[string] Extend StringShape to query the shared bit

Rename StringShape::full_representation_tag to
StringShape::representation_and_encoding_tag, since the full
representation tag now includes the shared bit.

There are no users of the new method in this CL; this is split out to
make subsequent shared string CLs smaller.

Bug: v8:12007
Change-Id: Ic4ac0241fd9846241e85b4a094dfee6d201ba42b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3313428Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Patrick Thier <pthier@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/main@{#78253}

include/v8-internal.h

@@ -218,7 +218,7 @@ class Internals {
   static const int kEmbedderDataSlotRawPayloadOffset = kApiTaggedSize;
 #endif
   static const int kNativeContextEmbedderDataOffset = 6 * kApiTaggedSize;
-  static const int kFullStringRepresentationMask = 0x0f;
+  static const int kStringRepresentationAndEncodingMask = 0x0f;
   static const int kStringEncodingMask = 0x8;
   static const int kExternalTwoByteRepresentationTag = 0x02;
   static const int kExternalOneByteRepresentationTag = 0x0a;
@@ -337,7 +337,7 @@ class Internals {
   }
 
   V8_INLINE static bool IsExternalTwoByteString(int instance_type) {
-    int representation = (instance_type & kFullStringRepresentationMask);
+    int representation = (instance_type & kStringRepresentationAndEncodingMask);
     return representation == kExternalTwoByteRepresentationTag;
   }
 

include/v8-primitive.h

@@ -806,7 +806,7 @@ String::ExternalStringResourceBase* String::GetExternalStringResourceBase(
   using A = internal::Address;
   using I = internal::Internals;
   A obj = *reinterpret_cast<const A*>(this);
-  int type = I::GetInstanceType(obj) & I::kFullStringRepresentationMask;
+  int type = I::GetInstanceType(obj) & I::kStringRepresentationAndEncodingMask;
   *encoding_out = static_cast<Encoding>(type & I::kStringEncodingMask);
   ExternalStringResourceBase* resource;
   if (type == I::kExternalOneByteRepresentationTag ||

src/api/api.cc

@@ -5870,7 +5870,8 @@ String::ExternalStringResourceBase* String::GetExternalStringResourceBaseSlow(
   }
 
   internal::Address string = str.ptr();
-  int type = I::GetInstanceType(string) & I::kFullStringRepresentationMask;
+  int type =
+      I::GetInstanceType(string) & I::kStringRepresentationAndEncodingMask;
   *encoding_out = static_cast<Encoding>(type & I::kStringEncodingMask);
   if (i::StringShape(str).IsExternalOneByte() ||
       i::StringShape(str).IsExternalTwoByte()) {

src/objects/instance-type.h

@@ -85,6 +85,9 @@ const uint32_t kInternalizedTag = 0;
 const uint32_t kSharedStringMask = 1 << 6;
 const uint32_t kSharedStringTag = 1 << 6;
 
+constexpr uint32_t kStringRepresentationEncodingAndSharedMask =
+    kStringRepresentationAndEncodingMask | kSharedStringMask;
+
 // A ConsString with an empty string as the right side is a candidate
 // for being shortcut by the garbage collector. We don't allocate any
 // non-flat internalized strings, so we do not shortcut them thereby

src/objects/string-inl.h

@@ -179,6 +179,24 @@ bool StringShape::IsShared() const {
          (FLAG_shared_string_table && IsInternalized());
 }
 
+bool StringShape::CanMigrateInParallel() const {
+  switch (representation_encoding_and_shared_tag()) {
+    case kSeqOneByteStringTag | kSharedStringTag:
+    case kSeqTwoByteStringTag | kSharedStringTag:
+      // Shared SeqStrings can migrate to ThinStrings.
+      return true;
+    case kThinStringTag | kOneByteStringTag | kSharedStringTag:
+    case kThinStringTag | kTwoByteStringTag | kSharedStringTag:
+      // Shared ThinStrings do not migrate.
+      return false;
+    default:
+      // If you crashed here, you probably added a new shared string
+      // type. Explicitly handle all shared string cases above.
+      DCHECK(!IsShared());
+      return false;
+  }
+}
+
 StringRepresentationTag StringShape::representation_tag() const {
   uint32_t tag = (type_ & kStringRepresentationMask);
   return static_cast<StringRepresentationTag>(tag);
@@ -188,26 +206,30 @@ uint32_t StringShape::encoding_tag() const {
   return type_ & kStringEncodingMask;
 }
 
-uint32_t StringShape::full_representation_tag() const {
+uint32_t StringShape::representation_and_encoding_tag() const {
   return (type_ & (kStringRepresentationAndEncodingMask));
 }
 
+uint32_t StringShape::representation_encoding_and_shared_tag() const {
+  return (type_ & (kStringRepresentationEncodingAndSharedMask));
+}
+
 STATIC_ASSERT((kStringRepresentationAndEncodingMask) ==
-              Internals::kFullStringRepresentationMask);
+              Internals::kStringRepresentationAndEncodingMask);
 
 STATIC_ASSERT(static_cast<uint32_t>(kStringEncodingMask) ==
               Internals::kStringEncodingMask);
 
 bool StringShape::IsSequentialOneByte() const {
-  return full_representation_tag() == kSeqOneByteStringTag;
+  return representation_and_encoding_tag() == kSeqOneByteStringTag;
 }
 
 bool StringShape::IsSequentialTwoByte() const {
-  return full_representation_tag() == kSeqTwoByteStringTag;
+  return representation_and_encoding_tag() == kSeqTwoByteStringTag;
 }
 
 bool StringShape::IsExternalOneByte() const {
-  return full_representation_tag() == kExternalOneByteStringTag;
+  return representation_and_encoding_tag() == kExternalOneByteStringTag;
 }
 
 STATIC_ASSERT(kExternalOneByteStringTag ==
@@ -216,7 +238,7 @@ STATIC_ASSERT(kExternalOneByteStringTag ==
 STATIC_ASSERT(v8::String::ONE_BYTE_ENCODING == kOneByteStringTag);
 
 bool StringShape::IsExternalTwoByte() const {
-  return full_representation_tag() == kExternalTwoByteStringTag;
+  return representation_and_encoding_tag() == kExternalTwoByteStringTag;
 }
 
 STATIC_ASSERT(kExternalTwoByteStringTag ==
@@ -226,7 +248,7 @@ STATIC_ASSERT(v8::String::TWO_BYTE_ENCODING == kTwoByteStringTag);
 
 template <typename TDispatcher, typename TResult, typename... TArgs>
 inline TResult StringShape::DispatchToSpecificTypeWithoutCast(TArgs&&... args) {
-  switch (full_representation_tag()) {
+  switch (representation_and_encoding_tag()) {
     case kSeqStringTag | kOneByteStringTag:
       return TDispatcher::HandleSeqOneByteString(std::forward<TArgs>(args)...);
     case kSeqStringTag | kTwoByteStringTag:
@@ -655,7 +677,7 @@ base::Optional<String::FlatContent> String::TryGetFlatContentFromDirectString(
   DCHECK_GE(offset, 0);
   DCHECK_GE(length, 0);
   DCHECK_LE(offset + length, string.length());
-  switch (StringShape{string, cage_base}.full_representation_tag()) {
+  switch (StringShape{string, cage_base}.representation_and_encoding_tag()) {
     case kSeqOneByteStringTag:
       return FlatContent(
           SeqOneByteString::cast(string).GetChars(no_gc, access_guard) + offset,
@@ -825,7 +847,8 @@ ConsString String::VisitFlat(
   DCHECK(offset <= length);
   PtrComprCageBase cage_base = GetPtrComprCageBase(string);
   while (true) {
-    int32_t tag = StringShape(string, cage_base).full_representation_tag();
+    int32_t tag =
+        StringShape(string, cage_base).representation_and_encoding_tag();
     switch (tag) {
       case kSeqOneByteStringTag:
         visitor->VisitOneByteString(

src/objects/string.cc

@@ -711,7 +711,7 @@ void String::WriteToFlat(String source, sinkchar* sink, int start, int length,
     DCHECK_LT(0, length);
     DCHECK_LE(0, start);
     DCHECK_LE(length, source.length());
-    switch (StringShape(source, cage_base).full_representation_tag()) {
+    switch (StringShape(source, cage_base).representation_and_encoding_tag()) {
       case kOneByteStringTag | kExternalStringTag:
         CopyChars(
             sink,
@@ -1815,7 +1815,7 @@ const byte* String::AddressOfCharacterAt(
   }
   CHECK_LE(0, start_index);
   CHECK_LE(start_index, subject.length());
-  switch (shape.full_representation_tag()) {
+  switch (shape.representation_and_encoding_tag()) {
     case kOneByteStringTag | kSeqStringTag:
       return reinterpret_cast<const byte*>(
           SeqOneByteString::cast(subject).GetChars(no_gc) + start_index);

src/objects/string.h

@@ -60,9 +60,11 @@ class StringShape {
   V8_INLINE bool IsSequentialTwoByte() const;
   V8_INLINE bool IsInternalized() const;
   V8_INLINE bool IsShared() const;
+  V8_INLINE bool CanMigrateInParallel() const;
   V8_INLINE StringRepresentationTag representation_tag() const;
   V8_INLINE uint32_t encoding_tag() const;
-  V8_INLINE uint32_t full_representation_tag() const;
+  V8_INLINE uint32_t representation_and_encoding_tag() const;
+  V8_INLINE uint32_t representation_encoding_and_shared_tag() const;
 #ifdef DEBUG
   inline uint32_t type() const { return type_; }
   inline void invalidate() { valid_ = false; }