Introduce short external strings without pointer cache.

BUG=
TEST=

Review URL: http://codereview.chromium.org/8635011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10049 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This commit is contained in:
yangguo@chromium.org 2011-11-23 09:58:58 +00:00
parent 46f0a651cc
commit 21edc7c30b
13 changed files with 168 additions and 77 deletions

View File

@ -4556,12 +4556,9 @@ bool v8::String::CanMakeExternal() {
i::Handle<i::String> obj = Utils::OpenHandle(this);
i::Isolate* isolate = obj->GetIsolate();
if (IsDeadCheck(isolate, "v8::String::CanMakeExternal()")) return false;
if (isolate->string_tracker()->IsFreshUnusedString(obj)) {
return false;
}
if (isolate->string_tracker()->IsFreshUnusedString(obj)) return false;
int size = obj->Size(); // Byte size of the original string.
if (size < i::ExternalString::kSize)
return false;
if (size < i::ExternalString::kShortSize) return false;
i::StringShape shape(*obj);
return !shape.IsExternal();
}

View File

@ -255,16 +255,11 @@ void Heap::FinalizeExternalString(String* string) {
ExternalString::kResourceOffset -
kHeapObjectTag);
// Clear pointer cache.
ExternalString::cast(string)->clear_data_cache();
// Dispose of the C++ object if it has not already been disposed.
if (*resource_addr != NULL) {
(*resource_addr)->Dispose();
*resource_addr = NULL;
}
// Clear the resource pointer in the string.
*resource_addr = NULL;
}

View File

@ -3997,6 +3997,15 @@ Map* Heap::SymbolMapForString(String* string) {
if (map == external_string_with_ascii_data_map()) {
return external_symbol_with_ascii_data_map();
}
if (map == short_external_string_map()) {
return short_external_symbol_map();
}
if (map == short_external_ascii_string_map()) {
return short_external_ascii_symbol_map();
}
if (map == short_external_string_with_ascii_data_map()) {
return short_external_symbol_with_ascii_data_map();
}
// No match found.
return NULL;

View File

@ -110,6 +110,16 @@ inline Heap* _inline_get_heap_();
V(Map, external_string_map, ExternalStringMap) \
V(Map, external_string_with_ascii_data_map, ExternalStringWithAsciiDataMap) \
V(Map, external_ascii_string_map, ExternalAsciiStringMap) \
V(Map, short_external_symbol_map, ShortExternalSymbolMap) \
V(Map, \
short_external_symbol_with_ascii_data_map, \
ShortExternalSymbolWithAsciiDataMap) \
V(Map, short_external_ascii_symbol_map, ShortExternalAsciiSymbolMap) \
V(Map, short_external_string_map, ShortExternalStringMap) \
V(Map, \
short_external_string_with_ascii_data_map, \
ShortExternalStringWithAsciiDataMap) \
V(Map, short_external_ascii_string_map, ShortExternalAsciiStringMap) \
V(Map, undetectable_string_map, UndetectableStringMap) \
V(Map, undetectable_ascii_string_map, UndetectableAsciiStringMap) \
V(Map, external_pixel_array_map, ExternalPixelArrayMap) \

View File

@ -563,15 +563,17 @@ void StringCharLoadGenerator::Generate(MacroAssembler* masm,
__ Assert(zero, "external string expected, but not found");
}
__ mov(result, FieldOperand(string, ExternalString::kResourceDataOffset));
// Assert that the external string has not been finalized yet.
__ test(result, result);
__ j(zero, call_runtime);
Register scratch = string;
__ mov(scratch, FieldOperand(string, HeapObject::kMapOffset));
__ cmp(scratch, Immediate(factory->external_ascii_string_map()));
__ j(equal, &ascii_external, Label::kNear);
__ cmp(scratch, Immediate(factory->external_ascii_symbol_map()));
__ j(equal, &ascii_external, Label::kNear);
__ movzx_b(scratch, FieldOperand(scratch, Map::kInstanceTypeOffset));
// Rule out short external strings.
STATIC_CHECK(kShortExternalStringTag != 0);
__ test_b(scratch, kShortExternalStringMask);
__ j(not_zero, call_runtime);
// Check encoding.
STATIC_ASSERT(kTwoByteStringTag == 0);
__ test_b(scratch, kStringEncodingMask);
__ j(not_equal, &ascii_external, Label::kNear);
// Two-byte string.
__ movzx_w(result, Operand(result, index, times_2, 0));
__ jmp(&done);

View File

@ -2297,8 +2297,9 @@ void ConsString::set_second(String* value, WriteBarrierMode mode) {
}
void ExternalString::clear_data_cache() {
WRITE_INTPTR_FIELD(this, kResourceDataOffset, 0);
bool ExternalString::is_short() {
InstanceType type = map()->instance_type();
return (type & kShortExternalStringMask) == kShortExternalStringTag;
}
@ -2307,19 +2308,24 @@ const ExternalAsciiString::Resource* ExternalAsciiString::resource() {
}
void ExternalAsciiString::update_data_cache() {
if (is_short()) return;
const char** data_field =
reinterpret_cast<const char**>(FIELD_ADDR(this, kResourceDataOffset));
*data_field = resource()->data();
}
void ExternalAsciiString::set_resource(
const ExternalAsciiString::Resource* resource) {
*reinterpret_cast<const Resource**>(
FIELD_ADDR(this, kResourceOffset)) = resource;
clear_data_cache();
if (resource != NULL) update_data_cache();
}
const char* ExternalAsciiString::GetChars() {
const char** data_field =
reinterpret_cast<const char**>(FIELD_ADDR(this, kResourceDataOffset));
if (*data_field == NULL) *data_field = resource()->data();
return *data_field;
return resource()->data();
}
@ -2334,19 +2340,24 @@ const ExternalTwoByteString::Resource* ExternalTwoByteString::resource() {
}
void ExternalTwoByteString::update_data_cache() {
if (is_short()) return;
const uint16_t** data_field =
reinterpret_cast<const uint16_t**>(FIELD_ADDR(this, kResourceDataOffset));
*data_field = resource()->data();
}
void ExternalTwoByteString::set_resource(
const ExternalTwoByteString::Resource* resource) {
*reinterpret_cast<const Resource**>(
FIELD_ADDR(this, kResourceOffset)) = resource;
clear_data_cache();
if (resource != NULL) update_data_cache();
}
const uint16_t* ExternalTwoByteString::GetChars() {
const uint16_t** data_field =
reinterpret_cast<const uint16_t**>(FIELD_ADDR(this, kResourceDataOffset));
if (*data_field == NULL) *data_field = resource()->data();
return *data_field;
return resource()->data();
}

View File

@ -446,6 +446,9 @@ static const char* TypeToString(InstanceType type) {
case EXTERNAL_ASCII_SYMBOL_TYPE:
case EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE:
case EXTERNAL_SYMBOL_TYPE: return "EXTERNAL_SYMBOL";
case SHORT_EXTERNAL_ASCII_SYMBOL_TYPE:
case SHORT_EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE:
case SHORT_EXTERNAL_SYMBOL_TYPE: return "SHORT_EXTERNAL_SYMBOL";
case ASCII_STRING_TYPE: return "ASCII_STRING";
case STRING_TYPE: return "TWO_BYTE_STRING";
case CONS_STRING_TYPE:
@ -453,6 +456,9 @@ static const char* TypeToString(InstanceType type) {
case EXTERNAL_ASCII_STRING_TYPE:
case EXTERNAL_STRING_WITH_ASCII_DATA_TYPE:
case EXTERNAL_STRING_TYPE: return "EXTERNAL_STRING";
case SHORT_EXTERNAL_ASCII_STRING_TYPE:
case SHORT_EXTERNAL_STRING_WITH_ASCII_DATA_TYPE:
case SHORT_EXTERNAL_STRING_TYPE: return "SHORT_EXTERNAL_STRING";
case FIXED_ARRAY_TYPE: return "FIXED_ARRAY";
case BYTE_ARRAY_TYPE: return "BYTE_ARRAY";
case FREE_SPACE_TYPE: return "FREE_SPACE";

View File

@ -64,7 +64,7 @@ StaticVisitorBase::VisitorId StaticVisitorBase::GetVisitorId(
case kExternalStringTag:
return GetVisitorIdForSize(kVisitDataObject,
kVisitDataObjectGeneric,
ExternalString::kSize);
instance_size);
}
UNREACHABLE();
}

View File

@ -952,33 +952,32 @@ bool String::MakeExternal(v8::String::ExternalStringResource* resource) {
#endif // DEBUG
Heap* heap = GetHeap();
int size = this->Size(); // Byte size of the original string.
if (size < ExternalString::kSize) {
if (size < ExternalString::kShortSize) {
return false;
}
ASSERT(size >= ExternalString::kSize);
bool is_ascii = this->IsAsciiRepresentation();
bool is_symbol = this->IsSymbol();
int length = this->length();
int hash_field = this->hash_field();
// Morph the object to an external string by adjusting the map and
// reinitializing the fields.
this->set_map(is_ascii ?
heap->external_string_with_ascii_data_map() :
heap->external_string_map());
ExternalTwoByteString* self = ExternalTwoByteString::cast(this);
self->set_length(length);
self->set_hash_field(hash_field);
self->set_resource(resource);
// Additionally make the object into an external symbol if the original string
// was a symbol to start with.
if (is_symbol) {
self->Hash(); // Force regeneration of the hash value.
// Now morph this external string into a external symbol.
this->set_map(is_ascii ?
heap->external_symbol_with_ascii_data_map() :
heap->external_symbol_map());
if (size >= ExternalString::kSize) {
this->set_map(
is_symbol
? (is_ascii ? heap->external_symbol_with_ascii_data_map()
: heap->external_symbol_map())
: (is_ascii ? heap->external_string_with_ascii_data_map()
: heap->external_string_map()));
} else {
this->set_map(
is_symbol
? (is_ascii ? heap->short_external_symbol_with_ascii_data_map()
: heap->short_external_symbol_map())
: (is_ascii ? heap->short_external_string_with_ascii_data_map()
: heap->short_external_string_map()));
}
ExternalTwoByteString* self = ExternalTwoByteString::cast(this);
self->set_resource(resource);
if (is_symbol) self->Hash(); // Force regeneration of the hash value.
// Fill the remainder of the string with dead wood.
int new_size = this->Size(); // Byte size of the external String object.
@ -1004,28 +1003,23 @@ bool String::MakeExternal(v8::String::ExternalAsciiStringResource* resource) {
#endif // DEBUG
Heap* heap = GetHeap();
int size = this->Size(); // Byte size of the original string.
if (size < ExternalString::kSize) {
if (size < ExternalString::kShortSize) {
return false;
}
ASSERT(size >= ExternalString::kSize);
bool is_symbol = this->IsSymbol();
int length = this->length();
int hash_field = this->hash_field();
// Morph the object to an external string by adjusting the map and
// reinitializing the fields.
this->set_map(heap->external_ascii_string_map());
ExternalAsciiString* self = ExternalAsciiString::cast(this);
self->set_length(length);
self->set_hash_field(hash_field);
self->set_resource(resource);
// Additionally make the object into an external symbol if the original string
// was a symbol to start with.
if (is_symbol) {
self->Hash(); // Force regeneration of the hash value.
// Now morph this external string into a external symbol.
this->set_map(heap->external_ascii_symbol_map());
// reinitializing the fields. Use short version if space is limited.
if (size >= ExternalString::kSize) {
this->set_map(is_symbol ? heap->external_ascii_symbol_map()
: heap->external_ascii_string_map());
} else {
this->set_map(is_symbol ? heap->short_external_ascii_symbol_map()
: heap->short_external_ascii_string_map());
}
ExternalAsciiString* self = ExternalAsciiString::cast(this);
self->set_resource(resource);
if (is_symbol) self->Hash(); // Force regeneration of the hash value.
// Fill the remainder of the string with dead wood.
int new_size = this->Size(); // Byte size of the external String object.
@ -1033,7 +1027,6 @@ bool String::MakeExternal(v8::String::ExternalAsciiStringResource* resource) {
if (Marking::IsBlack(Marking::MarkBitFrom(this))) {
MemoryChunk::IncrementLiveBytes(this->address(), new_size - size);
}
return true;
}

View File

@ -232,6 +232,9 @@ static const int kVariableSizeSentinel = 0;
V(EXTERNAL_SYMBOL_TYPE) \
V(EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE) \
V(EXTERNAL_ASCII_SYMBOL_TYPE) \
V(SHORT_EXTERNAL_SYMBOL_TYPE) \
V(SHORT_EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE) \
V(SHORT_EXTERNAL_ASCII_SYMBOL_TYPE) \
V(STRING_TYPE) \
V(ASCII_STRING_TYPE) \
V(CONS_STRING_TYPE) \
@ -240,6 +243,9 @@ static const int kVariableSizeSentinel = 0;
V(EXTERNAL_STRING_TYPE) \
V(EXTERNAL_STRING_WITH_ASCII_DATA_TYPE) \
V(EXTERNAL_ASCII_STRING_TYPE) \
V(SHORT_EXTERNAL_STRING_TYPE) \
V(SHORT_EXTERNAL_STRING_WITH_ASCII_DATA_TYPE) \
V(SHORT_EXTERNAL_ASCII_STRING_TYPE) \
V(PRIVATE_EXTERNAL_ASCII_STRING_TYPE) \
\
V(MAP_TYPE) \
@ -340,6 +346,18 @@ static const int kVariableSizeSentinel = 0;
ExternalAsciiString::kSize, \
external_ascii_symbol, \
ExternalAsciiSymbol) \
V(SHORT_EXTERNAL_SYMBOL_TYPE, \
ExternalTwoByteString::kShortSize, \
short_external_symbol, \
ShortExternalSymbol) \
V(SHORT_EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE, \
ExternalTwoByteString::kShortSize, \
short_external_symbol_with_ascii_data, \
ShortExternalSymbolWithAsciiData) \
V(SHORT_EXTERNAL_ASCII_SYMBOL_TYPE, \
ExternalAsciiString::kShortSize, \
short_external_ascii_symbol, \
ShortExternalAsciiSymbol) \
V(STRING_TYPE, \
kVariableSizeSentinel, \
string, \
@ -375,7 +393,19 @@ static const int kVariableSizeSentinel = 0;
V(EXTERNAL_ASCII_STRING_TYPE, \
ExternalAsciiString::kSize, \
external_ascii_string, \
ExternalAsciiString)
ExternalAsciiString) \
V(SHORT_EXTERNAL_STRING_TYPE, \
ExternalTwoByteString::kShortSize, \
short_external_string, \
ShortExternalString) \
V(SHORT_EXTERNAL_STRING_WITH_ASCII_DATA_TYPE, \
ExternalTwoByteString::kShortSize, \
short_external_string_with_ascii_data, \
ShortExternalStringWithAsciiData) \
V(SHORT_EXTERNAL_ASCII_STRING_TYPE, \
ExternalAsciiString::kShortSize, \
short_external_ascii_string, \
ShortExternalAsciiString)
// A struct is a simple object a set of object-valued fields. Including an
// object type in this causes the compiler to generate most of the boilerplate
@ -459,6 +489,11 @@ STATIC_ASSERT(IS_POWER_OF_TWO(kSlicedNotConsMask) && kSlicedNotConsMask != 0);
const uint32_t kAsciiDataHintMask = 0x08;
const uint32_t kAsciiDataHintTag = 0x08;
// If bit 7 is clear and string representation indicates an external string,
// then bit 4 indicates whether the data pointer is cached.
const uint32_t kShortExternalStringMask = 0x10;
const uint32_t kShortExternalStringTag = 0x10;
// A ConsString with an empty string as the right side is a candidate
// for being shortcut by the garbage collector unless it is a
@ -478,6 +513,13 @@ enum InstanceType {
ASCII_SYMBOL_TYPE = kAsciiStringTag | kSymbolTag | kSeqStringTag,
CONS_SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag | kConsStringTag,
CONS_ASCII_SYMBOL_TYPE = kAsciiStringTag | kSymbolTag | kConsStringTag,
SHORT_EXTERNAL_SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag |
kExternalStringTag | kShortExternalStringTag,
SHORT_EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE =
kTwoByteStringTag | kSymbolTag | kExternalStringTag |
kAsciiDataHintTag | kShortExternalStringTag,
SHORT_EXTERNAL_ASCII_SYMBOL_TYPE = kAsciiStringTag | kExternalStringTag |
kSymbolTag | kShortExternalStringTag,
EXTERNAL_SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag | kExternalStringTag,
EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE =
kTwoByteStringTag | kSymbolTag | kExternalStringTag | kAsciiDataHintTag,
@ -489,6 +531,13 @@ enum InstanceType {
CONS_ASCII_STRING_TYPE = kAsciiStringTag | kConsStringTag,
SLICED_STRING_TYPE = kTwoByteStringTag | kSlicedStringTag,
SLICED_ASCII_STRING_TYPE = kAsciiStringTag | kSlicedStringTag,
SHORT_EXTERNAL_STRING_TYPE =
kTwoByteStringTag | kExternalStringTag | kShortExternalStringTag,
SHORT_EXTERNAL_STRING_WITH_ASCII_DATA_TYPE =
kTwoByteStringTag | kExternalStringTag |
kAsciiDataHintTag | kShortExternalStringTag,
SHORT_EXTERNAL_ASCII_STRING_TYPE =
kAsciiStringTag | kExternalStringTag | kShortExternalStringTag,
EXTERNAL_STRING_TYPE = kTwoByteStringTag | kExternalStringTag,
EXTERNAL_STRING_WITH_ASCII_DATA_TYPE =
kTwoByteStringTag | kExternalStringTag | kAsciiDataHintTag,
@ -6755,12 +6804,12 @@ class ExternalString: public String {
// Layout description.
static const int kResourceOffset = POINTER_SIZE_ALIGN(String::kSize);
static const int kShortSize = kResourceOffset + kPointerSize;
static const int kResourceDataOffset = kResourceOffset + kPointerSize;
static const int kSize = kResourceDataOffset + kPointerSize;
// Clear the cached pointer to the character array provided by the resource.
// This cache is updated the first time the character array is accessed.
inline void clear_data_cache();
// Return whether external string is short (data pointer is not cached).
inline bool is_short();
STATIC_CHECK(kResourceOffset == Internals::kStringResourceOffset);
@ -6781,6 +6830,12 @@ class ExternalAsciiString: public ExternalString {
inline const Resource* resource();
inline void set_resource(const Resource* buffer);
// Update the pointer cache to the external character array.
// The cached pointer is always valid, as the external character array does =
// not move during lifetime. Deserialization is the only exception, after
// which the pointer cache has to be refreshed.
inline void update_data_cache();
inline const char* GetChars();
// Dispatched behavior.
@ -6820,6 +6875,12 @@ class ExternalTwoByteString: public ExternalString {
inline const Resource* resource();
inline void set_resource(const Resource* buffer);
// Update the pointer cache to the external character array.
// The cached pointer is always valid, as the external character array does =
// not move during lifetime. Deserialization is the only exception, after
// which the pointer cache has to be refreshed.
inline void update_data_cache();
inline const uint16_t* GetChars();
// Dispatched behavior.

View File

@ -669,6 +669,14 @@ void Deserializer::Deserialize() {
isolate_->heap()->set_global_contexts_list(
isolate_->heap()->undefined_value());
// Update data pointers to the external strings containing natives sources.
for (int i = 0; i < Natives::GetBuiltinsCount(); i++) {
Object* source = isolate_->heap()->natives_source_cache()->get(i);
if (!source->IsUndefined()) {
ExternalAsciiString::cast(source)->update_data_cache();
}
}
}
@ -1564,7 +1572,6 @@ void Serializer::ObjectSerializer::VisitExternalAsciiString(
sink_->Put(kNativesStringResource, "NativesStringResource");
sink_->PutSection(i, "NativesStringResourceEnd");
bytes_processed_so_far_ += sizeof(resource);
string->clear_data_cache();
return;
}
}

View File

@ -491,7 +491,7 @@ TEST(MakingExternalStringConditions) {
HEAP->CollectGarbage(i::NEW_SPACE);
HEAP->CollectGarbage(i::NEW_SPACE);
uint16_t* two_byte_string = AsciiToTwoByteString("abcdefghi");
uint16_t* two_byte_string = AsciiToTwoByteString("s1");
Local<String> small_string = String::New(two_byte_string);
i::DeleteArray(two_byte_string);
@ -503,7 +503,7 @@ TEST(MakingExternalStringConditions) {
// Old space strings should be accepted.
CHECK(small_string->CanMakeExternal());
two_byte_string = AsciiToTwoByteString("abcdefghi");
two_byte_string = AsciiToTwoByteString("small string 2");
small_string = String::New(two_byte_string);
i::DeleteArray(two_byte_string);
@ -537,7 +537,7 @@ TEST(MakingExternalAsciiStringConditions) {
HEAP->CollectGarbage(i::NEW_SPACE);
HEAP->CollectGarbage(i::NEW_SPACE);
Local<String> small_string = String::New("abcdefghi");
Local<String> small_string = String::New("s1");
// We should refuse to externalize newly created small string.
CHECK(!small_string->CanMakeExternal());
// Trigger GCs so that the newly allocated string moves to old gen.
@ -546,7 +546,7 @@ TEST(MakingExternalAsciiStringConditions) {
// Old space strings should be accepted.
CHECK(small_string->CanMakeExternal());
small_string = String::New("abcdefghi");
small_string = String::New("small string 2");
// We should refuse externalizing newly created small string.
CHECK(!small_string->CanMakeExternal());
for (int i = 0; i < 100; i++) {

View File

@ -39,7 +39,7 @@ function test() {
assertTrue(isAsciiString(str));
var twoByteExternalWithAsciiData =
"AAAAAAAA" + (function() { return "A"; })();
"AA" + (function() { return "A"; })();
externalizeString(twoByteExternalWithAsciiData, true /* force two-byte */);
assertFalse(isAsciiString(twoByteExternalWithAsciiData));