[string] Add PtrComprCageBase to WriteToFlat, HashString, and Get

Also a couple of microoptimizations and consistent formatting in
WriteToFlat.

Change-Id: Ie642a4b8e0819b04603ee5c5d12eebccf6a2d59c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3151963
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#76799}

src/heap/factory-base.cc

@@ -645,9 +645,10 @@ MaybeHandle<String> FactoryBase<Impl>::NewConsString(
     DisallowGarbageCollection no_gc;
     SharedStringAccessGuardIfNeeded access_guard(isolate());
     base::uc16* sink = result->GetChars(no_gc, access_guard);
-    String::WriteToFlat(*left, sink, 0, left->length(), access_guard);
-    String::WriteToFlat(*right, sink + left->length(), 0, right->length(),
+    String::WriteToFlat(*left, sink, 0, left->length(), isolate(),
                         access_guard);
+    String::WriteToFlat(*right, sink + left->length(), 0, right->length(),
+                        isolate(), access_guard);
     return result;
   }
 

src/numbers/conversions.cc

@@ -1425,7 +1425,7 @@ base::Optional<double> TryStringToDouble(LocalIsolate* isolate,
   const int flags = ALLOW_HEX | ALLOW_OCTAL | ALLOW_BINARY;
   auto buffer = std::make_unique<base::uc16[]>(max_length_for_conversion);
   SharedStringAccessGuardIfNeeded access_guard(isolate);
-  String::WriteToFlat(*object, buffer.get(), 0, length, access_guard);
+  String::WriteToFlat(*object, buffer.get(), 0, length, isolate, access_guard);
   base::Vector<const base::uc16> v(buffer.get(), length);
   return StringToDouble(v, flags);
 }

src/objects/lookup.cc

@@ -1507,7 +1507,8 @@ ConcurrentLookupIterator::Result ConcurrentLookupIterator::TryGetOwnChar(
   uint16_t charcode;
   {
     SharedStringAccessGuardIfNeeded access_guard(local_isolate);
-    charcode = string.Get(static_cast<int>(index));
+    charcode = string.Get(static_cast<int>(index), PtrComprCageBase(isolate),
+                          access_guard);
   }
 
   if (charcode > unibrow::Latin1::kMaxChar) return kGaveUp;

src/objects/string-inl.h

@@ -119,6 +119,12 @@ StringShape::StringShape(const String str)
   DCHECK_EQ(type_ & kIsNotStringMask, kStringTag);
 }
 
+StringShape::StringShape(const String str, PtrComprCageBase cage_base)
+    : type_(str.map(cage_base, kAcquireLoad).instance_type()) {
+  set_valid();
+  DCHECK_EQ(type_ & kIsNotStringMask, kStringTag);
+}
+
 StringShape::StringShape(Map map) : type_(map.instance_type()) {
   set_valid();
   DCHECK_EQ(type_ & kIsNotStringMask, kStringTag);
@@ -617,45 +623,53 @@ Handle<String> String::Flatten(LocalIsolate* isolate, Handle<String> string,
   return string;
 }
 
+uint16_t String::Get(int index) const {
+  DCHECK(!SharedStringAccessGuardIfNeeded::IsNeeded(*this));
+  return GetImpl(index, GetPtrComprCageBase(*this),
+                 SharedStringAccessGuardIfNeeded::NotNeeded());
+}
+
 uint16_t String::Get(int index, Isolate* isolate) const {
   SharedStringAccessGuardIfNeeded scope(isolate);
-  return GetImpl(index, scope);
+  return GetImpl(index, isolate, scope);
 }
 
 uint16_t String::Get(int index, LocalIsolate* local_isolate) const {
   SharedStringAccessGuardIfNeeded scope(local_isolate);
-  return GetImpl(index, scope);
+  return GetImpl(index, local_isolate, scope);
 }
 
 uint16_t String::Get(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
-  return GetImpl(index, access_guard);
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
+  return GetImpl(index, cage_base, access_guard);
 }
 
 uint16_t String::GetImpl(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
   DCHECK(index >= 0 && index < length());
 
   class StringGetDispatcher : public AllStatic {
    public:
 #define DEFINE_METHOD(Type)                                  \
   static inline uint16_t Handle##Type(                       \
-      Type str, int index,                                   \
+      Type str, int index, PtrComprCageBase cage_base,       \
       const SharedStringAccessGuardIfNeeded& access_guard) { \
-    return str.Get(index, access_guard);                     \
+    return str.Get(index, cage_base, access_guard);          \
   }
     STRING_CLASS_TYPES(DEFINE_METHOD)
 #undef DEFINE_METHOD
     static inline uint16_t HandleInvalidString(
-        String str, int index,
+        String str, int index, PtrComprCageBase cage_base,
         const SharedStringAccessGuardIfNeeded& access_guard) {
       UNREACHABLE();
     }
   };
 
   return StringShape(*this)
-      .DispatchToSpecificType<StringGetDispatcher, uint16_t>(*this, index,
-                                                             access_guard);
+      .DispatchToSpecificType<StringGetDispatcher, uint16_t>(
+          *this, index, cage_base, access_guard);
 }
 
 void String::Set(int index, uint16_t value) {
@@ -667,9 +681,11 @@ void String::Set(int index, uint16_t value) {
              : SeqTwoByteString::cast(*this).SeqTwoByteStringSet(index, value);
 }
 
-bool String::IsFlat() const {
-  if (!StringShape(*this).IsCons()) return true;
-  return ConsString::cast(*this).second().length() == 0;
+bool String::IsFlat() const { return IsFlat(GetPtrComprCageBase(*this)); }
+
+bool String::IsFlat(PtrComprCageBase cage_base) const {
+  if (!StringShape(*this, cage_base).IsCons()) return true;
+  return ConsString::cast(*this).second(cage_base).length() == 0;
 }
 
 String String::GetUnderlying() const {
@@ -771,11 +787,13 @@ inline base::Vector<const base::uc16> String::GetCharVector(
 
 uint8_t SeqOneByteString::Get(int index) const {
   DCHECK(!SharedStringAccessGuardIfNeeded::IsNeeded(*this));
-  return Get(index, SharedStringAccessGuardIfNeeded::NotNeeded());
+  return Get(index, GetPtrComprCageBase(*this),
+             SharedStringAccessGuardIfNeeded::NotNeeded());
 }
 
 uint8_t SeqOneByteString::Get(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
   USE(access_guard);
   DCHECK(index >= 0 && index < length());
   return ReadField<byte>(kHeaderSize + index * kCharSize);
@@ -825,7 +843,8 @@ base::uc16* SeqTwoByteString::GetChars(
 }
 
 uint16_t SeqTwoByteString::Get(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
   USE(access_guard);
   DCHECK(index >= 0 && index < length());
   return ReadField<uint16_t>(kHeaderSize + index * kShortSize);
@@ -977,7 +996,8 @@ const uint8_t* ExternalOneByteString::GetChars() const {
 }
 
 uint8_t ExternalOneByteString::Get(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
   USE(access_guard);
   DCHECK(index >= 0 && index < length());
   return GetChars()[index];
@@ -1044,7 +1064,8 @@ const uint16_t* ExternalTwoByteString::GetChars() const {
 }
 
 uint16_t ExternalTwoByteString::Get(
-    int index, const SharedStringAccessGuardIfNeeded& access_guard) const {
+    int index, PtrComprCageBase cage_base,
+    const SharedStringAccessGuardIfNeeded& access_guard) const {
   USE(access_guard);
   DCHECK(index >= 0 && index < length());
   return GetChars()[index];

src/objects/string.h

@@ -10,6 +10,7 @@
 #include "src/base/bits.h"
 #include "src/base/export-template.h"
 #include "src/base/strings.h"
+#include "src/common/globals.h"
 #include "src/objects/instance-type.h"
 #include "src/objects/name.h"
 #include "src/objects/smi.h"
@@ -43,6 +44,7 @@ enum RobustnessFlag { ROBUST_STRING_TRAVERSAL, FAST_STRING_TRAVERSAL };
 class StringShape {
  public:
   inline explicit StringShape(const String s);
+  inline explicit StringShape(const String s, PtrComprCageBase cage_base);
   inline explicit StringShape(Map s);
   inline explicit StringShape(InstanceType t);
   inline bool IsSequential() const;
@@ -220,13 +222,15 @@ class String : public TorqueGeneratedString<String, Name> {
   // to this method are not efficient unless the string is flat.
   // If it is called from a background thread, the LocalIsolate version should
   // be used.
-  V8_INLINE uint16_t Get(int index, Isolate* isolate = nullptr) const;
+  V8_INLINE uint16_t Get(int index) const;
+  V8_INLINE uint16_t Get(int index, Isolate* isolate) const;
   V8_INLINE uint16_t Get(int index, LocalIsolate* local_isolate) const;
   // Method to pass down the access_guard. Useful for recursive calls such as
   // ThinStrings where we go String::Get into ThinString::Get into String::Get
   // again for the internalized string.
   V8_INLINE uint16_t
-  Get(int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+  Get(int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   // ES6 section 7.1.3.1 ToNumber Applied to the String Type
   static Handle<Object> ToNumber(Isolate* isolate, Handle<String> subject);
@@ -429,6 +433,7 @@ class String : public TorqueGeneratedString<String, Name> {
   DECL_VERIFIER(String)
 
   inline bool IsFlat() const;
+  inline bool IsFlat(PtrComprCageBase cage_base) const;
 
   // Max char codes.
   static const int32_t kMaxOneByteCharCode = unibrow::Latin1::kMaxChar;
@@ -474,6 +479,7 @@ class String : public TorqueGeneratedString<String, Name> {
   static void WriteToFlat(String source, sinkchar* sink, int from, int to);
   template <typename sinkchar>
   static void WriteToFlat(String source, sinkchar* sink, int from, int to,
+                          PtrComprCageBase cage_base,
                           const SharedStringAccessGuardIfNeeded&);
 
   static inline bool IsAscii(const char* chars, int length) {
@@ -551,7 +557,8 @@ class String : public TorqueGeneratedString<String, Name> {
 
   // Implementation of the Get() public methods. Do not use directly.
   V8_INLINE uint16_t
-  GetImpl(int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+  GetImpl(int index, PtrComprCageBase cage_base,
+          const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   // Implementation of the IsEqualTo() public methods. Do not use directly.
   template <EqualityType kEqType, typename Char>
@@ -596,11 +603,13 @@ void String::WriteToFlat(String source, uint8_t* sink, int from, int to);
 extern template EXPORT_TEMPLATE_DECLARE(V8_EXPORT_PRIVATE)
 void String::WriteToFlat(String source, uint16_t* sink, int from, int to);
 extern template EXPORT_TEMPLATE_DECLARE(V8_EXPORT_PRIVATE)
-void String::WriteToFlat(String source, uint8_t* sink, int from, int to ,
-                        const SharedStringAccessGuardIfNeeded&);
+void String::WriteToFlat(String source, uint8_t* sink, int from, int to,
+                         PtrComprCageBase cage_base,
+                         const SharedStringAccessGuardIfNeeded&);
 extern template EXPORT_TEMPLATE_DECLARE(V8_EXPORT_PRIVATE)
 void String::WriteToFlat(String source, uint16_t* sink, int from, int to,
-                        const SharedStringAccessGuardIfNeeded&);
+                         PtrComprCageBase cage_base,
+                         const SharedStringAccessGuardIfNeeded&);
 // clang-format on
 
 class SubStringRange {
@@ -650,7 +659,7 @@ class SeqOneByteString
   // defined for convenience and it will check that the access guard is not
   // needed.
   inline uint8_t Get(int index) const;
-  inline uint8_t Get(int index,
+  inline uint8_t Get(int index, PtrComprCageBase cage_base,
                      const SharedStringAccessGuardIfNeeded& access_guard) const;
   inline void SeqOneByteStringSet(int index, uint16_t value);
 
@@ -698,7 +707,8 @@ class SeqTwoByteString
 
   // Dispatched behavior.
   inline uint16_t Get(
-      int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+      int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
   inline void SeqTwoByteStringSet(int index, uint16_t value);
 
   // Get the address of the characters in this string.
@@ -756,7 +766,8 @@ class ConsString : public TorqueGeneratedConsString<ConsString, String> {
 
   // Dispatched behavior.
   V8_EXPORT_PRIVATE uint16_t
-  Get(int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+  Get(int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   // Minimum length for a cons string.
   static const int kMinLength = 13;
@@ -780,7 +791,8 @@ class ThinString : public TorqueGeneratedThinString<ThinString, String> {
   DECL_GETTER(unchecked_actual, HeapObject)
 
   V8_EXPORT_PRIVATE uint16_t
-  Get(int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+  Get(int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   DECL_VERIFIER(ThinString)
 
@@ -805,7 +817,8 @@ class SlicedString : public TorqueGeneratedSlicedString<SlicedString, String> {
                          WriteBarrierMode mode = UPDATE_WRITE_BARRIER);
   // Dispatched behavior.
   V8_EXPORT_PRIVATE uint16_t
-  Get(int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+  Get(int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   // Minimum length for a sliced string.
   static const int kMinLength = 13;
@@ -891,7 +904,7 @@ class ExternalOneByteString
   inline const uint8_t* GetChars() const;
 
   // Dispatched behavior.
-  inline uint8_t Get(int index,
+  inline uint8_t Get(int index, PtrComprCageBase cage_base,
                      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   class BodyDescriptor;
@@ -935,7 +948,8 @@ class ExternalTwoByteString
 
   // Dispatched behavior.
   inline uint16_t Get(
-      int index, const SharedStringAccessGuardIfNeeded& access_guard) const;
+      int index, PtrComprCageBase cage_base,
+      const SharedStringAccessGuardIfNeeded& access_guard) const;
 
   // For regexp code.
   inline const uint16_t* ExternalTwoByteStringGetData(unsigned start);