AuroraMiddleware/v8
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity

[compiler] Don't remove OSR code cache if deoptimizing at out of loop

Browse Source 
The main purpose of OSR compilation is fasten inner loop execution, the
OSR code cache is still correct for loop if optimizing at out of loop,
keep OSR code cache can reduce unnecessary slow bytecode execution with
feedback collection and avoid re-OSR compilation.
This CL can improve JetStream2 case navier-stokes by ~6%.

Change-Id: I9518317fb922071b131cab5b56998a0fc198804a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3494981
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Tao Pan <tao.pan@intel.com>
Cr-Commit-Position: refs/heads/main@{#79413}
This commit is contained in:
Pan, Tao 2022-03-09 09:19:51 +08:00 committed by V8 LUCI CQ
parent 4e9ab3a0f5
commit 190b5d959b
12 changed files with 86 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/baseline/baseline-compiler.cc
View File

@ -1934,7 +1934,7 @@ void BaselineCompiler::VisitJumpLoop() {
__ LoadRegister(osr_level, interpreter::Register::bytecode_array());
__ LoadByteField(osr_level, osr_level,
BytecodeArray::kOsrLoopNestingLevelOffset);
int loop_depth = iterator().GetImmediateOperand(1);
int loop_depth = iterator().GetJumpLoopNestingLevel();
__ JumpIfByte(Condition::kUnsignedLessThanEqual, osr_level, loop_depth,
&osr_not_armed);
CallBuiltin<Builtin::kBaselineOnStackReplacement>();

1
src/compiler/backend/code-generator.cc
View File

@ -559,6 +559,7 @@ MaybeHandle<Code> CodeGenerator::FinalizeCode() {
.set_is_turbofanned()
.set_stack_slots(frame()->GetTotalFrameSlotCount())
.set_profiler_data(info()->profiler_data())
.set_osr_offset(info()->osr_offset())
.TryBuild();
Handle<Code> code;

17
src/deoptimizer/deoptimizer.cc
View File

@ -532,13 +532,13 @@ Deoptimizer::Deoptimizer(Isolate* isolate, JSFunction function,
function.shared().internal_formal_parameter_count_with_receiver();
input_ = new (size) FrameDescription(size, parameter_count);
DeoptimizationData deopt_data =
DeoptimizationData::cast(compiled_code_.deoptimization_data());
if (kSupportsFixedDeoptExitSizes) {
DCHECK_EQ(deopt_exit_index_, kFixedExitSizeMarker);
// Calculate the deopt exit index from return address.
DCHECK_GT(kNonLazyDeoptExitSize, 0);
DCHECK_GT(kLazyDeoptExitSize, 0);
DeoptimizationData deopt_data =
DeoptimizationData::cast(compiled_code_.deoptimization_data());
Address deopt_start = compiled_code_.raw_instruction_start() +
deopt_data.DeoptExitStart().value();
int eager_soft_and_bailout_deopt_count =
@ -580,6 +580,7 @@ Deoptimizer::Deoptimizer(Isolate* isolate, JSFunction function,
(offset / kEagerWithResumeDeoptExitSize);
}
}
deopt_exit_bytecode_offset_ = deopt_data.GetBytecodeOffset(deopt_exit_index_);
}
Code Deoptimizer::FindOptimizedCode() {
@ -713,8 +714,7 @@ int LookupCatchHandler(Isolate* isolate, TranslatedFrame* translated_frame,
} // namespace
void Deoptimizer::TraceDeoptBegin(int optimization_id,
BytecodeOffset bytecode_offset) {
void Deoptimizer::TraceDeoptBegin(int optimization_id) {
DCHECK(tracing_enabled());
FILE* file = trace_scope()->file();
Deoptimizer::DeoptInfo info =
@ -738,8 +738,9 @@ void Deoptimizer::TraceDeoptBegin(int optimization_id,
#ifdef DEBUG
info.node_id,
#endif // DEBUG
bytecode_offset.ToInt(), deopt_exit_index_, fp_to_sp_delta_,
caller_frame_top_, PointerAuthentication::StripPAC(from_));
deopt_exit_bytecode_offset_.ToInt(), deopt_exit_index_,
fp_to_sp_delta_, caller_frame_top_,
PointerAuthentication::StripPAC(from_));
if (verbose_tracing_enabled() && deopt_kind_ != DeoptimizeKind::kLazy) {
PrintF(file, " ;;; deoptimize at ");
OFStream outstr(file);
@ -867,15 +868,13 @@ void Deoptimizer::DoComputeOutputFrames() {
CHECK_GT(static_cast<uintptr_t>(caller_frame_top_),
stack_guard->real_jslimit());
BytecodeOffset bytecode_offset =
input_data.GetBytecodeOffset(deopt_exit_index_);
ByteArray translations = input_data.TranslationByteArray();
unsigned translation_index =
input_data.TranslationIndex(deopt_exit_index_).value();
if (tracing_enabled()) {
timer.Start();
TraceDeoptBegin(input_data.OptimizationId().value(), bytecode_offset);
TraceDeoptBegin(input_data.OptimizationId().value());
}
FILE* trace_file =

6
src/deoptimizer/deoptimizer.h
View File

@ -54,6 +54,9 @@ class Deoptimizer : public Malloced {
Handle<JSFunction> function() const;
Handle<Code> compiled_code() const;
DeoptimizeKind deopt_kind() const { return deopt_kind_; }
BytecodeOffset deopt_exit_bytecode_offset() const {
return deopt_exit_bytecode_offset_;
}
static Deoptimizer* New(Address raw_function, DeoptimizeKind kind,
unsigned deopt_exit_index, Address from,
@ -199,7 +202,7 @@ class Deoptimizer : public Malloced {
CodeTracer::Scope* verbose_trace_scope() const {
return FLAG_trace_deopt_verbose ? trace_scope() : nullptr;
}
void TraceDeoptBegin(int optimization_id, BytecodeOffset bytecode_offset);
void TraceDeoptBegin(int optimization_id);
void TraceDeoptEnd(double deopt_duration);
#ifdef DEBUG
static void TraceFoundActivation(Isolate* isolate, JSFunction function);
@ -211,6 +214,7 @@ class Deoptimizer : public Malloced {
JSFunction function_;
Code compiled_code_;
unsigned deopt_exit_index_;
BytecodeOffset deopt_exit_bytecode_offset_ = BytecodeOffset::None();
DeoptimizeKind deopt_kind_;
Address from_;
int fp_to_sp_delta_;

2
src/execution/tiering-manager.cc
View File

@ -297,7 +297,7 @@ OptimizationDecision TieringManager::ShouldOptimize(JSFunction function,
int jump_target_offset = iterator.GetJumpTargetOffset();
if (jump_offset >= current_offset &&
current_offset >= jump_target_offset) {
bytecode.set_osr_loop_nesting_level(iterator.GetImmediateOperand(1) +
bytecode.set_osr_loop_nesting_level(iterator.GetJumpLoopNestingLevel() +
1);
return OptimizationDecision::TurbofanHotAndStable();
}

1
src/heap/factory.cc
View File

@ -180,6 +180,7 @@ MaybeHandle<Code> Factory::CodeBuilder::BuildInternal(
// this field. We currently assume it's immutable thus a relaxed read (after
// passing IsPendingAllocation).
raw_code.set_inlined_bytecode_size(inlined_bytecode_size_);
raw_code.set_osr_offset(osr_offset_);
raw_code.set_code_data_container(*data_container, kReleaseStore);
if (kind_ == CodeKind::BASELINE) {
raw_code.set_bytecode_or_interpreter_data(*interpreter_data_);

7
src/heap/factory.h
View File

@ -913,6 +913,12 @@ class V8_EXPORT_PRIVATE Factory : public FactoryBase<Factory> {
return *this;
}
CodeBuilder& set_osr_offset(BytecodeOffset offset) {
DCHECK_IMPLIES(!offset.IsNone(), CodeKindIsOptimizedJSFunction(kind_));
osr_offset_ = offset;
return *this;
}
CodeBuilder& set_source_position_table(Handle<ByteArray> table) {
DCHECK_NE(kind_, CodeKind::BASELINE);
DCHECK(!table.is_null());
@ -990,6 +996,7 @@ class V8_EXPORT_PRIVATE Factory : public FactoryBase<Factory> {
MaybeHandle<Object> self_reference_;
Builtin builtin_ = Builtin::kNoBuiltinId;
uint32_t inlined_bytecode_size_ = 0;
BytecodeOffset osr_offset_ = BytecodeOffset::None();
int32_t kind_specific_flags_ = 0;
// Either source_position_table for non-baseline code
// or bytecode_offset_table for baseline code.

5
src/interpreter/bytecode-array-iterator.cc
View File

@ -244,6 +244,11 @@ int BytecodeArrayIterator::GetJumpTargetOffset() const {
return GetAbsoluteOffset(GetRelativeJumpTargetOffset());
}
int BytecodeArrayIterator::GetJumpLoopNestingLevel() const {
DCHECK_EQ(current_bytecode(), interpreter::Bytecode::kJumpLoop);
return GetImmediateOperand(1);
}
JumpTableTargetOffsets BytecodeArrayIterator::GetJumpTableTargetOffsets()
const {
uint32_t table_start, table_size;

4
src/interpreter/bytecode-array-iterator.h
View File

@ -141,6 +141,10 @@ class V8_EXPORT_PRIVATE BytecodeArrayIterator {
// bytecode is not a switch.
JumpTableTargetOffsets GetJumpTableTargetOffsets() const;
// Returns loop nesting level of current bytecode.
// It is an error to call this method if the bytecode is not for a JumpLoop.
int GetJumpLoopNestingLevel() const;
// Returns the absolute offset of the bytecode at the given relative offset
// from the current bytecode.
int GetAbsoluteOffset(int relative_offset) const;

9
src/objects/code-inl.h
View File

@ -694,6 +694,15 @@ void Code::set_inlined_bytecode_size(unsigned size) {
RELAXED_WRITE_UINT_FIELD(*this, kInlinedBytecodeSizeOffset, size);
}
BytecodeOffset Code::osr_offset() const {
int32_t offset = RELAXED_READ_INT32_FIELD(*this, kOsrOffsetOffset);
return BytecodeOffset(offset);
}
void Code::set_osr_offset(BytecodeOffset offset) {
RELAXED_WRITE_INT32_FIELD(*this, kOsrOffsetOffset, offset.ToInt());
}
bool Code::uses_safepoint_table() const {
return is_turbofanned() || is_maglevved() || is_wasm_code();
}

26
src/objects/code.h
View File

@ -455,6 +455,9 @@ class Code : public HeapObject {
inline unsigned inlined_bytecode_size() const;
inline void set_inlined_bytecode_size(unsigned size);
inline BytecodeOffset osr_offset() const;
inline void set_osr_offset(BytecodeOffset offset);
// [uses_safepoint_table]: Whether this Code object uses safepoint tables
// (note the table may still be empty, see has_safepoint_table).
inline bool uses_safepoint_table() const;
@ -633,6 +636,7 @@ class Code : public HeapObject {
V(kFlagsOffset, kInt32Size) \
V(kBuiltinIndexOffset, kIntSize) \
V(kInlinedBytecodeSizeOffset, kIntSize) \
V(kOsrOffsetOffset, kInt32Size) \
/* Offsets describing inline metadata tables, relative to MetadataStart. */ \
V(kHandlerTableOffsetOffset, kIntSize) \
V(kConstantPoolOffsetOffset, \
@ -652,28 +656,28 @@ class Code : public HeapObject {
// due to padding for code alignment.
#if V8_TARGET_ARCH_ARM64
static constexpr int kHeaderPaddingSize =
V8_EXTERNAL_CODE_SPACE_BOOL ? 8 : (COMPRESS_POINTERS_BOOL ? 12 : 24);
V8_EXTERNAL_CODE_SPACE_BOOL ? 4 : (COMPRESS_POINTERS_BOOL ? 8 : 20);
#elif V8_TARGET_ARCH_MIPS64
static constexpr int kHeaderPaddingSize = 24;
static constexpr int kHeaderPaddingSize = 20;
#elif V8_TARGET_ARCH_LOONG64
static constexpr int kHeaderPaddingSize = 24;
static constexpr int kHeaderPaddingSize = 20;
#elif V8_TARGET_ARCH_X64
static constexpr int kHeaderPaddingSize =
V8_EXTERNAL_CODE_SPACE_BOOL ? 8 : (COMPRESS_POINTERS_BOOL ? 12 : 56);
V8_EXTERNAL_CODE_SPACE_BOOL ? 4 : (COMPRESS_POINTERS_BOOL ? 8 : 52);
#elif V8_TARGET_ARCH_ARM
static constexpr int kHeaderPaddingSize = 12;
static constexpr int kHeaderPaddingSize = 8;
#elif V8_TARGET_ARCH_IA32
static constexpr int kHeaderPaddingSize = 12;
static constexpr int kHeaderPaddingSize = 8;
#elif V8_TARGET_ARCH_MIPS
static constexpr int kHeaderPaddingSize = 12;
static constexpr int kHeaderPaddingSize = 8;
#elif V8_TARGET_ARCH_PPC64
static constexpr int kHeaderPaddingSize =
FLAG_enable_embedded_constant_pool ? (COMPRESS_POINTERS_BOOL ? 8 : 20)
: (COMPRESS_POINTERS_BOOL ? 12 : 24);
FLAG_enable_embedded_constant_pool ? (COMPRESS_POINTERS_BOOL ? 4 : 16)
: (COMPRESS_POINTERS_BOOL ? 8 : 20);
#elif V8_TARGET_ARCH_S390X
static constexpr int kHeaderPaddingSize = COMPRESS_POINTERS_BOOL ? 12 : 24;
static constexpr int kHeaderPaddingSize = COMPRESS_POINTERS_BOOL ? 8 : 20;
#elif V8_TARGET_ARCH_RISCV64
static constexpr int kHeaderPaddingSize = (COMPRESS_POINTERS_BOOL ? 12 : 24);
static constexpr int kHeaderPaddingSize = (COMPRESS_POINTERS_BOOL ? 8 : 20);
#else
#error Unknown architecture.
#endif

30
src/runtime/runtime-compiler.cc
View File

@ -196,6 +196,8 @@ RUNTIME_FUNCTION(Runtime_NotifyDeoptimized) {
// Make sure to materialize objects before causing any allocation.
deoptimizer->MaterializeHeapObjects();
BytecodeOffset deopt_exit_bytecode_offset =
deoptimizer->deopt_exit_bytecode_offset();
delete deoptimizer;
// Ensure the context register is updated for materialized objects.
@ -205,7 +207,33 @@ RUNTIME_FUNCTION(Runtime_NotifyDeoptimized) {
// Invalidate the underlying optimized code on eager and soft deopts.
if (type == DeoptimizeKind::kEager || type == DeoptimizeKind::kSoft) {
Deoptimizer::DeoptimizeFunction(*function, *optimized_code);
// Non-OSR'd code is deoptimized unconditionally.
//
// For OSR'd code, we keep the optimized code around if deoptimization
// occurs outside the outermost loop (containing the loop that triggered OSR
// compilation). The reasoning is that OSR is intended to speed up the
// long-running loop; so if the deoptimization occurs outside this loop it
// is still worth jumping to the OSR'd code on the next run. The reduced
// cost of the loop should pay for the deoptimization costs.
if (!optimized_code->osr_offset().IsNone()) {
interpreter::BytecodeArrayIterator iterator(
Handle<BytecodeArray>(function->shared().GetBytecodeArray(isolate),
isolate),
optimized_code->osr_offset().ToInt());
// Iterate forward from current JumpLoop to the outermost JumpLoop.
for (; !iterator.done(); iterator.Advance()) {
if (deopt_exit_bytecode_offset.ToInt() <= iterator.current_offset()) {
Deoptimizer::DeoptimizeFunction(*function, *optimized_code);
break;
}
if (iterator.current_bytecode() == interpreter::Bytecode::kJumpLoop &&
iterator.GetJumpLoopNestingLevel() == 0) {
break;
}
}
} else {
Deoptimizer::DeoptimizeFunction(*function, *optimized_code);
}
}
return ReadOnlyRoots(isolate).undefined_value();