implement String.intern()

classpath/java/lang/String.java

@@ -33,10 +33,6 @@ public final class String implements Comparable<String> {
     }
   }
 
-  public static String valueOf(int v) {
-    return valueOf((long) v);
-  }
-
   public int length() {
     return length;
   }
@@ -217,7 +213,13 @@ public final class String implements Comparable<String> {
       return ((char[]) data)[index + offset];
     } else {
       return (char) ((byte[]) data)[index + offset];
-    }    
+    }
+  }
+
+  public native String intern();
+
+  public static String valueOf(int v) {
+    return valueOf((long) v);
   }
 
   public static String valueOf(long v) {

makefile

@@ -16,7 +16,7 @@ src = src
 classpath = classpath
 test = test
 
-input = $(cls)/Threads.class
+input = $(cls)/Exceptions.class
 
 cxx = g++
 cc = gcc

src/builtin.cpp

@@ -445,6 +445,12 @@ Array_makeObjectArray(Thread* t, jclass, jclass elementType, jint length)
   return pushReference(t, makeObjectArray(t, *elementType, length, true));
 }
 
+jobject
+String_intern(Thread* t, jobject this_)
+{
+  return pushReference(t, intern(t, *this_));
+}
+
 void
 System_arraycopy(Thread* t, jclass, jobject src, jint srcOffset, jobject dst,
                  jint dstOffset, jint length)
@@ -649,6 +655,9 @@ populateBuiltinMap(Thread* t, object map)
     { "Java_java_lang_Runtiime_exit",
       reinterpret_cast<void*>(::Runtime_exit) },
 
+    { "Java_java_lang_String_intern",
+      reinterpret_cast<void*>(::String_intern) },
+
     { "Java_java_lang_Thread_doStart",
       reinterpret_cast<void*>(::Thread_doStart) },
     { "Java_java_lang_Thread_interrupt",

src/machine.cpp

@@ -573,6 +573,7 @@ parsePool(Thread* t, Stream& s)
         object bytes = arrayBody(t, pool, intArrayBody(t, o, 1) - 1);
         object value = makeString
           (t, bytes, 0, byteArrayLength(t, bytes) - 1, 0);
+        value = intern(t, value);
         set(t, arrayBody(t, pool, i), value);
       } break;
 
@@ -1253,6 +1254,12 @@ removeMonitor(Thread* t, object o)
   static_cast<System::Monitor*>(pointerValue(t, p))->dispose();
 }
 
+void
+removeString(Thread* t, object o)
+{
+  hashMapRemove(t, t->vm->stringMap, o, stringHash, objectEqual);
+}
+
 } // namespace
 
 namespace vm {
@@ -1274,6 +1281,7 @@ Machine::Machine(System* system, Heap* heap, ClassFinder* classFinder):
   bootstrapClassMap(0),
   builtinMap(0),
   monitorMap(0),
+  stringMap(0),
   types(0),
   finalizers(0),
   tenuredFinalizers(0),
@@ -1373,6 +1381,7 @@ Thread::Thread(Machine* m, object javaThread, Thread* parent):
     m->classMap = makeHashMap(this, NormalMap, 0, 0);
     m->builtinMap = makeHashMap(this, NormalMap, 0, 0);
     m->monitorMap = makeHashMap(this, WeakMap, 0, 0);
+    m->stringMap = makeHashMap(this, WeakMap, 0, 0);
 
     populateBuiltinMap(t, m->builtinMap);
 
@@ -2234,6 +2243,23 @@ objectMonitor(Thread* t, object o)
   }
 }
 
+object
+intern(Thread* t, object s)
+{
+  ACQUIRE(t, t->vm->referenceLock);
+
+  object n = hashMapFindNode(t, t->vm->stringMap, s, stringHash, stringEqual);
+  if (n) {
+    return jreferenceTarget(t, tripleFirst(t, n));
+  } else {
+    PROTECT(t, s);
+
+    hashMapInsert(t, t->vm->stringMap, s, 0, stringHash);
+    addFinalizer(t, s, removeString);
+    return s;
+  }
+}
+
 void
 collect(Thread* t, Heap::CollectionType type)
 {
@@ -2248,6 +2274,7 @@ collect(Thread* t, Heap::CollectionType type)
       v->visit(&(m->bootstrapClassMap));
       v->visit(&(m->builtinMap));
       v->visit(&(m->monitorMap));
+      v->visit(&(m->stringMap));
       v->visit(&(m->types));
 
       for (Thread* t = m->rootThread; t; t = t->peer) {

src/machine.h

@@ -1114,6 +1114,7 @@ class Machine {
   object bootstrapClassMap;
   object builtinMap;
   object monitorMap;
+  object stringMap;
   object types;
   object finalizers;
   object tenuredFinalizers;
@@ -1777,6 +1778,14 @@ hash(const int8_t* s, unsigned length)
   return h;  
 }
 
+inline uint32_t
+hash(const uint16_t* s, unsigned length)
+{
+  uint32_t h = 0;
+  for (unsigned i = 0; i < length; ++i) h = (h * 31) + s[i];
+  return h;  
+}
+
 inline unsigned
 baseSize(Thread* t, object o, object class_)
 {
@@ -1847,6 +1856,12 @@ byteArrayHash(Thread* t, object array)
   return hash(&byteArrayBody(t, array, 0), byteArrayLength(t, array));
 }
 
+inline uint32_t
+charArrayHash(Thread* t, object array)
+{
+  return hash(&charArrayBody(t, array, 0), charArrayLength(t, array));
+}
+
 inline bool
 byteArrayEqual(Thread* t, object a, object b)
 {
@@ -1856,6 +1871,52 @@ byteArrayEqual(Thread* t, object a, object b)
             byteArrayLength(t, a)) == 0);
 }
 
+inline uint32_t
+stringHash(Thread* t, object s)
+{
+  if (stringHashCode(t, s) == 0) {
+    object data = stringData(t, s);
+    if (objectClass(t, data)
+        == arrayBody(t, t->vm->types, Machine::ByteArrayType))
+    {
+      stringHashCode(t, s) = byteArrayHash(t, data);
+    } else {
+      stringHashCode(t, s) = charArrayHash(t, data);
+    }
+  }
+  return stringHashCode(t, s);
+}
+
+inline uint16_t
+stringCharAt(Thread* t, object s, int i)
+{
+  object data = stringData(t, s);
+  if (objectClass(t, data)
+      == arrayBody(t, t->vm->types, Machine::ByteArrayType))
+  {
+    return byteArrayBody(t, data, i);
+  } else {
+    return charArrayBody(t, data, i);
+  }
+}
+
+inline bool
+stringEqual(Thread* t, object a, object b)
+{
+  if (a == b) {
+    return true;
+  } else if (stringLength(t, a) == stringLength(t, b)) {
+    for (int i = 0; i < stringLength(t, a); ++i) {
+      if (stringCharAt(t, a, i) != stringCharAt(t, b, i)) {
+        return false;
+      }
+    }
+    return true;
+  } else {
+    return false;
+  }
+}
+
 inline bool
 intArrayEqual(Thread* t, object a, object b)
 {
@@ -2122,6 +2183,9 @@ interrupt(Thread*, Thread* target)
   target->systemThread->interrupt();
 }
 
+object
+intern(Thread* t, object s);
+
 void
 exit(Thread* t);
 

src/types.def

@@ -121,7 +121,7 @@
   (object data)
   (int32_t offset)
   (int32_t length)
-  (int32_t hash))
+  (int32_t hashCode))
 
 (type thread java/lang/Thread
   (extends jobject)