v8/test/cctest/compiler/codegen-tester.h
Jakob Gruber c7cb9beca1 Reland "Reland "[deoptimizer] Change deopt entries into builtins""
This is a reland of fbfa9bf4ec

The arm64 was missing proper codegen for CFI, thus sizes were off.

Original change's description:
> Reland "[deoptimizer] Change deopt entries into builtins"
>
> This is a reland of 7f58ced72e
>
> It fixes the different exit size emitted on x64/Atom CPUs due to
> performance tuning in TurboAssembler::Call. Additionally, add
> cctests to verify the fixed size exits.
>
> Original change's description:
> > [deoptimizer] Change deopt entries into builtins
> >
> > While the overall goal of this commit is to change deoptimization
> > entries into builtins, there are multiple related things happening:
> >
> > - Deoptimization entries, formerly stubs (i.e. Code objects generated
> >   at runtime, guaranteed to be immovable), have been converted into
> >   builtins. The major restriction is that we now need to preserve the
> >   kRootRegister, which was formerly used on most architectures to pass
> >   the deoptimization id. The solution differs based on platform.
> > - Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
> > - Removed heap/ support for immovable Code generation.
> > - Removed the DeserializerData class (no longer needed).
> > - arm64: to preserve 4-byte deopt exits, introduced a new optimization
> >   in which the final jump to the deoptimization entry is generated
> >   once per Code object, and deopt exits can continue to emit a
> >   near-call.
> > - arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
> >   sizes by 4/8, 5, and 5 bytes, respectively.
> >
> > On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
> > by using the same strategy as on arm64 (recalc deopt id from return
> > address). Before:
> >
> >  e300a002       movw r10, <id>
> >  e59fc024       ldr ip, [pc, <entry offset>]
> >  e12fff3c       blx ip
> >
> > After:
> >
> >  e59acb35       ldr ip, [r10, <entry offset>]
> >  e12fff3c       blx ip
> >
> > On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
> > with CFI). Additionally, up to 4 builtin jumps are emitted per Code
> > object (max 32 bytes added overhead per Code object). Before:
> >
> >  9401cdae       bl <entry offset>
> >
> > After:
> >
> >  # eager deoptimization entry jump.
> >  f95b1f50       ldr x16, [x26, <eager entry offset>]
> >  d61f0200       br x16
> >  # lazy deoptimization entry jump.
> >  f95b2b50       ldr x16, [x26, <lazy entry offset>]
> >  d61f0200       br x16
> >  # the deopt exit.
> >  97fffffc       bl <eager deoptimization entry jump offset>
> >
> > On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
> >
> >  bb00000000     mov ebx,<id>
> >  e825f5372b     call <entry>
> >
> > After:
> >
> >  e8ea2256ba     call <entry>
> >
> > On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
> >
> >  49c7c511000000 REX.W movq r13,<id>
> >  e8ea2f0700     call <entry>
> >
> > After:
> >
> >  41ff9560360000 call [r13+<entry offset>]
> >
> > Bug: v8:8661,v8:8768
> > Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
> > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
> > Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> > Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#70597}
>
> Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
> Bug: v8:8661,v8:8768,chromium:1140165
> Change-Id: Ibcd5c39c58a70bf2b2ac221aa375fc68d495e144
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2485506
> Reviewed-by: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70655}

Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
Bug: v8:8661
Bug: v8:8768
Bug: chromium:1140165
Change-Id: I471cc94fc085e527dc9bfb5a84b96bd907c2333f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2488682
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70672}
2020-10-21 06:01:38 +00:00

454 lines
16 KiB
C++

// Copyright 2014 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.
#ifndef V8_CCTEST_COMPILER_CODEGEN_TESTER_H_
#define V8_CCTEST_COMPILER_CODEGEN_TESTER_H_
#include "src/codegen/optimized-compilation-info.h"
#include "src/compiler/backend/instruction-selector.h"
#include "src/compiler/pipeline.h"
#include "src/compiler/raw-machine-assembler.h"
#include "src/execution/simulator.h"
#include "test/cctest/cctest.h"
#include "test/cctest/compiler/call-tester.h"
namespace v8 {
namespace internal {
namespace compiler {
template <typename ReturnType>
class RawMachineAssemblerTester : public HandleAndZoneScope,
public CallHelper<ReturnType>,
public RawMachineAssembler {
public:
template <typename... ParamMachTypes>
explicit RawMachineAssemblerTester(ParamMachTypes... p)
: HandleAndZoneScope(kCompressGraphZone),
CallHelper<ReturnType>(
main_isolate(),
CSignature::New(main_zone(), MachineTypeForC<ReturnType>(), p...)),
RawMachineAssembler(
main_isolate(), main_zone()->template New<Graph>(main_zone()),
Linkage::GetSimplifiedCDescriptor(
main_zone(),
CSignature::New(main_zone(), MachineTypeForC<ReturnType>(),
p...),
CallDescriptor::kInitializeRootRegister),
MachineType::PointerRepresentation(),
InstructionSelector::SupportedMachineOperatorFlags(),
InstructionSelector::AlignmentRequirements()) {}
template <typename... ParamMachTypes>
RawMachineAssemblerTester(CodeKind kind, ParamMachTypes... p)
: HandleAndZoneScope(kCompressGraphZone),
CallHelper<ReturnType>(
main_isolate(),
CSignature::New(main_zone(), MachineTypeForC<ReturnType>(), p...)),
RawMachineAssembler(
main_isolate(), main_zone()->template New<Graph>(main_zone()),
Linkage::GetSimplifiedCDescriptor(
main_zone(),
CSignature::New(main_zone(), MachineTypeForC<ReturnType>(),
p...),
CallDescriptor::kInitializeRootRegister),
MachineType::PointerRepresentation(),
InstructionSelector::SupportedMachineOperatorFlags(),
InstructionSelector::AlignmentRequirements()),
kind_(kind) {}
~RawMachineAssemblerTester() override = default;
void CheckNumber(double expected, Object number) {
CHECK(this->isolate()->factory()->NewNumber(expected)->SameValue(number));
}
void CheckString(const char* expected, Object string) {
CHECK(
this->isolate()->factory()->InternalizeUtf8String(expected)->SameValue(
string));
}
void GenerateCode() { Generate(); }
Handle<Code> GetCode() {
Generate();
return code_.ToHandleChecked();
}
protected:
Address Generate() override {
if (code_.is_null()) {
Schedule* schedule = this->ExportForTest();
auto call_descriptor = this->call_descriptor();
Graph* graph = this->graph();
OptimizedCompilationInfo info(ArrayVector("testing"), main_zone(), kind_);
code_ = Pipeline::GenerateCodeForTesting(
&info, main_isolate(), call_descriptor, graph,
AssemblerOptions::Default(main_isolate()), schedule);
}
return this->code_.ToHandleChecked()->entry();
}
private:
CodeKind kind_ = CodeKind::FOR_TESTING;
MaybeHandle<Code> code_;
};
template <typename ReturnType>
class BufferedRawMachineAssemblerTester
: public RawMachineAssemblerTester<int32_t> {
public:
template <typename... ParamMachTypes>
explicit BufferedRawMachineAssemblerTester(ParamMachTypes... p)
: RawMachineAssemblerTester<int32_t>(
MachineType::Pointer(), ((void)p, MachineType::Pointer())...),
test_graph_signature_(
CSignature::New(this->main_zone(), MachineType::Int32(), p...)),
return_parameter_index_(sizeof...(p)) {
static_assert(sizeof...(p) <= arraysize(parameter_nodes_),
"increase parameter_nodes_ array");
std::array<MachineType, sizeof...(p)> p_arr{{p...}};
for (size_t i = 0; i < p_arr.size(); ++i) {
parameter_nodes_[i] = Load(p_arr[i], RawMachineAssembler::Parameter(i));
}
}
Address Generate() override { return RawMachineAssemblerTester::Generate(); }
// The BufferedRawMachineAssemblerTester does not pass parameters directly
// to the constructed IR graph. Instead it passes a pointer to the parameter
// to the IR graph, and adds Load nodes to the IR graph to load the
// parameters from memory. Thereby it is possible to pass 64 bit parameters
// to the IR graph.
Node* Parameter(size_t index) {
CHECK_GT(arraysize(parameter_nodes_), index);
return parameter_nodes_[index];
}
// The BufferedRawMachineAssemblerTester adds a Store node to the IR graph
// to store the graph's return value in memory. The memory address for the
// Store node is provided as a parameter. By storing the return value in
// memory it is possible to return 64 bit values.
void Return(Node* input) {
if (COMPRESS_POINTERS_BOOL && MachineTypeForC<ReturnType>().IsTagged()) {
// Since we are returning values via storing to off-heap location
// generate full-word store here.
Store(MachineType::PointerRepresentation(),
RawMachineAssembler::Parameter(return_parameter_index_),
BitcastTaggedToWord(input), kNoWriteBarrier);
} else {
Store(MachineTypeForC<ReturnType>().representation(),
RawMachineAssembler::Parameter(return_parameter_index_), input,
kNoWriteBarrier);
}
RawMachineAssembler::Return(Int32Constant(1234));
}
template <typename... Params>
ReturnType Call(Params... p) {
uintptr_t zap_data[] = {kZapValue, kZapValue};
ReturnType return_value;
STATIC_ASSERT(sizeof(return_value) <= sizeof(zap_data));
MemCopy(&return_value, &zap_data, sizeof(return_value));
CSignature::VerifyParams<Params...>(test_graph_signature_);
CallHelper<int32_t>::Call(reinterpret_cast<void*>(&p)...,
reinterpret_cast<void*>(&return_value));
return return_value;
}
private:
CSignature* test_graph_signature_;
Node* parameter_nodes_[4];
uint32_t return_parameter_index_;
};
template <>
class BufferedRawMachineAssemblerTester<void>
: public RawMachineAssemblerTester<void> {
public:
template <typename... ParamMachTypes>
explicit BufferedRawMachineAssemblerTester(ParamMachTypes... p)
: RawMachineAssemblerTester<void>(((void)p, MachineType::Pointer())...),
test_graph_signature_(
CSignature::New(RawMachineAssemblerTester<void>::main_zone(),
MachineType::None(), p...)) {
static_assert(sizeof...(p) <= arraysize(parameter_nodes_),
"increase parameter_nodes_ array");
std::array<MachineType, sizeof...(p)> p_arr{{p...}};
for (size_t i = 0; i < p_arr.size(); ++i) {
parameter_nodes_[i] = Load(p_arr[i], RawMachineAssembler::Parameter(i));
}
}
Address Generate() override { return RawMachineAssemblerTester::Generate(); }
// The BufferedRawMachineAssemblerTester does not pass parameters directly
// to the constructed IR graph. Instead it passes a pointer to the parameter
// to the IR graph, and adds Load nodes to the IR graph to load the
// parameters from memory. Thereby it is possible to pass 64 bit parameters
// to the IR graph.
Node* Parameter(size_t index) {
CHECK_GT(arraysize(parameter_nodes_), index);
return parameter_nodes_[index];
}
template <typename... Params>
void Call(Params... p) {
CSignature::VerifyParams<Params...>(test_graph_signature_);
CallHelper<void>::Call(reinterpret_cast<void*>(&p)...);
}
private:
CSignature* test_graph_signature_;
Node* parameter_nodes_[4];
};
static const bool USE_RESULT_BUFFER = true;
static const bool USE_RETURN_REGISTER = false;
static const int32_t CHECK_VALUE = 0x99BEEDCE;
// TODO(titzer): use the C-style calling convention, or any register-based
// calling convention for binop tests.
template <typename CType, bool use_result_buffer>
class BinopTester {
public:
explicit BinopTester(RawMachineAssemblerTester<int32_t>* tester,
MachineType type)
: T(tester),
param0(T->LoadFromPointer(&p0, type)),
param1(T->LoadFromPointer(&p1, type)),
type(type),
p0(static_cast<CType>(0)),
p1(static_cast<CType>(0)),
result(static_cast<CType>(0)) {}
RawMachineAssemblerTester<int32_t>* T;
Node* param0;
Node* param1;
CType call(CType a0, CType a1) {
p0 = a0;
p1 = a1;
if (use_result_buffer) {
CHECK_EQ(CHECK_VALUE, T->Call());
return result;
} else {
return static_cast<CType>(T->Call());
}
}
void AddReturn(Node* val) {
if (use_result_buffer) {
T->Store(type.representation(), T->PointerConstant(&result),
T->Int32Constant(0), val, kNoWriteBarrier);
T->Return(T->Int32Constant(CHECK_VALUE));
} else {
T->Return(val);
}
}
template <typename Ci, typename Cj, typename Fn>
void Run(const Ci& ci, const Cj& cj, const Fn& fn) {
typename Ci::const_iterator i;
typename Cj::const_iterator j;
for (i = ci.begin(); i != ci.end(); ++i) {
for (j = cj.begin(); j != cj.end(); ++j) {
CHECK_EQ(fn(*i, *j), this->call(*i, *j));
}
}
}
protected:
MachineType type;
CType p0;
CType p1;
CType result;
};
// A helper class for testing code sequences that take two int parameters and
// return an int value.
class Int32BinopTester : public BinopTester<int32_t, USE_RETURN_REGISTER> {
public:
explicit Int32BinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<int32_t, USE_RETURN_REGISTER>(tester,
MachineType::Int32()) {}
};
// A helper class for testing code sequences that take two int parameters and
// return an int value.
class Int64BinopTester : public BinopTester<int64_t, USE_RETURN_REGISTER> {
public:
explicit Int64BinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<int64_t, USE_RETURN_REGISTER>(tester,
MachineType::Int64()) {}
};
// A helper class for testing code sequences that take two uint parameters and
// return an uint value.
class Uint32BinopTester : public BinopTester<uint32_t, USE_RETURN_REGISTER> {
public:
explicit Uint32BinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<uint32_t, USE_RETURN_REGISTER>(tester,
MachineType::Uint32()) {}
uint32_t call(uint32_t a0, uint32_t a1) {
p0 = a0;
p1 = a1;
return static_cast<uint32_t>(T->Call());
}
};
// A helper class for testing code sequences that take two float parameters and
// return a float value.
class Float32BinopTester : public BinopTester<float, USE_RESULT_BUFFER> {
public:
explicit Float32BinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<float, USE_RESULT_BUFFER>(tester, MachineType::Float32()) {}
};
// A helper class for testing code sequences that take two double parameters and
// return a double value.
class Float64BinopTester : public BinopTester<double, USE_RESULT_BUFFER> {
public:
explicit Float64BinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<double, USE_RESULT_BUFFER>(tester, MachineType::Float64()) {
}
};
// A helper class for testing code sequences that take two pointer parameters
// and return a pointer value.
// TODO(titzer): pick word size of pointers based on V8_TARGET.
template <typename Type>
class PointerBinopTester : public BinopTester<Type, USE_RETURN_REGISTER> {
public:
explicit PointerBinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<Type, USE_RETURN_REGISTER>(tester, MachineType::Pointer()) {
}
};
// A helper class for testing code sequences that take two tagged parameters and
// return a tagged value.
template <typename Type>
class TaggedBinopTester : public BinopTester<Type, USE_RETURN_REGISTER> {
public:
explicit TaggedBinopTester(RawMachineAssemblerTester<int32_t>* tester)
: BinopTester<Type, USE_RETURN_REGISTER>(tester,
MachineType::AnyTagged()) {}
};
// A helper class for testing compares. Wraps a machine opcode and provides
// evaluation routines and the operators.
class CompareWrapper {
public:
explicit CompareWrapper(IrOpcode::Value op) : opcode(op) {}
Node* MakeNode(RawMachineAssemblerTester<int32_t>* m, Node* a, Node* b) {
return m->AddNode(op(m->machine()), a, b);
}
const Operator* op(MachineOperatorBuilder* machine) {
switch (opcode) {
case IrOpcode::kWord32Equal:
return machine->Word32Equal();
case IrOpcode::kInt32LessThan:
return machine->Int32LessThan();
case IrOpcode::kInt32LessThanOrEqual:
return machine->Int32LessThanOrEqual();
case IrOpcode::kUint32LessThan:
return machine->Uint32LessThan();
case IrOpcode::kUint32LessThanOrEqual:
return machine->Uint32LessThanOrEqual();
case IrOpcode::kFloat64Equal:
return machine->Float64Equal();
case IrOpcode::kFloat64LessThan:
return machine->Float64LessThan();
case IrOpcode::kFloat64LessThanOrEqual:
return machine->Float64LessThanOrEqual();
default:
UNREACHABLE();
}
return nullptr;
}
bool Int32Compare(int32_t a, int32_t b) {
switch (opcode) {
case IrOpcode::kWord32Equal:
return a == b;
case IrOpcode::kInt32LessThan:
return a < b;
case IrOpcode::kInt32LessThanOrEqual:
return a <= b;
case IrOpcode::kUint32LessThan:
return static_cast<uint32_t>(a) < static_cast<uint32_t>(b);
case IrOpcode::kUint32LessThanOrEqual:
return static_cast<uint32_t>(a) <= static_cast<uint32_t>(b);
default:
UNREACHABLE();
}
return false;
}
bool Float64Compare(double a, double b) {
switch (opcode) {
case IrOpcode::kFloat64Equal:
return a == b;
case IrOpcode::kFloat64LessThan:
return a < b;
case IrOpcode::kFloat64LessThanOrEqual:
return a <= b;
default:
UNREACHABLE();
}
return false;
}
IrOpcode::Value opcode;
};
// A small closure class to generate code for a function of two inputs that
// produces a single output so that it can be used in many different contexts.
// The {expected()} method should compute the expected output for a given
// pair of inputs.
template <typename T>
class BinopGen {
public:
virtual void gen(RawMachineAssemblerTester<int32_t>* m, Node* a, Node* b) = 0;
virtual T expected(T a, T b) = 0;
virtual ~BinopGen() = default;
};
// A helper class to generate various combination of input shape combinations
// and run the generated code to ensure it produces the correct results.
class Int32BinopInputShapeTester {
public:
explicit Int32BinopInputShapeTester(BinopGen<int32_t>* g)
: gen(g), input_a(0), input_b(0) {}
void TestAllInputShapes();
private:
BinopGen<int32_t>* gen;
int32_t input_a;
int32_t input_b;
void Run(RawMachineAssemblerTester<int32_t>* m);
void RunLeft(RawMachineAssemblerTester<int32_t>* m);
void RunRight(RawMachineAssemblerTester<int32_t>* m);
};
} // namespace compiler
} // namespace internal
} // namespace v8
#endif // V8_CCTEST_COMPILER_CODEGEN_TESTER_H_