Track ascii-ness of data in externalized strings.

src/objects.h

Index: src/objects.h
diff --git a/src/objects.h b/src/objects.h
index 095dd981fb8fa266be10e85555f12ba0e75694d0..e81b7f12f97032593f0ad4bcaf3035ebaf98b1c2 100644
--- a/src/objects.h
+++ b/src/objects.h
@@ -320,6 +320,10 @@ enum PropertyNormalizationMode {
     ExternalTwoByteString::kSize,                                              \
     external_symbol,                                                           \
     ExternalSymbol)                                                            \
+  V(EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE,                                      \
+    ExternalTwoByteString::kSize,                                              \
+    external_symbol_with_ascii_data,                                           \
+    ExternalSymbolWithAsciiData)                                               \
   V(EXTERNAL_ASCII_SYMBOL_TYPE,                                                \
     ExternalAsciiString::kSize,                                                \
     external_ascii_symbol,                                                     \
@@ -344,6 +348,10 @@ enum PropertyNormalizationMode {
     ExternalTwoByteString::kSize,                                              \
     external_string,                                                           \
     ExternalString)                                                            \
+  V(EXTERNAL_STRING_WITH_ASCII_DATA_TYPE,                                      \
+    ExternalTwoByteString::kSize,                                              \
+    external_string_with_ascii_data,                                           \
+    ExternalStringWithAsciiData)                                               \
   V(EXTERNAL_ASCII_STRING_TYPE,                                                \
     ExternalAsciiString::kSize,                                                \
     external_ascii_string,                                                     \
@@ -412,6 +420,11 @@ enum StringRepresentationTag {
 };
 const uint32_t kIsConsStringMask = 0x1;
 
+// If bit 7 is clear, then bit 3 indicates whether this two-byte
+// string actually contains ascii data.
+const uint32_t kAsciiDataHintMask = 0x08;
+const uint32_t kAsciiDataHintTag = 0x08;
+
 
 // A ConsString with an empty string as the right side is a candidate
 // for being shortcut by the garbage collector unless it is a
@@ -427,18 +440,22 @@ const uint32_t kShortcutTypeTag = kConsStringTag;
 
 enum InstanceType {
   // String types.
-  SYMBOL_TYPE = kSymbolTag | kSeqStringTag,
+  SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag | kSeqStringTag,
   ASCII_SYMBOL_TYPE = kAsciiStringTag | kSymbolTag | kSeqStringTag,
-  CONS_SYMBOL_TYPE = kSymbolTag | kConsStringTag,
+  CONS_SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag | kConsStringTag,
   CONS_ASCII_SYMBOL_TYPE = kAsciiStringTag | kSymbolTag | kConsStringTag,
-  EXTERNAL_SYMBOL_TYPE = kSymbolTag | kExternalStringTag,
+  EXTERNAL_SYMBOL_TYPE = kTwoByteStringTag | kSymbolTag | kExternalStringTag,
+  EXTERNAL_SYMBOL_WITH_ASCII_DATA_TYPE =
+      kTwoByteStringTag | kSymbolTag | kExternalStringTag | kAsciiDataHintTag,
   EXTERNAL_ASCII_SYMBOL_TYPE =
       kAsciiStringTag | kSymbolTag | kExternalStringTag,
-  STRING_TYPE = kSeqStringTag,
+  STRING_TYPE = kTwoByteStringTag | kSeqStringTag,
   ASCII_STRING_TYPE = kAsciiStringTag | kSeqStringTag,
-  CONS_STRING_TYPE = kConsStringTag,
+  CONS_STRING_TYPE = kTwoByteStringTag | kConsStringTag,
   CONS_ASCII_STRING_TYPE = kAsciiStringTag | kConsStringTag,
-  EXTERNAL_STRING_TYPE = kExternalStringTag,
+  EXTERNAL_STRING_TYPE = kTwoByteStringTag | kExternalStringTag,
+  EXTERNAL_STRING_WITH_ASCII_DATA_TYPE =
+      kTwoByteStringTag | kExternalStringTag | kAsciiDataHintTag,
   EXTERNAL_ASCII_STRING_TYPE = kAsciiStringTag | kExternalStringTag,
   PRIVATE_EXTERNAL_ASCII_STRING_TYPE = EXTERNAL_ASCII_STRING_TYPE,
 
@@ -4064,12 +4081,14 @@ class String: public HeapObject {
   inline bool IsAsciiRepresentation();
   inline bool IsTwoByteRepresentation();
 
-  // Check whether this string is an external two-byte string that in
-  // fact contains only ascii characters.
+  // Returns whether this string has ascii chars, i.e. all of them can
+  // be ascii encoded.  This might be the case even if the string is
+  // two-byte.  Such strings may appear when the embedder prefers
+  // two-byte external representations even for ascii data.
   //
-  // Such strings may appear when the embedder prefers two-byte
-  // representations even for ascii data.
-  inline bool IsExternalTwoByteStringWithAsciiChars();
+  // NOTE: this should be considered only a hint.  False negatives are
+  // possible.
+  inline bool HasAsciiChars();
 
   // Get and set individual two byte chars in the string.
   inline void Set(int index, uint16_t value);