changeset 3981:1e129851479e hs25-b15

Merge
author amurillo
date Fri, 11 Jan 2013 01:43:09 -0800
parents 11619f33cd68 f1c06dcee0b5
children b5e6bec76f4a
files
diffstat 127 files changed, 3221 insertions(+), 1982 deletions(-) [+]
line wrap: on
line diff
--- a/make/bsd/makefiles/mapfile-vers-debug	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/bsd/makefiles/mapfile-vers-debug	Fri Jan 11 01:43:09 2013 -0800
@@ -1,5 +1,5 @@
 #
-# @(#)mapfile-vers-debug	1.18 07/10/25 16:47:35
+# @(#)mapfile-vers-debug        1.18 07/10/25 16:47:35
 #
 
 #
@@ -126,8 +126,9 @@
                 JVM_GetClassModifiers;
                 JVM_GetClassName;
                 JVM_GetClassNameUTF;
-		JVM_GetClassSignature;
+                        JVM_GetClassSignature;
                 JVM_GetClassSigners;
+                JVM_GetClassTypeAnnotations;
                 JVM_GetComponentType;
                 JVM_GetDeclaredClasses;
                 JVM_GetDeclaringClass;
@@ -154,6 +155,7 @@
                 JVM_GetMethodIxNameUTF;
                 JVM_GetMethodIxSignatureUTF;
                 JVM_GetMethodParameterAnnotations;
+                JVM_GetMethodParameters;
                 JVM_GetPrimitiveArrayElement;
                 JVM_GetProtectionDomain;
                 JVM_GetSockName;
@@ -283,7 +285,7 @@
                 # This is for Forte Analyzer profiling support.
                 AsyncGetCallTrace;
 
-		# INSERT VTABLE SYMBOLS HERE
+                # INSERT VTABLE SYMBOLS HERE
 
         local:
                 *;
--- a/make/bsd/makefiles/mapfile-vers-product	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/bsd/makefiles/mapfile-vers-product	Fri Jan 11 01:43:09 2013 -0800
@@ -128,6 +128,7 @@
                 JVM_GetClassNameUTF;
                 JVM_GetClassSignature;
                 JVM_GetClassSigners;
+                JVM_GetClassTypeAnnotations;
                 JVM_GetComponentType;
                 JVM_GetDeclaredClasses;
                 JVM_GetDeclaringClass;
@@ -154,6 +155,7 @@
                 JVM_GetMethodIxNameUTF;
                 JVM_GetMethodIxSignatureUTF;
                 JVM_GetMethodParameterAnnotations;
+                JVM_GetMethodParameters;
                 JVM_GetPrimitiveArrayElement;
                 JVM_GetProtectionDomain;
                 JVM_GetSockName;
--- a/make/hotspot_version	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/hotspot_version	Fri Jan 11 01:43:09 2013 -0800
@@ -35,7 +35,7 @@
 
 HS_MAJOR_VER=25
 HS_MINOR_VER=0
-HS_BUILD_NUMBER=14
+HS_BUILD_NUMBER=15
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=8
--- a/make/linux/makefiles/mapfile-vers-debug	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/linux/makefiles/mapfile-vers-debug	Fri Jan 11 01:43:09 2013 -0800
@@ -124,6 +124,7 @@
                 JVM_GetClassNameUTF;
 		JVM_GetClassSignature;
                 JVM_GetClassSigners;
+                JVM_GetClassTypeAnnotations;
                 JVM_GetComponentType;
                 JVM_GetDeclaredClasses;
                 JVM_GetDeclaringClass;
@@ -150,6 +151,7 @@
                 JVM_GetMethodIxNameUTF;
                 JVM_GetMethodIxSignatureUTF;
                 JVM_GetMethodParameterAnnotations;
+                JVM_GetMethodParameters;
                 JVM_GetPrimitiveArrayElement;
                 JVM_GetProtectionDomain;
                 JVM_GetSockName;
--- a/make/linux/makefiles/mapfile-vers-product	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/linux/makefiles/mapfile-vers-product	Fri Jan 11 01:43:09 2013 -0800
@@ -124,6 +124,7 @@
                 JVM_GetClassNameUTF;
                 JVM_GetClassSignature;
                 JVM_GetClassSigners;
+                JVM_GetClassTypeAnnotations;
                 JVM_GetComponentType;
                 JVM_GetDeclaredClasses;
                 JVM_GetDeclaringClass;
@@ -150,6 +151,7 @@
                 JVM_GetMethodIxNameUTF;
                 JVM_GetMethodIxSignatureUTF;
                 JVM_GetMethodParameterAnnotations;
+                JVM_GetMethodParameters;
                 JVM_GetPrimitiveArrayElement;
                 JVM_GetProtectionDomain;
                 JVM_GetSockName;
--- a/make/solaris/makefiles/mapfile-vers	Thu Jan 10 09:55:30 2013 -0800
+++ b/make/solaris/makefiles/mapfile-vers	Fri Jan 11 01:43:09 2013 -0800
@@ -26,235 +26,237 @@
 
 SUNWprivate_1.1 {
         global:
-		# JNI
+                # JNI
                 JNI_CreateJavaVM;
                 JNI_GetCreatedJavaVMs;
                 JNI_GetDefaultJavaVMInitArgs;
-	
-		# JVM
-		JVM_Accept;
-		JVM_ActiveProcessorCount;
-		JVM_AllocateNewArray;
-		JVM_AllocateNewObject;
-		JVM_ArrayCopy;
-		JVM_AssertionStatusDirectives;
-		JVM_Available;
-		JVM_Bind;
-		JVM_ClassDepth;
-		JVM_ClassLoaderDepth;
-		JVM_Clone;
-		JVM_Close;
-		JVM_CX8Field;
-		JVM_CompileClass;
-		JVM_CompileClasses;
-		JVM_CompilerCommand;
-		JVM_Connect;
-		JVM_ConstantPoolGetClassAt;
-		JVM_ConstantPoolGetClassAtIfLoaded;
-		JVM_ConstantPoolGetDoubleAt;
-		JVM_ConstantPoolGetFieldAt;
-		JVM_ConstantPoolGetFieldAtIfLoaded;
-		JVM_ConstantPoolGetFloatAt;
-		JVM_ConstantPoolGetIntAt;
-		JVM_ConstantPoolGetLongAt;
-		JVM_ConstantPoolGetMethodAt;
-		JVM_ConstantPoolGetMethodAtIfLoaded;
-		JVM_ConstantPoolGetMemberRefInfoAt;
-		JVM_ConstantPoolGetSize;
-		JVM_ConstantPoolGetStringAt;
-		JVM_ConstantPoolGetUTF8At;
-		JVM_CountStackFrames;
-		JVM_CurrentClassLoader;
-		JVM_CurrentLoadedClass;
-		JVM_CurrentThread;
-		JVM_CurrentTimeMillis;
-		JVM_DefineClass;
-		JVM_DefineClassWithSource;
-		JVM_DefineClassWithSourceCond;
-		JVM_DesiredAssertionStatus;
-		JVM_DisableCompiler;
-		JVM_DoPrivileged;
-		JVM_DTraceGetVersion;
-		JVM_DTraceActivate;
-		JVM_DTraceIsProbeEnabled;
-		JVM_DTraceIsSupported;
-		JVM_DTraceDispose;
-		JVM_DumpAllStacks;
-		JVM_DumpThreads;
-		JVM_EnableCompiler;
-		JVM_Exit;
-		JVM_FillInStackTrace;
-		JVM_FindClassFromClass;
-		JVM_FindClassFromClassLoader;
-		JVM_FindClassFromBootLoader;
-		JVM_FindLibraryEntry;
-		JVM_FindLoadedClass;
-		JVM_FindPrimitiveClass;
-		JVM_FindSignal;
-		JVM_FreeMemory;
-		JVM_GC;
-		JVM_GetAllThreads;
-		JVM_GetArrayElement;
-		JVM_GetArrayLength;
-		JVM_GetCPClassNameUTF;
-		JVM_GetCPFieldClassNameUTF;
-		JVM_GetCPFieldModifiers;
-		JVM_GetCPFieldNameUTF;
-		JVM_GetCPFieldSignatureUTF;
-		JVM_GetCPMethodClassNameUTF;
-		JVM_GetCPMethodModifiers;
-		JVM_GetCPMethodNameUTF;
-		JVM_GetCPMethodSignatureUTF;
-		JVM_GetCallerClass;
-		JVM_GetClassAccessFlags;
-		JVM_GetClassAnnotations;
-		JVM_GetClassCPEntriesCount;
-		JVM_GetClassCPTypes;
-		JVM_GetClassConstantPool;
-		JVM_GetClassContext;
-		JVM_GetClassDeclaredConstructors;
-		JVM_GetClassDeclaredFields;
-		JVM_GetClassDeclaredMethods;
-		JVM_GetClassFieldsCount;
-		JVM_GetClassInterfaces;
-		JVM_GetClassLoader;
-		JVM_GetClassMethodsCount;
-		JVM_GetClassModifiers;
-		JVM_GetClassName;
-		JVM_GetClassNameUTF;
-		JVM_GetClassSignature;
-		JVM_GetClassSigners;
-		JVM_GetComponentType;
-		JVM_GetDeclaredClasses;
-		JVM_GetDeclaringClass;
-		JVM_GetEnclosingMethodInfo;
-	        JVM_GetFieldAnnotations;
-		JVM_GetFieldIxModifiers;
-		JVM_GetHostName;
-		JVM_GetInheritedAccessControlContext;
-		JVM_GetInterfaceVersion;
-		JVM_GetLastErrorString;
-		JVM_GetManagement;
-	        JVM_GetMethodAnnotations;
-	        JVM_GetMethodDefaultAnnotationValue;
-		JVM_GetMethodIxArgsSize;
-		JVM_GetMethodIxByteCode;
-		JVM_GetMethodIxByteCodeLength;
-		JVM_GetMethodIxExceptionIndexes;
-		JVM_GetMethodIxExceptionTableEntry;
-		JVM_GetMethodIxExceptionTableLength;
-		JVM_GetMethodIxExceptionsCount;
-		JVM_GetMethodIxLocalsCount;
-		JVM_GetMethodIxMaxStack;
-		JVM_GetMethodIxModifiers;
-		JVM_GetMethodIxNameUTF;
-		JVM_GetMethodIxSignatureUTF;
-	        JVM_GetMethodParameterAnnotations;
-		JVM_GetPrimitiveArrayElement;
-		JVM_GetProtectionDomain;
-		JVM_GetSockName;
-		JVM_GetSockOpt;
-		JVM_GetStackAccessControlContext;
-		JVM_GetStackTraceDepth;
-		JVM_GetStackTraceElement;
-		JVM_GetSystemPackage;
-		JVM_GetSystemPackages;
-		JVM_GetThreadStateNames;
-		JVM_GetThreadStateValues;
-		JVM_GetVersionInfo;
-		JVM_Halt;
-		JVM_HoldsLock;
-		JVM_IHashCode;
-		JVM_InitAgentProperties;
-		JVM_InitProperties;
-		JVM_InitializeCompiler;
-		JVM_InitializeSocketLibrary;
-		JVM_InternString;
-		JVM_Interrupt;
-		JVM_InvokeMethod;
-		JVM_IsArrayClass;
-		JVM_IsConstructorIx;
-		JVM_IsInterface;
-		JVM_IsInterrupted;
-		JVM_IsNaN;
-		JVM_IsPrimitiveClass;
-		JVM_IsSameClassPackage;
-		JVM_IsSilentCompiler;
-		JVM_IsSupportedJNIVersion;
-		JVM_IsThreadAlive;
-		JVM_LatestUserDefinedLoader;
-		JVM_Listen;
-		JVM_LoadClass0;
-		JVM_LoadLibrary;
-		JVM_Lseek;
-		JVM_MaxObjectInspectionAge;
-		JVM_MaxMemory;
-		JVM_MonitorNotify;
-		JVM_MonitorNotifyAll;
-		JVM_MonitorWait;
-		JVM_NativePath;
-		JVM_NanoTime;
-		JVM_NewArray;
-		JVM_NewInstanceFromConstructor;
-		JVM_NewMultiArray;
-		JVM_OnExit;
-		JVM_Open;
-		JVM_PrintStackTrace;
-		JVM_RaiseSignal;
-		JVM_RawMonitorCreate;
-		JVM_RawMonitorDestroy;
-		JVM_RawMonitorEnter;
-		JVM_RawMonitorExit;
-		JVM_Read;
-		JVM_Recv;
-		JVM_RecvFrom;
-		JVM_RegisterSignal;
-		JVM_ReleaseUTF;
-		JVM_ResolveClass;
-		JVM_ResumeThread;
-		JVM_Send;
-		JVM_SendTo;
-		JVM_SetArrayElement;
-		JVM_SetClassSigners;
-		JVM_SetLength;
+        
+                # JVM
+                JVM_Accept;
+                JVM_ActiveProcessorCount;
+                JVM_AllocateNewArray;
+                JVM_AllocateNewObject;
+                JVM_ArrayCopy;
+                JVM_AssertionStatusDirectives;
+                JVM_Available;
+                JVM_Bind;
+                JVM_ClassDepth;
+                JVM_ClassLoaderDepth;
+                JVM_Clone;
+                JVM_Close;
+                JVM_CX8Field;
+                JVM_CompileClass;
+                JVM_CompileClasses;
+                JVM_CompilerCommand;
+                JVM_Connect;
+                JVM_ConstantPoolGetClassAt;
+                JVM_ConstantPoolGetClassAtIfLoaded;
+                JVM_ConstantPoolGetDoubleAt;
+                JVM_ConstantPoolGetFieldAt;
+                JVM_ConstantPoolGetFieldAtIfLoaded;
+                JVM_ConstantPoolGetFloatAt;
+                JVM_ConstantPoolGetIntAt;
+                JVM_ConstantPoolGetLongAt;
+                JVM_ConstantPoolGetMethodAt;
+                JVM_ConstantPoolGetMethodAtIfLoaded;
+                JVM_ConstantPoolGetMemberRefInfoAt;
+                JVM_ConstantPoolGetSize;
+                JVM_ConstantPoolGetStringAt;
+                JVM_ConstantPoolGetUTF8At;
+                JVM_CountStackFrames;
+                JVM_CurrentClassLoader;
+                JVM_CurrentLoadedClass;
+                JVM_CurrentThread;
+                JVM_CurrentTimeMillis;
+                JVM_DefineClass;
+                JVM_DefineClassWithSource;
+                JVM_DefineClassWithSourceCond;
+                JVM_DesiredAssertionStatus;
+                JVM_DisableCompiler;
+                JVM_DoPrivileged;
+                JVM_DTraceGetVersion;
+                JVM_DTraceActivate;
+                JVM_DTraceIsProbeEnabled;
+                JVM_DTraceIsSupported;
+                JVM_DTraceDispose;
+                JVM_DumpAllStacks;
+                JVM_DumpThreads;
+                JVM_EnableCompiler;
+                JVM_Exit;
+                JVM_FillInStackTrace;
+                JVM_FindClassFromClass;
+                JVM_FindClassFromClassLoader;
+                JVM_FindClassFromBootLoader;
+                JVM_FindLibraryEntry;
+                JVM_FindLoadedClass;
+                JVM_FindPrimitiveClass;
+                JVM_FindSignal;
+                JVM_FreeMemory;
+                JVM_GC;
+                JVM_GetAllThreads;
+                JVM_GetArrayElement;
+                JVM_GetArrayLength;
+                JVM_GetCPClassNameUTF;
+                JVM_GetCPFieldClassNameUTF;
+                JVM_GetCPFieldModifiers;
+                JVM_GetCPFieldNameUTF;
+                JVM_GetCPFieldSignatureUTF;
+                JVM_GetCPMethodClassNameUTF;
+                JVM_GetCPMethodModifiers;
+                JVM_GetCPMethodNameUTF;
+                JVM_GetCPMethodSignatureUTF;
+                JVM_GetCallerClass;
+                JVM_GetClassAccessFlags;
+                JVM_GetClassAnnotations;
+                JVM_GetClassCPEntriesCount;
+                JVM_GetClassCPTypes;
+                JVM_GetClassConstantPool;
+                JVM_GetClassContext;
+                JVM_GetClassDeclaredConstructors;
+                JVM_GetClassDeclaredFields;
+                JVM_GetClassDeclaredMethods;
+                JVM_GetClassFieldsCount;
+                JVM_GetClassInterfaces;
+                JVM_GetClassLoader;
+                JVM_GetClassMethodsCount;
+                JVM_GetClassModifiers;
+                JVM_GetClassName;
+                JVM_GetClassNameUTF;
+                JVM_GetClassSignature;
+                JVM_GetClassSigners;
+                JVM_GetComponentType;
+                JVM_GetClassTypeAnnotations;
+                JVM_GetDeclaredClasses;
+                JVM_GetDeclaringClass;
+                JVM_GetEnclosingMethodInfo;
+                JVM_GetFieldAnnotations;
+                JVM_GetFieldIxModifiers;
+                JVM_GetHostName;
+                JVM_GetInheritedAccessControlContext;
+                JVM_GetInterfaceVersion;
+                JVM_GetLastErrorString;
+                JVM_GetManagement;
+                JVM_GetMethodAnnotations;
+                JVM_GetMethodDefaultAnnotationValue;
+                JVM_GetMethodIxArgsSize;
+                JVM_GetMethodIxByteCode;
+                JVM_GetMethodIxByteCodeLength;
+                JVM_GetMethodIxExceptionIndexes;
+                JVM_GetMethodIxExceptionTableEntry;
+                JVM_GetMethodIxExceptionTableLength;
+                JVM_GetMethodIxExceptionsCount;
+                JVM_GetMethodIxLocalsCount;
+                JVM_GetMethodIxMaxStack;
+                JVM_GetMethodIxModifiers;
+                JVM_GetMethodIxNameUTF;
+                JVM_GetMethodIxSignatureUTF;
+                JVM_GetMethodParameterAnnotations;
+                JVM_GetMethodParameters;
+                JVM_GetPrimitiveArrayElement;
+                JVM_GetProtectionDomain;
+                JVM_GetSockName;
+                JVM_GetSockOpt;
+                JVM_GetStackAccessControlContext;
+                JVM_GetStackTraceDepth;
+                JVM_GetStackTraceElement;
+                JVM_GetSystemPackage;
+                JVM_GetSystemPackages;
+                JVM_GetThreadStateNames;
+                JVM_GetThreadStateValues;
+                JVM_GetVersionInfo;
+                JVM_Halt;
+                JVM_HoldsLock;
+                JVM_IHashCode;
+                JVM_InitAgentProperties;
+                JVM_InitProperties;
+                JVM_InitializeCompiler;
+                JVM_InitializeSocketLibrary;
+                JVM_InternString;
+                JVM_Interrupt;
+                JVM_InvokeMethod;
+                JVM_IsArrayClass;
+                JVM_IsConstructorIx;
+                JVM_IsInterface;
+                JVM_IsInterrupted;
+                JVM_IsNaN;
+                JVM_IsPrimitiveClass;
+                JVM_IsSameClassPackage;
+                JVM_IsSilentCompiler;
+                JVM_IsSupportedJNIVersion;
+                JVM_IsThreadAlive;
+                JVM_LatestUserDefinedLoader;
+                JVM_Listen;
+                JVM_LoadClass0;
+                JVM_LoadLibrary;
+                JVM_Lseek;
+                JVM_MaxObjectInspectionAge;
+                JVM_MaxMemory;
+                JVM_MonitorNotify;
+                JVM_MonitorNotifyAll;
+                JVM_MonitorWait;
+                JVM_NativePath;
+                JVM_NanoTime;
+                JVM_NewArray;
+                JVM_NewInstanceFromConstructor;
+                JVM_NewMultiArray;
+                JVM_OnExit;
+                JVM_Open;
+                JVM_PrintStackTrace;
+                JVM_RaiseSignal;
+                JVM_RawMonitorCreate;
+                JVM_RawMonitorDestroy;
+                JVM_RawMonitorEnter;
+                JVM_RawMonitorExit;
+                JVM_Read;
+                JVM_Recv;
+                JVM_RecvFrom;
+                JVM_RegisterSignal;
+                JVM_ReleaseUTF;
+                JVM_ResolveClass;
+                JVM_ResumeThread;
+                JVM_Send;
+                JVM_SendTo;
+                JVM_SetArrayElement;
+                JVM_SetClassSigners;
+                JVM_SetLength;
                 JVM_SetNativeThreadName;
-		JVM_SetPrimitiveArrayElement;
-		JVM_SetProtectionDomain;
-		JVM_SetSockOpt;
-		JVM_SetThreadPriority;
-		JVM_Sleep;
-		JVM_Socket;
-		JVM_SocketAvailable;
-		JVM_SocketClose;
-		JVM_SocketShutdown;
-		JVM_StartThread;
-		JVM_StopThread;
-		JVM_SuspendThread;
-		JVM_SupportsCX8;
-		JVM_Sync;
-		JVM_Timeout;
-		JVM_TotalMemory;
-		JVM_TraceInstructions;
-		JVM_TraceMethodCalls;
-		JVM_UnloadLibrary;
-		JVM_Write;
-		JVM_Yield;
-		JVM_handle_solaris_signal;
+                JVM_SetPrimitiveArrayElement;
+                JVM_SetProtectionDomain;
+                JVM_SetSockOpt;
+                JVM_SetThreadPriority;
+                JVM_Sleep;
+                JVM_Socket;
+                JVM_SocketAvailable;
+                JVM_SocketClose;
+                JVM_SocketShutdown;
+                JVM_StartThread;
+                JVM_StopThread;
+                JVM_SuspendThread;
+                JVM_SupportsCX8;
+                JVM_Sync;
+                JVM_Timeout;
+                JVM_TotalMemory;
+                JVM_TraceInstructions;
+                JVM_TraceMethodCalls;
+                JVM_UnloadLibrary;
+                JVM_Write;
+                JVM_Yield;
+                JVM_handle_solaris_signal;
 
-		# miscellaneous functions
-		jio_fprintf;
-		jio_printf;
-		jio_snprintf;
-		jio_vfprintf;
-		jio_vsnprintf;
+                # miscellaneous functions
+                jio_fprintf;
+                jio_printf;
+                jio_snprintf;
+                jio_vfprintf;
+                jio_vsnprintf;
 
-		# Needed because there is no JVM interface for this.
-		sysThreadAvailableStackWithSlack;
+                # Needed because there is no JVM interface for this.
+                sysThreadAvailableStackWithSlack;
 
-		# This is for Forte Analyzer profiling support.
-		AsyncGetCallTrace;
+                # This is for Forte Analyzer profiling support.
+                AsyncGetCallTrace;
 
-		# INSERT VTABLE SYMBOLS HERE
+                # INSERT VTABLE SYMBOLS HERE
 
         local:
                 *;
--- a/src/cpu/sparc/vm/assembler_sparc.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/sparc/vm/assembler_sparc.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -675,8 +675,8 @@
     AbstractAssembler::flush();
   }
 
-  inline void emit_long(int);  // shadows AbstractAssembler::emit_long
-  inline void emit_data(int x) { emit_long(x); }
+  inline void emit_int32(int);  // shadows AbstractAssembler::emit_int32
+  inline void emit_data(int x) { emit_int32(x); }
   inline void emit_data(int, RelocationHolder const&);
   inline void emit_data(int, relocInfo::relocType rtype);
   // helper for above fcns
@@ -691,12 +691,12 @@
   inline void add(Register s1, Register s2, Register d );
   inline void add(Register s1, int simm13a, Register d );
 
-  void addcc(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void addcc(  Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void addc(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | rs2(s2) ); }
-  void addc(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void addccc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void addccc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void addcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void addcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void addc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | rs2(s2) ); }
+  void addc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void addccc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void addccc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
 
   // pp 136
@@ -749,76 +749,76 @@
   // at address s1 is swapped with the data in d. If the values are not equal,
   // the the contents of memory at s1 is loaded into d, without the swap.
 
-  void casa(  Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(casa_op3 ) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
-  void casxa( Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(casxa_op3) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
+  void casa(  Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(casa_op3 ) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
+  void casxa( Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(casxa_op3) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
 
   // pp 152
 
-  void udiv(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | rs2(s2)); }
-  void udiv(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void sdiv(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | rs2(s2)); }
-  void sdiv(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void udivcc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
-  void udivcc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void sdivcc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
-  void sdivcc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void udiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | rs2(s2)); }
+  void udiv(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void sdiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | rs2(s2)); }
+  void sdiv(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void udivcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
+  void udivcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void sdivcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
+  void sdivcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 155
 
-  void done()  { v9_only();  cti();  emit_long( op(arith_op) | fcn(0) | op3(done_op3) ); }
-  void retry() { v9_only();  cti();  emit_long( op(arith_op) | fcn(1) | op3(retry_op3) ); }
+  void done()  { v9_only();  cti();  emit_int32( op(arith_op) | fcn(0) | op3(done_op3) ); }
+  void retry() { v9_only();  cti();  emit_int32( op(arith_op) | fcn(1) | op3(retry_op3) ); }
 
   // pp 156
 
-  void fadd( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x40 + w) | fs2(s2, w)); }
-  void fsub( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x44 + w) | fs2(s2, w)); }
+  void fadd( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x40 + w) | fs2(s2, w)); }
+  void fsub( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x44 + w) | fs2(s2, w)); }
 
   // pp 157
 
-  void fcmp(  FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { v8_no_cc(cc);  emit_long( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x50 + w) | fs2(s2, w)); }
-  void fcmpe( FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { v8_no_cc(cc);  emit_long( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x54 + w) | fs2(s2, w)); }
+  void fcmp(  FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { v8_no_cc(cc);  emit_int32( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x50 + w) | fs2(s2, w)); }
+  void fcmpe( FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { v8_no_cc(cc);  emit_int32( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x54 + w) | fs2(s2, w)); }
 
   // pp 159
 
-  void ftox( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_long( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(fpop1_op3) | opf(0x80 + w) | fs2(s, w)); }
-  void ftoi( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_long( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(fpop1_op3) | opf(0xd0 + w) | fs2(s, w)); }
+  void ftox( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(fpop1_op3) | opf(0x80 + w) | fs2(s, w)); }
+  void ftoi( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(fpop1_op3) | opf(0xd0 + w) | fs2(s, w)); }
 
   // pp 160
 
-  void ftof( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw, FloatRegister s, FloatRegister d ) { emit_long( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | opf(0xc0 + sw + dw*4) | fs2(s, sw)); }
+  void ftof( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | opf(0xc0 + sw + dw*4) | fs2(s, sw)); }
 
   // pp 161
 
-  void fxtof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x80 + w*4) | fs2(s, FloatRegisterImpl::D)); }
-  void fitof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0xc0 + w*4) | fs2(s, FloatRegisterImpl::S)); }
+  void fxtof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x80 + w*4) | fs2(s, FloatRegisterImpl::D)); }
+  void fitof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0xc0 + w*4) | fs2(s, FloatRegisterImpl::S)); }
 
   // pp 162
 
-  void fmov( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x00 + w) | fs2(s, w)); }
+  void fmov( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x00 + w) | fs2(s, w)); }
 
-  void fneg( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x04 + w) | fs2(s, w)); }
+  void fneg( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x04 + w) | fs2(s, w)); }
 
   // page 144 sparc v8 architecture (double prec works on v8 if the source and destination registers are the same). fnegs is the only instruction available
   // on v8 to do negation of single, double and quad precision floats.
 
-  void fneg( FloatRegisterImpl::Width w, FloatRegister sd ) { if (VM_Version::v9_instructions_work()) emit_long( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x04 + w) | fs2(sd, w)); else emit_long( op(arith_op) | fd(sd, w) | op3(fpop1_op3) |  opf(0x05) | fs2(sd, w)); }
+  void fneg( FloatRegisterImpl::Width w, FloatRegister sd ) { if (VM_Version::v9_instructions_work()) emit_int32( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x04 + w) | fs2(sd, w)); else emit_int32( op(arith_op) | fd(sd, w) | op3(fpop1_op3) |  opf(0x05) | fs2(sd, w)); }
 
-  void fabs( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x08 + w) | fs2(s, w)); }
+  void fabs( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v8_s_only(w);  emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x08 + w) | fs2(s, w)); }
 
   // page 144 sparc v8 architecture (double prec works on v8 if the source and destination registers are the same). fabss is the only instruction available
   // on v8 to do abs operation on single/double/quad precision floats.
 
-  void fabs( FloatRegisterImpl::Width w, FloatRegister sd ) { if (VM_Version::v9_instructions_work()) emit_long( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x08 + w) | fs2(sd, w)); else emit_long( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x09) | fs2(sd, w)); }
+  void fabs( FloatRegisterImpl::Width w, FloatRegister sd ) { if (VM_Version::v9_instructions_work()) emit_int32( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x08 + w) | fs2(sd, w)); else emit_int32( op(arith_op) | fd(sd, w) | op3(fpop1_op3) | opf(0x09) | fs2(sd, w)); }
 
   // pp 163
 
-  void fmul( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_long( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x48 + w)         | fs2(s2, w)); }
-  void fmul( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw,  FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_long( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | fs1(s1, sw) | opf(0x60 + sw + dw*4) | fs2(s2, sw)); }
-  void fdiv( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_long( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x4c + w)         | fs2(s2, w)); }
+  void fmul( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x48 + w)         | fs2(s2, w)); }
+  void fmul( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw,  FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | fs1(s1, sw) | opf(0x60 + sw + dw*4) | fs2(s2, sw)); }
+  void fdiv( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x4c + w)         | fs2(s2, w)); }
 
   // pp 164
 
-  void fsqrt( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_long( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x28 + w) | fs2(s, w)); }
+  void fsqrt( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x28 + w) | fs2(s, w)); }
 
   // pp 165
 
@@ -827,22 +827,22 @@
 
   // pp 167
 
-  void flushw() { v9_only();  emit_long( op(arith_op) | op3(flushw_op3) ); }
+  void flushw() { v9_only();  emit_int32( op(arith_op) | op3(flushw_op3) ); }
 
   // pp 168
 
-  void illtrap( int const22a) { if (const22a != 0) v9_only();  emit_long( op(branch_op) | u_field(const22a, 21, 0) ); }
+  void illtrap( int const22a) { if (const22a != 0) v9_only();  emit_int32( op(branch_op) | u_field(const22a, 21, 0) ); }
   // v8 unimp == illtrap(0)
 
   // pp 169
 
-  void impdep1( int id1, int const19a ) { v9_only();  emit_long( op(arith_op) | fcn(id1) | op3(impdep1_op3) | u_field(const19a, 18, 0)); }
-  void impdep2( int id1, int const19a ) { v9_only();  emit_long( op(arith_op) | fcn(id1) | op3(impdep2_op3) | u_field(const19a, 18, 0)); }
+  void impdep1( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep1_op3) | u_field(const19a, 18, 0)); }
+  void impdep2( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep2_op3) | u_field(const19a, 18, 0)); }
 
   // pp 149 (v8)
 
-  void cpop1( int opc, int cr1, int cr2, int crd ) { v8_only();  emit_long( op(arith_op) | fcn(crd) | op3(impdep1_op3) | u_field(cr1, 18, 14) | opf(opc) | u_field(cr2, 4, 0)); }
-  void cpop2( int opc, int cr1, int cr2, int crd ) { v8_only();  emit_long( op(arith_op) | fcn(crd) | op3(impdep2_op3) | u_field(cr1, 18, 14) | opf(opc) | u_field(cr2, 4, 0)); }
+  void cpop1( int opc, int cr1, int cr2, int crd ) { v8_only();  emit_int32( op(arith_op) | fcn(crd) | op3(impdep1_op3) | u_field(cr1, 18, 14) | opf(opc) | u_field(cr2, 4, 0)); }
+  void cpop2( int opc, int cr1, int cr2, int crd ) { v8_only();  emit_int32( op(arith_op) | fcn(crd) | op3(impdep2_op3) | u_field(cr1, 18, 14) | opf(opc) | u_field(cr2, 4, 0)); }
 
   // pp 170
 
@@ -872,8 +872,8 @@
 
   // 173
 
-  void ldfa(  FloatRegisterImpl::Width w, Register s1, Register s2, int ia, FloatRegister d ) { v9_only();  emit_long( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldfa(  FloatRegisterImpl::Width w, Register s1, int simm13a,         FloatRegister d ) { v9_only();  emit_long( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldfa(  FloatRegisterImpl::Width w, Register s1, Register s2, int ia, FloatRegister d ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldfa(  FloatRegisterImpl::Width w, Register s1, int simm13a,         FloatRegister d ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 175, lduw is ld on v8
 
@@ -896,22 +896,22 @@
 
   // pp 177
 
-  void ldsba(  Register s1, Register s2, int ia, Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldsba(  Register s1, int simm13a,         Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void ldsha(  Register s1, Register s2, int ia, Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldsha(  Register s1, int simm13a,         Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void ldswa(  Register s1, Register s2, int ia, Register d ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldswa(  Register s1, int simm13a,         Register d ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void lduba(  Register s1, Register s2, int ia, Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void lduba(  Register s1, int simm13a,         Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void lduha(  Register s1, Register s2, int ia, Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void lduha(  Register s1, int simm13a,         Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void lduwa(  Register s1, Register s2, int ia, Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void lduwa(  Register s1, int simm13a,         Register d ) {             emit_long( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void ldxa(   Register s1, Register s2, int ia, Register d ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldxa(   Register s1, int simm13a,         Register d ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void ldda(   Register s1, Register s2, int ia, Register d ) { v9_dep();   emit_long( op(ldst_op) | rd(d) | op3(ldd_op3  | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldda(   Register s1, int simm13a,         Register d ) { v9_dep();   emit_long( op(ldst_op) | rd(d) | op3(ldd_op3  | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldsba(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldsba(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldsha(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldsha(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldswa(  Register s1, Register s2, int ia, Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldswa(  Register s1, int simm13a,         Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void lduba(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void lduba(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void lduha(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void lduha(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void lduwa(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void lduwa(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldxa(   Register s1, Register s2, int ia, Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldxa(   Register s1, int simm13a,         Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldda(   Register s1, Register s2, int ia, Register d ) { v9_dep();   emit_int32( op(ldst_op) | rd(d) | op3(ldd_op3  | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldda(   Register s1, int simm13a,         Register d ) { v9_dep();   emit_int32( op(ldst_op) | rd(d) | op3(ldd_op3  | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 179
 
@@ -920,111 +920,111 @@
 
   // pp 180
 
-  void ldstuba( Register s1, Register s2, int ia, Register d ) { emit_long( op(ldst_op) | rd(d) | op3(ldstub_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void ldstuba( Register s1, int simm13a,         Register d ) { emit_long( op(ldst_op) | rd(d) | op3(ldstub_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void ldstuba( Register s1, Register s2, int ia, Register d ) { emit_int32( op(ldst_op) | rd(d) | op3(ldstub_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void ldstuba( Register s1, int simm13a,         Register d ) { emit_int32( op(ldst_op) | rd(d) | op3(ldstub_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 181
 
-  void and3(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | rs2(s2) ); }
-  void and3(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void andcc(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void andcc(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void andn(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | rs2(s2) ); }
-  void andn(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void andncc(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void andncc(  Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void or3(     Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | rs2(s2) ); }
-  void or3(     Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void orcc(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void orcc(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void orn(     Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | rs2(s2) ); }
-  void orn(     Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void orncc(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void orncc(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void xor3(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | rs2(s2) ); }
-  void xor3(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void xorcc(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void xorcc(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void xnor(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | rs2(s2) ); }
-  void xnor(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void xnorcc(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void xnorcc(  Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void and3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | rs2(s2) ); }
+  void and3(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void andcc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void andcc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void andn(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | rs2(s2) ); }
+  void andn(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void andncc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void andncc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void or3(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | rs2(s2) ); }
+  void or3(     Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void orcc(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void orcc(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void orn(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | rs2(s2) ); }
+  void orn(     Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void orncc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void orncc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void xor3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | rs2(s2) ); }
+  void xor3(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void xorcc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void xorcc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void xnor(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | rs2(s2) ); }
+  void xnor(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void xnorcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void xnorcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 183
 
-  void membar( Membar_mask_bits const7a ) { v9_only(); emit_long( op(arith_op) | op3(membar_op3) | rs1(O7) | immed(true) | u_field( int(const7a), 6, 0)); }
+  void membar( Membar_mask_bits const7a ) { v9_only(); emit_int32( op(arith_op) | op3(membar_op3) | rs1(O7) | immed(true) | u_field( int(const7a), 6, 0)); }
 
   // pp 185
 
-  void fmov( FloatRegisterImpl::Width w, Condition c,  bool floatCC, CC cca, FloatRegister s2, FloatRegister d ) { v9_only();  emit_long( op(arith_op) | fd(d, w) | op3(fpop2_op3) | cond_mov(c) | opf_cc(cca, floatCC) | opf_low6(w) | fs2(s2, w)); }
+  void fmov( FloatRegisterImpl::Width w, Condition c,  bool floatCC, CC cca, FloatRegister s2, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop2_op3) | cond_mov(c) | opf_cc(cca, floatCC) | opf_low6(w) | fs2(s2, w)); }
 
   // pp 189
 
-  void fmov( FloatRegisterImpl::Width w, RCondition c, Register s1,  FloatRegister s2, FloatRegister d ) { v9_only();  emit_long( op(arith_op) | fd(d, w) | op3(fpop2_op3) | rs1(s1) | rcond(c) | opf_low5(4 + w) | fs2(s2, w)); }
+  void fmov( FloatRegisterImpl::Width w, RCondition c, Register s1,  FloatRegister s2, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop2_op3) | rs1(s1) | rcond(c) | opf_low5(4 + w) | fs2(s2, w)); }
 
   // pp 191
 
-  void movcc( Condition c, bool floatCC, CC cca, Register s2, Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | rs2(s2) ); }
-  void movcc( Condition c, bool floatCC, CC cca, int simm11a, Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | immed(true) | simm(simm11a, 11) ); }
+  void movcc( Condition c, bool floatCC, CC cca, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | rs2(s2) ); }
+  void movcc( Condition c, bool floatCC, CC cca, int simm11a, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | immed(true) | simm(simm11a, 11) ); }
 
   // pp 195
 
-  void movr( RCondition c, Register s1, Register s2,  Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | rs2(s2) ); }
-  void movr( RCondition c, Register s1, int simm10a,  Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | immed(true) | simm(simm10a, 10) ); }
+  void movr( RCondition c, Register s1, Register s2,  Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | rs2(s2) ); }
+  void movr( RCondition c, Register s1, int simm10a,  Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | immed(true) | simm(simm10a, 10) ); }
 
   // pp 196
 
-  void mulx(  Register s1, Register s2, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | rs2(s2) ); }
-  void mulx(  Register s1, int simm13a, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void sdivx( Register s1, Register s2, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | rs2(s2) ); }
-  void sdivx( Register s1, int simm13a, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void udivx( Register s1, Register s2, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | rs2(s2) ); }
-  void udivx( Register s1, int simm13a, Register d ) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void mulx(  Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | rs2(s2) ); }
+  void mulx(  Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void sdivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | rs2(s2) ); }
+  void sdivx( Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void udivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | rs2(s2) ); }
+  void udivx( Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 197
 
-  void umul(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | rs2(s2) ); }
-  void umul(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void smul(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | rs2(s2) ); }
-  void smul(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void umulcc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void umulcc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void smulcc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void smulcc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void umul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | rs2(s2) ); }
+  void umul(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void smul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | rs2(s2) ); }
+  void smul(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void umulcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void umulcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void smulcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void smulcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 199
 
-  void mulscc(   Register s1, Register s2, Register d ) { v9_dep();  emit_long( op(arith_op) | rd(d) | op3(mulscc_op3) | rs1(s1) | rs2(s2) ); }
-  void mulscc(   Register s1, int simm13a, Register d ) { v9_dep();  emit_long( op(arith_op) | rd(d) | op3(mulscc_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void mulscc(   Register s1, Register s2, Register d ) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(mulscc_op3) | rs1(s1) | rs2(s2) ); }
+  void mulscc(   Register s1, int simm13a, Register d ) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(mulscc_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 201
 
-  void nop() { emit_long( op(branch_op) | op2(sethi_op2) ); }
+  void nop() { emit_int32( op(branch_op) | op2(sethi_op2) ); }
 
 
   // pp 202
 
-  void popc( Register s,  Register d) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(popc_op3) | rs2(s)); }
-  void popc( int simm13a, Register d) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(popc_op3) | immed(true) | simm(simm13a, 13)); }
+  void popc( Register s,  Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | rs2(s)); }
+  void popc( int simm13a, Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | immed(true) | simm(simm13a, 13)); }
 
   // pp 203
 
-  void prefetch(   Register s1, Register s2, PrefetchFcn f) { v9_only();  emit_long( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | rs2(s2) ); }
+  void prefetch(   Register s1, Register s2, PrefetchFcn f) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | rs2(s2) ); }
   void prefetch(   Register s1, int simm13a, PrefetchFcn f) { v9_only();  emit_data( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-  void prefetcha(  Register s1, Register s2, int ia, PrefetchFcn f ) { v9_only();  emit_long( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void prefetcha(  Register s1, int simm13a,         PrefetchFcn f ) { v9_only();  emit_long( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void prefetcha(  Register s1, Register s2, int ia, PrefetchFcn f ) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void prefetcha(  Register s1, int simm13a,         PrefetchFcn f ) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 208
 
   // not implementing read privileged register
 
-  inline void rdy(    Register d) { v9_dep();  emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(0, 18, 14)); }
-  inline void rdccr(  Register d) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(2, 18, 14)); }
-  inline void rdasi(  Register d) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(3, 18, 14)); }
-  inline void rdtick( Register d) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(4, 18, 14)); } // Spoon!
-  inline void rdpc(   Register d) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(5, 18, 14)); }
-  inline void rdfprs( Register d) { v9_only(); emit_long( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(6, 18, 14)); }
+  inline void rdy(    Register d) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(0, 18, 14)); }
+  inline void rdccr(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(2, 18, 14)); }
+  inline void rdasi(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(3, 18, 14)); }
+  inline void rdtick( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(4, 18, 14)); } // Spoon!
+  inline void rdpc(   Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(5, 18, 14)); }
+  inline void rdfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(6, 18, 14)); }
 
   // pp 213
 
@@ -1033,47 +1033,47 @@
 
   // pp 214
 
-  void save(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | rs2(s2) ); }
+  void save(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | rs2(s2) ); }
   void save(    Register s1, int simm13a, Register d ) {
     // make sure frame is at least large enough for the register save area
     assert(-simm13a >= 16 * wordSize, "frame too small");
-    emit_long( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) );
+    emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) );
   }
 
-  void restore( Register s1 = G0,  Register s2 = G0, Register d = G0 ) { emit_long( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | rs2(s2) ); }
-  void restore( Register s1,       int simm13a,      Register d      ) { emit_long( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void restore( Register s1 = G0,  Register s2 = G0, Register d = G0 ) { emit_int32( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | rs2(s2) ); }
+  void restore( Register s1,       int simm13a,      Register d      ) { emit_int32( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 216
 
-  void saved()    { v9_only();  emit_long( op(arith_op) | fcn(0) | op3(saved_op3)); }
-  void restored() { v9_only();  emit_long( op(arith_op) | fcn(1) | op3(saved_op3)); }
+  void saved()    { v9_only();  emit_int32( op(arith_op) | fcn(0) | op3(saved_op3)); }
+  void restored() { v9_only();  emit_int32( op(arith_op) | fcn(1) | op3(saved_op3)); }
 
   // pp 217
 
   inline void sethi( int imm22a, Register d, RelocationHolder const& rspec = RelocationHolder() );
   // pp 218
 
-  void sll(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
-  void sll(  Register s1, int imm5a,   Register d ) { emit_long( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
-  void srl(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
-  void srl(  Register s1, int imm5a,   Register d ) { emit_long( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
-  void sra(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
-  void sra(  Register s1, int imm5a,   Register d ) { emit_long( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
+  void sll(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
+  void sll(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
+  void srl(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
+  void srl(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
+  void sra(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
+  void sra(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
 
-  void sllx( Register s1, Register s2, Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
-  void sllx( Register s1, int imm6a,   Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
-  void srlx( Register s1, Register s2, Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
-  void srlx( Register s1, int imm6a,   Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
-  void srax( Register s1, Register s2, Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
-  void srax( Register s1, int imm6a,   Register d ) { v9_only();  emit_long( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
+  void sllx( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
+  void sllx( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
+  void srlx( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
+  void srlx( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
+  void srax( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
+  void srax( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
 
   // pp 220
 
-  void sir( int simm13a ) { emit_long( op(arith_op) | fcn(15) | op3(sir_op3) | immed(true) | simm(simm13a, 13)); }
+  void sir( int simm13a ) { emit_int32( op(arith_op) | fcn(15) | op3(sir_op3) | immed(true) | simm(simm13a, 13)); }
 
   // pp 221
 
-  void stbar() { emit_long( op(arith_op) | op3(membar_op3) | u_field(15, 18, 14)); }
+  void stbar() { emit_int32( op(arith_op) | op3(membar_op3) | u_field(15, 18, 14)); }
 
   // pp 222
 
@@ -1087,8 +1087,8 @@
 
   //  pp 224
 
-  void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2, int ia ) { v9_only();  emit_long( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a         ) { v9_only();  emit_long( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2, int ia ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a         ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // p 226
 
@@ -1105,16 +1105,16 @@
 
   // pp 177
 
-  void stba(  Register d, Register s1, Register s2, int ia ) {             emit_long( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stba(  Register d, Register s1, int simm13a         ) {             emit_long( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void stha(  Register d, Register s1, Register s2, int ia ) {             emit_long( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stha(  Register d, Register s1, int simm13a         ) {             emit_long( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void stwa(  Register d, Register s1, Register s2, int ia ) {             emit_long( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stwa(  Register d, Register s1, int simm13a         ) {             emit_long( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void stxa(  Register d, Register s1, Register s2, int ia ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stxa(  Register d, Register s1, int simm13a         ) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void stda(  Register d, Register s1, Register s2, int ia ) {             emit_long( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void stda(  Register d, Register s1, int simm13a         ) {             emit_long( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stba(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stba(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stha(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stha(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stwa(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stwa(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stxa(  Register d, Register s1, Register s2, int ia ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stxa(  Register d, Register s1, int simm13a         ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void stda(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void stda(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 97 (v8)
 
@@ -1129,15 +1129,15 @@
 
   // pp 230
 
-  void sub(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | rs2(s2) ); }
-  void sub(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void sub(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | rs2(s2) ); }
+  void sub(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
-  void subcc(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | rs2(s2) ); }
-  void subcc(  Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void subc(   Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | rs2(s2) ); }
-  void subc(   Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void subccc( Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
-  void subccc( Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void subcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | rs2(s2) ); }
+  void subcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void subc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | rs2(s2) ); }
+  void subc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void subccc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
+  void subccc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 231
 
@@ -1146,55 +1146,55 @@
 
   // pp 232
 
-  void swapa(   Register s1, Register s2, int ia, Register d ) { v9_dep();  emit_long( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
-  void swapa(   Register s1, int simm13a,         Register d ) { v9_dep();  emit_long( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void swapa(   Register s1, Register s2, int ia, Register d ) { v9_dep();  emit_int32( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
+  void swapa(   Register s1, int simm13a,         Register d ) { v9_dep();  emit_int32( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 234, note op in book is wrong, see pp 268
 
-  void taddcc(    Register s1, Register s2, Register d ) {            emit_long( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | rs2(s2) ); }
-  void taddcc(    Register s1, int simm13a, Register d ) {            emit_long( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void taddcctv(  Register s1, Register s2, Register d ) { v9_dep();  emit_long( op(arith_op) | rd(d) | op3(taddcctv_op3) | rs1(s1) | rs2(s2) ); }
-  void taddcctv(  Register s1, int simm13a, Register d ) { v9_dep();  emit_long( op(arith_op) | rd(d) | op3(taddcctv_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void taddcc(    Register s1, Register s2, Register d ) {            emit_int32( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | rs2(s2) ); }
+  void taddcc(    Register s1, int simm13a, Register d ) {            emit_int32( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void taddcctv(  Register s1, Register s2, Register d ) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(taddcctv_op3) | rs1(s1) | rs2(s2) ); }
+  void taddcctv(  Register s1, int simm13a, Register d ) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(taddcctv_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 235
 
-  void tsubcc(    Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | rs2(s2) ); }
-  void tsubcc(    Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
-  void tsubcctv(  Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(tsubcctv_op3) | rs1(s1) | rs2(s2) ); }
-  void tsubcctv(  Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(tsubcctv_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void tsubcc(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | rs2(s2) ); }
+  void tsubcc(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+  void tsubcctv(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcctv_op3) | rs1(s1) | rs2(s2) ); }
+  void tsubcctv(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcctv_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
   // pp 237
 
-  void trap( Condition c, CC cc, Register s1, Register s2 ) { v8_no_cc(cc);  emit_long( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | rs2(s2)); }
-  void trap( Condition c, CC cc, Register s1, int trapa   ) { v8_no_cc(cc);  emit_long( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | immed(true) | u_field(trapa, 6, 0)); }
+  void trap( Condition c, CC cc, Register s1, Register s2 ) { v8_no_cc(cc);  emit_int32( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | rs2(s2)); }
+  void trap( Condition c, CC cc, Register s1, int trapa   ) { v8_no_cc(cc);  emit_int32( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | immed(true) | u_field(trapa, 6, 0)); }
   // simple uncond. trap
   void trap( int trapa ) { trap( always, icc, G0, trapa ); }
 
   // pp 239 omit write priv register for now
 
-  inline void wry(    Register d) { v9_dep();  emit_long( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(0, 29, 25)); }
-  inline void wrccr(Register s) { v9_only(); emit_long( op(arith_op) | rs1(s) | op3(wrreg_op3) | u_field(2, 29, 25)); }
-  inline void wrccr(Register s, int simm13a) { v9_only(); emit_long( op(arith_op) |
+  inline void wry(    Register d) { v9_dep();  emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(0, 29, 25)); }
+  inline void wrccr(Register s) { v9_only(); emit_int32( op(arith_op) | rs1(s) | op3(wrreg_op3) | u_field(2, 29, 25)); }
+  inline void wrccr(Register s, int simm13a) { v9_only(); emit_int32( op(arith_op) |
                                                                            rs1(s) |
                                                                            op3(wrreg_op3) |
                                                                            u_field(2, 29, 25) |
                                                                            immed(true) |
                                                                            simm(simm13a, 13)); }
-  inline void wrasi(Register d) { v9_only(); emit_long( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(3, 29, 25)); }
+  inline void wrasi(Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(3, 29, 25)); }
   // wrasi(d, imm) stores (d xor imm) to asi
-  inline void wrasi(Register d, int simm13a) { v9_only(); emit_long( op(arith_op) | rs1(d) | op3(wrreg_op3) |
+  inline void wrasi(Register d, int simm13a) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) |
                                                u_field(3, 29, 25) | immed(true) | simm(simm13a, 13)); }
-  inline void wrfprs( Register d) { v9_only(); emit_long( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(6, 29, 25)); }
+  inline void wrfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(6, 29, 25)); }
 
 
   // VIS3 instructions
 
-  void movstosw( FloatRegister s, Register d ) { vis3_only();  emit_long( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstosw_opf) | fs2(s, FloatRegisterImpl::S)); }
-  void movstouw( FloatRegister s, Register d ) { vis3_only();  emit_long( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstouw_opf) | fs2(s, FloatRegisterImpl::S)); }
-  void movdtox(  FloatRegister s, Register d ) { vis3_only();  emit_long( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mdtox_opf) | fs2(s, FloatRegisterImpl::D)); }
+  void movstosw( FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstosw_opf) | fs2(s, FloatRegisterImpl::S)); }
+  void movstouw( FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstouw_opf) | fs2(s, FloatRegisterImpl::S)); }
+  void movdtox(  FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mdtox_opf) | fs2(s, FloatRegisterImpl::D)); }
 
-  void movwtos( Register s, FloatRegister d ) { vis3_only();  emit_long( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(mftoi_op3) | opf(mwtos_opf) | rs2(s)); }
-  void movxtod( Register s, FloatRegister d ) { vis3_only();  emit_long( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(mftoi_op3) | opf(mxtod_opf) | rs2(s)); }
+  void movwtos( Register s, FloatRegister d ) { vis3_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(mftoi_op3) | opf(mwtos_opf) | rs2(s)); }
+  void movxtod( Register s, FloatRegister d ) { vis3_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(mftoi_op3) | opf(mxtod_opf) | rs2(s)); }
 
   // Creation
   Assembler(CodeBuffer* code) : AbstractAssembler(code) {
--- a/src/cpu/sparc/vm/assembler_sparc.inline.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/sparc/vm/assembler_sparc.inline.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -35,24 +35,24 @@
 # endif
 }
 
-inline void Assembler::emit_long(int x) {
+inline void Assembler::emit_int32(int x) {
   check_delay();
-  AbstractAssembler::emit_long(x);
+  AbstractAssembler::emit_int32(x);
 }
 
 inline void Assembler::emit_data(int x, relocInfo::relocType rtype) {
   relocate(rtype);
-  emit_long(x);
+  emit_int32(x);
 }
 
 inline void Assembler::emit_data(int x, RelocationHolder const& rspec) {
   relocate(rspec);
-  emit_long(x);
+  emit_int32(x);
 }
 
 
-inline void Assembler::add(Register s1, Register s2, Register d ) { emit_long( op(arith_op) | rd(d) | op3(add_op3) | rs1(s1) | rs2(s2) ); }
-inline void Assembler::add(Register s1, int simm13a, Register d ) { emit_long( op(arith_op) | rd(d) | op3(add_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
+inline void Assembler::add(Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::add(Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 
 inline void Assembler::bpr( RCondition c, bool a, Predict p, Register s1, address d, relocInfo::relocType rt ) { v9_only();  cti();  emit_data( op(branch_op) | annul(a) | cond(c) | op2(bpr_op2) | wdisp16(intptr_t(d), intptr_t(pc())) | predict(p) | rs1(s1), rt);  has_delay_slot(); }
 inline void Assembler::bpr( RCondition c, bool a, Predict p, Register s1, Label& L) { bpr( c, a, p, s1, target(L)); }
@@ -79,93 +79,93 @@
 inline void Assembler::call( address d,  relocInfo::relocType rt ) { cti();  emit_data( op(call_op) | wdisp(intptr_t(d), intptr_t(pc()), 30), rt);  has_delay_slot(); assert(rt != relocInfo::virtual_call_type, "must use virtual_call_Relocation::spec"); }
 inline void Assembler::call( Label& L,   relocInfo::relocType rt ) { call( target(L), rt); }
 
-inline void Assembler::flush( Register s1, Register s2) { emit_long( op(arith_op) | op3(flush_op3) | rs1(s1) | rs2(s2)); }
+inline void Assembler::flush( Register s1, Register s2) { emit_int32( op(arith_op) | op3(flush_op3) | rs1(s1) | rs2(s2)); }
 inline void Assembler::flush( Register s1, int simm13a) { emit_data( op(arith_op) | op3(flush_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::jmpl( Register s1, Register s2, Register d ) { cti();  emit_long( op(arith_op) | rd(d) | op3(jmpl_op3) | rs1(s1) | rs2(s2));  has_delay_slot(); }
+inline void Assembler::jmpl( Register s1, Register s2, Register d ) { cti();  emit_int32( op(arith_op) | rd(d) | op3(jmpl_op3) | rs1(s1) | rs2(s2));  has_delay_slot(); }
 inline void Assembler::jmpl( Register s1, int simm13a, Register d, RelocationHolder const& rspec ) { cti();  emit_data( op(arith_op) | rd(d) | op3(jmpl_op3) | rs1(s1) | immed(true) | simm(simm13a, 13), rspec);  has_delay_slot(); }
 
-inline void Assembler::ldf(FloatRegisterImpl::Width w, Register s1, Register s2, FloatRegister d) { emit_long( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3, w) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldf(FloatRegisterImpl::Width w, Register s1, Register s2, FloatRegister d) { emit_int32( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3, w) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldf(FloatRegisterImpl::Width w, Register s1, int simm13a, FloatRegister d, RelocationHolder const& rspec) { emit_data( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13), rspec); }
 
-inline void Assembler::ldfsr(  Register s1, Register s2) { v9_dep();   emit_long( op(ldst_op) |             op3(ldfsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldfsr(  Register s1, Register s2) { v9_dep();   emit_int32( op(ldst_op) |             op3(ldfsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldfsr(  Register s1, int simm13a) { v9_dep();   emit_data( op(ldst_op) |             op3(ldfsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::ldxfsr( Register s1, Register s2) { v9_only();  emit_long( op(ldst_op) | rd(G1)    | op3(ldfsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldxfsr( Register s1, Register s2) { v9_only();  emit_int32( op(ldst_op) | rd(G1)    | op3(ldfsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldxfsr( Register s1, int simm13a) { v9_only();  emit_data( op(ldst_op) | rd(G1)    | op3(ldfsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::ldc(   Register s1, Register s2, int crd) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(ldc_op3  ) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldc(   Register s1, Register s2, int crd) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(ldc_op3  ) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldc(   Register s1, int simm13a, int crd) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(ldc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::lddc(  Register s1, Register s2, int crd) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(lddc_op3 ) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::lddc(  Register s1, Register s2, int crd) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(lddc_op3 ) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::lddc(  Register s1, int simm13a, int crd) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(lddc_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::ldcsr( Register s1, Register s2, int crd) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(ldcsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldcsr( Register s1, Register s2, int crd) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(ldcsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldcsr( Register s1, int simm13a, int crd) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(ldcsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::ldsb(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(ldsb_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldsb(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(ldsb_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldsb(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(ldsb_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::ldsh(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(ldsh_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldsh(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(ldsh_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldsh(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(ldsh_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::ldsw(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(ldsw_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldsw(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(ldsw_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldsw(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(ldsw_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::ldub(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(ldub_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldub(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(ldub_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldub(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(ldub_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::lduh(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(lduh_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::lduh(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(lduh_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::lduh(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(lduh_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::lduw(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(lduw_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::lduw(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(lduw_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::lduw(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(lduw_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::ldx(   Register s1, Register s2, Register d) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(ldx_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldx(   Register s1, Register s2, Register d) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldx_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldx(   Register s1, int simm13a, Register d) { v9_only();  emit_data( op(ldst_op) | rd(d) | op3(ldx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::ldd(   Register s1, Register s2, Register d) { v9_dep(); assert(d->is_even(), "not even"); emit_long( op(ldst_op) | rd(d) | op3(ldd_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldd(   Register s1, Register s2, Register d) { v9_dep(); assert(d->is_even(), "not even"); emit_int32( op(ldst_op) | rd(d) | op3(ldd_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldd(   Register s1, int simm13a, Register d) { v9_dep(); assert(d->is_even(), "not even"); emit_data( op(ldst_op) | rd(d) | op3(ldd_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::ldstub(  Register s1, Register s2, Register d) { emit_long( op(ldst_op) | rd(d) | op3(ldstub_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::ldstub(  Register s1, Register s2, Register d) { emit_int32( op(ldst_op) | rd(d) | op3(ldstub_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::ldstub(  Register s1, int simm13a, Register d) { emit_data( op(ldst_op) | rd(d) | op3(ldstub_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::rett( Register s1, Register s2                         ) { cti();  emit_long( op(arith_op) | op3(rett_op3) | rs1(s1) | rs2(s2));  has_delay_slot(); }
+inline void Assembler::rett( Register s1, Register s2                         ) { cti();  emit_int32( op(arith_op) | op3(rett_op3) | rs1(s1) | rs2(s2));  has_delay_slot(); }
 inline void Assembler::rett( Register s1, int simm13a, relocInfo::relocType rt) { cti();  emit_data( op(arith_op) | op3(rett_op3) | rs1(s1) | immed(true) | simm(simm13a, 13), rt);  has_delay_slot(); }
 
 inline void Assembler::sethi( int imm22a, Register d, RelocationHolder const& rspec ) { emit_data( op(branch_op) | rd(d) | op2(sethi_op2) | hi22(imm22a), rspec); }
 
   // pp 222
 
-inline void Assembler::stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2) { emit_long( op(ldst_op) | fd(d, w) | alt_op3(stf_op3, w) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2) { emit_int32( op(ldst_op) | fd(d, w) | alt_op3(stf_op3, w) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a) { emit_data( op(ldst_op) | fd(d, w) | alt_op3(stf_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::stfsr(  Register s1, Register s2) { v9_dep();   emit_long( op(ldst_op) |             op3(stfsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stfsr(  Register s1, Register s2) { v9_dep();   emit_int32( op(ldst_op) |             op3(stfsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stfsr(  Register s1, int simm13a) { v9_dep();   emit_data( op(ldst_op) |             op3(stfsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::stxfsr( Register s1, Register s2) { v9_only();  emit_long( op(ldst_op) | rd(G1)    | op3(stfsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stxfsr( Register s1, Register s2) { v9_only();  emit_int32( op(ldst_op) | rd(G1)    | op3(stfsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stxfsr( Register s1, int simm13a) { v9_only();  emit_data( op(ldst_op) | rd(G1)    | op3(stfsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
   // p 226
 
-inline void Assembler::stb(  Register d, Register s1, Register s2) { emit_long( op(ldst_op) | rd(d) | op3(stb_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stb(  Register d, Register s1, Register s2) { emit_int32( op(ldst_op) | rd(d) | op3(stb_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stb(  Register d, Register s1, int simm13a) { emit_data( op(ldst_op) | rd(d) | op3(stb_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::sth(  Register d, Register s1, Register s2) { emit_long( op(ldst_op) | rd(d) | op3(sth_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::sth(  Register d, Register s1, Register s2) { emit_int32( op(ldst_op) | rd(d) | op3(sth_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::sth(  Register d, Register s1, int simm13a) { emit_data( op(ldst_op) | rd(d) | op3(sth_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::stw(  Register d, Register s1, Register s2) { emit_long( op(ldst_op) | rd(d) | op3(stw_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stw(  Register d, Register s1, Register s2) { emit_int32( op(ldst_op) | rd(d) | op3(stw_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stw(  Register d, Register s1, int simm13a) { emit_data( op(ldst_op) | rd(d) | op3(stw_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
 
-inline void Assembler::stx(  Register d, Register s1, Register s2) { v9_only();  emit_long( op(ldst_op) | rd(d) | op3(stx_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stx(  Register d, Register s1, Register s2) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(stx_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stx(  Register d, Register s1, int simm13a) { v9_only();  emit_data( op(ldst_op) | rd(d) | op3(stx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::std(  Register d, Register s1, Register s2) { v9_dep(); assert(d->is_even(), "not even"); emit_long( op(ldst_op) | rd(d) | op3(std_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::std(  Register d, Register s1, Register s2) { v9_dep(); assert(d->is_even(), "not even"); emit_int32( op(ldst_op) | rd(d) | op3(std_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::std(  Register d, Register s1, int simm13a) { v9_dep(); assert(d->is_even(), "not even"); emit_data( op(ldst_op) | rd(d) | op3(std_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
 // v8 p 99
 
-inline void Assembler::stc(    int crd, Register s1, Register s2) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(stc_op3 ) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stc(    int crd, Register s1, Register s2) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(stc_op3 ) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stc(    int crd, Register s1, int simm13a) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(stc_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::stdc(   int crd, Register s1, Register s2) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(stdc_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stdc(   int crd, Register s1, Register s2) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(stdc_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stdc(   int crd, Register s1, int simm13a) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(stdc_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::stcsr(  int crd, Register s1, Register s2) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(stcsr_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stcsr(  int crd, Register s1, Register s2) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(stcsr_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stcsr(  int crd, Register s1, int simm13a) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(stcsr_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
-inline void Assembler::stdcq(  int crd, Register s1, Register s2) { v8_only();  emit_long( op(ldst_op) | fcn(crd) | op3(stdcq_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::stdcq(  int crd, Register s1, Register s2) { v8_only();  emit_int32( op(ldst_op) | fcn(crd) | op3(stdcq_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::stdcq(  int crd, Register s1, int simm13a) { v8_only();  emit_data( op(ldst_op) | fcn(crd) | op3(stdcq_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
 // pp 231
 
-inline void Assembler::swap(    Register s1, Register s2, Register d) { v9_dep();  emit_long( op(ldst_op) | rd(d) | op3(swap_op3) | rs1(s1) | rs2(s2) ); }
+inline void Assembler::swap(    Register s1, Register s2, Register d) { v9_dep();  emit_int32( op(ldst_op) | rd(d) | op3(swap_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::swap(    Register s1, int simm13a, Register d) { v9_dep();  emit_data( op(ldst_op) | rd(d) | op3(swap_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
 #endif // CPU_SPARC_VM_ASSEMBLER_SPARC_INLINE_HPP
--- a/src/cpu/sparc/vm/cppInterpreter_sparc.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/sparc/vm/cppInterpreter_sparc.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -137,7 +137,7 @@
   }
   __ ret();                           // return from interpreter activation
   __ delayed()->restore(I5_savedSP, G0, SP);  // remove interpreter frame
-  NOT_PRODUCT(__ emit_long(0);)       // marker for disassembly
+  NOT_PRODUCT(__ emit_int32(0);)       // marker for disassembly
   return entry;
 }
 
@@ -232,7 +232,7 @@
   }
   __ retl();                          // return from interpreter activation
   __ delayed()->nop();                // schedule this better
-  NOT_PRODUCT(__ emit_long(0);)       // marker for disassembly
+  NOT_PRODUCT(__ emit_int32(0);)       // marker for disassembly
   return entry;
 }
 
@@ -1473,7 +1473,7 @@
     __ brx(Assembler::equal, false, Assembler::pt, skip);         \
     __ delayed()->nop();                                          \
     __ breakpoint_trap();                                         \
-    __ emit_long(marker);                                         \
+    __ emit_int32(marker);                                         \
     __ bind(skip);                                                \
   }
 #else
--- a/src/cpu/sparc/vm/macroAssembler_sparc.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/sparc/vm/macroAssembler_sparc.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1224,7 +1224,7 @@
   // Relocation with special format (see relocInfo_sparc.hpp).
   relocate(rspec, 1);
   // Assembler::sethi(0x3fffff, d);
-  emit_long( op(branch_op) | rd(d) | op2(sethi_op2) | hi22(0x3fffff) );
+  emit_int32( op(branch_op) | rd(d) | op2(sethi_op2) | hi22(0x3fffff) );
   // Don't add relocation for 'add'. Do patching during 'sethi' processing.
   add(d, 0x3ff, d);
 
@@ -1240,7 +1240,7 @@
   // Relocation with special format (see relocInfo_sparc.hpp).
   relocate(rspec, 1);
   // Assembler::sethi(encoded_k, d);
-  emit_long( op(branch_op) | rd(d) | op2(sethi_op2) | hi22(encoded_k) );
+  emit_int32( op(branch_op) | rd(d) | op2(sethi_op2) | hi22(encoded_k) );
   // Don't add relocation for 'add'. Do patching during 'sethi' processing.
   add(d, low10(encoded_k), d);
 
--- a/src/cpu/sparc/vm/templateInterpreter_sparc.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/sparc/vm/templateInterpreter_sparc.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -259,7 +259,7 @@
   }
   __ ret();                           // return from interpreter activation
   __ delayed()->restore(I5_savedSP, G0, SP);  // remove interpreter frame
-  NOT_PRODUCT(__ emit_long(0);)       // marker for disassembly
+  NOT_PRODUCT(__ emit_int32(0);)       // marker for disassembly
   return entry;
 }
 
--- a/src/cpu/x86/vm/assembler_x86.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/assembler_x86.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -182,7 +182,7 @@
 // make this go away someday
 void Assembler::emit_data(jint data, relocInfo::relocType rtype, int format) {
   if (rtype == relocInfo::none)
-        emit_long(data);
+        emit_int32(data);
   else  emit_data(data, Relocation::spec_simple(rtype), format);
 }
 
@@ -202,7 +202,7 @@
     else
       code_section()->relocate(inst_mark(), rspec, format);
   }
-  emit_long(data);
+  emit_int32(data);
 }
 
 static int encode(Register r) {
@@ -243,7 +243,7 @@
   } else {
     emit_int8(op1);
     emit_int8(op2 | encode(dst));
-    emit_long(imm32);
+    emit_int32(imm32);
   }
 }
 
@@ -254,7 +254,7 @@
   assert((op1 & 0x02) == 0, "sign-extension bit should not be set");
   emit_int8(op1);
   emit_int8(op2 | encode(dst));
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 // immediate-to-memory forms
@@ -268,7 +268,7 @@
   } else {
     emit_int8(op1);
     emit_operand(rm, adr, 4);
-    emit_long(imm32);
+    emit_int32(imm32);
   }
 }
 
@@ -976,7 +976,7 @@
   emit_int8(0x1F);
   emit_int8((unsigned char)0x80);
                    // emit_rm(cbuf, 0x2, EAX_enc, EAX_enc);
-  emit_long(0);    // 32-bits offset (4 bytes)
+  emit_int32(0);   // 32-bits offset (4 bytes)
 }
 
 void Assembler::addr_nop_8() {
@@ -987,7 +987,7 @@
   emit_int8((unsigned char)0x84);
                    // emit_rm(cbuf, 0x2, EAX_enc, 0x4);
   emit_int8(0x00); // emit_rm(cbuf, 0x0, EAX_enc, EAX_enc);
-  emit_long(0);    // 32-bits offset (4 bytes)
+  emit_int32(0);   // 32-bits offset (4 bytes)
 }
 
 void Assembler::addsd(XMMRegister dst, XMMRegister src) {
@@ -1076,7 +1076,7 @@
   prefix(dst);
   emit_int8((unsigned char)0x81);
   emit_operand(rsp, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::andl(Register dst, int32_t imm32) {
@@ -1204,7 +1204,7 @@
   prefix(dst);
   emit_int8((unsigned char)0x81);
   emit_operand(rdi, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::cmpl(Register dst, int32_t imm32) {
@@ -1408,7 +1408,7 @@
   } else {
     emit_int8(0x69);
     emit_int8((unsigned char)(0xC0 | encode));
-    emit_long(value);
+    emit_int32(value);
   }
 }
 
@@ -1440,7 +1440,7 @@
              "must be 32bit offset (call4)");
       emit_int8(0x0F);
       emit_int8((unsigned char)(0x80 | cc));
-      emit_long(offs - long_size);
+      emit_int32(offs - long_size);
     }
   } else {
     // Note: could eliminate cond. jumps to this jump if condition
@@ -1450,7 +1450,7 @@
     L.add_patch_at(code(), locator());
     emit_int8(0x0F);
     emit_int8((unsigned char)(0x80 | cc));
-    emit_long(0);
+    emit_int32(0);
   }
 }
 
@@ -1498,7 +1498,7 @@
       emit_int8((offs - short_size) & 0xFF);
     } else {
       emit_int8((unsigned char)0xE9);
-      emit_long(offs - long_size);
+      emit_int32(offs - long_size);
     }
   } else {
     // By default, forward jumps are always 32-bit displacements, since
@@ -1508,7 +1508,7 @@
     InstructionMark im(this);
     L.add_patch_at(code(), locator());
     emit_int8((unsigned char)0xE9);
-    emit_long(0);
+    emit_int32(0);
   }
 }
 
@@ -1732,7 +1732,7 @@
 void Assembler::movl(Register dst, int32_t imm32) {
   int encode = prefix_and_encode(dst->encoding());
   emit_int8((unsigned char)(0xB8 | encode));
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::movl(Register dst, Register src) {
@@ -1753,7 +1753,7 @@
   prefix(dst);
   emit_int8((unsigned char)0xC7);
   emit_operand(rax, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::movl(Address dst, Register src) {
@@ -2468,6 +2468,26 @@
   emit_int8((unsigned char)(0xC0 | encode));
 }
 
+void Assembler::vptest(XMMRegister dst, Address src) {
+  assert(VM_Version::supports_avx(), "");
+  InstructionMark im(this);
+  bool vector256 = true;
+  assert(dst != xnoreg, "sanity");
+  int dst_enc = dst->encoding();
+  // swap src<->dst for encoding
+  vex_prefix(src, dst_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F_38, false, vector256);
+  emit_int8(0x17);
+  emit_operand(dst, src);
+}
+
+void Assembler::vptest(XMMRegister dst, XMMRegister src) {
+  assert(VM_Version::supports_avx(), "");
+  bool vector256 = true;
+  int encode = vex_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, vector256, VEX_OPCODE_0F_38);
+  emit_int8(0x17);
+  emit_int8((unsigned char)(0xC0 | encode));
+}
+
 void Assembler::punpcklbw(XMMRegister dst, Address src) {
   NOT_LP64(assert(VM_Version::supports_sse2(), ""));
   assert((UseAVX > 0), "SSE mode requires address alignment 16 bytes");
@@ -2499,7 +2519,7 @@
   // in 64bits we push 64bits onto the stack but only
   // take a 32bit immediate
   emit_int8(0x68);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::push(Register src) {
@@ -2544,12 +2564,18 @@
   emit_int8((unsigned char)0xA5);
 }
 
+// sets rcx bytes with rax, value at [edi]
+void Assembler::rep_stosb() {
+  emit_int8((unsigned char)0xF3); // REP
+  LP64_ONLY(prefix(REX_W));
+  emit_int8((unsigned char)0xAA); // STOSB
+}
+
 // sets rcx pointer sized words with rax, value at [edi]
 // generic
-void Assembler::rep_set() { // rep_set
-  emit_int8((unsigned char)0xF3);
-  // STOSQ
-  LP64_ONLY(prefix(REX_W));
+void Assembler::rep_stos() {
+  emit_int8((unsigned char)0xF3); // REP
+  LP64_ONLY(prefix(REX_W));       // LP64:STOSQ, LP32:STOSD
   emit_int8((unsigned char)0xAB);
 }
 
@@ -2785,7 +2811,7 @@
     emit_int8((unsigned char)0xF7);
     emit_int8((unsigned char)(0xC0 | encode));
   }
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::testl(Register dst, Register src) {
@@ -3650,6 +3676,15 @@
   emit_int8(0x01);
 }
 
+// duplicate 4-bytes integer data from src into 8 locations in dest
+void Assembler::vpbroadcastd(XMMRegister dst, XMMRegister src) {
+  assert(VM_Version::supports_avx2(), "");
+  bool vector256 = true;
+  int encode = vex_prefix_and_encode(dst, xnoreg, src, VEX_SIMD_66, vector256, VEX_OPCODE_0F_38);
+  emit_int8(0x58);
+  emit_int8((unsigned char)(0xC0 | encode));
+}
+
 void Assembler::vzeroupper() {
   assert(VM_Version::supports_avx(), "");
   (void)vex_prefix_and_encode(xmm0, xmm0, xmm0, VEX_SIMD_NONE);
@@ -4720,7 +4755,7 @@
   prefixq(dst);
   emit_int8((unsigned char)0x81);
   emit_operand(rsp, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::andq(Register dst, int32_t imm32) {
@@ -4793,7 +4828,7 @@
   prefixq(dst);
   emit_int8((unsigned char)0x81);
   emit_operand(rdi, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::cmpq(Register dst, int32_t imm32) {
@@ -4932,7 +4967,7 @@
   } else {
     emit_int8(0x69);
     emit_int8((unsigned char)(0xC0 | encode));
-    emit_long(value);
+    emit_int32(value);
   }
 }
 
@@ -5085,7 +5120,7 @@
   InstructionMark im(this);
   int encode = prefixq_and_encode(dst->encoding());
   emit_int8((unsigned char)(0xC7 | encode));
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::movslq(Address dst, int32_t imm32) {
@@ -5094,7 +5129,7 @@
   prefixq(dst);
   emit_int8((unsigned char)0xC7);
   emit_operand(rax, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::movslq(Register dst, Address src) {
@@ -5172,7 +5207,7 @@
   prefixq(dst);
   emit_int8((unsigned char)0x81);
   emit_operand(rcx, dst, 4);
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::orq(Register dst, int32_t imm32) {
@@ -5407,7 +5442,7 @@
     emit_int8((unsigned char)0xF7);
     emit_int8((unsigned char)(0xC0 | encode));
   }
-  emit_long(imm32);
+  emit_int32(imm32);
 }
 
 void Assembler::testq(Register dst, Register src) {
--- a/src/cpu/x86/vm/assembler_x86.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/assembler_x86.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -832,7 +832,8 @@
 
   // These do register sized moves/scans
   void rep_mov();
-  void rep_set();
+  void rep_stos();
+  void rep_stosb();
   void repne_scan();
 #ifdef _LP64
   void repne_scanl();
@@ -1443,9 +1444,12 @@
   // Shift Right by bytes Logical DoubleQuadword Immediate
   void psrldq(XMMRegister dst, int shift);
 
-  // Logical Compare Double Quadword
+  // Logical Compare 128bit
   void ptest(XMMRegister dst, XMMRegister src);
   void ptest(XMMRegister dst, Address src);
+  // Logical Compare 256bit
+  void vptest(XMMRegister dst, XMMRegister src);
+  void vptest(XMMRegister dst, Address src);
 
   // Interleave Low Bytes
   void punpcklbw(XMMRegister dst, XMMRegister src);
@@ -1753,6 +1757,9 @@
   void vextractf128h(Address dst, XMMRegister src);
   void vextracti128h(Address dst, XMMRegister src);
 
+  // duplicate 4-bytes integer data from src into 8 locations in dest
+  void vpbroadcastd(XMMRegister dst, XMMRegister src);
+
   // AVX instruction which is used to clear upper 128 bits of YMM registers and
   // to avoid transaction penalty between AVX and SSE states. There is no
   // penalty if legacy SSE instructions are encoded using VEX prefix because
--- a/src/cpu/x86/vm/globals_x86.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/globals_x86.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -120,6 +120,9 @@
   product(bool, UseUnalignedLoadStores, false,                              \
           "Use SSE2 MOVDQU instruction for Arraycopy")                      \
                                                                             \
+  product(bool, UseFastStosb, false,                                        \
+          "Use fast-string operation for zeroing: rep stosb")               \
+                                                                            \
   /* assembler */                                                           \
   product(bool, Use486InstrsOnly, false,                                    \
           "Use 80486 Compliant instruction subset")                         \
--- a/src/cpu/x86/vm/macroAssembler_x86.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/macroAssembler_x86.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -2540,7 +2540,7 @@
       // 0000 1111 1000 tttn #32-bit disp
       emit_int8(0x0F);
       emit_int8((unsigned char)(0x80 | cc));
-      emit_long(offs - long_size);
+      emit_int32(offs - long_size);
     }
   } else {
 #ifdef ASSERT
@@ -5224,6 +5224,22 @@
 
 }
 
+void MacroAssembler::clear_mem(Register base, Register cnt, Register tmp) {
+  // cnt - number of qwords (8-byte words).
+  // base - start address, qword aligned.
+  assert(base==rdi, "base register must be edi for rep stos");
+  assert(tmp==rax,   "tmp register must be eax for rep stos");
+  assert(cnt==rcx,   "cnt register must be ecx for rep stos");
+
+  xorptr(tmp, tmp);
+  if (UseFastStosb) {
+    shlptr(cnt,3); // convert to number of bytes
+    rep_stosb();
+  } else {
+    NOT_LP64(shlptr(cnt,1);) // convert to number of dwords for 32-bit VM
+    rep_stos();
+  }
+}
 
 // IndexOf for constant substrings with size >= 8 chars
 // which don't need to be loaded through stack.
@@ -5659,42 +5675,114 @@
   testl(cnt2, cnt2);
   jcc(Assembler::zero, LENGTH_DIFF_LABEL);
 
-  // Load first characters
+  // Compare first characters
   load_unsigned_short(result, Address(str1, 0));
   load_unsigned_short(cnt1, Address(str2, 0));
-
-  // Compare first characters
   subl(result, cnt1);
   jcc(Assembler::notZero,  POP_LABEL);
-  decrementl(cnt2);
-  jcc(Assembler::zero, LENGTH_DIFF_LABEL);
-
-  {
-    // Check after comparing first character to see if strings are equivalent
-    Label LSkip2;
-    // Check if the strings start at same location
-    cmpptr(str1, str2);
-    jccb(Assembler::notEqual, LSkip2);
-
-    // Check if the length difference is zero (from stack)
-    cmpl(Address(rsp, 0), 0x0);
-    jcc(Assembler::equal,  LENGTH_DIFF_LABEL);
-
-    // Strings might not be equivalent
-    bind(LSkip2);
-  }
+  cmpl(cnt2, 1);
+  jcc(Assembler::equal, LENGTH_DIFF_LABEL);
+
+  // Check if the strings start at the same location.
+  cmpptr(str1, str2);
+  jcc(Assembler::equal, LENGTH_DIFF_LABEL);
 
   Address::ScaleFactor scale = Address::times_2;
   int stride = 8;
 
-  // Advance to next element
-  addptr(str1, 16/stride);
-  addptr(str2, 16/stride);
-
-  if (UseSSE42Intrinsics) {
+  if (UseAVX >= 2) {
+    Label COMPARE_WIDE_VECTORS, VECTOR_NOT_EQUAL, COMPARE_WIDE_TAIL, COMPARE_SMALL_STR;
+    Label COMPARE_WIDE_VECTORS_LOOP, COMPARE_16_CHARS, COMPARE_INDEX_CHAR;
+    Label COMPARE_TAIL_LONG;
+    int pcmpmask = 0x19;
+
+    // Setup to compare 16-chars (32-bytes) vectors,
+    // start from first character again because it has aligned address.
+    int stride2 = 16;
+    int adr_stride  = stride  << scale;
+    int adr_stride2 = stride2 << scale;
+
+    assert(result == rax && cnt2 == rdx && cnt1 == rcx, "pcmpestri");
+    // rax and rdx are used by pcmpestri as elements counters
+    movl(result, cnt2);
+    andl(cnt2, ~(stride2-1));   // cnt2 holds the vector count
+    jcc(Assembler::zero, COMPARE_TAIL_LONG);
+
+    // fast path : compare first 2 8-char vectors.
+    bind(COMPARE_16_CHARS);
+    movdqu(vec1, Address(str1, 0));
+    pcmpestri(vec1, Address(str2, 0), pcmpmask);
+    jccb(Assembler::below, COMPARE_INDEX_CHAR);
+
+    movdqu(vec1, Address(str1, adr_stride));
+    pcmpestri(vec1, Address(str2, adr_stride), pcmpmask);
+    jccb(Assembler::aboveEqual, COMPARE_WIDE_VECTORS);
+    addl(cnt1, stride);
+
+    // Compare the characters at index in cnt1
+    bind(COMPARE_INDEX_CHAR); //cnt1 has the offset of the mismatching character
+    load_unsigned_short(result, Address(str1, cnt1, scale));
+    load_unsigned_short(cnt2, Address(str2, cnt1, scale));
+    subl(result, cnt2);
+    jmp(POP_LABEL);
+
+    // Setup the registers to start vector comparison loop
+    bind(COMPARE_WIDE_VECTORS);
+    lea(str1, Address(str1, result, scale));
+    lea(str2, Address(str2, result, scale));
+    subl(result, stride2);
+    subl(cnt2, stride2);
+    jccb(Assembler::zero, COMPARE_WIDE_TAIL);
+    negptr(result);
+
+    //  In a loop, compare 16-chars (32-bytes) at once using (vpxor+vptest)
+    bind(COMPARE_WIDE_VECTORS_LOOP);
+    vmovdqu(vec1, Address(str1, result, scale));
+    vpxor(vec1, Address(str2, result, scale));
+    vptest(vec1, vec1);
+    jccb(Assembler::notZero, VECTOR_NOT_EQUAL);
+    addptr(result, stride2);
+    subl(cnt2, stride2);
+    jccb(Assembler::notZero, COMPARE_WIDE_VECTORS_LOOP);
+
+    // compare wide vectors tail
+    bind(COMPARE_WIDE_TAIL);
+    testptr(result, result);
+    jccb(Assembler::zero, LENGTH_DIFF_LABEL);
+
+    movl(result, stride2);
+    movl(cnt2, result);
+    negptr(result);
+    jmpb(COMPARE_WIDE_VECTORS_LOOP);
+
+    // Identifies the mismatching (higher or lower)16-bytes in the 32-byte vectors.
+    bind(VECTOR_NOT_EQUAL);
+    lea(str1, Address(str1, result, scale));
+    lea(str2, Address(str2, result, scale));
+    jmp(COMPARE_16_CHARS);
+
+    // Compare tail chars, length between 1 to 15 chars
+    bind(COMPARE_TAIL_LONG);
+    movl(cnt2, result);
+    cmpl(cnt2, stride);
+    jccb(Assembler::less, COMPARE_SMALL_STR);
+
+    movdqu(vec1, Address(str1, 0));
+    pcmpestri(vec1, Address(str2, 0), pcmpmask);
+    jcc(Assembler::below, COMPARE_INDEX_CHAR);
+    subptr(cnt2, stride);
+    jccb(Assembler::zero, LENGTH_DIFF_LABEL);
+    lea(str1, Address(str1, result, scale));
+    lea(str2, Address(str2, result, scale));
+    negptr(cnt2);
+    jmpb(WHILE_HEAD_LABEL);
+
+    bind(COMPARE_SMALL_STR);
+  } else if (UseSSE42Intrinsics) {
     Label COMPARE_WIDE_VECTORS, VECTOR_NOT_EQUAL, COMPARE_TAIL;
     int pcmpmask = 0x19;
-    // Setup to compare 16-byte vectors
+    // Setup to compare 8-char (16-byte) vectors,
+    // start from first character again because it has aligned address.
     movl(result, cnt2);
     andl(cnt2, ~(stride - 1));   // cnt2 holds the vector count
     jccb(Assembler::zero, COMPARE_TAIL);
@@ -5726,7 +5814,7 @@
     jccb(Assembler::notZero, COMPARE_WIDE_VECTORS);
 
     // compare wide vectors tail
-    testl(result, result);
+    testptr(result, result);
     jccb(Assembler::zero, LENGTH_DIFF_LABEL);
 
     movl(cnt2, stride);
@@ -5738,21 +5826,20 @@
 
     // Mismatched characters in the vectors
     bind(VECTOR_NOT_EQUAL);
-    addptr(result, cnt1);
-    movptr(cnt2, result);
-    load_unsigned_short(result, Address(str1, cnt2, scale));
-    load_unsigned_short(cnt1, Address(str2, cnt2, scale));
-    subl(result, cnt1);
+    addptr(cnt1, result);
+    load_unsigned_short(result, Address(str1, cnt1, scale));
+    load_unsigned_short(cnt2, Address(str2, cnt1, scale));
+    subl(result, cnt2);
     jmpb(POP_LABEL);
 
     bind(COMPARE_TAIL); // limit is zero
     movl(cnt2, result);
     // Fallthru to tail compare
   }
-
   // Shift str2 and str1 to the end of the arrays, negate min
-  lea(str1, Address(str1, cnt2, scale, 0));
-  lea(str2, Address(str2, cnt2, scale, 0));
+  lea(str1, Address(str1, cnt2, scale));
+  lea(str2, Address(str2, cnt2, scale));
+  decrementl(cnt2);  // first character was compared already
   negptr(cnt2);
 
   // Compare the rest of the elements
@@ -5817,7 +5904,44 @@
   shll(limit, 1);      // byte count != 0
   movl(result, limit); // copy
 
-  if (UseSSE42Intrinsics) {
+  if (UseAVX >= 2) {
+    // With AVX2, use 32-byte vector compare
+    Label COMPARE_WIDE_VECTORS, COMPARE_TAIL;
+
+    // Compare 32-byte vectors
+    andl(result, 0x0000001e);  //   tail count (in bytes)
+    andl(limit, 0xffffffe0);   // vector count (in bytes)
+    jccb(Assembler::zero, COMPARE_TAIL);
+
+    lea(ary1, Address(ary1, limit, Address::times_1));
+    lea(ary2, Address(ary2, limit, Address::times_1));
+    negptr(limit);
+
+    bind(COMPARE_WIDE_VECTORS);
+    vmovdqu(vec1, Address(ary1, limit, Address::times_1));
+    vmovdqu(vec2, Address(ary2, limit, Address::times_1));
+    vpxor(vec1, vec2);
+
+    vptest(vec1, vec1);
+    jccb(Assembler::notZero, FALSE_LABEL);
+    addptr(limit, 32);
+    jcc(Assembler::notZero, COMPARE_WIDE_VECTORS);
+
+    testl(result, result);
+    jccb(Assembler::zero, TRUE_LABEL);
+
+    vmovdqu(vec1, Address(ary1, result, Address::times_1, -32));
+    vmovdqu(vec2, Address(ary2, result, Address::times_1, -32));
+    vpxor(vec1, vec2);
+
+    vptest(vec1, vec1);
+    jccb(Assembler::notZero, FALSE_LABEL);
+    jmpb(TRUE_LABEL);
+
+    bind(COMPARE_TAIL); // limit is zero
+    movl(limit, result);
+    // Fallthru to tail compare
+  } else if (UseSSE42Intrinsics) {
     // With SSE4.2, use double quad vector compare
     Label COMPARE_WIDE_VECTORS, COMPARE_TAIL;
 
@@ -5995,29 +6119,53 @@
     {
       assert( UseSSE >= 2, "supported cpu only" );
       Label L_fill_32_bytes_loop, L_check_fill_8_bytes, L_fill_8_bytes_loop, L_fill_8_bytes;
-      // Fill 32-byte chunks
       movdl(xtmp, value);
-      pshufd(xtmp, xtmp, 0);
-
-      subl(count, 8 << shift);
-      jcc(Assembler::less, L_check_fill_8_bytes);
-      align(16);
-
-      BIND(L_fill_32_bytes_loop);
-
-      if (UseUnalignedLoadStores) {
-        movdqu(Address(to, 0), xtmp);
-        movdqu(Address(to, 16), xtmp);
+      if (UseAVX >= 2 && UseUnalignedLoadStores) {
+        // Fill 64-byte chunks
+        Label L_fill_64_bytes_loop, L_check_fill_32_bytes;
+        vpbroadcastd(xtmp, xtmp);
+
+        subl(count, 16 << shift);
+        jcc(Assembler::less, L_check_fill_32_bytes);
+        align(16);
+
+        BIND(L_fill_64_bytes_loop);
+        vmovdqu(Address(to, 0), xtmp);
+        vmovdqu(Address(to, 32), xtmp);
+        addptr(to, 64);
+        subl(count, 16 << shift);
+        jcc(Assembler::greaterEqual, L_fill_64_bytes_loop);
+
+        BIND(L_check_fill_32_bytes);
+        addl(count, 8 << shift);
+        jccb(Assembler::less, L_check_fill_8_bytes);
+        vmovdqu(Address(to, 0), xtmp);
+        addptr(to, 32);
+        subl(count, 8 << shift);
       } else {
-        movq(Address(to, 0), xtmp);
-        movq(Address(to, 8), xtmp);
-        movq(Address(to, 16), xtmp);
-        movq(Address(to, 24), xtmp);
+        // Fill 32-byte chunks
+        pshufd(xtmp, xtmp, 0);
+
+        subl(count, 8 << shift);
+        jcc(Assembler::less, L_check_fill_8_bytes);
+        align(16);
+
+        BIND(L_fill_32_bytes_loop);
+
+        if (UseUnalignedLoadStores) {
+          movdqu(Address(to, 0), xtmp);
+          movdqu(Address(to, 16), xtmp);
+        } else {
+          movq(Address(to, 0), xtmp);
+          movq(Address(to, 8), xtmp);
+          movq(Address(to, 16), xtmp);
+          movq(Address(to, 24), xtmp);
+        }
+
+        addptr(to, 32);
+        subl(count, 8 << shift);
+        jcc(Assembler::greaterEqual, L_fill_32_bytes_loop);
       }
-
-      addptr(to, 32);
-      subl(count, 8 << shift);
-      jcc(Assembler::greaterEqual, L_fill_32_bytes_loop);
       BIND(L_check_fill_8_bytes);
       addl(count, 8 << shift);
       jccb(Assembler::zero, L_exit);
--- a/src/cpu/x86/vm/macroAssembler_x86.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/macroAssembler_x86.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1011,6 +1011,10 @@
       Assembler::vxorpd(dst, nds, src, vector256);
   }
 
+  // Simple version for AVX2 256bit vectors
+  void vpxor(XMMRegister dst, XMMRegister src) { Assembler::vpxor(dst, dst, src, true); }
+  void vpxor(XMMRegister dst, Address src) { Assembler::vpxor(dst, dst, src, true); }
+
   // Move packed integer values from low 128 bit to hign 128 bit in 256 bit vector.
   void vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src) {
     if (UseAVX > 1) // vinserti128h is available only in AVX2
@@ -1096,6 +1100,9 @@
   // C2 compiled method's prolog code.
   void verified_entry(int framesize, bool stack_bang, bool fp_mode_24b);
 
+  // clear memory of size 'cnt' qwords, starting at 'base'.
+  void clear_mem(Register base, Register cnt, Register rtmp);
+
   // IndexOf strings.
   // Small strings are loaded through stack if they cross page boundary.
   void string_indexof(Register str1, Register str2,
--- a/src/cpu/x86/vm/stubGenerator_x86_32.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/stubGenerator_x86_32.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -796,16 +796,22 @@
     __ align(OptoLoopAlignment);
   __ BIND(L_copy_64_bytes_loop);
 
-    if(UseUnalignedLoadStores) {
-      __ movdqu(xmm0, Address(from, 0));
-      __ movdqu(Address(from, to_from, Address::times_1, 0), xmm0);
-      __ movdqu(xmm1, Address(from, 16));
-      __ movdqu(Address(from, to_from, Address::times_1, 16), xmm1);
-      __ movdqu(xmm2, Address(from, 32));
-      __ movdqu(Address(from, to_from, Address::times_1, 32), xmm2);
-      __ movdqu(xmm3, Address(from, 48));
-      __ movdqu(Address(from, to_from, Address::times_1, 48), xmm3);
-
+    if (UseUnalignedLoadStores) {
+      if (UseAVX >= 2) {
+        __ vmovdqu(xmm0, Address(from,  0));
+        __ vmovdqu(Address(from, to_from, Address::times_1,  0), xmm0);
+        __ vmovdqu(xmm1, Address(from, 32));
+        __ vmovdqu(Address(from, to_from, Address::times_1, 32), xmm1);
+      } else {
+        __ movdqu(xmm0, Address(from, 0));
+        __ movdqu(Address(from, to_from, Address::times_1, 0), xmm0);
+        __ movdqu(xmm1, Address(from, 16));
+        __ movdqu(Address(from, to_from, Address::times_1, 16), xmm1);
+        __ movdqu(xmm2, Address(from, 32));
+        __ movdqu(Address(from, to_from, Address::times_1, 32), xmm2);
+        __ movdqu(xmm3, Address(from, 48));
+        __ movdqu(Address(from, to_from, Address::times_1, 48), xmm3);
+      }
     } else {
       __ movq(xmm0, Address(from, 0));
       __ movq(Address(from, to_from, Address::times_1, 0), xmm0);
--- a/src/cpu/x86/vm/stubGenerator_x86_64.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/stubGenerator_x86_64.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1286,23 +1286,54 @@
   //   end_to       - destination array end address
   //   qword_count  - 64-bits element count, negative
   //   to           - scratch
-  //   L_copy_32_bytes - entry label
+  //   L_copy_bytes - entry label
   //   L_copy_8_bytes  - exit  label
   //
-  void copy_32_bytes_forward(Register end_from, Register end_to,
+  void copy_bytes_forward(Register end_from, Register end_to,
                              Register qword_count, Register to,
-                             Label& L_copy_32_bytes, Label& L_copy_8_bytes) {
+                             Label& L_copy_bytes, Label& L_copy_8_bytes) {
     DEBUG_ONLY(__ stop("enter at entry label, not here"));
     Label L_loop;
     __ align(OptoLoopAlignment);
-  __ BIND(L_loop);
-    if(UseUnalignedLoadStores) {
-      __ movdqu(xmm0, Address(end_from, qword_count, Address::times_8, -24));
-      __ movdqu(Address(end_to, qword_count, Address::times_8, -24), xmm0);
-      __ movdqu(xmm1, Address(end_from, qword_count, Address::times_8, - 8));
-      __ movdqu(Address(end_to, qword_count, Address::times_8, - 8), xmm1);
-
+    if (UseUnalignedLoadStores) {
+      Label L_end;
+      // Copy 64-bytes per iteration
+      __ BIND(L_loop);
+      if (UseAVX >= 2) {
+        __ vmovdqu(xmm0, Address(end_from, qword_count, Address::times_8, -56));
+        __ vmovdqu(Address(end_to, qword_count, Address::times_8, -56), xmm0);
+        __ vmovdqu(xmm1, Address(end_from, qword_count, Address::times_8, -24));
+        __ vmovdqu(Address(end_to, qword_count, Address::times_8, -24), xmm1);
+      } else {
+        __ movdqu(xmm0, Address(end_from, qword_count, Address::times_8, -56));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, -56), xmm0);
+        __ movdqu(xmm1, Address(end_from, qword_count, Address::times_8, -40));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, -40), xmm1);
+        __ movdqu(xmm2, Address(end_from, qword_count, Address::times_8, -24));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, -24), xmm2);
+        __ movdqu(xmm3, Address(end_from, qword_count, Address::times_8, - 8));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, - 8), xmm3);
+      }
+      __ BIND(L_copy_bytes);
+      __ addptr(qword_count, 8);
+      __ jcc(Assembler::lessEqual, L_loop);
+      __ subptr(qword_count, 4);  // sub(8) and add(4)
+      __ jccb(Assembler::greater, L_end);
+      // Copy trailing 32 bytes
+      if (UseAVX >= 2) {
+        __ vmovdqu(xmm0, Address(end_from, qword_count, Address::times_8, -24));
+        __ vmovdqu(Address(end_to, qword_count, Address::times_8, -24), xmm0);
+      } else {
+        __ movdqu(xmm0, Address(end_from, qword_count, Address::times_8, -24));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, -24), xmm0);
+        __ movdqu(xmm1, Address(end_from, qword_count, Address::times_8, - 8));
+        __ movdqu(Address(end_to, qword_count, Address::times_8, - 8), xmm1);
+      }
+      __ addptr(qword_count, 4);
+      __ BIND(L_end);
     } else {
+      // Copy 32-bytes per iteration
+      __ BIND(L_loop);
       __ movq(to, Address(end_from, qword_count, Address::times_8, -24));
       __ movq(Address(end_to, qword_count, Address::times_8, -24), to);
       __ movq(to, Address(end_from, qword_count, Address::times_8, -16));
@@ -1311,15 +1342,15 @@
       __ movq(Address(end_to, qword_count, Address::times_8, - 8), to);
       __ movq(to, Address(end_from, qword_count, Address::times_8, - 0));
       __ movq(Address(end_to, qword_count, Address::times_8, - 0), to);
+
+      __ BIND(L_copy_bytes);
+      __ addptr(qword_count, 4);
+      __ jcc(Assembler::lessEqual, L_loop);
     }
-  __ BIND(L_copy_32_bytes);
-    __ addptr(qword_count, 4);
-    __ jcc(Assembler::lessEqual, L_loop);
     __ subptr(qword_count, 4);
     __ jcc(Assembler::less, L_copy_8_bytes); // Copy trailing qwords
   }
 
-
   // Copy big chunks backward
   //
   // Inputs:
@@ -1327,23 +1358,55 @@
   //   dest         - destination array address
   //   qword_count  - 64-bits element count
   //   to           - scratch
-  //   L_copy_32_bytes - entry label
+  //   L_copy_bytes - entry label
   //   L_copy_8_bytes  - exit  label
   //
-  void copy_32_bytes_backward(Register from, Register dest,
+  void copy_bytes_backward(Register from, Register dest,
                               Register qword_count, Register to,
-                              Label& L_copy_32_bytes, Label& L_copy_8_bytes) {
+                              Label& L_copy_bytes, Label& L_copy_8_bytes) {
     DEBUG_ONLY(__ stop("enter at entry label, not here"));
     Label L_loop;
     __ align(OptoLoopAlignment);
-  __ BIND(L_loop);
-    if(UseUnalignedLoadStores) {
-      __ movdqu(xmm0, Address(from, qword_count, Address::times_8, 16));
-      __ movdqu(Address(dest, qword_count, Address::times_8, 16), xmm0);
-      __ movdqu(xmm1, Address(from, qword_count, Address::times_8,  0));
-      __ movdqu(Address(dest, qword_count, Address::times_8,  0), xmm1);
-
+    if (UseUnalignedLoadStores) {
+      Label L_end;
+      // Copy 64-bytes per iteration
+      __ BIND(L_loop);
+      if (UseAVX >= 2) {
+        __ vmovdqu(xmm0, Address(from, qword_count, Address::times_8, 32));
+        __ vmovdqu(Address(dest, qword_count, Address::times_8, 32), xmm0);
+        __ vmovdqu(xmm1, Address(from, qword_count, Address::times_8,  0));
+        __ vmovdqu(Address(dest, qword_count, Address::times_8,  0), xmm1);
+      } else {
+        __ movdqu(xmm0, Address(from, qword_count, Address::times_8, 48));
+        __ movdqu(Address(dest, qword_count, Address::times_8, 48), xmm0);
+        __ movdqu(xmm1, Address(from, qword_count, Address::times_8, 32));
+        __ movdqu(Address(dest, qword_count, Address::times_8, 32), xmm1);
+        __ movdqu(xmm2, Address(from, qword_count, Address::times_8, 16));
+        __ movdqu(Address(dest, qword_count, Address::times_8, 16), xmm2);
+        __ movdqu(xmm3, Address(from, qword_count, Address::times_8,  0));
+        __ movdqu(Address(dest, qword_count, Address::times_8,  0), xmm3);
+      }
+      __ BIND(L_copy_bytes);
+      __ subptr(qword_count, 8);
+      __ jcc(Assembler::greaterEqual, L_loop);
+
+      __ addptr(qword_count, 4);  // add(8) and sub(4)
+      __ jccb(Assembler::less, L_end);
+      // Copy trailing 32 bytes
+      if (UseAVX >= 2) {
+        __ vmovdqu(xmm0, Address(from, qword_count, Address::times_8, 0));
+        __ vmovdqu(Address(dest, qword_count, Address::times_8, 0), xmm0);
+      } else {
+        __ movdqu(xmm0, Address(from, qword_count, Address::times_8, 16));
+        __ movdqu(Address(dest, qword_count, Address::times_8, 16), xmm0);
+        __ movdqu(xmm1, Address(from, qword_count, Address::times_8,  0));
+        __ movdqu(Address(dest, qword_count, Address::times_8,  0), xmm1);
+      }
+      __ subptr(qword_count, 4);
+      __ BIND(L_end);
     } else {
+      // Copy 32-bytes per iteration
+      __ BIND(L_loop);
       __ movq(to, Address(from, qword_count, Address::times_8, 24));
       __ movq(Address(dest, qword_count, Address::times_8, 24), to);
       __ movq(to, Address(from, qword_count, Address::times_8, 16));
@@ -1352,10 +1415,11 @@
       __ movq(Address(dest, qword_count, Address::times_8,  8), to);
       __ movq(to, Address(from, qword_count, Address::times_8,  0));
       __ movq(Address(dest, qword_count, Address::times_8,  0), to);
+
+      __ BIND(L_copy_bytes);
+      __ subptr(qword_count, 4);
+      __ jcc(Assembler::greaterEqual, L_loop);
     }
-  __ BIND(L_copy_32_bytes);
-    __ subptr(qword_count, 4);
-    __ jcc(Assembler::greaterEqual, L_loop);
     __ addptr(qword_count, 4);
     __ jcc(Assembler::greater, L_copy_8_bytes); // Copy trailing qwords
   }
@@ -1385,7 +1449,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_4_bytes, L_copy_2_bytes;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_4_bytes, L_copy_2_bytes;
     Label L_copy_byte, L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
@@ -1417,7 +1481,7 @@
     __ lea(end_from, Address(from, qword_count, Address::times_8, -8));
     __ lea(end_to,   Address(to,   qword_count, Address::times_8, -8));
     __ negptr(qword_count); // make the count negative
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1460,8 +1524,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_forward(end_from, end_to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_forward(end_from, end_to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
     __ jmp(L_copy_4_bytes);
 
     return start;
@@ -1488,7 +1552,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_4_bytes, L_copy_2_bytes;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_4_bytes, L_copy_2_bytes;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register count       = rdx;  // elements count
@@ -1531,10 +1595,10 @@
     // Check for and copy trailing dword
   __ BIND(L_copy_4_bytes);
     __ testl(byte_count, 4);
-    __ jcc(Assembler::zero, L_copy_32_bytes);
+    __ jcc(Assembler::zero, L_copy_bytes);
     __ movl(rax, Address(from, qword_count, Address::times_8));
     __ movl(Address(to, qword_count, Address::times_8), rax);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1549,8 +1613,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_backward(from, to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_backward(from, to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
 
     restore_arg_regs();
     inc_counter_np(SharedRuntime::_jbyte_array_copy_ctr); // Update counter after rscratch1 is free
@@ -1585,7 +1649,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_4_bytes,L_copy_2_bytes,L_exit;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_4_bytes,L_copy_2_bytes,L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register count       = rdx;  // elements count
@@ -1616,7 +1680,7 @@
     __ lea(end_from, Address(from, qword_count, Address::times_8, -8));
     __ lea(end_to,   Address(to,   qword_count, Address::times_8, -8));
     __ negptr(qword_count);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1652,8 +1716,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_forward(end_from, end_to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_forward(end_from, end_to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
     __ jmp(L_copy_4_bytes);
 
     return start;
@@ -1700,7 +1764,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_4_bytes;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_4_bytes;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register count       = rdx;  // elements count
@@ -1735,10 +1799,10 @@
     // Check for and copy trailing dword
   __ BIND(L_copy_4_bytes);
     __ testl(word_count, 2);
-    __ jcc(Assembler::zero, L_copy_32_bytes);
+    __ jcc(Assembler::zero, L_copy_bytes);
     __ movl(rax, Address(from, qword_count, Address::times_8));
     __ movl(Address(to, qword_count, Address::times_8), rax);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1753,8 +1817,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_backward(from, to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_backward(from, to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
 
     restore_arg_regs();
     inc_counter_np(SharedRuntime::_jshort_array_copy_ctr); // Update counter after rscratch1 is free
@@ -1790,7 +1854,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_4_bytes, L_exit;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_4_bytes, L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register count       = rdx;  // elements count
@@ -1826,7 +1890,7 @@
     __ lea(end_from, Address(from, qword_count, Address::times_8, -8));
     __ lea(end_to,   Address(to,   qword_count, Address::times_8, -8));
     __ negptr(qword_count);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1853,8 +1917,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy 32-bytes chunks
-    copy_32_bytes_forward(end_from, end_to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_forward(end_from, end_to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
     __ jmp(L_copy_4_bytes);
 
     return start;
@@ -1882,7 +1946,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_copy_2_bytes, L_exit;
+    Label L_copy_bytes, L_copy_8_bytes, L_copy_2_bytes, L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register count       = rdx;  // elements count
@@ -1916,10 +1980,10 @@
 
     // Check for and copy trailing dword
     __ testl(dword_count, 1);
-    __ jcc(Assembler::zero, L_copy_32_bytes);
+    __ jcc(Assembler::zero, L_copy_bytes);
     __ movl(rax, Address(from, dword_count, Address::times_4, -4));
     __ movl(Address(to, dword_count, Address::times_4, -4), rax);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -1937,8 +2001,8 @@
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_backward(from, to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_backward(from, to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
 
    __ bind(L_exit);
      if (is_oop) {
@@ -1976,7 +2040,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_exit;
+    Label L_copy_bytes, L_copy_8_bytes, L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register qword_count = rdx;  // elements count
@@ -2008,7 +2072,7 @@
     __ lea(end_from, Address(from, qword_count, Address::times_8, -8));
     __ lea(end_to,   Address(to,   qword_count, Address::times_8, -8));
     __ negptr(qword_count);
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -2027,8 +2091,8 @@
       __ ret(0);
     }
 
-    // Copy 64-byte chunks
-    copy_32_bytes_forward(end_from, end_to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_forward(end_from, end_to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
 
     if (is_oop) {
     __ BIND(L_exit);
@@ -2065,7 +2129,7 @@
     StubCodeMark mark(this, "StubRoutines", name);
     address start = __ pc();
 
-    Label L_copy_32_bytes, L_copy_8_bytes, L_exit;
+    Label L_copy_bytes, L_copy_8_bytes, L_exit;
     const Register from        = rdi;  // source array address
     const Register to          = rsi;  // destination array address
     const Register qword_count = rdx;  // elements count
@@ -2091,7 +2155,7 @@
       gen_write_ref_array_pre_barrier(to, saved_count, dest_uninitialized);
     }
 
-    __ jmp(L_copy_32_bytes);
+    __ jmp(L_copy_bytes);
 
     // Copy trailing qwords
   __ BIND(L_copy_8_bytes);
@@ -2110,8 +2174,8 @@
       __ ret(0);
     }
 
-    // Copy in 32-bytes chunks
-    copy_32_bytes_backward(from, to, qword_count, rax, L_copy_32_bytes, L_copy_8_bytes);
+    // Copy in multi-bytes chunks
+    copy_bytes_backward(from, to, qword_count, rax, L_copy_bytes, L_copy_8_bytes);
 
     if (is_oop) {
     __ BIND(L_exit);
--- a/src/cpu/x86/vm/vm_version_x86.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/vm_version_x86.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -429,7 +429,7 @@
   }
 
   char buf[256];
-  jio_snprintf(buf, sizeof(buf), "(%u cores per cpu, %u threads per core) family %d model %d stepping %d%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
+  jio_snprintf(buf, sizeof(buf), "(%u cores per cpu, %u threads per core) family %d model %d stepping %d%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
                cores_per_cpu(), threads_per_core(),
                cpu_family(), _model, _stepping,
                (supports_cmov() ? ", cmov" : ""),
@@ -446,6 +446,7 @@
                (supports_avx()    ? ", avx" : ""),
                (supports_avx2()   ? ", avx2" : ""),
                (supports_aes()    ? ", aes" : ""),
+               (supports_erms()   ? ", erms" : ""),
                (supports_mmx_ext() ? ", mmxext" : ""),
                (supports_3dnow_prefetch() ? ", 3dnowpref" : ""),
                (supports_lzcnt()   ? ", lzcnt": ""),
@@ -671,6 +672,16 @@
     FLAG_SET_DEFAULT(UsePopCountInstruction, false);
   }
 
+  // Use fast-string operations if available.
+  if (supports_erms()) {
+    if (FLAG_IS_DEFAULT(UseFastStosb)) {
+      UseFastStosb = true;
+    }
+  } else if (UseFastStosb) {
+    warning("fast-string operations are not available on this CPU");
+    FLAG_SET_DEFAULT(UseFastStosb, false);
+  }
+
 #ifdef COMPILER2
   if (FLAG_IS_DEFAULT(AlignVector)) {
     // Modern processors allow misaligned memory operations for vectors.
--- a/src/cpu/x86/vm/vm_version_x86.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/vm_version_x86.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -204,7 +204,8 @@
                    avx2 : 1,
                         : 2,
                    bmi2 : 1,
-                        : 23;
+                   erms : 1,
+                        : 22;
     } bits;
   };
 
@@ -247,7 +248,8 @@
     CPU_TSCINV = (1 << 16),
     CPU_AVX    = (1 << 17),
     CPU_AVX2   = (1 << 18),
-    CPU_AES    = (1 << 19)
+    CPU_AES    = (1 << 19),
+    CPU_ERMS   = (1 << 20) // enhanced 'rep movsb/stosb' instructions
   } cpuFeatureFlags;
 
   enum {
@@ -425,6 +427,8 @@
       result |= CPU_TSCINV;
     if (_cpuid_info.std_cpuid1_ecx.bits.aes != 0)
       result |= CPU_AES;
+    if (_cpuid_info.sef_cpuid7_ebx.bits.erms != 0)
+      result |= CPU_ERMS;
 
     // AMD features.
     if (is_amd()) {
@@ -489,7 +493,7 @@
     return (_cpuid_info.std_max_function >= 0xB) &&
            // eax[4:0] | ebx[0:15] == 0 indicates invalid topology level.
            // Some cpus have max cpuid >= 0xB but do not support processor topology.
-           ((_cpuid_info.tpl_cpuidB0_eax & 0x1f | _cpuid_info.tpl_cpuidB0_ebx.bits.logical_cpus) != 0);
+           (((_cpuid_info.tpl_cpuidB0_eax & 0x1f) | _cpuid_info.tpl_cpuidB0_ebx.bits.logical_cpus) != 0);
   }
 
   static uint cores_per_cpu()  {
@@ -550,6 +554,7 @@
   static bool supports_avx2()     { return (_cpuFeatures & CPU_AVX2) != 0; }
   static bool supports_tsc()      { return (_cpuFeatures & CPU_TSC)    != 0; }
   static bool supports_aes()      { return (_cpuFeatures & CPU_AES) != 0; }
+  static bool supports_erms()     { return (_cpuFeatures & CPU_ERMS) != 0; }
 
   // Intel features
   static bool is_intel_family_core() { return is_intel() &&
--- a/src/cpu/x86/vm/x86_32.ad	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/x86_32.ad	Fri Jan 11 01:43:09 2013 -0800
@@ -11572,15 +11572,28 @@
 // =======================================================================
 // fast clearing of an array
 instruct rep_stos(eCXRegI cnt, eDIRegP base, eAXRegI zero, Universe dummy, eFlagsReg cr) %{
+  predicate(!UseFastStosb);
   match(Set dummy (ClearArray cnt base));
   effect(USE_KILL cnt, USE_KILL base, KILL zero, KILL cr);
-  format %{ "SHL    ECX,1\t# Convert doublewords to words\n\t"
-            "XOR    EAX,EAX\n\t"
+  format %{ "XOR    EAX,EAX\t# ClearArray:\n\t"
+            "SHL    ECX,1\t# Convert doublewords to words\n\t"
             "REP STOS\t# store EAX into [EDI++] while ECX--" %}
-  opcode(0,0x4);
-  ins_encode( Opcode(0xD1), RegOpc(ECX),
-              OpcRegReg(0x33,EAX,EAX),
-              Opcode(0xF3), Opcode(0xAB) );
+  ins_encode %{ 
+    __ clear_mem($base$$Register, $cnt$$Register, $zero$$Register);
+  %}
+  ins_pipe( pipe_slow );
+%}
+
+instruct rep_fast_stosb(eCXRegI cnt, eDIRegP base, eAXRegI zero, Universe dummy, eFlagsReg cr) %{
+  predicate(UseFastStosb);
+  match(Set dummy (ClearArray cnt base));
+  effect(USE_KILL cnt, USE_KILL base, KILL zero, KILL cr);
+  format %{ "XOR    EAX,EAX\t# ClearArray:\n\t"
+            "SHL    ECX,3\t# Convert doublewords to bytes\n\t"
+            "REP STOSB\t# store EAX into [EDI++] while ECX--" %}
+  ins_encode %{ 
+    __ clear_mem($base$$Register, $cnt$$Register, $zero$$Register);
+  %}
   ins_pipe( pipe_slow );
 %}
 
--- a/src/cpu/x86/vm/x86_64.ad	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/cpu/x86/vm/x86_64.ad	Fri Jan 11 01:43:09 2013 -0800
@@ -10374,16 +10374,33 @@
 instruct rep_stos(rcx_RegL cnt, rdi_RegP base, rax_RegI zero, Universe dummy,
                   rFlagsReg cr)
 %{
+  predicate(!UseFastStosb);
   match(Set dummy (ClearArray cnt base));
   effect(USE_KILL cnt, USE_KILL base, KILL zero, KILL cr);
 
-  format %{ "xorl    rax, rax\t# ClearArray:\n\t"
-            "rep stosq\t# Store rax to *rdi++ while rcx--" %}
-  ins_encode(opc_reg_reg(0x33, RAX, RAX), // xorl %eax, %eax
-             Opcode(0xF3), Opcode(0x48), Opcode(0xAB)); // rep REX_W stos
+  format %{ "xorq    rax, rax\t# ClearArray:\n\t"
+            "rep     stosq\t# Store rax to *rdi++ while rcx--" %}
+  ins_encode %{ 
+    __ clear_mem($base$$Register, $cnt$$Register, $zero$$Register);
+  %}
   ins_pipe(pipe_slow);
 %}
 
+instruct rep_fast_stosb(rcx_RegL cnt, rdi_RegP base, rax_RegI zero, Universe dummy,
+                        rFlagsReg cr)
+%{
+  predicate(UseFastStosb);
+  match(Set dummy (ClearArray cnt base));
+  effect(USE_KILL cnt, USE_KILL base, KILL zero, KILL cr);
+  format %{ "xorq    rax, rax\t# ClearArray:\n\t"
+            "shlq    rcx,3\t# Convert doublewords to bytes\n\t"
+            "rep     stosb\t# Store rax to *rdi++ while rcx--" %}
+  ins_encode %{ 
+    __ clear_mem($base$$Register, $cnt$$Register, $zero$$Register);
+  %}
+  ins_pipe( pipe_slow );
+%}
+
 instruct string_compare(rdi_RegP str1, rcx_RegI cnt1, rsi_RegP str2, rdx_RegI cnt2,
                         rax_RegI result, regD tmp1, rFlagsReg cr)
 %{
--- a/src/os/bsd/vm/os_bsd.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/os/bsd/vm/os_bsd.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -298,12 +298,12 @@
 
   // The next steps are taken in the product version:
   //
-  // Obtain the JAVA_HOME value from the location of libjvm[_g].so.
+  // Obtain the JAVA_HOME value from the location of libjvm.so.
   // This library should be located at:
-  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm[_g].so.
+  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm.so.
   //
   // If "/jre/lib/" appears at the right place in the path, then we
-  // assume libjvm[_g].so is installed in a JDK and we use this path.
+  // assume libjvm.so is installed in a JDK and we use this path.
   //
   // Otherwise exit with message: "Could not create the Java virtual machine."
   //
@@ -313,9 +313,9 @@
   // instead of exit check for $JAVA_HOME environment variable.
   //
   // If it is defined and we are able to locate $JAVA_HOME/jre/lib/<arch>,
-  // then we append a fake suffix "hotspot/libjvm[_g].so" to this path so
-  // it looks like libjvm[_g].so is installed there
-  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm[_g].so.
+  // then we append a fake suffix "hotspot/libjvm.so" to this path so
+  // it looks like libjvm.so is installed there
+  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm.so.
   //
   // Otherwise exit.
   //
@@ -1228,7 +1228,7 @@
   return getcwd(buf, buflen);
 }
 
-// check if addr is inside libjvm[_g].so
+// check if addr is inside libjvm.so
 bool os::address_is_in_vm(address addr) {
   static address libjvm_base_addr;
   Dl_info dlinfo;
@@ -1689,7 +1689,7 @@
 
 static char saved_jvm_path[MAXPATHLEN] = {0};
 
-// Find the full path to the current module, libjvm or libjvm_g
+// Find the full path to the current module, libjvm
 void os::jvm_path(char *buf, jint buflen) {
   // Error checking.
   if (buflen < MAXPATHLEN) {
@@ -1732,10 +1732,9 @@
         char* jrelib_p;
         int len;
 
-        // Check the current module name "libjvm" or "libjvm_g".
+        // Check the current module name "libjvm"
         p = strrchr(buf, '/');
         assert(strstr(p, "/libjvm") == p, "invalid library name");
-        p = strstr(p, "_g") ? "_g" : "";
 
         rp = realpath(java_home_var, buf);
         if (rp == NULL)
@@ -1764,11 +1763,9 @@
         // to complete the path to JVM being overridden.  Otherwise fallback
         // to the path to the current library.
         if (0 == access(buf, F_OK)) {
-          // Use current module name "libjvm[_g]" instead of
-          // "libjvm"debug_only("_g")"" since for fastdebug version
-          // we should have "libjvm" but debug_only("_g") adds "_g"!
+          // Use current module name "libjvm"
           len = strlen(buf);
-          snprintf(buf + len, buflen-len, "/libjvm%s%s", p, JNI_LIB_SUFFIX);
+          snprintf(buf + len, buflen-len, "/libjvm%s", JNI_LIB_SUFFIX);
         } else {
           // Fall back to path of current library
           rp = realpath(dli_fname, buf);
--- a/src/os/linux/vm/os_linux.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/os/linux/vm/os_linux.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -321,12 +321,12 @@
 
   // The next steps are taken in the product version:
   //
-  // Obtain the JAVA_HOME value from the location of libjvm[_g].so.
+  // Obtain the JAVA_HOME value from the location of libjvm.so.
   // This library should be located at:
-  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm[_g].so.
+  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm.so.
   //
   // If "/jre/lib/" appears at the right place in the path, then we
-  // assume libjvm[_g].so is installed in a JDK and we use this path.
+  // assume libjvm.so is installed in a JDK and we use this path.
   //
   // Otherwise exit with message: "Could not create the Java virtual machine."
   //
@@ -336,9 +336,9 @@
   // instead of exit check for $JAVA_HOME environment variable.
   //
   // If it is defined and we are able to locate $JAVA_HOME/jre/lib/<arch>,
-  // then we append a fake suffix "hotspot/libjvm[_g].so" to this path so
-  // it looks like libjvm[_g].so is installed there
-  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm[_g].so.
+  // then we append a fake suffix "hotspot/libjvm.so" to this path so
+  // it looks like libjvm.so is installed there
+  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm.so.
   //
   // Otherwise exit.
   //
@@ -1679,7 +1679,7 @@
   return getcwd(buf, buflen);
 }
 
-// check if addr is inside libjvm[_g].so
+// check if addr is inside libjvm.so
 bool os::address_is_in_vm(address addr) {
   static address libjvm_base_addr;
   Dl_info dlinfo;
@@ -2180,7 +2180,7 @@
 
 static char saved_jvm_path[MAXPATHLEN] = {0};
 
-// Find the full path to the current module, libjvm.so or libjvm_g.so
+// Find the full path to the current module, libjvm.so
 void os::jvm_path(char *buf, jint buflen) {
   // Error checking.
   if (buflen < MAXPATHLEN) {
@@ -2223,10 +2223,9 @@
         char* jrelib_p;
         int len;
 
-        // Check the current module name "libjvm.so" or "libjvm_g.so".
+        // Check the current module name "libjvm.so".
         p = strrchr(buf, '/');
         assert(strstr(p, "/libjvm") == p, "invalid library name");
-        p = strstr(p, "_g") ? "_g" : "";
 
         rp = realpath(java_home_var, buf);
         if (rp == NULL)
@@ -2242,11 +2241,9 @@
         }
 
         if (0 == access(buf, F_OK)) {
-          // Use current module name "libjvm[_g].so" instead of
-          // "libjvm"debug_only("_g")".so" since for fastdebug version
-          // we should have "libjvm.so" but debug_only("_g") adds "_g"!
+          // Use current module name "libjvm.so"
           len = strlen(buf);
-          snprintf(buf + len, buflen-len, "/hotspot/libjvm%s.so", p);
+          snprintf(buf + len, buflen-len, "/hotspot/libjvm.so");
         } else {
           // Go back to path of .so
           rp = realpath(dli_fname, buf);
--- a/src/os/solaris/vm/os_solaris.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/os/solaris/vm/os_solaris.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -734,12 +734,12 @@
 
   // The next steps are taken in the product version:
   //
-  // Obtain the JAVA_HOME value from the location of libjvm[_g].so.
+  // Obtain the JAVA_HOME value from the location of libjvm.so.
   // This library should be located at:
-  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm[_g].so.
+  // <JAVA_HOME>/jre/lib/<arch>/{client|server}/libjvm.so.
   //
   // If "/jre/lib/" appears at the right place in the path, then we
-  // assume libjvm[_g].so is installed in a JDK and we use this path.
+  // assume libjvm.so is installed in a JDK and we use this path.
   //
   // Otherwise exit with message: "Could not create the Java virtual machine."
   //
@@ -749,9 +749,9 @@
   // instead of exit check for $JAVA_HOME environment variable.
   //
   // If it is defined and we are able to locate $JAVA_HOME/jre/lib/<arch>,
-  // then we append a fake suffix "hotspot/libjvm[_g].so" to this path so
-  // it looks like libjvm[_g].so is installed there
-  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm[_g].so.
+  // then we append a fake suffix "hotspot/libjvm.so" to this path so
+  // it looks like libjvm.so is installed there
+  // <JAVA_HOME>/jre/lib/<arch>/hotspot/libjvm.so.
   //
   // Otherwise exit.
   //
@@ -1934,7 +1934,7 @@
   return getcwd(buf, buflen);
 }
 
-// check if addr is inside libjvm[_g].so
+// check if addr is inside libjvm.so
 bool os::address_is_in_vm(address addr) {
   static address libjvm_base_addr;
   Dl_info dlinfo;
@@ -2474,7 +2474,7 @@
 
 static char saved_jvm_path[MAXPATHLEN] = { 0 };
 
-// Find the full path to the current module, libjvm.so or libjvm_g.so
+// Find the full path to the current module, libjvm.so
 void os::jvm_path(char *buf, jint buflen) {
   // Error checking.
   if (buflen < MAXPATHLEN) {
@@ -2522,10 +2522,9 @@
           strcpy(cpu_arch, "amd64");
         }
 #endif
-        // Check the current module name "libjvm.so" or "libjvm_g.so".
+        // Check the current module name "libjvm.so".
         p = strrchr(buf, '/');
         assert(strstr(p, "/libjvm") == p, "invalid library name");
-        p = strstr(p, "_g") ? "_g" : "";
 
         realpath(java_home_var, buf);
         // determine if this is a legacy image or modules image
@@ -2538,11 +2537,9 @@
         }
 
         if (0 == access(buf, F_OK)) {
-          // Use current module name "libjvm[_g].so" instead of
-          // "libjvm"debug_only("_g")".so" since for fastdebug version
-          // we should have "libjvm.so" but debug_only("_g") adds "_g"!
+          // Use current module name "libjvm.so"
           len = strlen(buf);
-          snprintf(buf + len, buflen-len, "/hotspot/libjvm%s.so", p);
+          snprintf(buf + len, buflen-len, "/hotspot/libjvm.so");
         } else {
           // Go back to path of .so
           realpath((char *)dlinfo.dli_fname, buf);
--- a/src/os/windows/vm/os_windows.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/os/windows/vm/os_windows.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -182,7 +182,7 @@
 
       if (!getenv("_ALT_JAVA_HOME_DIR", home_dir, MAX_PATH)) {
           os::jvm_path(home_dir, sizeof(home_dir));
-          // Found the full path to jvm[_g].dll.
+          // Found the full path to jvm.dll.
           // Now cut the path to <java_home>/jre if we can.
           *(strrchr(home_dir, '\\')) = '\0';  /* get rid of \jvm.dll */
           pslash = strrchr(home_dir, '\\');
@@ -1715,7 +1715,7 @@
 
 static char saved_jvm_path[MAX_PATH] = {0};
 
-// Find the full path to the current module, jvm.dll or jvm_g.dll
+// Find the full path to the current module, jvm.dll
 void os::jvm_path(char *buf, jint buflen) {
   // Error checking.
   if (buflen < MAX_PATH) {
--- a/src/share/tools/ProjectCreator/ProjectCreator.java	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/tools/ProjectCreator/ProjectCreator.java	Fri Jan 11 01:43:09 2013 -0800
@@ -36,7 +36,7 @@
             + "into .dsp file, substituting for path given in "
             + "-sourceBase. Example: HotSpotWorkSpace>");
       System.err.println("  -dllLoc <path to directory in which to put "
-            + "jvm.dll and jvm_g.dll; no trailing slash>");
+            + "jvm.dll; no trailing slash>");
       System.err.println("  If any of the above are specified, "
             + "they must all be.");
       System.err.println("  Additional, optional arguments, which can be "
--- a/src/share/vm/asm/assembler.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/asm/assembler.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -216,8 +216,6 @@
   bool isByte(int x) const             { return 0 <= x && x < 0x100; }
   bool isShiftCount(int x) const       { return 0 <= x && x < 32; }
 
-  void emit_long(jint x) { emit_int32(x); }  // deprecated
-
   // Instruction boundaries (required when emitting relocatable values).
   class InstructionMark: public StackObj {
    private:
--- a/src/share/vm/c1/c1_LIR.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/c1/c1_LIR.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -2259,7 +2259,7 @@
   typedef enum { inputMode, firstMode = inputMode, tempMode, outputMode, numModes, invalidMode = -1 } OprMode;
 
   enum {
-    maxNumberOfOperands = 16,
+    maxNumberOfOperands = 20,
     maxNumberOfInfos = 4
   };
 
--- a/src/share/vm/classfile/classFileParser.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/classFileParser.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -906,6 +906,7 @@
                                              bool* is_synthetic_addr,
                                              u2* generic_signature_index_addr,
                                              AnnotationArray** field_annotations,
+                                             AnnotationArray** field_type_annotations,
                                              ClassFileParser::FieldAnnotationCollector* parsed_annotations,
                                              TRAPS) {
   ClassFileStream* cfs = stream();
@@ -917,6 +918,10 @@
   int runtime_visible_annotations_length = 0;
   u1* runtime_invisible_annotations = NULL;
   int runtime_invisible_annotations_length = 0;
+  u1* runtime_visible_type_annotations = NULL;
+  int runtime_visible_type_annotations_length = 0;
+  u1* runtime_invisible_type_annotations = NULL;
+  int runtime_invisible_type_annotations_length = 0;
   while (attributes_count--) {
     cfs->guarantee_more(6, CHECK);  // attribute_name_index, attribute_length
     u2 attribute_name_index = cfs->get_u2_fast();
@@ -971,6 +976,16 @@
         runtime_invisible_annotations = cfs->get_u1_buffer();
         assert(runtime_invisible_annotations != NULL, "null invisible annotations");
         cfs->skip_u1(runtime_invisible_annotations_length, CHECK);
+      } else if (attribute_name == vmSymbols::tag_runtime_visible_type_annotations()) {
+        runtime_visible_type_annotations_length = attribute_length;
+        runtime_visible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_visible_type_annotations != NULL, "null visible type annotations");
+        cfs->skip_u1(runtime_visible_type_annotations_length, CHECK);
+      } else if (PreserveAllAnnotations && attribute_name == vmSymbols::tag_runtime_invisible_type_annotations()) {
+        runtime_invisible_type_annotations_length = attribute_length;
+        runtime_invisible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_invisible_type_annotations != NULL, "null invisible type annotations");
+        cfs->skip_u1(runtime_invisible_type_annotations_length, CHECK);
       } else {
         cfs->skip_u1(attribute_length, CHECK);  // Skip unknown attributes
       }
@@ -988,6 +1003,12 @@
                                             runtime_invisible_annotations,
                                             runtime_invisible_annotations_length,
                                             CHECK);
+  *field_type_annotations = assemble_annotations(loader_data,
+                                            runtime_visible_type_annotations,
+                                            runtime_visible_type_annotations_length,
+                                            runtime_invisible_type_annotations,
+                                            runtime_invisible_type_annotations_length,
+                                            CHECK);
   return;
 }
 
@@ -1084,6 +1105,7 @@
                                          bool is_interface,
                                          FieldAllocationCount *fac,
                                          Array<AnnotationArray*>** fields_annotations,
+                                         Array<AnnotationArray*>** fields_type_annotations,
                                          u2* java_fields_count_ptr, TRAPS) {
   ClassFileStream* cfs = stream();
   cfs->guarantee_more(2, CHECK_NULL);  // length
@@ -1119,6 +1141,7 @@
              THREAD, u2, total_fields * (FieldInfo::field_slots + 1));
 
   AnnotationArray* field_annotations = NULL;
+  AnnotationArray* field_type_annotations = NULL;
   // The generic signature slots start after all other fields' data.
   int generic_signature_slot = total_fields * FieldInfo::field_slots;
   int num_generic_signature = 0;
@@ -1160,7 +1183,7 @@
                              cp, attributes_count, is_static, signature_index,
                              &constantvalue_index, &is_synthetic,
                              &generic_signature_index, &field_annotations,
-                             &parsed_annotations,
+                             &field_type_annotations, &parsed_annotations,
                              CHECK_NULL);
       if (field_annotations != NULL) {
         if (*fields_annotations == NULL) {
@@ -1170,6 +1193,14 @@
         }
         (*fields_annotations)->at_put(n, field_annotations);
       }
+      if (field_type_annotations != NULL) {
+        if (*fields_type_annotations == NULL) {
+          *fields_type_annotations = MetadataFactory::new_array<AnnotationArray*>(
+                                                  loader_data, length, NULL,
+                                                  CHECK_NULL);
+        }
+        (*fields_type_annotations)->at_put(n, field_type_annotations);
+      }
       if (is_synthetic) {
         access_flags.set_is_synthetic();
       }
@@ -1831,6 +1862,7 @@
                                            AnnotationArray** method_annotations,
                                            AnnotationArray** method_parameter_annotations,
                                            AnnotationArray** method_default_annotations,
+                                           AnnotationArray** method_type_annotations,
                                            TRAPS) {
   ClassFileStream* cfs = stream();
   methodHandle nullHandle;
@@ -1903,6 +1935,8 @@
   u2** localvariable_table_start;
   u2* localvariable_type_table_length;
   u2** localvariable_type_table_start;
+  u2 method_parameters_length = 0;
+  u1* method_parameters_data = NULL;
   bool parsed_code_attribute = false;
   bool parsed_checked_exceptions_attribute = false;
   bool parsed_stackmap_attribute = false;
@@ -1918,6 +1952,10 @@
   int runtime_visible_parameter_annotations_length = 0;
   u1* runtime_invisible_parameter_annotations = NULL;
   int runtime_invisible_parameter_annotations_length = 0;
+  u1* runtime_visible_type_annotations = NULL;
+  int runtime_visible_type_annotations_length = 0;
+  u1* runtime_invisible_type_annotations = NULL;
+  int runtime_invisible_type_annotations_length = 0;
   u1* annotation_default = NULL;
   int annotation_default_length = 0;
 
@@ -2108,6 +2146,14 @@
             parse_checked_exceptions(&checked_exceptions_length,
                                      method_attribute_length,
                                      cp, CHECK_(nullHandle));
+    } else if (method_attribute_name == vmSymbols::tag_method_parameters()) {
+      method_parameters_length = cfs->get_u1_fast();
+      method_parameters_data = cfs->get_u1_buffer();
+      cfs->skip_u2_fast(method_parameters_length);
+      cfs->skip_u4_fast(method_parameters_length);
+      // ignore this attribute if it cannot be reflected
+      if (!SystemDictionary::Parameter_klass_loaded())
+        method_parameters_length = 0;
     } else if (method_attribute_name == vmSymbols::tag_synthetic()) {
       if (method_attribute_length != 0) {
         classfile_parse_error(
@@ -2159,6 +2205,17 @@
         annotation_default = cfs->get_u1_buffer();
         assert(annotation_default != NULL, "null annotation default");
         cfs->skip_u1(annotation_default_length, CHECK_(nullHandle));
+      } else if (method_attribute_name == vmSymbols::tag_runtime_visible_type_annotations()) {
+        runtime_visible_type_annotations_length = method_attribute_length;
+        runtime_visible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_visible_type_annotations != NULL, "null visible type annotations");
+        // No need for the VM to parse Type annotations
+        cfs->skip_u1(runtime_visible_type_annotations_length, CHECK_(nullHandle));
+      } else if (PreserveAllAnnotations && method_attribute_name == vmSymbols::tag_runtime_invisible_type_annotations()) {
+        runtime_invisible_type_annotations_length = method_attribute_length;
+        runtime_invisible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_invisible_type_annotations != NULL, "null invisible type annotations");
+        cfs->skip_u1(runtime_invisible_type_annotations_length, CHECK_(nullHandle));
       } else {
         // Skip unknown attributes
         cfs->skip_u1(method_attribute_length, CHECK_(nullHandle));
@@ -2184,7 +2241,8 @@
   Method* m = Method::allocate(
       loader_data, code_length, access_flags, linenumber_table_length,
       total_lvt_length, exception_table_length, checked_exceptions_length,
-      generic_signature_index, ConstMethod::NORMAL, CHECK_(nullHandle));
+      method_parameters_length, generic_signature_index,
+      ConstMethod::NORMAL, CHECK_(nullHandle));
 
   ClassLoadingService::add_class_method_size(m->size()*HeapWordSize);
 
@@ -2232,6 +2290,18 @@
                              exception_table_start, size);
   }
 
+  // Copy method parameters
+  if (method_parameters_length > 0) {
+    MethodParametersElement* elem = m->constMethod()->method_parameters_start();
+    for(int i = 0; i < method_parameters_length; i++) {
+      elem[i].name_cp_index =
+        Bytes::get_Java_u2(method_parameters_data);
+      method_parameters_data += 2;
+      elem[i].flags = Bytes::get_Java_u4(method_parameters_data);
+      method_parameters_data += 4;
+    }
+  }
+
   // Copy checked exceptions
   if (checked_exceptions_length > 0) {
     int size = checked_exceptions_length * sizeof(CheckedExceptionElement) / sizeof(u2);
@@ -2333,6 +2403,12 @@
                                                      NULL,
                                                      0,
                                                      CHECK_(nullHandle));
+  *method_type_annotations = assemble_annotations(loader_data,
+                                                  runtime_visible_type_annotations,
+                                                  runtime_visible_type_annotations_length,
+                                                  runtime_invisible_type_annotations,
+                                                  runtime_invisible_type_annotations_length,
+                                                  CHECK_(nullHandle));
 
   if (name == vmSymbols::finalize_method_name() &&
       signature == vmSymbols::void_method_signature()) {
@@ -2364,12 +2440,14 @@
                                                Array<AnnotationArray*>** methods_annotations,
                                                Array<AnnotationArray*>** methods_parameter_annotations,
                                                Array<AnnotationArray*>** methods_default_annotations,
+                                               Array<AnnotationArray*>** methods_type_annotations,
                                                bool* has_default_methods,
                                                TRAPS) {
   ClassFileStream* cfs = stream();
   AnnotationArray* method_annotations = NULL;
   AnnotationArray* method_parameter_annotations = NULL;
   AnnotationArray* method_default_annotations = NULL;
+  AnnotationArray* method_type_annotations = NULL;
   cfs->guarantee_more(2, CHECK_NULL);  // length
   u2 length = cfs->get_u2_fast();
   if (length == 0) {
@@ -2386,6 +2464,7 @@
                                          &method_annotations,
                                          &method_parameter_annotations,
                                          &method_default_annotations,
+                                         &method_type_annotations,
                                          CHECK_NULL);
 
       if (method->is_final()) {
@@ -2411,7 +2490,13 @@
             MetadataFactory::new_array<AnnotationArray*>(loader_data, length, NULL, CHECK_NULL);
       }
       (*methods_default_annotations)->at_put(index, method_default_annotations);
+      if (*methods_type_annotations == NULL) {
+        *methods_type_annotations =
+             MetadataFactory::new_array<AnnotationArray*>(loader_data, length, NULL, CHECK_NULL);
+      }
+      (*methods_type_annotations)->at_put(index, method_type_annotations);
     }
+
     if (_need_verify && length > 1) {
       // Check duplicated methods
       ResourceMark rm(THREAD);
@@ -2445,6 +2530,7 @@
                                           Array<AnnotationArray*>* methods_annotations,
                                           Array<AnnotationArray*>* methods_parameter_annotations,
                                           Array<AnnotationArray*>* methods_default_annotations,
+                                          Array<AnnotationArray*>* methods_type_annotations,
                                               TRAPS) {
   int length = methods->length();
   // If JVMTI original method ordering or sharing is enabled we have to
@@ -2463,7 +2549,8 @@
   // Note that the ordering is not alphabetical, see Symbol::fast_compare
   Method::sort_methods(methods, methods_annotations,
                        methods_parameter_annotations,
-                       methods_default_annotations);
+                       methods_default_annotations,
+                       methods_type_annotations);
 
   // If JVMTI original method ordering or sharing is enabled construct int
   // array remembering the original ordering
@@ -2728,6 +2815,10 @@
   int runtime_visible_annotations_length = 0;
   u1* runtime_invisible_annotations = NULL;
   int runtime_invisible_annotations_length = 0;
+  u1* runtime_visible_type_annotations = NULL;
+  int runtime_visible_type_annotations_length = 0;
+  u1* runtime_invisible_type_annotations = NULL;
+  int runtime_invisible_type_annotations_length = 0;
   u1* inner_classes_attribute_start = NULL;
   u4  inner_classes_attribute_length = 0;
   u2  enclosing_method_class_index = 0;
@@ -2834,6 +2925,17 @@
           classfile_parse_error("Multiple BootstrapMethods attributes in class file %s", CHECK);
         parsed_bootstrap_methods_attribute = true;
         parse_classfile_bootstrap_methods_attribute(loader_data, cp, attribute_length, CHECK);
+      } else if (tag == vmSymbols::tag_runtime_visible_type_annotations()) {
+        runtime_visible_type_annotations_length = attribute_length;
+        runtime_visible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_visible_type_annotations != NULL, "null visible type annotations");
+        // No need for the VM to parse Type annotations
+        cfs->skip_u1(runtime_visible_type_annotations_length, CHECK);
+      } else if (PreserveAllAnnotations && tag == vmSymbols::tag_runtime_invisible_type_annotations()) {
+        runtime_invisible_type_annotations_length = attribute_length;
+        runtime_invisible_type_annotations = cfs->get_u1_buffer();
+        assert(runtime_invisible_type_annotations != NULL, "null invisible type annotations");
+        cfs->skip_u1(runtime_invisible_type_annotations_length, CHECK);
       } else {
         // Unknown attribute
         cfs->skip_u1(attribute_length, CHECK);
@@ -2850,6 +2952,13 @@
                                                       runtime_invisible_annotations_length,
                                                       CHECK);
   set_class_annotations(annotations);
+  AnnotationArray* type_annotations = assemble_annotations(loader_data,
+                                                           runtime_visible_type_annotations,
+                                                           runtime_visible_type_annotations_length,
+                                                           runtime_invisible_type_annotations,
+                                                           runtime_invisible_type_annotations_length,
+                                                           CHECK);
+  set_class_type_annotations(type_annotations);
 
   if (parsed_innerclasses_attribute || parsed_enclosingmethod_attribute) {
     u2 num_of_classes = parse_classfile_inner_classes_attribute(
@@ -2956,6 +3065,7 @@
                                                     TempNewSymbol& parsed_name,
                                                     bool verify,
                                                     TRAPS) {
+
   // When a retransformable agent is attached, JVMTI caches the
   // class bytes that existed before the first retransformation.
   // If RedefineClasses() was used before the retransformable
@@ -3190,7 +3300,9 @@
     // Fields (offsets are filled in later)
     FieldAllocationCount fac;
     Array<AnnotationArray*>* fields_annotations = NULL;
+    Array<AnnotationArray*>* fields_type_annotations = NULL;
     Array<u2>* fields = parse_fields(loader_data, class_name, cp, access_flags.is_interface(), &fac, &fields_annotations,
+                                          &fields_type_annotations,
                                           &java_fields_count,
                                           CHECK_(nullHandle));
     // Methods
@@ -3202,6 +3314,7 @@
     Array<AnnotationArray*>* methods_annotations = NULL;
     Array<AnnotationArray*>* methods_parameter_annotations = NULL;
     Array<AnnotationArray*>* methods_default_annotations = NULL;
+    Array<AnnotationArray*>* methods_type_annotations = NULL;
     Array<Method*>* methods = parse_methods(loader_data,
                                             cp, access_flags.is_interface(),
                                             &promoted_flags,
@@ -3209,6 +3322,7 @@
                                             &methods_annotations,
                                             &methods_parameter_annotations,
                                             &methods_default_annotations,
+                                            &methods_type_annotations,
                                             &has_default_methods,
                                             CHECK_(nullHandle));
 
@@ -3270,6 +3384,7 @@
                                                methods_annotations,
                                                methods_parameter_annotations,
                                                methods_default_annotations,
+                                               methods_type_annotations,
                                                CHECK_(nullHandle));
 
     // promote flags from parse_methods() to the klass' flags
@@ -3687,11 +3802,13 @@
     if (is_anonymous())  // I am well known to myself
       cp->klass_at_put(this_class_index, this_klass()); // eagerly resolve
 
+    // Allocate an annotation type if needed.
     if (fields_annotations != NULL ||
         methods_annotations != NULL ||
         methods_parameter_annotations != NULL ||
-        methods_default_annotations != NULL) {
-      // Allocate an annotation type if needed.
+        methods_default_annotations != NULL ||
+        fields_type_annotations != NULL ||
+        methods_type_annotations != NULL) {
       Annotations* anno = Annotations::allocate(loader_data,
                             fields_annotations, methods_annotations,
                             methods_parameter_annotations,
@@ -3701,6 +3818,16 @@
       this_klass->set_annotations(NULL);
     }
 
+    if (fields_type_annotations != NULL ||
+        methods_type_annotations != NULL) {
+      assert(this_klass->annotations() != NULL, "annotations should have been allocated");
+      Annotations* anno = Annotations::allocate(loader_data,
+                                                fields_type_annotations,
+                                                methods_type_annotations,
+                                                NULL,
+                                                NULL, CHECK_(nullHandle));
+      this_klass->annotations()->set_type_annotations(anno);
+    }
 
     this_klass->set_minor_version(minor_version);
     this_klass->set_major_version(major_version);
@@ -3725,6 +3852,7 @@
     // Fill in field values obtained by parse_classfile_attributes
     if (parsed_annotations.has_any_annotations())
       parsed_annotations.apply_to(this_klass);
+
     // Create annotations
     if (_annotations != NULL && this_klass->annotations() == NULL) {
       Annotations* anno = Annotations::allocate(loader_data, CHECK_NULL);
@@ -3732,6 +3860,19 @@
     }
     apply_parsed_class_attributes(this_klass);
 
+    // Create type annotations
+    if (_type_annotations != NULL) {
+      if (this_klass->annotations() == NULL) {
+        Annotations* anno = Annotations::allocate(loader_data, CHECK_NULL);
+        this_klass->set_annotations(anno);
+      }
+      if (this_klass->annotations()->type_annotations() == NULL) {
+        Annotations* anno = Annotations::allocate(loader_data, CHECK_NULL);
+        this_klass->annotations()->set_type_annotations(anno);
+      }
+      this_klass->annotations()->type_annotations()->set_class_annotations(_type_annotations);
+    }
+
     // Miranda methods
     if ((num_miranda_methods > 0) ||
         // if this class introduced new miranda methods or
--- a/src/share/vm/classfile/classFileParser.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/classFileParser.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -64,6 +64,7 @@
   int        _sde_length;
   Array<u2>* _inner_classes;
   AnnotationArray* _annotations;
+  AnnotationArray* _type_annotations;
 
   void set_class_synthetic_flag(bool x)           { _synthetic_flag = x; }
   void set_class_sourcefile(Symbol* x)            { _sourcefile = x; }
@@ -71,12 +72,14 @@
   void set_class_sde_buffer(char* x, int len)     { _sde_buffer = x; _sde_length = len; }
   void set_class_inner_classes(Array<u2>* x)      { _inner_classes = x; }
   void set_class_annotations(AnnotationArray* x)  { _annotations = x; }
+  void set_class_type_annotations(AnnotationArray* x)  { _type_annotations = x; }
   void init_parsed_class_attributes() {
     _synthetic_flag = false;
     _sourcefile = NULL;
     _generic_signature = NULL;
     _sde_buffer = NULL;
     _sde_length = 0;
+    _annotations = _type_annotations = NULL;
     // initialize the other flags too:
     _has_finalizer = _has_empty_finalizer = _has_vanilla_constructor = false;
     _max_bootstrap_specifier_index = -1;
@@ -163,6 +166,7 @@
                               bool* is_synthetic_addr,
                               u2* generic_signature_index_addr,
                               AnnotationArray** field_annotations,
+                              AnnotationArray** field_type_annotations,
                               FieldAnnotationCollector* parsed_annotations,
                               TRAPS);
   Array<u2>* parse_fields(ClassLoaderData* loader_data,
@@ -170,6 +174,7 @@
                           constantPoolHandle cp, bool is_interface,
                           FieldAllocationCount *fac,
                           Array<AnnotationArray*>** fields_annotations,
+                          Array<AnnotationArray*>** fields_type_annotations,
                           u2* java_fields_count_ptr, TRAPS);
 
   // Method parsing
@@ -180,6 +185,7 @@
                             AnnotationArray** method_annotations,
                             AnnotationArray** method_parameter_annotations,
                             AnnotationArray** method_default_annotations,
+                            AnnotationArray** method_type_annotations,
                             TRAPS);
   Array<Method*>* parse_methods(ClassLoaderData* loader_data,
                                 constantPoolHandle cp,
@@ -189,6 +195,7 @@
                                 Array<AnnotationArray*>** methods_annotations,
                                 Array<AnnotationArray*>** methods_parameter_annotations,
                                 Array<AnnotationArray*>** methods_default_annotations,
+                                Array<AnnotationArray*>** methods_type_annotations,
                                 bool* has_default_method,
                                 TRAPS);
   Array<int>* sort_methods(ClassLoaderData* loader_data,
@@ -196,6 +203,7 @@
                            Array<AnnotationArray*>* methods_annotations,
                            Array<AnnotationArray*>* methods_parameter_annotations,
                            Array<AnnotationArray*>* methods_default_annotations,
+                           Array<AnnotationArray*>* methods_type_annotations,
                                 TRAPS);
   u2* parse_exception_table(ClassLoaderData* loader_data,
                             u4 code_length, u4 exception_table_length,
--- a/src/share/vm/classfile/classFileStream.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/classFileStream.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -93,3 +93,10 @@
   }
   _current += length * 2;
 }
+
+void ClassFileStream::skip_u4(int length, TRAPS) {
+  if (_need_verify) {
+    guarantee_more(length * 4, CHECK);
+  }
+  _current += length * 4;
+}
--- a/src/share/vm/classfile/classFileStream.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/classFileStream.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -133,6 +133,11 @@
     _current += 2 * length;
   }
 
+  void skip_u4(int length, TRAPS);
+  void skip_u4_fast(int length) {
+    _current += 4 * length;
+  }
+
   // Tells whether eos is reached
   bool at_eos() const          { return _current == _buffer_end; }
 };
--- a/src/share/vm/classfile/classLoaderData.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/classLoaderData.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -330,10 +330,19 @@
     }
     if (this == the_null_class_loader_data()) {
       assert (class_loader() == NULL, "Must be");
-      size_t word_size = Metaspace::first_chunk_word_size();
-      set_metaspace(new Metaspace(_metaspace_lock, word_size));
+      set_metaspace(new Metaspace(_metaspace_lock, Metaspace::BootMetaspaceType));
+    } else if (is_anonymous()) {
+      if (TraceClassLoaderData && Verbose && class_loader() != NULL) {
+        tty->print_cr("is_anonymous: %s", class_loader()->klass()->internal_name());
+      }
+      set_metaspace(new Metaspace(_metaspace_lock, Metaspace::AnonymousMetaspaceType));
+    } else if (class_loader()->is_a(SystemDictionary::reflect_DelegatingClassLoader_klass())) {
+      if (TraceClassLoaderData && Verbose && class_loader() != NULL) {
+        tty->print_cr("is_reflection: %s", class_loader()->klass()->internal_name());
+      }
+      set_metaspace(new Metaspace(_metaspace_lock, Metaspace::ReflectionMetaspaceType));
     } else {
-      set_metaspace(new Metaspace(_metaspace_lock));  // default size for now.
+      set_metaspace(new Metaspace(_metaspace_lock, Metaspace::StandardMetaspaceType));
     }
   }
   return _metaspace;
@@ -672,8 +681,8 @@
          "only supported for null loader data for now");
   assert (!_shared_metaspaces_initialized, "only initialize once");
   MutexLockerEx ml(metaspace_lock(),  Mutex::_no_safepoint_check_flag);
-  _ro_metaspace = new Metaspace(_metaspace_lock, SharedReadOnlySize/wordSize);
-  _rw_metaspace = new Metaspace(_metaspace_lock, SharedReadWriteSize/wordSize);
+  _ro_metaspace = new Metaspace(_metaspace_lock, Metaspace::ROMetaspaceType);
+  _rw_metaspace = new Metaspace(_metaspace_lock, Metaspace::ReadWriteMetaspaceType);
   _shared_metaspaces_initialized = true;
 }
 
--- a/src/share/vm/classfile/defaultMethods.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/defaultMethods.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1148,7 +1148,8 @@
   int code_length = bytecodes->length();
 
   Method* m = Method::allocate(cp->pool_holder()->class_loader_data(),
-      code_length, flags, 0, 0, 0, 0, 0, mt, CHECK_NULL);
+                               code_length, flags, 0, 0, 0, 0, 0, 0,
+                               mt, CHECK_NULL);
 
   m->set_constants(NULL); // This will get filled in later
   m->set_name_index(cp->utf8(name));
--- a/src/share/vm/classfile/javaClasses.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/javaClasses.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1813,10 +1813,12 @@
   annotations_offset = -1;
   parameter_annotations_offset = -1;
   annotation_default_offset = -1;
+  type_annotations_offset = -1;
   compute_optional_offset(signature_offset,             k, vmSymbols::signature_name(),             vmSymbols::string_signature());
   compute_optional_offset(annotations_offset,           k, vmSymbols::annotations_name(),           vmSymbols::byte_array_signature());
   compute_optional_offset(parameter_annotations_offset, k, vmSymbols::parameter_annotations_name(), vmSymbols::byte_array_signature());
   compute_optional_offset(annotation_default_offset,    k, vmSymbols::annotation_default_name(),    vmSymbols::byte_array_signature());
+  compute_optional_offset(type_annotations_offset,      k, vmSymbols::type_annotations_name(),      vmSymbols::byte_array_signature());
 }
 
 Handle java_lang_reflect_Method::create(TRAPS) {
@@ -1962,6 +1964,22 @@
   method->obj_field_put(annotation_default_offset, value);
 }
 
+bool java_lang_reflect_Method::has_type_annotations_field() {
+  return (type_annotations_offset >= 0);
+}
+
+oop java_lang_reflect_Method::type_annotations(oop method) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  return method->obj_field(type_annotations_offset);
+}
+
+void java_lang_reflect_Method::set_type_annotations(oop method, oop value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  method->obj_field_put(type_annotations_offset, value);
+}
+
 void java_lang_reflect_Constructor::compute_offsets() {
   Klass* k = SystemDictionary::reflect_Constructor_klass();
   compute_offset(clazz_offset,          k, vmSymbols::clazz_name(),          vmSymbols::class_signature());
@@ -1973,9 +1991,11 @@
   signature_offset = -1;
   annotations_offset = -1;
   parameter_annotations_offset = -1;
+  type_annotations_offset = -1;
   compute_optional_offset(signature_offset,             k, vmSymbols::signature_name(),             vmSymbols::string_signature());
   compute_optional_offset(annotations_offset,           k, vmSymbols::annotations_name(),           vmSymbols::byte_array_signature());
   compute_optional_offset(parameter_annotations_offset, k, vmSymbols::parameter_annotations_name(), vmSymbols::byte_array_signature());
+  compute_optional_offset(type_annotations_offset,      k, vmSymbols::type_annotations_name(),      vmSymbols::byte_array_signature());
 }
 
 Handle java_lang_reflect_Constructor::create(TRAPS) {
@@ -2086,6 +2106,22 @@
   method->obj_field_put(parameter_annotations_offset, value);
 }
 
+bool java_lang_reflect_Constructor::has_type_annotations_field() {
+  return (type_annotations_offset >= 0);
+}
+
+oop java_lang_reflect_Constructor::type_annotations(oop constructor) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  return constructor->obj_field(type_annotations_offset);
+}
+
+void java_lang_reflect_Constructor::set_type_annotations(oop constructor, oop value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  constructor->obj_field_put(type_annotations_offset, value);
+}
+
 void java_lang_reflect_Field::compute_offsets() {
   Klass* k = SystemDictionary::reflect_Field_klass();
   compute_offset(clazz_offset,     k, vmSymbols::clazz_name(),     vmSymbols::class_signature());
@@ -2096,8 +2132,10 @@
   // The generic signature and annotations fields are only present in 1.5
   signature_offset = -1;
   annotations_offset = -1;
+  type_annotations_offset = -1;
   compute_optional_offset(signature_offset, k, vmSymbols::signature_name(), vmSymbols::string_signature());
   compute_optional_offset(annotations_offset,  k, vmSymbols::annotations_name(),  vmSymbols::byte_array_signature());
+  compute_optional_offset(type_annotations_offset,  k, vmSymbols::type_annotations_name(),  vmSymbols::byte_array_signature());
 }
 
 Handle java_lang_reflect_Field::create(TRAPS) {
@@ -2192,6 +2230,21 @@
   field->obj_field_put(annotations_offset, value);
 }
 
+bool java_lang_reflect_Field::has_type_annotations_field() {
+  return (type_annotations_offset >= 0);
+}
+
+oop java_lang_reflect_Field::type_annotations(oop field) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  return field->obj_field(type_annotations_offset);
+}
+
+void java_lang_reflect_Field::set_type_annotations(oop field, oop value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  assert(has_type_annotations_field(), "type_annotations field must be present");
+  field->obj_field_put(type_annotations_offset, value);
+}
 
 void sun_reflect_ConstantPool::compute_offsets() {
   Klass* k = SystemDictionary::reflect_ConstantPool_klass();
@@ -2202,6 +2255,66 @@
   }
 }
 
+void java_lang_reflect_Parameter::compute_offsets() {
+  Klass* k = SystemDictionary::reflect_Parameter_klass();
+  if(NULL != k) {
+    compute_offset(name_offset,        k, vmSymbols::name_name(),        vmSymbols::string_signature());
+    compute_offset(modifiers_offset,   k, vmSymbols::modifiers_name(),   vmSymbols::int_signature());
+    compute_offset(index_offset,       k, vmSymbols::index_name(),       vmSymbols::int_signature());
+    compute_offset(executable_offset,  k, vmSymbols::executable_name(),  vmSymbols::executable_signature());
+  }
+}
+
+Handle java_lang_reflect_Parameter::create(TRAPS) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  Symbol* name = vmSymbols::java_lang_reflect_Parameter();
+  Klass* k = SystemDictionary::resolve_or_fail(name, true, CHECK_NH);
+  instanceKlassHandle klass (THREAD, k);
+  // Ensure it is initialized
+  klass->initialize(CHECK_NH);
+  return klass->allocate_instance_handle(CHECK_NH);
+}
+
+oop java_lang_reflect_Parameter::name(oop param) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  return param->obj_field(name_offset);
+}
+
+void java_lang_reflect_Parameter::set_name(oop param, oop value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  param->obj_field_put(name_offset, value);
+}
+
+int java_lang_reflect_Parameter::modifiers(oop param) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  return param->int_field(modifiers_offset);
+}
+
+void java_lang_reflect_Parameter::set_modifiers(oop param, int value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  param->int_field_put(modifiers_offset, value);
+}
+
+int java_lang_reflect_Parameter::index(oop param) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  return param->int_field(index_offset);
+}
+
+void java_lang_reflect_Parameter::set_index(oop param, int value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  param->int_field_put(index_offset, value);
+}
+
+oop java_lang_reflect_Parameter::executable(oop param) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  return param->obj_field(executable_offset);
+}
+
+void java_lang_reflect_Parameter::set_executable(oop param, oop value) {
+  assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
+  param->obj_field_put(executable_offset, value);
+}
+
 
 Handle sun_reflect_ConstantPool::create(TRAPS) {
   assert(Universe::is_fully_initialized(), "Need to find another solution to the reflection problem");
@@ -2857,6 +2970,7 @@
 int java_lang_reflect_Method::annotations_offset;
 int java_lang_reflect_Method::parameter_annotations_offset;
 int java_lang_reflect_Method::annotation_default_offset;
+int java_lang_reflect_Method::type_annotations_offset;
 int java_lang_reflect_Constructor::clazz_offset;
 int java_lang_reflect_Constructor::parameterTypes_offset;
 int java_lang_reflect_Constructor::exceptionTypes_offset;
@@ -2865,6 +2979,7 @@
 int java_lang_reflect_Constructor::signature_offset;
 int java_lang_reflect_Constructor::annotations_offset;
 int java_lang_reflect_Constructor::parameter_annotations_offset;
+int java_lang_reflect_Constructor::type_annotations_offset;
 int java_lang_reflect_Field::clazz_offset;
 int java_lang_reflect_Field::name_offset;
 int java_lang_reflect_Field::type_offset;
@@ -2872,6 +2987,11 @@
 int java_lang_reflect_Field::modifiers_offset;
 int java_lang_reflect_Field::signature_offset;
 int java_lang_reflect_Field::annotations_offset;
+int java_lang_reflect_Field::type_annotations_offset;
+int java_lang_reflect_Parameter::name_offset;
+int java_lang_reflect_Parameter::modifiers_offset;
+int java_lang_reflect_Parameter::index_offset;
+int java_lang_reflect_Parameter::executable_offset;
 int java_lang_boxing_object::value_offset;
 int java_lang_boxing_object::long_value_offset;
 int java_lang_ref_Reference::referent_offset;
@@ -3056,6 +3176,8 @@
     sun_reflect_ConstantPool::compute_offsets();
     sun_reflect_UnsafeStaticFieldAccessorImpl::compute_offsets();
   }
+  if (JDK_Version::is_jdk18x_version())
+    java_lang_reflect_Parameter::compute_offsets();
 
   // generated interpreter code wants to know about the offsets we just computed:
   AbstractAssembler::update_delayed_values();
--- a/src/share/vm/classfile/javaClasses.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/javaClasses.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -554,6 +554,7 @@
   static int annotations_offset;
   static int parameter_annotations_offset;
   static int annotation_default_offset;
+  static int type_annotations_offset;
 
   static void compute_offsets();
 
@@ -599,6 +600,10 @@
   static oop annotation_default(oop method);
   static void set_annotation_default(oop method, oop value);
 
+  static bool has_type_annotations_field();
+  static oop type_annotations(oop method);
+  static void set_type_annotations(oop method, oop value);
+
   // Debugging
   friend class JavaClasses;
 };
@@ -618,6 +623,7 @@
   static int signature_offset;
   static int annotations_offset;
   static int parameter_annotations_offset;
+  static int type_annotations_offset;
 
   static void compute_offsets();
 
@@ -653,6 +659,10 @@
   static oop parameter_annotations(oop method);
   static void set_parameter_annotations(oop method, oop value);
 
+  static bool has_type_annotations_field();
+  static oop type_annotations(oop constructor);
+  static void set_type_annotations(oop constructor, oop value);
+
   // Debugging
   friend class JavaClasses;
 };
@@ -671,6 +681,7 @@
   static int modifiers_offset;
   static int signature_offset;
   static int annotations_offset;
+  static int type_annotations_offset;
 
   static void compute_offsets();
 
@@ -710,10 +721,45 @@
   static oop annotation_default(oop method);
   static void set_annotation_default(oop method, oop value);
 
+  static bool has_type_annotations_field();
+  static oop type_annotations(oop field);
+  static void set_type_annotations(oop field, oop value);
+
   // Debugging
   friend class JavaClasses;
 };
 
+class java_lang_reflect_Parameter {
+ private:
+  // Note that to reduce dependencies on the JDK we compute these
+  // offsets at run-time.
+  static int name_offset;
+  static int modifiers_offset;
+  static int index_offset;
+  static int executable_offset;
+
+  static void compute_offsets();
+
+ public:
+  // Allocation
+  static Handle create(TRAPS);
+
+  // Accessors
+  static oop name(oop field);
+  static void set_name(oop field, oop value);
+
+  static int index(oop reflect);
+  static void set_index(oop reflect, int value);
+
+  static int modifiers(oop reflect);
+  static void set_modifiers(oop reflect, int value);
+
+  static oop executable(oop constructor);
+  static void set_executable(oop constructor, oop value);
+
+  friend class JavaClasses;
+};
+
 // Interface to sun.reflect.ConstantPool objects
 class sun_reflect_ConstantPool {
  private:
--- a/src/share/vm/classfile/systemDictionary.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/systemDictionary.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -131,6 +131,7 @@
   do_klass(Properties_klass,                            java_util_Properties,                      Pre                 ) \
   do_klass(reflect_AccessibleObject_klass,              java_lang_reflect_AccessibleObject,        Pre                 ) \
   do_klass(reflect_Field_klass,                         java_lang_reflect_Field,                   Pre                 ) \
+  do_klass(reflect_Parameter_klass,                     java_lang_reflect_Parameter,               Opt                 ) \
   do_klass(reflect_Method_klass,                        java_lang_reflect_Method,                  Pre                 ) \
   do_klass(reflect_Constructor_klass,                   java_lang_reflect_Constructor,             Pre                 ) \
                                                                                                                          \
@@ -459,6 +460,7 @@
   // Tells whether ClassLoader.checkPackageAccess is present
   static bool has_checkPackageAccess()      { return _has_checkPackageAccess; }
 
+  static bool Parameter_klass_loaded()      { return WK_KLASS(reflect_Parameter_klass) != NULL; }
   static bool Class_klass_loaded()          { return WK_KLASS(Class_klass) != NULL; }
   static bool Cloneable_klass_loaded()      { return WK_KLASS(Cloneable_klass) != NULL; }
   static bool Object_klass_loaded()         { return WK_KLASS(Object_klass) != NULL; }
--- a/src/share/vm/classfile/vmSymbols.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/classfile/vmSymbols.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -86,6 +86,7 @@
   template(java_lang_reflect_Method,                  "java/lang/reflect/Method")                 \
   template(java_lang_reflect_Constructor,             "java/lang/reflect/Constructor")            \
   template(java_lang_reflect_Field,                   "java/lang/reflect/Field")                  \
+  template(java_lang_reflect_Parameter,               "java/lang/reflect/Parameter")              \
   template(java_lang_reflect_Array,                   "java/lang/reflect/Array")                  \
   template(java_lang_StringBuffer,                    "java/lang/StringBuffer")                   \
   template(java_lang_StringBuilder,                   "java/lang/StringBuilder")                  \
@@ -126,6 +127,7 @@
   template(tag_line_number_table,                     "LineNumberTable")                          \
   template(tag_local_variable_table,                  "LocalVariableTable")                       \
   template(tag_local_variable_type_table,             "LocalVariableTypeTable")                   \
+  template(tag_method_parameters,                     "MethodParameters")                         \
   template(tag_stack_map_table,                       "StackMapTable")                            \
   template(tag_synthetic,                             "Synthetic")                                \
   template(tag_deprecated,                            "Deprecated")                               \
@@ -136,6 +138,8 @@
   template(tag_runtime_visible_parameter_annotations, "RuntimeVisibleParameterAnnotations")       \
   template(tag_runtime_invisible_parameter_annotations,"RuntimeInvisibleParameterAnnotations")    \
   template(tag_annotation_default,                    "AnnotationDefault")                        \
+  template(tag_runtime_visible_type_annotations,      "RuntimeVisibleTypeAnnotations")            \
+  template(tag_runtime_invisible_type_annotations,    "RuntimeInvisibleTypeAnnotations")          \
   template(tag_enclosing_method,                      "EnclosingMethod")                          \
   template(tag_bootstrap_methods,                     "BootstrapMethods")                         \
                                                                                                   \
@@ -233,12 +237,17 @@
   /* Support for annotations (JDK 1.5 and above) */                                               \
                                                                                                   \
   template(annotations_name,                          "annotations")                              \
+  template(index_name,                                "index")                                    \
+  template(executable_name,                           "executable")                               \
   template(parameter_annotations_name,                "parameterAnnotations")                     \
   template(annotation_default_name,                   "annotationDefault")                        \
   template(sun_reflect_ConstantPool,                  "sun/reflect/ConstantPool")                 \
   template(ConstantPool_name,                         "constantPoolOop")                          \
   template(sun_reflect_UnsafeStaticFieldAccessorImpl, "sun/reflect/UnsafeStaticFieldAccessorImpl")\
   template(base_name,                                 "base")                                     \
+  /* Type Annotations (JDK 8 and above) */                                                        \
+  template(type_annotations_name,                     "typeAnnotations")                          \
+                                                                                                  \
                                                                                                   \
   /* Support for JSR 292 & invokedynamic (JDK 1.7 and above) */                                   \
   template(java_lang_invoke_CallSite,                 "java/lang/invoke/CallSite")                \
@@ -470,6 +479,7 @@
   template(class_signature,                           "Ljava/lang/Class;")                                        \
   template(string_signature,                          "Ljava/lang/String;")                                       \
   template(reference_signature,                       "Ljava/lang/ref/Reference;")                                \
+  template(executable_signature,                      "Ljava/lang/reflect/Executable;")                           \
   template(concurrenthashmap_signature,               "Ljava/util/concurrent/ConcurrentHashMap;")                 \
   template(String_StringBuilder_signature,            "(Ljava/lang/String;)Ljava/lang/StringBuilder;")            \
   template(int_StringBuilder_signature,               "(I)Ljava/lang/StringBuilder;")                             \
--- a/src/share/vm/gc_implementation/concurrentMarkSweep/cmsCollectorPolicy.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/concurrentMarkSweep/cmsCollectorPolicy.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -56,7 +56,7 @@
   if (_generations == NULL)
     vm_exit_during_initialization("Unable to allocate gen spec");
 
-  if (ParNewGeneration::in_use()) {
+  if (UseParNewGC) {
     if (UseAdaptiveSizePolicy) {
       _generations[0] = new GenerationSpec(Generation::ASParNew,
                                            _initial_gen0_size, _max_gen0_size);
@@ -96,7 +96,7 @@
 
 void ConcurrentMarkSweepPolicy::initialize_gc_policy_counters() {
   // initialize the policy counters - 2 collectors, 3 generations
-  if (ParNewGeneration::in_use()) {
+  if (UseParNewGC) {
     _gc_policy_counters = new GCPolicyCounters("ParNew:CMS", 2, 3);
   }
   else {
@@ -119,7 +119,7 @@
 
   assert(size_policy() != NULL, "A size policy is required");
   // initialize the policy counters - 2 collectors, 3 generations
-  if (ParNewGeneration::in_use()) {
+  if (UseParNewGC) {
     _gc_policy_counters = new CMSGCAdaptivePolicyCounters("ParNew:CMS", 2, 3,
       size_policy());
   }
--- a/src/share/vm/gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -214,7 +214,6 @@
     assert(q->forwardee() == NULL, "should be forwarded to NULL");
   }
 
-  VALIDATE_MARK_SWEEP_ONLY(MarkSweep::register_live_oop(q, adjusted_size));
   compact_top += adjusted_size;
 
   // we need to update the offset table so that the beginnings of objects can be
--- a/src/share/vm/gc_implementation/concurrentMarkSweep/concurrentMarkSweepGeneration.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/concurrentMarkSweep/concurrentMarkSweepGeneration.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -827,10 +827,10 @@
   GenCollectedHeap* gch = GenCollectedHeap::heap();
   if (PrintGCDetails) {
     if (Verbose) {
-      gclog_or_tty->print(" [%d %s-%s: "SIZE_FORMAT"("SIZE_FORMAT")]",
+      gclog_or_tty->print("[%d %s-%s: "SIZE_FORMAT"("SIZE_FORMAT")]",
         level(), short_name(), s, used(), capacity());
     } else {
-      gclog_or_tty->print(" [%d %s-%s: "SIZE_FORMAT"K("SIZE_FORMAT"K)]",
+      gclog_or_tty->print("[%d %s-%s: "SIZE_FORMAT"K("SIZE_FORMAT"K)]",
         level(), short_name(), s, used() / K, capacity() / K);
     }
   }
--- a/src/share/vm/gc_implementation/g1/collectionSetChooser.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/collectionSetChooser.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -85,7 +85,7 @@
     _curr_index(0), _length(0), _first_par_unreserved_idx(0),
     _region_live_threshold_bytes(0), _remaining_reclaimable_bytes(0) {
   _region_live_threshold_bytes =
-    HeapRegion::GrainBytes * (size_t) G1OldCSetRegionLiveThresholdPercent / 100;
+    HeapRegion::GrainBytes * (size_t) G1MixedGCLiveThresholdPercent / 100;
 }
 
 #ifndef PRODUCT
--- a/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2001, 2013, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -192,6 +192,7 @@
   setEmpty();
   _capacity = (jint) capacity;
   _saved_index = -1;
+  _should_expand = false;
   NOT_PRODUCT(_max_depth = 0);
   return true;
 }
@@ -747,8 +748,8 @@
   assert(_heap_end != NULL, "heap bounds should look ok");
   assert(_heap_start < _heap_end, "heap bounds should look ok");
 
-  // reset all the marking data structures and any necessary flags
-  clear_marking_state();
+  // Reset all the marking data structures and any necessary flags
+  reset_marking_state();
 
   if (verbose_low()) {
     gclog_or_tty->print_cr("[global] resetting");
@@ -766,6 +767,23 @@
   set_concurrent_marking_in_progress();
 }
 
+
+void ConcurrentMark::reset_marking_state(bool clear_overflow) {
+  _markStack.set_should_expand();
+  _markStack.setEmpty();        // Also clears the _markStack overflow flag
+  if (clear_overflow) {
+    clear_has_overflown();
+  } else {
+    assert(has_overflown(), "pre-condition");
+  }
+  _finger = _heap_start;
+
+  for (uint i = 0; i < _max_worker_id; ++i) {
+    CMTaskQueue* queue = _task_queues->queue(i);
+    queue->set_empty();
+  }
+}
+
 void ConcurrentMark::set_phase(uint active_tasks, bool concurrent) {
   assert(active_tasks <= _max_worker_id, "we should not have more");
 
@@ -796,7 +814,7 @@
 void ConcurrentMark::set_non_marking_state() {
   // We set the global marking state to some default values when we're
   // not doing marking.
-  clear_marking_state();
+  reset_marking_state();
   _active_tasks = 0;
   clear_concurrent_marking_in_progress();
 }
@@ -963,7 +981,7 @@
     // not clear the overflow flag since we rely on it being true when
     // we exit this method to abort the pause and restart concurent
     // marking.
-    clear_marking_state(concurrent() /* clear_overflow */);
+    reset_marking_state(concurrent() /* clear_overflow */);
     force_overflow()->update();
 
     if (G1Log::fine()) {
@@ -1257,8 +1275,9 @@
   if (has_overflown()) {
     // Oops.  We overflowed.  Restart concurrent marking.
     _restart_for_overflow = true;
-    // Clear the flag. We do not need it any more.
-    clear_has_overflown();
+    // Clear the marking state because we will be restarting
+    // marking due to overflowing the global mark stack.
+    reset_marking_state();
     if (G1TraceMarkStackOverflow) {
       gclog_or_tty->print_cr("\nRemark led to restart for overflow.");
     }
@@ -1282,6 +1301,8 @@
                        /* option */ VerifyOption_G1UseNextMarking);
     }
     assert(!restart_for_overflow(), "sanity");
+    // Completely reset the marking state since marking completed
+    set_non_marking_state();
   }
 
   // Expand the marking stack, if we have to and if we can.
@@ -1289,11 +1310,6 @@
     _markStack.expand();
   }
 
-  // Reset the marking state if marking completed
-  if (!restart_for_overflow()) {
-    set_non_marking_state();
-  }
-
 #if VERIFY_OBJS_PROCESSED
   _scan_obj_cl.objs_processed = 0;
   ThreadLocalObjQueue::objs_enqueued = 0;
@@ -2963,22 +2979,6 @@
 }
 #endif // PRODUCT
 
-void ConcurrentMark::clear_marking_state(bool clear_overflow) {
-  _markStack.set_should_expand();
-  _markStack.setEmpty();        // Also clears the _markStack overflow flag
-  if (clear_overflow) {
-    clear_has_overflown();
-  } else {
-    assert(has_overflown(), "pre-condition");
-  }
-  _finger = _heap_start;
-
-  for (uint i = 0; i < _max_worker_id; ++i) {
-    CMTaskQueue* queue = _task_queues->queue(i);
-    queue->set_empty();
-  }
-}
-
 // Aggregate the counting data that was constructed concurrently
 // with marking.
 class AggregateCountDataHRClosure: public HeapRegionClosure {
@@ -3185,7 +3185,7 @@
   // Clear the liveness counting data
   clear_all_count_data();
   // Empty mark stack
-  clear_marking_state();
+  reset_marking_state();
   for (uint i = 0; i < _max_worker_id; ++i) {
     _tasks[i]->clear_region_fields();
   }
--- a/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2001, 2013, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -478,15 +478,18 @@
   // It resets the global marking data structures, as well as the
   // task local ones; should be called during initial mark.
   void reset();
-  // It resets all the marking data structures.
-  void clear_marking_state(bool clear_overflow = true);
+
+  // Resets all the marking data structures. Called when we have to restart
+  // marking or when marking completes (via set_non_marking_state below).
+  void reset_marking_state(bool clear_overflow = true);
+
+  // We do this after we're done with marking so that the marking data
+  // structures are initialised to a sensible and predictable state.
+  void set_non_marking_state();
 
   // It should be called to indicate which phase we're in (concurrent
   // mark or remark) and how many threads are currently active.
   void set_phase(uint active_tasks, bool concurrent);
-  // We do this after we're done with marking so that the marking data
-  // structures are initialised to a sensible and predictable state.
-  void set_non_marking_state();
 
   // prints all gathered CM-related statistics
   void print_stats();
--- a/src/share/vm/gc_implementation/g1/concurrentMarkThread.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/concurrentMarkThread.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2001, 2013, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -159,13 +159,11 @@
           VM_CGC_Operation op(&final_cl, verbose_str, true /* needs_pll */);
           VMThread::execute(&op);
         }
-        if (cm()->restart_for_overflow() &&
-            G1TraceMarkStackOverflow) {
-          gclog_or_tty->print_cr("Restarting conc marking because of MS overflow "
-                                 "in remark (restart #%d).", iter);
-        }
-
         if (cm()->restart_for_overflow()) {
+          if (G1TraceMarkStackOverflow) {
+            gclog_or_tty->print_cr("Restarting conc marking because of MS overflow "
+                                   "in remark (restart #%d).", iter);
+          }
           if (G1Log::fine()) {
             gclog_or_tty->date_stamp(PrintGCDateStamps);
             gclog_or_tty->stamp(PrintGCTimeStamps);
--- a/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -3668,7 +3668,7 @@
   gclog_or_tty->stamp(PrintGCTimeStamps);
 
   GCCauseString gc_cause_str = GCCauseString("GC pause", gc_cause())
-    .append(g1_policy()->gcs_are_young() ? " (young)" : " (mixed)")
+    .append(g1_policy()->gcs_are_young() ? "(young)" : "(mixed)")
     .append(g1_policy()->during_initial_mark_pause() ? " (initial-mark)" : "");
 
   gclog_or_tty->print("[%s", (const char*)gc_cause_str);
--- a/src/share/vm/gc_implementation/g1/g1CollectorPolicy.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/g1CollectorPolicy.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -309,9 +309,9 @@
 }
 
 G1YoungGenSizer::G1YoungGenSizer() : _sizer_kind(SizerDefaults), _adaptive_size(true) {
-  assert(G1DefaultMinNewGenPercent <= G1DefaultMaxNewGenPercent, "Min larger than max");
-  assert(G1DefaultMinNewGenPercent > 0 && G1DefaultMinNewGenPercent < 100, "Min out of bounds");
-  assert(G1DefaultMaxNewGenPercent > 0 && G1DefaultMaxNewGenPercent < 100, "Max out of bounds");
+  assert(G1NewSizePercent <= G1MaxNewSizePercent, "Min larger than max");
+  assert(G1NewSizePercent > 0 && G1NewSizePercent < 100, "Min out of bounds");
+  assert(G1MaxNewSizePercent > 0 && G1MaxNewSizePercent < 100, "Max out of bounds");
 
   if (FLAG_IS_CMDLINE(NewRatio)) {
     if (FLAG_IS_CMDLINE(NewSize) || FLAG_IS_CMDLINE(MaxNewSize)) {
@@ -344,12 +344,12 @@
 }
 
 uint G1YoungGenSizer::calculate_default_min_length(uint new_number_of_heap_regions) {
-  uint default_value = (new_number_of_heap_regions * G1DefaultMinNewGenPercent) / 100;
+  uint default_value = (new_number_of_heap_regions * G1NewSizePercent) / 100;
   return MAX2(1U, default_value);
 }
 
 uint G1YoungGenSizer::calculate_default_max_length(uint new_number_of_heap_regions) {
-  uint default_value = (new_number_of_heap_regions * G1DefaultMaxNewGenPercent) / 100;
+  uint default_value = (new_number_of_heap_regions * G1MaxNewSizePercent) / 100;
   return MAX2(1U, default_value);
 }
 
--- a/src/share/vm/gc_implementation/g1/g1CollectorPolicy.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/g1CollectorPolicy.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -94,18 +94,18 @@
 // will occur.
 //
 // If nothing related to the the young gen size is set on the command
-// line we should allow the young gen to be between
-// G1DefaultMinNewGenPercent and G1DefaultMaxNewGenPercent of the
-// heap size. This means that every time the heap size changes the
-// limits for the young gen size will be updated.
+// line we should allow the young gen to be between G1NewSizePercent
+// and G1MaxNewSizePercent of the heap size. This means that every time
+// the heap size changes, the limits for the young gen size will be
+// recalculated.
 //
 // If only -XX:NewSize is set we should use the specified value as the
-// minimum size for young gen. Still using G1DefaultMaxNewGenPercent
-// of the heap as maximum.
+// minimum size for young gen. Still using G1MaxNewSizePercent of the
+// heap as maximum.
 //
 // If only -XX:MaxNewSize is set we should use the specified value as the
-// maximum size for young gen. Still using G1DefaultMinNewGenPercent
-// of the heap as minimum.
+// maximum size for young gen. Still using G1NewSizePercent of the heap
+// as minimum.
 //
 // If -XX:NewSize and -XX:MaxNewSize are both specified we use these values.
 // No updates when the heap size changes. There is a special case when
--- a/src/share/vm/gc_implementation/g1/g1MarkSweep.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/g1MarkSweep.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -282,10 +282,8 @@
       if (r->startsHumongous()) {
         // We must adjust the pointers on the single H object.
         oop obj = oop(r->bottom());
-        debug_only(GenMarkSweep::track_interior_pointers(obj));
         // point all the oops to the new location
         obj->adjust_pointers();
-        debug_only(GenMarkSweep::check_interior_pointers());
       }
     } else {
       // This really ought to be "as_CompactibleSpace"...
--- a/src/share/vm/gc_implementation/g1/g1_globals.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/g1/g1_globals.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -287,17 +287,18 @@
           "The number of times we'll force an overflow during "             \
           "concurrent marking")                                             \
                                                                             \
-  experimental(uintx, G1DefaultMinNewGenPercent, 20,                        \
-          "Percentage (0-100) of the heap size to use as minimum "          \
-          "young gen size.")                                                \
+  experimental(uintx, G1NewSizePercent, 20,                                 \
+          "Percentage (0-100) of the heap size to use as default "          \
+          "minimum young gen size.")                                        \
                                                                             \
-  experimental(uintx, G1DefaultMaxNewGenPercent, 80,                        \
-          "Percentage (0-100) of the heap size to use as maximum "          \
-          "young gen size.")                                                \
+  experimental(uintx, G1MaxNewSizePercent, 80,                              \
+          "Percentage (0-100) of the heap size to use as default "          \
+          " maximum young gen size.")                                       \
                                                                             \
-  experimental(uintx, G1OldCSetRegionLiveThresholdPercent, 90,              \
-          "Threshold for regions to be added to the collection set. "       \
-          "Regions with more live bytes than this will not be collected.")  \
+  experimental(uintx, G1MixedGCLiveThresholdPercent, 90,                    \
+          "Threshold for regions to be considered for inclusion in the "    \
+          "collection set of mixed GCs. "                                   \
+          "Regions with live bytes exceeding this will not be collected.")  \
                                                                             \
   product(uintx, G1HeapWastePercent, 5,                                     \
           "Amount of space, expressed as a percentage of the heap size, "   \
--- a/src/share/vm/gc_implementation/parNew/parNewGeneration.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parNew/parNewGeneration.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1623,7 +1623,3 @@
 const char* ParNewGeneration::name() const {
   return "par new generation";
 }
-
-bool ParNewGeneration::in_use() {
-  return UseParNewGC && ParallelGCThreads > 0;
-}
--- a/src/share/vm/gc_implementation/parNew/parNewGeneration.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parNew/parNewGeneration.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -361,8 +361,6 @@
     delete _task_queues;
   }
 
-  static bool in_use();
-
   virtual void ref_processor_init();
   virtual Generation::Name kind()        { return Generation::ParNew; }
   virtual const char* name() const;
--- a/src/share/vm/gc_implementation/parallelScavenge/psMarkSweepDecorator.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parallelScavenge/psMarkSweepDecorator.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -164,7 +164,6 @@
         start_array->allocate_block(compact_top);
       }
 
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::register_live_oop(oop(q), size));
       compact_top += size;
       assert(compact_top <= dest->space()->end(),
         "Exceeding space in destination");
@@ -225,7 +224,6 @@
             start_array->allocate_block(compact_top);
           }
 
-          VALIDATE_MARK_SWEEP_ONLY(MarkSweep::register_live_oop(oop(q), sz));
           compact_top += sz;
           assert(compact_top <= dest->space()->end(),
             "Exceeding space in destination");
@@ -304,11 +302,8 @@
     HeapWord* end = _first_dead;
 
     while (q < end) {
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::track_interior_pointers(oop(q)));
       // point all the oops to the new location
       size_t size = oop(q)->adjust_pointers();
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::check_interior_pointers());
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::validate_live_oop(oop(q), size));
       q += size;
     }
 
@@ -328,11 +323,8 @@
     Prefetch::write(q, interval);
     if (oop(q)->is_gc_marked()) {
       // q is alive
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::track_interior_pointers(oop(q)));
       // point all the oops to the new location
       size_t size = oop(q)->adjust_pointers();
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::check_interior_pointers());
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::validate_live_oop(oop(q), size));
       debug_only(prev_q = q);
       q += size;
     } else {
@@ -366,7 +358,6 @@
     while (q < end) {
       size_t size = oop(q)->size();
       assert(!oop(q)->is_gc_marked(), "should be unmarked (special dense prefix handling)");
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::live_oop_moved_to(q, size, q));
       debug_only(prev_q = q);
       q += size;
     }
@@ -401,7 +392,6 @@
       Prefetch::write(compaction_top, copy_interval);
 
       // copy object and reinit its mark
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::live_oop_moved_to(q, size, compaction_top));
       assert(q != compaction_top, "everything in this pass should be moving");
       Copy::aligned_conjoint_words(q, compaction_top, size);
       oop(compaction_top)->init_mark();
--- a/src/share/vm/gc_implementation/parallelScavenge/psParallelCompact.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parallelScavenge/psParallelCompact.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -99,25 +99,6 @@
 bool   PSParallelCompact::_dwl_initialized = false;
 #endif  // #ifdef ASSERT
 
-#ifdef VALIDATE_MARK_SWEEP
-GrowableArray<void*>*   PSParallelCompact::_root_refs_stack = NULL;
-GrowableArray<oop> *    PSParallelCompact::_live_oops = NULL;
-GrowableArray<oop> *    PSParallelCompact::_live_oops_moved_to = NULL;
-GrowableArray<size_t>*  PSParallelCompact::_live_oops_size = NULL;
-size_t                  PSParallelCompact::_live_oops_index = 0;
-GrowableArray<void*>*   PSParallelCompact::_other_refs_stack = NULL;
-GrowableArray<void*>*   PSParallelCompact::_adjusted_pointers = NULL;
-bool                    PSParallelCompact::_pointer_tracking = false;
-bool                    PSParallelCompact::_root_tracking = true;
-
-GrowableArray<HeapWord*>* PSParallelCompact::_cur_gc_live_oops = NULL;
-GrowableArray<HeapWord*>* PSParallelCompact::_cur_gc_live_oops_moved_to = NULL;
-GrowableArray<size_t>   * PSParallelCompact::_cur_gc_live_oops_size = NULL;
-GrowableArray<HeapWord*>* PSParallelCompact::_last_gc_live_oops = NULL;
-GrowableArray<HeapWord*>* PSParallelCompact::_last_gc_live_oops_moved_to = NULL;
-GrowableArray<size_t>   * PSParallelCompact::_last_gc_live_oops_size = NULL;
-#endif
-
 void SplitInfo::record(size_t src_region_idx, size_t partial_obj_size,
                        HeapWord* destination)
 {
@@ -2715,151 +2696,6 @@
 }
 #endif  // #ifdef ASSERT
 
-
-#ifdef VALIDATE_MARK_SWEEP
-
-void PSParallelCompact::track_adjusted_pointer(void* p, bool isroot) {
-  if (!ValidateMarkSweep)
-    return;
-
-  if (!isroot) {
-    if (_pointer_tracking) {
-      guarantee(_adjusted_pointers->contains(p), "should have seen this pointer");
-      _adjusted_pointers->remove(p);
-    }
-  } else {
-    ptrdiff_t index = _root_refs_stack->find(p);
-    if (index != -1) {
-      int l = _root_refs_stack->length();
-      if (l > 0 && l - 1 != index) {
-        void* last = _root_refs_stack->pop();
-        assert(last != p, "should be different");
-        _root_refs_stack->at_put(index, last);
-      } else {
-        _root_refs_stack->remove(p);
-      }
-    }
-  }
-}
-
-
-void PSParallelCompact::check_adjust_pointer(void* p) {
-  _adjusted_pointers->push(p);
-}
-
-
-class AdjusterTracker: public OopClosure {
- public:
-  AdjusterTracker() {};
-  void do_oop(oop* o)         { PSParallelCompact::check_adjust_pointer(o); }
-  void do_oop(narrowOop* o)   { PSParallelCompact::check_adjust_pointer(o); }
-};
-
-
-void PSParallelCompact::track_interior_pointers(oop obj) {
-  if (ValidateMarkSweep) {
-    _adjusted_pointers->clear();
-    _pointer_tracking = true;
-
-    AdjusterTracker checker;
-    obj->oop_iterate_no_header(&checker);
-  }
-}
-
-
-void PSParallelCompact::check_interior_pointers() {
-  if (ValidateMarkSweep) {
-    _pointer_tracking = false;
-    guarantee(_adjusted_pointers->length() == 0, "should have processed the same pointers");
-  }
-}
-
-
-void PSParallelCompact::reset_live_oop_tracking() {
-  if (ValidateMarkSweep) {
-    guarantee((size_t)_live_oops->length() == _live_oops_index, "should be at end of live oops");
-    _live_oops_index = 0;
-  }
-}
-
-
-void PSParallelCompact::register_live_oop(oop p, size_t size) {
-  if (ValidateMarkSweep) {
-    _live_oops->push(p);
-    _live_oops_size->push(size);
-    _live_oops_index++;
-  }
-}
-
-void PSParallelCompact::validate_live_oop(oop p, size_t size) {
-  if (ValidateMarkSweep) {
-    oop obj = _live_oops->at((int)_live_oops_index);
-    guarantee(obj == p, "should be the same object");
-    guarantee(_live_oops_size->at((int)_live_oops_index) == size, "should be the same size");
-    _live_oops_index++;
-  }
-}
-
-void PSParallelCompact::live_oop_moved_to(HeapWord* q, size_t size,
-                                  HeapWord* compaction_top) {
-  assert(oop(q)->forwardee() == NULL || oop(q)->forwardee() == oop(compaction_top),
-         "should be moved to forwarded location");
-  if (ValidateMarkSweep) {
-    PSParallelCompact::validate_live_oop(oop(q), size);
-    _live_oops_moved_to->push(oop(compaction_top));
-  }
-  if (RecordMarkSweepCompaction) {
-    _cur_gc_live_oops->push(q);
-    _cur_gc_live_oops_moved_to->push(compaction_top);
-    _cur_gc_live_oops_size->push(size);
-  }
-}
-
-
-void PSParallelCompact::compaction_complete() {
-  if (RecordMarkSweepCompaction) {
-    GrowableArray<HeapWord*>* _tmp_live_oops          = _cur_gc_live_oops;
-    GrowableArray<HeapWord*>* _tmp_live_oops_moved_to = _cur_gc_live_oops_moved_to;
-    GrowableArray<size_t>   * _tmp_live_oops_size     = _cur_gc_live_oops_size;
-
-    _cur_gc_live_oops           = _last_gc_live_oops;
-    _cur_gc_live_oops_moved_to  = _last_gc_live_oops_moved_to;
-    _cur_gc_live_oops_size      = _last_gc_live_oops_size;
-    _last_gc_live_oops          = _tmp_live_oops;
-    _last_gc_live_oops_moved_to = _tmp_live_oops_moved_to;
-    _last_gc_live_oops_size     = _tmp_live_oops_size;
-  }
-}
-
-
-void PSParallelCompact::print_new_location_of_heap_address(HeapWord* q) {
-  if (!RecordMarkSweepCompaction) {
-    tty->print_cr("Requires RecordMarkSweepCompaction to be enabled");
-    return;
-  }
-
-  if (_last_gc_live_oops == NULL) {
-    tty->print_cr("No compaction information gathered yet");
-    return;
-  }
-
-  for (int i = 0; i < _last_gc_live_oops->length(); i++) {
-    HeapWord* old_oop = _last_gc_live_oops->at(i);
-    size_t    sz      = _last_gc_live_oops_size->at(i);
-    if (old_oop <= q && q < (old_oop + sz)) {
-      HeapWord* new_oop = _last_gc_live_oops_moved_to->at(i);
-      size_t offset = (q - old_oop);
-      tty->print_cr("Address " PTR_FORMAT, q);
-      tty->print_cr(" Was in oop " PTR_FORMAT ", size %d, at offset %d", old_oop, sz, offset);
-      tty->print_cr(" Now in oop " PTR_FORMAT ", actual address " PTR_FORMAT, new_oop, new_oop + offset);
-      return;
-    }
-  }
-
-  tty->print_cr("Address " PTR_FORMAT " not found in live oop information from last GC", q);
-}
-#endif //VALIDATE_MARK_SWEEP
-
 // Update interior oops in the ranges of regions [beg_region, end_region).
 void
 PSParallelCompact::update_and_deadwood_in_dense_prefix(ParCompactionManager* cm,
--- a/src/share/vm/gc_implementation/parallelScavenge/psParallelCompact.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parallelScavenge/psParallelCompact.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1006,34 +1006,6 @@
   // Reset time since last full gc
   static void reset_millis_since_last_gc();
 
- protected:
-#ifdef VALIDATE_MARK_SWEEP
-  static GrowableArray<void*>*           _root_refs_stack;
-  static GrowableArray<oop> *            _live_oops;
-  static GrowableArray<oop> *            _live_oops_moved_to;
-  static GrowableArray<size_t>*          _live_oops_size;
-  static size_t                          _live_oops_index;
-  static size_t                          _live_oops_index_at_perm;
-  static GrowableArray<void*>*           _other_refs_stack;
-  static GrowableArray<void*>*           _adjusted_pointers;
-  static bool                            _pointer_tracking;
-  static bool                            _root_tracking;
-
-  // The following arrays are saved since the time of the last GC and
-  // assist in tracking down problems where someone has done an errant
-  // store into the heap, usually to an oop that wasn't properly
-  // handleized across a GC. If we crash or otherwise fail before the
-  // next GC, we can query these arrays to find out the object we had
-  // intended to do the store to (assuming it is still alive) and the
-  // offset within that object. Covered under RecordMarkSweepCompaction.
-  static GrowableArray<HeapWord*> *      _cur_gc_live_oops;
-  static GrowableArray<HeapWord*> *      _cur_gc_live_oops_moved_to;
-  static GrowableArray<size_t>*          _cur_gc_live_oops_size;
-  static GrowableArray<HeapWord*> *      _last_gc_live_oops;
-  static GrowableArray<HeapWord*> *      _last_gc_live_oops_moved_to;
-  static GrowableArray<size_t>*          _last_gc_live_oops_size;
-#endif
-
  public:
   class MarkAndPushClosure: public OopClosure {
    private:
@@ -1191,25 +1163,6 @@
   // Time since last full gc (in milliseconds).
   static jlong millis_since_last_gc();
 
-#ifdef VALIDATE_MARK_SWEEP
-  static void track_adjusted_pointer(void* p, bool isroot);
-  static void check_adjust_pointer(void* p);
-  static void track_interior_pointers(oop obj);
-  static void check_interior_pointers();
-
-  static void reset_live_oop_tracking();
-  static void register_live_oop(oop p, size_t size);
-  static void validate_live_oop(oop p, size_t size);
-  static void live_oop_moved_to(HeapWord* q, size_t size, HeapWord* compaction_top);
-  static void compaction_complete();
-
-  // Querying operation of RecordMarkSweepCompaction results.
-  // Finds and prints the current base oop and offset for a word
-  // within an oop that was live during the last GC. Helpful for
-  // tracking down heap stomps.
-  static void print_new_location_of_heap_address(HeapWord* q);
-#endif  // #ifdef VALIDATE_MARK_SWEEP
-
 #ifndef PRODUCT
   // Debugging support.
   static const char* space_names[last_space_id];
@@ -1250,12 +1203,7 @@
 inline void PSParallelCompact::follow_root(ParCompactionManager* cm, T* p) {
   assert(!Universe::heap()->is_in_reserved(p),
          "roots shouldn't be things within the heap");
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    guarantee(!_root_refs_stack->contains(p), "should only be in here once");
-    _root_refs_stack->push(p);
-  }
-#endif
+
   T heap_oop = oopDesc::load_heap_oop(p);
   if (!oopDesc::is_null(heap_oop)) {
     oop obj = oopDesc::decode_heap_oop_not_null(heap_oop);
@@ -1294,20 +1242,10 @@
       oopDesc::encode_store_heap_oop_not_null(p, new_obj);
     }
   }
-  VALIDATE_MARK_SWEEP_ONLY(track_adjusted_pointer(p, isroot));
 }
 
 template <class T>
 inline void PSParallelCompact::KeepAliveClosure::do_oop_work(T* p) {
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    if (!Universe::heap()->is_in_reserved(p)) {
-      _root_refs_stack->push(p);
-    } else {
-      _other_refs_stack->push(p);
-    }
-  }
-#endif
   mark_and_push(_compaction_manager, p);
 }
 
--- a/src/share/vm/gc_implementation/parallelScavenge/psYoungGen.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/parallelScavenge/psYoungGen.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -808,8 +808,9 @@
   st->print("  to  "); to_space()->print_on(st);
 }
 
+// Note that a space is not printed before the [NAME:
 void PSYoungGen::print_used_change(size_t prev_used) const {
-  gclog_or_tty->print(" [%s:", name());
+  gclog_or_tty->print("[%s:", name());
   gclog_or_tty->print(" "  SIZE_FORMAT "K"
                       "->" SIZE_FORMAT "K"
                       "("  SIZE_FORMAT "K)",
--- a/src/share/vm/gc_implementation/shared/markSweep.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/shared/markSweep.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -42,26 +42,6 @@
 PreservedMark*          MarkSweep::_preserved_marks = NULL;
 ReferenceProcessor*     MarkSweep::_ref_processor   = NULL;
 
-#ifdef VALIDATE_MARK_SWEEP
-GrowableArray<void*>*   MarkSweep::_root_refs_stack = NULL;
-GrowableArray<oop> *    MarkSweep::_live_oops = NULL;
-GrowableArray<oop> *    MarkSweep::_live_oops_moved_to = NULL;
-GrowableArray<size_t>*  MarkSweep::_live_oops_size = NULL;
-size_t                  MarkSweep::_live_oops_index = 0;
-size_t                  MarkSweep::_live_oops_index_at_perm = 0;
-GrowableArray<void*>*   MarkSweep::_other_refs_stack = NULL;
-GrowableArray<void*>*   MarkSweep::_adjusted_pointers = NULL;
-bool                         MarkSweep::_pointer_tracking = false;
-bool                         MarkSweep::_root_tracking = true;
-
-GrowableArray<HeapWord*>* MarkSweep::_cur_gc_live_oops = NULL;
-GrowableArray<HeapWord*>* MarkSweep::_cur_gc_live_oops_moved_to = NULL;
-GrowableArray<size_t>   * MarkSweep::_cur_gc_live_oops_size = NULL;
-GrowableArray<HeapWord*>* MarkSweep::_last_gc_live_oops = NULL;
-GrowableArray<HeapWord*>* MarkSweep::_last_gc_live_oops_moved_to = NULL;
-GrowableArray<size_t>   * MarkSweep::_last_gc_live_oops_size = NULL;
-#endif
-
 MarkSweep::FollowRootClosure  MarkSweep::follow_root_closure;
 CodeBlobToOopClosure MarkSweep::follow_code_root_closure(&MarkSweep::follow_root_closure, /*do_marking=*/ true);
 
@@ -185,142 +165,6 @@
   }
 }
 
-#ifdef VALIDATE_MARK_SWEEP
-
-void MarkSweep::track_adjusted_pointer(void* p, bool isroot) {
-  if (!ValidateMarkSweep)
-    return;
-
-  if (!isroot) {
-    if (_pointer_tracking) {
-      guarantee(_adjusted_pointers->contains(p), "should have seen this pointer");
-      _adjusted_pointers->remove(p);
-    }
-  } else {
-    ptrdiff_t index = _root_refs_stack->find(p);
-    if (index != -1) {
-      int l = _root_refs_stack->length();
-      if (l > 0 && l - 1 != index) {
-        void* last = _root_refs_stack->pop();
-        assert(last != p, "should be different");
-        _root_refs_stack->at_put(index, last);
-      } else {
-        _root_refs_stack->remove(p);
-      }
-    }
-  }
-}
-
-void MarkSweep::check_adjust_pointer(void* p) {
-  _adjusted_pointers->push(p);
-}
-
-class AdjusterTracker: public OopClosure {
- public:
-  AdjusterTracker() {}
-  void do_oop(oop* o)       { MarkSweep::check_adjust_pointer(o); }
-  void do_oop(narrowOop* o) { MarkSweep::check_adjust_pointer(o); }
-};
-
-void MarkSweep::track_interior_pointers(oop obj) {
-  if (ValidateMarkSweep) {
-    _adjusted_pointers->clear();
-    _pointer_tracking = true;
-
-    AdjusterTracker checker;
-    obj->oop_iterate_no_header(&checker);
-  }
-}
-
-void MarkSweep::check_interior_pointers() {
-  if (ValidateMarkSweep) {
-    _pointer_tracking = false;
-    guarantee(_adjusted_pointers->length() == 0, "should have processed the same pointers");
-  }
-}
-
-void MarkSweep::reset_live_oop_tracking() {
-  if (ValidateMarkSweep) {
-    guarantee((size_t)_live_oops->length() == _live_oops_index, "should be at end of live oops");
-    _live_oops_index = 0;
-  }
-}
-
-void MarkSweep::register_live_oop(oop p, size_t size) {
-  if (ValidateMarkSweep) {
-    _live_oops->push(p);
-    _live_oops_size->push(size);
-    _live_oops_index++;
-  }
-}
-
-void MarkSweep::validate_live_oop(oop p, size_t size) {
-  if (ValidateMarkSweep) {
-    oop obj = _live_oops->at((int)_live_oops_index);
-    guarantee(obj == p, "should be the same object");
-    guarantee(_live_oops_size->at((int)_live_oops_index) == size, "should be the same size");
-    _live_oops_index++;
-  }
-}
-
-void MarkSweep::live_oop_moved_to(HeapWord* q, size_t size,
-                                  HeapWord* compaction_top) {
-  assert(oop(q)->forwardee() == NULL || oop(q)->forwardee() == oop(compaction_top),
-         "should be moved to forwarded location");
-  if (ValidateMarkSweep) {
-    MarkSweep::validate_live_oop(oop(q), size);
-    _live_oops_moved_to->push(oop(compaction_top));
-  }
-  if (RecordMarkSweepCompaction) {
-    _cur_gc_live_oops->push(q);
-    _cur_gc_live_oops_moved_to->push(compaction_top);
-    _cur_gc_live_oops_size->push(size);
-  }
-}
-
-void MarkSweep::compaction_complete() {
-  if (RecordMarkSweepCompaction) {
-    GrowableArray<HeapWord*>* _tmp_live_oops          = _cur_gc_live_oops;
-    GrowableArray<HeapWord*>* _tmp_live_oops_moved_to = _cur_gc_live_oops_moved_to;
-    GrowableArray<size_t>   * _tmp_live_oops_size     = _cur_gc_live_oops_size;
-
-    _cur_gc_live_oops           = _last_gc_live_oops;
-    _cur_gc_live_oops_moved_to  = _last_gc_live_oops_moved_to;
-    _cur_gc_live_oops_size      = _last_gc_live_oops_size;
-    _last_gc_live_oops          = _tmp_live_oops;
-    _last_gc_live_oops_moved_to = _tmp_live_oops_moved_to;
-    _last_gc_live_oops_size     = _tmp_live_oops_size;
-  }
-}
-
-void MarkSweep::print_new_location_of_heap_address(HeapWord* q) {
-  if (!RecordMarkSweepCompaction) {
-    tty->print_cr("Requires RecordMarkSweepCompaction to be enabled");
-    return;
-  }
-
-  if (_last_gc_live_oops == NULL) {
-    tty->print_cr("No compaction information gathered yet");
-    return;
-  }
-
-  for (int i = 0; i < _last_gc_live_oops->length(); i++) {
-    HeapWord* old_oop = _last_gc_live_oops->at(i);
-    size_t    sz      = _last_gc_live_oops_size->at(i);
-    if (old_oop <= q && q < (old_oop + sz)) {
-      HeapWord* new_oop = _last_gc_live_oops_moved_to->at(i);
-      size_t offset = (q - old_oop);
-      tty->print_cr("Address " PTR_FORMAT, q);
-      tty->print_cr(" Was in oop " PTR_FORMAT ", size " SIZE_FORMAT ", at offset " SIZE_FORMAT, old_oop, sz, offset);
-      tty->print_cr(" Now in oop " PTR_FORMAT ", actual address " PTR_FORMAT, new_oop, new_oop + offset);
-      return;
-    }
-  }
-
-  tty->print_cr("Address " PTR_FORMAT " not found in live oop information from last GC", q);
-}
-#endif //VALIDATE_MARK_SWEEP
-
 MarkSweep::IsAliveClosure   MarkSweep::is_alive;
 
 void MarkSweep::IsAliveClosure::do_object(oop p)   { ShouldNotReachHere(); }
--- a/src/share/vm/gc_implementation/shared/markSweep.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/shared/markSweep.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -44,21 +44,6 @@
 //
 // Class unloading will only occur when a full gc is invoked.
 
-// If VALIDATE_MARK_SWEEP is defined, the -XX:+ValidateMarkSweep flag will
-// be operational, and will provide slow but comprehensive self-checks within
-// the GC.  This is not enabled by default in product or release builds,
-// since the extra call to track_adjusted_pointer() in _adjust_pointer()
-// would be too much overhead, and would disturb performance measurement.
-// However, debug builds are sometimes way too slow to run GC tests!
-#ifdef ASSERT
-#define VALIDATE_MARK_SWEEP 1
-#endif
-#ifdef VALIDATE_MARK_SWEEP
-#define VALIDATE_MARK_SWEEP_ONLY(code) code
-#else
-#define VALIDATE_MARK_SWEEP_ONLY(code)
-#endif
-
 // declared at end
 class PreservedMark;
 
@@ -147,33 +132,6 @@
   // Reference processing (used in ...follow_contents)
   static ReferenceProcessor*             _ref_processor;
 
-#ifdef VALIDATE_MARK_SWEEP
-  static GrowableArray<void*>*           _root_refs_stack;
-  static GrowableArray<oop> *            _live_oops;
-  static GrowableArray<oop> *            _live_oops_moved_to;
-  static GrowableArray<size_t>*          _live_oops_size;
-  static size_t                          _live_oops_index;
-  static size_t                          _live_oops_index_at_perm;
-  static GrowableArray<void*>*           _other_refs_stack;
-  static GrowableArray<void*>*           _adjusted_pointers;
-  static bool                            _pointer_tracking;
-  static bool                            _root_tracking;
-
-  // The following arrays are saved since the time of the last GC and
-  // assist in tracking down problems where someone has done an errant
-  // store into the heap, usually to an oop that wasn't properly
-  // handleized across a GC. If we crash or otherwise fail before the
-  // next GC, we can query these arrays to find out the object we had
-  // intended to do the store to (assuming it is still alive) and the
-  // offset within that object. Covered under RecordMarkSweepCompaction.
-  static GrowableArray<HeapWord*> *      _cur_gc_live_oops;
-  static GrowableArray<HeapWord*> *      _cur_gc_live_oops_moved_to;
-  static GrowableArray<size_t>*          _cur_gc_live_oops_size;
-  static GrowableArray<HeapWord*> *      _last_gc_live_oops;
-  static GrowableArray<HeapWord*> *      _last_gc_live_oops_moved_to;
-  static GrowableArray<size_t>*          _last_gc_live_oops_size;
-#endif
-
   // Non public closures
   static KeepAliveClosure keep_alive;
 
@@ -227,24 +185,6 @@
   static void adjust_pointer(oop* p)       { adjust_pointer(p, false); }
   static void adjust_pointer(narrowOop* p) { adjust_pointer(p, false); }
 
-#ifdef VALIDATE_MARK_SWEEP
-  static void track_adjusted_pointer(void* p, bool isroot);
-  static void check_adjust_pointer(void* p);
-  static void track_interior_pointers(oop obj);
-  static void check_interior_pointers();
-
-  static void reset_live_oop_tracking();
-  static void register_live_oop(oop p, size_t size);
-  static void validate_live_oop(oop p, size_t size);
-  static void live_oop_moved_to(HeapWord* q, size_t size, HeapWord* compaction_top);
-  static void compaction_complete();
-
-  // Querying operation of RecordMarkSweepCompaction results.
-  // Finds and prints the current base oop and offset for a word
-  // within an oop that was live during the last GC. Helpful for
-  // tracking down heap stomps.
-  static void print_new_location_of_heap_address(HeapWord* q);
-#endif
 };
 
 class PreservedMark VALUE_OBJ_CLASS_SPEC {
--- a/src/share/vm/gc_implementation/shared/markSweep.inline.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_implementation/shared/markSweep.inline.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -46,12 +46,6 @@
 template <class T> inline void MarkSweep::follow_root(T* p) {
   assert(!Universe::heap()->is_in_reserved(p),
          "roots shouldn't be things within the heap");
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    guarantee(!_root_refs_stack->contains(p), "should only be in here once");
-    _root_refs_stack->push(p);
-  }
-#endif
   T heap_oop = oopDesc::load_heap_oop(p);
   if (!oopDesc::is_null(heap_oop)) {
     oop obj = oopDesc::decode_heap_oop_not_null(heap_oop);
@@ -97,19 +91,9 @@
       oopDesc::encode_store_heap_oop_not_null(p, new_obj);
     }
   }
-  VALIDATE_MARK_SWEEP_ONLY(track_adjusted_pointer(p, isroot));
 }
 
 template <class T> inline void MarkSweep::KeepAliveClosure::do_oop_work(T* p) {
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    if (!Universe::heap()->is_in_reserved(p)) {
-      _root_refs_stack->push(p);
-    } else {
-      _other_refs_stack->push(p);
-    }
-  }
-#endif
   mark_and_push(p);
 }
 
--- a/src/share/vm/gc_interface/gcCause.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/gc_interface/gcCause.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -99,9 +99,9 @@
  public:
    GCCauseString(const char* prefix, GCCause::Cause cause) {
      if (PrintGCCause) {
-      _position = jio_snprintf(_buffer, _length, "%s (%s)", prefix, GCCause::to_string(cause));
+      _position = jio_snprintf(_buffer, _length, "%s (%s) ", prefix, GCCause::to_string(cause));
      } else {
-      _position = jio_snprintf(_buffer, _length, "%s", prefix);
+      _position = jio_snprintf(_buffer, _length, "%s ", prefix);
      }
      assert(_position >= 0 && _position <= _length,
        err_msg("Need to increase the buffer size in GCCauseString? %d", _position));
--- a/src/share/vm/interpreter/interpreterRuntime.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/interpreter/interpreterRuntime.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -417,7 +417,7 @@
 
     // exception handler lookup
     KlassHandle h_klass(THREAD, h_exception->klass());
-    handler_bci = h_method->fast_exception_handler_bci_for(h_klass, current_bci, THREAD);
+    handler_bci = Method::fast_exception_handler_bci_for(h_method, h_klass, current_bci, THREAD);
     if (HAS_PENDING_EXCEPTION) {
       // We threw an exception while trying to find the exception handler.
       // Transfer the new exception to the exception handle which will
--- a/src/share/vm/interpreter/rewriter.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/interpreter/rewriter.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -27,13 +27,8 @@
 #include "interpreter/interpreter.hpp"
 #include "interpreter/rewriter.hpp"
 #include "memory/gcLocker.hpp"
-#include "memory/metadataFactory.hpp"
-#include "memory/oopFactory.hpp"
 #include "memory/resourceArea.hpp"
 #include "oops/generateOopMap.hpp"
-#include "oops/objArrayOop.hpp"
-#include "oops/oop.inline.hpp"
-#include "prims/methodComparator.hpp"
 #include "prims/methodHandles.hpp"
 
 // Computes a CPC map (new_index -> original_index) for constant pool entries
@@ -402,13 +397,6 @@
 }
 
 
-void Rewriter::rewrite(instanceKlassHandle klass, constantPoolHandle cpool, Array<Method*>* methods, TRAPS) {
-  ResourceMark rm(THREAD);
-  Rewriter     rw(klass, cpool, methods, CHECK);
-  // (That's all, folks.)
-}
-
-
 Rewriter::Rewriter(instanceKlassHandle klass, constantPoolHandle cpool, Array<Method*>* methods, TRAPS)
   : _klass(klass),
     _pool(cpool),
@@ -453,46 +441,25 @@
     restore_bytecodes();
     return;
   }
-}
 
-// Relocate jsr/rets in a method.  This can't be done with the rewriter
-// stage because it can throw other exceptions, leaving the bytecodes
-// pointing at constant pool cache entries.
-// Link and check jvmti dependencies while we're iterating over the methods.
-// JSR292 code calls with a different set of methods, so two entry points.
-void Rewriter::relocate_and_link(instanceKlassHandle this_oop, TRAPS) {
-  relocate_and_link(this_oop, this_oop->methods(), THREAD);
-}
-
-void Rewriter::relocate_and_link(instanceKlassHandle this_oop,
-                                 Array<Method*>* methods, TRAPS) {
-  int len = methods->length();
+  // Relocate after everything, but still do this under the is_rewritten flag,
+  // so methods with jsrs in custom class lists in aren't attempted to be
+  // rewritten in the RO section of the shared archive.
+  // Relocated bytecodes don't have to be restored, only the cp cache entries
   for (int i = len-1; i >= 0; i--) {
-    methodHandle m(THREAD, methods->at(i));
+    methodHandle m(THREAD, _methods->at(i));
 
     if (m->has_jsrs()) {
-      m = rewrite_jsrs(m, CHECK);
+      m = rewrite_jsrs(m, THREAD);
+      // Restore bytecodes to their unrewritten state if there are exceptions
+      // relocating bytecodes.  If some are relocated, that is ok because that
+      // doesn't affect constant pool to cpCache rewriting.
+      if (HAS_PENDING_EXCEPTION) {
+        restore_bytecodes();
+        return;
+      }
       // Method might have gotten rewritten.
       methods->at_put(i, m());
     }
-
-    // Set up method entry points for compiler and interpreter    .
-    m->link_method(m, CHECK);
-
-    // This is for JVMTI and unrelated to relocator but the last thing we do
-#ifdef ASSERT
-    if (StressMethodComparator) {
-      static int nmc = 0;
-      for (int j = i; j >= 0 && j >= i-4; j--) {
-        if ((++nmc % 1000) == 0)  tty->print_cr("Have run MethodComparator %d times...", nmc);
-        bool z = MethodComparator::methods_EMCP(m(),
-                   methods->at(j));
-        if (j == i && !z) {
-          tty->print("MethodComparator FAIL: "); m->print(); m->print_codes();
-          assert(z, "method must compare equal to itself");
-        }
-      }
-    }
-#endif //ASSERT
   }
 }
--- a/src/share/vm/interpreter/rewriter.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/interpreter/rewriter.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -158,14 +158,6 @@
  public:
   // Driver routine:
   static void rewrite(instanceKlassHandle klass, TRAPS);
-  static void rewrite(instanceKlassHandle klass, constantPoolHandle cpool, Array<Method*>* methods, TRAPS);
-
-  // Second pass, not gated by is_rewritten flag
-  static void relocate_and_link(instanceKlassHandle klass, TRAPS);
-  // JSR292 version to call with it's own methods.
-  static void relocate_and_link(instanceKlassHandle klass,
-                                Array<Method*>* methods, TRAPS);
-
 };
 
 #endif // SHARE_VM_INTERPRETER_REWRITER_HPP
--- a/src/share/vm/memory/binaryTreeDictionary.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/binaryTreeDictionary.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -67,7 +67,8 @@
 }
 
 template <class Chunk_t, template <class> class FreeList_t>
-TreeList<Chunk_t, FreeList_t>::TreeList() {}
+TreeList<Chunk_t, FreeList_t>::TreeList() : _parent(NULL),
+  _left(NULL), _right(NULL) {}
 
 template <class Chunk_t, template <class> class FreeList_t>
 TreeList<Chunk_t, FreeList_t>*
@@ -82,7 +83,7 @@
   tl->link_head(tc);
   tl->link_tail(tc);
   tl->set_count(1);
-
+  assert(tl->parent() == NULL, "Should be clear");
   return tl;
 }
 
--- a/src/share/vm/memory/collectorPolicy.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/collectorPolicy.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -777,6 +777,15 @@
                                        full_gc_count,
                                        GCCause::_metadata_GC_threshold);
     VMThread::execute(&op);
+
+    // If GC was locked out, try again.  Check
+    // before checking success because the prologue
+    // could have succeeded and the GC still have
+    // been locked out.
+    if (op.gc_locked()) {
+      continue;
+    }
+
     if (op.prologue_succeeded()) {
       return op.result();
     }
@@ -818,7 +827,7 @@
   if (_generations == NULL)
     vm_exit_during_initialization("Unable to allocate gen spec");
 
-  if (UseParNewGC && ParallelGCThreads > 0) {
+  if (UseParNewGC) {
     _generations[0] = new GenerationSpec(Generation::ParNew, _initial_gen0_size, _max_gen0_size);
   } else {
     _generations[0] = new GenerationSpec(Generation::DefNew, _initial_gen0_size, _max_gen0_size);
@@ -831,10 +840,9 @@
 
 void MarkSweepPolicy::initialize_gc_policy_counters() {
   // initialize the policy counters - 2 collectors, 3 generations
-  if (UseParNewGC && ParallelGCThreads > 0) {
+  if (UseParNewGC) {
     _gc_policy_counters = new GCPolicyCounters("ParNew:MSC", 2, 3);
-  }
-  else {
+  } else {
     _gc_policy_counters = new GCPolicyCounters("Copy:MSC", 2, 3);
   }
 }
--- a/src/share/vm/memory/filemap.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/filemap.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -119,6 +119,7 @@
   _header._magic = 0xf00baba2;
   _header._version = _current_version;
   _header._alignment = alignment;
+  _header._obj_alignment = ObjectAlignmentInBytes;
 
   // The following fields are for sanity checks for whether this archive
   // will function correctly with this JVM and the bootclasspath it's
@@ -211,7 +212,11 @@
 
   // Remove the existing file in case another process has it open.
   remove(_full_path);
+#ifdef _WINDOWS  // if 0444 is used on Windows, then remove() will fail.
+  int fd = open(_full_path, O_RDWR | O_CREAT | O_TRUNC | O_BINARY, 0744);
+#else
   int fd = open(_full_path, O_RDWR | O_CREAT | O_TRUNC | O_BINARY, 0444);
+#endif
   if (fd < 0) {
     fail_stop("Unable to create shared archive file %s.", _full_path);
   }
@@ -370,9 +375,8 @@
     return rs;
   }
   // the reserved virtual memory is for mapping class data sharing archive
-  if (MemTracker::is_on()) {
-    MemTracker::record_virtual_memory_type((address)rs.base(), mtClassShared);
-  }
+  MemTracker::record_virtual_memory_type((address)rs.base(), mtClassShared);
+
   return rs;
 }
 
@@ -394,6 +398,11 @@
     fail_continue(err_msg("Unable to map %s shared space at required address.", shared_region_name[i]));
     return NULL;
   }
+#ifdef _WINDOWS
+  // This call is Windows-only because the memory_type gets recorded for the other platforms
+  // in method FileMapInfo::reserve_shared_memory(), which is not called on Windows.
+  MemTracker::record_virtual_memory_type((address)base, mtClassShared);
+#endif
   return base;
 }
 
@@ -465,6 +474,12 @@
                   " version or build of HotSpot.");
     return false;
   }
+  if (_header._obj_alignment != ObjectAlignmentInBytes) {
+    fail_continue("The shared archive file's ObjectAlignmentInBytes of %d"
+                  " does not equal the current ObjectAlignmentInBytes of %d.",
+                  _header._obj_alignment, ObjectAlignmentInBytes);
+    return false;
+  }
 
   // Cannot verify interpreter yet, as it can only be created after the GC
   // heap has been initialized.
--- a/src/share/vm/memory/filemap.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/filemap.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -63,6 +63,7 @@
     int    _magic;                    // identify file type.
     int    _version;                  // (from enum, above.)
     size_t _alignment;                // how shared archive should be aligned
+    int    _obj_alignment;            // value of ObjectAlignmentInBytes
 
     struct space_info {
       int    _file_offset;   // sizeof(this) rounded to vm page size
--- a/src/share/vm/memory/genMarkSweep.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/genMarkSweep.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -100,21 +100,8 @@
 
   mark_sweep_phase3(level);
 
-  VALIDATE_MARK_SWEEP_ONLY(
-    if (ValidateMarkSweep) {
-      guarantee(_root_refs_stack->length() == 0, "should be empty by now");
-    }
-  )
-
   mark_sweep_phase4();
 
-  VALIDATE_MARK_SWEEP_ONLY(
-    if (ValidateMarkSweep) {
-      guarantee(_live_oops->length() == _live_oops_moved_to->length(),
-                "should be the same size");
-    }
-  )
-
   restore_marks();
 
   // Set saved marks for allocation profiler (and other things? -- dld)
@@ -187,31 +174,6 @@
 
   _preserved_marks = (PreservedMark*)scratch;
   _preserved_count = 0;
-
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    _root_refs_stack    = new (ResourceObj::C_HEAP, mtGC) GrowableArray<void*>(100, true);
-    _other_refs_stack   = new (ResourceObj::C_HEAP, mtGC) GrowableArray<void*>(100, true);
-    _adjusted_pointers  = new (ResourceObj::C_HEAP, mtGC) GrowableArray<void*>(100, true);
-    _live_oops          = new (ResourceObj::C_HEAP, mtGC) GrowableArray<oop>(100, true);
-    _live_oops_moved_to = new (ResourceObj::C_HEAP, mtGC) GrowableArray<oop>(100, true);
-    _live_oops_size     = new (ResourceObj::C_HEAP, mtGC) GrowableArray<size_t>(100, true);
-  }
-  if (RecordMarkSweepCompaction) {
-    if (_cur_gc_live_oops == NULL) {
-      _cur_gc_live_oops           = new(ResourceObj::C_HEAP, mtGC) GrowableArray<HeapWord*>(100, true);
-      _cur_gc_live_oops_moved_to  = new(ResourceObj::C_HEAP, mtGC) GrowableArray<HeapWord*>(100, true);
-      _cur_gc_live_oops_size      = new(ResourceObj::C_HEAP, mtGC) GrowableArray<size_t>(100, true);
-      _last_gc_live_oops          = new(ResourceObj::C_HEAP, mtGC) GrowableArray<HeapWord*>(100, true);
-      _last_gc_live_oops_moved_to = new(ResourceObj::C_HEAP, mtGC) GrowableArray<HeapWord*>(100, true);
-      _last_gc_live_oops_size     = new(ResourceObj::C_HEAP, mtGC) GrowableArray<size_t>(100, true);
-    } else {
-      _cur_gc_live_oops->clear();
-      _cur_gc_live_oops_moved_to->clear();
-      _cur_gc_live_oops_size->clear();
-    }
-  }
-#endif
 }
 
 
@@ -225,19 +187,6 @@
   _preserved_oop_stack.clear(true);
   _marking_stack.clear();
   _objarray_stack.clear(true);
-
-#ifdef VALIDATE_MARK_SWEEP
-  if (ValidateMarkSweep) {
-    delete _root_refs_stack;
-    delete _other_refs_stack;
-    delete _adjusted_pointers;
-    delete _live_oops;
-    delete _live_oops_size;
-    delete _live_oops_moved_to;
-    _live_oops_index = 0;
-    _live_oops_index_at_perm = 0;
-  }
-#endif
 }
 
 void GenMarkSweep::mark_sweep_phase1(int level,
@@ -246,8 +195,6 @@
   TraceTime tm("phase 1", PrintGC && Verbose, true, gclog_or_tty);
   trace(" 1");
 
-  VALIDATE_MARK_SWEEP_ONLY(reset_live_oop_tracking());
-
   GenCollectedHeap* gch = GenCollectedHeap::heap();
 
   // Because follow_root_closure is created statically, cannot
@@ -315,8 +262,6 @@
   TraceTime tm("phase 2", PrintGC && Verbose, true, gclog_or_tty);
   trace("2");
 
-  VALIDATE_MARK_SWEEP_ONLY(reset_live_oop_tracking());
-
   gch->prepare_for_compaction();
 }
 
@@ -337,8 +282,6 @@
   // Need new claim bits for the pointer adjustment tracing.
   ClassLoaderDataGraph::clear_claimed_marks();
 
-  VALIDATE_MARK_SWEEP_ONLY(reset_live_oop_tracking());
-
   // Because the two closures below are created statically, cannot
   // use OopsInGenClosure constructor which takes a generation,
   // as the Universe has not been created when the static constructors
@@ -393,10 +336,6 @@
   TraceTime tm("phase 4", PrintGC && Verbose, true, gclog_or_tty);
   trace("4");
 
-  VALIDATE_MARK_SWEEP_ONLY(reset_live_oop_tracking());
-
   GenCompactClosure blk;
   gch->generation_iterate(&blk, true);
-
-  VALIDATE_MARK_SWEEP_ONLY(compaction_complete());
 }
--- a/src/share/vm/memory/metachunk.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/metachunk.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -56,6 +56,7 @@
   assert(chunk_end > chunk_bottom, "Chunk must be too small");
   chunk->set_end(chunk_end);
   chunk->set_next(NULL);
+  chunk->set_prev(NULL);
   chunk->set_word_size(word_size);
 #ifdef ASSERT
   size_t data_word_size = pointer_delta(chunk_end, chunk_bottom, sizeof(MetaWord));
@@ -76,15 +77,15 @@
 }
 
 // _bottom points to the start of the chunk including the overhead.
-size_t Metachunk::used_word_size() {
+size_t Metachunk::used_word_size() const {
   return pointer_delta(_top, _bottom, sizeof(MetaWord));
 }
 
-size_t Metachunk::free_word_size() {
+size_t Metachunk::free_word_size() const {
   return pointer_delta(_end, _top, sizeof(MetaWord));
 }
 
-size_t Metachunk::capacity_word_size() {
+size_t Metachunk::capacity_word_size() const {
   return pointer_delta(_end, _bottom, sizeof(MetaWord));
 }
 
@@ -93,6 +94,10 @@
                " bottom " PTR_FORMAT " top " PTR_FORMAT
                " end " PTR_FORMAT " size " SIZE_FORMAT,
                bottom(), top(), end(), word_size());
+  if (Verbose) {
+    st->print_cr("    used " SIZE_FORMAT " free " SIZE_FORMAT,
+                 used_word_size(), free_word_size());
+  }
 }
 
 #ifndef PRODUCT
--- a/src/share/vm/memory/metachunk.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/metachunk.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -67,9 +67,11 @@
   void set_word_size(size_t v) { _word_size = v; }
  public:
 #ifdef ASSERT
-  Metachunk() : _bottom(NULL), _end(NULL), _top(NULL), _is_free(false) {}
+  Metachunk() : _bottom(NULL), _end(NULL), _top(NULL), _is_free(false),
+    _next(NULL), _prev(NULL) {}
 #else
-  Metachunk() : _bottom(NULL), _end(NULL), _top(NULL) {}
+  Metachunk() : _bottom(NULL), _end(NULL), _top(NULL),
+    _next(NULL), _prev(NULL) {}
 #endif
 
   // Used to add a Metachunk to a list of Metachunks
@@ -102,15 +104,15 @@
   }
 
   // Reset top to bottom so chunk can be reused.
-  void reset_empty() { _top = (_bottom + _overhead); }
+  void reset_empty() { _top = (_bottom + _overhead); _next = NULL; _prev = NULL; }
   bool is_empty() { return _top == (_bottom + _overhead); }
 
   // used (has been allocated)
   // free (available for future allocations)
   // capacity (total size of chunk)
-  size_t used_word_size();
-  size_t free_word_size();
-  size_t capacity_word_size();
+  size_t used_word_size() const;
+  size_t free_word_size() const;
+  size_t capacity_word_size()const;
 
   // Debug support
 #ifdef ASSERT
--- a/src/share/vm/memory/metaspace.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/metaspace.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -58,11 +58,23 @@
 
 // Used in declarations in SpaceManager and ChunkManager
 enum ChunkIndex {
-  SmallIndex = 0,
-  MediumIndex = 1,
-  HumongousIndex = 2,
-  NumberOfFreeLists = 2,
-  NumberOfInUseLists = 3
+  ZeroIndex = 0,
+  SpecializedIndex = ZeroIndex,
+  SmallIndex = SpecializedIndex + 1,
+  MediumIndex = SmallIndex + 1,
+  HumongousIndex = MediumIndex + 1,
+  NumberOfFreeLists = 3,
+  NumberOfInUseLists = 4
+};
+
+enum ChunkSizes {    // in words.
+  ClassSpecializedChunk = 128,
+  SpecializedChunk = 128,
+  ClassSmallChunk = 256,
+  SmallChunk = 512,
+  ClassMediumChunk = 1 * K,
+  MediumChunk = 8 * K,
+  HumongousChunkGranularity = 8
 };
 
 static ChunkIndex next_chunk_index(ChunkIndex i) {
@@ -126,6 +138,7 @@
   //   HumongousChunk
   ChunkList _free_chunks[NumberOfFreeLists];
 
+
   //   HumongousChunk
   ChunkTreeDictionary _humongous_dictionary;
 
@@ -169,6 +182,10 @@
   Metachunk* chunk_freelist_allocate(size_t word_size);
   void chunk_freelist_deallocate(Metachunk* chunk);
 
+  // Map a size to a list index assuming that there are lists
+  // for special, small, medium, and humongous chunks.
+  static ChunkIndex list_index(size_t size);
+
   // Total of the space in the free chunks list
   size_t free_chunks_total();
   size_t free_chunks_total_in_bytes();
@@ -180,8 +197,6 @@
     Atomic::add_ptr(count, &_free_chunks_count);
     Atomic::add_ptr(v, &_free_chunks_total);
   }
-  ChunkList* free_medium_chunks() { return &_free_chunks[1]; }
-  ChunkList* free_small_chunks() { return &_free_chunks[0]; }
   ChunkTreeDictionary* humongous_dictionary() {
     return &_humongous_dictionary;
   }
@@ -400,7 +415,14 @@
   VirtualSpaceList(size_t word_size);
   VirtualSpaceList(ReservedSpace rs);
 
-  Metachunk* get_new_chunk(size_t word_size, size_t grow_chunks_by_words);
+  Metachunk* get_new_chunk(size_t word_size,
+                           size_t grow_chunks_by_words,
+                           size_t medium_chunk_bunch);
+
+  // Get the first chunk for a Metaspace.  Used for
+  // special cases such as the boot class loader, reflection
+  // class loader and anonymous class loader.
+  Metachunk* get_initialization_chunk(size_t word_size, size_t chunk_bunch);
 
   VirtualSpaceNode* current_virtual_space() {
     return _current_virtual_space;
@@ -501,9 +523,13 @@
   friend class Metadebug;
 
  private:
+
   // protects allocations and contains.
   Mutex* const _lock;
 
+  // Chunk related size
+  size_t _medium_chunk_bunch;
+
   // List of chunks in use by this SpaceManager.  Allocations
   // are done from the current chunk.  The list is used for deallocating
   // chunks when the SpaceManager is freed.
@@ -532,6 +558,7 @@
   static const int    _expand_lock_rank;
   static Mutex* const _expand_lock;
 
+ private:
   // Accessors
   Metachunk* chunks_in_use(ChunkIndex index) const { return _chunks_in_use[index]; }
   void set_chunks_in_use(ChunkIndex index, Metachunk* v) { _chunks_in_use[index] = v; }
@@ -554,23 +581,37 @@
 
   Mutex* lock() const { return _lock; }
 
+  const char* chunk_size_name(ChunkIndex index) const;
+
+ protected:
+  void initialize();
+
  public:
-  SpaceManager(Mutex* lock, VirtualSpaceList* vs_list);
+  SpaceManager(Mutex* lock,
+               VirtualSpaceList* vs_list);
   ~SpaceManager();
 
-  enum ChunkSizes {    // in words.
-    SmallChunk = 512,
-    MediumChunk = 8 * K,
-    MediumChunkBunch = 4 * MediumChunk
+  enum ChunkMultiples {
+    MediumChunkMultiple = 4
   };
 
   // Accessors
+  size_t specialized_chunk_size() { return SpecializedChunk; }
+  size_t small_chunk_size() { return (size_t) vs_list()->is_class() ? ClassSmallChunk : SmallChunk; }
+  size_t medium_chunk_size() { return (size_t) vs_list()->is_class() ? ClassMediumChunk : MediumChunk; }
+  size_t medium_chunk_bunch() { return medium_chunk_size() * MediumChunkMultiple; }
+
   size_t allocation_total() const { return _allocation_total; }
   void inc_allocation_total(size_t v) { Atomic::add_ptr(v, &_allocation_total); }
-  static bool is_humongous(size_t word_size) { return word_size > MediumChunk; }
+  bool is_humongous(size_t word_size) { return word_size > medium_chunk_size(); }
 
   static Mutex* expand_lock() { return _expand_lock; }
 
+  // Set the sizes for the initial chunks.
+  void get_initial_chunk_sizes(Metaspace::MetaspaceType type,
+                               size_t* chunk_word_size,
+                               size_t* class_chunk_word_size);
+
   size_t sum_capacity_in_chunks_in_use() const;
   size_t sum_used_in_chunks_in_use() const;
   size_t sum_free_in_chunks_in_use() const;
@@ -580,6 +621,8 @@
   size_t sum_count_in_chunks_in_use();
   size_t sum_count_in_chunks_in_use(ChunkIndex i);
 
+  Metachunk* get_new_chunk(size_t word_size, size_t grow_chunks_by_words);
+
   // Block allocation and deallocation.
   // Allocates a block from the current chunk
   MetaWord* allocate(size_t word_size);
@@ -772,8 +815,10 @@
     return false;
   }
 
-  // Commit only 1 page instead of the whole reserved space _rs.size()
-  size_t committed_byte_size = os::vm_page_size();
+  // An allocation out of this Virtualspace that is larger
+  // than an initial commit size can waste that initial committed
+  // space.
+  size_t committed_byte_size = 0;
   bool result = virtual_space()->initialize(_rs, committed_byte_size);
   if (result) {
     set_top((MetaWord*)virtual_space()->low());
@@ -799,7 +844,8 @@
   st->print_cr("   space @ " PTR_FORMAT " " SIZE_FORMAT "K, %3d%% used "
            "[" PTR_FORMAT ", " PTR_FORMAT ", "
            PTR_FORMAT ", " PTR_FORMAT ")",
-           vs, capacity / K, used * 100 / capacity,
+           vs, capacity / K,
+           capacity == 0 ? 0 : used * 100 / capacity,
            bottom(), top(), end(),
            vs->high_boundary());
 }
@@ -922,7 +968,8 @@
 }
 
 Metachunk* VirtualSpaceList::get_new_chunk(size_t word_size,
-                                           size_t grow_chunks_by_words) {
+                                           size_t grow_chunks_by_words,
+                                           size_t medium_chunk_bunch) {
 
   // Get a chunk from the chunk freelist
   Metachunk* next = chunk_manager()->chunk_freelist_allocate(grow_chunks_by_words);
@@ -935,8 +982,8 @@
   if (next == NULL) {
     // Not enough room in current virtual space.  Try to commit
     // more space.
-    size_t expand_vs_by_words = MAX2((size_t)SpaceManager::MediumChunkBunch,
-                                       grow_chunks_by_words);
+    size_t expand_vs_by_words = MAX2(medium_chunk_bunch,
+                                     grow_chunks_by_words);
     size_t page_size_words = os::vm_page_size() / BytesPerWord;
     size_t aligned_expand_vs_by_words = align_size_up(expand_vs_by_words,
                                                         page_size_words);
@@ -954,12 +1001,6 @@
           // Got it.  It's on the list now.  Get a chunk from it.
           next = current_virtual_space()->get_chunk_vs_with_expand(grow_chunks_by_words);
         }
-        if (TraceMetadataHumongousAllocation && SpaceManager::is_humongous(word_size)) {
-          gclog_or_tty->print_cr("  aligned_expand_vs_by_words " PTR_FORMAT,
-                                 aligned_expand_vs_by_words);
-          gclog_or_tty->print_cr("  grow_vs_words " PTR_FORMAT,
-                                 grow_vs_words);
-        }
       } else {
         // Allocation will fail and induce a GC
         if (TraceMetadataChunkAllocation && Verbose) {
@@ -974,9 +1015,20 @@
     }
   }
 
+  assert(next == NULL || (next->next() == NULL && next->prev() == NULL),
+         "New chunk is still on some list");
   return next;
 }
 
+Metachunk* VirtualSpaceList::get_initialization_chunk(size_t chunk_word_size,
+                                                      size_t chunk_bunch) {
+  // Get a chunk from the chunk freelist
+  Metachunk* new_chunk = get_new_chunk(chunk_word_size,
+                                       chunk_word_size,
+                                       chunk_bunch);
+  return new_chunk;
+}
+
 void VirtualSpaceList::print_on(outputStream* st) const {
   if (TraceMetadataChunkAllocation && Verbose) {
     VirtualSpaceListIterator iter(virtual_space_list());
@@ -1373,16 +1425,17 @@
 
 void ChunkList::add_at_head(Metachunk* head, Metachunk* tail) {
   assert_lock_strong(SpaceManager::expand_lock());
-  assert(tail->next() == NULL, "Not the tail");
+  assert(head == tail || tail->next() == NULL,
+         "Not the tail or the head has already been added to a list");
 
   if (TraceMetadataChunkAllocation && Verbose) {
-    tty->print("ChunkList::add_at_head: ");
+    gclog_or_tty->print("ChunkList::add_at_head(head, tail): ");
     Metachunk* cur = head;
     while (cur != NULL) {
-    tty->print(PTR_FORMAT " (" SIZE_FORMAT ") ", cur, cur->word_size());
+      gclog_or_tty->print(PTR_FORMAT " (" SIZE_FORMAT ") ", cur, cur->word_size());
       cur = cur->next();
     }
-    tty->print_cr("");
+    gclog_or_tty->print_cr("");
   }
 
   if (tail != NULL) {
@@ -1486,13 +1539,13 @@
 
 void ChunkManager::locked_print_free_chunks(outputStream* st) {
   assert_lock_strong(SpaceManager::expand_lock());
-  st->print_cr("Free chunk total 0x%x  count 0x%x",
+  st->print_cr("Free chunk total " SIZE_FORMAT "  count " SIZE_FORMAT,
                 _free_chunks_total, _free_chunks_count);
 }
 
 void ChunkManager::locked_print_sum_free_chunks(outputStream* st) {
   assert_lock_strong(SpaceManager::expand_lock());
-  st->print_cr("Sum free chunk total 0x%x  count 0x%x",
+  st->print_cr("Sum free chunk total " SIZE_FORMAT "  count " SIZE_FORMAT,
                 sum_free_chunks(), sum_free_chunks_count());
 }
 ChunkList* ChunkManager::free_chunks(ChunkIndex index) {
@@ -1504,7 +1557,7 @@
 size_t ChunkManager::sum_free_chunks() {
   assert_lock_strong(SpaceManager::expand_lock());
   size_t result = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfFreeLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfFreeLists; i = next_chunk_index(i)) {
     ChunkList* list = free_chunks(i);
 
     if (list == NULL) {
@@ -1520,7 +1573,7 @@
 size_t ChunkManager::sum_free_chunks_count() {
   assert_lock_strong(SpaceManager::expand_lock());
   size_t count = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfFreeLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfFreeLists; i = next_chunk_index(i)) {
     ChunkList* list = free_chunks(i);
     if (list == NULL) {
       continue;
@@ -1532,15 +1585,9 @@
 }
 
 ChunkList* ChunkManager::find_free_chunks_list(size_t word_size) {
-  switch (word_size) {
-  case SpaceManager::SmallChunk :
-      return &_free_chunks[0];
-  case SpaceManager::MediumChunk :
-      return &_free_chunks[1];
-  default:
-    assert(word_size > SpaceManager::MediumChunk, "List inconsistency");
-    return &_free_chunks[2];
-  }
+  ChunkIndex index = list_index(word_size);
+  assert(index < HumongousIndex, "No humongous list");
+  return free_chunks(index);
 }
 
 void ChunkManager::free_chunks_put(Metachunk* chunk) {
@@ -1574,7 +1621,7 @@
   slow_locked_verify();
 
   Metachunk* chunk = NULL;
-  if (!SpaceManager::is_humongous(word_size)) {
+  if (list_index(word_size) != HumongousIndex) {
     ChunkList* free_list = find_free_chunks_list(word_size);
     assert(free_list != NULL, "Sanity check");
 
@@ -1587,8 +1634,8 @@
 
     // Remove the chunk as the head of the list.
     free_list->set_head(chunk->next());
-    chunk->set_next(NULL);
-    // Chunk has been removed from the chunks free list.
+
+    // Chunk is being removed from the chunks free list.
     dec_free_chunks_total(chunk->capacity_word_size());
 
     if (TraceMetadataChunkAllocation && Verbose) {
@@ -1614,8 +1661,14 @@
 #ifdef ASSERT
       chunk->set_is_free(false);
 #endif
+    } else {
+      return NULL;
     }
   }
+
+  // Remove it from the links to this freelist
+  chunk->set_next(NULL);
+  chunk->set_prev(NULL);
   slow_locked_verify();
   return chunk;
 }
@@ -1630,13 +1683,20 @@
     return NULL;
   }
 
-  assert(word_size <= chunk->word_size() ||
-           SpaceManager::is_humongous(chunk->word_size()),
-           "Non-humongous variable sized chunk");
+  assert((word_size <= chunk->word_size()) ||
+         list_index(chunk->word_size() == HumongousIndex),
+         "Non-humongous variable sized chunk");
   if (TraceMetadataChunkAllocation) {
-    tty->print("ChunkManager::chunk_freelist_allocate: chunk "
-               PTR_FORMAT "  size " SIZE_FORMAT " ",
-               chunk, chunk->word_size());
+    size_t list_count;
+    if (list_index(word_size) < HumongousIndex) {
+      ChunkList* list = find_free_chunks_list(word_size);
+      list_count = list->sum_list_count();
+    } else {
+      list_count = humongous_dictionary()->total_count();
+    }
+    tty->print("ChunkManager::chunk_freelist_allocate: " PTR_FORMAT " chunk "
+               PTR_FORMAT "  size " SIZE_FORMAT " count " SIZE_FORMAT " ",
+               this, chunk, chunk->word_size(), list_count);
     locked_print_free_chunks(tty);
   }
 
@@ -1651,10 +1711,42 @@
 
 // SpaceManager methods
 
+void SpaceManager::get_initial_chunk_sizes(Metaspace::MetaspaceType type,
+                                           size_t* chunk_word_size,
+                                           size_t* class_chunk_word_size) {
+  switch (type) {
+  case Metaspace::BootMetaspaceType:
+    *chunk_word_size = Metaspace::first_chunk_word_size();
+    *class_chunk_word_size = Metaspace::first_class_chunk_word_size();
+    break;
+  case Metaspace::ROMetaspaceType:
+    *chunk_word_size = SharedReadOnlySize / wordSize;
+    *class_chunk_word_size = ClassSpecializedChunk;
+    break;
+  case Metaspace::ReadWriteMetaspaceType:
+    *chunk_word_size = SharedReadWriteSize / wordSize;
+    *class_chunk_word_size = ClassSpecializedChunk;
+    break;
+  case Metaspace::AnonymousMetaspaceType:
+  case Metaspace::ReflectionMetaspaceType:
+    *chunk_word_size = SpecializedChunk;
+    *class_chunk_word_size = ClassSpecializedChunk;
+    break;
+  default:
+    *chunk_word_size = SmallChunk;
+    *class_chunk_word_size = ClassSmallChunk;
+    break;
+  }
+  assert(chunk_word_size != 0 && class_chunk_word_size != 0,
+    err_msg("Initial chunks sizes bad: data  " SIZE_FORMAT
+            " class " SIZE_FORMAT,
+            chunk_word_size, class_chunk_word_size));
+}
+
 size_t SpaceManager::sum_free_in_chunks_in_use() const {
   MutexLockerEx cl(lock(), Mutex::_no_safepoint_check_flag);
   size_t free = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
     Metachunk* chunk = chunks_in_use(i);
     while (chunk != NULL) {
       free += chunk->free_word_size();
@@ -1667,9 +1759,7 @@
 size_t SpaceManager::sum_waste_in_chunks_in_use() const {
   MutexLockerEx cl(lock(), Mutex::_no_safepoint_check_flag);
   size_t result = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
-
-
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
    result += sum_waste_in_chunks_in_use(i);
   }
 
@@ -1678,7 +1768,6 @@
 
 size_t SpaceManager::sum_waste_in_chunks_in_use(ChunkIndex index) const {
   size_t result = 0;
-  size_t count = 0;
   Metachunk* chunk = chunks_in_use(index);
   // Count the free space in all the chunk but not the
   // current chunk from which allocations are still being done.
@@ -1688,7 +1777,6 @@
       result += chunk->free_word_size();
       prev = chunk;
       chunk = chunk->next();
-      count++;
     }
   }
   return result;
@@ -1697,7 +1785,7 @@
 size_t SpaceManager::sum_capacity_in_chunks_in_use() const {
   MutexLockerEx cl(lock(), Mutex::_no_safepoint_check_flag);
   size_t sum = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
     Metachunk* chunk = chunks_in_use(i);
     while (chunk != NULL) {
       // Just changed this sum += chunk->capacity_word_size();
@@ -1711,7 +1799,7 @@
 
 size_t SpaceManager::sum_count_in_chunks_in_use() {
   size_t count = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
     count = count + sum_count_in_chunks_in_use(i);
   }
 
@@ -1732,7 +1820,7 @@
 size_t SpaceManager::sum_used_in_chunks_in_use() const {
   MutexLockerEx cl(lock(), Mutex::_no_safepoint_check_flag);
   size_t used = 0;
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
     Metachunk* chunk = chunks_in_use(i);
     while (chunk != NULL) {
       used += chunk->used_word_size();
@@ -1744,19 +1832,17 @@
 
 void SpaceManager::locked_print_chunks_in_use_on(outputStream* st) const {
 
-  Metachunk* small_chunk = chunks_in_use(SmallIndex);
-  st->print_cr("SpaceManager: small chunk " PTR_FORMAT
-               " free " SIZE_FORMAT,
-               small_chunk,
-               small_chunk->free_word_size());
-
-  Metachunk* medium_chunk = chunks_in_use(MediumIndex);
-  st->print("medium chunk " PTR_FORMAT, medium_chunk);
-  Metachunk* tail = current_chunk();
-  st->print_cr(" current chunk " PTR_FORMAT, tail);
-
-  Metachunk* head = chunks_in_use(HumongousIndex);
-  st->print_cr("humongous chunk " PTR_FORMAT, head);
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+    Metachunk* chunk = chunks_in_use(i);
+    st->print("SpaceManager: %s " PTR_FORMAT,
+                 chunk_size_name(i), chunk);
+    if (chunk != NULL) {
+      st->print_cr(" free " SIZE_FORMAT,
+                   chunk->free_word_size());
+    } else {
+      st->print_cr("");
+    }
+  }
 
   vs_list()->chunk_manager()->locked_print_free_chunks(st);
   vs_list()->chunk_manager()->locked_print_sum_free_chunks(st);
@@ -1772,18 +1858,28 @@
   if (chunks_in_use(MediumIndex) == NULL &&
       (!has_small_chunk_limit() ||
        sum_count_in_chunks_in_use(SmallIndex) < _small_chunk_limit)) {
-    chunk_word_size = (size_t) SpaceManager::SmallChunk;
-    if (word_size + Metachunk::overhead() > SpaceManager::SmallChunk) {
-      chunk_word_size = MediumChunk;
+    chunk_word_size = (size_t) small_chunk_size();
+    if (word_size + Metachunk::overhead() > small_chunk_size()) {
+      chunk_word_size = medium_chunk_size();
     }
   } else {
-    chunk_word_size = MediumChunk;
+    chunk_word_size = medium_chunk_size();
   }
 
-  // Might still need a humongous chunk
+  // Might still need a humongous chunk.  Enforce an
+  // eight word granularity to facilitate reuse (some
+  // wastage but better chance of reuse).
+  size_t if_humongous_sized_chunk =
+    align_size_up(word_size + Metachunk::overhead(),
+                  HumongousChunkGranularity);
   chunk_word_size =
-    MAX2((size_t) chunk_word_size, word_size + Metachunk::overhead());
-
+    MAX2((size_t) chunk_word_size, if_humongous_sized_chunk);
+
+  assert(!SpaceManager::is_humongous(word_size) ||
+         chunk_word_size == if_humongous_sized_chunk,
+         err_msg("Size calculation is wrong, word_size " SIZE_FORMAT
+                 " chunk_word_size " SIZE_FORMAT,
+                 word_size, chunk_word_size));
   if (TraceMetadataHumongousAllocation &&
       SpaceManager::is_humongous(word_size)) {
     gclog_or_tty->print_cr("Metadata humongous allocation:");
@@ -1805,15 +1901,21 @@
   MutexLockerEx cl(SpaceManager::expand_lock(), Mutex::_no_safepoint_check_flag);
 
   if (TraceMetadataChunkAllocation && Verbose) {
+    size_t words_left = 0;
+    size_t words_used = 0;
+    if (current_chunk() != NULL) {
+      words_left = current_chunk()->free_word_size();
+      words_used = current_chunk()->used_word_size();
+    }
     gclog_or_tty->print_cr("SpaceManager::grow_and_allocate for " SIZE_FORMAT
-                           " words " SIZE_FORMAT " space left",
-                            word_size, current_chunk() != NULL ?
-                              current_chunk()->free_word_size() : 0);
+                           " words " SIZE_FORMAT " words used " SIZE_FORMAT
+                           " words left",
+                            word_size, words_used, words_left);
   }
 
   // Get another chunk out of the virtual space
   size_t grow_chunks_by_words = calc_chunk_size(word_size);
-  Metachunk* next = vs_list()->get_new_chunk(word_size, grow_chunks_by_words);
+  Metachunk* next = get_new_chunk(word_size, grow_chunks_by_words);
 
   // If a chunk was available, add it to the in-use chunk list
   // and do an allocation from it.
@@ -1828,7 +1930,7 @@
 
 void SpaceManager::print_on(outputStream* st) const {
 
-  for (ChunkIndex i = SmallIndex;
+  for (ChunkIndex i = ZeroIndex;
        i < NumberOfInUseLists ;
        i = next_chunk_index(i) ) {
     st->print_cr("  chunks_in_use " PTR_FORMAT " chunk size " PTR_FORMAT,
@@ -1847,12 +1949,18 @@
   }
 }
 
-SpaceManager::SpaceManager(Mutex* lock, VirtualSpaceList* vs_list) :
+SpaceManager::SpaceManager(Mutex* lock,
+                           VirtualSpaceList* vs_list) :
   _vs_list(vs_list),
   _allocation_total(0),
-  _lock(lock) {
+  _lock(lock)
+{
+  initialize();
+}
+
+void SpaceManager::initialize() {
   Metadebug::init_allocation_fail_alot_count();
-  for (ChunkIndex i = SmallIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+  for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
     _chunks_in_use[i] = NULL;
   }
   _current_chunk = NULL;
@@ -1885,30 +1993,37 @@
   // Add all the chunks in use by this space manager
   // to the global list of free chunks.
 
-  // Small chunks.  There is one _current_chunk for each
-  // Metaspace.  It could point to a small or medium chunk.
-  // Rather than determine which it is, follow the list of
-  // small chunks to add them to the free list
-  Metachunk* small_chunk = chunks_in_use(SmallIndex);
-  chunk_manager->free_small_chunks()->add_at_head(small_chunk);
-  set_chunks_in_use(SmallIndex, NULL);
-
-  // After the small chunk are the medium chunks
-  Metachunk* medium_chunk = chunks_in_use(MediumIndex);
-  assert(medium_chunk == NULL ||
-         medium_chunk->word_size() == MediumChunk,
-         "Chunk is on the wrong list");
-
-  if (medium_chunk != NULL) {
-    Metachunk* head = medium_chunk;
-    // If there is a medium chunk then the _current_chunk can only
-    // point to the last medium chunk.
-    Metachunk* tail = current_chunk();
-    chunk_manager->free_medium_chunks()->add_at_head(head, tail);
-    set_chunks_in_use(MediumIndex, NULL);
+  // Follow each list of chunks-in-use and add them to the
+  // free lists.  Each list is NULL terminated.
+
+  for (ChunkIndex i = ZeroIndex; i < HumongousIndex; i = next_chunk_index(i)) {
+    if (TraceMetadataChunkAllocation && Verbose) {
+      gclog_or_tty->print_cr("returned %d %s chunks to freelist",
+                             sum_count_in_chunks_in_use(i),
+                             chunk_size_name(i));
+    }
+    Metachunk* chunks = chunks_in_use(i);
+    chunk_manager->free_chunks(i)->add_at_head(chunks);
+    set_chunks_in_use(i, NULL);
+    if (TraceMetadataChunkAllocation && Verbose) {
+      gclog_or_tty->print_cr("updated freelist count %d %s",
+                             chunk_manager->free_chunks(i)->sum_list_count(),
+                             chunk_size_name(i));
+    }
+    assert(i != HumongousIndex, "Humongous chunks are handled explicitly later");
   }
 
+  // The medium chunk case may be optimized by passing the head and
+  // tail of the medium chunk list to add_at_head().  The tail is often
+  // the current chunk but there are probably exceptions.
+
   // Humongous chunks
+  if (TraceMetadataChunkAllocation && Verbose) {
+    gclog_or_tty->print_cr("returned %d %s humongous chunks to dictionary",
+                            sum_count_in_chunks_in_use(HumongousIndex),
+                            chunk_size_name(HumongousIndex));
+    gclog_or_tty->print("Humongous chunk dictionary: ");
+  }
   // Humongous chunks are never the current chunk.
   Metachunk* humongous_chunks = chunks_in_use(HumongousIndex);
 
@@ -1916,14 +2031,65 @@
 #ifdef ASSERT
     humongous_chunks->set_is_free(true);
 #endif
+    if (TraceMetadataChunkAllocation && Verbose) {
+      gclog_or_tty->print(PTR_FORMAT " (" SIZE_FORMAT ") ",
+                          humongous_chunks,
+                          humongous_chunks->word_size());
+    }
+    assert(humongous_chunks->word_size() == (size_t)
+           align_size_up(humongous_chunks->word_size(),
+                             HumongousChunkGranularity),
+           err_msg("Humongous chunk size is wrong: word size " SIZE_FORMAT
+                   " granularity " SIZE_FORMAT,
+                   humongous_chunks->word_size(), HumongousChunkGranularity));
     Metachunk* next_humongous_chunks = humongous_chunks->next();
     chunk_manager->humongous_dictionary()->return_chunk(humongous_chunks);
     humongous_chunks = next_humongous_chunks;
   }
+  if (TraceMetadataChunkAllocation && Verbose) {
+    gclog_or_tty->print_cr("");
+    gclog_or_tty->print_cr("updated dictionary count %d %s",
+                     chunk_manager->humongous_dictionary()->total_count(),
+                     chunk_size_name(HumongousIndex));
+  }
   set_chunks_in_use(HumongousIndex, NULL);
   chunk_manager->slow_locked_verify();
 }
 
+const char* SpaceManager::chunk_size_name(ChunkIndex index) const {
+  switch (index) {
+    case SpecializedIndex:
+      return "Specialized";
+    case SmallIndex:
+      return "Small";
+    case MediumIndex:
+      return "Medium";
+    case HumongousIndex:
+      return "Humongous";
+    default:
+      return NULL;
+  }
+}
+
+ChunkIndex ChunkManager::list_index(size_t size) {
+  switch (size) {
+    case SpecializedChunk:
+      assert(SpecializedChunk == ClassSpecializedChunk,
+             "Need branch for ClassSpecializedChunk");
+      return SpecializedIndex;
+    case SmallChunk:
+    case ClassSmallChunk:
+      return SmallIndex;
+    case MediumChunk:
+    case ClassMediumChunk:
+      return MediumIndex;
+    default:
+      assert(size > MediumChunk || size > ClassMediumChunk,
+             "Not a humongous chunk");
+      return HumongousIndex;
+  }
+}
+
 void SpaceManager::deallocate(MetaWord* p, size_t word_size) {
   assert_lock_strong(_lock);
   size_t min_size = TreeChunk<Metablock, FreeList>::min_size();
@@ -1942,52 +2108,13 @@
 
   // Find the correct list and and set the current
   // chunk for that list.
-  switch (new_chunk->word_size()) {
-  case SpaceManager::SmallChunk :
-    if (chunks_in_use(SmallIndex) == NULL) {
-      // First chunk to add to the list
-      set_chunks_in_use(SmallIndex, new_chunk);
-    } else {
-      assert(current_chunk()->word_size() == SpaceManager::SmallChunk,
-        err_msg( "Incorrect mix of sizes in chunk list "
-        SIZE_FORMAT " new chunk " SIZE_FORMAT,
-        current_chunk()->word_size(), new_chunk->word_size()));
-      current_chunk()->set_next(new_chunk);
-    }
-    // Make current chunk
+  ChunkIndex index = ChunkManager::list_index(new_chunk->word_size());
+
+  if (index != HumongousIndex) {
     set_current_chunk(new_chunk);
-    break;
-  case SpaceManager::MediumChunk :
-    if (chunks_in_use(MediumIndex) == NULL) {
-      // About to add the first medium chunk so teminate the
-      // small chunk list.  In general once medium chunks are
-      // being added, we're past the need for small chunks.
-      if (current_chunk() != NULL) {
-        // Only a small chunk or the initial chunk could be
-        // the current chunk if this is the first medium chunk.
-        assert(current_chunk()->word_size() == SpaceManager::SmallChunk ||
-          chunks_in_use(SmallIndex) == NULL,
-          err_msg("Should be a small chunk or initial chunk, current chunk "
-          SIZE_FORMAT " new chunk " SIZE_FORMAT,
-          current_chunk()->word_size(), new_chunk->word_size()));
-        current_chunk()->set_next(NULL);
-      }
-      // First chunk to add to the list
-      set_chunks_in_use(MediumIndex, new_chunk);
-
-    } else {
-      // As a minimum the first medium chunk added would
-      // have become the _current_chunk
-      // so the _current_chunk has to be non-NULL here
-      // (although not necessarily still the first medium chunk).
-      assert(current_chunk()->word_size() == SpaceManager::MediumChunk,
-             "A medium chunk should the current chunk");
-      current_chunk()->set_next(new_chunk);
-    }
-    // Make current chunk
-    set_current_chunk(new_chunk);
-    break;
-  default: {
+    new_chunk->set_next(chunks_in_use(index));
+    set_chunks_in_use(index, new_chunk);
+  } else {
     // For null class loader data and DumpSharedSpaces, the first chunk isn't
     // small, so small will be null.  Link this first chunk as the current
     // chunk.
@@ -2002,8 +2129,7 @@
     new_chunk->set_next(chunks_in_use(HumongousIndex));
     set_chunks_in_use(HumongousIndex, new_chunk);
 
-    assert(new_chunk->word_size() > MediumChunk, "List inconsistency");
-  }
+    assert(new_chunk->word_size() > medium_chunk_size(), "List inconsistency");
   }
 
   assert(new_chunk->is_empty(), "Not ready for reuse");
@@ -2015,6 +2141,22 @@
   }
 }
 
+Metachunk* SpaceManager::get_new_chunk(size_t word_size,
+                                       size_t grow_chunks_by_words) {
+
+  Metachunk* next = vs_list()->get_new_chunk(word_size,
+                                             grow_chunks_by_words,
+                                             medium_chunk_bunch());
+
+  if (TraceMetadataHumongousAllocation &&
+      SpaceManager::is_humongous(next->word_size())) {
+    gclog_or_tty->print_cr("  new humongous chunk word size " PTR_FORMAT,
+                           next->word_size());
+  }
+
+  return next;
+}
+
 MetaWord* SpaceManager::allocate(size_t word_size) {
   MutexLockerEx cl(lock(), Mutex::_no_safepoint_check_flag);
 
@@ -2090,12 +2232,7 @@
   // verfication of chunks does not work since
   // being in the dictionary alters a chunk.
   if (block_freelists()->total_size() == 0) {
-    // Skip the small chunks because their next link points to
-    // medium chunks.  This is because the small chunk is the
-    // current chunk (for allocations) until it is full and the
-    // the addition of the next chunk does not NULL the next
-    // like of the small chunk.
-    for (ChunkIndex i = MediumIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
+    for (ChunkIndex i = ZeroIndex; i < NumberOfInUseLists; i = next_chunk_index(i)) {
       Metachunk* curr = chunks_in_use(i);
       while (curr != NULL) {
         curr->verify();
@@ -2108,15 +2245,15 @@
 
 void SpaceManager::verify_chunk_size(Metachunk* chunk) {
   assert(is_humongous(chunk->word_size()) ||
-         chunk->word_size() == MediumChunk ||
-         chunk->word_size() == SmallChunk,
+         chunk->word_size() == medium_chunk_size() ||
+         chunk->word_size() == small_chunk_size() ||
+         chunk->word_size() == specialized_chunk_size(),
          "Chunk size is wrong");
   return;
 }
 
 #ifdef ASSERT
 void SpaceManager::verify_allocation_total() {
-#if 0
   // Verification is only guaranteed at a safepoint.
   if (SafepointSynchronize::is_at_safepoint()) {
     gclog_or_tty->print_cr("Chunk " PTR_FORMAT " allocation_total " SIZE_FORMAT
@@ -2129,7 +2266,6 @@
   assert(allocation_total() == sum_used_in_chunks_in_use(),
     err_msg("allocation total is not consistent %d vs %d",
             allocation_total(), sum_used_in_chunks_in_use()));
-#endif
 }
 
 #endif
@@ -2142,7 +2278,7 @@
   size_t capacity = 0;
 
   // Add up statistics for all chunks in this SpaceManager.
-  for (ChunkIndex index = SmallIndex;
+  for (ChunkIndex index = ZeroIndex;
        index < NumberOfInUseLists;
        index = next_chunk_index(index)) {
     for (Metachunk* curr = chunks_in_use(index);
@@ -2160,7 +2296,7 @@
     }
   }
 
-  size_t free = current_chunk()->free_word_size();
+  size_t free = current_chunk() == NULL ? 0 : current_chunk()->free_word_size();
   // Free space isn't wasted.
   waste -= free;
 
@@ -2171,25 +2307,18 @@
 
 #ifndef PRODUCT
 void SpaceManager::mangle_freed_chunks() {
-  for (ChunkIndex index = SmallIndex;
+  for (ChunkIndex index = ZeroIndex;
        index < NumberOfInUseLists;
        index = next_chunk_index(index)) {
     for (Metachunk* curr = chunks_in_use(index);
          curr != NULL;
          curr = curr->next()) {
-      // Try to detect incorrectly terminated small chunk
-      // list.
-      assert(index == MediumIndex || curr != chunks_in_use(MediumIndex),
-             err_msg("Mangling medium chunks in small chunks? "
-                     "curr " PTR_FORMAT " medium list " PTR_FORMAT,
-                     curr, chunks_in_use(MediumIndex)));
       curr->mangle();
     }
   }
 }
 #endif // PRODUCT
 
-
 // MetaspaceAux
 
 size_t MetaspaceAux::used_in_bytes(Metaspace::MetadataType mdtype) {
@@ -2236,7 +2365,7 @@
   return reserved * BytesPerWord;
 }
 
-size_t MetaspaceAux::min_chunk_size() { return SpaceManager::MediumChunk; }
+size_t MetaspaceAux::min_chunk_size() { return Metaspace::first_chunk_word_size(); }
 
 size_t MetaspaceAux::free_chunks_total(Metaspace::MetadataType mdtype) {
   ChunkManager* chunk = (mdtype == Metaspace::ClassType) ?
@@ -2316,26 +2445,44 @@
 // Print total fragmentation for class and data metaspaces separately
 void MetaspaceAux::print_waste(outputStream* out) {
 
-  size_t small_waste = 0, medium_waste = 0, large_waste = 0;
-  size_t cls_small_waste = 0, cls_medium_waste = 0, cls_large_waste = 0;
+  size_t specialized_waste = 0, small_waste = 0, medium_waste = 0, large_waste = 0;
+  size_t specialized_count = 0, small_count = 0, medium_count = 0, large_count = 0;
+  size_t cls_specialized_waste = 0, cls_small_waste = 0, cls_medium_waste = 0, cls_large_waste = 0;
+  size_t cls_specialized_count = 0, cls_small_count = 0, cls_medium_count = 0, cls_large_count = 0;
 
   ClassLoaderDataGraphMetaspaceIterator iter;
   while (iter.repeat()) {
     Metaspace* msp = iter.get_next();
     if (msp != NULL) {
+      specialized_waste += msp->vsm()->sum_waste_in_chunks_in_use(SpecializedIndex);
+      specialized_count += msp->vsm()->sum_count_in_chunks_in_use(SpecializedIndex);
       small_waste += msp->vsm()->sum_waste_in_chunks_in_use(SmallIndex);
+      small_count += msp->vsm()->sum_count_in_chunks_in_use(SmallIndex);
       medium_waste += msp->vsm()->sum_waste_in_chunks_in_use(MediumIndex);
+      medium_count += msp->vsm()->sum_count_in_chunks_in_use(MediumIndex);
       large_waste += msp->vsm()->sum_waste_in_chunks_in_use(HumongousIndex);
-
+      large_count += msp->vsm()->sum_count_in_chunks_in_use(HumongousIndex);
+
+      cls_specialized_waste += msp->class_vsm()->sum_waste_in_chunks_in_use(SpecializedIndex);
+      cls_specialized_count += msp->class_vsm()->sum_count_in_chunks_in_use(SpecializedIndex);
       cls_small_waste += msp->class_vsm()->sum_waste_in_chunks_in_use(SmallIndex);
+      cls_small_count += msp->class_vsm()->sum_count_in_chunks_in_use(SmallIndex);
       cls_medium_waste += msp->class_vsm()->sum_waste_in_chunks_in_use(MediumIndex);
+      cls_medium_count += msp->class_vsm()->sum_count_in_chunks_in_use(MediumIndex);
       cls_large_waste += msp->class_vsm()->sum_waste_in_chunks_in_use(HumongousIndex);
+      cls_large_count += msp->class_vsm()->sum_count_in_chunks_in_use(HumongousIndex);
     }
   }
   out->print_cr("Total fragmentation waste (words) doesn't count free space");
-  out->print("  data: small " SIZE_FORMAT " medium " SIZE_FORMAT,
-             small_waste, medium_waste);
-  out->print_cr(" class: small " SIZE_FORMAT, cls_small_waste);
+  out->print_cr("  data: " SIZE_FORMAT " specialized(s) " SIZE_FORMAT ", "
+                        SIZE_FORMAT " small(s) " SIZE_FORMAT ", "
+                        SIZE_FORMAT " medium(s) " SIZE_FORMAT,
+             specialized_count, specialized_waste, small_count,
+             small_waste, medium_count, medium_waste);
+  out->print_cr(" class: " SIZE_FORMAT " specialized(s) " SIZE_FORMAT ", "
+                           SIZE_FORMAT " small(s) " SIZE_FORMAT,
+             cls_specialized_count, cls_specialized_waste,
+             cls_small_count, cls_small_waste);
 }
 
 // Dump global metaspace things from the end of ClassLoaderDataGraph
@@ -2354,13 +2501,10 @@
 // Metaspace methods
 
 size_t Metaspace::_first_chunk_word_size = 0;
-
-Metaspace::Metaspace(Mutex* lock, size_t word_size) {
-  initialize(lock, word_size);
-}
-
-Metaspace::Metaspace(Mutex* lock) {
-  initialize(lock);
+size_t Metaspace::_first_class_chunk_word_size = 0;
+
+Metaspace::Metaspace(Mutex* lock, MetaspaceType type) {
+  initialize(lock, type);
 }
 
 Metaspace::~Metaspace() {
@@ -2412,11 +2556,18 @@
       }
     }
 
-    // Initialize this before initializing the VirtualSpaceList
+    // Initialize these before initializing the VirtualSpaceList
     _first_chunk_word_size = InitialBootClassLoaderMetaspaceSize / BytesPerWord;
+    _first_chunk_word_size = align_word_size_up(_first_chunk_word_size);
+    // Make the first class chunk bigger than a medium chunk so it's not put
+    // on the medium chunk list.   The next chunk will be small and progress
+    // from there.  This size calculated by -version.
+    _first_class_chunk_word_size = MIN2((size_t)MediumChunk*6,
+                                       (ClassMetaspaceSize/BytesPerWord)*2);
+    _first_class_chunk_word_size = align_word_size_up(_first_class_chunk_word_size);
     // Arbitrarily set the initial virtual space to a multiple
     // of the boot class loader size.
-    size_t word_size = VIRTUALSPACEMULTIPLIER * Metaspace::first_chunk_word_size();
+    size_t word_size = VIRTUALSPACEMULTIPLIER * first_chunk_word_size();
     // Initialize the list of virtual spaces.
     _space_list = new VirtualSpaceList(word_size);
   }
@@ -2431,23 +2582,8 @@
   _class_space_list = new VirtualSpaceList(rs);
 }
 
-
-void Metaspace::initialize(Mutex* lock, size_t initial_size) {
-  // Use SmallChunk size if not specified.   If specified, use this size for
-  // the data metaspace.
-  size_t word_size;
-  size_t class_word_size;
-  if (initial_size == 0) {
-    word_size = (size_t) SpaceManager::SmallChunk;
-    class_word_size = (size_t) SpaceManager::SmallChunk;
-  } else {
-    word_size = initial_size;
-    // Make the first class chunk bigger than a medium chunk so it's not put
-    // on the medium chunk list.   The next chunk will be small and progress
-    // from there.  This size calculated by -version.
-    class_word_size = MIN2((size_t)SpaceManager::MediumChunk*5,
-                           (ClassMetaspaceSize/BytesPerWord)*2);
-  }
+void Metaspace::initialize(Mutex* lock,
+                           MetaspaceType type) {
 
   assert(space_list() != NULL,
     "Metadata VirtualSpaceList has not been initialized");
@@ -2456,6 +2592,11 @@
   if (_vsm == NULL) {
     return;
   }
+  size_t word_size;
+  size_t class_word_size;
+  vsm()->get_initial_chunk_sizes(type,
+                                 &word_size,
+                                 &class_word_size);
 
   assert(class_space_list() != NULL,
     "Class VirtualSpaceList has not been initialized");
@@ -2470,7 +2611,8 @@
 
   // Allocate chunk for metadata objects
   Metachunk* new_chunk =
-     space_list()->current_virtual_space()->get_chunk_vs_with_expand(word_size);
+     space_list()->get_initialization_chunk(word_size,
+                                            vsm()->medium_chunk_bunch());
   assert(!DumpSharedSpaces || new_chunk != NULL, "should have enough space for both chunks");
   if (new_chunk != NULL) {
     // Add to this manager's list of chunks in use and current_chunk().
@@ -2479,12 +2621,18 @@
 
   // Allocate chunk for class metadata objects
   Metachunk* class_chunk =
-     class_space_list()->current_virtual_space()->get_chunk_vs_with_expand(class_word_size);
+     class_space_list()->get_initialization_chunk(class_word_size,
+                                                  class_vsm()->medium_chunk_bunch());
   if (class_chunk != NULL) {
     class_vsm()->add_chunk(class_chunk, true);
   }
 }
 
+size_t Metaspace::align_word_size_up(size_t word_size) {
+  size_t byte_size = word_size * wordSize;
+  return ReservedSpace::allocation_align_size_up(byte_size) / wordSize;
+}
+
 MetaWord* Metaspace::allocate(size_t word_size, MetadataType mdtype) {
   // DumpSharedSpaces doesn't use class metadata area (yet)
   if (mdtype == ClassType && !DumpSharedSpaces) {
@@ -2610,6 +2758,12 @@
 
     // If result is still null, we are out of memory.
     if (result == NULL) {
+      if (Verbose && TraceMetadataChunkAllocation) {
+        gclog_or_tty->print_cr("Metaspace allocation failed for size "
+          SIZE_FORMAT, word_size);
+        if (loader_data->metaspace_or_null() != NULL) loader_data->metaspace_or_null()->dump(gclog_or_tty);
+        MetaspaceAux::dump(gclog_or_tty);
+      }
       // -XX:+HeapDumpOnOutOfMemoryError and -XX:OnOutOfMemoryError support
       report_java_out_of_memory("Metadata space");
 
--- a/src/share/vm/memory/metaspace.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/metaspace.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -87,11 +87,23 @@
 
  public:
   enum MetadataType {ClassType, NonClassType};
+  enum MetaspaceType {
+    StandardMetaspaceType,
+    BootMetaspaceType,
+    ROMetaspaceType,
+    ReadWriteMetaspaceType,
+    AnonymousMetaspaceType,
+    ReflectionMetaspaceType
+  };
 
  private:
-  void initialize(Mutex* lock, size_t initial_size = 0);
+  void initialize(Mutex* lock, MetaspaceType type);
+
+  // Align up the word size to the allocation word size
+  static size_t align_word_size_up(size_t);
 
   static size_t _first_chunk_word_size;
+  static size_t _first_class_chunk_word_size;
 
   SpaceManager* _vsm;
   SpaceManager* vsm() const { return _vsm; }
@@ -110,8 +122,7 @@
 
  public:
 
-  Metaspace(Mutex* lock, size_t initial_size);
-  Metaspace(Mutex* lock);
+  Metaspace(Mutex* lock, MetaspaceType type);
   ~Metaspace();
 
   // Initialize globals for Metaspace
@@ -119,6 +130,7 @@
   static void initialize_class_space(ReservedSpace rs);
 
   static size_t first_chunk_word_size() { return _first_chunk_word_size; }
+  static size_t first_class_chunk_word_size() { return _first_class_chunk_word_size; }
 
   char*  bottom() const;
   size_t used_words(MetadataType mdtype) const;
--- a/src/share/vm/memory/metaspaceShared.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/metaspaceShared.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -689,9 +689,15 @@
 bool MetaspaceShared::map_shared_spaces(FileMapInfo* mapinfo) {
   size_t image_alignment = mapinfo->alignment();
 
-  // Map in the shared memory and then map the regions on top of it
+#ifndef _WINDOWS
+  // Map in the shared memory and then map the regions on top of it.
+  // On Windows, don't map the memory here because it will cause the
+  // mappings of the regions to fail.
   ReservedSpace shared_rs = mapinfo->reserve_shared_memory();
   if (!shared_rs.is_reserved()) return false;
+#endif
+
+  assert(!DumpSharedSpaces, "Should not be called with DumpSharedSpaces");
 
   // Map each shared region
   if ((_ro_base = mapinfo->map_region(ro)) != NULL &&
@@ -708,8 +714,10 @@
     if (_rw_base != NULL) mapinfo->unmap_region(rw);
     if (_md_base != NULL) mapinfo->unmap_region(md);
     if (_mc_base != NULL) mapinfo->unmap_region(mc);
+#ifndef _WINDOWS
     // Release the entire mapped region
     shared_rs.release();
+#endif
     // If -Xshare:on is specified, print out the error message and exit VM,
     // otherwise, set UseSharedSpaces to false and continue.
     if (RequireSharedSpaces) {
--- a/src/share/vm/memory/space.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/space.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -411,7 +411,6 @@
     assert(q->forwardee() == NULL, "should be forwarded to NULL");
   }
 
-  VALIDATE_MARK_SWEEP_ONLY(MarkSweep::register_live_oop(q, size));
   compact_top += size;
 
   // we need to update the offset table so that the beginnings of objects can be
@@ -470,13 +469,10 @@
     if (oop(q)->is_gc_marked()) {
       // q is alive
 
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::track_interior_pointers(oop(q)));
       // point all the oops to the new location
       size_t size = oop(q)->adjust_pointers();
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::check_interior_pointers());
 
       debug_only(prev_q = q);
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::validate_live_oop(oop(q), size));
 
       q += size;
     } else {
--- a/src/share/vm/memory/space.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/space.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -655,16 +655,10 @@
       assert(block_is_obj(q),                                                   \
              "should be at block boundaries, and should be looking at objs");   \
                                                                                 \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::track_interior_pointers(oop(q)));     \
-                                                                                \
       /* point all the oops to the new location */                              \
       size_t size = oop(q)->adjust_pointers();                                  \
       size = adjust_obj_size(size);                                             \
                                                                                 \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::check_interior_pointers());           \
-                                                                                \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::validate_live_oop(oop(q), size));     \
-                                                                                \
       q += size;                                                                \
     }                                                                           \
                                                                                 \
@@ -685,12 +679,9 @@
     Prefetch::write(q, interval);                                               \
     if (oop(q)->is_gc_marked()) {                                               \
       /* q is alive */                                                          \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::track_interior_pointers(oop(q)));     \
       /* point all the oops to the new location */                              \
       size_t size = oop(q)->adjust_pointers();                                  \
       size = adjust_obj_size(size);                                             \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::check_interior_pointers());           \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::validate_live_oop(oop(q), size));     \
       debug_only(prev_q = q);                                                   \
       q += size;                                                                \
     } else {                                                                    \
@@ -725,7 +716,6 @@
       size_t size = obj_size(q);                                                \
       assert(!oop(q)->is_gc_marked(),                                           \
              "should be unmarked (special dense prefix handling)");             \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::live_oop_moved_to(q, size, q));       \
       debug_only(prev_q = q);                                                   \
       q += size;                                                                \
     }                                                                           \
@@ -759,8 +749,6 @@
       Prefetch::write(compaction_top, copy_interval);                           \
                                                                                 \
       /* copy object and reinit its mark */                                     \
-      VALIDATE_MARK_SWEEP_ONLY(MarkSweep::live_oop_moved_to(q, size,            \
-                                                            compaction_top));   \
       assert(q != compaction_top, "everything in this pass should be moving");  \
       Copy::aligned_conjoint_words(q, compaction_top, size);                    \
       oop(compaction_top)->init_mark();                                         \
--- a/src/share/vm/memory/tenuredGeneration.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/memory/tenuredGeneration.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -62,7 +62,7 @@
                                        _virtual_space.reserved_size(),
                                        _the_space, _gen_counters);
 #ifndef SERIALGC
-  if (UseParNewGC && ParallelGCThreads > 0) {
+  if (UseParNewGC) {
     typedef ParGCAllocBufferWithBOT* ParGCAllocBufferWithBOTPtr;
     _alloc_buffers = NEW_C_HEAP_ARRAY(ParGCAllocBufferWithBOTPtr,
                                       ParallelGCThreads, mtGC);
--- a/src/share/vm/oops/annotations.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/annotations.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -61,6 +61,9 @@
   free_contents(loader_data, methods_annotations());
   free_contents(loader_data, methods_parameter_annotations());
   free_contents(loader_data, methods_default_annotations());
+
+  // Recursively deallocate optional Annotations linked through this one
+  MetadataFactory::free_metadata(loader_data, type_annotations());
 }
 
 // Set the annotation at 'idnum' to 'anno'.
--- a/src/share/vm/oops/annotations.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/annotations.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -38,7 +38,8 @@
 typedef Array<u1> AnnotationArray;
 
 // Class to hold the various types of annotations. The only metadata that points
-// to this is InstanceKlass.
+// to this is InstanceKlass, or another Annotations instance if this is a
+// a type_annotation instance.
 
 class Annotations: public MetaspaceObj {
 
@@ -58,6 +59,8 @@
   // such annotations.
   // Index is the idnum, which is initially the same as the methods array index.
   Array<AnnotationArray*>*     _methods_default_annotations;
+  // Type annotations for this class, or null if none.
+  Annotations*                 _type_annotations;
 
   // Constructor where some some values are known to not be null
   Annotations(Array<AnnotationArray*>* fa, Array<AnnotationArray*>* ma,
@@ -66,7 +69,8 @@
                  _fields_annotations(fa),
                  _methods_annotations(ma),
                  _methods_parameter_annotations(mpa),
-                 _methods_default_annotations(mda) {}
+                 _methods_default_annotations(mda),
+                 _type_annotations(NULL) {}
 
  public:
   // Allocate instance of this class
@@ -81,22 +85,26 @@
   static int size()    { return sizeof(Annotations) / wordSize; }
 
   // Constructor to initialize to null
-  Annotations() : _class_annotations(NULL), _fields_annotations(NULL),
+  Annotations() : _class_annotations(NULL),
+                  _fields_annotations(NULL),
                   _methods_annotations(NULL),
                   _methods_parameter_annotations(NULL),
-                  _methods_default_annotations(NULL) {}
+                  _methods_default_annotations(NULL),
+                  _type_annotations(NULL) {}
 
   AnnotationArray* class_annotations() const                       { return _class_annotations; }
   Array<AnnotationArray*>* fields_annotations() const              { return _fields_annotations; }
   Array<AnnotationArray*>* methods_annotations() const             { return _methods_annotations; }
   Array<AnnotationArray*>* methods_parameter_annotations() const   { return _methods_parameter_annotations; }
   Array<AnnotationArray*>* methods_default_annotations() const     { return _methods_default_annotations; }
+  Annotations* type_annotations() const                            { return _type_annotations; }
 
   void set_class_annotations(AnnotationArray* md)                     { _class_annotations = md; }
   void set_fields_annotations(Array<AnnotationArray*>* md)            { _fields_annotations = md; }
   void set_methods_annotations(Array<AnnotationArray*>* md)           { _methods_annotations = md; }
   void set_methods_parameter_annotations(Array<AnnotationArray*>* md) { _methods_parameter_annotations = md; }
   void set_methods_default_annotations(Array<AnnotationArray*>* md)   { _methods_default_annotations = md; }
+  void set_type_annotations(Annotations* annos)                       { _type_annotations = annos; }
 
   // Redefine classes support
   AnnotationArray* get_method_annotations_of(int idnum)
@@ -129,6 +137,7 @@
   inline AnnotationArray* get_method_annotations_from(int idnum, Array<AnnotationArray*>* annos);
   void set_annotations(Array<AnnotationArray*>* md, Array<AnnotationArray*>** md_p)  { *md_p = md; }
 
+  bool is_klass() const { return false; }
  private:
   void set_methods_annotations_of(instanceKlassHandle ik,
                                   int idnum, AnnotationArray* anno,
--- a/src/share/vm/oops/constMethod.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/constMethod.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -39,18 +39,21 @@
                                    int localvariable_table_length,
                                    int exception_table_length,
                                    int checked_exceptions_length,
+                                   int method_parameters_length,
                                    u2  generic_signature_index,
                                    MethodType method_type,
                                    TRAPS) {
   int size = ConstMethod::size(byte_code_size,
-                                      compressed_line_number_size,
-                                      localvariable_table_length,
-                                      exception_table_length,
-                                      checked_exceptions_length,
-                                      generic_signature_index);
+                               compressed_line_number_size,
+                               localvariable_table_length,
+                               exception_table_length,
+                               checked_exceptions_length,
+                               method_parameters_length,
+                               generic_signature_index);
   return new (loader_data, size, true, THREAD) ConstMethod(
       byte_code_size, compressed_line_number_size, localvariable_table_length,
-      exception_table_length, checked_exceptions_length, generic_signature_index,
+      exception_table_length, checked_exceptions_length,
+      method_parameters_length, generic_signature_index,
       method_type, size);
 }
 
@@ -59,6 +62,7 @@
                          int localvariable_table_length,
                          int exception_table_length,
                          int checked_exceptions_length,
+                         int method_parameters_length,
                          u2  generic_signature_index,
                          MethodType method_type,
                          int size) {
@@ -74,7 +78,8 @@
                             checked_exceptions_length,
                             compressed_line_number_size,
                             localvariable_table_length,
-                            exception_table_length);
+                            exception_table_length,
+                            method_parameters_length);
   set_method_type(method_type);
   assert(this->size() == size, "wrong size for object");
 }
@@ -92,11 +97,12 @@
 // How big must this constMethodObject be?
 
 int ConstMethod::size(int code_size,
-                                    int compressed_line_number_size,
-                                    int local_variable_table_length,
-                                    int exception_table_length,
-                                    int checked_exceptions_length,
-                                    u2  generic_signature_index) {
+                      int compressed_line_number_size,
+                      int local_variable_table_length,
+                      int exception_table_length,
+                      int checked_exceptions_length,
+                      int method_parameters_length,
+                      u2  generic_signature_index) {
   int extra_bytes = code_size;
   if (compressed_line_number_size > 0) {
     extra_bytes += compressed_line_number_size;
@@ -117,6 +123,10 @@
   if (generic_signature_index != 0) {
     extra_bytes += sizeof(u2);
   }
+  if (method_parameters_length > 0) {
+    extra_bytes += sizeof(u2);
+    extra_bytes += method_parameters_length * sizeof(MethodParametersElement);
+  }
   int extra_words = align_size_up(extra_bytes, BytesPerWord) / BytesPerWord;
   return align_object_size(header_size() + extra_words);
 }
@@ -143,6 +153,18 @@
 u2* ConstMethod::checked_exceptions_length_addr() const {
   // Located immediately before the generic signature index.
   assert(has_checked_exceptions(), "called only if table is present");
+  if(has_method_parameters()) {
+    // If method parameters present, locate immediately before them.
+    return (u2*)method_parameters_start() - 1;
+  } else {
+    // Else, the exception table is at the end of the constMethod.
+    return has_generic_signature() ? (last_u2_element() - 1) :
+                                     last_u2_element();
+  }
+}
+
+u2* ConstMethod::method_parameters_length_addr() const {
+  assert(has_method_parameters(), "called only if table is present");
   return has_generic_signature() ? (last_u2_element() - 1) :
                                     last_u2_element();
 }
@@ -153,11 +175,15 @@
     // If checked_exception present, locate immediately before them.
     return (u2*) checked_exceptions_start() - 1;
   } else {
-    // Else, the exception table is at the end of the constMethod or
-    // immediately before the generic signature index.
+    if(has_method_parameters()) {
+      // If method parameters present, locate immediately before them.
+      return (u2*)method_parameters_start() - 1;
+    } else {
+      // Else, the exception table is at the end of the constMethod.
     return has_generic_signature() ? (last_u2_element() - 1) :
                                       last_u2_element();
   }
+  }
 }
 
 u2* ConstMethod::localvariable_table_length_addr() const {
@@ -170,12 +196,16 @@
       // If checked_exception present, locate immediately before them.
       return (u2*) checked_exceptions_start() - 1;
     } else {
-      // Else, the linenumber table is at the end of the constMethod or
-      // immediately before the generic signature index.
+      if(has_method_parameters()) {
+        // If method parameters present, locate immediately before them.
+        return (u2*)method_parameters_start() - 1;
+      } else {
+        // Else, the exception table is at the end of the constMethod.
       return has_generic_signature() ? (last_u2_element() - 1) :
                                         last_u2_element();
     }
   }
+  }
 }
 
 // Update the flags to indicate the presence of these optional fields.
@@ -183,29 +213,57 @@
                                             int checked_exceptions_len,
                                             int compressed_line_number_size,
                                             int localvariable_table_len,
-                                            int exception_table_len) {
-  // Must be done in the order below, otherwise length_addr accessors
-  // will not work. Only set bit in header if length is positive.
+                                            int exception_table_len,
+                                            int method_parameters_len) {
   assert(_flags == 0, "Error");
-  if (compressed_line_number_size > 0) {
+  if (compressed_line_number_size > 0)
     _flags |= _has_linenumber_table;
-  }
-  if (generic_signature_index != 0) {
+  if (generic_signature_index != 0)
     _flags |= _has_generic_signature;
+  if (method_parameters_len > 0)
+    _flags |= _has_method_parameters;
+  if (checked_exceptions_len > 0)
+    _flags |= _has_checked_exceptions;
+  if (exception_table_len > 0)
+    _flags |= _has_exception_table;
+  if (localvariable_table_len > 0)
+    _flags |= _has_localvariable_table;
+
+  // This code is extremely brittle and should possibly be revised.
+  // The *_length_addr functions walk backwards through the
+  // constMethod data, using each of the length indexes ahead of them,
+  // as well as the flags variable.  Therefore, the indexes must be
+  // initialized in reverse order, or else they will compute the wrong
+  // offsets.  Moving the initialization of _flags into a separate
+  // block solves *half* of the problem, but the following part will
+  // still break if the order is not exactly right.
+  //
+  // Also, the servicability agent needs to be informed anytime
+  // anything is added here.  It might be advisable to have some sort
+  // of indication of this inline.
+  if (generic_signature_index != 0)
     *(generic_signature_index_addr()) = generic_signature_index;
-  }
-  if (checked_exceptions_len > 0) {
-    _flags |= _has_checked_exceptions;
+  // New data should probably go here.
+  if (method_parameters_len > 0)
+    *(method_parameters_length_addr()) = method_parameters_len;
+  if (checked_exceptions_len > 0)
     *(checked_exceptions_length_addr()) = checked_exceptions_len;
-  }
-  if (exception_table_len > 0) {
-    _flags |= _has_exception_table;
+  if (exception_table_len > 0)
     *(exception_table_length_addr()) = exception_table_len;
-  }
-  if (localvariable_table_len > 0) {
-    _flags |= _has_localvariable_table;
+  if (localvariable_table_len > 0)
     *(localvariable_table_length_addr()) = localvariable_table_len;
-  }
+}
+
+int ConstMethod::method_parameters_length() const {
+  return has_method_parameters() ? *(method_parameters_length_addr()) : 0;
+}
+
+MethodParametersElement* ConstMethod::method_parameters_start() const {
+  u2* addr = method_parameters_length_addr();
+  u2 length = *addr;
+  assert(length > 0, "should only be called if table is present");
+  addr -= length * sizeof(MethodParametersElement) / sizeof(u2);
+  return (MethodParametersElement*) addr;
 }
 
 
@@ -298,6 +356,10 @@
   }
   guarantee(compressed_table_end <= m_end, "invalid method layout");
   // Verify checked exceptions, exception table and local variable tables
+  if (has_method_parameters()) {
+    u2* addr = method_parameters_length_addr();
+    guarantee(*addr > 0 && (address) addr >= compressed_table_end && (address) addr < m_end, "invalid method layout");
+  }
   if (has_checked_exceptions()) {
     u2* addr = checked_exceptions_length_addr();
     guarantee(*addr > 0 && (address) addr >= compressed_table_end && (address) addr < m_end, "invalid method layout");
@@ -318,6 +380,8 @@
     uncompressed_table_start = (u2*) exception_table_start();
   } else if (has_checked_exceptions()) {
       uncompressed_table_start = (u2*) checked_exceptions_start();
+  } else if (has_method_parameters()) {
+      uncompressed_table_start = (u2*) method_parameters_start();
   } else {
       uncompressed_table_start = (u2*) m_end;
   }
--- a/src/share/vm/oops/constMethod.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/constMethod.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -77,9 +77,18 @@
 // |  (access flags bit tells whether table is present)   |
 // |  (indexed from end of ConstMethod*)                  |
 // |------------------------------------------------------|
+// | method parameters elements + length (length last)    |
+// |  (length is u2, elements are u2, u4 structures)      |
+// |  (see class MethodParametersElement)                 |
+// |  (access flags bit tells whether table is present)   |
+// |  (indexed from end of ConstMethod*)                  |
+// |------------------------------------------------------|
 // | generic signature index (u2)                         |
 // |  (indexed from start of constMethodOop)              |
 // |------------------------------------------------------|
+//
+// IMPORTANT: If anything gets added here, there need to be changes to
+// ensure that ServicabilityAgent doesn't get broken as a result!
 
 
 // Utitily class decribing elements in checked exceptions table inlined in Method*.
@@ -109,6 +118,13 @@
   u2 catch_type_index;
 };
 
+// Utility class describing elements in method parameters
+class MethodParametersElement VALUE_OBJ_CLASS_SPEC {
+ public:
+  u2 name_cp_index;
+  u4 flags;
+};
+
 
 class ConstMethod : public MetaspaceObj {
   friend class VMStructs;
@@ -123,7 +139,8 @@
     _has_localvariable_table = 4,
     _has_exception_table = 8,
     _has_generic_signature = 16,
-    _is_overpass = 32
+    _has_method_parameters = 32,
+    _is_overpass = 64
   };
 
   // Bit vector of signature
@@ -160,6 +177,7 @@
               int localvariable_table_length,
               int exception_table_length,
               int checked_exceptions_length,
+              int method_parameters_length,
               u2  generic_signature_index,
               MethodType is_overpass,
               int size);
@@ -171,6 +189,7 @@
                                int localvariable_table_length,
                                int exception_table_length,
                                int checked_exceptions_length,
+                               int method_parameters_length,
                                u2  generic_signature_index,
                                MethodType mt,
                                TRAPS);
@@ -182,7 +201,8 @@
                                  int checked_exceptions_len,
                                  int compressed_line_number_size,
                                  int localvariable_table_len,
-                                 int exception_table_len);
+                                 int exception_table_len,
+                                 int method_parameters_length);
 
   bool has_generic_signature() const
     { return (_flags & _has_generic_signature) != 0; }
@@ -199,6 +219,9 @@
   bool has_exception_handler() const
     { return (_flags & _has_exception_table) != 0; }
 
+  bool has_method_parameters() const
+    { return (_flags & _has_method_parameters) != 0; }
+
   MethodType method_type() const {
     return ((_flags & _is_overpass) == 0) ? NORMAL : OVERPASS;
   }
@@ -284,10 +307,11 @@
 
   // Size needed
   static int size(int code_size, int compressed_line_number_size,
-                         int local_variable_table_length,
-                         int exception_table_length,
-                         int checked_exceptions_length,
-                         u2  generic_signature_index);
+                  int local_variable_table_length,
+                  int exception_table_length,
+                  int checked_exceptions_length,
+                  int method_parameters_length,
+                  u2  generic_signature_index);
 
   int size() const                    { return _constMethod_size;}
   void set_constMethod_size(int size)     { _constMethod_size = size; }
@@ -308,6 +332,7 @@
   u2* checked_exceptions_length_addr() const;
   u2* localvariable_table_length_addr() const;
   u2* exception_table_length_addr() const;
+  u2* method_parameters_length_addr() const;
 
   // checked exceptions
   int checked_exceptions_length() const;
@@ -321,6 +346,10 @@
   int exception_table_length() const;
   ExceptionTableElement* exception_table_start() const;
 
+  // method parameters table
+  int method_parameters_length() const;
+  MethodParametersElement* method_parameters_start() const;
+
   // byte codes
   void    set_code(address code) {
     if (code_size() > 0) {
--- a/src/share/vm/oops/instanceKlass.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/instanceKlass.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -47,6 +47,7 @@
 #include "oops/symbol.hpp"
 #include "prims/jvmtiExport.hpp"
 #include "prims/jvmtiRedefineClassesTrace.hpp"
+#include "prims/methodComparator.hpp"
 #include "runtime/fieldDescriptor.hpp"
 #include "runtime/handles.inline.hpp"
 #include "runtime/javaCalls.hpp"
@@ -160,6 +161,8 @@
 
 #endif //  ndef DTRACE_ENABLED
 
+volatile int InstanceKlass::_total_instanceKlass_count = 0;
+
 Klass* InstanceKlass::allocate_instance_klass(ClassLoaderData* loader_data,
                                                 int vtable_len,
                                                 int itable_len,
@@ -203,6 +206,7 @@
         access_flags, !host_klass.is_null());
   }
 
+  Atomic::inc(&_total_instanceKlass_count);
   return ik;
 }
 
@@ -361,6 +365,9 @@
   set_protection_domain(NULL);
   set_signers(NULL);
   set_init_lock(NULL);
+
+  // We should deallocate the Annotations instance
+  MetadataFactory::free_metadata(loader_data, annotations());
   set_annotations(NULL);
 }
 
@@ -599,7 +606,7 @@
       }
 
       // relocate jsrs and link methods after they are all rewritten
-      this_oop->relocate_and_link_methods(CHECK_false);
+      this_oop->link_methods(CHECK_false);
 
       // Initialize the vtable and interface table after
       // methods have been rewritten since rewrite may
@@ -647,10 +654,31 @@
 // Now relocate and link method entry points after class is rewritten.
 // This is outside is_rewritten flag. In case of an exception, it can be
 // executed more than once.
-void InstanceKlass::relocate_and_link_methods(TRAPS) {
-  assert(is_loaded(), "must be loaded");
-  instanceKlassHandle this_oop(THREAD, this);
-  Rewriter::relocate_and_link(this_oop, CHECK);
+void InstanceKlass::link_methods(TRAPS) {
+  int len = methods()->length();
+  for (int i = len-1; i >= 0; i--) {
+    methodHandle m(THREAD, methods()->at(i));
+
+    // Set up method entry points for compiler and interpreter    .
+    m->link_method(m, CHECK);
+
+    // This is for JVMTI and unrelated to relocator but the last thing we do
+#ifdef ASSERT
+    if (StressMethodComparator) {
+      ResourceMark rm(THREAD);
+      static int nmc = 0;
+      for (int j = i; j >= 0 && j >= i-4; j--) {
+        if ((++nmc % 1000) == 0)  tty->print_cr("Have run MethodComparator %d times...", nmc);
+        bool z = MethodComparator::methods_EMCP(m(),
+                   methods()->at(j));
+        if (j == i && !z) {
+          tty->print("MethodComparator FAIL: "); m->print(); m->print_codes();
+          assert(z, "method must compare equal to itself");
+        }
+      }
+    }
+#endif //ASSERT
+  }
 }
 
 
@@ -2306,6 +2334,9 @@
   if (_array_name != NULL)  _array_name->decrement_refcount();
   if (_source_file_name != NULL) _source_file_name->decrement_refcount();
   if (_source_debug_extension != NULL) FREE_C_HEAP_ARRAY(char, _source_debug_extension, mtClass);
+
+  assert(_total_instanceKlass_count >= 1, "Sanity check");
+  Atomic::dec(&_total_instanceKlass_count);
 }
 
 void InstanceKlass::set_source_file_name(Symbol* n) {
--- a/src/share/vm/oops/instanceKlass.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/instanceKlass.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -31,6 +31,7 @@
 #include "oops/fieldInfo.hpp"
 #include "oops/instanceOop.hpp"
 #include "oops/klassVtable.hpp"
+#include "runtime/atomic.hpp"
 #include "runtime/handles.hpp"
 #include "runtime/os.hpp"
 #include "utilities/accessFlags.hpp"
@@ -170,6 +171,11 @@
     initialization_error                // error happened during initialization
   };
 
+  static int number_of_instance_classes() { return _total_instanceKlass_count; }
+
+ private:
+  static volatile int _total_instanceKlass_count;
+
  protected:
   // Protection domain.
   oop             _protection_domain;
@@ -454,7 +460,7 @@
   bool link_class_or_fail(TRAPS); // returns false on failure
   void unlink_class();
   void rewrite_class(TRAPS);
-  void relocate_and_link_methods(TRAPS);
+  void link_methods(TRAPS);
   Method* class_initializer();
 
   // set the class to initialized if no static initializer is present
@@ -657,6 +663,10 @@
     if (annotations() == NULL) return NULL;
     return annotations()->fields_annotations();
   }
+  Annotations* type_annotations() const {
+    if (annotations() == NULL) return NULL;
+    return annotations()->type_annotations();
+  }
 
   // allocation
   instanceOop allocate_instance(TRAPS);
--- a/src/share/vm/oops/method.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/method.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -64,6 +64,7 @@
                          int localvariable_table_length,
                          int exception_table_length,
                          int checked_exceptions_length,
+                         int method_parameters_length,
                          u2  generic_signature_index,
                          ConstMethod::MethodType method_type,
                          TRAPS) {
@@ -75,6 +76,7 @@
                                           localvariable_table_length,
                                           exception_table_length,
                                           checked_exceptions_length,
+                                          method_parameters_length,
                                           generic_signature_index,
                                           method_type,
                                           CHECK_NULL);
@@ -192,16 +194,16 @@
   return buf;
 }
 
-int  Method::fast_exception_handler_bci_for(KlassHandle ex_klass, int throw_bci, TRAPS) {
+int Method::fast_exception_handler_bci_for(methodHandle mh, KlassHandle ex_klass, int throw_bci, TRAPS) {
   // exception table holds quadruple entries of the form (beg_bci, end_bci, handler_bci, klass_index)
   // access exception table
-  ExceptionTable table(this);
+  ExceptionTable table(mh());
   int length = table.length();
   // iterate through all entries sequentially
-  constantPoolHandle pool(THREAD, constants());
+  constantPoolHandle pool(THREAD, mh->constants());
   for (int i = 0; i < length; i ++) {
     //reacquire the table in case a GC happened
-    ExceptionTable table(this);
+    ExceptionTable table(mh());
     int beg_bci = table.start_pc(i);
     int end_bci = table.end_pc(i);
     assert(beg_bci <= end_bci, "inconsistent exception table");
@@ -1035,8 +1037,10 @@
 
   methodHandle m;
   {
-    Method* m_oop = Method::allocate(loader_data, 0, accessFlags_from(flags_bits),
-             0, 0, 0, 0, 0, ConstMethod::NORMAL, CHECK_(empty));
+    Method* m_oop = Method::allocate(loader_data, 0,
+                                     accessFlags_from(flags_bits),
+                                     0, 0, 0, 0, 0, 0,
+                                     ConstMethod::NORMAL, CHECK_(empty));
     m = methodHandle(THREAD, m_oop);
   }
   m->set_constants(cp());
@@ -1088,6 +1092,7 @@
   int checked_exceptions_len = m->checked_exceptions_length();
   int localvariable_len = m->localvariable_table_length();
   int exception_table_len = m->exception_table_length();
+  int method_parameters_len = m->method_parameters_length();
 
   ClassLoaderData* loader_data = m->method_holder()->class_loader_data();
   Method* newm_oop = Method::allocate(loader_data,
@@ -1097,6 +1102,7 @@
                                       localvariable_len,
                                       exception_table_len,
                                       checked_exceptions_len,
+                                      method_parameters_len,
                                       generic_signature_index,
                                       m->method_type(),
                                       CHECK_(methodHandle()));
@@ -1331,13 +1337,15 @@
                                  Array<AnnotationArray*>* methods_annotations,
                                  Array<AnnotationArray*>* methods_parameter_annotations,
                                  Array<AnnotationArray*>* methods_default_annotations,
+                                 Array<AnnotationArray*>* methods_type_annotations,
                                  bool idempotent) {
   int length = methods->length();
   if (length > 1) {
     bool do_annotations = false;
     if (methods_annotations != NULL ||
         methods_parameter_annotations != NULL ||
-        methods_default_annotations != NULL) {
+        methods_default_annotations != NULL ||
+        methods_type_annotations != NULL) {
       do_annotations = true;
     }
     if (do_annotations) {
@@ -1356,6 +1364,7 @@
     assert(methods_annotations == NULL           || methods_annotations->length() == methods->length(), "");
     assert(methods_parameter_annotations == NULL || methods_parameter_annotations->length() == methods->length(), "");
     assert(methods_default_annotations == NULL   || methods_default_annotations->length() == methods->length(), "");
+    assert(methods_type_annotations == NULL   || methods_type_annotations->length() == methods->length(), "");
     if (do_annotations) {
       ResourceMark rm;
       // Allocate temporary storage
@@ -1363,6 +1372,7 @@
       reorder_based_on_method_index(methods, methods_annotations, temp_array);
       reorder_based_on_method_index(methods, methods_parameter_annotations, temp_array);
       reorder_based_on_method_index(methods, methods_default_annotations, temp_array);
+      reorder_based_on_method_index(methods, methods_type_annotations, temp_array);
     }
 
     // Reset method ordering
--- a/src/share/vm/oops/method.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/oops/method.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -160,6 +160,7 @@
                           int localvariable_table_length,
                           int exception_table_length,
                           int checked_exceptions_length,
+                          int method_parameters_length,
                           u2 generic_signature_index,
                           ConstMethod::MethodType method_type,
                           TRAPS);
@@ -225,6 +226,13 @@
     }
     return ik->annotations()->get_method_default_annotations_of(method_idnum());
   }
+  AnnotationArray* type_annotations() const {
+  InstanceKlass* ik = method_holder();
+  Annotations* type_annos = ik->type_annotations();
+  if (type_annos == NULL)
+    return NULL;
+  return type_annos->get_method_annotations_of(method_idnum());
+}
 
 #ifdef CC_INTERP
   void set_result_index(BasicType type);
@@ -343,7 +351,7 @@
   // exception handler which caused the exception to be thrown, which
   // is needed for proper retries. See, for example,
   // InterpreterRuntime::exception_handler_for_exception.
-  int fast_exception_handler_bci_for(KlassHandle ex_klass, int throw_bci, TRAPS);
+  static int fast_exception_handler_bci_for(methodHandle mh, KlassHandle ex_klass, int throw_bci, TRAPS);
 
   // method data access
   MethodData* method_data() const              {
@@ -473,6 +481,12 @@
   void print_codes_on(outputStream* st) const                      PRODUCT_RETURN;
   void print_codes_on(int from, int to, outputStream* st) const    PRODUCT_RETURN;
 
+  // method parameters
+  int method_parameters_length() const
+                         { return constMethod()->method_parameters_length(); }
+  MethodParametersElement* method_parameters_start() const
+                          { return constMethod()->method_parameters_start(); }
+
   // checked exceptions
   int checked_exceptions_length() const
                          { return constMethod()->checked_exceptions_length(); }
@@ -790,6 +804,7 @@
                            Array<AnnotationArray*>* methods_annotations,
                            Array<AnnotationArray*>* methods_parameter_annotations,
                            Array<AnnotationArray*>* methods_default_annotations,
+                           Array<AnnotationArray*>* methods_type_annotations,
                            bool idempotent = false);
 
   // Deallocation function for redefine classes or if an error occurs
--- a/src/share/vm/opto/bytecodeInfo.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/bytecodeInfo.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -46,7 +46,8 @@
   _method(callee),
   _site_invoke_ratio(site_invoke_ratio),
   _max_inline_level(max_inline_level),
-  _count_inline_bcs(method()->code_size_for_inlining())
+  _count_inline_bcs(method()->code_size_for_inlining()),
+  _subtrees(c->comp_arena(), 2, 0, NULL)
 {
   NOT_PRODUCT(_count_inlines = 0;)
   if (_caller_jvms != NULL) {
@@ -209,16 +210,18 @@
   if ( callee_method->dont_inline())                        return "don't inline by annotation";
   if ( callee_method->has_unloaded_classes_in_signature())  return "unloaded signature classes";
 
-  if (callee_method->force_inline() || callee_method->should_inline()) {
+  if (callee_method->should_inline()) {
     // ignore heuristic controls on inlining
     return NULL;
   }
 
   // Now perform checks which are heuristic
 
-  if (callee_method->has_compiled_code() &&
-      callee_method->instructions_size() > InlineSmallCode) {
+  if (!callee_method->force_inline()) {
+    if (callee_method->has_compiled_code() &&
+        callee_method->instructions_size() > InlineSmallCode) {
     return "already compiled into a big method";
+    }
   }
 
   // don't inline exception code unless the top method belongs to an
@@ -277,12 +280,15 @@
 //-----------------------------try_to_inline-----------------------------------
 // return NULL if ok, reason for not inlining otherwise
 // Relocated from "InliningClosure::try_to_inline"
-const char* InlineTree::try_to_inline(ciMethod* callee_method, ciMethod* caller_method, int caller_bci, ciCallProfile& profile, WarmCallInfo* wci_result) {
-
+const char* InlineTree::try_to_inline(ciMethod* callee_method, ciMethod* caller_method, int caller_bci, ciCallProfile& profile, WarmCallInfo* wci_result, bool& should_delay) {
   // Old algorithm had funny accumulating BC-size counters
   if (UseOldInlining && ClipInlining
       && (int)count_inline_bcs() >= DesiredMethodLimit) {
-    return "size > DesiredMethodLimit";
+    if (!callee_method->force_inline() || !IncrementalInline) {
+      return "size > DesiredMethodLimit";
+    } else if (!C->inlining_incrementally()) {
+      should_delay = true;
+    }
   }
 
   const char *msg = NULL;
@@ -303,8 +309,13 @@
   if (callee_method->code_size() > MaxTrivialSize) {
 
     // don't inline into giant methods
-    if (C->unique() > (uint)NodeCountInliningCutoff) {
-      return "NodeCountInliningCutoff";
+    if (C->over_inlining_cutoff()) {
+      if ((!callee_method->force_inline() && !caller_method->is_compiled_lambda_form())
+          || !IncrementalInline) {
+        return "NodeCountInliningCutoff";
+      } else {
+        should_delay = true;
+      }
     }
 
     if ((!UseInterpreter || CompileTheWorld) &&
@@ -323,7 +334,11 @@
     return "not an accessor";
   }
   if (inline_level() > _max_inline_level) {
-    return "inlining too deep";
+    if (!callee_method->force_inline() || !IncrementalInline) {
+      return "inlining too deep";
+    } else if (!C->inlining_incrementally()) {
+      should_delay = true;
+    }
   }
 
   // detect direct and indirect recursive inlining
@@ -348,7 +363,11 @@
 
   if (UseOldInlining && ClipInlining
       && (int)count_inline_bcs() + size >= DesiredMethodLimit) {
-    return "size > DesiredMethodLimit";
+    if (!callee_method->force_inline() || !IncrementalInline) {
+      return "size > DesiredMethodLimit";
+    } else if (!C->inlining_incrementally()) {
+      should_delay = true;
+    }
   }
 
   // ok, inline this method
@@ -413,8 +432,9 @@
 }
 
 //------------------------------ok_to_inline-----------------------------------
-WarmCallInfo* InlineTree::ok_to_inline(ciMethod* callee_method, JVMState* jvms, ciCallProfile& profile, WarmCallInfo* initial_wci) {
+WarmCallInfo* InlineTree::ok_to_inline(ciMethod* callee_method, JVMState* jvms, ciCallProfile& profile, WarmCallInfo* initial_wci, bool& should_delay) {
   assert(callee_method != NULL, "caller checks for optimized virtual!");
+  assert(!should_delay, "should be initialized to false");
 #ifdef ASSERT
   // Make sure the incoming jvms has the same information content as me.
   // This means that we can eventually make this whole class AllStatic.
@@ -444,7 +464,7 @@
 
   // Check if inlining policy says no.
   WarmCallInfo wci = *(initial_wci);
-  failure_msg = try_to_inline(callee_method, caller_method, caller_bci, profile, &wci);
+  failure_msg = try_to_inline(callee_method, caller_method, caller_bci, profile, &wci, should_delay);
   if (failure_msg != NULL && C->log() != NULL) {
     C->log()->inline_fail(failure_msg);
   }
--- a/src/share/vm/opto/c2_globals.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/c2_globals.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -606,6 +606,16 @@
                                                                             \
   develop(bool, VerifyAliases, false,                                       \
           "perform extra checks on the results of alias analysis")          \
+                                                                            \
+  product(bool, IncrementalInline, true,                                    \
+          "do post parse inlining")                                         \
+                                                                            \
+  develop(bool, AlwaysIncrementalInline, false,                             \
+          "do all inlining incrementally")                                  \
+                                                                            \
+  product(intx, LiveNodeCountInliningCutoff, 20000,                         \
+          "max number of live nodes in a method")                           \
+
 
 C2_FLAGS(DECLARE_DEVELOPER_FLAG, DECLARE_PD_DEVELOPER_FLAG, DECLARE_PRODUCT_FLAG, DECLARE_PD_PRODUCT_FLAG, DECLARE_DIAGNOSTIC_FLAG, DECLARE_EXPERIMENTAL_FLAG, DECLARE_NOTPRODUCT_FLAG)
 
--- a/src/share/vm/opto/callGenerator.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/callGenerator.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -262,8 +262,11 @@
 
 // Allow inlining decisions to be delayed
 class LateInlineCallGenerator : public DirectCallGenerator {
+ protected:
   CallGenerator* _inline_cg;
 
+  virtual bool do_late_inline_check(JVMState* jvms) { return true; }
+
  public:
   LateInlineCallGenerator(ciMethod* method, CallGenerator* inline_cg) :
     DirectCallGenerator(method, true), _inline_cg(inline_cg) {}
@@ -279,7 +282,9 @@
 
     // Record that this call site should be revisited once the main
     // parse is finished.
-    Compile::current()->add_late_inline(this);
+    if (!is_mh_late_inline()) {
+      C->add_late_inline(this);
+    }
 
     // Emit the CallStaticJava and request separate projections so
     // that the late inlining logic can distinguish between fall
@@ -287,15 +292,34 @@
     // as is done for allocations and macro expansion.
     return DirectCallGenerator::generate(jvms);
   }
+
+  virtual void print_inlining_late(const char* msg) {
+    CallNode* call = call_node();
+    Compile* C = Compile::current();
+    C->print_inlining_insert(this);
+    C->print_inlining(method(), call->jvms()->depth()-1, call->jvms()->bci(), msg);
+  }
+
 };
 
-
 void LateInlineCallGenerator::do_late_inline() {
   // Can't inline it
   if (call_node() == NULL || call_node()->outcnt() == 0 ||
       call_node()->in(0) == NULL || call_node()->in(0)->is_top())
     return;
 
+  for (int i1 = 0; i1 < method()->arg_size(); i1++) {
+    if (call_node()->in(TypeFunc::Parms + i1)->is_top()) {
+      assert(Compile::current()->inlining_incrementally(), "shouldn't happen during parsing");
+      return;
+    }
+  }
+
+  if (call_node()->in(TypeFunc::Memory)->is_top()) {
+    assert(Compile::current()->inlining_incrementally(), "shouldn't happen during parsing");
+    return;
+  }
+
   CallStaticJavaNode* call = call_node();
 
   // Make a clone of the JVMState that appropriate to use for driving a parse
@@ -324,6 +348,11 @@
     }
   }
 
+  if (!do_late_inline_check(jvms)) {
+    map->disconnect_inputs(NULL, C);
+    return;
+  }
+
   C->print_inlining_insert(this);
 
   CompileLog* log = C->log();
@@ -360,6 +389,10 @@
     result = (result_size == 1) ? kit.pop() : kit.pop_pair();
   }
 
+  C->set_has_loops(C->has_loops() || _inline_cg->method()->has_loops());
+  C->env()->notice_inlined_method(_inline_cg->method());
+  C->set_inlining_progress(true);
+
   kit.replace_call(call, result);
 }
 
@@ -368,6 +401,83 @@
   return new LateInlineCallGenerator(method, inline_cg);
 }
 
+class LateInlineMHCallGenerator : public LateInlineCallGenerator {
+  ciMethod* _caller;
+  int _attempt;
+  bool _input_not_const;
+
+  virtual bool do_late_inline_check(JVMState* jvms);
+  virtual bool already_attempted() const { return _attempt > 0; }
+
+ public:
+  LateInlineMHCallGenerator(ciMethod* caller, ciMethod* callee, bool input_not_const) :
+    LateInlineCallGenerator(callee, NULL), _caller(caller), _attempt(0), _input_not_const(input_not_const) {}
+
+  virtual bool is_mh_late_inline() const { return true; }
+
+  virtual JVMState* generate(JVMState* jvms) {
+    JVMState* new_jvms = LateInlineCallGenerator::generate(jvms);
+    if (_input_not_const) {
+      // inlining won't be possible so no need to enqueue right now.
+      call_node()->set_generator(this);
+    } else {
+      Compile::current()->add_late_inline(this);
+    }
+    return new_jvms;
+  }
+
+  virtual void print_inlining_late(const char* msg) {
+    if (!_input_not_const) return;
+    LateInlineCallGenerator::print_inlining_late(msg);
+  }
+};
+
+bool LateInlineMHCallGenerator::do_late_inline_check(JVMState* jvms) {
+
+  CallGenerator* cg = for_method_handle_inline(jvms, _caller, method(), _input_not_const);
+
+  if (!_input_not_const) {
+    _attempt++;
+  }
+
+  if (cg != NULL) {
+    assert(!cg->is_late_inline() && cg->is_inline(), "we're doing late inlining");
+    _inline_cg = cg;
+    Compile::current()->dec_number_of_mh_late_inlines();
+    return true;
+  }
+
+  call_node()->set_generator(this);
+  return false;
+}
+
+CallGenerator* CallGenerator::for_mh_late_inline(ciMethod* caller, ciMethod* callee, bool input_not_const) {
+  Compile::current()->inc_number_of_mh_late_inlines();
+  CallGenerator* cg = new LateInlineMHCallGenerator(caller, callee, input_not_const);
+  return cg;
+}
+
+class LateInlineStringCallGenerator : public LateInlineCallGenerator {
+
+ public:
+  LateInlineStringCallGenerator(ciMethod* method, CallGenerator* inline_cg) :
+    LateInlineCallGenerator(method, inline_cg) {}
+
+  virtual JVMState* generate(JVMState* jvms) {
+    Compile *C = Compile::current();
+    C->print_inlining_skip(this);
+
+    C->add_string_late_inline(this);
+
+    JVMState* new_jvms =  DirectCallGenerator::generate(jvms);
+    return new_jvms;
+  }
+};
+
+CallGenerator* CallGenerator::for_string_late_inline(ciMethod* method, CallGenerator* inline_cg) {
+  return new LateInlineStringCallGenerator(method, inline_cg);
+}
+
 
 //---------------------------WarmCallGenerator--------------------------------
 // Internal class which handles initial deferral of inlining decisions.
@@ -586,35 +696,53 @@
 }
 
 
-CallGenerator* CallGenerator::for_method_handle_call(JVMState* jvms, ciMethod* caller, ciMethod* callee) {
+CallGenerator* CallGenerator::for_method_handle_call(JVMState* jvms, ciMethod* caller, ciMethod* callee, bool delayed_forbidden) {
   assert(callee->is_method_handle_intrinsic() ||
          callee->is_compiled_lambda_form(), "for_method_handle_call mismatch");
-  CallGenerator* cg = CallGenerator::for_method_handle_inline(jvms, caller, callee);
-  if (cg != NULL)
-    return cg;
-  return CallGenerator::for_direct_call(callee);
+  bool input_not_const;
+  CallGenerator* cg = CallGenerator::for_method_handle_inline(jvms, caller, callee, input_not_const);
+  Compile* C = Compile::current();
+  if (cg != NULL) {
+    if (!delayed_forbidden && AlwaysIncrementalInline) {
+      return CallGenerator::for_late_inline(callee, cg);
+    } else {
+      return cg;
+    }
+  }
+  int bci = jvms->bci();
+  ciCallProfile profile = caller->call_profile_at_bci(bci);
+  int call_site_count = caller->scale_count(profile.count());
+
+  if (IncrementalInline && call_site_count > 0 &&
+      (input_not_const || !C->inlining_incrementally() || C->over_inlining_cutoff())) {
+    return CallGenerator::for_mh_late_inline(caller, callee, input_not_const);
+  } else {
+    // Out-of-line call.
+    return CallGenerator::for_direct_call(callee);
+  }
 }
 
-CallGenerator* CallGenerator::for_method_handle_inline(JVMState* jvms, ciMethod* caller, ciMethod* callee) {
+CallGenerator* CallGenerator::for_method_handle_inline(JVMState* jvms, ciMethod* caller, ciMethod* callee, bool& input_not_const) {
   GraphKit kit(jvms);
   PhaseGVN& gvn = kit.gvn();
   Compile* C = kit.C;
   vmIntrinsics::ID iid = callee->intrinsic_id();
+  input_not_const = true;
   switch (iid) {
   case vmIntrinsics::_invokeBasic:
     {
       // Get MethodHandle receiver:
       Node* receiver = kit.argument(0);
       if (receiver->Opcode() == Op_ConP) {
+        input_not_const = false;
         const TypeOopPtr* oop_ptr = receiver->bottom_type()->is_oopptr();
         ciMethod* target = oop_ptr->const_oop()->as_method_handle()->get_vmtarget();
         guarantee(!target->is_method_handle_intrinsic(), "should not happen");  // XXX remove
         const int vtable_index = Method::invalid_vtable_index;
-        CallGenerator* cg = C->call_generator(target, vtable_index, false, jvms, true, PROB_ALWAYS);
+        CallGenerator* cg = C->call_generator(target, vtable_index, false, jvms, true, PROB_ALWAYS, true, true);
+        assert(!cg->is_late_inline() || cg->is_mh_late_inline(), "no late inline here");
         if (cg != NULL && cg->is_inline())
           return cg;
-      } else {
-        if (PrintInlining)  C->print_inlining(callee, jvms->depth() - 1, jvms->bci(), "receiver not constant");
       }
     }
     break;
@@ -627,6 +755,7 @@
       // Get MemberName argument:
       Node* member_name = kit.argument(callee->arg_size() - 1);
       if (member_name->Opcode() == Op_ConP) {
+        input_not_const = false;
         const TypeOopPtr* oop_ptr = member_name->bottom_type()->is_oopptr();
         ciMethod* target = oop_ptr->const_oop()->as_member_name()->get_vmtarget();
 
@@ -659,9 +788,25 @@
             }
           }
         }
-        const int vtable_index = Method::invalid_vtable_index;
-        const bool call_is_virtual = target->is_abstract();  // FIXME workaround
-        CallGenerator* cg = C->call_generator(target, vtable_index, call_is_virtual, jvms, true, PROB_ALWAYS);
+
+        // Try to get the most accurate receiver type
+        const bool is_virtual              = (iid == vmIntrinsics::_linkToVirtual);
+        const bool is_virtual_or_interface = (is_virtual || iid == vmIntrinsics::_linkToInterface);
+        int  vtable_index       = Method::invalid_vtable_index;
+        bool call_does_dispatch = false;
+
+        if (is_virtual_or_interface) {
+          ciInstanceKlass* klass = target->holder();
+          Node*             receiver_node = kit.argument(0);
+          const TypeOopPtr* receiver_type = gvn.type(receiver_node)->isa_oopptr();
+          // call_does_dispatch and vtable_index are out-parameters.  They might be changed.
+          target = C->optimize_virtual_call(caller, jvms->bci(), klass, target, receiver_type,
+                                            is_virtual,
+                                            call_does_dispatch, vtable_index);  // out-parameters
+        }
+
+        CallGenerator* cg = C->call_generator(target, vtable_index, call_does_dispatch, jvms, true, PROB_ALWAYS, true, true);
+        assert(!cg->is_late_inline() || cg->is_mh_late_inline(), "no late inline here");
         if (cg != NULL && cg->is_inline())
           return cg;
       }
--- a/src/share/vm/opto/callGenerator.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/callGenerator.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -68,6 +68,12 @@
 
   // is_late_inline: supports conversion of call into an inline
   virtual bool      is_late_inline() const      { return false; }
+  // same but for method handle calls
+  virtual bool      is_mh_late_inline() const   { return false; }
+
+  // for method handle calls: have we tried inlinining the call already?
+  virtual bool      already_attempted() const   { ShouldNotReachHere(); return false; }
+
   // Replace the call with an inline version of the code
   virtual void do_late_inline() { ShouldNotReachHere(); }
 
@@ -112,11 +118,13 @@
   static CallGenerator* for_virtual_call(ciMethod* m, int vtable_index);  // virtual, interface
   static CallGenerator* for_dynamic_call(ciMethod* m);   // invokedynamic
 
-  static CallGenerator* for_method_handle_call(  JVMState* jvms, ciMethod* caller, ciMethod* callee);
-  static CallGenerator* for_method_handle_inline(JVMState* jvms, ciMethod* caller, ciMethod* callee);
+  static CallGenerator* for_method_handle_call(  JVMState* jvms, ciMethod* caller, ciMethod* callee, bool delayed_forbidden);
+  static CallGenerator* for_method_handle_inline(JVMState* jvms, ciMethod* caller, ciMethod* callee, bool& input_not_const);
 
   // How to generate a replace a direct call with an inline version
   static CallGenerator* for_late_inline(ciMethod* m, CallGenerator* inline_cg);
+  static CallGenerator* for_mh_late_inline(ciMethod* caller, ciMethod* callee, bool input_not_const);
+  static CallGenerator* for_string_late_inline(ciMethod* m, CallGenerator* inline_cg);
 
   // How to make a call but defer the decision whether to inline or not.
   static CallGenerator* for_warm_call(WarmCallInfo* ci,
@@ -147,6 +155,8 @@
                                                 CallGenerator* cg);
   virtual Node* generate_predicate(JVMState* jvms) { return NULL; };
 
+  virtual void print_inlining_late(const char* msg) { ShouldNotReachHere(); }
+
   static void print_inlining(Compile* C, ciMethod* callee, int inline_level, int bci, const char* msg) {
     if (PrintInlining)
       C->print_inlining(callee, inline_level, bci, msg);
--- a/src/share/vm/opto/callnode.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/callnode.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -25,6 +25,7 @@
 #include "precompiled.hpp"
 #include "ci/bcEscapeAnalyzer.hpp"
 #include "compiler/oopMap.hpp"
+#include "opto/callGenerator.hpp"
 #include "opto/callnode.hpp"
 #include "opto/escape.hpp"
 #include "opto/locknode.hpp"
@@ -775,16 +776,38 @@
   // and the exception object may not exist if an exception handler
   // swallows the exception but all the other must exist and be found.
   assert(projs->fallthrough_proj      != NULL, "must be found");
-  assert(projs->fallthrough_catchproj != NULL, "must be found");
-  assert(projs->fallthrough_memproj   != NULL, "must be found");
-  assert(projs->fallthrough_ioproj    != NULL, "must be found");
-  assert(projs->catchall_catchproj    != NULL, "must be found");
+  assert(Compile::current()->inlining_incrementally() || projs->fallthrough_catchproj != NULL, "must be found");
+  assert(Compile::current()->inlining_incrementally() || projs->fallthrough_memproj   != NULL, "must be found");
+  assert(Compile::current()->inlining_incrementally() || projs->fallthrough_ioproj    != NULL, "must be found");
+  assert(Compile::current()->inlining_incrementally() || projs->catchall_catchproj    != NULL, "must be found");
   if (separate_io_proj) {
-    assert(projs->catchall_memproj      != NULL, "must be found");
-    assert(projs->catchall_ioproj       != NULL, "must be found");
+    assert(Compile::current()->inlining_incrementally() || projs->catchall_memproj    != NULL, "must be found");
+    assert(Compile::current()->inlining_incrementally() || projs->catchall_ioproj     != NULL, "must be found");
   }
 }
 
+Node *CallNode::Ideal(PhaseGVN *phase, bool can_reshape) {
+  CallGenerator* cg = generator();
+  if (can_reshape && cg != NULL && cg->is_mh_late_inline() && !cg->already_attempted()) {
+    // Check whether this MH handle call becomes a candidate for inlining
+    ciMethod* callee = cg->method();
+    vmIntrinsics::ID iid = callee->intrinsic_id();
+    if (iid == vmIntrinsics::_invokeBasic) {
+      if (in(TypeFunc::Parms)->Opcode() == Op_ConP) {
+        phase->C->prepend_late_inline(cg);
+        set_generator(NULL);
+      }
+    } else {
+      assert(callee->has_member_arg(), "wrong type of call?");
+      if (in(TypeFunc::Parms + callee->arg_size() - 1)->Opcode() == Op_ConP) {
+        phase->C->prepend_late_inline(cg);
+        set_generator(NULL);
+      }
+    }
+  }
+  return SafePointNode::Ideal(phase, can_reshape);
+}
+
 
 //=============================================================================
 uint CallJavaNode::size_of() const { return sizeof(*this); }
--- a/src/share/vm/opto/callnode.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/callnode.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -507,6 +507,7 @@
   Node* exobj;
 };
 
+class CallGenerator;
 
 //------------------------------CallNode---------------------------------------
 // Call nodes now subsume the function of debug nodes at callsites, so they
@@ -517,26 +518,31 @@
   const TypeFunc *_tf;        // Function type
   address      _entry_point;  // Address of method being called
   float        _cnt;          // Estimate of number of times called
+  CallGenerator* _generator;  // corresponding CallGenerator for some late inline calls
 
   CallNode(const TypeFunc* tf, address addr, const TypePtr* adr_type)
     : SafePointNode(tf->domain()->cnt(), NULL, adr_type),
       _tf(tf),
       _entry_point(addr),
-      _cnt(COUNT_UNKNOWN)
+      _cnt(COUNT_UNKNOWN),
+      _generator(NULL)
   {
     init_class_id(Class_Call);
   }
 
-  const TypeFunc* tf()        const { return _tf; }
-  const address entry_point() const { return _entry_point; }
-  const float   cnt()         const { return _cnt; }
+  const TypeFunc* tf()         const { return _tf; }
+  const address  entry_point() const { return _entry_point; }
+  const float    cnt()         const { return _cnt; }
+  CallGenerator* generator()   const { return _generator; }
 
-  void set_tf(const TypeFunc* tf) { _tf = tf; }
-  void set_entry_point(address p) { _entry_point = p; }
-  void set_cnt(float c)           { _cnt = c; }
+  void set_tf(const TypeFunc* tf)       { _tf = tf; }
+  void set_entry_point(address p)       { _entry_point = p; }
+  void set_cnt(float c)                 { _cnt = c; }
+  void set_generator(CallGenerator* cg) { _generator = cg; }
 
   virtual const Type *bottom_type() const;
   virtual const Type *Value( PhaseTransform *phase ) const;
+  virtual Node *Ideal(PhaseGVN *phase, bool can_reshape);
   virtual Node *Identity( PhaseTransform *phase ) { return this; }
   virtual uint        cmp( const Node &n ) const;
   virtual uint        size_of() const = 0;
--- a/src/share/vm/opto/cfgnode.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/cfgnode.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -363,6 +363,49 @@
   return true; // The Region node is unreachable - it is dead.
 }
 
+bool RegionNode::try_clean_mem_phi(PhaseGVN *phase) {
+  // Incremental inlining + PhaseStringOpts sometimes produce:
+  //
+  // cmpP with 1 top input
+  //           |
+  //          If
+  //         /  \
+  //   IfFalse  IfTrue  /- Some Node
+  //         \  /      /    /
+  //        Region    / /-MergeMem
+  //             \---Phi
+  //
+  //
+  // It's expected by PhaseStringOpts that the Region goes away and is
+  // replaced by If's control input but because there's still a Phi,
+  // the Region stays in the graph. The top input from the cmpP is
+  // propagated forward and a subgraph that is useful goes away. The
+  // code below replaces the Phi with the MergeMem so that the Region
+  // is simplified.
+
+  PhiNode* phi = has_unique_phi();
+  if (phi && phi->type() == Type::MEMORY && req() == 3 && phi->is_diamond_phi(true)) {
+    MergeMemNode* m = NULL;
+    assert(phi->req() == 3, "same as region");
+    for (uint i = 1; i < 3; ++i) {
+      Node *mem = phi->in(i);
+      if (mem && mem->is_MergeMem() && in(i)->outcnt() == 1) {
+        // Nothing is control-dependent on path #i except the region itself.
+        m = mem->as_MergeMem();
+        uint j = 3 - i;
+        Node* other = phi->in(j);
+        if (other && other == m->base_memory()) {
+          // m is a successor memory to other, and is not pinned inside the diamond, so push it out.
+          // This will allow the diamond to collapse completely.
+          phase->is_IterGVN()->replace_node(phi, m);
+          return true;
+        }
+      }
+    }
+  }
+  return false;
+}
+
 //------------------------------Ideal------------------------------------------
 // Return a node which is more "ideal" than the current node.  Must preserve
 // the CFG, but we can still strip out dead paths.
@@ -375,6 +418,10 @@
   bool has_phis = false;
   if (can_reshape) {            // Need DU info to check for Phi users
     has_phis = (has_phi() != NULL);       // Cache result
+    if (has_phis && try_clean_mem_phi(phase)) {
+      has_phis = false;
+    }
+
     if (!has_phis) {            // No Phi users?  Nothing merging?
       for (uint i = 1; i < req()-1; i++) {
         Node *if1 = in(i);
@@ -1005,7 +1052,9 @@
 //------------------------------is_diamond_phi---------------------------------
 // Does this Phi represent a simple well-shaped diamond merge?  Return the
 // index of the true path or 0 otherwise.
-int PhiNode::is_diamond_phi() const {
+// If check_control_only is true, do not inspect the If node at the
+// top, and return -1 (not an edge number) on success.
+int PhiNode::is_diamond_phi(bool check_control_only) const {
   // Check for a 2-path merge
   Node *region = in(0);
   if( !region ) return 0;
@@ -1018,6 +1067,7 @@
   Node *iff = ifp1->in(0);
   if( !iff || !iff->is_If() ) return 0;
   if( iff != ifp2->in(0) ) return 0;
+  if (check_control_only)  return -1;
   // Check for a proper bool/cmp
   const Node *b = iff->in(1);
   if( !b->is_Bool() ) return 0;
--- a/src/share/vm/opto/cfgnode.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/cfgnode.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -95,6 +95,7 @@
   virtual Node *Identity( PhaseTransform *phase );
   virtual Node *Ideal(PhaseGVN *phase, bool can_reshape);
   virtual const RegMask &out_RegMask() const;
+  bool try_clean_mem_phi(PhaseGVN *phase);
 };
 
 //------------------------------JProjNode--------------------------------------
@@ -181,7 +182,7 @@
   LoopSafety simple_data_loop_check(Node *in) const;
   // Is it unsafe data loop? It becomes a dead loop if this phi node removed.
   bool is_unsafe_data_reference(Node *in) const;
-  int  is_diamond_phi() const;
+  int  is_diamond_phi(bool check_control_only = false) const;
   virtual int Opcode() const;
   virtual bool pinned() const { return in(0) != 0; }
   virtual const TypePtr *adr_type() const { verify_adr_type(true); return _adr_type; }
--- a/src/share/vm/opto/compile.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/compile.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -136,7 +136,7 @@
 
 void Compile::register_intrinsic(CallGenerator* cg) {
   if (_intrinsics == NULL) {
-    _intrinsics = new GrowableArray<CallGenerator*>(60);
+    _intrinsics = new (comp_arena())GrowableArray<CallGenerator*>(comp_arena(), 60, 0, NULL);
   }
   // This code is stolen from ciObjectFactory::insert.
   // Really, GrowableArray should have methods for
@@ -365,6 +365,21 @@
   }
 }
 
+void Compile::remove_useless_late_inlines(GrowableArray<CallGenerator*>* inlines, Unique_Node_List &useful) {
+  int shift = 0;
+  for (int i = 0; i < inlines->length(); i++) {
+    CallGenerator* cg = inlines->at(i);
+    CallNode* call = cg->call_node();
+    if (shift > 0) {
+      inlines->at_put(i-shift, cg);
+    }
+    if (!useful.member(call)) {
+      shift++;
+    }
+  }
+  inlines->trunc_to(inlines->length()-shift);
+}
+
 // Disconnect all useless nodes by disconnecting those at the boundary.
 void Compile::remove_useless_nodes(Unique_Node_List &useful) {
   uint next = 0;
@@ -394,6 +409,9 @@
       remove_macro_node(n);
     }
   }
+  // clean up the late inline lists
+  remove_useless_late_inlines(&_string_late_inlines, useful);
+  remove_useless_late_inlines(&_late_inlines, useful);
   debug_only(verify_graph_edges(true/*check for no_dead_code*/);)
 }
 
@@ -611,6 +629,12 @@
                   _printer(IdealGraphPrinter::printer()),
 #endif
                   _congraph(NULL),
+                  _late_inlines(comp_arena(), 2, 0, NULL),
+                  _string_late_inlines(comp_arena(), 2, 0, NULL),
+                  _late_inlines_pos(0),
+                  _number_of_mh_late_inlines(0),
+                  _inlining_progress(false),
+                  _inlining_incrementally(false),
                   _print_inlining_list(NULL),
                   _print_inlining(0) {
   C = this;
@@ -737,29 +761,13 @@
       rethrow_exceptions(kit.transfer_exceptions_into_jvms());
     }
 
-    if (!failing() && has_stringbuilder()) {
-      {
-        // remove useless nodes to make the usage analysis simpler
-        ResourceMark rm;
-        PhaseRemoveUseless pru(initial_gvn(), &for_igvn);
-      }
-
-      {
-        ResourceMark rm;
-        print_method("Before StringOpts", 3);
-        PhaseStringOpts pso(initial_gvn(), &for_igvn);
-        print_method("After StringOpts", 3);
-      }
-
-      // now inline anything that we skipped the first time around
-      while (_late_inlines.length() > 0) {
-        CallGenerator* cg = _late_inlines.pop();
-        cg->do_late_inline();
-        if (failing())  return;
-      }
+    assert(IncrementalInline || (_late_inlines.length() == 0 && !has_mh_late_inlines()), "incremental inlining is off");
+
+    if (_late_inlines.length() == 0 && !has_mh_late_inlines() && !failing() && has_stringbuilder()) {
+      inline_string_calls(true);
     }
-    assert(_late_inlines.length() == 0, "should have been processed");
-    dump_inlining();
+
+    if (failing())  return;
 
     print_method("Before RemoveUseless", 3);
 
@@ -906,6 +914,9 @@
     _dead_node_list(comp_arena()),
     _dead_node_count(0),
     _congraph(NULL),
+    _number_of_mh_late_inlines(0),
+    _inlining_progress(false),
+    _inlining_incrementally(false),
     _print_inlining_list(NULL),
     _print_inlining(0) {
   C = this;
@@ -1760,6 +1771,124 @@
   assert(predicate_count()==0, "should be clean!");
 }
 
+// StringOpts and late inlining of string methods
+void Compile::inline_string_calls(bool parse_time) {
+  {
+    // remove useless nodes to make the usage analysis simpler
+    ResourceMark rm;
+    PhaseRemoveUseless pru(initial_gvn(), for_igvn());
+  }
+
+  {
+    ResourceMark rm;
+    print_method("Before StringOpts", 3);
+    PhaseStringOpts pso(initial_gvn(), for_igvn());
+    print_method("After StringOpts", 3);
+  }
+
+  // now inline anything that we skipped the first time around
+  if (!parse_time) {
+    _late_inlines_pos = _late_inlines.length();
+  }
+
+  while (_string_late_inlines.length() > 0) {
+    CallGenerator* cg = _string_late_inlines.pop();
+    cg->do_late_inline();
+    if (failing())  return;
+  }
+  _string_late_inlines.trunc_to(0);
+}
+
+void Compile::inline_incrementally_one(PhaseIterGVN& igvn) {
+  assert(IncrementalInline, "incremental inlining should be on");
+  PhaseGVN* gvn = initial_gvn();
+
+  set_inlining_progress(false);
+  for_igvn()->clear();
+  gvn->replace_with(&igvn);
+
+  int i = 0;
+
+  for (; i <_late_inlines.length() && !inlining_progress(); i++) {
+    CallGenerator* cg = _late_inlines.at(i);
+    _late_inlines_pos = i+1;
+    cg->do_late_inline();
+    if (failing())  return;
+  }
+  int j = 0;
+  for (; i < _late_inlines.length(); i++, j++) {
+    _late_inlines.at_put(j, _late_inlines.at(i));
+  }
+  _late_inlines.trunc_to(j);
+
+  {
+    ResourceMark rm;
+    PhaseRemoveUseless pru(C->initial_gvn(), C->for_igvn());
+  }
+
+  igvn = PhaseIterGVN(gvn);
+}
+
+// Perform incremental inlining until bound on number of live nodes is reached
+void Compile::inline_incrementally(PhaseIterGVN& igvn) {
+  PhaseGVN* gvn = initial_gvn();
+
+  set_inlining_incrementally(true);
+  set_inlining_progress(true);
+  uint low_live_nodes = 0;
+
+  while(inlining_progress() && _late_inlines.length() > 0) {
+
+    if (live_nodes() > (uint)LiveNodeCountInliningCutoff) {
+      if (low_live_nodes < (uint)LiveNodeCountInliningCutoff * 8 / 10) {
+        // PhaseIdealLoop is expensive so we only try it once we are
+        // out of loop and we only try it again if the previous helped
+        // got the number of nodes down significantly
+        PhaseIdealLoop ideal_loop( igvn, false, true );
+        if (failing())  return;
+        low_live_nodes = live_nodes();
+        _major_progress = true;
+      }
+
+      if (live_nodes() > (uint)LiveNodeCountInliningCutoff) {
+        break;
+      }
+    }
+
+    inline_incrementally_one(igvn);
+
+    if (failing())  return;
+
+    igvn.optimize();
+
+    if (failing())  return;
+  }
+
+  assert( igvn._worklist.size() == 0, "should be done with igvn" );
+
+  if (_string_late_inlines.length() > 0) {
+    assert(has_stringbuilder(), "inconsistent");
+    for_igvn()->clear();
+    initial_gvn()->replace_with(&igvn);
+
+    inline_string_calls(false);
+
+    if (failing())  return;
+
+    {
+      ResourceMark rm;
+      PhaseRemoveUseless pru(initial_gvn(), for_igvn());
+    }
+
+    igvn = PhaseIterGVN(gvn);
+
+    igvn.optimize();
+  }
+
+  set_inlining_incrementally(false);
+}
+
+
 //------------------------------Optimize---------------------------------------
 // Given a graph, optimize it.
 void Compile::Optimize() {
@@ -1792,6 +1921,12 @@
 
   if (failing())  return;
 
+  inline_incrementally(igvn);
+
+  print_method("Incremental Inline", 2);
+
+  if (failing())  return;
+
   // Perform escape analysis
   if (_do_escape_analysis && ConnectionGraph::has_candidates(this)) {
     if (has_loops()) {
@@ -1914,6 +2049,7 @@
 
  } // (End scope of igvn; run destructor if necessary for asserts.)
 
+ dump_inlining();
   // A method with only infinite loops has no edges entering loops from root
   {
     NOT_PRODUCT( TracePhase t2("graphReshape", &_t_graphReshaping, TimeCompiler); )
@@ -3362,6 +3498,28 @@
 
 void Compile::dump_inlining() {
   if (PrintInlining) {
+    // Print inlining message for candidates that we couldn't inline
+    // for lack of space or non constant receiver
+    for (int i = 0; i < _late_inlines.length(); i++) {
+      CallGenerator* cg = _late_inlines.at(i);
+      cg->print_inlining_late("live nodes > LiveNodeCountInliningCutoff");
+    }
+    Unique_Node_List useful;
+    useful.push(root());
+    for (uint next = 0; next < useful.size(); ++next) {
+      Node* n  = useful.at(next);
+      if (n->is_Call() && n->as_Call()->generator() != NULL && n->as_Call()->generator()->call_node() == n) {
+        CallNode* call = n->as_Call();
+        CallGenerator* cg = call->generator();
+        cg->print_inlining_late("receiver not constant");
+      }
+      uint max = n->len();
+      for ( uint i = 0; i < max; ++i ) {
+        Node *m = n->in(i);
+        if ( m == NULL ) continue;
+        useful.push(m);
+      }
+    }
     for (int i = 0; i < _print_inlining_list->length(); i++) {
       tty->print(_print_inlining_list->at(i).ss()->as_string());
     }
--- a/src/share/vm/opto/compile.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/compile.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -72,6 +72,7 @@
 class JVMState;
 class TypeData;
 class TypePtr;
+class TypeOopPtr;
 class TypeFunc;
 class Unique_Node_List;
 class nmethod;
@@ -280,6 +281,8 @@
   int                   _orig_pc_slot_offset_in_bytes;
 
   int                   _major_progress;        // Count of something big happening
+  bool                  _inlining_progress;     // progress doing incremental inlining?
+  bool                  _inlining_incrementally;// Are we doing incremental inlining (post parse)
   bool                  _has_loops;             // True if the method _may_ have some loops
   bool                  _has_split_ifs;         // True if the method _may_ have some split-if
   bool                  _has_unsafe_access;     // True if the method _may_ produce faults in unsafe loads or stores.
@@ -367,8 +370,13 @@
   Unique_Node_List*     _for_igvn;              // Initial work-list for next round of Iterative GVN
   WarmCallInfo*         _warm_calls;            // Sorted work-list for heat-based inlining.
 
-  GrowableArray<CallGenerator*> _late_inlines;  // List of CallGenerators to be revisited after
-                                                // main parsing has finished.
+  GrowableArray<CallGenerator*> _late_inlines;        // List of CallGenerators to be revisited after
+                                                      // main parsing has finished.
+  GrowableArray<CallGenerator*> _string_late_inlines; // same but for string operations
+
+  int                           _late_inlines_pos;    // Where in the queue should the next late inlining candidate go (emulate depth first inlining)
+  uint                          _number_of_mh_late_inlines; // number of method handle late inlining still pending
+
 
   // Inlining may not happen in parse order which would make
   // PrintInlining output confusing. Keep track of PrintInlining
@@ -491,6 +499,10 @@
   int               fixed_slots() const         { assert(_fixed_slots >= 0, "");         return _fixed_slots; }
   void          set_fixed_slots(int n)          { _fixed_slots = n; }
   int               major_progress() const      { return _major_progress; }
+  void          set_inlining_progress(bool z)   { _inlining_progress = z; }
+  int               inlining_progress() const   { return _inlining_progress; }
+  void          set_inlining_incrementally(bool z) { _inlining_incrementally = z; }
+  int               inlining_incrementally() const { return _inlining_incrementally; }
   void          set_major_progress()            { _major_progress++; }
   void        clear_major_progress()            { _major_progress = 0; }
   int               num_loop_opts() const       { return _num_loop_opts; }
@@ -729,9 +741,17 @@
 
   // Decide how to build a call.
   // The profile factor is a discount to apply to this site's interp. profile.
-  CallGenerator*    call_generator(ciMethod* call_method, int vtable_index, bool call_is_virtual, JVMState* jvms, bool allow_inline, float profile_factor, bool allow_intrinsics = true);
+  CallGenerator*    call_generator(ciMethod* call_method, int vtable_index, bool call_does_dispatch, JVMState* jvms, bool allow_inline, float profile_factor, bool allow_intrinsics = true, bool delayed_forbidden = false);
   bool should_delay_inlining(ciMethod* call_method, JVMState* jvms);
 
+  // Helper functions to identify inlining potential at call-site
+  ciMethod* optimize_virtual_call(ciMethod* caller, int bci, ciInstanceKlass* klass,
+                                  ciMethod* callee, const TypeOopPtr* receiver_type,
+                                  bool is_virtual,
+                                  bool &call_does_dispatch, int &vtable_index);
+  ciMethod* optimize_inlining(ciMethod* caller, int bci, ciInstanceKlass* klass,
+                              ciMethod* callee, const TypeOopPtr* receiver_type);
+
   // Report if there were too many traps at a current method and bci.
   // Report if a trap was recorded, and/or PerMethodTrapLimit was exceeded.
   // If there is no MDO at all, report no trap unless told to assume it.
@@ -765,10 +785,39 @@
   WarmCallInfo* pop_warm_call();
 
   // Record this CallGenerator for inlining at the end of parsing.
-  void              add_late_inline(CallGenerator* cg) { _late_inlines.push(cg); }
+  void              add_late_inline(CallGenerator* cg)        {
+    _late_inlines.insert_before(_late_inlines_pos, cg);
+    _late_inlines_pos++;
+  }
+
+  void              prepend_late_inline(CallGenerator* cg)    {
+    _late_inlines.insert_before(0, cg);
+  }
+
+  void              add_string_late_inline(CallGenerator* cg) {
+    _string_late_inlines.push(cg);
+  }
+
+  void remove_useless_late_inlines(GrowableArray<CallGenerator*>* inlines, Unique_Node_List &useful);
 
   void dump_inlining();
 
+  bool over_inlining_cutoff() const {
+    if (!inlining_incrementally()) {
+      return unique() > (uint)NodeCountInliningCutoff;
+    } else {
+      return live_nodes() > (uint)LiveNodeCountInliningCutoff;
+    }
+  }
+
+  void inc_number_of_mh_late_inlines() { _number_of_mh_late_inlines++; }
+  void dec_number_of_mh_late_inlines() { assert(_number_of_mh_late_inlines > 0, "_number_of_mh_late_inlines < 0 !"); _number_of_mh_late_inlines--; }
+  bool has_mh_late_inlines() const     { return _number_of_mh_late_inlines > 0; }
+
+  void inline_incrementally_one(PhaseIterGVN& igvn);
+  void inline_incrementally(PhaseIterGVN& igvn);
+  void inline_string_calls(bool parse_time);
+
   // Matching, CFG layout, allocation, code generation
   PhaseCFG*         cfg()                       { return _cfg; }
   bool              select_24_bit_instr() const { return _select_24_bit_instr; }
--- a/src/share/vm/opto/doCall.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/doCall.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -61,9 +61,9 @@
   }
 }
 
-CallGenerator* Compile::call_generator(ciMethod* callee, int vtable_index, bool call_is_virtual,
+CallGenerator* Compile::call_generator(ciMethod* callee, int vtable_index, bool call_does_dispatch,
                                        JVMState* jvms, bool allow_inline,
-                                       float prof_factor, bool allow_intrinsics) {
+                                       float prof_factor, bool allow_intrinsics, bool delayed_forbidden) {
   ciMethod*       caller   = jvms->method();
   int             bci      = jvms->bci();
   Bytecodes::Code bytecode = caller->java_code_at_bci(bci);
@@ -82,7 +82,7 @@
   // See how many times this site has been invoked.
   int site_count = profile.count();
   int receiver_count = -1;
-  if (call_is_virtual && UseTypeProfile && profile.has_receiver(0)) {
+  if (call_does_dispatch && UseTypeProfile && profile.has_receiver(0)) {
     // Receivers in the profile structure are ordered by call counts
     // so that the most called (major) receiver is profile.receiver(0).
     receiver_count = profile.receiver_count(0);
@@ -94,7 +94,7 @@
     int r2id = (rid != -1 && profile.has_receiver(1))? log->identify(profile.receiver(1)):-1;
     log->begin_elem("call method='%d' count='%d' prof_factor='%g'",
                     log->identify(callee), site_count, prof_factor);
-    if (call_is_virtual)  log->print(" virtual='1'");
+    if (call_does_dispatch)  log->print(" virtual='1'");
     if (allow_inline)     log->print(" inline='1'");
     if (receiver_count >= 0) {
       log->print(" receiver='%d' receiver_count='%d'", rid, receiver_count);
@@ -111,12 +111,12 @@
   // We do this before the strict f.p. check below because the
   // intrinsics handle strict f.p. correctly.
   if (allow_inline && allow_intrinsics) {
-    CallGenerator* cg = find_intrinsic(callee, call_is_virtual);
+    CallGenerator* cg = find_intrinsic(callee, call_does_dispatch);
     if (cg != NULL) {
       if (cg->is_predicted()) {
         // Code without intrinsic but, hopefully, inlined.
         CallGenerator* inline_cg = this->call_generator(callee,
-              vtable_index, call_is_virtual, jvms, allow_inline, prof_factor, false);
+              vtable_index, call_does_dispatch, jvms, allow_inline, prof_factor, false);
         if (inline_cg != NULL) {
           cg = CallGenerator::for_predicted_intrinsic(cg, inline_cg);
         }
@@ -130,7 +130,9 @@
   // MethodHandle.invoke* are native methods which obviously don't
   // have bytecodes and so normal inlining fails.
   if (callee->is_method_handle_intrinsic()) {
-    return CallGenerator::for_method_handle_call(jvms, caller, callee);
+    CallGenerator* cg = CallGenerator::for_method_handle_call(jvms, caller, callee, delayed_forbidden);
+    assert(cg == NULL || !delayed_forbidden || !cg->is_late_inline() || cg->is_mh_late_inline(), "unexpected CallGenerator");
+    return cg;
   }
 
   // Do not inline strict fp into non-strict code, or the reverse
@@ -147,7 +149,7 @@
     float expected_uses = past_uses;
 
     // Try inlining a bytecoded method:
-    if (!call_is_virtual) {
+    if (!call_does_dispatch) {
       InlineTree* ilt;
       if (UseOldInlining) {
         ilt = InlineTree::find_subtree_from_root(this->ilt(), jvms->caller(), jvms->method());
@@ -161,32 +163,39 @@
       WarmCallInfo scratch_ci;
       if (!UseOldInlining)
         scratch_ci.init(jvms, callee, profile, prof_factor);
-      WarmCallInfo* ci = ilt->ok_to_inline(callee, jvms, profile, &scratch_ci);
+      bool should_delay = false;
+      WarmCallInfo* ci = ilt->ok_to_inline(callee, jvms, profile, &scratch_ci, should_delay);
       assert(ci != &scratch_ci, "do not let this pointer escape");
       bool allow_inline   = (ci != NULL && !ci->is_cold());
       bool require_inline = (allow_inline && ci->is_hot());
 
       if (allow_inline) {
         CallGenerator* cg = CallGenerator::for_inline(callee, expected_uses);
-        if (require_inline && cg != NULL && should_delay_inlining(callee, jvms)) {
+
+        if (require_inline && cg != NULL) {
           // Delay the inlining of this method to give us the
           // opportunity to perform some high level optimizations
           // first.
-          return CallGenerator::for_late_inline(callee, cg);
+          if (should_delay_inlining(callee, jvms)) {
+            assert(!delayed_forbidden, "strange");
+            return CallGenerator::for_string_late_inline(callee, cg);
+          } else if ((should_delay || AlwaysIncrementalInline) && !delayed_forbidden) {
+            return CallGenerator::for_late_inline(callee, cg);
+          }
         }
-        if (cg == NULL) {
+        if (cg == NULL || should_delay) {
           // Fall through.
         } else if (require_inline || !InlineWarmCalls) {
           return cg;
         } else {
-          CallGenerator* cold_cg = call_generator(callee, vtable_index, call_is_virtual, jvms, false, prof_factor);
+          CallGenerator* cold_cg = call_generator(callee, vtable_index, call_does_dispatch, jvms, false, prof_factor);
           return CallGenerator::for_warm_call(ci, cold_cg, cg);
         }
       }
     }
 
     // Try using the type profile.
-    if (call_is_virtual && site_count > 0 && receiver_count > 0) {
+    if (call_does_dispatch && site_count > 0 && receiver_count > 0) {
       // The major receiver's count >= TypeProfileMajorReceiverPercent of site_count.
       bool have_major_receiver = (100.*profile.receiver_prob(0) >= (float)TypeProfileMajorReceiverPercent);
       ciMethod* receiver_method = NULL;
@@ -200,7 +209,7 @@
       if (receiver_method != NULL) {
         // The single majority receiver sufficiently outweighs the minority.
         CallGenerator* hit_cg = this->call_generator(receiver_method,
-              vtable_index, !call_is_virtual, jvms, allow_inline, prof_factor);
+              vtable_index, !call_does_dispatch, jvms, allow_inline, prof_factor);
         if (hit_cg != NULL) {
           // Look up second receiver.
           CallGenerator* next_hit_cg = NULL;
@@ -210,7 +219,7 @@
                                                                profile.receiver(1));
             if (next_receiver_method != NULL) {
               next_hit_cg = this->call_generator(next_receiver_method,
-                                  vtable_index, !call_is_virtual, jvms,
+                                  vtable_index, !call_does_dispatch, jvms,
                                   allow_inline, prof_factor);
               if (next_hit_cg != NULL && !next_hit_cg->is_inline() &&
                   have_major_receiver && UseOnlyInlinedBimorphic) {
@@ -256,7 +265,7 @@
 
   // There was no special inlining tactic, or it bailed out.
   // Use a more generic tactic, like a simple call.
-  if (call_is_virtual) {
+  if (call_does_dispatch) {
     return CallGenerator::for_virtual_call(callee, vtable_index);
   } else {
     // Class Hierarchy Analysis or Type Profile reveals a unique target,
@@ -388,6 +397,7 @@
   // orig_callee is the resolved callee which's signature includes the
   // appendix argument.
   const int nargs = orig_callee->arg_size();
+  const bool is_signature_polymorphic = MethodHandles::is_signature_polymorphic(orig_callee->intrinsic_id());
 
   // Push appendix argument (MethodType, CallSite, etc.), if one.
   if (iter().has_appendix()) {
@@ -404,25 +414,18 @@
   // Then we may introduce a run-time check and inline on the path where it succeeds.
   // The other path may uncommon_trap, check for another receiver, or do a v-call.
 
-  // Choose call strategy.
-  bool call_is_virtual = is_virtual_or_interface;
-  int vtable_index = Method::invalid_vtable_index;
-  ciMethod* callee = orig_callee;
+  // Try to get the most accurate receiver type
+  ciMethod* callee             = orig_callee;
+  int       vtable_index       = Method::invalid_vtable_index;
+  bool      call_does_dispatch = false;
 
-  // Try to get the most accurate receiver type
   if (is_virtual_or_interface) {
     Node*             receiver_node = stack(sp() - nargs);
     const TypeOopPtr* receiver_type = _gvn.type(receiver_node)->isa_oopptr();
-    ciMethod* optimized_virtual_method = optimize_inlining(method(), bci(), klass, orig_callee, receiver_type);
-
-    // Have the call been sufficiently improved such that it is no longer a virtual?
-    if (optimized_virtual_method != NULL) {
-      callee          = optimized_virtual_method;
-      call_is_virtual = false;
-    } else if (!UseInlineCaches && is_virtual && callee->is_loaded()) {
-      // We can make a vtable call at this site
-      vtable_index = callee->resolve_vtable_index(method()->holder(), klass);
-    }
+    // call_does_dispatch and vtable_index are out-parameters.  They might be changed.
+    callee = C->optimize_virtual_call(method(), bci(), klass, orig_callee, receiver_type,
+                                      is_virtual,
+                                      call_does_dispatch, vtable_index);  // out-parameters
   }
 
   // Note:  It's OK to try to inline a virtual call.
@@ -438,7 +441,7 @@
   // Decide call tactic.
   // This call checks with CHA, the interpreter profile, intrinsics table, etc.
   // It decides whether inlining is desirable or not.
-  CallGenerator* cg = C->call_generator(callee, vtable_index, call_is_virtual, jvms, try_inline, prof_factor());
+  CallGenerator* cg = C->call_generator(callee, vtable_index, call_does_dispatch, jvms, try_inline, prof_factor());
 
   // NOTE:  Don't use orig_callee and callee after this point!  Use cg->method() instead.
   orig_callee = callee = NULL;
@@ -478,7 +481,7 @@
     // the call site, perhaps because it did not match a pattern the
     // intrinsic was expecting to optimize. Should always be possible to
     // get a normal java call that may inline in that case
-    cg = C->call_generator(cg->method(), vtable_index, call_is_virtual, jvms, try_inline, prof_factor(), /* allow_intrinsics= */ false);
+    cg = C->call_generator(cg->method(), vtable_index, call_does_dispatch, jvms, try_inline, prof_factor(), /* allow_intrinsics= */ false);
     if ((new_jvms = cg->generate(jvms)) == NULL) {
       guarantee(failing(), "call failed to generate:  calls should work");
       return;
@@ -513,55 +516,44 @@
     round_double_result(cg->method());
 
     ciType* rtype = cg->method()->return_type();
-    if (Bytecodes::has_optional_appendix(iter().cur_bc_raw())) {
+    ciType* ctype = declared_signature->return_type();
+
+    if (Bytecodes::has_optional_appendix(iter().cur_bc_raw()) || is_signature_polymorphic) {
       // Be careful here with return types.
-      ciType* ctype = declared_signature->return_type();
       if (ctype != rtype) {
         BasicType rt = rtype->basic_type();
         BasicType ct = ctype->basic_type();
-        Node* retnode = peek();
         if (ct == T_VOID) {
           // It's OK for a method  to return a value that is discarded.
           // The discarding does not require any special action from the caller.
           // The Java code knows this, at VerifyType.isNullConversion.
           pop_node(rt);  // whatever it was, pop it
-          retnode = top();
         } else if (rt == T_INT || is_subword_type(rt)) {
-          // FIXME: This logic should be factored out.
-          if (ct == T_BOOLEAN) {
-            retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0x1)) );
-          } else if (ct == T_CHAR) {
-            retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0xFFFF)) );
-          } else if (ct == T_BYTE) {
-            retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(24)) );
-            retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(24)) );
-          } else if (ct == T_SHORT) {
-            retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(16)) );
-            retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(16)) );
-          } else {
-            assert(ct == T_INT, err_msg_res("rt=%s, ct=%s", type2name(rt), type2name(ct)));
-          }
+          // Nothing.  These cases are handled in lambda form bytecode.
+          assert(ct == T_INT || is_subword_type(ct), err_msg_res("must match: rt=%s, ct=%s", type2name(rt), type2name(ct)));
         } else if (rt == T_OBJECT || rt == T_ARRAY) {
           assert(ct == T_OBJECT || ct == T_ARRAY, err_msg_res("rt=%s, ct=%s", type2name(rt), type2name(ct)));
           if (ctype->is_loaded()) {
             const TypeOopPtr* arg_type = TypeOopPtr::make_from_klass(rtype->as_klass());
             const Type*       sig_type = TypeOopPtr::make_from_klass(ctype->as_klass());
             if (arg_type != NULL && !arg_type->higher_equal(sig_type)) {
+              Node* retnode = pop();
               Node* cast_obj = _gvn.transform(new (C) CheckCastPPNode(control(), retnode, sig_type));
-              pop();
               push(cast_obj);
             }
           }
         } else {
-          assert(ct == rt, err_msg("unexpected mismatch rt=%d, ct=%d", rt, ct));
+          assert(rt == ct, err_msg_res("unexpected mismatch: rt=%s, ct=%s", type2name(rt), type2name(ct)));
           // push a zero; it's better than getting an oop/int mismatch
-          retnode = pop_node(rt);
-          retnode = zerocon(ct);
+          pop_node(rt);
+          Node* retnode = zerocon(ct);
           push_node(ct, retnode);
         }
         // Now that the value is well-behaved, continue with the call-site type.
         rtype = ctype;
       }
+    } else {
+      assert(rtype == ctype, "mismatched return types");  // symbolic resolution enforces this
     }
 
     // If the return type of the method is not loaded, assert that the
@@ -879,17 +871,39 @@
 #endif //PRODUCT
 
 
+ciMethod* Compile::optimize_virtual_call(ciMethod* caller, int bci, ciInstanceKlass* klass,
+                                         ciMethod* callee, const TypeOopPtr* receiver_type,
+                                         bool is_virtual,
+                                         bool& call_does_dispatch, int& vtable_index) {
+  // Set default values for out-parameters.
+  call_does_dispatch = true;
+  vtable_index       = Method::invalid_vtable_index;
+
+  // Choose call strategy.
+  ciMethod* optimized_virtual_method = optimize_inlining(caller, bci, klass, callee, receiver_type);
+
+  // Have the call been sufficiently improved such that it is no longer a virtual?
+  if (optimized_virtual_method != NULL) {
+    callee             = optimized_virtual_method;
+    call_does_dispatch = false;
+  } else if (!UseInlineCaches && is_virtual && callee->is_loaded()) {
+    // We can make a vtable call at this site
+    vtable_index = callee->resolve_vtable_index(caller->holder(), klass);
+  }
+  return callee;
+}
+
 // Identify possible target method and inlining style
-ciMethod* Parse::optimize_inlining(ciMethod* caller, int bci, ciInstanceKlass* klass,
-                                   ciMethod *dest_method, const TypeOopPtr* receiver_type) {
+ciMethod* Compile::optimize_inlining(ciMethod* caller, int bci, ciInstanceKlass* klass,
+                                     ciMethod* callee, const TypeOopPtr* receiver_type) {
   // only use for virtual or interface calls
 
   // If it is obviously final, do not bother to call find_monomorphic_target,
   // because the class hierarchy checks are not needed, and may fail due to
   // incompletely loaded classes.  Since we do our own class loading checks
   // in this module, we may confidently bind to any method.
-  if (dest_method->can_be_statically_bound()) {
-    return dest_method;
+  if (callee->can_be_statically_bound()) {
+    return callee;
   }
 
   // Attempt to improve the receiver
@@ -898,8 +912,8 @@
   if (receiver_type != NULL) {
     // Array methods are all inherited from Object, and are monomorphic.
     if (receiver_type->isa_aryptr() &&
-        dest_method->holder() == env()->Object_klass()) {
-      return dest_method;
+        callee->holder() == env()->Object_klass()) {
+      return callee;
     }
 
     // All other interesting cases are instance klasses.
@@ -919,7 +933,7 @@
   }
 
   ciInstanceKlass*   calling_klass = caller->holder();
-  ciMethod* cha_monomorphic_target = dest_method->find_monomorphic_target(calling_klass, klass, actual_receiver);
+  ciMethod* cha_monomorphic_target = callee->find_monomorphic_target(calling_klass, klass, actual_receiver);
   if (cha_monomorphic_target != NULL) {
     assert(!cha_monomorphic_target->is_abstract(), "");
     // Look at the method-receiver type.  Does it add "too much information"?
@@ -937,10 +951,10 @@
         cha_monomorphic_target->print();
         tty->cr();
       }
-      if (C->log() != NULL) {
-        C->log()->elem("missed_CHA_opportunity klass='%d' method='%d'",
-                       C->log()->identify(klass),
-                       C->log()->identify(cha_monomorphic_target));
+      if (log() != NULL) {
+        log()->elem("missed_CHA_opportunity klass='%d' method='%d'",
+                       log()->identify(klass),
+                       log()->identify(cha_monomorphic_target));
       }
       cha_monomorphic_target = NULL;
     }
@@ -952,7 +966,7 @@
     // by dynamic class loading.  Be sure to test the "static" receiver
     // dest_method here, as opposed to the actual receiver, which may
     // falsely lead us to believe that the receiver is final or private.
-    C->dependencies()->assert_unique_concrete_method(actual_receiver, cha_monomorphic_target);
+    dependencies()->assert_unique_concrete_method(actual_receiver, cha_monomorphic_target);
     return cha_monomorphic_target;
   }
 
@@ -961,7 +975,7 @@
   if (actual_receiver_is_exact) {
     // In case of evolution, there is a dependence on every inlined method, since each
     // such method can be changed when its class is redefined.
-    ciMethod* exact_method = dest_method->resolve_invoke(calling_klass, actual_receiver);
+    ciMethod* exact_method = callee->resolve_invoke(calling_klass, actual_receiver);
     if (exact_method != NULL) {
 #ifndef PRODUCT
       if (PrintOpto) {
--- a/src/share/vm/opto/graphKit.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/graphKit.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1794,10 +1794,15 @@
 
   if (ejvms == NULL) {
     // No exception edges to simply kill off those paths
-    C->gvn_replace_by(callprojs.catchall_catchproj, C->top());
-    C->gvn_replace_by(callprojs.catchall_memproj,   C->top());
-    C->gvn_replace_by(callprojs.catchall_ioproj,    C->top());
-
+    if (callprojs.catchall_catchproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_catchproj, C->top());
+    }
+    if (callprojs.catchall_memproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_memproj,   C->top());
+    }
+    if (callprojs.catchall_ioproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_ioproj,    C->top());
+    }
     // Replace the old exception object with top
     if (callprojs.exobj != NULL) {
       C->gvn_replace_by(callprojs.exobj, C->top());
@@ -1809,10 +1814,15 @@
     SafePointNode* ex_map = ekit.combine_and_pop_all_exception_states();
 
     Node* ex_oop = ekit.use_exception_state(ex_map);
-
-    C->gvn_replace_by(callprojs.catchall_catchproj, ekit.control());
-    C->gvn_replace_by(callprojs.catchall_memproj,   ekit.reset_memory());
-    C->gvn_replace_by(callprojs.catchall_ioproj,    ekit.i_o());
+    if (callprojs.catchall_catchproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_catchproj, ekit.control());
+    }
+    if (callprojs.catchall_memproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_memproj,   ekit.reset_memory());
+    }
+    if (callprojs.catchall_ioproj != NULL) {
+      C->gvn_replace_by(callprojs.catchall_ioproj,    ekit.i_o());
+    }
 
     // Replace the old exception object with the newly created one
     if (callprojs.exobj != NULL) {
--- a/src/share/vm/opto/memnode.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/memnode.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -2725,10 +2725,8 @@
     zend  = phase->transform( new(C) URShiftXNode(zend,  shift) );
   }
 
+  // Bulk clear double-words
   Node* zsize = phase->transform( new(C) SubXNode(zend, zbase) );
-  Node* zinit = phase->zerocon((unit == BytesPerLong) ? T_LONG : T_INT);
-
-  // Bulk clear double-words
   Node* adr = phase->transform( new(C) AddPNode(dest, dest, start_offset) );
   mem = new (C) ClearArrayNode(ctl, mem, zsize, adr);
   return phase->transform(mem);
--- a/src/share/vm/opto/parse.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/parse.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -70,7 +70,7 @@
   InlineTree *build_inline_tree_for_callee(ciMethod* callee_method,
                                            JVMState* caller_jvms,
                                            int caller_bci);
-  const char* try_to_inline(ciMethod* callee_method, ciMethod* caller_method, int caller_bci, ciCallProfile& profile, WarmCallInfo* wci_result);
+  const char* try_to_inline(ciMethod* callee_method, ciMethod* caller_method, int caller_bci, ciCallProfile& profile, WarmCallInfo* wci_result, bool& should_delay);
   const char* should_inline(ciMethod* callee_method, ciMethod* caller_method, int caller_bci, ciCallProfile& profile, WarmCallInfo* wci_result) const;
   const char* should_not_inline(ciMethod* callee_method, ciMethod* caller_method, WarmCallInfo* wci_result) const;
   void        print_inlining(ciMethod *callee_method, int caller_bci, const char *failure_msg) const;
@@ -107,7 +107,7 @@
   // and may be accessed by find_subtree_from_root.
   // The call_method is the dest_method for a special or static invocation.
   // The call_method is an optimized virtual method candidate otherwise.
-  WarmCallInfo* ok_to_inline(ciMethod *call_method, JVMState* caller_jvms, ciCallProfile& profile, WarmCallInfo* wci);
+  WarmCallInfo* ok_to_inline(ciMethod *call_method, JVMState* caller_jvms, ciCallProfile& profile, WarmCallInfo* wci, bool& should_delay);
 
   // Information about inlined method
   JVMState*   caller_jvms()       const { return _caller_jvms; }
@@ -469,10 +469,6 @@
   // Helper function to uncommon-trap or bailout for non-compilable call-sites
   bool can_not_compile_call_site(ciMethod *dest_method, ciInstanceKlass *klass);
 
-  // Helper function to identify inlining potential at call-site
-  ciMethod* optimize_inlining(ciMethod* caller, int bci, ciInstanceKlass* klass,
-                              ciMethod *dest_method, const TypeOopPtr* receiver_type);
-
   // Helper function to setup for type-profile based inlining
   bool prepare_type_profile_inline(ciInstanceKlass* prof_klass, ciMethod* prof_method);
 
--- a/src/share/vm/opto/parse1.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/parse1.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1404,7 +1404,8 @@
 
     do_one_bytecode();
 
-    assert(!have_se || stopped() || failing() || (sp() - pre_bc_sp) == depth, "correct depth prediction");
+    assert(!have_se || stopped() || failing() || (sp() - pre_bc_sp) == depth,
+           err_msg_res("incorrect depth prediction: sp=%d, pre_bc_sp=%d, depth=%d", sp(), pre_bc_sp, depth));
 
     do_exceptions();
 
--- a/src/share/vm/opto/phaseX.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/phaseX.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -75,6 +75,13 @@
   // nh->_sentinel must be in the current node space
 }
 
+void NodeHash::replace_with(NodeHash *nh) {
+  debug_only(_table = (Node**)badAddress);   // interact correctly w/ operator=
+  // just copy in all the fields
+  *this = *nh;
+  // nh->_sentinel must be in the current node space
+}
+
 //------------------------------hash_find--------------------------------------
 // Find in hash table
 Node *NodeHash::hash_find( const Node *n ) {
--- a/src/share/vm/opto/phaseX.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/phaseX.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -92,6 +92,7 @@
   }
 
   void   remove_useless_nodes(VectorSet &useful); // replace with sentinel
+  void replace_with(NodeHash* nh);
 
   Node  *sentinel() { return _sentinel; }
 
@@ -386,6 +387,11 @@
   Node  *transform( Node *n );
   Node  *transform_no_reclaim( Node *n );
 
+  void replace_with(PhaseGVN* gvn) {
+    _table.replace_with(&gvn->_table);
+    _types = gvn->_types;
+  }
+
   // Check for a simple dead loop when a data node references itself.
   DEBUG_ONLY(void dead_loop_check(Node *n);)
 };
--- a/src/share/vm/opto/stringopts.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/opto/stringopts.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -265,7 +265,8 @@
     } else if (n->is_IfTrue()) {
       Compile* C = _stringopts->C;
       C->gvn_replace_by(n, n->in(0)->in(0));
-      C->gvn_replace_by(n->in(0), C->top());
+      // get rid of the other projection
+      C->gvn_replace_by(n->in(0)->as_If()->proj_out(false), C->top());
     }
   }
 }
@@ -439,7 +440,7 @@
       }
       // Find the constructor call
       Node* result = alloc->result_cast();
-      if (result == NULL || !result->is_CheckCastPP()) {
+      if (result == NULL || !result->is_CheckCastPP() || alloc->in(TypeFunc::Memory)->is_top()) {
         // strange looking allocation
 #ifndef PRODUCT
         if (PrintOptimizeStringConcat) {
@@ -834,6 +835,9 @@
           ptr->in(1)->in(0) != NULL && ptr->in(1)->in(0)->is_If()) {
         // Simple diamond.
         // XXX should check for possibly merging stores.  simple data merges are ok.
+        // The IGVN will make this simple diamond go away when it
+        // transforms the Region. Make sure it sees it.
+        Compile::current()->record_for_igvn(ptr);
         ptr = ptr->in(1)->in(0)->in(0);
         continue;
       }
--- a/src/share/vm/prims/jvm.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/prims/jvm.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1515,7 +1515,7 @@
 JVM_END
 
 
-static Method* jvm_get_method_common(jobject method, TRAPS) {
+static Method* jvm_get_method_common(jobject method) {
   // some of this code was adapted from from jni_FromReflectedMethod
 
   oop reflected = JNIHandles::resolve_non_null(method);
@@ -1533,8 +1533,7 @@
   }
   Klass* k = java_lang_Class::as_Klass(mirror);
 
-  KlassHandle kh(THREAD, k);
-  Method* m = InstanceKlass::cast(kh())->method_with_idnum(slot);
+  Method* m = InstanceKlass::cast(k)->method_with_idnum(slot);
   if (m == NULL) {
     assert(false, "cannot find method");
     return NULL;  // robustness
@@ -1548,7 +1547,7 @@
   JVMWrapper("JVM_GetMethodAnnotations");
 
   // method is a handle to a java.lang.reflect.Method object
-  Method* m = jvm_get_method_common(method, CHECK_NULL);
+  Method* m = jvm_get_method_common(method);
   return (jbyteArray) JNIHandles::make_local(env,
     Annotations::make_java_array(m->annotations(), THREAD));
 JVM_END
@@ -1558,7 +1557,7 @@
   JVMWrapper("JVM_GetMethodDefaultAnnotationValue");
 
   // method is a handle to a java.lang.reflect.Method object
-  Method* m = jvm_get_method_common(method, CHECK_NULL);
+  Method* m = jvm_get_method_common(method);
   return (jbyteArray) JNIHandles::make_local(env,
     Annotations::make_java_array(m->annotation_default(), THREAD));
 JVM_END
@@ -1568,11 +1567,54 @@
   JVMWrapper("JVM_GetMethodParameterAnnotations");
 
   // method is a handle to a java.lang.reflect.Method object
-  Method* m = jvm_get_method_common(method, CHECK_NULL);
+  Method* m = jvm_get_method_common(method);
   return (jbyteArray) JNIHandles::make_local(env,
     Annotations::make_java_array(m->parameter_annotations(), THREAD));
 JVM_END
 
+/* Type use annotations support (JDK 1.8) */
+
+JVM_ENTRY(jbyteArray, JVM_GetClassTypeAnnotations(JNIEnv *env, jclass cls))
+  assert (cls != NULL, "illegal class");
+  JVMWrapper("JVM_GetClassTypeAnnotations");
+  ResourceMark rm(THREAD);
+  // Return null for arrays and primitives
+  if (!java_lang_Class::is_primitive(JNIHandles::resolve(cls))) {
+    Klass* k = java_lang_Class::as_Klass(JNIHandles::resolve(cls));
+    if (k->oop_is_instance()) {
+      typeArrayOop a = Annotations::make_java_array(InstanceKlass::cast(k)->type_annotations()->class_annotations(), CHECK_NULL);
+      return (jbyteArray) JNIHandles::make_local(env, a);
+    }
+  }
+  return NULL;
+JVM_END
+
+JVM_ENTRY(jobjectArray, JVM_GetMethodParameters(JNIEnv *env, jobject method))
+{
+  JVMWrapper("JVM_GetMethodParameters");
+  // method is a handle to a java.lang.reflect.Method object
+  Method* method_ptr = jvm_get_method_common(method);
+  methodHandle mh (THREAD, method_ptr);
+  Handle reflected_method (THREAD, JNIHandles::resolve_non_null(method));
+  const int num_params = mh->method_parameters_length();
+
+  if(0 != num_params) {
+    objArrayOop result_oop = oopFactory::new_objArray(SystemDictionary::reflect_Parameter_klass(), num_params, CHECK_NULL);
+    objArrayHandle result (THREAD, result_oop);
+
+    for(int i = 0; i < num_params; i++) {
+      MethodParametersElement* params = mh->method_parameters_start();
+      Symbol* const sym = mh->constants()->symbol_at(params[i].name_cp_index);
+      oop param = Reflection::new_parameter(reflected_method, i, sym,
+                                            params[i].flags, CHECK_NULL);
+      result->obj_at_put(i, param);
+    }
+    return (jobjectArray)JNIHandles::make_local(env, result());
+  } else {
+    return (jobjectArray)NULL;
+  }
+}
+JVM_END
 
 // New (JDK 1.4) reflection implementation /////////////////////////////////////
 
--- a/src/share/vm/prims/jvm.h	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/prims/jvm.h	Fri Jan 11 01:43:09 2013 -0800
@@ -86,6 +86,8 @@
 
 #define JVM_INTERFACE_VERSION 4
 
+JNIEXPORT jobjectArray JNICALL
+JVM_GetMethodParameters(JNIEnv *env, jobject method);
 
 JNIEXPORT jint JNICALL
 JVM_GetInterfaceVersion(void);
@@ -519,6 +521,10 @@
 JNIEXPORT jbyteArray JNICALL
 JVM_GetMethodParameterAnnotations(JNIEnv *env, jobject method);
 
+/* Type use annotations support (JDK 1.8) */
+
+JNIEXPORT jbyteArray JNICALL
+JVM_GetClassTypeAnnotations(JNIEnv *env, jclass cls);
 
 /*
  * New (JDK 1.4) reflection implementation
--- a/src/share/vm/prims/jvmtiExport.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/prims/jvmtiExport.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1305,15 +1305,21 @@
         vframeStream st(thread);
         assert(!st.at_end(), "cannot be at end");
         Method* current_method = NULL;
+        // A GC may occur during the Method::fast_exception_handler_bci_for()
+        // call below if it needs to load the constraint class. Using a
+        // methodHandle to keep the 'current_method' from being deallocated
+        // if GC happens.
+        methodHandle current_mh = methodHandle(thread, current_method);
         int current_bci = -1;
         do {
           current_method = st.method();
+          current_mh = methodHandle(thread, current_method);
           current_bci = st.bci();
           do {
             should_repeat = false;
             KlassHandle eh_klass(thread, exception_handle()->klass());
-            current_bci = current_method->fast_exception_handler_bci_for(
-              eh_klass, current_bci, THREAD);
+            current_bci = Method::fast_exception_handler_bci_for(
+              current_mh, eh_klass, current_bci, THREAD);
             if (HAS_PENDING_EXCEPTION) {
               exception_handle = Handle(thread, PENDING_EXCEPTION);
               CLEAR_PENDING_EXCEPTION;
@@ -1328,8 +1334,7 @@
           catch_jmethodID = 0;
           current_bci = 0;
         } else {
-          catch_jmethodID = jem.to_jmethodID(
-                                     methodHandle(thread, current_method));
+          catch_jmethodID = jem.to_jmethodID(current_mh);
         }
 
         JvmtiJavaThreadEventTransition jet(thread);
--- a/src/share/vm/prims/jvmtiRedefineClasses.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/prims/jvmtiRedefineClasses.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1043,7 +1043,7 @@
 
     Rewriter::rewrite(scratch_class, THREAD);
     if (!HAS_PENDING_EXCEPTION) {
-      Rewriter::relocate_and_link(scratch_class, THREAD);
+      scratch_class->link_methods(THREAD);
     }
     if (HAS_PENDING_EXCEPTION) {
       Symbol* ex_name = PENDING_EXCEPTION->klass()->name();
@@ -3338,7 +3338,20 @@
     the_class->set_access_flags(flags);
   }
 
-  // Replace annotation fields value
+  // Since there is currently no rewriting of type annotations indexes
+  // into the CP, we null out type annotations on scratch_class before
+  // we swap annotations with the_class rather than facing the
+  // possibility of shipping annotations with broken indexes to
+  // Java-land.
+  Annotations* new_annotations = scratch_class->annotations();
+  if (new_annotations != NULL) {
+    Annotations* new_type_annotations = new_annotations->type_annotations();
+    if (new_type_annotations != NULL) {
+      MetadataFactory::free_metadata(scratch_class->class_loader_data(), new_type_annotations);
+      new_annotations->set_type_annotations(NULL);
+    }
+  }
+  // Swap annotation fields values
   Annotations* old_annotations = the_class->annotations();
   the_class->set_annotations(scratch_class->annotations());
   scratch_class->set_annotations(old_annotations);
--- a/src/share/vm/runtime/arguments.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/runtime/arguments.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1083,10 +1083,6 @@
   }
 }
 
-// If the user has chosen ParallelGCThreads > 0, we set UseParNewGC
-// if it's not explictly set or unset. If the user has chosen
-// UseParNewGC and not explicitly set ParallelGCThreads we
-// set it, unless this is a single cpu machine.
 void Arguments::set_parnew_gc_flags() {
   assert(!UseSerialGC && !UseParallelOldGC && !UseParallelGC && !UseG1GC,
          "control point invariant");
@@ -1095,42 +1091,41 @@
   // Turn off AdaptiveSizePolicy for parnew until it is complete.
   disable_adaptive_size_policy("UseParNewGC");
 
-  if (ParallelGCThreads == 0) {
-    FLAG_SET_DEFAULT(ParallelGCThreads,
-                     Abstract_VM_Version::parallel_worker_threads());
-    if (ParallelGCThreads == 1) {
-      FLAG_SET_DEFAULT(UseParNewGC, false);
-      FLAG_SET_DEFAULT(ParallelGCThreads, 0);
+  if (FLAG_IS_DEFAULT(ParallelGCThreads)) {
+    FLAG_SET_DEFAULT(ParallelGCThreads, Abstract_VM_Version::parallel_worker_threads());
+    assert(ParallelGCThreads > 0, "We should always have at least one thread by default");
+  } else if (ParallelGCThreads == 0) {
+    jio_fprintf(defaultStream::error_stream(),
+        "The ParNew GC can not be combined with -XX:ParallelGCThreads=0\n");
+    vm_exit(1);
+  }
+
+  // By default YoungPLABSize and OldPLABSize are set to 4096 and 1024 respectively,
+  // these settings are default for Parallel Scavenger. For ParNew+Tenured configuration
+  // we set them to 1024 and 1024.
+  // See CR 6362902.
+  if (FLAG_IS_DEFAULT(YoungPLABSize)) {
+    FLAG_SET_DEFAULT(YoungPLABSize, (intx)1024);
+  }
+  if (FLAG_IS_DEFAULT(OldPLABSize)) {
+    FLAG_SET_DEFAULT(OldPLABSize, (intx)1024);
+  }
+
+  // AlwaysTenure flag should make ParNew promote all at first collection.
+  // See CR 6362902.
+  if (AlwaysTenure) {
+    FLAG_SET_CMDLINE(uintx, MaxTenuringThreshold, 0);
+  }
+  // When using compressed oops, we use local overflow stacks,
+  // rather than using a global overflow list chained through
+  // the klass word of the object's pre-image.
+  if (UseCompressedOops && !ParGCUseLocalOverflow) {
+    if (!FLAG_IS_DEFAULT(ParGCUseLocalOverflow)) {
+      warning("Forcing +ParGCUseLocalOverflow: needed if using compressed references");
     }
+    FLAG_SET_DEFAULT(ParGCUseLocalOverflow, true);
   }
-  if (UseParNewGC) {
-    // By default YoungPLABSize and OldPLABSize are set to 4096 and 1024 respectively,
-    // these settings are default for Parallel Scavenger. For ParNew+Tenured configuration
-    // we set them to 1024 and 1024.
-    // See CR 6362902.
-    if (FLAG_IS_DEFAULT(YoungPLABSize)) {
-      FLAG_SET_DEFAULT(YoungPLABSize, (intx)1024);
-    }
-    if (FLAG_IS_DEFAULT(OldPLABSize)) {
-      FLAG_SET_DEFAULT(OldPLABSize, (intx)1024);
-    }
-
-    // AlwaysTenure flag should make ParNew promote all at first collection.
-    // See CR 6362902.
-    if (AlwaysTenure) {
-      FLAG_SET_CMDLINE(uintx, MaxTenuringThreshold, 0);
-    }
-    // When using compressed oops, we use local overflow stacks,
-    // rather than using a global overflow list chained through
-    // the klass word of the object's pre-image.
-    if (UseCompressedOops && !ParGCUseLocalOverflow) {
-      if (!FLAG_IS_DEFAULT(ParGCUseLocalOverflow)) {
-        warning("Forcing +ParGCUseLocalOverflow: needed if using compressed references");
-      }
-      FLAG_SET_DEFAULT(ParGCUseLocalOverflow, true);
-    }
-    assert(ParGCUseLocalOverflow || !UseCompressedOops, "Error");
-  }
+  assert(ParGCUseLocalOverflow || !UseCompressedOops, "Error");
 }
 
 // Adjust some sizes to suit CMS and/or ParNew needs; these work well on
@@ -1331,14 +1326,14 @@
   // then a saved space from compressed oops.
   if ((int)ObjectAlignmentInBytes > 256) {
     jio_fprintf(defaultStream::error_stream(),
-                "error: ObjectAlignmentInBytes=%d must not be greater then 256\n",
+                "error: ObjectAlignmentInBytes=%d must not be greater than 256\n",
                 (int)ObjectAlignmentInBytes);
     return false;
   }
   // In case page size is very small.
   if ((int)ObjectAlignmentInBytes >= os::vm_page_size()) {
     jio_fprintf(defaultStream::error_stream(),
-                "error: ObjectAlignmentInBytes=%d must be less then page size %d\n",
+                "error: ObjectAlignmentInBytes=%d must be less than page size %d\n",
                 (int)ObjectAlignmentInBytes, os::vm_page_size());
     return false;
   }
@@ -1459,30 +1454,34 @@
 
   // If no heap maximum was requested explicitly, use some reasonable fraction
   // of the physical memory, up to a maximum of 1GB.
-  if (UseParallelGC) {
-    FLAG_SET_DEFAULT(ParallelGCThreads,
-                     Abstract_VM_Version::parallel_worker_threads());
-
-    // If InitialSurvivorRatio or MinSurvivorRatio were not specified, but the
-    // SurvivorRatio has been set, reset their default values to SurvivorRatio +
-    // 2.  By doing this we make SurvivorRatio also work for Parallel Scavenger.
-    // See CR 6362902 for details.
-    if (!FLAG_IS_DEFAULT(SurvivorRatio)) {
-      if (FLAG_IS_DEFAULT(InitialSurvivorRatio)) {
-         FLAG_SET_DEFAULT(InitialSurvivorRatio, SurvivorRatio + 2);
-      }
-      if (FLAG_IS_DEFAULT(MinSurvivorRatio)) {
-        FLAG_SET_DEFAULT(MinSurvivorRatio, SurvivorRatio + 2);
-      }
+  FLAG_SET_DEFAULT(ParallelGCThreads,
+                   Abstract_VM_Version::parallel_worker_threads());
+  if (ParallelGCThreads == 0) {
+    jio_fprintf(defaultStream::error_stream(),
+        "The Parallel GC can not be combined with -XX:ParallelGCThreads=0\n");
+    vm_exit(1);
+  }
+
+
+  // If InitialSurvivorRatio or MinSurvivorRatio were not specified, but the
+  // SurvivorRatio has been set, reset their default values to SurvivorRatio +
+  // 2.  By doing this we make SurvivorRatio also work for Parallel Scavenger.
+  // See CR 6362902 for details.
+  if (!FLAG_IS_DEFAULT(SurvivorRatio)) {
+    if (FLAG_IS_DEFAULT(InitialSurvivorRatio)) {
+       FLAG_SET_DEFAULT(InitialSurvivorRatio, SurvivorRatio + 2);
     }
-
-    if (UseParallelOldGC) {
-      // Par compact uses lower default values since they are treated as
-      // minimums.  These are different defaults because of the different
-      // interpretation and are not ergonomically set.
-      if (FLAG_IS_DEFAULT(MarkSweepDeadRatio)) {
-        FLAG_SET_DEFAULT(MarkSweepDeadRatio, 1);
-      }
+    if (FLAG_IS_DEFAULT(MinSurvivorRatio)) {
+      FLAG_SET_DEFAULT(MinSurvivorRatio, SurvivorRatio + 2);
+    }
+  }
+
+  if (UseParallelOldGC) {
+    // Par compact uses lower default values since they are treated as
+    // minimums.  These are different defaults because of the different
+    // interpretation and are not ergonomically set.
+    if (FLAG_IS_DEFAULT(MarkSweepDeadRatio)) {
+      FLAG_SET_DEFAULT(MarkSweepDeadRatio, 1);
     }
   }
 }
@@ -1783,6 +1782,24 @@
   return status;
 }
 
+void Arguments::check_deprecated_gcs() {
+  if (UseConcMarkSweepGC && !UseParNewGC) {
+    warning("Using the DefNew young collector with the CMS collector is deprecated "
+        "and will likely be removed in a future release");
+  }
+
+  if (UseParNewGC && !UseConcMarkSweepGC) {
+    // !UseConcMarkSweepGC means that we are using serial old gc. Unfortunately we don't
+    // set up UseSerialGC properly, so that can't be used in the check here.
+    warning("Using the ParNew young collector with the Serial old collector is deprecated "
+        "and will likely be removed in a future release");
+  }
+
+  if (CMSIncrementalMode) {
+    warning("Using incremental CMS is deprecated and will likely be removed in a future release");
+  }
+}
+
 // Check stack pages settings
 bool Arguments::check_stack_pages()
 {
@@ -2997,11 +3014,6 @@
     FLAG_SET_DEFAULT(UseLargePages, false);
   }
 
-  // Add 2M to any size for SharedReadOnlySize to get around the JPRT setting
-  if (DumpSharedSpaces && !FLAG_IS_DEFAULT(SharedReadOnlySize)) {
-    SharedReadOnlySize = 14*M;
-  }
-
   if (DumpSharedSpaces) {
     if (RequireSharedSpaces) {
       warning("cannot dump shared archive while using shared archive");
@@ -3038,7 +3050,6 @@
   strcpy(shared_archive_path, jvm_path);
   strcat(shared_archive_path, os::file_separator());
   strcat(shared_archive_path, "classes");
-  DEBUG_ONLY(strcat(shared_archive_path, "_g");)
   strcat(shared_archive_path, ".jsa");
   SharedArchivePath = shared_archive_path;
 
@@ -3247,6 +3258,7 @@
   } else if (UseG1GC) {
     set_g1_gc_flags();
   }
+  check_deprecated_gcs();
 #endif // INCLUDE_ALTERNATE_GCS
 
 #ifdef SERIALGC
@@ -3291,6 +3303,18 @@
   if (!EliminateLocks) {
     EliminateNestedLocks = false;
   }
+  if (!Inline) {
+    IncrementalInline = false;
+  }
+#ifndef PRODUCT
+  if (!IncrementalInline) {
+    AlwaysIncrementalInline = false;
+  }
+#endif
+  if (IncrementalInline && FLAG_IS_DEFAULT(MaxNodeLimit)) {
+    // incremental inlining: bump MaxNodeLimit
+    FLAG_SET_DEFAULT(MaxNodeLimit, (intx)75000);
+  }
 #endif
 
   if (PrintAssembly && FLAG_IS_DEFAULT(DebugNonSafepoints)) {
--- a/src/share/vm/runtime/arguments.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/runtime/arguments.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -413,6 +413,7 @@
   static jint adjust_after_os();
   // Check for consistency in the selection of the garbage collector.
   static bool check_gc_consistency();
+  static void check_deprecated_gcs();
   // Check consistecy or otherwise of VM argument settings
   static bool check_vm_args_consistency();
   // Check stack pages settings
--- a/src/share/vm/runtime/fieldDescriptor.cpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/runtime/fieldDescriptor.cpp	Fri Jan 11 01:43:09 2013 -0800
@@ -65,6 +65,17 @@
   return md->at(index());
 }
 
+AnnotationArray* fieldDescriptor::type_annotations() const {
+  InstanceKlass* ik = field_holder();
+  Annotations* type_annos = ik->type_annotations();
+  if (type_annos == NULL)
+    return NULL;
+  Array<AnnotationArray*>* md = type_annos->fields_annotations();
+  if (md == NULL)
+    return NULL;
+  return md->at(index());
+}
+
 constantTag fieldDescriptor::initial_value_tag() const {
   return constants()->tag_at(initial_value_index());
 }
--- a/src/share/vm/runtime/fieldDescriptor.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/runtime/fieldDescriptor.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -68,6 +68,7 @@
   Symbol* generic_signature()     const;
   int index()                     const    { return _index; }
   AnnotationArray* annotations()  const;
+  AnnotationArray* type_annotations()  const;
 
   // Initial field value
   bool has_initial_value()        const    { return field()->initval_index() != 0; }
--- a/src/share/vm/runtime/globals.hpp	Thu Jan 10 09:55:30 2013 -0800
+++ b/src/share/vm/runtime/globals.hpp	Fri Jan 11 01:43:09 2013 -0800
@@ -1101,13 +1101,6 @@
   product(bool, ReduceSignalUsage, false,                                   \
           "Reduce the use of OS signals in Java and/or the VM")             \
                                                                             \
-  notproduct(bool, ValidateMarkSweep, false,                                \
-          "Do extra validation during MarkSweep collection")                \
-                                                                            \
-  notproduct(bool, RecordMarkSweepCompaction, false,                        \
-          "Enable GC-to-GC recording and querying of compaction during "    \
-          "MarkSweep")                                                      \
-                                                                            \
   develop_pd(bool, ShareVtableStubs,                                        \
           "Share vtable stubs (smaller code but worse branch prediction")   \
                                                                             \
@@ -1606,7 +1599,7 @@
   develop(bool, CMSTraceThreadState, false,                                 \
           "Trace the CMS thread state (enable the trace_state() method)")   \
                                                                             \
-  product(bool, CMSClassUnloadingEnabled, false,                            \
+  product(bool, CMSClassUnloadingEnabled, true,                             \
           "Whether class unloading enabled when using CMS GC")              \
                                                                             \
   product(uintx, CMSClassUnloadingMaxInterval, 0,                           \
@@ -1830,7 +1823,7 @@
                                                                             \
   product(intx, CMSIsTooFullPercentage, 98,                                 \
           "An absolute ceiling above which CMS will always consider the "   \
-          "perm gen ripe for collection")                                   \
+          "unloading of classes when class unloading is enabled")           \
                                                                             \
   develop(bool, CMSTestInFreeList, false,                                   \
           "Check if the coalesced range is already in the "                 \
@@ -1899,13 +1892,13 @@
           "Metadata deallocation alot interval")                            \
                                                                             \
   develop(bool, TraceMetadataChunkAllocation, false,                        \
-          "Trace humongous metadata allocations")                           \
+          "Trace chunk metadata allocations")                               \
                                                                             \