_handler_labels_.size() == 0) return;
CodeAssemblerExceptionHandlerLabel* label =
state_->exception_handler_labels_.back();
if (node->op()->HasProperty(Operator::kNoThrow)) {
return;
}
Label success(this), exception(this, Label::kDeferred);
success.MergeVariables();
exception.MergeVariables();
raw_assembler()->Continuations(node, success.label_, exception.label_);
Bind(&exception);
const Operator* op = raw_assembler()->common()->IfException();
Node* exception_value = raw_assembler()->AddNode(op, node, node);
label->AddInputs({UncheckedCast<Object>(exception_value)});
Goto(label->plain_label());
Bind(&success);
raw_assembler()->AddNode(raw_assembler()->common()->IfSuccess(), node);
}
namespace {
template <size_t kMaxSize>
class NodeArray {
public:
void Add(Node* node) {
DCHECK_GT(kMaxSize, size());
*ptr_++ = node;
}
Node* const* data() const { return arr_; }
int size() const { return static_cast<int>(ptr_ - arr_); }
private:
Node* arr_[kMaxSize];
Node** ptr_ = arr_;
};
} // namespace
TNode<Object> CodeAssembler::CallRuntimeImpl(
Runtime::FunctionId function, TNode<Object> context,
std::initializer_list<TNode<Object>> args) {
int result_size = Runtime::FunctionForId(function)->result_size;
TNode<Code> centry =
HeapConstant(CodeFactory::RuntimeCEntry(isolate(), result_size));
return CallRuntimeWithCEntryImpl(function, centry, context, args);
}
TNode<Object> CodeAssembler::CallRuntimeWithCEntryImpl(
Runtime::FunctionId function, TNode<Code> centry, TNode<Object> context,
std::initializer_list<TNode<Object>> args) {
constexpr size_t kMaxNumArgs = 6;
DCHECK_GE(kMaxNumArgs, args.size());
int argc = static_cast<int>(args.size());
auto call_descriptor = Linkage::GetRuntimeCallDescriptor(
zone(), function, argc, Operator::kNoProperties,
Reland: [csa] verify skipped write-barriers in MemoryOptimizer With very few exceptions, this verifies all skipped write-barriers in CSA and Torque, showing that the MemoryOptimizer together with some type information on the stored value are enough to avoid unsafe skipped write-barriers. Changes to CSA: SKIP_WRITE_BARRIER and Store*NoWriteBarrier are verified by the MemoryOptimizer by default. Type information about the stored values (TNode<Smi>) is exploited to safely skip write barriers for stored Smi values. In some cases, the code is re-structured to make it easier to consume for the MemoryOptimizer (manual branch and load elimination). Changes to the MemoryOptimizer: Improve the MemoryOptimizer to remove write barriers: - When the store happens to a CSA-generated InnerAllocate, by ignoring Bitcasts and additions. - When the stored value is the HeapConstant of an immortal immovable root. - When the stored value is a SmiConstant (recognized by BitcastToTaggedSigned). - Fast C-calls are treated as non-allocating. - Runtime calls can be white-listed as non-allocating. Remaining missing cases: - C++-style iterator loops with inner pointers. - Inner allocates that are reloaded from a field where they were just stored (for example an elements backing store). Load elimination would fix that. - Safe stored value types that cannot be expressed in CSA (e.g., Smi|Hole). We could handle that in Torque. - Double-aligned allocations, which are not lowered in the MemoryOptimizer but in CSA. Drive-by change: Avoid Smi suffix for StoreFixedArrayElement since this can be handled by overload resolution (in Torque and C++). Reland Change: Support pointer compression operands. R=jarin@chromium.org TBR=mvstanton@chromium.org Bug: v8:7793 Change-Id: I84e1831eb6bf9be14f36db3f8b485ee4fab6b22e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1612904 Auto-Submit: Tobias Tebbi <tebbi@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#61522}
2019-05-15 08:45:36 +00:00
Runtime::MayAllocate(function) ? CallDescriptor::kNoFlags
: CallDescriptor::kNoAllocate);
Reland: [refactoring] Remove the isolate from signatures of ExternalReferences I missed one required change which was hidden behind an #if. The fix is in the diff between Patch 1 and Patch 3. Original message: In this CL I remove the isolate from signatures of ExternalReference accessor functions where the isolate is not used. The uses of the isolate were already removed in previous CLs. Changes: * I split the ExternalReference list in external-reference.h into those which need the isolate for initialization and those which do not. * I removed the public constructors and replaced them by ExternalReference::Create(). The reason is to separate external creation more clearly from internal creation, because externally created ExternalReferences sometimes need redirection, whereas internally created ExternalReferences are just stored as they are. In addition, by removing the isolate from the signature of the public constructors, they suddenly exactly matched the interal constructor. * Replace all uses of the public constructors with ExternalReference::Create(). * Remove the isolate from all call sites where necessary. This is a step towards making WebAssembly compilation independent of the isolate. R=mstarzinger@chromium.org Bug: v8:7570 Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng Change-Id: I750c162f5d58ed32e866722b0db920f8b9bd8057 Reviewed-on: https://chromium-review.googlesource.com/1026673 Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#52777}
2018-04-25 07:28:14 +00:00
Node* ref = ExternalConstant(ExternalReference::Create(function));
Node* arity = Int32Constant(argc);
NodeArray<kMaxNumArgs + 4> inputs;
inputs.Add(centry);
for (auto arg : args) inputs.Add(arg);
inputs.Add(ref);
inputs.Add(arity);
inputs.Add(context);
CallPrologue();
Node* return_value =
raw_assembler()->CallN(call_descriptor, inputs.size(), inputs.data());
HandleException(return_value);
CallEpilogue();
return UncheckedCast<Object>(return_value);
}
void CodeAssembler::TailCallRuntimeImpl(
Runtime::FunctionId function, TNode<Int32T> arity, TNode<Object> context,
std::initializer_list<TNode<Object>> args) {
int result_size = Runtime::FunctionForId(function)->result_size;
TNode<Code> centry =
HeapConstant(CodeFactory::RuntimeCEntry(isolate(), result_size));
return TailCallRuntimeWithCEntryImpl(function, arity, centry, context, args);
}
void CodeAssembler::TailCallRuntimeWithCEntryImpl(
Runtime::FunctionId function, TNode<Int32T> arity, TNode<Code> centry,
TNode<Object> context, std::initializer_list<TNode<Object>> args) {
constexpr size_t kMaxNumArgs = 6;
DCHECK_GE(kMaxNumArgs, args.size());
int argc = static_cast<int>(args.size());
auto call_descriptor = Linkage::GetRuntimeCallDescriptor(
zone(), function, argc, Operator::kNoProperties,
CallDescriptor::kNoFlags);
Reland: [refactoring] Remove the isolate from signatures of ExternalReferences I missed one required change which was hidden behind an #if. The fix is in the diff between Patch 1 and Patch 3. Original message: In this CL I remove the isolate from signatures of ExternalReference accessor functions where the isolate is not used. The uses of the isolate were already removed in previous CLs. Changes: * I split the ExternalReference list in external-reference.h into those which need the isolate for initialization and those which do not. * I removed the public constructors and replaced them by ExternalReference::Create(). The reason is to separate external creation more clearly from internal creation, because externally created ExternalReferences sometimes need redirection, whereas internally created ExternalReferences are just stored as they are. In addition, by removing the isolate from the signature of the public constructors, they suddenly exactly matched the interal constructor. * Replace all uses of the public constructors with ExternalReference::Create(). * Remove the isolate from all call sites where necessary. This is a step towards making WebAssembly compilation independent of the isolate. R=mstarzinger@chromium.org Bug: v8:7570 Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng Change-Id: I750c162f5d58ed32e866722b0db920f8b9bd8057 Reviewed-on: https://chromium-review.googlesource.com/1026673 Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#52777}
2018-04-25 07:28:14 +00:00
Node* ref = ExternalConstant(ExternalReference::Create(function));
NodeArray<kMaxNumArgs + 4> inputs;
inputs.Add(centry);
for (auto arg : args) inputs.Add(arg);
inputs.Add(ref);
inputs.Add(arity);
inputs.Add(context);
raw_assembler()->TailCallN(call_descriptor, inputs.size(), inputs.data());
}
Node* CodeAssembler::CallStubN(StubCallMode call_mode,
const CallInterfaceDescriptor& descriptor,
size_t result_size, int input_count,
Node* const* inputs) {
DCHECK(call_mode == StubCallMode::kCallCodeObject ||
call_mode == StubCallMode::kCallBuiltinPointer);
// implicit nodes are target and optionally context.
int implicit_nodes = descriptor.HasContextParameter() ? 2 : 1;
DCHECK_LE(implicit_nodes, input_count);
int argc = input_count - implicit_nodes;
DCHECK_LE(descriptor.GetParameterCount(), argc);
// Extra arguments not mentioned in the descriptor are passed on the stack.
int stack_parameter_count = argc - descriptor.GetRegisterParameterCount();
DCHECK_LE(descriptor.GetStackParameterCount(), stack_parameter_count);
DCHECK_EQ(result_size, descriptor.GetReturnCount());
auto call_descriptor = Linkage::GetStubCallDescriptor(
zone(), descriptor, stack_parameter_count, CallDescriptor::kNoFlags,
Operator::kNoProperties, call_mode);
CallPrologue();
Node* return_value =
raw_assembler()->CallN(call_descriptor, input_count, inputs);
HandleException(return_value);
CallEpilogue();
return return_value;
}
void CodeAssembler::TailCallStubImpl(const CallInterfaceDescriptor& descriptor,
TNode<Code> target, TNode<Object> context,
std::initializer_list<Node*> args) {
constexpr size_t kMaxNumArgs = 11;
DCHECK_GE(kMaxNumArgs, args.size());
DCHECK_EQ(descriptor.GetParameterCount(), args.size());
auto call_descriptor = Linkage::GetStubCallDescriptor(
zone(), descriptor, descriptor.GetStackParameterCount(),
CallDescriptor::kNoFlags, Operator::kNoProperties);
NodeArray<kMaxNumArgs + 2> inputs;
inputs.Add(target);
for (auto arg : args) inputs.Add(arg);
if (descriptor.HasContextParameter()) {
inputs.Add(context);
}
raw_assembler()->TailCallN(call_descriptor, inputs.size(), inputs.data());
}
Node* CodeAssembler::CallStubRImpl(StubCallMode call_mode,
const CallInterfaceDescriptor& descriptor,
size_t result_size, Node* target,
SloppyTNode<Object> context,
std::initializer_list<Node*> args) {
DCHECK(call_mode == StubCallMode::kCallCodeObject ||
call_mode == StubCallMode::kCallBuiltinPointer);
constexpr size_t kMaxNumArgs = 10;
DCHECK_GE(kMaxNumArgs, args.size());
NodeArray<kMaxNumArgs + 2> inputs;
inputs.Add(target);
for (auto arg : args) inputs.Add(arg);
if (descriptor.HasContextParameter()) {
inputs.Add(context);
}
return CallStubN(call_mode, descriptor, result_size, inputs.size(),
inputs.data());
}
Node* CodeAssembler::TailCallStubThenBytecodeDispatchImpl(
const CallInterfaceDescriptor& descriptor, Node* target, Node* context,
std::initializer_list<Node*> args) {
constexpr size_t kMaxNumArgs = 6;
DCHECK_GE(kMaxNumArgs, args.size());
DCHECK_LE(descriptor.GetParameterCount(), args.size());
int argc = static_cast<int>(args.size());
// Extra arguments not mentioned in the descriptor are passed on the stack.
int stack_parameter_count = argc - descriptor.GetRegisterParameterCount();
DCHECK_LE(descriptor.GetStackParameterCount(), stack_parameter_count);
auto call_descriptor = Linkage::GetStubCallDescriptor(
zone(), descriptor, stack_parameter_count, CallDescriptor::kNoFlags,
Operator::kNoProperties);
NodeArray<kMaxNumArgs + 2> inputs;
inputs.Add(target);
for (auto arg : args) inputs.Add(arg);
inputs.Add(context);
return raw_assembler()->TailCallN(call_descriptor, inputs.size(),
inputs.data());
}
template <class... TArgs>
Node* CodeAssembler::TailCallBytecodeDispatch(
const CallInterfaceDescriptor& descriptor, Node* target, TArgs... args) {
DCHECK_EQ(descriptor.GetParameterCount(), sizeof...(args));
auto call_descriptor = Linkage::GetBytecodeDispatchCallDescriptor(
zone(), descriptor, descriptor.GetStackParameterCount());
Node* nodes[] = {target, args...};
CHECK_EQ(descriptor.GetParameterCount() + 1, arraysize(nodes));
return raw_assembler()->TailCallN(call_descriptor, arraysize(nodes), nodes);
}
// Instantiate TailCallBytecodeDispatch() for argument counts used by
// CSA-generated code
template V8_EXPORT_PRIVATE Node* CodeAssembler::TailCallBytecodeDispatch(
const CallInterfaceDescriptor& descriptor, Node* target, Node*, Node*,
Node*, Node*);
TNode<Object> CodeAssembler::TailCallJSCode(TNode<Code> code,
TNode<Context> context,
TNode<JSFunction> function,
TNode<Object> new_target,
TNode<Int32T> arg_count) {
JSTrampolineDescriptor descriptor;
auto call_descriptor = Linkage::GetStubCallDescriptor(
zone(), descriptor, descriptor.GetStackParameterCount(),
CallDescriptor::kFixedTargetRegister, Operator::kNoProperties);
Node* nodes[] = {code, function, new_target, arg_count, context};
CHECK_EQ(descriptor.GetParameterCount() + 2, arraysize(nodes));
return UncheckedCast<Object>(
raw_assembler()->TailCallN(call_descriptor, arraysize(nodes), nodes));
}
Node* CodeAssembler::CallCFunctionN(Signature<MachineType>* signature,
int input_count, Node* const* inputs) {
auto call_descriptor = Linkage::GetSimplifiedCDescriptor(zone(), signature);
return raw_assembler()->CallN(call_descriptor, input_count, inputs);
}
Node* CodeAssembler::CallCFunction(
Node* function, MachineType return_type,
std::initializer_list<CodeAssembler::CFunctionArg> args) {
return raw_assembler()->CallCFunction(function, return_type, args);
}
Node* CodeAssembler::CallCFunctionWithCallerSavedRegisters(
Node* function, MachineType return_type, SaveFPRegsMode mode,
std::initializer_list<CodeAssembler::CFunctionArg> args) {
DCHECK(return_type.LessThanOrEqualPointerSize());
return raw_assembler()->CallCFunctionWithCallerSavedRegisters(
function, return_type, mode, args);
}
void CodeAssembler::Goto(Label* label) {
label->MergeVariables();
raw_assembler()->Goto(label->label_);
}
void CodeAssembler::GotoIf(SloppyTNode<IntegralT> condition,
Label* true_label) {
Label false_label(this);
Branch(condition, true_label, &false_label);
Bind(&false_label);
}
void CodeAssembler::GotoIfNot(SloppyTNode<IntegralT> condition,
Label* false_label) {
Label true_label(this);
Branch(condition, &true_label, false_label);
Bind(&true_label);
}
void CodeAssembler::Branch(SloppyTNode<IntegralT> condition, Label* true_label,
Label* false_label) {
int32_t constant;
if (ToInt32Constant(condition, constant)) {
if ((true_label->is_used() || true_label->is_bound()) &&
(false_label->is_used() || false_label->is_bound())) {
return Goto(constant ? true_label : false_label);
}
}
true_label->MergeVariables();
false_label->MergeVariables();
return raw_assembler()->Branch(condition, true_label->label_,
false_label->label_);
}
void CodeAssembler::Branch(TNode<BoolT> condition,
const std::function<void()>& true_body,
const std::function<void()>& false_body) {
int32_t constant;
if (ToInt32Constant(condition, constant)) {
return constant ? true_body() : false_body();
}
Label vtrue(this), vfalse(this);
Branch(condition, &vtrue, &vfalse);
Bind(&vtrue);
true_body();
Bind(&vfalse);
false_body();
}
void CodeAssembler::Branch(TNode<BoolT> condition, Label* true_label,
const std::function<void()>& false_body) {
int32_t constant;
if (ToInt32Constant(condition, constant)) {
return constant ? Goto(true_label) : false_body();
}
Label vfalse(this);
Branch(condition, true_label, &vfalse);
Bind(&vfalse);
false_body();
}
void CodeAssembler::Branch(TNode<BoolT> condition,
const std::function<void()>& true_body,
Label* false_label) {
int32_t constant;
if (ToInt32Constant(condition, constant)) {
return constant ? true_body() : Goto(false_label);
}
Label vtrue(this);
Branch(condition, &vtrue, false_label);
Bind(&vtrue);
true_body();
}
void CodeAssembler::Switch(Node* index, Label* default_label,
const int32_t* case_values, Label** case_labels,
size_t case_count) {
RawMachineLabel** labels =
new (zone()->New(sizeof(RawMachineLabel*) * case_count))
RawMachineLabel*[case_count];
for (size_t i = 0; i < case_count; ++i) {
labels[i] = case_labels[i]->label_;
case_labels[i]->MergeVariables();
}
default_label->MergeVariables();
return raw_assembler()->Switch(index, default_label->label_, case_values,
labels, case_count);
}
bool CodeAssembler::UnalignedLoadSupported(MachineRepresentation rep) const {
return raw_assembler()->machine()->UnalignedLoadSupported(rep);
}
bool CodeAssembler::UnalignedStoreSupported(MachineRepresentation rep) const {
return raw_assembler()->machine()->UnalignedStoreSupported(rep);
}
// RawMachineAssembler delegate helpers:
Isolate* CodeAssembler::isolate() const { return raw_assembler()->isolate(); }
Factory* CodeAssembler::factory() const { return isolate()->factory(); }
Zone* CodeAssembler::zone() const { return raw_assembler()->zone(); }
bool CodeAssembler::IsExceptionHandlerActive() const {
return state_->exception_handler_labels_.size() != 0;
}
RawMachineAssembler* CodeAssembler::raw_assembler() const {
return state_->raw_assembler_.get();
}
// The core implementation of Variable is stored through an indirection so
// that it can outlive the often block-scoped Variable declarations. This is
// needed to ensure that variable binding and merging through phis can
// properly be verified.
class CodeAssemblerVariable::Impl : public ZoneObject {
public:
explicit Impl(MachineRepresentation rep, CodeAssemblerState::VariableId id)
:
#if DEBUG
debug_info_(AssemblerDebugInfo(nullptr, nullptr, -1)),
#endif
value_(nullptr),
rep_(rep),
var_id_(id) {
}
#if DEBUG
AssemblerDebugInfo debug_info() const { return debug_info_; }
void set_debug_info(AssemblerDebugInfo debug_info) {
debug_info_ = debug_info;
}
AssemblerDebugInfo debug_info_;
#endif // DEBUG
bool operator<(const CodeAssemblerVariable::Impl& other) const {
return var_id_ < other.var_id_;
}
Node* value_;
MachineRepresentation rep_;
CodeAssemblerState::VariableId var_id_;
};
bool CodeAssemblerVariable::ImplComparator::operator()(
const CodeAssemblerVariable::Impl* a,
const CodeAssemblerVariable::Impl* b) const {
return *a < *b;
}
CodeAssemblerVariable::CodeAssemblerVariable(CodeAssembler* assembler,
MachineRepresentation rep)
: impl_(new (assembler->zone())
Impl(rep, assembler->state()->NextVariableId())),
state_(assembler->state()) {
state_->variables_.insert(impl_);
}
CodeAssemblerVariable::CodeAssemblerVariable(CodeAssembler* assembler,
MachineRepresentation rep,
Node* initial_value)
: CodeAssemblerVariable(assembler, rep) {
Bind(initial_value);
}
#if DEBUG
CodeAssemblerVariable::CodeAssemblerVariable(CodeAssembler* assembler,
AssemblerDebugInfo debug_info,
MachineRepresentation rep)
: impl_(new (assembler->zone())
Impl(rep, assembler->state()->NextVariableId())),
state_(assembler->state()) {
impl_->set_debug_info(debug_info);
state_->variables_.insert(impl_);
}
CodeAssemblerVariable::CodeAssemblerVariable(CodeAssembler* assembler,
AssemblerDebugInfo debug_info,
MachineRepresentation rep,
Node* initial_value)
: CodeAssemblerVariable(assembler, debug_info, rep) {
impl_->set_debug_info(debug_info);
Bind(initial_value);
}
#endif // DEBUG
CodeAssemblerVariable::~CodeAssemblerVariable() {
state_->variables_.erase(impl_);
}
void CodeAssemblerVariable::Bind(Node* value) { impl_->value_ = value; }
Node* CodeAssemblerVariable::value() const {
#if DEBUG
if (!IsBound()) {
std::stringstream str;
str << "#Use of unbound variable:"
<< "#\n Variable: " << *this << "#\n Current Block: ";
state_->PrintCurrentBlock(str);
FATAL("%s", str.str().c_str());
}
if (!state_->InsideBlock()) {
std::stringstream str;
str << "#Accessing variable value outside a block:"
<< "#\n Variable: " << *this;
FATAL("%s", str.str().c_str());
}
#endif // DEBUG
return impl_->value_;
}
MachineRepresentation CodeAssemblerVariable::rep() const { return impl_->rep_; }
bool CodeAssemblerVariable::IsBound() const { return impl_->value_ != nullptr; }
std::ostream& operator<<(std::ostream& os,
const CodeAssemblerVariable::Impl& impl) {
#if DEBUG
AssemblerDebugInfo info = impl.debug_info();
if (info.name) os << "V" << info;
#endif // DEBUG
return os;
}
std::ostream& operator<<(std::ostream& os,
const CodeAssemblerVariable& variable) {
os << *variable.impl_;
return os;
}
CodeAssemblerLabel::CodeAssemblerLabel(CodeAssembler* assembler,
size_t vars_count,
CodeAssemblerVariable* const* vars,
CodeAssemblerLabel::Type type)
: bound_(false),
merge_count_(0),
state_(assembler->state()),
label_(nullptr) {
void* buffer = assembler->zone()->New(sizeof(RawMachineLabel));
label_ = new (buffer)
RawMachineLabel(type == kDeferred ? RawMachineLabel::kDeferred
: RawMachineLabel::kNonDeferred);
for (size_t i = 0; i < vars_count; ++i) {
variable_phis_[vars[i]->impl_] = nullptr;
}
}
CodeAssemblerLabel::~CodeAssemblerLabel() { label_->~RawMachineLabel(); }
void CodeAssemblerLabel::MergeVariables() {
++merge_count_;
for (CodeAssemblerVariable::Impl* var : state_->variables_) {
size_t count = 0;
Node* node = var->value_;
if (node != nullptr) {
auto i = variable_merges_.find(var);
if (i != variable_merges_.end()) {
i->second.push_back(node);
count = i->second.size();
} else {
count = 1;
variable_merges_[var] = std::vector<Node*>(1, node);
}
}
// If the following asserts, then you've jumped to a label without a bound
// variable along that path that expects to merge its value into a phi.
DCHECK(variable_phis_.find(var) == variable_phis_.end() ||
count == merge_count_);
USE(count);
// If the label is already bound, we already know the set of variables to
// merge and phi nodes have already been created.
if (bound_) {
auto phi = variable_phis_.find(var);
if (phi != variable_phis_.end()) {
DCHECK_NOT_NULL(phi->second);
state_->raw_assembler_->AppendPhiInput(phi->second, node);
} else {
auto i = variable_merges_.find(var);
if (i != variable_merges_.end()) {
// If the following assert fires, then you've declared a variable that
// has the same bound value along all paths up until the point you
// bound this label, but then later merged a path with a new value for
// the variable after the label bind (it's not possible to add phis to
// the bound label after the fact, just make sure to list the variable
// in the label's constructor's list of merged variables).
#if DEBUG
if (find_if(i->second.begin(), i->second.end(),
[node](Node* e) -> bool { return node != e; }) !=
i->second.end()) {
std::stringstream str;
str << "Unmerged variable found when jumping to block. \n"
<< "# Variable: " << *var;
if (bound_) {
str << "\n# Target block: " << *label_->block();
}
str << "\n# Current Block: ";
state_->PrintCurrentBlock(str);
FATAL("%s", str.str().c_str());
}
#endif // DEBUG
}
}
}
}
}
#if DEBUG
void CodeAssemblerLabel::Bind(AssemblerDebugInfo debug_info) {
if (bound_) {
std::stringstream str;
str << "Cannot bind the same label twice:"
<< "\n# current: " << debug_info
<< "\n# previous: " << *label_->block();
FATAL("%s", str.str().c_str());
}
2019-02-18 11:41:56 +00:00
if (FLAG_enable_source_at_csa_bind) {
state_->raw_assembler_->SetSourcePosition(debug_info.file, debug_info.line);
}
state_->raw_assembler_->Bind(label_, debug_info);
UpdateVariablesAfterBind();
}
#endif // DEBUG
void CodeAssemblerLabel::Bind() {
DCHECK(!bound_);
state_->raw_assembler_->Bind(label_);
UpdateVariablesAfterBind();
}
void CodeAssemblerLabel::UpdateVariablesAfterBind() {
// Make sure that all variables that have changed along any path up to this
// point are marked as merge variables.
for (auto var : state_->variables_) {
Node* shared_value = nullptr;
auto i = variable_merges_.find(var);
if (i != variable_merges_.end()) {
for (auto value : i->second) {
DCHECK_NOT_NULL(value);
if (value != shared_value) {
if (shared_value == nullptr) {
shared_value = value;
} else {
variable_phis_[var] = nullptr;
}
}
}
}
}
for (auto var : variable_phis_) {
CodeAssemblerVariable::Impl* var_impl = var.first;
auto i = variable_merges_.find(var_impl);
#if DEBUG
bool not_found = i == variable_merges_.end();
if (not_found || i->second.size() != merge_count_) {
std::stringstream str;
str << "A variable that has been marked as beeing merged at the label"
<< "\n# doesn't have a bound value along all of the paths that "
<< "\n# have been merged into the label up to this point."
<< "\n#"
<< "\n# This can happen in the following cases:"
<< "\n# - By explicitly marking it so in the label constructor"
<< "\n# - By having seen different bound values at branches"
<< "\n#"
<< "\n# Merge count: expected=" << merge_count_
<< " vs. found=" << (not_found ? 0 : i->second.size())
<< "\n# Variable: " << *var_impl
<< "\n# Current Block: " << *label_->block();
FATAL("%s", str.str().c_str());
}
#endif // DEBUG
Node* phi = state_->raw_assembler_->Phi(
var.first->rep_, static_cast<int>(merge_count_), &(i->second[0]));
variable_phis_[var_impl] = phi;
}
// Bind all variables to a merge phi, the common value along all paths or
// null.
for (auto var : state_->variables_) {
auto i = variable_phis_.find(var);
if (i != variable_phis_.end()) {
var->value_ = i->second;
} else {
auto j = variable_merges_.find(var);
if (j != variable_merges_.end() && j->second.size() == merge_count_) {
var->value_ = j->second.back();
} else {
var->value_ = nullptr;
}
}
}
bound_ = true;
}
void CodeAssemblerParameterizedLabelBase::AddInputs(std::vector<Node*> inputs) {
if (!phi_nodes_.empty()) {
DCHECK_EQ(inputs.size(), phi_nodes_.size());
for (size_t i = 0; i < inputs.size(); ++i) {
// We use {nullptr} as a sentinel for an uninitialized value.
if (phi_nodes_[i] == nullptr) continue;
state_->raw_assembler_->AppendPhiInput(phi_nodes_[i], inputs[i]);
}
} else {
DCHECK_EQ(inputs.size(), phi_inputs_.size());
for (size_t i = 0; i < inputs.size(); ++i) {
phi_inputs_[i].push_back(inputs[i]);
}
}
}
Node* CodeAssemblerParameterizedLabelBase::CreatePhi(
MachineRepresentation rep, const std::vector<Node*>& inputs) {
for (Node* input : inputs) {
// We use {nullptr} as a sentinel for an uninitialized value. We must not
// create phi nodes for these.
if (input == nullptr) return nullptr;
}
return state_->raw_assembler_->Phi(rep, static_cast<int>(inputs.size()),
&inputs.front());
}
const std::vector<Node*>& CodeAssemblerParameterizedLabelBase::CreatePhis(
std::vector<MachineRepresentation> representations) {
DCHECK(is_used());
DCHECK(phi_nodes_.empty());
phi_nodes_.reserve(phi_inputs_.size());
DCHECK_EQ(representations.size(), phi_inputs_.size());
for (size_t i = 0; i < phi_inputs_.size(); ++i) {
phi_nodes_.push_back(CreatePhi(representations[i], phi_inputs_[i]));
}
return phi_nodes_;
}
void CodeAssemblerState::PushExceptionHandler(
CodeAssemblerExceptionHandlerLabel* label) {
exception_handler_labels_.push_back(label);
}
void CodeAssemblerState::PopExceptionHandler() {
exception_handler_labels_.pop_back();
}
CodeAssemblerScopedExceptionHandler::CodeAssemblerScopedExceptionHandler(
CodeAssembler* assembler, CodeAssemblerExceptionHandlerLabel* label)
: has_handler_(label != nullptr),
assembler_(assembler),
compatibility_label_(nullptr),
exception_(nullptr) {
if (has_handler_) {
assembler_->state()->PushExceptionHandler(label);
}
}
CodeAssemblerScopedExceptionHandler::CodeAssemblerScopedExceptionHandler(
CodeAssembler* assembler, CodeAssemblerLabel* label,
TypedCodeAssemblerVariable<Object>* exception)
: has_handler_(label != nullptr),
assembler_(assembler),
compatibility_label_(label),
exception_(exception) {
if (has_handler_) {
label_ = base::make_unique<CodeAssemblerExceptionHandlerLabel>(
assembler, CodeAssemblerLabel::kDeferred);
assembler_->state()->PushExceptionHandler(label_.get());
}
}
CodeAssemblerScopedExceptionHandler::~CodeAssemblerScopedExceptionHandler() {
if (has_handler_) {
assembler_->state()->PopExceptionHandler();
}
if (label_ && label_->is_used()) {
CodeAssembler::Label skip(assembler_);
bool inside_block = assembler_->state()->InsideBlock();
if (inside_block) {
assembler_->Goto(&skip);
}
TNode<Object> e;
assembler_->Bind(label_.get(), &e);
*exception_ = e;
assembler_->Goto(compatibility_label_);
if (inside_block) {
assembler_->Bind(&skip);
}
}
}
} // namespace compiler
Address CheckObjectType(Address raw_value, Address raw_type,
Address raw_location) {
#ifdef DEBUG
Object value(raw_value);
Smi type(raw_type);
String location = String::cast(Object(raw_location));
const char* expected;
switch (static_cast<ObjectType>(type->value())) {
#define TYPE_CASE(Name) \
case ObjectType::k##Name: \
if (value->Is##Name()) return Smi::FromInt(0).ptr(); \
expected = #Name; \
break;
#define TYPE_STRUCT_CASE(NAME, Name, name) \
case ObjectType::k##Name: \
if (value->Is##Name()) return Smi::FromInt(0).ptr(); \
expected = #Name; \
break;
TYPE_CASE(Object)
TYPE_CASE(Smi)
TYPE_CASE(HeapObject)
OBJECT_TYPE_LIST(TYPE_CASE)
HEAP_OBJECT_TYPE_LIST(TYPE_CASE)
STRUCT_LIST(TYPE_STRUCT_CASE)
#undef TYPE_CASE
#undef TYPE_STRUCT_CASE
}
std::stringstream value_description;
value->Print(value_description);
FATAL(
"Type cast failed in %s\n"
" Expected %s but found %s",
location->ToAsciiArray(), expected, value_description.str().c_str());
#else
UNREACHABLE();
#endif
}
} // namespace internal
} // namespace v8