blob: 8851a6a2dfa869be2807326930cffe52cf970f0e [file] [log] [blame]
// 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.
#include "test/unittests/compiler/backend/instruction-selector-unittest.h"
#include "src/objects/objects-inl.h"
namespace v8 {
namespace internal {
namespace compiler {
namespace {
// Immediates (random subset).
const int32_t kImmediates[] = {kMinInt, -42, -1, 0, 1, 2,
3, 4, 5, 6, 7, 8,
16, 42, 0xFF, 0xFFFF, 0x0F0F0F0F, kMaxInt};
} // namespace
TEST_F(InstructionSelectorTest, Int32AddWithParameter) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
m.Return(m.Int32Add(m.Parameter(0), m.Parameter(1)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
}
TEST_F(InstructionSelectorTest, Int32AddWithImmediate) {
TRACED_FOREACH(int32_t, imm, kImmediates) {
{
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32());
m.Return(m.Int32Add(m.Parameter(0), m.Int32Constant(imm)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
if (imm == 0) {
ASSERT_EQ(1U, s[0]->InputCount());
} else {
ASSERT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(imm, s.ToInt32(s[0]->InputAt(1)));
}
}
{
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32());
m.Return(m.Int32Add(m.Int32Constant(imm), m.Parameter(0)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
if (imm == 0) {
ASSERT_EQ(1U, s[0]->InputCount());
} else {
ASSERT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(imm, s.ToInt32(s[0]->InputAt(1)));
}
}
}
}
TEST_F(InstructionSelectorTest, Int32SubWithParameter) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
m.Return(m.Int32Sub(m.Parameter(0), m.Parameter(1)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Sub, s[0]->arch_opcode());
EXPECT_EQ(1U, s[0]->OutputCount());
}
TEST_F(InstructionSelectorTest, Int32SubWithImmediate) {
TRACED_FOREACH(int32_t, imm, kImmediates) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32());
m.Return(m.Int32Sub(m.Parameter(0), m.Int32Constant(imm)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Sub, s[0]->arch_opcode());
ASSERT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(imm, s.ToInt32(s[0]->InputAt(1)));
}
}
// -----------------------------------------------------------------------------
// Conversions.
TEST_F(InstructionSelectorTest, ChangeFloat32ToFloat64WithParameter) {
StreamBuilder m(this, MachineType::Float32(), MachineType::Float64());
m.Return(m.ChangeFloat32ToFloat64(m.Parameter(0)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kSSEFloat32ToFloat64, s[0]->arch_opcode());
EXPECT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(1U, s[0]->OutputCount());
}
TEST_F(InstructionSelectorTest, TruncateFloat64ToFloat32WithParameter) {
StreamBuilder m(this, MachineType::Float64(), MachineType::Float32());
m.Return(m.TruncateFloat64ToFloat32(m.Parameter(0)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kSSEFloat64ToFloat32, s[0]->arch_opcode());
EXPECT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(1U, s[0]->OutputCount());
}
// -----------------------------------------------------------------------------
// Better left operand for commutative binops
TEST_F(InstructionSelectorTest, BetterLeftOperandTestAddBinop) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* param1 = m.Parameter(0);
Node* param2 = m.Parameter(1);
Node* add = m.Int32Add(param1, param2);
m.Return(m.Int32Add(add, param1));
Stream s = m.Build();
ASSERT_EQ(2U, s.size());
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
ASSERT_EQ(2U, s[0]->InputCount());
ASSERT_TRUE(s[0]->InputAt(0)->IsUnallocated());
EXPECT_EQ(s.ToVreg(param1), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(param2), s.ToVreg(s[0]->InputAt(1)));
ASSERT_EQ(2U, s[1]->InputCount());
EXPECT_EQ(s.ToVreg(param1), s.ToVreg(s[0]->InputAt(0)));
}
TEST_F(InstructionSelectorTest, BetterLeftOperandTestMulBinop) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* param1 = m.Parameter(0);
Node* param2 = m.Parameter(1);
Node* mul = m.Int32Mul(param1, param2);
m.Return(m.Int32Mul(mul, param1));
Stream s = m.Build();
ASSERT_EQ(2U, s.size());
EXPECT_EQ(kIA32Imul, s[0]->arch_opcode());
ASSERT_EQ(2U, s[0]->InputCount());
ASSERT_TRUE(s[0]->InputAt(0)->IsUnallocated());
EXPECT_EQ(s.ToVreg(param2), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(param1), s.ToVreg(s[0]->InputAt(1)));
}
// -----------------------------------------------------------------------------
// Conversions.
TEST_F(InstructionSelectorTest, ChangeUint32ToFloat64WithParameter) {
StreamBuilder m(this, MachineType::Float64(), MachineType::Uint32());
m.Return(m.ChangeUint32ToFloat64(m.Parameter(0)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kSSEUint32ToFloat64, s[0]->arch_opcode());
}
// -----------------------------------------------------------------------------
// Loads and stores
namespace {
struct MemoryAccess {
MachineType type;
ArchOpcode load_opcode;
ArchOpcode store_opcode;
};
std::ostream& operator<<(std::ostream& os, const MemoryAccess& memacc) {
return os << memacc.type;
}
static const MemoryAccess kMemoryAccesses[] = {
{MachineType::Int8(), kIA32Movsxbl, kIA32Movb},
{MachineType::Uint8(), kIA32Movzxbl, kIA32Movb},
{MachineType::Int16(), kIA32Movsxwl, kIA32Movw},
{MachineType::Uint16(), kIA32Movzxwl, kIA32Movw},
{MachineType::Int32(), kIA32Movl, kIA32Movl},
{MachineType::Uint32(), kIA32Movl, kIA32Movl},
{MachineType::Float32(), kIA32Movss, kIA32Movss},
{MachineType::Float64(), kIA32Movsd, kIA32Movsd}};
} // namespace
using InstructionSelectorMemoryAccessTest =
InstructionSelectorTestWithParam<MemoryAccess>;
TEST_P(InstructionSelectorMemoryAccessTest, LoadWithParameters) {
const MemoryAccess memacc = GetParam();
StreamBuilder m(this, memacc.type, MachineType::Pointer(),
MachineType::Int32());
m.Return(m.Load(memacc.type, m.Parameter(0), m.Parameter(1)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.load_opcode, s[0]->arch_opcode());
EXPECT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(1U, s[0]->OutputCount());
}
TEST_P(InstructionSelectorMemoryAccessTest, LoadWithImmediateBase) {
const MemoryAccess memacc = GetParam();
TRACED_FOREACH(int32_t, base, kImmediates) {
StreamBuilder m(this, memacc.type, MachineType::Pointer());
m.Return(m.Load(memacc.type, m.Int32Constant(base), m.Parameter(0)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.load_opcode, s[0]->arch_opcode());
if (base == 0) {
ASSERT_EQ(1U, s[0]->InputCount());
} else {
ASSERT_EQ(2U, s[0]->InputCount());
ASSERT_EQ(InstructionOperand::IMMEDIATE, s[0]->InputAt(1)->kind());
EXPECT_EQ(base, s.ToInt32(s[0]->InputAt(1)));
}
EXPECT_EQ(1U, s[0]->OutputCount());
}
}
TEST_P(InstructionSelectorMemoryAccessTest, LoadWithImmediateIndex) {
const MemoryAccess memacc = GetParam();
TRACED_FOREACH(int32_t, index, kImmediates) {
StreamBuilder m(this, memacc.type, MachineType::Pointer());
m.Return(m.Load(memacc.type, m.Parameter(0), m.Int32Constant(index)));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.load_opcode, s[0]->arch_opcode());
if (index == 0) {
ASSERT_EQ(1U, s[0]->InputCount());
} else {
ASSERT_EQ(2U, s[0]->InputCount());
ASSERT_EQ(InstructionOperand::IMMEDIATE, s[0]->InputAt(1)->kind());
EXPECT_EQ(index, s.ToInt32(s[0]->InputAt(1)));
}
EXPECT_EQ(1U, s[0]->OutputCount());
}
}
TEST_P(InstructionSelectorMemoryAccessTest, StoreWithParameters) {
const MemoryAccess memacc = GetParam();
StreamBuilder m(this, MachineType::Int32(), MachineType::Pointer(),
MachineType::Int32(), memacc.type);
m.Store(memacc.type.representation(), m.Parameter(0), m.Parameter(1),
m.Parameter(2), kNoWriteBarrier);
m.Return(m.Int32Constant(0));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.store_opcode, s[0]->arch_opcode());
EXPECT_EQ(3U, s[0]->InputCount());
EXPECT_EQ(0U, s[0]->OutputCount());
}
TEST_P(InstructionSelectorMemoryAccessTest, StoreWithImmediateBase) {
const MemoryAccess memacc = GetParam();
TRACED_FOREACH(int32_t, base, kImmediates) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
memacc.type);
m.Store(memacc.type.representation(), m.Int32Constant(base), m.Parameter(0),
m.Parameter(1), kNoWriteBarrier);
m.Return(m.Int32Constant(0));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.store_opcode, s[0]->arch_opcode());
if (base == 0) {
ASSERT_EQ(2U, s[0]->InputCount());
} else {
ASSERT_EQ(3U, s[0]->InputCount());
ASSERT_EQ(InstructionOperand::IMMEDIATE, s[0]->InputAt(1)->kind());
EXPECT_EQ(base, s.ToInt32(s[0]->InputAt(1)));
}
EXPECT_EQ(0U, s[0]->OutputCount());
}
}
TEST_P(InstructionSelectorMemoryAccessTest, StoreWithImmediateIndex) {
const MemoryAccess memacc = GetParam();
TRACED_FOREACH(int32_t, index, kImmediates) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Pointer(),
memacc.type);
m.Store(memacc.type.representation(), m.Parameter(0),
m.Int32Constant(index), m.Parameter(1), kNoWriteBarrier);
m.Return(m.Int32Constant(0));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(memacc.store_opcode, s[0]->arch_opcode());
if (index == 0) {
ASSERT_EQ(2U, s[0]->InputCount());
} else {
ASSERT_EQ(3U, s[0]->InputCount());
ASSERT_EQ(InstructionOperand::IMMEDIATE, s[0]->InputAt(1)->kind());
EXPECT_EQ(index, s.ToInt32(s[0]->InputAt(1)));
}
EXPECT_EQ(0U, s[0]->OutputCount());
}
}
INSTANTIATE_TEST_SUITE_P(InstructionSelectorTest,
InstructionSelectorMemoryAccessTest,
::testing::ValuesIn(kMemoryAccesses));
// -----------------------------------------------------------------------------
// AddressingMode for loads and stores.
class AddressingModeUnitTest : public InstructionSelectorTest {
public:
AddressingModeUnitTest() : m(NULL) { Reset(); }
~AddressingModeUnitTest() { delete m; }
void Run(Node* base, Node* load_index, Node* store_index,
AddressingMode mode) {
Node* load = m->Load(MachineType::Int32(), base, load_index);
m->Store(MachineRepresentation::kWord32, base, store_index, load,
kNoWriteBarrier);
m->Return(m->Int32Constant(0));
Stream s = m->Build();
ASSERT_EQ(2U, s.size());
EXPECT_EQ(mode, s[0]->addressing_mode());
EXPECT_EQ(mode, s[1]->addressing_mode());
}
Node* zero;
Node* null_ptr;
Node* non_zero;
Node* base_reg; // opaque value to generate base as register
Node* index_reg; // opaque value to generate index as register
Node* scales[4];
StreamBuilder* m;
void Reset() {
delete m;
m = new StreamBuilder(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
zero = m->Int32Constant(0);
null_ptr = m->Int32Constant(0);
non_zero = m->Int32Constant(127);
base_reg = m->Parameter(0);
index_reg = m->Parameter(0);
scales[0] = m->Int32Constant(1);
scales[1] = m->Int32Constant(2);
scales[2] = m->Int32Constant(4);
scales[3] = m->Int32Constant(8);
}
};
TEST_F(AddressingModeUnitTest, AddressingMode_MR) {
Node* base = base_reg;
Node* index = zero;
Run(base, index, index, kMode_MR);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MRI) {
Node* base = base_reg;
Node* index = non_zero;
Run(base, index, index, kMode_MRI);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MR1) {
Node* base = base_reg;
Node* index = index_reg;
Run(base, index, index, kMode_MR1);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MRN) {
AddressingMode expected[] = {kMode_MR1, kMode_MR2, kMode_MR4, kMode_MR8};
for (size_t i = 0; i < arraysize(scales); ++i) {
Reset();
Node* base = base_reg;
Node* load_index = m->Int32Mul(index_reg, scales[i]);
Node* store_index = m->Int32Mul(index_reg, scales[i]);
Run(base, load_index, store_index, expected[i]);
}
}
TEST_F(AddressingModeUnitTest, AddressingMode_MR1I) {
Node* base = base_reg;
Node* load_index = m->Int32Add(index_reg, non_zero);
Node* store_index = m->Int32Add(index_reg, non_zero);
Run(base, load_index, store_index, kMode_MR1I);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MRNI) {
AddressingMode expected[] = {kMode_MR1I, kMode_MR2I, kMode_MR4I, kMode_MR8I};
for (size_t i = 0; i < arraysize(scales); ++i) {
Reset();
Node* base = base_reg;
Node* load_index = m->Int32Add(m->Int32Mul(index_reg, scales[i]), non_zero);
Node* store_index =
m->Int32Add(m->Int32Mul(index_reg, scales[i]), non_zero);
Run(base, load_index, store_index, expected[i]);
}
}
TEST_F(AddressingModeUnitTest, AddressingMode_M1ToMR) {
Node* base = null_ptr;
Node* index = index_reg;
// M1 maps to MR
Run(base, index, index, kMode_MR);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MN) {
AddressingMode expected[] = {kMode_MR, kMode_M2, kMode_M4, kMode_M8};
for (size_t i = 0; i < arraysize(scales); ++i) {
Reset();
Node* base = null_ptr;
Node* load_index = m->Int32Mul(index_reg, scales[i]);
Node* store_index = m->Int32Mul(index_reg, scales[i]);
Run(base, load_index, store_index, expected[i]);
}
}
TEST_F(AddressingModeUnitTest, AddressingMode_M1IToMRI) {
Node* base = null_ptr;
Node* load_index = m->Int32Add(index_reg, non_zero);
Node* store_index = m->Int32Add(index_reg, non_zero);
// M1I maps to MRI
Run(base, load_index, store_index, kMode_MRI);
}
TEST_F(AddressingModeUnitTest, AddressingMode_MNI) {
AddressingMode expected[] = {kMode_MRI, kMode_M2I, kMode_M4I, kMode_M8I};
for (size_t i = 0; i < arraysize(scales); ++i) {
Reset();
Node* base = null_ptr;
Node* load_index = m->Int32Add(m->Int32Mul(index_reg, scales[i]), non_zero);
Node* store_index =
m->Int32Add(m->Int32Mul(index_reg, scales[i]), non_zero);
Run(base, load_index, store_index, expected[i]);
}
}
TEST_F(AddressingModeUnitTest, AddressingMode_MI) {
Node* bases[] = {null_ptr, non_zero};
Node* indices[] = {zero, non_zero};
for (size_t i = 0; i < arraysize(bases); ++i) {
for (size_t j = 0; j < arraysize(indices); ++j) {
Reset();
Node* base = bases[i];
Node* index = indices[j];
Run(base, index, index, kMode_MI);
}
}
}
// -----------------------------------------------------------------------------
// Multiplication.
namespace {
struct MultParam {
int value;
bool lea_expected;
AddressingMode addressing_mode;
};
std::ostream& operator<<(std::ostream& os, const MultParam& m) {
return os << m.value << "." << m.lea_expected << "." << m.addressing_mode;
}
const MultParam kMultParams[] = {{-1, false, kMode_None},
{0, false, kMode_None},
{1, true, kMode_MR},
{2, true, kMode_M2},
{3, true, kMode_MR2},
{4, true, kMode_M4},
{5, true, kMode_MR4},
{6, false, kMode_None},
{7, false, kMode_None},
{8, true, kMode_M8},
{9, true, kMode_MR8},
{10, false, kMode_None},
{11, false, kMode_None}};
} // namespace
using InstructionSelectorMultTest = InstructionSelectorTestWithParam<MultParam>;
static unsigned InputCountForLea(AddressingMode mode) {
switch (mode) {
case kMode_MR1I:
case kMode_MR2I:
case kMode_MR4I:
case kMode_MR8I:
return 3U;
case kMode_M1I:
case kMode_M2I:
case kMode_M4I:
case kMode_M8I:
return 2U;
case kMode_MR1:
case kMode_MR2:
case kMode_MR4:
case kMode_MR8:
case kMode_MRI:
return 2U;
case kMode_M1:
case kMode_M2:
case kMode_M4:
case kMode_M8:
case kMode_MI:
case kMode_MR:
return 1U;
default:
UNREACHABLE();
}
}
static AddressingMode AddressingModeForAddMult(int32_t imm,
const MultParam& m) {
if (imm == 0) return m.addressing_mode;
switch (m.addressing_mode) {
case kMode_MR1:
return kMode_MR1I;
case kMode_MR2:
return kMode_MR2I;
case kMode_MR4:
return kMode_MR4I;
case kMode_MR8:
return kMode_MR8I;
case kMode_M1:
return kMode_M1I;
case kMode_M2:
return kMode_M2I;
case kMode_M4:
return kMode_M4I;
case kMode_M8:
return kMode_M8I;
case kMode_MR:
return kMode_MRI;
default:
UNREACHABLE();
}
}
TEST_P(InstructionSelectorMultTest, Mult32) {
const MultParam m_param = GetParam();
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32());
Node* param = m.Parameter(0);
Node* mult = m.Int32Mul(param, m.Int32Constant(m_param.value));
m.Return(mult);
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(m_param.addressing_mode, s[0]->addressing_mode());
if (m_param.lea_expected) {
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
ASSERT_EQ(InputCountForLea(s[0]->addressing_mode()), s[0]->InputCount());
} else {
EXPECT_EQ(kIA32Imul, s[0]->arch_opcode());
ASSERT_EQ(2U, s[0]->InputCount());
}
EXPECT_EQ(s.ToVreg(param), s.ToVreg(s[0]->InputAt(0)));
}
TEST_P(InstructionSelectorMultTest, MultAdd32) {
TRACED_FOREACH(int32_t, imm, kImmediates) {
const MultParam m_param = GetParam();
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32());
Node* param = m.Parameter(0);
Node* mult = m.Int32Add(m.Int32Mul(param, m.Int32Constant(m_param.value)),
m.Int32Constant(imm));
m.Return(mult);
Stream s = m.Build();
if (m_param.lea_expected) {
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Lea, s[0]->arch_opcode());
EXPECT_EQ(AddressingModeForAddMult(imm, m_param),
s[0]->addressing_mode());
unsigned input_count = InputCountForLea(s[0]->addressing_mode());
ASSERT_EQ(input_count, s[0]->InputCount());
if (imm != 0) {
ASSERT_EQ(InstructionOperand::IMMEDIATE,
s[0]->InputAt(input_count - 1)->kind());
EXPECT_EQ(imm, s.ToInt32(s[0]->InputAt(input_count - 1)));
}
} else {
ASSERT_EQ(2U, s.size());
EXPECT_EQ(kIA32Imul, s[0]->arch_opcode());
EXPECT_EQ(kIA32Lea, s[1]->arch_opcode());
}
}
}
INSTANTIATE_TEST_SUITE_P(InstructionSelectorTest, InstructionSelectorMultTest,
::testing::ValuesIn(kMultParams));
TEST_F(InstructionSelectorTest, Int32MulHigh) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
Node* const n = m.Int32MulHigh(p0, p1);
m.Return(n);
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32ImulHigh, s[0]->arch_opcode());
ASSERT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
EXPECT_TRUE(s.IsFixed(s[0]->InputAt(0), eax));
EXPECT_EQ(s.ToVreg(p1), s.ToVreg(s[0]->InputAt(1)));
EXPECT_TRUE(!s.IsUsedAtStart(s[0]->InputAt(1)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
EXPECT_TRUE(s.IsFixed(s[0]->OutputAt(0), edx));
}
// -----------------------------------------------------------------------------
// Binops with a memory operand.
TEST_F(InstructionSelectorTest, LoadAnd32) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
m.Return(
m.Word32And(p0, m.Load(MachineType::Int32(), p1, m.Int32Constant(127))));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32And, s[0]->arch_opcode());
ASSERT_EQ(3U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(p1), s.ToVreg(s[0]->InputAt(1)));
}
TEST_F(InstructionSelectorTest, LoadOr32) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
m.Return(
m.Word32Or(p0, m.Load(MachineType::Int32(), p1, m.Int32Constant(127))));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Or, s[0]->arch_opcode());
ASSERT_EQ(3U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(p1), s.ToVreg(s[0]->InputAt(1)));
}
TEST_F(InstructionSelectorTest, LoadXor32) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
m.Return(
m.Word32Xor(p0, m.Load(MachineType::Int32(), p1, m.Int32Constant(127))));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Xor, s[0]->arch_opcode());
ASSERT_EQ(3U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(p1), s.ToVreg(s[0]->InputAt(1)));
}
TEST_F(InstructionSelectorTest, LoadAdd32) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
m.Return(
m.Int32Add(p0, m.Load(MachineType::Int32(), p1, m.Int32Constant(127))));
Stream s = m.Build();
// Use lea instead of add, so memory operand is invalid.
ASSERT_EQ(2U, s.size());
EXPECT_EQ(kIA32Movl, s[0]->arch_opcode());
EXPECT_EQ(kIA32Lea, s[1]->arch_opcode());
}
TEST_F(InstructionSelectorTest, LoadSub32) {
StreamBuilder m(this, MachineType::Int32(), MachineType::Int32(),
MachineType::Int32());
Node* const p0 = m.Parameter(0);
Node* const p1 = m.Parameter(1);
m.Return(
m.Int32Sub(p0, m.Load(MachineType::Int32(), p1, m.Int32Constant(127))));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Sub, s[0]->arch_opcode());
ASSERT_EQ(3U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
EXPECT_EQ(s.ToVreg(p1), s.ToVreg(s[0]->InputAt(1)));
}
// -----------------------------------------------------------------------------
// Floating point operations.
TEST_F(InstructionSelectorTest, Float32Abs) {
{
StreamBuilder m(this, MachineType::Float32(), MachineType::Float32());
Node* const p0 = m.Parameter(0);
Node* const n = m.Float32Abs(p0);
m.Return(n);
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kSSEFloat32Abs, s[0]->arch_opcode());
ASSERT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_TRUE(s.IsSameAsFirst(s[0]->Output()));
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
EXPECT_EQ(kFlags_none, s[0]->flags_mode());
}
{
StreamBuilder m(this, MachineType::Float32(), MachineType::Float32());
Node* const p0 = m.Parameter(0);
Node* const n = m.Float32Abs(p0);
m.Return(n);
Stream s = m.Build(AVX);
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kAVXFloat32Abs, s[0]->arch_opcode());
ASSERT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
EXPECT_EQ(kFlags_none, s[0]->flags_mode());
}
}
TEST_F(InstructionSelectorTest, Float64Abs) {
{
StreamBuilder m(this, MachineType::Float64(), MachineType::Float64());
Node* const p0 = m.Parameter(0);
Node* const n = m.Float64Abs(p0);
m.Return(n);
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kSSEFloat64Abs, s[0]->arch_opcode());
ASSERT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_TRUE(s.IsSameAsFirst(s[0]->Output()));
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
EXPECT_EQ(kFlags_none, s[0]->flags_mode());
}
{
StreamBuilder m(this, MachineType::Float64(), MachineType::Float64());
Node* const p0 = m.Parameter(0);
Node* const n = m.Float64Abs(p0);
m.Return(n);
Stream s = m.Build(AVX);
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kAVXFloat64Abs, s[0]->arch_opcode());
ASSERT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
EXPECT_EQ(kFlags_none, s[0]->flags_mode());
}
}
TEST_F(InstructionSelectorTest, Float64BinopArithmetic) {
{
StreamBuilder m(this, MachineType::Float64(), MachineType::Float64(),
MachineType::Float64());
Node* add = m.Float64Add(m.Parameter(0), m.Parameter(1));
Node* mul = m.Float64Mul(add, m.Parameter(1));
Node* sub = m.Float64Sub(mul, add);
Node* ret = m.Float64Div(mul, sub);
m.Return(ret);
Stream s = m.Build(AVX);
ASSERT_EQ(4U, s.size());
EXPECT_EQ(kAVXFloat64Add, s[0]->arch_opcode());
EXPECT_EQ(kAVXFloat64Mul, s[1]->arch_opcode());
EXPECT_EQ(kAVXFloat64Sub, s[2]->arch_opcode());
EXPECT_EQ(kAVXFloat64Div, s[3]->arch_opcode());
}
{
StreamBuilder m(this, MachineType::Float64(), MachineType::Float64(),
MachineType::Float64());
Node* add = m.Float64Add(m.Parameter(0), m.Parameter(1));
Node* mul = m.Float64Mul(add, m.Parameter(1));
Node* sub = m.Float64Sub(mul, add);
Node* ret = m.Float64Div(mul, sub);
m.Return(ret);
Stream s = m.Build();
ASSERT_EQ(4U, s.size());
EXPECT_EQ(kSSEFloat64Add, s[0]->arch_opcode());
EXPECT_EQ(kSSEFloat64Mul, s[1]->arch_opcode());
EXPECT_EQ(kSSEFloat64Sub, s[2]->arch_opcode());
EXPECT_EQ(kSSEFloat64Div, s[3]->arch_opcode());
}
}
// -----------------------------------------------------------------------------
// Miscellaneous.
TEST_F(InstructionSelectorTest, Word32Clz) {
StreamBuilder m(this, MachineType::Uint32(), MachineType::Uint32());
Node* const p0 = m.Parameter(0);
Node* const n = m.Word32Clz(p0);
m.Return(n);
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Lzcnt, s[0]->arch_opcode());
ASSERT_EQ(1U, s[0]->InputCount());
EXPECT_EQ(s.ToVreg(p0), s.ToVreg(s[0]->InputAt(0)));
ASSERT_EQ(1U, s[0]->OutputCount());
EXPECT_EQ(s.ToVreg(n), s.ToVreg(s[0]->Output()));
}
TEST_F(InstructionSelectorTest, StackCheck0) {
ExternalReference js_stack_limit =
ExternalReference::Create(isolate()->stack_guard()->address_of_jslimit());
StreamBuilder m(this, MachineType::Int32());
Node* const sp = m.LoadStackPointer();
Node* const stack_limit =
m.Load(MachineType::Pointer(), m.ExternalConstant(js_stack_limit));
Node* const interrupt = m.UintPtrLessThan(sp, stack_limit);
RawMachineLabel if_true, if_false;
m.Branch(interrupt, &if_true, &if_false);
m.Bind(&if_true);
m.Return(m.Int32Constant(1));
m.Bind(&if_false);
m.Return(m.Int32Constant(0));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32Cmp, s[0]->arch_opcode());
EXPECT_EQ(4U, s[0]->InputCount());
EXPECT_EQ(0U, s[0]->OutputCount());
}
TEST_F(InstructionSelectorTest, StackCheck1) {
ExternalReference js_stack_limit =
ExternalReference::Create(isolate()->stack_guard()->address_of_jslimit());
StreamBuilder m(this, MachineType::Int32());
Node* const sp = m.LoadStackPointer();
Node* const stack_limit =
m.Load(MachineType::Pointer(), m.ExternalConstant(js_stack_limit));
Node* const sp_within_limit = m.UintPtrLessThan(stack_limit, sp);
RawMachineLabel if_true, if_false;
m.Branch(sp_within_limit, &if_true, &if_false);
m.Bind(&if_true);
m.Return(m.Int32Constant(1));
m.Bind(&if_false);
m.Return(m.Int32Constant(0));
Stream s = m.Build();
ASSERT_EQ(1U, s.size());
EXPECT_EQ(kIA32StackCheck, s[0]->arch_opcode());
EXPECT_EQ(2U, s[0]->InputCount());
EXPECT_EQ(0U, s[0]->OutputCount());
}
} // namespace compiler
} // namespace internal
} // namespace v8