v8/test/unittests/interpreter/bytecode-array-builder-unittest.cc

798 lines
28 KiB
C++
Raw Normal View History

// 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 "src/v8.h"
#include "src/ast/scopes.h"
#include "src/interpreter/bytecode-array-builder.h"
#include "src/interpreter/bytecode-array-iterator.h"
#include "src/interpreter/bytecode-label.h"
#include "src/interpreter/bytecode-register-allocator.h"
#include "src/objects-inl.h"
#include "test/unittests/test-utils.h"
namespace v8 {
namespace internal {
namespace interpreter {
class BytecodeArrayBuilderTest : public TestWithIsolateAndZone {
public:
BytecodeArrayBuilderTest() {}
~BytecodeArrayBuilderTest() override {}
};
TEST_F(BytecodeArrayBuilderTest, AllBytecodesGenerated) {
CanonicalHandleScope canonical(isolate());
BytecodeArrayBuilder builder(isolate(), zone(), 0, 1, 131);
Factory* factory = isolate()->factory();
AstValueFactory ast_factory(zone(), isolate()->ast_string_constants(),
isolate()->heap()->HashSeed());
DeclarationScope scope(zone(), &ast_factory);
CHECK_EQ(builder.locals_count(), 131);
CHECK_EQ(builder.context_count(), 1);
CHECK_EQ(builder.fixed_register_count(), 132);
Register reg(0);
Register other(reg.index() + 1);
Register wide(128);
RegisterList reg_list;
RegisterList single(0, 1), pair(0, 2), triple(0, 3);
[runtime] Optimize and unify rest parameters. Replace the somewhat awkward RestParamAccessStub, which would always call into the runtime anyway with a proper FastNewRestParameterStub, which is basically based on the code that was already there for strict arguments object materialization. But for rest parameters we could optimize even further (leading to 8-10x improvements for functions with rest parameters), by fixing the internal formal parameter count: Every SharedFunctionInfo has a formal_parameter_count field, which specifies the number of formal parameters, and is used to decide whether we need to create an arguments adaptor frame when calling a function (i.e. if there's a mismatch between the actual and expected parameters). Previously the formal_parameter_count included the rest parameter, which was sort of unfortunate, as that meant that calling a function with only the non-rest parameters still required an arguments adaptor (plus some other oddities). Now with this CL we fix, so that we do no longer include the rest parameter in that count. Thereby checking for rest parameters is very efficient, as we only need to check whether there is an arguments adaptor frame, and if not create an empty array, otherwise check whether the arguments adaptor frame has more parameters than specified by the formal_parameter_count. The FastNewRestParameterStub is written in a way that it can be directly used by Ignition as well, and with some tweaks to the TurboFan backends and the CodeStubAssembler, we should be able to rewrite it as TurboFanCodeStub in the near future. Drive-by-fix: Refactor and unify the CreateArgumentsType which was different in TurboFan and Ignition; now we have a single enum class which is used in both TurboFan and Ignition. R=jarin@chromium.org, rmcilroy@chromium.org TBR=rossberg@chromium.org BUG=v8:2159 LOG=n Review URL: https://codereview.chromium.org/1676883002 Cr-Commit-Position: refs/heads/master@{#33809}
2016-02-08 10:08:21 +00:00
// Emit argument creation operations.
builder.CreateArguments(CreateArgumentsType::kMappedArguments)
.CreateArguments(CreateArgumentsType::kUnmappedArguments)
[runtime] Optimize and unify rest parameters. Replace the somewhat awkward RestParamAccessStub, which would always call into the runtime anyway with a proper FastNewRestParameterStub, which is basically based on the code that was already there for strict arguments object materialization. But for rest parameters we could optimize even further (leading to 8-10x improvements for functions with rest parameters), by fixing the internal formal parameter count: Every SharedFunctionInfo has a formal_parameter_count field, which specifies the number of formal parameters, and is used to decide whether we need to create an arguments adaptor frame when calling a function (i.e. if there's a mismatch between the actual and expected parameters). Previously the formal_parameter_count included the rest parameter, which was sort of unfortunate, as that meant that calling a function with only the non-rest parameters still required an arguments adaptor (plus some other oddities). Now with this CL we fix, so that we do no longer include the rest parameter in that count. Thereby checking for rest parameters is very efficient, as we only need to check whether there is an arguments adaptor frame, and if not create an empty array, otherwise check whether the arguments adaptor frame has more parameters than specified by the formal_parameter_count. The FastNewRestParameterStub is written in a way that it can be directly used by Ignition as well, and with some tweaks to the TurboFan backends and the CodeStubAssembler, we should be able to rewrite it as TurboFanCodeStub in the near future. Drive-by-fix: Refactor and unify the CreateArgumentsType which was different in TurboFan and Ignition; now we have a single enum class which is used in both TurboFan and Ignition. R=jarin@chromium.org, rmcilroy@chromium.org TBR=rossberg@chromium.org BUG=v8:2159 LOG=n Review URL: https://codereview.chromium.org/1676883002 Cr-Commit-Position: refs/heads/master@{#33809}
2016-02-08 10:08:21 +00:00
.CreateArguments(CreateArgumentsType::kRestParameter);
// Emit constant loads.
builder.LoadLiteral(Smi::kZero)
.StoreAccumulatorInRegister(reg)
.LoadLiteral(Smi::FromInt(8))
.CompareOperation(Token::Value::EQ, reg,
1) // Prevent peephole optimization
// LdaSmi, Star -> LdrSmi.
.StoreAccumulatorInRegister(reg)
.LoadLiteral(Smi::FromInt(10000000))
.StoreAccumulatorInRegister(reg)
.LoadLiteral(
ast_factory.NewString(ast_factory.GetOneByteString("A constant")))
.StoreAccumulatorInRegister(reg)
.LoadUndefined()
.StoreAccumulatorInRegister(reg)
.LoadNull()
.StoreAccumulatorInRegister(reg)
.LoadTheHole()
.StoreAccumulatorInRegister(reg)
.LoadTrue()
.StoreAccumulatorInRegister(reg)
.LoadFalse()
.StoreAccumulatorInRegister(wide);
// Emit Ldar and Star taking care to foil the register optimizer.
builder.StackCheck(0)
.LoadAccumulatorWithRegister(other)
.BinaryOperation(Token::ADD, reg, 1)
.StoreAccumulatorInRegister(reg)
.LoadNull();
// Emit register-register transfer.
builder.MoveRegister(reg, other);
builder.MoveRegister(reg, wide);
// Emit global load / store operations.
const AstRawString* name = ast_factory.GetOneByteString("var_name");
builder.LoadGlobal(name, 1, TypeofMode::NOT_INSIDE_TYPEOF)
.LoadGlobal(name, 1, TypeofMode::INSIDE_TYPEOF)
.StoreGlobal(name, 1, LanguageMode::SLOPPY)
.StoreGlobal(name, 1, LanguageMode::STRICT);
// Emit context operations.
builder.PushContext(reg)
.PopContext(reg)
Reland of Thread maybe-assigned through the bytecodes. (patchset #1 id:1 of https://codereview.chromium.org/2680923003/ ) Reason for revert: False alarm, bot hiccup Original issue's description: > Revert of Thread maybe-assigned through the bytecodes. (patchset #5 id:80001 of https://codereview.chromium.org/2655733003/ ) > > Reason for revert: > needed for properly reverting https://chromium.googlesource.com/v8/v8/+/f3ae5ccf57690d8c2d87c4fe1d10b103ad6a4ab3 > > Original issue's description: > > Thread maybe-assigned through the bytecodes. > > > > This introduces LoadImmutableContextSlot and LoadImmutableCurrentContextSlot > > bytecodes, which are emitted when reading from never-assigned context slot. > > > > There is a subtlety here: the slot are not immutable, the meaning is > > actually undefined-or-hole-or-immutable. > > > > Review-Url: https://codereview.chromium.org/2655733003 > > Cr-Commit-Position: refs/heads/master@{#43000} > > Committed: https://chromium.googlesource.com/v8/v8/+/17c2dd388697da626224c371c8181d20d2016d82 > > TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > > Review-Url: https://codereview.chromium.org/2680923003 > Cr-Commit-Position: refs/heads/master@{#43011} > Committed: https://chromium.googlesource.com/v8/v8/+/ece4e54a31ac76ee9c450270d65270562db716e1 TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Review-Url: https://codereview.chromium.org/2679953003 Cr-Commit-Position: refs/heads/master@{#43012}
2017-02-07 20:42:03 +00:00
.LoadContextSlot(reg, 1, 0, BytecodeArrayBuilder::kMutableSlot)
.StoreContextSlot(reg, 1, 0)
.LoadContextSlot(reg, 2, 0, BytecodeArrayBuilder::kImmutableSlot)
.StoreContextSlot(reg, 3, 0);
// Emit context operations which operate on the local context.
Reland of Thread maybe-assigned through the bytecodes. (patchset #1 id:1 of https://codereview.chromium.org/2680923003/ ) Reason for revert: False alarm, bot hiccup Original issue's description: > Revert of Thread maybe-assigned through the bytecodes. (patchset #5 id:80001 of https://codereview.chromium.org/2655733003/ ) > > Reason for revert: > needed for properly reverting https://chromium.googlesource.com/v8/v8/+/f3ae5ccf57690d8c2d87c4fe1d10b103ad6a4ab3 > > Original issue's description: > > Thread maybe-assigned through the bytecodes. > > > > This introduces LoadImmutableContextSlot and LoadImmutableCurrentContextSlot > > bytecodes, which are emitted when reading from never-assigned context slot. > > > > There is a subtlety here: the slot are not immutable, the meaning is > > actually undefined-or-hole-or-immutable. > > > > Review-Url: https://codereview.chromium.org/2655733003 > > Cr-Commit-Position: refs/heads/master@{#43000} > > Committed: https://chromium.googlesource.com/v8/v8/+/17c2dd388697da626224c371c8181d20d2016d82 > > TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > > Review-Url: https://codereview.chromium.org/2680923003 > Cr-Commit-Position: refs/heads/master@{#43011} > Committed: https://chromium.googlesource.com/v8/v8/+/ece4e54a31ac76ee9c450270d65270562db716e1 TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Review-Url: https://codereview.chromium.org/2679953003 Cr-Commit-Position: refs/heads/master@{#43012}
2017-02-07 20:42:03 +00:00
builder
.LoadContextSlot(Register::current_context(), 1, 0,
BytecodeArrayBuilder::kMutableSlot)
.StoreContextSlot(Register::current_context(), 1, 0)
.LoadContextSlot(Register::current_context(), 2, 0,
BytecodeArrayBuilder::kImmutableSlot)
.StoreContextSlot(Register::current_context(), 3, 0);
// Emit load / store property operations.
builder.LoadNamedProperty(reg, name, 0)
.LoadKeyedProperty(reg, 0)
.StoreNamedProperty(reg, name, 0, LanguageMode::SLOPPY)
.StoreKeyedProperty(reg, reg, 0, LanguageMode::SLOPPY)
.StoreNamedProperty(reg, name, 0, LanguageMode::STRICT)
.StoreKeyedProperty(reg, reg, 0, LanguageMode::STRICT)
.StoreNamedOwnProperty(reg, name, 0);
// Emit load / store lookup slots.
builder.LoadLookupSlot(name, TypeofMode::NOT_INSIDE_TYPEOF)
.LoadLookupSlot(name, TypeofMode::INSIDE_TYPEOF)
.StoreLookupSlot(name, LanguageMode::SLOPPY)
.StoreLookupSlot(name, LanguageMode::STRICT);
// Emit load / store lookup slots with context fast paths.
builder.LoadLookupContextSlot(name, TypeofMode::NOT_INSIDE_TYPEOF, 1, 0)
.LoadLookupContextSlot(name, TypeofMode::INSIDE_TYPEOF, 1, 0);
// Emit load / store lookup slots with global fast paths.
builder.LoadLookupGlobalSlot(name, TypeofMode::NOT_INSIDE_TYPEOF, 1, 0)
.LoadLookupGlobalSlot(name, TypeofMode::INSIDE_TYPEOF, 1, 0);
// Emit closure operations.
builder.CreateClosure(0, 1, NOT_TENURED);
// Emit create context operation.
builder.CreateBlockContext(&scope);
builder.CreateCatchContext(reg, name, &scope);
builder.CreateFunctionContext(1);
builder.CreateEvalContext(1);
builder.CreateWithContext(reg, &scope);
// Emit literal creation operations.
builder.CreateRegExpLiteral(ast_factory.GetOneByteString("a"), 0, 0);
builder.CreateArrayLiteral(0, 0, 0);
builder.CreateObjectLiteral(0, 0, 0, reg);
// Call operations.
builder.Call(reg, reg_list, 1, Call::GLOBAL_CALL)
.Call(reg, single, 1, Call::GLOBAL_CALL)
.Call(reg, pair, 1, Call::GLOBAL_CALL)
.Call(reg, triple, 1, Call::GLOBAL_CALL)
.Call(reg, reg_list, 1, Call::NAMED_PROPERTY_CALL,
TailCallMode::kDisallow)
.Call(reg, single, 1, Call::NAMED_PROPERTY_CALL)
.Call(reg, pair, 1, Call::NAMED_PROPERTY_CALL)
.Call(reg, triple, 1, Call::NAMED_PROPERTY_CALL)
.Call(reg, reg_list, 1, Call::GLOBAL_CALL, TailCallMode::kAllow)
.CallRuntime(Runtime::kIsArray, reg)
.CallRuntimeForPair(Runtime::kLoadLookupSlotForCall, reg_list, pair)
.CallJSRuntime(Context::SPREAD_ITERABLE_INDEX, reg_list)
.CallWithSpread(reg, reg_list);
// Emit binary operator invocations.
builder.BinaryOperation(Token::Value::ADD, reg, 1)
.BinaryOperation(Token::Value::SUB, reg, 2)
.BinaryOperation(Token::Value::MUL, reg, 3)
.BinaryOperation(Token::Value::DIV, reg, 4)
.BinaryOperation(Token::Value::MOD, reg, 5);
// Emit bitwise operator invocations
builder.BinaryOperation(Token::Value::BIT_OR, reg, 6)
.BinaryOperation(Token::Value::BIT_XOR, reg, 7)
.BinaryOperation(Token::Value::BIT_AND, reg, 8);
// Emit shift operator invocations
builder.BinaryOperation(Token::Value::SHL, reg, 9)
.BinaryOperation(Token::Value::SAR, reg, 10)
.BinaryOperation(Token::Value::SHR, reg, 11);
// Emit peephole optimizations of LdaSmi followed by binary operation.
builder.LoadLiteral(Smi::FromInt(1))
.BinaryOperation(Token::Value::ADD, reg, 1)
.LoadLiteral(Smi::FromInt(2))
.BinaryOperation(Token::Value::SUB, reg, 2)
.LoadLiteral(Smi::FromInt(3))
.BinaryOperation(Token::Value::BIT_AND, reg, 3)
.LoadLiteral(Smi::FromInt(4))
.BinaryOperation(Token::Value::BIT_OR, reg, 4)
.LoadLiteral(Smi::FromInt(5))
.BinaryOperation(Token::Value::SHL, reg, 5)
.LoadLiteral(Smi::FromInt(6))
.BinaryOperation(Token::Value::SAR, reg, 6);
// Emit count operatior invocations
builder.CountOperation(Token::Value::ADD, 1)
.CountOperation(Token::Value::SUB, 1);
// Emit unary operator invocations.
builder
.LogicalNot() // ToBooleanLogicalNot
.LogicalNot() // non-ToBoolean LogicalNot
.TypeOf();
// Emit delete
builder.Delete(reg, LanguageMode::SLOPPY).Delete(reg, LanguageMode::STRICT);
// Emit construct.
builder.Construct(reg, reg_list, 1).ConstructWithSpread(reg, reg_list);
// Emit test operator invocations.
builder.CompareOperation(Token::Value::EQ, reg, 1)
.CompareOperation(Token::Value::EQ_STRICT, reg, 2)
.CompareOperation(Token::Value::LT, reg, 3)
.CompareOperation(Token::Value::GT, reg, 4)
.CompareOperation(Token::Value::LTE, reg, 5)
.CompareOperation(Token::Value::GTE, reg, 6)
.CompareOperation(Token::Value::INSTANCEOF, reg, 7)
.CompareOperation(Token::Value::IN, reg, 8)
.CompareTypeOf(TestTypeOfFlags::LiteralFlag::kNumber);
// Emit peephole optimizations of equality with Null or Undefined.
builder.LoadUndefined()
.CompareOperation(Token::Value::EQ, reg, 1)
.LoadNull()
.CompareOperation(Token::Value::EQ, reg, 1)
.LoadUndefined()
.CompareOperation(Token::Value::EQ_STRICT, reg, 1)
.LoadNull()
.CompareOperation(Token::Value::EQ_STRICT, reg, 1);
// Emit conversion operator invocations.
builder.ConvertAccumulatorToNumber(reg)
.ConvertAccumulatorToObject(reg)
.ConvertAccumulatorToName(reg);
// Emit GetSuperConstructor.
builder.GetSuperConstructor(reg);
// Short jumps with Imm8 operands
{
[ignition] desugar GetIterator() via bytecode rather than via AST Introduces: - a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions. - a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan. The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST. This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value. BUG=v8:4280 R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.org TBR=rossberg@chromium.org Review-Url: https://codereview.chromium.org/2557593004 Cr-Commit-Position: refs/heads/master@{#41555}
2016-12-07 15:19:52 +00:00
BytecodeLabel start, after_jump1, after_jump2, after_jump3, after_jump4,
after_jump5;
builder.Bind(&start)
.Jump(&after_jump1)
.Bind(&after_jump1)
.JumpIfNull(&after_jump2)
.Bind(&after_jump2)
.JumpIfUndefined(&after_jump3)
.Bind(&after_jump3)
.JumpIfNotHole(&after_jump4)
.Bind(&after_jump4)
[ignition] desugar GetIterator() via bytecode rather than via AST Introduces: - a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions. - a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan. The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST. This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value. BUG=v8:4280 R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.org TBR=rossberg@chromium.org Review-Url: https://codereview.chromium.org/2557593004 Cr-Commit-Position: refs/heads/master@{#41555}
2016-12-07 15:19:52 +00:00
.JumpIfJSReceiver(&after_jump5)
.Bind(&after_jump5)
.JumpLoop(&start, 0);
}
// Longer jumps with constant operands
[ignition] desugar GetIterator() via bytecode rather than via AST Introduces: - a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions. - a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan. The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST. This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value. BUG=v8:4280 R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.org TBR=rossberg@chromium.org Review-Url: https://codereview.chromium.org/2557593004 Cr-Commit-Position: refs/heads/master@{#41555}
2016-12-07 15:19:52 +00:00
BytecodeLabel end[9];
{
BytecodeLabel after_jump;
builder.Jump(&end[0])
.Bind(&after_jump)
.LoadTrue()
.JumpIfTrue(&end[1])
.LoadTrue()
.JumpIfFalse(&end[2])
.LoadLiteral(Smi::kZero)
.JumpIfTrue(&end[3])
.LoadLiteral(Smi::kZero)
.JumpIfFalse(&end[4])
.JumpIfNull(&end[5])
.JumpIfUndefined(&end[6])
[ignition] desugar GetIterator() via bytecode rather than via AST Introduces: - a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions. - a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan. The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST. This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value. BUG=v8:4280 R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.org TBR=rossberg@chromium.org Review-Url: https://codereview.chromium.org/2557593004 Cr-Commit-Position: refs/heads/master@{#41555}
2016-12-07 15:19:52 +00:00
.JumpIfNotHole(&end[7])
.LoadLiteral(ast_factory.prototype_string())
[ignition] desugar GetIterator() via bytecode rather than via AST Introduces: - a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions. - a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan. The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST. This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value. BUG=v8:4280 R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.org TBR=rossberg@chromium.org Review-Url: https://codereview.chromium.org/2557593004 Cr-Commit-Position: refs/heads/master@{#41555}
2016-12-07 15:19:52 +00:00
.JumpIfJSReceiver(&end[8]);
}
// Perform an operation that returns boolean value to
// generate JumpIfTrue/False
{
BytecodeLabel after_jump1, after_jump2;
builder.CompareOperation(Token::Value::EQ, reg, 1)
.JumpIfTrue(&after_jump1)
.Bind(&after_jump1)
.CompareOperation(Token::Value::EQ, reg, 2)
.JumpIfFalse(&after_jump2)
.Bind(&after_jump2);
}
// Perform an operation that returns a non-boolean operation to
// generate JumpIfToBooleanTrue/False.
{
BytecodeLabel after_jump1, after_jump2;
builder.BinaryOperation(Token::Value::ADD, reg, 1)
.JumpIfTrue(&after_jump1)
.Bind(&after_jump1)
.BinaryOperation(Token::Value::ADD, reg, 2)
.JumpIfFalse(&after_jump2)
.Bind(&after_jump2);
}
// Emit set pending message bytecode.
builder.SetPendingMessage();
// Emit stack check bytecode.
builder.StackCheck(0);
// Emit throw and re-throw in it's own basic block so that the rest of the
// code isn't omitted due to being dead.
BytecodeLabel after_throw;
builder.Throw().Bind(&after_throw);
BytecodeLabel after_rethrow;
builder.ReThrow().Bind(&after_rethrow);
builder.ForInPrepare(reg, triple)
.ForInContinue(reg, reg)
.ForInNext(reg, reg, pair, 1)
.ForInStep(reg);
// Wide constant pool loads
for (int i = 0; i < 256; i++) {
// Emit junk in constant pool to force wide constant pool index.
builder.LoadLiteral(ast_factory.NewNumber(2.5321 + i));
}
builder.LoadLiteral(Smi::FromInt(20000000));
const AstRawString* wide_name = ast_factory.GetOneByteString("var_wide_name");
// Emit wide global load / store operations.
builder.LoadGlobal(name, 1024, TypeofMode::NOT_INSIDE_TYPEOF)
.LoadGlobal(name, 1024, TypeofMode::INSIDE_TYPEOF)
.LoadGlobal(name, 1024, TypeofMode::INSIDE_TYPEOF)
.StoreGlobal(name, 1024, LanguageMode::SLOPPY)
.StoreGlobal(wide_name, 1, LanguageMode::STRICT);
// Emit extra wide global load.
builder.LoadGlobal(name, 1024 * 1024, TypeofMode::NOT_INSIDE_TYPEOF);
// Emit wide load / store property operations.
builder.LoadNamedProperty(reg, wide_name, 0)
.LoadKeyedProperty(reg, 2056)
.StoreNamedProperty(reg, wide_name, 0, LanguageMode::SLOPPY)
.StoreKeyedProperty(reg, reg, 2056, LanguageMode::SLOPPY)
.StoreNamedProperty(reg, wide_name, 0, LanguageMode::STRICT)
.StoreKeyedProperty(reg, reg, 2056, LanguageMode::STRICT)
.StoreNamedOwnProperty(reg, wide_name, 0);
builder.StoreDataPropertyInLiteral(reg, reg,
DataPropertyInLiteralFlag::kNoFlags, 0);
// Emit wide context operations.
Reland of Thread maybe-assigned through the bytecodes. (patchset #1 id:1 of https://codereview.chromium.org/2680923003/ ) Reason for revert: False alarm, bot hiccup Original issue's description: > Revert of Thread maybe-assigned through the bytecodes. (patchset #5 id:80001 of https://codereview.chromium.org/2655733003/ ) > > Reason for revert: > needed for properly reverting https://chromium.googlesource.com/v8/v8/+/f3ae5ccf57690d8c2d87c4fe1d10b103ad6a4ab3 > > Original issue's description: > > Thread maybe-assigned through the bytecodes. > > > > This introduces LoadImmutableContextSlot and LoadImmutableCurrentContextSlot > > bytecodes, which are emitted when reading from never-assigned context slot. > > > > There is a subtlety here: the slot are not immutable, the meaning is > > actually undefined-or-hole-or-immutable. > > > > Review-Url: https://codereview.chromium.org/2655733003 > > Cr-Commit-Position: refs/heads/master@{#43000} > > Committed: https://chromium.googlesource.com/v8/v8/+/17c2dd388697da626224c371c8181d20d2016d82 > > TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > > Review-Url: https://codereview.chromium.org/2680923003 > Cr-Commit-Position: refs/heads/master@{#43011} > Committed: https://chromium.googlesource.com/v8/v8/+/ece4e54a31ac76ee9c450270d65270562db716e1 TBR=rmcilroy@chromium.org,bmeurer@chromium.org,neis@chromium.org,jarin@chromium.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Review-Url: https://codereview.chromium.org/2679953003 Cr-Commit-Position: refs/heads/master@{#43012}
2017-02-07 20:42:03 +00:00
builder.LoadContextSlot(reg, 1024, 0, BytecodeArrayBuilder::kMutableSlot)
.StoreContextSlot(reg, 1024, 0);
// Emit wide load / store lookup slots.
builder.LoadLookupSlot(wide_name, TypeofMode::NOT_INSIDE_TYPEOF)
.LoadLookupSlot(wide_name, TypeofMode::INSIDE_TYPEOF)
.StoreLookupSlot(wide_name, LanguageMode::SLOPPY)
.StoreLookupSlot(wide_name, LanguageMode::STRICT);
// CreateClosureWide
builder.CreateClosure(1000, 321, NOT_TENURED);
// Emit wide variant of literal creation operations.
builder
.CreateRegExpLiteral(ast_factory.GetOneByteString("wide_literal"), 0, 0)
.CreateArrayLiteral(0, 0, 0)
.CreateObjectLiteral(0, 0, 0, reg);
// Emit load and store operations for module variables.
builder.LoadModuleVariable(-1, 42)
.LoadModuleVariable(0, 42)
.LoadModuleVariable(1, 42)
.StoreModuleVariable(-1, 42)
.StoreModuleVariable(0, 42)
.StoreModuleVariable(1, 42);
// Emit generator operations.
builder.SuspendGenerator(reg)
.ResumeGenerator(reg);
// Intrinsics handled by the interpreter.
builder.CallRuntime(Runtime::kInlineIsArray, reg_list);
// Emit debugger bytecode.
builder.Debugger();
// Insert dummy ops to force longer jumps.
for (int i = 0; i < 256; i++) {
builder.Debugger();
}
// Bind labels for long jumps at the very end.
for (size_t i = 0; i < arraysize(end); i++) {
builder.Bind(&end[i]);
}
// Return must be the last instruction.
builder.Return();
// Generate BytecodeArray.
scope.SetScriptScopeInfo(factory->NewScopeInfo(1));
ast_factory.Internalize(isolate());
Handle<BytecodeArray> the_array = builder.ToBytecodeArray(isolate());
CHECK_EQ(the_array->frame_size(),
builder.total_register_count() * kPointerSize);
// Build scorecard of bytecodes encountered in the BytecodeArray.
std::vector<int> scorecard(Bytecodes::ToByte(Bytecode::kLast) + 1);
Bytecode final_bytecode = Bytecode::kLdaZero;
int i = 0;
while (i < the_array->length()) {
uint8_t code = the_array->get(i);
scorecard[code] += 1;
final_bytecode = Bytecodes::FromByte(code);
OperandScale operand_scale = OperandScale::kSingle;
int prefix_offset = 0;
if (Bytecodes::IsPrefixScalingBytecode(final_bytecode)) {
operand_scale = Bytecodes::PrefixBytecodeToOperandScale(final_bytecode);
prefix_offset = 1;
code = the_array->get(i + 1);
final_bytecode = Bytecodes::FromByte(code);
}
i += prefix_offset + Bytecodes::Size(final_bytecode, operand_scale);
}
// Insert entry for illegal bytecode as this is never willingly emitted.
scorecard[Bytecodes::ToByte(Bytecode::kIllegal)] = 1;
// Insert entry for nop bytecode as this often gets optimized out.
scorecard[Bytecodes::ToByte(Bytecode::kNop)] = 1;
if (!FLAG_ignition_peephole) {
// Insert entries for bytecodes only emitted by peephole optimizer.
scorecard[Bytecodes::ToByte(Bytecode::kLogicalNot)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kJump)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kJumpIfTrue)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kJumpIfFalse)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kJumpIfTrueConstant)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kJumpIfFalseConstant)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kAddSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kSubSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kBitwiseAndSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kBitwiseOrSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kShiftLeftSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kShiftRightSmi)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kTestUndetectable)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kTestUndefined)] = 1;
scorecard[Bytecodes::ToByte(Bytecode::kTestNull)] = 1;
}
Collect type profile for DevTools Collect type information for JavaScript variables and display it in Chrome DevTools. Design Doc: https://docs.google.com/a/google.com/document/d/1O1uepXZXBI6IwiawTrYC3ohhiNgzkyTdjn3R8ysbYgk/edit?usp=sharing When debugging JavaScript, it’s helpful to know the type of a variable, parameter, and return values. JavaScript is dynamically typed, and for complex source code it’s often hard to infer types. With type profiling, we can provide type information to JavaScript developers. This CL is a proof of concept. It collects type profile for assignments and simply prints the types to stdout. The output looks something like this: #my_var1 #Object #number #string #number #undefined #string #Object #Object We use an extra slot in the feedback vector of assignments to carry the list of types for that assignment. The extra slot is only added when the flag --type-profile is given. Missing work: * Collect data for parameters and return values (currently only assignments). * Remove duplicates from the list of collected types and use a common base class. * Add line numbers or source position instead of the variable name. For now, has a test that compares the stdout of --type-profile in test/message. We will remove this test when --type-profile is fully integrated in the debugger protocol. Adding the test in test/inspector does not work, because the inspector test itself consists of JavaScript code that would convolute the output and be non-deterministic under stress. BUG=v8:5935 Review-Url: https://codereview.chromium.org/2707873002 Cr-Commit-Position: refs/heads/master@{#43866}
2017-03-16 15:01:31 +00:00
if (!FLAG_type_profile) {
// Bytecode for CollectTypeProfile is only emitted when
// Type Information for DevTools is turned on.
scorecard[Bytecodes::ToByte(Bytecode::kCollectTypeProfile)] = 1;
}
// Check return occurs at the end and only once in the BytecodeArray.
CHECK_EQ(final_bytecode, Bytecode::kReturn);
CHECK_EQ(scorecard[Bytecodes::ToByte(final_bytecode)], 1);
#define CHECK_BYTECODE_PRESENT(Name, ...) \
/* Check Bytecode is marked in scorecard, unless it's a debug break */ \
if (!Bytecodes::IsDebugBreak(Bytecode::k##Name)) { \
CHECK_GE(scorecard[Bytecodes::ToByte(Bytecode::k##Name)], 1); \
}
BYTECODE_LIST(CHECK_BYTECODE_PRESENT)
#undef CHECK_BYTECODE_PRESENT
}
TEST_F(BytecodeArrayBuilderTest, FrameSizesLookGood) {
CanonicalHandleScope canonical(isolate());
for (int locals = 0; locals < 5; locals++) {
for (int contexts = 0; contexts < 4; contexts++) {
for (int temps = 0; temps < 3; temps++) {
BytecodeArrayBuilder builder(isolate(), zone(), 0, contexts, locals);
BytecodeRegisterAllocator* allocator(builder.register_allocator());
for (int i = 0; i < locals + contexts; i++) {
builder.LoadLiteral(Smi::kZero);
builder.StoreAccumulatorInRegister(Register(i));
}
for (int i = 0; i < temps; i++) {
Register temp = allocator->NewRegister();
builder.LoadLiteral(Smi::kZero);
builder.StoreAccumulatorInRegister(temp);
// Ensure temporaries are used so not optimized away by the
// register optimizer.
builder.ConvertAccumulatorToName(temp);
}
builder.Return();
Handle<BytecodeArray> the_array = builder.ToBytecodeArray(isolate());
int total_registers = locals + contexts + temps;
CHECK_EQ(the_array->frame_size(), total_registers * kPointerSize);
}
}
}
}
TEST_F(BytecodeArrayBuilderTest, RegisterValues) {
CanonicalHandleScope canonical(isolate());
int index = 1;
Register the_register(index);
CHECK_EQ(the_register.index(), index);
int actual_operand = the_register.ToOperand();
int actual_index = Register::FromOperand(actual_operand).index();
CHECK_EQ(actual_index, index);
}
TEST_F(BytecodeArrayBuilderTest, Parameters) {
CanonicalHandleScope canonical(isolate());
BytecodeArrayBuilder builder(isolate(), zone(), 10, 0, 0);
Register param0(builder.Parameter(0));
Register param9(builder.Parameter(9));
CHECK_EQ(param9.index() - param0.index(), 9);
}
TEST_F(BytecodeArrayBuilderTest, Constants) {
CanonicalHandleScope canonical(isolate());
BytecodeArrayBuilder builder(isolate(), zone(), 0, 0, 0);
AstValueFactory ast_factory(zone(), isolate()->ast_string_constants(),
isolate()->heap()->HashSeed());
const AstValue* heap_num_1 = ast_factory.NewNumber(3.14);
const AstValue* heap_num_2 = ast_factory.NewNumber(5.2);
const AstValue* string =
ast_factory.NewString(ast_factory.GetOneByteString("foo"));
const AstValue* string_copy =
ast_factory.NewString(ast_factory.GetOneByteString("foo"));
builder.LoadLiteral(heap_num_1)
.LoadLiteral(heap_num_2)
.LoadLiteral(string)
.LoadLiteral(heap_num_1)
.LoadLiteral(heap_num_1)
.LoadLiteral(string_copy)
.Return();
ast_factory.Internalize(isolate());
Handle<BytecodeArray> array = builder.ToBytecodeArray(isolate());
// Should only have one entry for each identical constant.
CHECK_EQ(array->constant_pool()->length(), 3);
}
static Bytecode PeepholeToBoolean(Bytecode jump_bytecode) {
return FLAG_ignition_peephole
? Bytecodes::GetJumpWithoutToBoolean(jump_bytecode)
: jump_bytecode;
}
TEST_F(BytecodeArrayBuilderTest, ForwardJumps) {
CanonicalHandleScope canonical(isolate());
static const int kFarJumpDistance = 256 + 20;
BytecodeArrayBuilder builder(isolate(), zone(), 0, 0, 1);
Register reg(0);
BytecodeLabel far0, far1, far2, far3, far4;
BytecodeLabel near0, near1, near2, near3, near4;
BytecodeLabel after_jump0, after_jump1;
builder.Jump(&near0)
.Bind(&after_jump0)
.CompareOperation(Token::Value::EQ, reg, 1)
.JumpIfTrue(&near1)
.CompareOperation(Token::Value::EQ, reg, 2)
.JumpIfFalse(&near2)
.BinaryOperation(Token::Value::ADD, reg, 1)
.JumpIfTrue(&near3)
.BinaryOperation(Token::Value::ADD, reg, 2)
.JumpIfFalse(&near4)
.Bind(&near0)
.Bind(&near1)
.Bind(&near2)
.Bind(&near3)
.Bind(&near4)
.Jump(&far0)
.Bind(&after_jump1)
.CompareOperation(Token::Value::EQ, reg, 3)
.JumpIfTrue(&far1)
.CompareOperation(Token::Value::EQ, reg, 4)
.JumpIfFalse(&far2)
.BinaryOperation(Token::Value::ADD, reg, 3)
.JumpIfTrue(&far3)
.BinaryOperation(Token::Value::ADD, reg, 4)
.JumpIfFalse(&far4);
for (int i = 0; i < kFarJumpDistance - 22; i++) {
builder.Debugger();
}
builder.Bind(&far0).Bind(&far1).Bind(&far2).Bind(&far3).Bind(&far4);
builder.Return();
Handle<BytecodeArray> array = builder.ToBytecodeArray(isolate());
DCHECK_EQ(array->length(), 44 + kFarJumpDistance - 22 + 1);
BytecodeArrayIterator iterator(array);
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJump);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 22);
iterator.Advance();
// Ignore compare operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(),
PeepholeToBoolean(Bytecode::kJumpIfToBooleanTrue));
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 17);
iterator.Advance();
// Ignore compare operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(),
PeepholeToBoolean(Bytecode::kJumpIfToBooleanFalse));
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 12);
iterator.Advance();
// Ignore add operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpIfToBooleanTrue);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 7);
iterator.Advance();
// Ignore add operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpIfToBooleanFalse);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 2);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpConstant);
CHECK_EQ(*iterator.GetConstantForIndexOperand(0),
Smi::FromInt(kFarJumpDistance));
iterator.Advance();
// Ignore compare operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(),
PeepholeToBoolean(Bytecode::kJumpIfToBooleanTrueConstant));
CHECK_EQ(*iterator.GetConstantForIndexOperand(0),
Smi::FromInt(kFarJumpDistance - 5));
iterator.Advance();
// Ignore compare operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(),
PeepholeToBoolean(Bytecode::kJumpIfToBooleanFalseConstant));
CHECK_EQ(*iterator.GetConstantForIndexOperand(0),
Smi::FromInt(kFarJumpDistance - 10));
iterator.Advance();
// Ignore add operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpIfToBooleanTrueConstant);
CHECK_EQ(*iterator.GetConstantForIndexOperand(0),
Smi::FromInt(kFarJumpDistance - 15));
iterator.Advance();
// Ignore add operation.
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(),
Bytecode::kJumpIfToBooleanFalseConstant);
CHECK_EQ(*iterator.GetConstantForIndexOperand(0),
Smi::FromInt(kFarJumpDistance - 20));
iterator.Advance();
}
TEST_F(BytecodeArrayBuilderTest, BackwardJumps) {
CanonicalHandleScope canonical(isolate());
BytecodeArrayBuilder builder(isolate(), zone(), 0, 0, 1);
Register reg(0);
BytecodeLabel label0;
builder.Bind(&label0).JumpLoop(&label0, 0);
for (int i = 0; i < 42; i++) {
BytecodeLabel after_jump;
builder.JumpLoop(&label0, 0).Bind(&after_jump);
}
// Add padding to force wide backwards jumps.
for (int i = 0; i < 256; i++) {
builder.Debugger();
}
builder.JumpLoop(&label0, 0);
BytecodeLabel end;
builder.Bind(&end);
builder.Return();
Handle<BytecodeArray> array = builder.ToBytecodeArray(isolate());
BytecodeArrayIterator iterator(array);
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 0);
iterator.Advance();
for (unsigned i = 0; i < 42; i++) {
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.current_operand_scale(), OperandScale::kSingle);
// offset of 3 (because kJumpLoop takes two immediate operands)
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), i * 3 + 3);
iterator.Advance();
}
// Check padding to force wide backwards jumps.
for (int i = 0; i < 256; i++) {
CHECK_EQ(iterator.current_bytecode(), Bytecode::kDebugger);
iterator.Advance();
}
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.current_operand_scale(), OperandScale::kDouble);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 386);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kReturn);
iterator.Advance();
CHECK(iterator.done());
}
TEST_F(BytecodeArrayBuilderTest, LabelReuse) {
CanonicalHandleScope canonical(isolate());
BytecodeArrayBuilder builder(isolate(), zone(), 0, 0, 0);
// Labels can only have 1 forward reference, but
// can be referred to mulitple times once bound.
BytecodeLabel label, after_jump0, after_jump1;
builder.Jump(&label)
.Bind(&label)
.JumpLoop(&label, 0)
.Bind(&after_jump0)
.JumpLoop(&label, 0)
.Bind(&after_jump1)
.Return();
Handle<BytecodeArray> array = builder.ToBytecodeArray(isolate());
BytecodeArrayIterator iterator(array);
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJump);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 2);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 0);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 3);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kReturn);
iterator.Advance();
CHECK(iterator.done());
}
TEST_F(BytecodeArrayBuilderTest, LabelAddressReuse) {
CanonicalHandleScope canonical(isolate());
static const int kRepeats = 3;
BytecodeArrayBuilder builder(isolate(), zone(), 0, 0, 0);
for (int i = 0; i < kRepeats; i++) {
BytecodeLabel label, after_jump0, after_jump1;
builder.Jump(&label)
.Bind(&label)
.JumpLoop(&label, 0)
.Bind(&after_jump0)
.JumpLoop(&label, 0)
.Bind(&after_jump1);
}
builder.Return();
Handle<BytecodeArray> array = builder.ToBytecodeArray(isolate());
BytecodeArrayIterator iterator(array);
for (int i = 0; i < kRepeats; i++) {
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJump);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 2);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 0);
iterator.Advance();
CHECK_EQ(iterator.current_bytecode(), Bytecode::kJumpLoop);
CHECK_EQ(iterator.GetUnsignedImmediateOperand(0), 3);
iterator.Advance();
}
CHECK_EQ(iterator.current_bytecode(), Bytecode::kReturn);
iterator.Advance();
CHECK(iterator.done());
}
} // namespace interpreter
} // namespace internal
} // namespace v8