Merge

.hgtags

@@ -172,3 +172,8 @@ b820143a6f1ce993c6e6f31db4d64de990f42654 jdk8-b47
 086271e35b0a419b38e8bda9bebd70693811df0a jdk8-b48
 cecd7026f30cbd83b0601925a7a5e059aec98138 jdk8-b49
 38fe5ab028908cf64dd73a43336ba3211577bfc3 jdk8-b50
+382651d28f2502d371eca751962232c0e535e57a jdk8-b51
+b67041a6cb508da18d2f5c7687e6a31e08bea4fc jdk8-b52
+c7aa5cca1c01689a7b1a92411daf83684af05a33 jdk8-b53
+7c6aa31ff1b2ae48c1c686ebe1aadf0c3da5be15 jdk8-b54
+319f583f66db47395fa86127dd3ddb729eb7c64f jdk8-b55

.hgtags-top-repo

@@ -172,3 +172,8 @@ e4f81a817447c3a4f6868f083c81c2fb1b15d44c jdk8-b44
 3f6c72d1c2a6e5c9e7d81c3dc984886678a128ad jdk8-b48
 c97b99424815c43818e3cc3ffcdd1a60f3198b52 jdk8-b49
 2fd67618b9a3c847780ed7b9d228e862b6e2824c jdk8-b50
+57c0aee7309050b9d6cfcbd202dc704e9260b377 jdk8-b51
+8d24def5ceb3b8f2e857f2e18b2804fc59eecf8d jdk8-b52
+febd7ff5280067ca482faaeb9418ae88764c1a35 jdk8-b53
+c1a277c6022affbc6855bdfb039511e73fbe2395 jdk8-b54
+b85b44cced2406792cfb9baab1377ff03e7001d8 jdk8-b55

corba/.hgtags

@@ -172,3 +172,8 @@ cd879aff5d3cc1f58829aab3116880aa19525b78 jdk8-b43
 7e2b179a5b4dbd3f097e28daa00abfcc72ba3e0b jdk8-b48
 fe44e58a6bdbeae350ce96aafb49770a5dca5d8a jdk8-b49
 d20d9eb9f093adbf392918c703960ad24c93a331 jdk8-b50
+9b0f841ca9f7ee9bacf16a5ab41c4f829276bc6b jdk8-b51
+80689ff9cb499837513f18a1136dac7f0686cd55 jdk8-b52
+63aeb7a2472fb299134ad7388e0a111a5340b02d jdk8-b53
+16c82fc74695bab9b9e0fb05c086a5a08ba0082f jdk8-b54
+e8a0e84383d6fbd303ce44bd355fb25972b13286 jdk8-b55

corba/src/share/classes/com/sun/corba/se/impl/transport/CorbaResponseWaitingRoomImpl.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001, 2004, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2001, 2012, 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
@@ -25,7 +25,10 @@
 
 package com.sun.corba.se.impl.transport;
 
-import java.util.Hashtable;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.Map;
 
 import org.omg.CORBA.CompletionStatus;
 import org.omg.CORBA.SystemException;
@@ -68,7 +71,7 @@ public class CorbaResponseWaitingRoomImpl
 
     private CorbaConnection connection;
     // Maps requestId to an OutCallDesc.
-    private Hashtable out_calls = null; // REVISIT - use int hastable/map
+    final private Map<Integer, OutCallDesc> out_calls;
 
     public CorbaResponseWaitingRoomImpl(ORB orb, CorbaConnection connection)
     {
@@ -76,7 +79,8 @@ public class CorbaResponseWaitingRoomImpl
         wrapper = ORBUtilSystemException.get( orb,
             CORBALogDomains.RPC_TRANSPORT ) ;
         this.connection = connection;
-        out_calls = new Hashtable();
+        out_calls =
+            Collections.synchronizedMap(new HashMap<Integer, OutCallDesc>());
     }
 
     ////////////////////////////////////////////////////
@@ -139,7 +143,7 @@ public class CorbaResponseWaitingRoomImpl
             return null;
         }
 
-        OutCallDesc call = (OutCallDesc)out_calls.get(requestId);
+        OutCallDesc call = out_calls.get(requestId);
         if (call == null) {
             throw wrapper.nullOutCall(CompletionStatus.COMPLETED_MAYBE);
         }
@@ -197,7 +201,7 @@ public class CorbaResponseWaitingRoomImpl
         LocateReplyOrReplyMessage header = (LocateReplyOrReplyMessage)
             inputObject.getMessageHeader();
         Integer requestId = new Integer(header.getRequestId());
-        OutCallDesc call = (OutCallDesc) out_calls.get(requestId);
+        OutCallDesc call = out_calls.get(requestId);
 
         if (orb.transportDebugFlag) {
             dprint(".responseReceived: id/"
@@ -248,7 +252,6 @@ public class CorbaResponseWaitingRoomImpl
 
     public int numberRegistered()
     {
-        // Note: Hashtable.size() is not synchronized
         return out_calls.size();
     }
 
@@ -264,12 +267,19 @@ public class CorbaResponseWaitingRoomImpl
             dprint(".signalExceptionToAllWaiters: " + systemException);
         }
 
-        OutCallDesc call;
+        synchronized (out_calls) {
-        java.util.Enumeration e = out_calls.elements();
+            if (orb.transportDebugFlag) {
-        while(e.hasMoreElements()) {
+                dprint(".signalExceptionToAllWaiters: out_calls size :" +
-            call = (OutCallDesc) e.nextElement();
+                       out_calls.size());
+            }
 
+            for (OutCallDesc call : out_calls.values()) {
+                if (orb.transportDebugFlag) {
+                    dprint(".signalExceptionToAllWaiters: signaling " +
+                            call);
+                }
                 synchronized(call.done) {
+                    try {
                         // anything waiting for BufferManagerRead's fragment queue
                         // needs to be cancelled
                         CorbaMessageMediator corbaMsgMediator =
@@ -284,9 +294,14 @@ public class CorbaResponseWaitingRoomImpl
                             int requestId = corbaMsgMediator.getRequestId();
                             bufferManager.cancelProcessing(requestId);
                         }
+                    } catch (Exception e) {
+                    } finally {
+                        // attempt to wake up waiting threads in all cases
                         call.inputObject = null;
                         call.exception = systemException;
-                call.done.notify();
+                        call.done.notifyAll();
+                    }
+                }
             }
         }
     }
@@ -294,7 +309,7 @@ public class CorbaResponseWaitingRoomImpl
     public MessageMediator getMessageMediator(int requestId)
     {
         Integer id = new Integer(requestId);
-        OutCallDesc call = (OutCallDesc) out_calls.get(id);
+        OutCallDesc call = out_calls.get(id);
         if (call == null) {
             // This can happen when getting early reply fragments for a
             // request which has completed (e.g., client marshaling error).

corba/src/share/classes/com/sun/corba/se/impl/transport/SocketOrChannelConnectionImpl.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2001, 2012, 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
@@ -1521,7 +1521,7 @@ public class SocketOrChannelConnectionImpl
             // connection and give them the SystemException;
 
             responseWaitingRoom.signalExceptionToAllWaiters(systemException);
-
+        } finally {
             if (contactInfo != null) {
                 ((OutboundConnectionCache)getConnectionCache()).remove(contactInfo);
             } else if (acceptor != null) {
@@ -1542,7 +1542,6 @@ public class SocketOrChannelConnectionImpl
 
             writeUnlock();
 
-        } finally {
             if (orb.transportDebugFlag) {
                 dprint(".purgeCalls<-: "
                        + minor_code + "/" + die + "/" + lockHeld

hotspot/.hgtags

@@ -268,3 +268,10 @@ e3619706a7253540a2d94e9e841acaab8ace7038 jdk8-b49
 58f237a9e83af6ded0d2e2c81d252cd47c0f4c45 jdk8-b50
 3b3ad16429701b2eb6712851c2f7c5a726eb2cbe hs24-b19
 663fc23da8d51c4c0552cbcb17ffc85f5869d4fd jdk8-b51
+4c8f2a12e757e7a808aa85827573e09f75d7459f hs24-b20
+6d0436885201db3f581523344a734793bb989549 jdk8-b52
+54240c1b8e87758f28da2c6a569a926fd9e0910a jdk8-b53
+9e3ae661284dc04185b029d85440fe7811f1ed07 hs24-b21
+e8fb566b94667f88462164defa654203f0ab6820 jdk8-b54
+09ea7e0752b306b8ae74713aeb4eb6263e1c6836 hs24-b22
+af0c8a0808516317333dcf9af15567cdd52761ce jdk8-b55

hotspot/agent/make/saenv.sh

@@ -26,7 +26,7 @@
 # This file sets common environment variables for all SA scripts
 
 OS=`uname`
-STARTDIR=`dirname $0`
+STARTDIR=`(cd \`dirname $0 \`; pwd)`
 ARCH=`uname -m`
 
 if [ "x$SA_JAVA" = "x" ]; then

hotspot/agent/make/start-debug-server-proc.sh

@@ -25,10 +25,11 @@
 
 . `dirname $0`/saenv.sh
 
-if [ -f $STARTDIR/sa.jar ] ; then
+if [ -f $STARTDIR/../lib/sa-jdi.jar ] ; then
-  CP=$STARTDIR/sa.jar
+  CP=$STARTDIR/../lib/sa-jdi.jar
 else
   CP=$STARTDIR/../build/classes
 fi
 
-$SA_JAVA -classpath $CP ${OPTIONS} -Djava.rmi.server.codebase=file:/$CP -Djava.security.policy=$STARTDIR\/grantAll.policy sun.jvm.hotspot.DebugServer $*
+$STARTDIR/java -classpath $CP ${OPTIONS} -Djava.rmi.server.codebase=file://$CP -Djava.security.policy=${STARTDIR}/grantAll.policy sun.jvm.hotspot.DebugServer $*
+

hotspot/agent/src/os/linux/LinuxDebuggerLocal.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2007, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2012, 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
@@ -55,11 +55,11 @@ static jmethodID listAdd_ID = 0;
 #define THROW_NEW_DEBUGGER_EXCEPTION_(str, value) { throw_new_debugger_exception(env, str); return value; }
 #define THROW_NEW_DEBUGGER_EXCEPTION(str) { throw_new_debugger_exception(env, str); return;}
 
-static void throw_new_debugger_exception(JNIEnv* env, const char* errMsg) {
+void throw_new_debugger_exception(JNIEnv* env, const char* errMsg) {
   (*env)->ThrowNew(env, (*env)->FindClass(env, "sun/jvm/hotspot/debugger/DebuggerException"), errMsg);
 }
 
-static struct ps_prochandle* get_proc_handle(JNIEnv* env, jobject this_obj) {
+struct ps_prochandle* get_proc_handle(JNIEnv* env, jobject this_obj) {
   jlong ptr = (*env)->GetLongField(env, this_obj, p_ps_prochandle_ID);
   return (struct ps_prochandle*)(intptr_t)ptr;
 }
@@ -280,6 +280,7 @@ JNIEXPORT jbyteArray JNICALL Java_sun_jvm_hotspot_debugger_linux_LinuxDebuggerLo
   return (err == PS_OK)? array : 0;
 }
 
+#if defined(i386) || defined(ia64) || defined(amd64) || defined(sparc) || defined(sparcv9)
 JNIEXPORT jlongArray JNICALL Java_sun_jvm_hotspot_debugger_linux_LinuxDebuggerLocal_getThreadIntegerRegisterSet0
   (JNIEnv *env, jobject this_obj, jint lwp_id) {
 
@@ -410,3 +411,4 @@ JNIEXPORT jlongArray JNICALL Java_sun_jvm_hotspot_debugger_linux_LinuxDebuggerLo
   (*env)->ReleaseLongArrayElements(env, array, regs, JNI_COMMIT);
   return array;
 }
+#endif

hotspot/agent/src/os/linux/libproc.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2003, 2007, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2003, 2012, 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
@@ -25,10 +25,15 @@
 #ifndef _LIBPROC_H_
 #define _LIBPROC_H_
 
+#include <jni.h>
 #include <unistd.h>
 #include <stdint.h>
 #include "proc_service.h"
 
+#if defined(arm) || defined(ppc)
+#include "libproc_md.h"
+#endif
+
 #if defined(sparc) || defined(sparcv9)
 /*
   If _LP64 is defined ptrace.h should be taken from /usr/include/asm-sparc64
@@ -139,4 +144,8 @@ uintptr_t lookup_symbol(struct ps_prochandle* ph,  const char* object_name,
 // address->nearest symbol lookup. return NULL for no symbol
 const char* symbol_for_pc(struct ps_prochandle* ph, uintptr_t addr, uintptr_t* poffset);
 
+struct ps_prochandle* get_proc_handle(JNIEnv* env, jobject this_obj);
+
+void throw_new_debugger_exception(JNIEnv* env, const char* errMsg);
+
 #endif //__LIBPROC_H_

hotspot/agent/src/share/classes/sun/jvm/hotspot/HotSpotAgent.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2000, 2012, 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
@@ -549,7 +549,13 @@ public class HotSpotAgent {
                     machDesc = new MachineDescriptionSPARC32Bit();
             }
         } else {
-            throw new DebuggerException("Linux only supported on x86/ia64/amd64/sparc/sparc64");
+          try {
+            machDesc = (MachineDescription)
+              Class.forName("sun.jvm.hotspot.debugger.MachineDescription" +
+                            cpu.toUpperCase()).newInstance();
+          } catch (Exception e) {
+            throw new DebuggerException("Linux not supported on machine type " + cpu);
+          }
         }
 
         LinuxDebuggerLocal dbg =

hotspot/agent/src/share/classes/sun/jvm/hotspot/bugspot/BugSpotAgent.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2012, 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
@@ -737,8 +737,15 @@ public class BugSpotAgent {
                machDesc = new MachineDescriptionSPARC32Bit();
             }
         } else {
-            throw new DebuggerException("Linux only supported on x86/ia64/amd64/sparc/sparc64");
+          try {
+            machDesc = (MachineDescription)
+              Class.forName("sun.jvm.hotspot.debugger.MachineDescription" +
+              cpu.toUpperCase()).newInstance();
+          } catch (Exception e) {
+            throw new DebuggerException("unsupported machine type");
           }
+        }
+
 
         // Note we do not use a cache for the local debugger in server
         // mode; it will be taken care of on the client side (once remote

hotspot/agent/src/share/classes/sun/jvm/hotspot/code/CodeBlob.java

@@ -93,7 +93,6 @@ public class CodeBlob extends VMObject {
   public boolean isUncommonTrapStub()   { return false; }
   public boolean isExceptionStub()      { return false; }
   public boolean isSafepointStub()      { return false; }
-  public boolean isRicochetBlob()       { return false; }
   public boolean isAdapterBlob()        { return false; }
 
   // Fine grain nmethod support: isNmethod() == isJavaMethod() || isNativeMethod() || isOSRMethod()

hotspot/agent/src/share/classes/sun/jvm/hotspot/code/CodeCache.java

@@ -57,7 +57,6 @@ public class CodeCache {
     virtualConstructor.addMapping("BufferBlob", BufferBlob.class);
     virtualConstructor.addMapping("nmethod", NMethod.class);
     virtualConstructor.addMapping("RuntimeStub", RuntimeStub.class);
-    virtualConstructor.addMapping("RicochetBlob", RicochetBlob.class);
     virtualConstructor.addMapping("AdapterBlob", AdapterBlob.class);
     virtualConstructor.addMapping("MethodHandlesAdapterBlob", MethodHandlesAdapterBlob.class);
     virtualConstructor.addMapping("SafepointBlob", SafepointBlob.class);
@@ -127,10 +126,6 @@ public class CodeCache {
       Assert.that(result.blobContains(start) || result.blobContains(start.addOffsetTo(8)),
                                                                     "found wrong CodeBlob");
     }
-    if (result.isRicochetBlob()) {
-      // This should probably be done for other SingletonBlobs
-      return VM.getVM().ricochetBlob();
-    }
     return result;
   }
 

hotspot/agent/src/share/classes/sun/jvm/hotspot/code/RicochetBlob.java

@@ -1,70 +0,0 @@
-/*
- * Copyright (c) 2011, 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
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-package sun.jvm.hotspot.code;
-
-import java.util.*;
-import sun.jvm.hotspot.debugger.*;
-import sun.jvm.hotspot.runtime.*;
-import sun.jvm.hotspot.types.*;
-
-/** RicochetBlob (currently only used by Compiler 2) */
-
-public class RicochetBlob extends SingletonBlob {
-  static {
-    VM.registerVMInitializedObserver(new Observer() {
-        public void update(Observable o, Object data) {
-          initialize(VM.getVM().getTypeDataBase());
-        }
-      });
-  }
-
-  private static void initialize(TypeDataBase db) {
-    Type type = db.lookupType("RicochetBlob");
-
-    bounceOffsetField                = type.getCIntegerField("_bounce_offset");
-    exceptionOffsetField             = type.getCIntegerField("_exception_offset");
-  }
-
-  private static CIntegerField bounceOffsetField;
-  private static CIntegerField exceptionOffsetField;
-
-  public RicochetBlob(Address addr) {
-    super(addr);
-  }
-
-  public boolean isRicochetBlob() {
-    return true;
-  }
-
-  public Address bounceAddr() {
-    return codeBegin().addOffsetTo(bounceOffsetField.getValue(addr));
-  }
-
-  public boolean returnsToBounceAddr(Address pc) {
-    Address bouncePc = bounceAddr();
-    return (pc.equals(bouncePc) || pc.addOffsetTo(Frame.pcReturnOffset()).equals(bouncePc));
-  }
-
-}

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/ThreadContext.java

@@ -24,6 +24,8 @@
 
 package sun.jvm.hotspot.debugger;
 
+import sun.jvm.hotspot.debugger.cdbg.*;
+
 /** This is a placeholder interface for a thread's context, containing
     only integer registers (no floating-point ones). What it contains
     is platform-dependent. Not all registers are guaranteed to be
@@ -54,4 +56,6 @@ public interface ThreadContext {
   /** Set the value of the specified register (0..getNumRegisters() -
       1) as an Address */
   public void setRegisterAsAddress(int index, Address value);
+
+  public CFrame getTopFrame(Debugger dbg);
 }

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/amd64/AMD64ThreadContext.java

@@ -25,6 +25,7 @@
 package sun.jvm.hotspot.debugger.amd64;
 
 import sun.jvm.hotspot.debugger.*;
+import sun.jvm.hotspot.debugger.cdbg.*;
 
 /** Specifies the thread context on amd64 platforms; only a sub-portion
  * of the context is guaranteed to be present on all operating
@@ -98,6 +99,10 @@ public abstract class AMD64ThreadContext implements ThreadContext {
         return data[index];
     }
 
+    public CFrame getTopFrame(Debugger dbg) {
+        return null;
+    }
+
     /** This can't be implemented in this class since we would have to
      * tie the implementation to, for example, the debugging system */
     public abstract void setRegisterAsAddress(int index, Address value);

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/ia64/IA64ThreadContext.java

@@ -25,6 +25,7 @@
 package sun.jvm.hotspot.debugger.ia64;
 
 import sun.jvm.hotspot.debugger.*;
+import sun.jvm.hotspot.debugger.cdbg.*;
 
 /** Specifies the thread context on ia64 platform; only a sub-portion
     of the context is guaranteed to be present on all operating
@@ -172,6 +173,10 @@ public abstract class IA64ThreadContext implements ThreadContext {
     return data[index];
   }
 
+  public CFrame getTopFrame(Debugger dbg) {
+    return null;
+  }
+
   /** This can't be implemented in this class since we would have to
       tie the implementation to, for example, the debugging system */
   public abstract void setRegisterAsAddress(int index, Address value);

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/linux/LinuxCDebugger.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2003, 2006, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2003, 2012, 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
@@ -107,7 +107,9 @@ class LinuxCDebugger implements CDebugger {
        if (pc == null) return null;
        return new LinuxSPARCCFrame(dbg, sp, pc, LinuxDebuggerLocal.getAddressSize());
     } else {
-       throw new DebuggerException(cpu + " is not yet supported");
+       // Runtime exception thrown by LinuxThreadContextFactory if unknown cpu
+       ThreadContext context = (ThreadContext) thread.getContext();
+       return context.getTopFrame(dbg);
     }
   }
 

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/linux/LinuxThreadContextFactory.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2006, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2012, 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
@@ -24,6 +24,7 @@
 
 package sun.jvm.hotspot.debugger.linux;
 
+import java.lang.reflect.*;
 import sun.jvm.hotspot.debugger.*;
 import sun.jvm.hotspot.debugger.linux.amd64.*;
 import sun.jvm.hotspot.debugger.linux.ia64.*;
@@ -42,7 +43,15 @@ class LinuxThreadContextFactory {
       } else if (cpu.equals("sparc")) {
          return new LinuxSPARCThreadContext(dbg);
       } else  {
+        try {
+          Class tcc = Class.forName("sun.jvm.hotspot.debugger.linux." +
+             cpu.toLowerCase() + ".Linux" + cpu.toUpperCase() +
+             "ThreadContext");
+          Constructor[] ctcc = tcc.getConstructors();
+          return (ThreadContext)ctcc[0].newInstance(dbg);
+        } catch (Exception e) {
           throw new RuntimeException("cpu " + cpu + " is not yet supported");
         }
       }
    }
+}

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/proc/ProcDebuggerLocal.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2008, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2012, 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
@@ -27,6 +27,7 @@ package sun.jvm.hotspot.debugger.proc;
 import java.io.*;
 import java.net.*;
 import java.util.*;
+import java.lang.reflect.*;
 import sun.jvm.hotspot.debugger.*;
 import sun.jvm.hotspot.debugger.cdbg.*;
 import sun.jvm.hotspot.debugger.proc.amd64.*;
@@ -86,7 +87,16 @@ public class ProcDebuggerLocal extends DebuggerBase implements ProcDebugger {
             pcRegIndex = AMD64ThreadContext.RIP;
             fpRegIndex = AMD64ThreadContext.RBP;
         } else {
+          try {
+            Class tfc = Class.forName("sun.jvm.hotspot.debugger.proc." +
+               cpu.toLowerCase() + ".Proc" + cpu.toUpperCase() +
+               "ThreadFactory");
+            Constructor[] ctfc = tfc.getConstructors();
+            threadFactory = (ProcThreadFactory)ctfc[0].newInstance(this);
+          } catch (Exception e) {
             throw new RuntimeException("Thread access for CPU architecture " + PlatformInfo.getCPU() + " not yet supported");
+            // Note: pcRegIndex and fpRegIndex do not appear to be referenced
+          }
         }
         if (useCache) {
             // Cache portion of the remote process's address space.
@@ -375,7 +385,11 @@ public class ProcDebuggerLocal extends DebuggerBase implements ProcDebugger {
         int pagesize = getPageSize0();
         if (pagesize == -1) {
             // return the hard coded default value.
-            pagesize = (PlatformInfo.getCPU().equals("x86"))? 4096 : 8192;
+            if (PlatformInfo.getCPU().equals("sparc") ||
+                PlatformInfo.getCPU().equals("amd64") )
+               pagesize = 8196;
+            else
+               pagesize = 4096;
         }
         return pagesize;
     }

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/remote/RemoteDebuggerClient.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2009, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2012, 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
@@ -26,6 +26,7 @@ package sun.jvm.hotspot.debugger.remote;
 
 import java.rmi.*;
 import java.util.*;
+import java.lang.reflect.*;
 
 import sun.jvm.hotspot.debugger.*;
 import sun.jvm.hotspot.debugger.cdbg.*;
@@ -70,8 +71,19 @@ public class RemoteDebuggerClient extends DebuggerBase implements JVMDebugger {
         cacheNumPages = parseCacheNumPagesProperty(cacheSize / cachePageSize);
         unalignedAccessesOkay = true;
       } else {
+        try {
+          Class tf = Class.forName("sun.jvm.hotspot.debugger.remote." +
+            cpu.toLowerCase() + ".Remote" + cpu.toUpperCase() +
+            "ThreadFactory");
+          Constructor[] ctf = tf.getConstructors();
+          threadFactory = (RemoteThreadFactory)ctf[0].newInstance(this);
+        } catch (Exception e) {
           throw new DebuggerException("Thread access for CPU architecture " + cpu + " not yet supported");
         }
+        cachePageSize = 4096;
+        cacheNumPages = parseCacheNumPagesProperty(cacheSize / cachePageSize);
+        unalignedAccessesOkay = false;
+      }
 
       // Cache portion of the remote process's address space.
       initCache(cachePageSize, cacheNumPages);

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/sparc/SPARCThreadContext.java

@@ -25,6 +25,7 @@
 package sun.jvm.hotspot.debugger.sparc;
 
 import sun.jvm.hotspot.debugger.*;
+import sun.jvm.hotspot.debugger.cdbg.*;
 
 /** Currently provides just the minimal information necessary to get
     stack traces working. FIXME: currently hardwired for v9 -- will
@@ -124,6 +125,10 @@ public abstract class SPARCThreadContext implements ThreadContext {
     return data[index];
   }
 
+  public CFrame getTopFrame(Debugger dbg) {
+    return null;
+  }
+
   /** This can't be implemented in this class since we would have to
       tie the implementation to, for example, the debugging system */
   public abstract void setRegisterAsAddress(int index, Address value);

hotspot/agent/src/share/classes/sun/jvm/hotspot/debugger/x86/X86ThreadContext.java

@@ -25,6 +25,7 @@
 package sun.jvm.hotspot.debugger.x86;
 
 import sun.jvm.hotspot.debugger.*;
+import sun.jvm.hotspot.debugger.cdbg.*;
 
 /** Specifies the thread context on x86 platforms; only a sub-portion
     of the context is guaranteed to be present on all operating
@@ -109,6 +110,10 @@ public abstract class X86ThreadContext implements ThreadContext {
     return data[index];
   }
 
+  public CFrame getTopFrame(Debugger dbg) {
+    return null;
+  }
+
   /** This can't be implemented in this class since we would have to
       tie the implementation to, for example, the debugging system */
   public abstract void setRegisterAsAddress(int index, Address value);

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/Frame.java

@@ -147,12 +147,6 @@ public abstract class Frame implements Cloneable {
     }
   }
 
-  public boolean isRicochetFrame() {
-    CodeBlob cb = VM.getVM().getCodeCache().findBlob(getPC());
-    RicochetBlob rcb = VM.getVM().ricochetBlob();
-    return (cb == rcb && rcb != null && rcb.returnsToBounceAddr(getPC()));
-  }
-
   public boolean isCompiledFrame() {
     if (Assert.ASSERTS_ENABLED) {
       Assert.that(!VM.getVM().isCore(), "noncore builds only");
@@ -216,8 +210,7 @@ public abstract class Frame implements Cloneable {
   public Frame realSender(RegisterMap map) {
     if (!VM.getVM().isCore()) {
       Frame result = sender(map);
-      while (result.isRuntimeFrame() ||
+      while (result.isRuntimeFrame()) {
-             result.isRicochetFrame()) {
         result = result.sender(map);
       }
       return result;
@@ -631,9 +624,6 @@ public abstract class Frame implements Cloneable {
     if (Assert.ASSERTS_ENABLED) {
       Assert.that(cb != null, "sanity check");
     }
-    if (cb == VM.getVM().ricochetBlob()) {
-      oopsRicochetDo(oopVisitor, regMap);
-    }
     if (cb.getOopMaps() != null) {
       OopMapSet.oopsDo(this, cb, regMap, oopVisitor, VM.getVM().isDebugging());
 
@@ -650,10 +640,6 @@ public abstract class Frame implements Cloneable {
     //    }
   }
 
-  private void oopsRicochetDo      (AddressVisitor oopVisitor, RegisterMap regMap) {
-    // XXX Empty for now
-  }
-
   // FIXME: implement the above routines, plus add
   // oops_interpreted_arguments_do and oops_compiled_arguments_do
 }

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/Threads.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2000, 2012, 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
@@ -91,6 +91,16 @@ public class Threads {
                 access = new LinuxAMD64JavaThreadPDAccess();
             } else if (cpu.equals("sparc")) {
                 access = new LinuxSPARCJavaThreadPDAccess();
+            } else {
+              try {
+                access = (JavaThreadPDAccess)
+                  Class.forName("sun.jvm.hotspot.runtime.linux_" +
+                     cpu.toLowerCase() + ".Linux" + cpu.toUpperCase() +
+                     "JavaThreadPDAccess").newInstance();
+              } catch (Exception e) {
+                throw new RuntimeException("OS/CPU combination " + os + "/" + cpu +
+                                           " not yet supported");
+              }
             }
         } else if (os.equals("bsd")) {
             if (cpu.equals("x86")) {

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/VM.java

@@ -87,13 +87,13 @@ public class VM {
   private StubRoutines stubRoutines;
   private Bytes        bytes;
 
-  private RicochetBlob ricochetBlob;
-
   /** Flags indicating whether we are attached to a core, C1, or C2 build */
   private boolean      usingClientCompiler;
   private boolean      usingServerCompiler;
   /** Flag indicating whether UseTLAB is turned on */
   private boolean      useTLAB;
+  /** Flag indicating whether invokedynamic support is on */
+  private boolean      enableInvokeDynamic;
   /** alignment constants */
   private boolean      isLP64;
   private int          bytesPerLong;
@@ -319,6 +319,7 @@ public class VM {
     }
 
     useTLAB = (db.lookupIntConstant("UseTLAB").intValue() != 0);
+    enableInvokeDynamic = (db.lookupIntConstant("EnableInvokeDynamic").intValue() != 0);
 
     if (debugger != null) {
       isLP64 = debugger.getMachineDescription().isLP64();
@@ -554,6 +555,10 @@ public class VM {
     return useTLAB;
   }
 
+  public boolean getEnableInvokeDynamic() {
+    return enableInvokeDynamic;
+  }
+
   public TypeDataBase getTypeDataBase() {
     return db;
   }
@@ -628,18 +633,6 @@ public class VM {
     return stubRoutines;
   }
 
-  public RicochetBlob ricochetBlob() {
-    if (ricochetBlob == null) {
-      Type ricochetType  = db.lookupType("SharedRuntime");
-      AddressField ricochetBlobAddress = ricochetType.getAddressField("_ricochet_blob");
-      Address addr = ricochetBlobAddress.getValue();
-      if (addr != null) {
-        ricochetBlob = new RicochetBlob(addr);
-      }
-    }
-    return ricochetBlob;
-  }
-
   public VMRegImpl getVMRegImplInfo() {
     if (vmregImpl == null) {
       vmregImpl = new VMRegImpl();

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/sparc/SPARCFrame.java

@@ -571,8 +571,6 @@ public class SPARCFrame extends Frame {
     //        registers callee-saved, then we will have to copy over
     //        the RegisterMap update logic from the Intel code.
 
-    if (isRicochetFrame()) return senderForRicochetFrame(map);
-
     // The constructor of the sender must know whether this frame is interpreted so it can set the
     // sender's _interpreter_sp_adjustment field.
     if (VM.getVM().getInterpreter().contains(pc)) {
@@ -945,20 +943,6 @@ public class SPARCFrame extends Frame {
   }
 
 
-  private Frame senderForRicochetFrame(SPARCRegisterMap map) {
-    if (DEBUG) {
-      System.out.println("senderForRicochetFrame");
-    }
-    //RicochetFrame* f = RicochetFrame::from_frame(fr);
-    // Cf. is_interpreted_frame path of frame::sender
-    Address youngerSP = getSP();
-    Address sp        = getSenderSP();
-    map.makeIntegerRegsUnsaved();
-    map.shiftWindow(sp, youngerSP);
-    boolean thisFrameAdjustedStack = true;  // I5_savedSP is live in this RF
-    return new SPARCFrame(biasSP(sp), biasSP(youngerSP), thisFrameAdjustedStack);
-  }
-
   private Frame senderForEntryFrame(RegisterMap regMap) {
     SPARCRegisterMap map = (SPARCRegisterMap) regMap;
 

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/sparc/SPARCRicochetFrame.java

@@ -1,77 +0,0 @@
-/*
- * Copyright (c) 2011, 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
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-package sun.jvm.hotspot.runtime.sparc;
-
-import java.util.*;
-import sun.jvm.hotspot.asm.sparc.SPARCRegister;
-import sun.jvm.hotspot.asm.sparc.SPARCRegisters;
-import sun.jvm.hotspot.debugger.*;
-import sun.jvm.hotspot.runtime.*;
-import sun.jvm.hotspot.types.*;
-
-public class SPARCRicochetFrame {
-  static {
-    VM.registerVMInitializedObserver(new Observer() {
-        public void update(Observable o, Object data) {
-          initialize(VM.getVM().getTypeDataBase());
-        }
-      });
-  }
-
-  private SPARCFrame frame;
-
-  private static void initialize(TypeDataBase db) {
-    // Type type = db.lookupType("MethodHandles::RicochetFrame");
-
-  }
-
-  static SPARCRicochetFrame fromFrame(SPARCFrame f) {
-    return new SPARCRicochetFrame(f);
-  }
-
-  private SPARCRicochetFrame(SPARCFrame f) {
-    frame = f;
-  }
-
-  private Address registerValue(SPARCRegister reg) {
-    return frame.getSP().addOffsetTo(reg.spOffsetInSavedWindow()).getAddressAt(0);
-  }
-
-  public Address savedArgsBase() {
-    return registerValue(SPARCRegisters.L4);
-  }
-  public Address exactSenderSP() {
-    return registerValue(SPARCRegisters.I5);
-  }
-  public Address senderLink() {
-    return frame.getSenderSP();
-  }
-  public Address senderPC() {
-    return frame.getSenderPC();
-  }
-  public Address extendedSenderSP() {
-    return savedArgsBase();
-  }
-}

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/x86/X86Frame.java

@@ -269,7 +269,6 @@ public class X86Frame extends Frame {
 
     if (isEntryFrame())       return senderForEntryFrame(map);
     if (isInterpretedFrame()) return senderForInterpreterFrame(map);
-    if (isRicochetFrame())    return senderForRicochetFrame(map);
 
     if(cb == null) {
       cb = VM.getVM().getCodeCache().findBlob(getPC());
@@ -288,16 +287,6 @@ public class X86Frame extends Frame {
     return new X86Frame(getSenderSP(), getLink(), getSenderPC());
   }
 
-  private Frame senderForRicochetFrame(X86RegisterMap map) {
-    if (DEBUG) {
-      System.out.println("senderForRicochetFrame");
-    }
-    X86RicochetFrame f = X86RicochetFrame.fromFrame(this);
-    if (map.getUpdateMap())
-      updateMapWithSavedLink(map, f.senderLinkAddress());
-    return new X86Frame(f.extendedSenderSP(), f.exactSenderSP(), f.senderLink(), f.senderPC());
-  }
-
   private Frame senderForEntryFrame(X86RegisterMap map) {
     if (DEBUG) {
       System.out.println("senderForEntryFrame");

hotspot/agent/src/share/classes/sun/jvm/hotspot/runtime/x86/X86RicochetFrame.java

@@ -1,81 +0,0 @@
-/*
- * Copyright (c) 2011, 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
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-package sun.jvm.hotspot.runtime.x86;
-
-import java.util.*;
-import sun.jvm.hotspot.debugger.*;
-import sun.jvm.hotspot.runtime.*;
-import sun.jvm.hotspot.types.*;
-
-public class X86RicochetFrame extends VMObject {
-  static {
-    VM.registerVMInitializedObserver(new Observer() {
-        public void update(Observable o, Object data) {
-          initialize(VM.getVM().getTypeDataBase());
-        }
-      });
-  }
-
-  private static void initialize(TypeDataBase db) {
-    Type type = db.lookupType("MethodHandles::RicochetFrame");
-
-    senderLinkField    = type.getAddressField("_sender_link");
-    savedArgsBaseField = type.getAddressField("_saved_args_base");
-    exactSenderSPField = type.getAddressField("_exact_sender_sp");
-    senderPCField      = type.getAddressField("_sender_pc");
-  }
-
-  private static AddressField senderLinkField;
-  private static AddressField savedArgsBaseField;
-  private static AddressField exactSenderSPField;
-  private static AddressField senderPCField;
-
-  static X86RicochetFrame fromFrame(X86Frame f) {
-    return new X86RicochetFrame(f.getFP().addOffsetTo(- senderLinkField.getOffset()));
-  }
-
-  private X86RicochetFrame(Address addr) {
-    super(addr);
-  }
-
-  public Address senderLink() {
-    return senderLinkField.getValue(addr);
-  }
-  public Address senderLinkAddress() {
-    return addr.addOffsetTo(senderLinkField.getOffset());
-  }
-  public Address savedArgsBase() {
-    return savedArgsBaseField.getValue(addr);
-  }
-  public Address extendedSenderSP() {
-    return savedArgsBase();
-  }
-  public Address exactSenderSP() {
-    return exactSenderSPField.getValue(addr);
-  }
-  public Address senderPC() {
-    return senderPCField.getValue(addr);
-  }
-}

hotspot/agent/src/share/classes/sun/jvm/hotspot/ui/classbrowser/HTMLGenerator.java

@@ -204,9 +204,15 @@ public class HTMLGenerator implements /* imports */ ClassConstants {
       } else if (cpu.equals("ia64")) {
          cpuHelper = new IA64Helper();
       } else {
+        try {
+          cpuHelper = (CPUHelper)Class.forName("sun.jvm.hotspot.asm." +
+             cpu.toLowerCase() + "." + cpu.toUpperCase() +
+             "Helper").newInstance();
+        } catch (Exception e) {
           throw new RuntimeException("cpu '" + cpu + "' is not yet supported!");
         }
       }
+   }
 
    protected static synchronized CPUHelper getCPUHelper() {
       return cpuHelper;

hotspot/agent/src/share/classes/sun/jvm/hotspot/utilities/AltPlatformInfo.java

@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2000, 2012, 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
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+package sun.jvm.hotspot.utilities;
+
+public interface AltPlatformInfo {
+  // Additional cpu types can be tested via this interface
+
+  public boolean knownCPU(String cpu);
+}

hotspot/agent/src/share/classes/sun/jvm/hotspot/utilities/PlatformInfo.java

@@ -64,6 +64,13 @@ public class PlatformInfo {
     } else if (cpu.equals("ia64") || cpu.equals("amd64") || cpu.equals("x86_64")) {
       return cpu;
     } else {
+      try {
+        Class pic = Class.forName("sun.jvm.hotspot.utilities.PlatformInfoClosed");
+        AltPlatformInfo api = (AltPlatformInfo)pic.newInstance();
+        if (api.knownCPU(cpu)) {
+          return cpu;
+        }
+      } catch (Exception e) {}
       throw new UnsupportedPlatformException("CPU type " + cpu + " not yet supported");
     }
   }

hotspot/make/defs.make

@@ -22,6 +22,14 @@
 #  
 #
 
+ifeq ($(HS_ALT_MAKE),)
+  ifneq ($(OPENJDK),true)
+    HS_ALT_MAKE=$(GAMMADIR)/make/closed
+  else
+    HS_ALT_MAKE=NO_SUCH_PATH
+  endif
+endif
+
 # The common definitions for hotspot builds.
 
 # Optionally include SPEC file generated by configure.
@@ -327,3 +335,4 @@ EXPORT_LIST += $(EXPORT_INCLUDE_DIR)/jmm.h
 ifndef JAVASE_EMBEDDED
 EXPORT_LIST += $(EXPORT_INCLUDE_DIR)/jfr.h
 endif
+

hotspot/make/hotspot_version

@@ -35,7 +35,7 @@ HOTSPOT_VM_COPYRIGHT=Copyright 2012
 
 HS_MAJOR_VER=24
 HS_MINOR_VER=0
-HS_BUILD_NUMBER=19
+HS_BUILD_NUMBER=22
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=8

hotspot/make/jprt.properties

@@ -38,7 +38,7 @@ jprt.need.sibling.build=false
 
 # This tells jprt what default release we want to build
 
-jprt.hotspot.default.release=jdk7
+jprt.hotspot.default.release=jdk8
 
 jprt.tools.default.release=${jprt.submit.option.release?${jprt.submit.option.release}:${jprt.hotspot.default.release}}
 
@@ -54,77 +54,77 @@ jprt.sync.push=false
 # Define the Solaris platforms we want for the various releases
 jprt.my.solaris.sparc.jdk8=solaris_sparc_5.10
 jprt.my.solaris.sparc.jdk7=solaris_sparc_5.10
-jprt.my.solaris.sparc.jdk7u6=${jprt.my.solaris.sparc.jdk7}
+jprt.my.solaris.sparc.jdk7u8=${jprt.my.solaris.sparc.jdk7}
 jprt.my.solaris.sparc=${jprt.my.solaris.sparc.${jprt.tools.default.release}}
 
 jprt.my.solaris.sparcv9.jdk8=solaris_sparcv9_5.10
 jprt.my.solaris.sparcv9.jdk7=solaris_sparcv9_5.10
-jprt.my.solaris.sparcv9.jdk7u6=${jprt.my.solaris.sparcv9.jdk7}
+jprt.my.solaris.sparcv9.jdk7u8=${jprt.my.solaris.sparcv9.jdk7}
 jprt.my.solaris.sparcv9=${jprt.my.solaris.sparcv9.${jprt.tools.default.release}}
 
 jprt.my.solaris.i586.jdk8=solaris_i586_5.10
 jprt.my.solaris.i586.jdk7=solaris_i586_5.10
-jprt.my.solaris.i586.jdk7u6=${jprt.my.solaris.i586.jdk7}
+jprt.my.solaris.i586.jdk7u8=${jprt.my.solaris.i586.jdk7}
 jprt.my.solaris.i586=${jprt.my.solaris.i586.${jprt.tools.default.release}}
 
 jprt.my.solaris.x64.jdk8=solaris_x64_5.10
 jprt.my.solaris.x64.jdk7=solaris_x64_5.10
-jprt.my.solaris.x64.jdk7u6=${jprt.my.solaris.x64.jdk7}
+jprt.my.solaris.x64.jdk7u8=${jprt.my.solaris.x64.jdk7}
 jprt.my.solaris.x64=${jprt.my.solaris.x64.${jprt.tools.default.release}}
 
 jprt.my.linux.i586.jdk8=linux_i586_2.6
 jprt.my.linux.i586.jdk7=linux_i586_2.6
-jprt.my.linux.i586.jdk7u6=${jprt.my.linux.i586.jdk7}
+jprt.my.linux.i586.jdk7u8=${jprt.my.linux.i586.jdk7}
 jprt.my.linux.i586=${jprt.my.linux.i586.${jprt.tools.default.release}}
 
 jprt.my.linux.x64.jdk8=linux_x64_2.6
 jprt.my.linux.x64.jdk7=linux_x64_2.6
-jprt.my.linux.x64.jdk7u6=${jprt.my.linux.x64.jdk7}
+jprt.my.linux.x64.jdk7u8=${jprt.my.linux.x64.jdk7}
 jprt.my.linux.x64=${jprt.my.linux.x64.${jprt.tools.default.release}}
 
 jprt.my.linux.ppc.jdk8=linux_ppc_2.6
 jprt.my.linux.ppc.jdk7=linux_ppc_2.6
-jprt.my.linux.ppc.jdk7u6=${jprt.my.linux.ppc.jdk7}
+jprt.my.linux.ppc.jdk7u8=${jprt.my.linux.ppc.jdk7}
 jprt.my.linux.ppc=${jprt.my.linux.ppc.${jprt.tools.default.release}}
 
 jprt.my.linux.ppcv2.jdk8=linux_ppcv2_2.6
 jprt.my.linux.ppcv2.jdk7=linux_ppcv2_2.6
-jprt.my.linux.ppcv2.jdk7u6=${jprt.my.linux.ppcv2.jdk7}
+jprt.my.linux.ppcv2.jdk7u8=${jprt.my.linux.ppcv2.jdk7}
 jprt.my.linux.ppcv2=${jprt.my.linux.ppcv2.${jprt.tools.default.release}}
 
 jprt.my.linux.ppcsflt.jdk8=linux_ppcsflt_2.6
 jprt.my.linux.ppcsflt.jdk7=linux_ppcsflt_2.6
-jprt.my.linux.ppcsflt.jdk7u6=${jprt.my.linux.ppcsflt.jdk7}
+jprt.my.linux.ppcsflt.jdk7u8=${jprt.my.linux.ppcsflt.jdk7}
 jprt.my.linux.ppcsflt=${jprt.my.linux.ppcsflt.${jprt.tools.default.release}}
 
 jprt.my.linux.armvfp.jdk8=linux_armvfp_2.6
 jprt.my.linux.armvfp.jdk7=linux_armvfp_2.6
-jprt.my.linux.armvfp.jdk7u6=${jprt.my.linux.armvfp.jdk7}
+jprt.my.linux.armvfp.jdk7u8=${jprt.my.linux.armvfp.jdk7}
 jprt.my.linux.armvfp=${jprt.my.linux.armvfp.${jprt.tools.default.release}}
 
 jprt.my.linux.armv6.jdk8=linux_armv6_2.6
 jprt.my.linux.armv6.jdk7=linux_armv6_2.6
-jprt.my.linux.armv6.jdk7u6=${jprt.my.linux.armv6.jdk7}
+jprt.my.linux.armv6.jdk7u8=${jprt.my.linux.armv6.jdk7}
 jprt.my.linux.armv6=${jprt.my.linux.armv6.${jprt.tools.default.release}}
 
 jprt.my.linux.armsflt.jdk8=linux_armsflt_2.6
 jprt.my.linux.armsflt.jdk7=linux_armsflt_2.6
-jprt.my.linux.armsflt.jdk7u6=${jprt.my.linux.armsflt.jdk7}
+jprt.my.linux.armsflt.jdk7u8=${jprt.my.linux.armsflt.jdk7}
 jprt.my.linux.armsflt=${jprt.my.linux.armsflt.${jprt.tools.default.release}}
 
 jprt.my.macosx.x64.jdk8=macosx_x64_10.7
 jprt.my.macosx.x64.jdk7=macosx_x64_10.7
-jprt.my.macosx.x64.jdk7u6=${jprt.my.macosx.x64.jdk7}
+jprt.my.macosx.x64.jdk7u8=${jprt.my.macosx.x64.jdk7}
 jprt.my.macosx.x64=${jprt.my.macosx.x64.${jprt.tools.default.release}}
 
 jprt.my.windows.i586.jdk8=windows_i586_5.1
 jprt.my.windows.i586.jdk7=windows_i586_5.1
-jprt.my.windows.i586.jdk7u6=${jprt.my.windows.i586.jdk7}
+jprt.my.windows.i586.jdk7u8=${jprt.my.windows.i586.jdk7}
 jprt.my.windows.i586=${jprt.my.windows.i586.${jprt.tools.default.release}}
 
 jprt.my.windows.x64.jdk8=windows_x64_5.2
 jprt.my.windows.x64.jdk7=windows_x64_5.2
-jprt.my.windows.x64.jdk7u6=${jprt.my.windows.x64.jdk7}
+jprt.my.windows.x64.jdk7u8=${jprt.my.windows.x64.jdk7}
 jprt.my.windows.x64=${jprt.my.windows.x64.${jprt.tools.default.release}}
 
 # Standard list of jprt build targets for this source tree
@@ -159,7 +159,7 @@ jprt.build.targets.all=${jprt.build.targets.standard}, \
 
 jprt.build.targets.jdk8=${jprt.build.targets.all}
 jprt.build.targets.jdk7=${jprt.build.targets.all}
-jprt.build.targets.jdk7u6=${jprt.build.targets.all}
+jprt.build.targets.jdk7u8=${jprt.build.targets.all}
 jprt.build.targets=${jprt.build.targets.${jprt.tools.default.release}}
 
 # Subset lists of test targets for this source tree
@@ -452,7 +452,7 @@ jprt.test.targets.embedded= 		\
 
 jprt.test.targets.jdk8=${jprt.test.targets.standard}
 jprt.test.targets.jdk7=${jprt.test.targets.standard}
-jprt.test.targets.jdk7u6=${jprt.test.targets.jdk7}
+jprt.test.targets.jdk7u8=${jprt.test.targets.jdk7}
 jprt.test.targets=${jprt.test.targets.${jprt.tools.default.release}}
 
 # The default test/Makefile targets that should be run
@@ -512,7 +512,7 @@ jprt.make.rule.test.targets.embedded = \
 
 jprt.make.rule.test.targets.jdk8=${jprt.make.rule.test.targets.standard}
 jprt.make.rule.test.targets.jdk7=${jprt.make.rule.test.targets.standard}
-jprt.make.rule.test.targets.jdk7u6=${jprt.make.rule.test.targets.jdk7}
+jprt.make.rule.test.targets.jdk7u8=${jprt.make.rule.test.targets.jdk7}
 jprt.make.rule.test.targets=${jprt.make.rule.test.targets.${jprt.tools.default.release}}
 
 # 7155453: Work-around to prevent popups on OSX from blocking test completion

hotspot/make/linux/makefiles/adlc.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 1999, 2011, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 1999, 2012, 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
@@ -133,8 +133,10 @@ $(GENERATEDFILES): refresh_adfiles
 # Note that product files are updated via "mv", which is atomic.
 TEMPDIR := $(OUTDIR)/mktmp$(shell echo $$$$)
 
-# Debuggable by default
+ifneq ($(DEBUG_BINARIES), true)
+  # Debuggable by default (unless already done by DEBUG_BINARIES)
   CFLAGS += -g
+endif
 
 # Pass -D flags into ADLC.
 ADLCFLAGS += $(SYSDEFS)

hotspot/make/linux/makefiles/defs.make

@@ -295,6 +295,8 @@ ADD_SA_BINARIES/ia64  =
 ADD_SA_BINARIES/arm   = 
 ADD_SA_BINARIES/zero  = 
 
+-include $(HS_ALT_MAKE)/linux/makefiles/defs.make
+
 EXPORT_LIST += $(ADD_SA_BINARIES/$(HS_ARCH))
 
 

hotspot/make/linux/makefiles/gcc.make

@@ -215,6 +215,10 @@ AOUT_FLAGS += -Xlinker -export-dynamic
 #------------------------------------------------------------------------
 # Debug flags
 
+# DEBUG_BINARIES uses full -g debug information for all configs
+ifeq ($(DEBUG_BINARIES), true)
+  CFLAGS += -g
+else
   # Use the stabs format for debugging information (this is the default
   # on gcc-2.91). It's good enough, has all the information about line
   # numbers and local variables, and libjvm_g.so is only about 16M.
@@ -249,11 +253,6 @@ ifeq ($(ENABLE_FULL_DEBUG_SYMBOLS),1)
       OPT_CFLAGS += -gstabs
     endif
   endif
-
-# DEBUG_BINARIES overrides everything, use full -g debug information
-ifeq ($(DEBUG_BINARIES), true)
-  DEBUG_CFLAGS = -g
-  CFLAGS += $(DEBUG_CFLAGS)
 endif
 
 # If we are building HEADLESS, pass on to VM

hotspot/make/linux/makefiles/sa.make

@@ -30,10 +30,16 @@
 
 include $(GAMMADIR)/make/linux/makefiles/rules.make
 
+include $(GAMMADIR)/make/defs.make
+include $(GAMMADIR)/make/altsrc.make
+
 AGENT_DIR = $(GAMMADIR)/agent
 
 include $(GAMMADIR)/make/sa.files
 
+-include $(HS_ALT_MAKE)/linux/makefiles/sa.make
+
+
 TOPDIR    = $(shell echo `pwd`)
 GENERATED = $(TOPDIR)/../generated
 
@@ -52,17 +58,15 @@ SA_BUILD_VERSION_PROP = "sun.jvm.hotspot.runtime.VM.saBuildVersion=$(SA_BUILD_VE
 SA_PROPERTIES = $(SA_CLASSDIR)/sa.properties
 
 # if $(AGENT_DIR) does not exist, we don't build SA
-# also, we don't build SA on Itanium, PowerPC, ARM or zero.
+# also, we don't build SA on Itanium or zero.
 
 all: 
 	if [ -d $(AGENT_DIR) -a "$(SRCARCH)" != "ia64" \
-             -a "$(SRCARCH)" != "arm" \
-             -a "$(SRCARCH)" != "ppc" \
              -a "$(SRCARCH)" != "zero" ] ; then \
 	   $(MAKE) -f sa.make $(GENERATED)/sa-jdi.jar; \
 	fi
 
-$(GENERATED)/sa-jdi.jar: $(AGENT_FILES)
+$(GENERATED)/sa-jdi.jar:: $(AGENT_FILES)
 	$(QUIETLY) echo "Making $@"
 	$(QUIETLY) if [ "$(BOOT_JAVA_HOME)" = "" ]; then \
 	  echo "ALT_BOOTDIR, BOOTDIR or JAVA_HOME needs to be defined to build SA"; \
@@ -111,3 +115,5 @@ clean:
 	rm -rf $(SA_CLASSDIR)
 	rm -rf $(GENERATED)/sa-jdi.jar
 	rm -rf $(AGENT_FILES_LIST)
+
+-include $(HS_ALT_MAKE)/linux/makefiles/sa-rules.make

hotspot/make/linux/makefiles/saproc.make

@@ -21,6 +21,8 @@
 # questions.
 #  
 #
+include $(GAMMADIR)/make/defs.make
+include $(GAMMADIR)/make/altsrc.make
 
 # Rules to build serviceability agent library, used by vm.make
 
@@ -48,6 +50,8 @@ SASRCFILES = $(SASRCDIR)/salibelf.c                   \
              $(SASRCDIR)/ps_core.c                    \
              $(SASRCDIR)/LinuxDebuggerLocal.c
 
+-include $(HS_ALT_MAKE)/linux/makefiles/saproc.make
+
 SAMAPFILE = $(SASRCDIR)/mapfile
 
 DEST_SAPROC           = $(JDK_LIBDIR)/$(LIBSAPROC)
@@ -60,15 +64,19 @@ ifeq ($(DEBUG_BINARIES), true)
 endif
 
 # if $(AGENT_DIR) does not exist, we don't build SA
-# also, we don't build SA on Itanium, PPC, ARM or zero.
+# also, we don't build SA on Itanium or zero.
 
 ifneq ($(wildcard $(AGENT_DIR)),)
-ifneq ($(filter-out ia64 arm ppc zero,$(SRCARCH)),)
+ifneq ($(filter-out ia64 zero,$(SRCARCH)),)
   BUILDLIBSAPROC = $(LIBSAPROC)
 endif
 endif
 
-
+ifneq ($(ALT_SASRCDIR),)
+ALT_SAINCDIR=-I$(ALT_SASRCDIR)
+else
+ALT_SAINCDIR=
+endif
 SA_LFLAGS = $(MAPFLAG:FILENAME=$(SAMAPFILE)) $(LDFLAGS_HASH_STYLE)
 
 $(LIBSAPROC): $(SASRCFILES) $(SAMAPFILE)
@@ -84,6 +92,7 @@ $(LIBSAPROC): $(SASRCFILES) $(SAMAPFILE)
 	           -I$(GENERATED)                                       \
 	           -I$(BOOT_JAVA_HOME)/include                          \
 	           -I$(BOOT_JAVA_HOME)/include/$(Platform_os_family)    \
+			   $(ALT_SAINCDIR) 										\
 	           $(SASRCFILES)                                        \
 	           $(SA_LFLAGS)                                         \
 	           $(SA_DEBUG_CFLAGS)                                   \

hotspot/make/pic.make

@@ -32,7 +32,7 @@ ifneq ($(OSNAME), windows)
   ifndef LP64
     PARTIAL_NONPIC=1
   endif
-  PIC_ARCH = ppc
+  PIC_ARCH = ppc arm
   ifneq ("$(filter $(PIC_ARCH),$(BUILDARCH))","")
     PARTIAL_NONPIC=0
   endif

hotspot/make/windows/makefiles/defs.make

@@ -188,14 +188,22 @@ ifdef COOKED_BUILD_NUMBER
   MAKE_ARGS += JDK_BUILD_NUMBER=$(COOKED_BUILD_NUMBER)
 endif
 
-NMAKE= MAKEFLAGS= MFLAGS= nmake /NOLOGO
+NMAKE= MAKEFLAGS= MFLAGS= nmake -NOLOGO
+ifndef SYSTEM_UNAME
+  SYSTEM_UNAME := $(shell uname)
+  export SYSTEM_UNAME
+endif
 
 # Check for CYGWIN
-ifneq (,$(findstring CYGWIN,$(shell uname)))
+ifneq (,$(findstring CYGWIN,$(SYSTEM_UNAME)))
   USING_CYGWIN=true
 else
   USING_CYGWIN=false
 endif
+# Check for MinGW
+ifneq (,$(findstring MINGW,$(SYSTEM_UNAME)))
+  USING_MINGW=true
+endif
 # FIXUP: The subdirectory for a debug build is NOT the same on all platforms
 VM_DEBUG=debug
 
@@ -208,6 +216,11 @@ ifeq ($(USING_CYGWIN), true)
   ABS_BOOTDIR     := $(subst /,\\,$(shell /bin/cygpath -m -a "$(BOOTDIR)"))
   ABS_GAMMADIR    := $(subst /,\\,$(shell /bin/cygpath -m -a "$(GAMMADIR)"))
   ABS_OS_MAKEFILE := $(shell /bin/cygpath -m -a "$(HS_MAKE_DIR)/$(OSNAME)")/build.make
+else ifeq ($(USING_MINGW), true)
+    ABS_OUTPUTDIR   := $(shell $(CD) $(OUTPUTDIR);$(PWD))
+    ABS_BOOTDIR     := $(shell $(CD) $(BOOTDIR);$(PWD))
+    ABS_GAMMADIR    := $(shell $(CD) $(GAMMADIR);$(PWD))
+    ABS_OS_MAKEFILE := $(shell $(CD) $(HS_MAKE_DIR)/$(OSNAME);$(PWD))/build.make
   else
     ABS_OUTPUTDIR   := $(subst /,\\,$(shell $(CD) $(OUTPUTDIR);$(PWD)))
     ABS_BOOTDIR     := $(subst /,\\,$(shell $(CD) $(BOOTDIR);$(PWD)))

hotspot/make/windows/makefiles/rules.make

@@ -23,14 +23,15 @@
 #
 
 # These are the commands used externally to compile and run.
-
+# The \ are used here for traditional Windows apps and " quoted to get
+# past the Unix-like shell:
 !ifdef BootStrapDir
-RUN_JAVA=$(BootStrapDir)\bin\java
+RUN_JAVA="$(BootStrapDir)\bin\java"
-RUN_JAVAP=$(BootStrapDir)\bin\javap
+RUN_JAVAP="$(BootStrapDir)\bin\javap"
-RUN_JAVAH=$(BootStrapDir)\bin\javah
+RUN_JAVAH="$(BootStrapDir)\bin\javah"
-RUN_JAR=$(BootStrapDir)\bin\jar
+RUN_JAR="$(BootStrapDir)\bin\jar"
-COMPILE_JAVAC=$(BootStrapDir)\bin\javac $(BOOTSTRAP_JAVAC_FLAGS)
+COMPILE_JAVAC="$(BootStrapDir)\bin\javac" $(BOOTSTRAP_JAVAC_FLAGS)
-COMPILE_RMIC=$(BootStrapDir)\bin\rmic
+COMPILE_RMIC="$(BootStrapDir)\bin\rmic"
 BOOT_JAVA_HOME=$(BootStrapDir)
 !else
 RUN_JAVA=java

hotspot/make/windows/makefiles/sa.make

@@ -36,37 +36,37 @@ checkAndBuildSA::
 !include $(WorkSpace)/make/windows/makefiles/rules.make
 !include $(WorkSpace)/make/sa.files
 
-GENERATED = ..\generated
+GENERATED = ../generated
 
 # tools.jar is needed by the JDI - SA binding
-SA_CLASSPATH = $(BOOT_JAVA_HOME)\lib\tools.jar
+SA_CLASSPATH = $(BOOT_JAVA_HOME)/lib/tools.jar
 
-SA_CLASSDIR = $(GENERATED)\saclasses
+SA_CLASSDIR = $(GENERATED)/saclasses
 
 SA_BUILD_VERSION_PROP = sun.jvm.hotspot.runtime.VM.saBuildVersion=$(SA_BUILD_VERSION)
 
-SA_PROPERTIES = $(SA_CLASSDIR)\sa.properties
+SA_PROPERTIES = $(SA_CLASSDIR)/sa.properties
 
-default::  $(GENERATED)\sa-jdi.jar
+default::  $(GENERATED)/sa-jdi.jar
 
 # Remove the space between $(SA_BUILD_VERSION_PROP) and > below as it adds a white space
 # at the end of SA version string and causes a version mismatch with the target VM version.
 
-$(GENERATED)\sa-jdi.jar: $(AGENT_FILES:/=\)
+$(GENERATED)/sa-jdi.jar: $(AGENT_FILES)
-	@if not exist $(SA_CLASSDIR) mkdir $(SA_CLASSDIR)
+	$(QUIETLY) mkdir -p $(SA_CLASSDIR)
-	@echo ...Building sa-jdi.jar
+	@echo ...Building sa-jdi.jar into $(SA_CLASSDIR)
 	@echo ...$(COMPILE_JAVAC) -classpath $(SA_CLASSPATH) -d $(SA_CLASSDIR) ....
-	@$(COMPILE_JAVAC) -classpath $(SA_CLASSPATH) -sourcepath $(AGENT_SRC_DIR) -d $(SA_CLASSDIR) $(AGENT_FILES:/=\)
+	@$(COMPILE_JAVAC) -classpath $(SA_CLASSPATH) -sourcepath $(AGENT_SRC_DIR) -d $(SA_CLASSDIR) $(AGENT_FILES)
 	$(COMPILE_RMIC) -classpath $(SA_CLASSDIR) -d $(SA_CLASSDIR) sun.jvm.hotspot.debugger.remote.RemoteDebuggerServer
 	$(QUIETLY) echo $(SA_BUILD_VERSION_PROP)> $(SA_PROPERTIES)
 	$(QUIETLY) rm -f $(SA_CLASSDIR)/sun/jvm/hotspot/utilities/soql/sa.js
 	$(QUIETLY) cp $(AGENT_SRC_DIR)/sun/jvm/hotspot/utilities/soql/sa.js $(SA_CLASSDIR)/sun/jvm/hotspot/utilities/soql
 	$(QUIETLY) rm -rf $(SA_CLASSDIR)/sun/jvm/hotspot/ui/resources
-	$(QUIETLY) mkdir $(SA_CLASSDIR)\sun\jvm\hotspot\ui\resources
+	$(QUIETLY) mkdir $(SA_CLASSDIR)/sun/jvm/hotspot/ui/resources
 	$(QUIETLY) cp $(AGENT_SRC_DIR)/sun/jvm/hotspot/ui/resources/*.png $(SA_CLASSDIR)/sun/jvm/hotspot/ui/resources
 	$(QUIETLY) cp -r $(AGENT_SRC_DIR)/images/* $(SA_CLASSDIR)
 	$(RUN_JAR) cf $@ -C $(SA_CLASSDIR) .
-	$(RUN_JAR) uf $@ -C $(AGENT_SRC_DIR:/=\) META-INF\services\com.sun.jdi.connect.Connector
+	$(RUN_JAR) uf $@ -C $(AGENT_SRC_DIR) META-INF/services/com.sun.jdi.connect.Connector
 	$(RUN_JAVAH) -classpath $(SA_CLASSDIR) -jni sun.jvm.hotspot.debugger.windbg.WindbgDebuggerLocal
 	$(RUN_JAVAH) -classpath $(SA_CLASSDIR) -jni sun.jvm.hotspot.debugger.x86.X86ThreadContext 
 	$(RUN_JAVAH) -classpath $(SA_CLASSDIR) -jni sun.jvm.hotspot.debugger.ia64.IA64ThreadContext 
@@ -85,27 +85,27 @@ checkAndBuildSA:: $(SAWINDBG)
 # will be useful to have the assertion checks in place
 
 !if "$(BUILDARCH)" == "ia64"
-SA_CFLAGS = /nologo $(MS_RUNTIME_OPTION) /W3 $(GX_OPTION) /Od /D "WIN32" /D "WIN64" /D "_WINDOWS" /D "_DEBUG" /D "_CONSOLE" /D "_MBCS" /YX /FD /c
+SA_CFLAGS = -nologo $(MS_RUNTIME_OPTION) -W3 $(GX_OPTION) -Od -D "WIN32" -D "WIN64" -D "_WINDOWS" -D "_DEBUG" -D "_CONSOLE" -D "_MBCS" -YX -FD -c
 !elseif "$(BUILDARCH)" == "amd64"
-SA_CFLAGS = /nologo $(MS_RUNTIME_OPTION) /W3 $(GX_OPTION) /Od /D "WIN32" /D "WIN64" /D "_WINDOWS" /D "_DEBUG" /D "_CONSOLE" /D "_MBCS" /YX /FD /c
+SA_CFLAGS = -nologo $(MS_RUNTIME_OPTION) -W3 $(GX_OPTION) -Od -D "WIN32" -D "WIN64" -D "_WINDOWS" -D "_DEBUG" -D "_CONSOLE" -D "_MBCS" -YX -FD -c
 !if "$(COMPILER_NAME)" == "VS2005"
 # On amd64, VS2005 compiler requires bufferoverflowU.lib on the link command line, 
 # otherwise we get missing __security_check_cookie externals at link time. 
 SA_LD_FLAGS = bufferoverflowU.lib
 !endif
 !else
-SA_CFLAGS = /nologo $(MS_RUNTIME_OPTION) /W3 /Gm $(GX_OPTION) /Od /D "WIN32" /D "_WINDOWS" /D "_DEBUG" /D "_CONSOLE" /D "_MBCS" /YX /FD /GZ /c
+SA_CFLAGS = -nologo $(MS_RUNTIME_OPTION) -W3 -Gm $(GX_OPTION) -Od -D "WIN32" -D "_WINDOWS" -D "_DEBUG" -D "_CONSOLE" -D "_MBCS" -YX -FD -GZ -c
 !if "$(ENABLE_FULL_DEBUG_SYMBOLS)" == "1"
-SA_CFLAGS = $(SA_CFLAGS) /ZI
+SA_CFLAGS = $(SA_CFLAGS) -ZI
 !endif
 !endif
 !if "$(MT)" != ""
-SA_LD_FLAGS = /manifest $(SA_LD_FLAGS)
+SA_LD_FLAGS = -manifest $(SA_LD_FLAGS)
 !endif
 SASRCFILE = $(AGENT_DIR)/src/os/win32/windbg/sawindbg.cpp
-SA_LFLAGS = $(SA_LD_FLAGS) /nologo /subsystem:console /machine:$(MACHINE)
+SA_LFLAGS = $(SA_LD_FLAGS) -nologo -subsystem:console -machine:$(MACHINE)
 !if "$(ENABLE_FULL_DEBUG_SYMBOLS)" == "1"
-SA_LFLAGS = $(SA_LFLAGS) /map /debug
+SA_LFLAGS = $(SA_LFLAGS) -map -debug
 !endif
 
 # Note that we do not keep sawindbj.obj around as it would then
@@ -117,15 +117,15 @@ SA_LFLAGS = $(SA_LFLAGS) /map /debug
 $(SAWINDBG): $(SASRCFILE)
 	set INCLUDE=$(SA_INCLUDE)$(INCLUDE)
 	$(CXX) @<<
-	  /I"$(BootStrapDir)/include" /I"$(BootStrapDir)/include/win32" 
+	  -I"$(BootStrapDir)/include" -I"$(BootStrapDir)/include/win32" 
-	  /I"$(GENERATED)" $(SA_CFLAGS)
+	  -I"$(GENERATED)" $(SA_CFLAGS)
 	  $(SASRCFILE)
-	  /out:$*.obj
+	  -out:$*.obj
 <<
 	set LIB=$(SA_LIB)$(LIB)
-	$(LD) /out:$@ /DLL $*.obj dbgeng.lib $(SA_LFLAGS)
+	$(LD) -out:$@ -DLL $*.obj dbgeng.lib $(SA_LFLAGS)
 !if "$(MT)" != ""
-	$(MT) /manifest $(@F).manifest /outputresource:$(@F);#2
+	$(MT) -manifest $(@F).manifest -outputresource:$(@F);#2
 !endif
 !if "$(ENABLE_FULL_DEBUG_SYMBOLS)" == "1"
 !if "$(ZIP_DEBUGINFO_FILES)" == "1"
@@ -136,6 +136,6 @@ $(SAWINDBG): $(SASRCFILE)
 	-@rm -f $*.obj
 
 cleanall :
-	rm -rf $(GENERATED:\=/)/saclasses
+	rm -rf $(GENERATED)/saclasses
-	rm -rf $(GENERATED:\=/)/sa-jdi.jar
+	rm -rf $(GENERATED)/sa-jdi.jar
 !endif

hotspot/make/windows/makefiles/shared.make

@@ -36,11 +36,12 @@ CXX=cl.exe
 
 
 !ifdef SUBDIRS
+# \ is used below because $(MAKE) is nmake here, which expects Windows paths
 $(SUBDIRS): FORCE
 	@if not exist $@ mkdir $@
-	@if not exist $@\local.make echo # Empty > $@\local.make
+	@if not exist $@/local.make echo # Empty > $@/local.make
-	@echo nmake $(ACTION) in $(DIR)\$@
+	@echo nmake $(ACTION) in $(DIR)/$@
-	cd $@ && $(MAKE) /NOLOGO /f $(WorkSpace)\make\windows\makefiles\$@.make $(ACTION) DIR=$(DIR)\$@ BUILD_FLAVOR=$(BUILD_FLAVOR)
+	cd $@ && $(MAKE) -NOLOGO -f $(WorkSpace)\make\windows\makefiles\$@.make $(ACTION) DIR=$(DIR)\$@ BUILD_FLAVOR=$(BUILD_FLAVOR)
 !endif
 
 # Creates the needed directory

hotspot/make/windows/projectfiles/common/Makefile

@@ -108,7 +108,7 @@ ProjectCreatorIDEOptions =       $(ProjectCreatorIDEOptions) \
       -define              HOTSPOT_VM_DISTRO=\\\"$(HOTSPOT_VM_DISTRO)\\\"
 
 $(HOTSPOTBUILDSPACE)/$(ProjectFile): $(HOTSPOTBUILDSPACE)/classes/ProjectCreator.class
-	@$(RUN_JAVA) -Djava.class.path=$(HOTSPOTBUILDSPACE)/classes ProjectCreator WinGammaPlatform$(VcVersion) $(ProjectCreatorIDEOptions)
+	@$(RUN_JAVA) -Djava.class.path="$(HOTSPOTBUILDSPACE)/classes" ProjectCreator WinGammaPlatform$(VcVersion) $(ProjectCreatorIDEOptions)
 
 clean:
 	@rm -rf $(HOTSPOTBUILDSPACE)/classes

hotspot/src/cpu/sparc/vm/assembler_sparc.cpp

@@ -44,8 +44,10 @@
 
 #ifdef PRODUCT
 #define BLOCK_COMMENT(str) /* nothing */
+#define STOP(error) stop(error)
 #else
 #define BLOCK_COMMENT(str) block_comment(str)
+#define STOP(error) block_comment(error); stop(error)
 #endif
 
 // Convert the raw encoding form into the form expected by the
@@ -992,7 +994,7 @@ void MacroAssembler::set_last_Java_frame(Register last_java_sp, Register last_Ja
   save_frame(0);                // to avoid clobbering O0
   ld_ptr(pc_addr, L0);
   br_null_short(L0, Assembler::pt, PcOk);
-  stop("last_Java_pc not zeroed before leaving Java");
+  STOP("last_Java_pc not zeroed before leaving Java");
   bind(PcOk);
 
   // Verify that flags was zeroed on return to Java
@@ -1001,7 +1003,7 @@ void MacroAssembler::set_last_Java_frame(Register last_java_sp, Register last_Ja
   tst(L0);
   br(Assembler::zero, false, Assembler::pt, FlagsOk);
   delayed() -> restore();
-  stop("flags not zeroed before leaving Java");
+  STOP("flags not zeroed before leaving Java");
   bind(FlagsOk);
 #endif /* ASSERT */
   //
@@ -1021,7 +1023,7 @@ void MacroAssembler::set_last_Java_frame(Register last_java_sp, Register last_Ja
   andcc(last_java_sp, 0x01, G0);
   br(Assembler::notZero, false, Assembler::pt, StackOk);
   delayed()->nop();
-  stop("Stack Not Biased in set_last_Java_frame");
+  STOP("Stack Not Biased in set_last_Java_frame");
   bind(StackOk);
 #endif // ASSERT
   assert( last_java_sp != G4_scratch, "bad register usage in set_last_Java_frame");
@@ -1650,23 +1652,28 @@ void MacroAssembler::safepoint() {
 
 
 void RegistersForDebugging::print(outputStream* s) {
+  FlagSetting fs(Debugging, true);
   int j;
-  for ( j = 0;  j < 8;  ++j )
+  for (j = 0; j < 8; ++j) {
-    if ( j != 6 ) s->print_cr("i%d = 0x%.16lx", j, i[j]);
+    if (j != 6) { s->print("i%d = ", j); os::print_location(s, i[j]); }
-    else          s->print_cr( "fp = 0x%.16lx",    i[j]);
+    else        { s->print( "fp = "   ); os::print_location(s, i[j]); }
+  }
   s->cr();
 
-  for ( j = 0;  j < 8;  ++j )
+  for (j = 0;  j < 8;  ++j) {
-    s->print_cr("l%d = 0x%.16lx", j, l[j]);
+    s->print("l%d = ", j); os::print_location(s, l[j]);
+  }
   s->cr();
 
-  for ( j = 0;  j < 8;  ++j )
+  for (j = 0; j < 8; ++j) {
-    if ( j != 6 ) s->print_cr("o%d = 0x%.16lx", j, o[j]);
+    if (j != 6) { s->print("o%d = ", j); os::print_location(s, o[j]); }
-    else          s->print_cr( "sp = 0x%.16lx",    o[j]);
+    else        { s->print( "sp = "   ); os::print_location(s, o[j]); }
+  }
   s->cr();
 
-  for ( j = 0;  j < 8;  ++j )
+  for (j = 0; j < 8; ++j) {
-    s->print_cr("g%d = 0x%.16lx", j, g[j]);
+    s->print("g%d = ", j); os::print_location(s, g[j]);
+  }
   s->cr();
 
   // print out floats with compression
@@ -2020,8 +2027,8 @@ void MacroAssembler::untested(const char* what) {
   char* b = new char[1024];
   sprintf(b, "untested: %s", what);
 
-  if ( ShowMessageBoxOnError )   stop(b);
+  if (ShowMessageBoxOnError) { STOP(b); }
-  else                           warn(b);
+  else                       { warn(b); }
 }
 
 
@@ -2998,26 +3005,60 @@ void MacroAssembler::lookup_interface_method(Register recv_klass,
 }
 
 
+// virtual method calling
+void MacroAssembler::lookup_virtual_method(Register recv_klass,
+                                           RegisterOrConstant vtable_index,
+                                           Register method_result) {
+  assert_different_registers(recv_klass, method_result, vtable_index.register_or_noreg());
+  Register sethi_temp = method_result;
+  const int base = (instanceKlass::vtable_start_offset() * wordSize +
+                    // method pointer offset within the vtable entry:
+                    vtableEntry::method_offset_in_bytes());
+  RegisterOrConstant vtable_offset = vtable_index;
+  // Each of the following three lines potentially generates an instruction.
+  // But the total number of address formation instructions will always be
+  // at most two, and will often be zero.  In any case, it will be optimal.
+  // If vtable_index is a register, we will have (sll_ptr N,x; inc_ptr B,x; ld_ptr k,x).
+  // If vtable_index is a constant, we will have at most (set B+X<<N,t; ld_ptr k,t).
+  vtable_offset = regcon_sll_ptr(vtable_index, exact_log2(vtableEntry::size() * wordSize), vtable_offset);
+  vtable_offset = regcon_inc_ptr(vtable_offset, base, vtable_offset, sethi_temp);
+  Address vtable_entry_addr(recv_klass, ensure_simm13_or_reg(vtable_offset, sethi_temp));
+  ld_ptr(vtable_entry_addr, method_result);
+}
+
+
 void MacroAssembler::check_klass_subtype(Register sub_klass,
                                          Register super_klass,
                                          Register temp_reg,
                                          Register temp2_reg,
                                          Label& L_success) {
-  Label L_failure, L_pop_to_failure;
-  check_klass_subtype_fast_path(sub_klass, super_klass,
-                                temp_reg, temp2_reg,
-                                &L_success, &L_failure, NULL);
   Register sub_2 = sub_klass;
   Register sup_2 = super_klass;
   if (!sub_2->is_global())  sub_2 = L0;
   if (!sup_2->is_global())  sup_2 = L1;
+  bool did_save = false;
+  if (temp_reg == noreg || temp2_reg == noreg) {
+    temp_reg = L2;
+    temp2_reg = L3;
+    save_frame_and_mov(0, sub_klass, sub_2, super_klass, sup_2);
+    sub_klass = sub_2;
+    super_klass = sup_2;
+    did_save = true;
+  }
+  Label L_failure, L_pop_to_failure, L_pop_to_success;
+  check_klass_subtype_fast_path(sub_klass, super_klass,
+                                temp_reg, temp2_reg,
+                                (did_save ? &L_pop_to_success : &L_success),
+                                (did_save ? &L_pop_to_failure : &L_failure), NULL);
 
+  if (!did_save)
     save_frame_and_mov(0, sub_klass, sub_2, super_klass, sup_2);
   check_klass_subtype_slow_path(sub_2, sup_2,
                                 L2, L3, L4, L5,
                                 NULL, &L_pop_to_failure);
 
   // on success:
+  bind(L_pop_to_success);
   restore();
   ba_short(L_success);
 
@@ -3234,54 +3275,6 @@ void MacroAssembler::check_klass_subtype_slow_path(Register sub_klass,
 }
 
 
-void MacroAssembler::check_method_handle_type(Register mtype_reg, Register mh_reg,
-                                              Register temp_reg,
-                                              Label& wrong_method_type) {
-  assert_different_registers(mtype_reg, mh_reg, temp_reg);
-  // compare method type against that of the receiver
-  RegisterOrConstant mhtype_offset = delayed_value(java_lang_invoke_MethodHandle::type_offset_in_bytes, temp_reg);
-  load_heap_oop(mh_reg, mhtype_offset, temp_reg);
-  cmp_and_brx_short(temp_reg, mtype_reg, Assembler::notEqual, Assembler::pn, wrong_method_type);
-}
-
-
-// A method handle has a "vmslots" field which gives the size of its
-// argument list in JVM stack slots.  This field is either located directly
-// in every method handle, or else is indirectly accessed through the
-// method handle's MethodType.  This macro hides the distinction.
-void MacroAssembler::load_method_handle_vmslots(Register vmslots_reg, Register mh_reg,
-                                                Register temp_reg) {
-  assert_different_registers(vmslots_reg, mh_reg, temp_reg);
-  // load mh.type.form.vmslots
-  Register temp2_reg = vmslots_reg;
-  load_heap_oop(Address(mh_reg,    delayed_value(java_lang_invoke_MethodHandle::type_offset_in_bytes, temp_reg)),      temp2_reg);
-  load_heap_oop(Address(temp2_reg, delayed_value(java_lang_invoke_MethodType::form_offset_in_bytes, temp_reg)),        temp2_reg);
-  ld(           Address(temp2_reg, delayed_value(java_lang_invoke_MethodTypeForm::vmslots_offset_in_bytes, temp_reg)), vmslots_reg);
-}
-
-
-void MacroAssembler::jump_to_method_handle_entry(Register mh_reg, Register temp_reg, bool emit_delayed_nop) {
-  assert(mh_reg == G3_method_handle, "caller must put MH object in G3");
-  assert_different_registers(mh_reg, temp_reg);
-
-  // pick out the interpreted side of the handler
-  // NOTE: vmentry is not an oop!
-  ld_ptr(mh_reg, delayed_value(java_lang_invoke_MethodHandle::vmentry_offset_in_bytes, temp_reg), temp_reg);
-
-  // off we go...
-  ld_ptr(temp_reg, MethodHandleEntry::from_interpreted_entry_offset_in_bytes(), temp_reg);
-  jmp(temp_reg, 0);
-
-  // for the various stubs which take control at this point,
-  // see MethodHandles::generate_method_handle_stub
-
-  // Some callers can fill the delay slot.
-  if (emit_delayed_nop) {
-    delayed()->nop();
-  }
-}
-
-
 RegisterOrConstant MacroAssembler::argument_offset(RegisterOrConstant arg_slot,
                                                    Register temp_reg,
                                                    int extra_slot_offset) {
@@ -3914,7 +3907,7 @@ void MacroAssembler::verify_tlab() {
     ld_ptr(G2_thread, in_bytes(JavaThread::tlab_start_offset()), t2);
     or3(t1, t2, t3);
     cmp_and_br_short(t1, t2, Assembler::greaterEqual, Assembler::pn, next);
-    stop("assert(top >= start)");
+    STOP("assert(top >= start)");
     should_not_reach_here();
 
     bind(next);
@@ -3922,13 +3915,13 @@ void MacroAssembler::verify_tlab() {
     ld_ptr(G2_thread, in_bytes(JavaThread::tlab_end_offset()), t2);
     or3(t3, t2, t3);
     cmp_and_br_short(t1, t2, Assembler::lessEqual, Assembler::pn, next2);
-    stop("assert(top <= end)");
+    STOP("assert(top <= end)");
     should_not_reach_here();
 
     bind(next2);
     and3(t3, MinObjAlignmentInBytesMask, t3);
     cmp_and_br_short(t3, 0, Assembler::lessEqual, Assembler::pn, ok);
-    stop("assert(aligned)");
+    STOP("assert(aligned)");
     should_not_reach_here();
 
     bind(ok);
@@ -3976,7 +3969,7 @@ void MacroAssembler::eden_allocate(
       btst(MinObjAlignmentInBytesMask, obj);
       br(Assembler::zero, false, Assembler::pt, L);
       delayed()->nop();
-      stop("eden top is not properly aligned");
+      STOP("eden top is not properly aligned");
       bind(L);
     }
 #endif // ASSERT
@@ -4013,7 +4006,7 @@ void MacroAssembler::eden_allocate(
       btst(MinObjAlignmentInBytesMask, top_addr);
       br(Assembler::zero, false, Assembler::pt, L);
       delayed()->nop();
-      stop("eden top is not properly aligned");
+      STOP("eden top is not properly aligned");
       bind(L);
     }
 #endif // ASSERT
@@ -4066,7 +4059,7 @@ void MacroAssembler::tlab_allocate(
     btst(MinObjAlignmentInBytesMask, free);
     br(Assembler::zero, false, Assembler::pt, L);
     delayed()->nop();
-    stop("updated TLAB free is not properly aligned");
+    STOP("updated TLAB free is not properly aligned");
     bind(L);
   }
 #endif // ASSERT
@@ -4164,7 +4157,7 @@ void MacroAssembler::tlab_refill(Label& retry, Label& try_eden, Label& slow_case
     ld_ptr(G2_thread, in_bytes(JavaThread::tlab_size_offset()), t2);
     sll_ptr(t2, LogHeapWordSize, t2);
     cmp_and_br_short(t1, t2, Assembler::equal, Assembler::pt, ok);
-    stop("assert(t1 == tlab_size)");
+    STOP("assert(t1 == tlab_size)");
     should_not_reach_here();
 
     bind(ok);

hotspot/src/cpu/sparc/vm/assembler_sparc.hpp

@@ -2538,6 +2538,11 @@ public:
                                Register temp_reg, Register temp2_reg,
                                Label& no_such_interface);
 
+  // virtual method calling
+  void lookup_virtual_method(Register recv_klass,
+                             RegisterOrConstant vtable_index,
+                             Register method_result);
+
   // Test sub_klass against super_klass, with fast and slow paths.
 
   // The fast path produces a tri-state answer: yes / no / maybe-slow.
@@ -2577,12 +2582,6 @@ public:
                            Label& L_success);
 
   // method handles (JSR 292)
-  void check_method_handle_type(Register mtype_reg, Register mh_reg,
-                                Register temp_reg,
-                                Label& wrong_method_type);
-  void load_method_handle_vmslots(Register vmslots_reg, Register mh_reg,
-                                  Register temp_reg);
-  void jump_to_method_handle_entry(Register mh_reg, Register temp_reg, bool emit_delayed_nop = true);
   // offset relative to Gargs of argument at tos[arg_slot].
   // (arg_slot == 0 means the last argument, not the first).
   RegisterOrConstant argument_offset(RegisterOrConstant arg_slot,
@@ -2590,7 +2589,7 @@ public:
                                      int extra_slot_offset = 0);
   // Address of Gargs and argument_offset.
   Address            argument_address(RegisterOrConstant arg_slot,
-                                      Register temp_reg,
+                                      Register temp_reg = noreg,
                                       int extra_slot_offset = 0);
 
   // Stack overflow checking

hotspot/src/cpu/sparc/vm/c1_CodeStubs_sparc.cpp

@@ -435,85 +435,6 @@ void G1PreBarrierStub::emit_code(LIR_Assembler* ce) {
 
 }
 
-void G1UnsafeGetObjSATBBarrierStub::emit_code(LIR_Assembler* ce) {
-  // At this point we know that offset == referent_offset.
-  //
-  // So we might have to emit:
-  //   if (src == null) goto continuation.
-  //
-  // and we definitely have to emit:
-  //   if (klass(src).reference_type == REF_NONE) goto continuation
-  //   if (!marking_active) goto continuation
-  //   if (pre_val == null) goto continuation
-  //   call pre_barrier(pre_val)
-  //   goto continuation
-  //
-  __ bind(_entry);
-
-  assert(src()->is_register(), "sanity");
-  Register src_reg = src()->as_register();
-
-  if (gen_src_check()) {
-    // The original src operand was not a constant.
-    // Generate src == null?
-    if (__ is_in_wdisp16_range(_continuation)) {
-      __ br_null(src_reg, /*annul*/false, Assembler::pt, _continuation);
-    } else {
-      __ cmp(src_reg, G0);
-      __ brx(Assembler::equal, false, Assembler::pt, _continuation);
-    }
-    __ delayed()->nop();
-  }
-
-  // Generate src->_klass->_reference_type() == REF_NONE)?
-  assert(tmp()->is_register(), "sanity");
-  Register tmp_reg = tmp()->as_register();
-
-  __ load_klass(src_reg, tmp_reg);
-
-  Address ref_type_adr(tmp_reg, instanceKlass::reference_type_offset());
-  __ ldub(ref_type_adr, tmp_reg);
-
-  // _reference_type field is of type ReferenceType (enum)
-  assert(REF_NONE == 0, "check this code");
-  __ cmp_zero_and_br(Assembler::equal, tmp_reg, _continuation, /*annul*/false, Assembler::pt);
-  __ delayed()->nop();
-
-  // Is marking active?
-  assert(thread()->is_register(), "precondition");
-  Register thread_reg = thread()->as_pointer_register();
-
-  Address in_progress(thread_reg, in_bytes(JavaThread::satb_mark_queue_offset() +
-                                       PtrQueue::byte_offset_of_active()));
-
-  if (in_bytes(PtrQueue::byte_width_of_active()) == 4) {
-    __ ld(in_progress, tmp_reg);
-  } else {
-    assert(in_bytes(PtrQueue::byte_width_of_active()) == 1, "Assumption");
-    __ ldsb(in_progress, tmp_reg);
-  }
-
-  __ cmp_zero_and_br(Assembler::equal, tmp_reg, _continuation, /*annul*/false, Assembler::pt);
-  __ delayed()->nop();
-
-  // val == null?
-  assert(val()->is_register(), "Precondition.");
-  Register val_reg = val()->as_register();
-
-  if (__ is_in_wdisp16_range(_continuation)) {
-    __ br_null(val_reg, /*annul*/false, Assembler::pt, _continuation);
-  } else {
-    __ cmp(val_reg, G0);
-    __ brx(Assembler::equal, false, Assembler::pt, _continuation);
-  }
-  __ delayed()->nop();
-
-  __ call(Runtime1::entry_for(Runtime1::Runtime1::g1_pre_barrier_slow_id));
-  __ delayed()->mov(val_reg, G4);
-  __ br(Assembler::always, false, Assembler::pt, _continuation);
-  __ delayed()->nop();
-}
-
 jbyte* G1PostBarrierStub::_byte_map_base = NULL;
 
 jbyte* G1PostBarrierStub::byte_map_base_slow() {

hotspot/src/cpu/sparc/vm/c1_LIRAssembler_sparc.cpp

@@ -2956,6 +2956,7 @@ void LIR_Assembler::emit_lock(LIR_OpLock* op) {
 void LIR_Assembler::emit_profile_call(LIR_OpProfileCall* op) {
   ciMethod* method = op->profiled_method();
   int bci          = op->profiled_bci();
+  ciMethod* callee = op->profiled_callee();
 
   // Update counter for all call types
   ciMethodData* md = method->method_data_or_null();
@@ -2984,9 +2985,11 @@ void LIR_Assembler::emit_profile_call(LIR_OpProfileCall* op) {
 
   Address counter_addr(mdo, md->byte_offset_of_slot(data, CounterData::count_offset()) - mdo_offset_bias);
   Bytecodes::Code bc = method->java_code_at_bci(bci);
+  const bool callee_is_static = callee->is_loaded() && callee->is_static();
   // Perform additional virtual call profiling for invokevirtual and
   // invokeinterface bytecodes
   if ((bc == Bytecodes::_invokevirtual || bc == Bytecodes::_invokeinterface) &&
+      !callee_is_static &&  // required for optimized MH invokes
       C1ProfileVirtualCalls) {
     assert(op->recv()->is_single_cpu(), "recv must be allocated");
     Register recv = op->recv()->as_register();

hotspot/src/cpu/sparc/vm/cppInterpreter_sparc.cpp

@@ -515,9 +515,9 @@ address InterpreterGenerator::generate_accessor_entry(void) {
     // Need to differentiate between igetfield, agetfield, bgetfield etc.
     // because they are different sizes.
     // Get the type from the constant pool cache
-    __ srl(G1_scratch, ConstantPoolCacheEntry::tosBits, G1_scratch);
+    __ srl(G1_scratch, ConstantPoolCacheEntry::tos_state_shift, G1_scratch);
-    // Make sure we don't need to mask G1_scratch for tosBits after the above shift
+    // Make sure we don't need to mask G1_scratch after the above shift
-    ConstantPoolCacheEntry::verify_tosBits();
+    ConstantPoolCacheEntry::verify_tos_state_shift();
     __ cmp(G1_scratch, atos );
     __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
     __ delayed()->ld_ptr(Otos_i, G3_scratch, Otos_i);

hotspot/src/cpu/sparc/vm/frame_sparc.cpp

@@ -514,7 +514,6 @@ frame frame::sender(RegisterMap* map) const {
   // interpreted but its pc is in the code cache (for c1 -> osr_frame_return_id stub), so it must be
   // explicitly recognized.
 
-  if (is_ricochet_frame())    return sender_for_ricochet_frame(map);
 
   bool frame_is_interpreted = is_interpreted_frame();
   if (frame_is_interpreted) {
@@ -821,9 +820,7 @@ void frame::describe_pd(FrameValues& values, int frame_no) {
     values.describe(frame_no, sp() + w, err_msg("register save area word %d", w), 1);
   }
 
-  if (is_ricochet_frame()) {
+  if (is_interpreted_frame()) {
-    MethodHandles::RicochetFrame::describe(this, values, frame_no);
-  } else if (is_interpreted_frame()) {
     DESCRIBE_FP_OFFSET(interpreter_frame_d_scratch_fp);
     DESCRIBE_FP_OFFSET(interpreter_frame_l_scratch_fp);
     DESCRIBE_FP_OFFSET(interpreter_frame_padding);

hotspot/src/cpu/sparc/vm/globals_sparc.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2000, 2012, 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
@@ -75,4 +75,43 @@ define_pd_global(bool, UseMembar,            false);
 
 // GC Ergo Flags
 define_pd_global(intx, CMSYoungGenPerWorker, 16*M);  // default max size of CMS young gen, per GC worker thread
+
+#define ARCH_FLAGS(develop, product, diagnostic, experimental, notproduct) \
+                                                                            \
+  product(intx, UseVIS, 99,                                                 \
+          "Highest supported VIS instructions set on Sparc")                \
+                                                                            \
+  product(bool, UseCBCond, false,                                           \
+          "Use compare and branch instruction on SPARC")                    \
+                                                                            \
+  product(bool, UseBlockZeroing, false,                                     \
+          "Use special cpu instructions for block zeroing")                 \
+                                                                            \
+  product(intx, BlockZeroingLowLimit, 2048,                                 \
+          "Minimum size in bytes when block zeroing will be used")          \
+                                                                            \
+  product(bool, UseBlockCopy, false,                                        \
+          "Use special cpu instructions for block copy")                    \
+                                                                            \
+  product(intx, BlockCopyLowLimit, 2048,                                    \
+          "Minimum size in bytes when block copy will be used")             \
+                                                                            \
+  develop(bool, UseV8InstrsOnly, false,                                     \
+          "Use SPARC-V8 Compliant instruction subset")                      \
+                                                                            \
+  product(bool, UseNiagaraInstrs, false,                                    \
+          "Use Niagara-efficient instruction subset")                       \
+                                                                            \
+  develop(bool, UseCASForSwap, false,                                       \
+          "Do not use swap instructions, but only CAS (in a loop) on SPARC")\
+                                                                            \
+  product(uintx,  ArraycopySrcPrefetchDistance, 0,                          \
+          "Distance to prefetch source array in arracopy")                  \
+                                                                            \
+  product(uintx,  ArraycopyDstPrefetchDistance, 0,                          \
+          "Distance to prefetch destination array in arracopy")             \
+                                                                            \
+  develop(intx, V8AtomicOperationUnderLockSpinCount,    50,                 \
+          "Number of times to spin wait on a v8 atomic operation lock")     \
+
 #endif // CPU_SPARC_VM_GLOBALS_SPARC_HPP

hotspot/src/cpu/sparc/vm/interp_masm_sparc.cpp

@@ -505,7 +505,7 @@ void InterpreterMacroAssembler::store_ptr(int n, Register val) {
 void InterpreterMacroAssembler::load_receiver(Register param_count,
                                               Register recv) {
   sll(param_count, Interpreter::logStackElementSize, param_count);
-  ld_ptr(Lesp, param_count, recv);                      // gets receiver Oop
+  ld_ptr(Lesp, param_count, recv);  // gets receiver oop
 }
 
 void InterpreterMacroAssembler::empty_expression_stack() {
@@ -767,8 +767,12 @@ void InterpreterMacroAssembler::get_cache_and_index_and_bytecode_at_bcp(Register
   get_cache_and_index_at_bcp(cache, temp, bcp_offset, index_size);
   ld_ptr(cache, constantPoolCacheOopDesc::base_offset() + ConstantPoolCacheEntry::indices_offset(), bytecode);
   const int shift_count = (1 + byte_no) * BitsPerByte;
+  assert((byte_no == TemplateTable::f1_byte && shift_count == ConstantPoolCacheEntry::bytecode_1_shift) ||
+         (byte_no == TemplateTable::f2_byte && shift_count == ConstantPoolCacheEntry::bytecode_2_shift),
+         "correct shift count");
   srl(bytecode, shift_count, bytecode);
-  and3(bytecode,        0xFF, bytecode);
+  assert(ConstantPoolCacheEntry::bytecode_1_mask == ConstantPoolCacheEntry::bytecode_2_mask, "common mask");
+  and3(bytecode, ConstantPoolCacheEntry::bytecode_1_mask, bytecode);
 }
 
 

hotspot/src/cpu/sparc/vm/interpreterGenerator_sparc.hpp

@@ -32,7 +32,6 @@
   address generate_normal_entry(bool synchronized);
   address generate_native_entry(bool synchronized);
   address generate_abstract_entry(void);
-  address generate_method_handle_entry(void);
   address generate_math_entry(AbstractInterpreter::MethodKind kind);
   address generate_empty_entry(void);
   address generate_accessor_entry(void);

hotspot/src/cpu/sparc/vm/interpreter_sparc.cpp

@@ -255,17 +255,6 @@ address InterpreterGenerator::generate_abstract_entry(void) {
 }
 
 
-// Method handle invoker
-// Dispatch a method of the form java.lang.invoke.MethodHandles::invoke(...)
-address InterpreterGenerator::generate_method_handle_entry(void) {
-  if (!EnableInvokeDynamic) {
-    return generate_abstract_entry();
-  }
-
-  return MethodHandles::generate_method_handle_interpreter_entry(_masm);
-}
-
-
 //----------------------------------------------------------------------------------------------------
 // Entry points & stack frame layout
 //
@@ -395,7 +384,7 @@ address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter:
     case Interpreter::empty                  : entry_point = ((InterpreterGenerator*)this)->generate_empty_entry();        break;
     case Interpreter::accessor               : entry_point = ((InterpreterGenerator*)this)->generate_accessor_entry();     break;
     case Interpreter::abstract               : entry_point = ((InterpreterGenerator*)this)->generate_abstract_entry();     break;
-    case Interpreter::method_handle          : entry_point = ((InterpreterGenerator*)this)->generate_method_handle_entry(); break;
+
     case Interpreter::java_lang_math_sin     :                                                                             break;
     case Interpreter::java_lang_math_cos     :                                                                             break;
     case Interpreter::java_lang_math_tan     :                                                                             break;
@@ -407,7 +396,9 @@ address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter:
     case Interpreter::java_lang_math_exp     :                                                                             break;
     case Interpreter::java_lang_ref_reference_get
                                              : entry_point = ((InterpreterGenerator*)this)->generate_Reference_get_entry(); break;
-    default                                  : ShouldNotReachHere();                                                       break;
+    default:
+      fatal(err_msg("unexpected method kind: %d", kind));
+      break;
   }
 
   if (entry_point) return entry_point;

hotspot/src/cpu/sparc/vm/methodHandles_sparc.hpp

@@ -30,186 +30,9 @@ enum /* platform_dependent_constants */ {
   adapter_code_size = NOT_LP64(23000 DEBUG_ONLY(+ 40000)) LP64_ONLY(35000 DEBUG_ONLY(+ 50000))
 };
 
-public:
-
-class RicochetFrame : public ResourceObj {
-  friend class MethodHandles;
-
- private:
-  /*
-    RF field            x86                 SPARC
-    sender_pc           *(rsp+0)            I7-0x8
-    sender_link         rbp                 I6+BIAS
-    exact_sender_sp     rsi/r13             I5_savedSP
-    conversion          *(rcx+&amh_conv)    L5_conv
-    saved_args_base     rax                 L4_sab (cf. Gargs = G4)
-    saved_args_layout   #NULL               L3_sal
-    saved_target        *(rcx+&mh_vmtgt)    L2_stgt
-    continuation        #STUB_CON           L1_cont
-   */
-  static const Register L1_continuation     ;  // what to do when control gets back here
-  static const Register L2_saved_target     ;  // target method handle to invoke on saved_args
-  static const Register L3_saved_args_layout;  // caching point for MethodTypeForm.vmlayout cookie
-  static const Register L4_saved_args_base  ;  // base of pushed arguments (slot 0, arg N) (-3)
-  static const Register L5_conversion       ;  // misc. information from original AdapterMethodHandle (-2)
-
-  frame _fr;
-
-  RicochetFrame(const frame& fr) : _fr(fr) { }
-
-  intptr_t* register_addr(Register reg) const  {
-    assert((_fr.sp() + reg->sp_offset_in_saved_window()) == _fr.register_addr(reg), "must agree");
-    return _fr.register_addr(reg);
-  }
-  intptr_t  register_value(Register reg) const { return *register_addr(reg); }
-
- public:
-  intptr_t* continuation() const        { return (intptr_t*) register_value(L1_continuation); }
-  oop       saved_target() const        { return (oop)       register_value(L2_saved_target); }
-  oop       saved_args_layout() const   { return (oop)       register_value(L3_saved_args_layout); }
-  intptr_t* saved_args_base() const     { return (intptr_t*) register_value(L4_saved_args_base); }
-  intptr_t  conversion() const          { return             register_value(L5_conversion); }
-  intptr_t* exact_sender_sp() const     { return (intptr_t*) register_value(I5_savedSP); }
-  intptr_t* sender_link() const         { return _fr.sender_sp(); }  // XXX
-  address   sender_pc() const           { return _fr.sender_pc(); }
-
-  // This value is not used for much, but it apparently must be nonzero.
-  static int frame_size_in_bytes()              { return wordSize * 4; }
-
-  intptr_t* extended_sender_sp() const  { return saved_args_base(); }
-
-  intptr_t  return_value_slot_number() const {
-    return adapter_conversion_vminfo(conversion());
-  }
-  BasicType return_value_type() const {
-    return adapter_conversion_dest_type(conversion());
-  }
-  bool has_return_value_slot() const {
-    return return_value_type() != T_VOID;
-  }
-  intptr_t* return_value_slot_addr() const {
-    assert(has_return_value_slot(), "");
-    return saved_arg_slot_addr(return_value_slot_number());
-  }
-  intptr_t* saved_target_slot_addr() const {
-    return saved_arg_slot_addr(saved_args_length());
-  }
-  intptr_t* saved_arg_slot_addr(int slot) const {
-    assert(slot >= 0, "");
-    return (intptr_t*)( (address)saved_args_base() + (slot * Interpreter::stackElementSize) );
-  }
-
-  jint      saved_args_length() const;
-  jint      saved_arg_offset(int arg) const;
-
-  // GC interface
-  oop*  saved_target_addr()                     { return (oop*)register_addr(L2_saved_target); }
-  oop*  saved_args_layout_addr()                { return (oop*)register_addr(L3_saved_args_layout); }
-
-  oop  compute_saved_args_layout(bool read_cache, bool write_cache);
-
-#ifdef ASSERT
-  // The magic number is supposed to help find ricochet frames within the bytes of stack dumps.
-  enum { MAGIC_NUMBER_1 = 0xFEED03E, MAGIC_NUMBER_2 = 0xBEEF03E };
-  static const Register L0_magic_number_1   ;  // cookie for debugging, at start of RSA
-  static Address magic_number_2_addr()  { return Address(L4_saved_args_base, -wordSize); }
-  intptr_t magic_number_1() const       { return register_value(L0_magic_number_1); }
-  intptr_t magic_number_2() const       { return saved_args_base()[-1]; }
-#endif //ASSERT
-
- public:
-  enum { RETURN_VALUE_PLACEHOLDER = (NOT_DEBUG(0) DEBUG_ONLY(42)) };
-
-  void verify() const NOT_DEBUG_RETURN; // check for MAGIC_NUMBER, etc.
-
-  static void generate_ricochet_blob(MacroAssembler* _masm,
-                                     // output params:
-                                     int* bounce_offset,
-                                     int* exception_offset,
-                                     int* frame_size_in_words);
-
-  static void enter_ricochet_frame(MacroAssembler* _masm,
-                                   Register recv_reg,
-                                   Register argv_reg,
-                                   address return_handler);
-
-  static void leave_ricochet_frame(MacroAssembler* _masm,
-                                   Register recv_reg,
-                                   Register new_sp_reg,
-                                   Register sender_pc_reg);
-
-  static RicochetFrame* from_frame(const frame& fr) {
-    RicochetFrame* rf = new RicochetFrame(fr);
-    rf->verify();
-    return rf;
-  }
-
-  static void verify_clean(MacroAssembler* _masm) NOT_DEBUG_RETURN;
-
-  static void describe(const frame* fr, FrameValues& values, int frame_no) PRODUCT_RETURN;
-};
-
 // Additional helper methods for MethodHandles code generation:
 public:
   static void load_klass_from_Class(MacroAssembler* _masm, Register klass_reg, Register temp_reg, Register temp2_reg);
-  static void load_conversion_vminfo(MacroAssembler* _masm, Address conversion_field_addr, Register reg);
-  static void extract_conversion_vminfo(MacroAssembler* _masm, Register conversion_field_reg, Register reg);
-  static void extract_conversion_dest_type(MacroAssembler* _masm, Register conversion_field_reg, Register reg);
-
-  static void load_stack_move(MacroAssembler* _masm,
-                              Address G3_amh_conversion,
-                              Register G5_stack_move);
-
-  static void insert_arg_slots(MacroAssembler* _masm,
-                               RegisterOrConstant arg_slots,
-                               Register argslot_reg,
-                               Register temp_reg, Register temp2_reg, Register temp3_reg);
-
-  static void remove_arg_slots(MacroAssembler* _masm,
-                               RegisterOrConstant arg_slots,
-                               Register argslot_reg,
-                               Register temp_reg, Register temp2_reg, Register temp3_reg);
-
-  static void push_arg_slots(MacroAssembler* _masm,
-                             Register argslot_reg,
-                             RegisterOrConstant slot_count,
-                             Register temp_reg, Register temp2_reg);
-
-  static void move_arg_slots_up(MacroAssembler* _masm,
-                                Register bottom_reg,  // invariant
-                                Address  top_addr,    // can use temp_reg
-                                RegisterOrConstant positive_distance_in_slots,
-                                Register temp_reg, Register temp2_reg);
-
-  static void move_arg_slots_down(MacroAssembler* _masm,
-                                  Address  bottom_addr,  // can use temp_reg
-                                  Register top_reg,      // invariant
-                                  RegisterOrConstant negative_distance_in_slots,
-                                  Register temp_reg, Register temp2_reg);
-
-  static void move_typed_arg(MacroAssembler* _masm,
-                             BasicType type, bool is_element,
-                             Address value_src, Address slot_dest,
-                             Register temp_reg);
-
-  static void move_return_value(MacroAssembler* _masm, BasicType type,
-                                Address return_slot);
-
-  static void verify_argslot(MacroAssembler* _masm, Register argslot_reg,
-                             Register temp_reg,
-                             const char* error_message) NOT_DEBUG_RETURN;
-
-  static void verify_argslots(MacroAssembler* _masm,
-                              RegisterOrConstant argslot_count,
-                              Register argslot_reg,
-                              Register temp_reg,
-                              Register temp2_reg,
-                              bool negate_argslot,
-                              const char* error_message) NOT_DEBUG_RETURN;
-
-  static void verify_stack_move(MacroAssembler* _masm,
-                                RegisterOrConstant arg_slots,
-                                int direction) NOT_DEBUG_RETURN;
 
   static void verify_klass(MacroAssembler* _masm,
                            Register obj_reg, KlassHandle klass,
@@ -223,8 +46,17 @@ public:
                  "reference is a MH");
   }
 
+  static void verify_ref_kind(MacroAssembler* _masm, int ref_kind, Register member_reg, Register temp) NOT_DEBUG_RETURN;
+
   // Similar to InterpreterMacroAssembler::jump_from_interpreted.
   // Takes care of special dispatch from single stepping too.
-  static void jump_from_method_handle(MacroAssembler* _masm, Register method, Register temp, Register temp2);
+  static void jump_from_method_handle(MacroAssembler* _masm, Register method,
+                                      Register temp, Register temp2,
+                                      bool for_compiler_entry);
+
+  static void jump_to_lambda_form(MacroAssembler* _masm,
+                                  Register recv, Register method_temp,
+                                  Register temp2, Register temp3,
+                                  bool for_compiler_entry);
 
   static void trace_method_handle(MacroAssembler* _masm, const char* adaptername) PRODUCT_RETURN;

hotspot/src/cpu/sparc/vm/sharedRuntime_sparc.cpp

@@ -436,7 +436,6 @@ int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
   // This is where the longs/doubles start on the stack.
   stk_reg_pairs = (stk_reg_pairs+1) & ~1; // Round
 
-  int int_reg_pairs = (int_reg_cnt+1) & ~1; // 32-bit 2-reg longs only
   int flt_reg_pairs = (flt_reg_cnt+1) & ~1;
 
   // int stk_reg = frame::register_save_words*(wordSize>>2);
@@ -517,17 +516,8 @@ int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
           stk_reg_pairs += 2;
         }
 #else // COMPILER2
-        if (int_reg_pairs + 1 < int_reg_max) {
-          if (is_outgoing) {
-            regs[i].set_pair(as_oRegister(int_reg_pairs + 1)->as_VMReg(), as_oRegister(int_reg_pairs)->as_VMReg());
-          } else {
-            regs[i].set_pair(as_iRegister(int_reg_pairs + 1)->as_VMReg(), as_iRegister(int_reg_pairs)->as_VMReg());
-          }
-          int_reg_pairs += 2;
-        } else {
           regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
           stk_reg_pairs += 2;
-        }
 #endif // COMPILER2
 #endif // _LP64
       break;
@@ -886,6 +876,20 @@ void AdapterGenerator::gen_c2i_adapter(
   __ delayed()->add(SP, G1, Gargs);
 }
 
+static void range_check(MacroAssembler* masm, Register pc_reg, Register temp_reg, Register temp2_reg,
+                        address code_start, address code_end,
+                        Label& L_ok) {
+  Label L_fail;
+  __ set(ExternalAddress(code_start), temp_reg);
+  __ set(pointer_delta(code_end, code_start, 1), temp2_reg);
+  __ cmp(pc_reg, temp_reg);
+  __ brx(Assembler::lessEqualUnsigned, false, Assembler::pn, L_fail);
+  __ delayed()->add(temp_reg, temp2_reg, temp_reg);
+  __ cmp(pc_reg, temp_reg);
+  __ cmp_and_brx_short(pc_reg, temp_reg, Assembler::lessUnsigned, Assembler::pt, L_ok);
+  __ bind(L_fail);
+}
+
 void AdapterGenerator::gen_i2c_adapter(
                             int total_args_passed,
                             // VMReg max_arg,
@@ -907,6 +911,51 @@ void AdapterGenerator::gen_i2c_adapter(
   // This removes all sorts of headaches on the x86 side and also eliminates
   // the possibility of having c2i -> i2c -> c2i -> ... endless transitions.
 
+  // More detail:
+  // Adapters can be frameless because they do not require the caller
+  // to perform additional cleanup work, such as correcting the stack pointer.
+  // An i2c adapter is frameless because the *caller* frame, which is interpreted,
+  // routinely repairs its own stack pointer (from interpreter_frame_last_sp),
+  // even if a callee has modified the stack pointer.
+  // A c2i adapter is frameless because the *callee* frame, which is interpreted,
+  // routinely repairs its caller's stack pointer (from sender_sp, which is set
+  // up via the senderSP register).
+  // In other words, if *either* the caller or callee is interpreted, we can
+  // get the stack pointer repaired after a call.
+  // This is why c2i and i2c adapters cannot be indefinitely composed.
+  // In particular, if a c2i adapter were to somehow call an i2c adapter,
+  // both caller and callee would be compiled methods, and neither would
+  // clean up the stack pointer changes performed by the two adapters.
+  // If this happens, control eventually transfers back to the compiled
+  // caller, but with an uncorrected stack, causing delayed havoc.
+
+  if (VerifyAdapterCalls &&
+      (Interpreter::code() != NULL || StubRoutines::code1() != NULL)) {
+    // So, let's test for cascading c2i/i2c adapters right now.
+    //  assert(Interpreter::contains($return_addr) ||
+    //         StubRoutines::contains($return_addr),
+    //         "i2c adapter must return to an interpreter frame");
+    __ block_comment("verify_i2c { ");
+    Label L_ok;
+    if (Interpreter::code() != NULL)
+      range_check(masm, O7, O0, O1,
+                  Interpreter::code()->code_start(), Interpreter::code()->code_end(),
+                  L_ok);
+    if (StubRoutines::code1() != NULL)
+      range_check(masm, O7, O0, O1,
+                  StubRoutines::code1()->code_begin(), StubRoutines::code1()->code_end(),
+                  L_ok);
+    if (StubRoutines::code2() != NULL)
+      range_check(masm, O7, O0, O1,
+                  StubRoutines::code2()->code_begin(), StubRoutines::code2()->code_end(),
+                  L_ok);
+    const char* msg = "i2c adapter must return to an interpreter frame";
+    __ block_comment(msg);
+    __ stop(msg);
+    __ bind(L_ok);
+    __ block_comment("} verify_i2ce ");
+  }
+
   // As you can see from the list of inputs & outputs there are not a lot
   // of temp registers to work with: mostly G1, G3 & G4.
 
@@ -1937,12 +1986,125 @@ static void unpack_array_argument(MacroAssembler* masm, VMRegPair reg, BasicType
   __ bind(done);
 }
 
+static void verify_oop_args(MacroAssembler* masm,
+                            int total_args_passed,
+                            const BasicType* sig_bt,
+                            const VMRegPair* regs) {
+  Register temp_reg = G5_method;  // not part of any compiled calling seq
+  if (VerifyOops) {
+    for (int i = 0; i < total_args_passed; i++) {
+      if (sig_bt[i] == T_OBJECT ||
+          sig_bt[i] == T_ARRAY) {
+        VMReg r = regs[i].first();
+        assert(r->is_valid(), "bad oop arg");
+        if (r->is_stack()) {
+          RegisterOrConstant ld_off = reg2offset(r) + STACK_BIAS;
+          ld_off = __ ensure_simm13_or_reg(ld_off, temp_reg);
+          __ ld_ptr(SP, ld_off, temp_reg);
+          __ verify_oop(temp_reg);
+        } else {
+          __ verify_oop(r->as_Register());
+        }
+      }
+    }
+  }
+}
+
+static void gen_special_dispatch(MacroAssembler* masm,
+                                 int total_args_passed,
+                                 int comp_args_on_stack,
+                                 vmIntrinsics::ID special_dispatch,
+                                 const BasicType* sig_bt,
+                                 const VMRegPair* regs) {
+  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+
+  // Now write the args into the outgoing interpreter space
+  bool     has_receiver   = false;
+  Register receiver_reg   = noreg;
+  int      member_arg_pos = -1;
+  Register member_reg     = noreg;
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  if (ref_kind != 0) {
+    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_reg = G5_method;  // known to be free at this point
+    has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
+  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+    has_receiver = true;
+  } else {
+    fatal(err_msg("special_dispatch=%d", special_dispatch));
+  }
+
+  if (member_reg != noreg) {
+    // Load the member_arg into register, if necessary.
+    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
+    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    VMReg r = regs[member_arg_pos].first();
+    assert(r->is_valid(), "bad member arg");
+    if (r->is_stack()) {
+      RegisterOrConstant ld_off = reg2offset(r) + STACK_BIAS;
+      ld_off = __ ensure_simm13_or_reg(ld_off, member_reg);
+      __ ld_ptr(SP, ld_off, member_reg);
+    } else {
+      // no data motion is needed
+      member_reg = r->as_Register();
+    }
+  }
+
+  if (has_receiver) {
+    // Make sure the receiver is loaded into a register.
+    assert(total_args_passed > 0, "oob");
+    assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
+    VMReg r = regs[0].first();
+    assert(r->is_valid(), "bad receiver arg");
+    if (r->is_stack()) {
+      // Porting note:  This assumes that compiled calling conventions always
+      // pass the receiver oop in a register.  If this is not true on some
+      // platform, pick a temp and load the receiver from stack.
+      assert(false, "receiver always in a register");
+      receiver_reg = G3_scratch;  // known to be free at this point
+      RegisterOrConstant ld_off = reg2offset(r) + STACK_BIAS;
+      ld_off = __ ensure_simm13_or_reg(ld_off, member_reg);
+      __ ld_ptr(SP, ld_off, receiver_reg);
+    } else {
+      // no data motion is needed
+      receiver_reg = r->as_Register();
+    }
+  }
+
+  // Figure out which address we are really jumping to:
+  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+                                                 receiver_reg, member_reg, /*for_compiler_entry:*/ true);
+}
+
 // ---------------------------------------------------------------------------
 // Generate a native wrapper for a given method.  The method takes arguments
 // in the Java compiled code convention, marshals them to the native
 // convention (handlizes oops, etc), transitions to native, makes the call,
 // returns to java state (possibly blocking), unhandlizes any result and
 // returns.
+//
+// Critical native functions are a shorthand for the use of
+// GetPrimtiveArrayCritical and disallow the use of any other JNI
+// functions.  The wrapper is expected to unpack the arguments before
+// passing them to the callee and perform checks before and after the
+// native call to ensure that they GC_locker
+// lock_critical/unlock_critical semantics are followed.  Some other
+// parts of JNI setup are skipped like the tear down of the JNI handle
+// block and the check for pending exceptions it's impossible for them
+// to be thrown.
+//
+// They are roughly structured like this:
+//    if (GC_locker::needs_gc())
+//      SharedRuntime::block_for_jni_critical();
+//    tranistion to thread_in_native
+//    unpack arrray arguments and call native entry point
+//    check for safepoint in progress
+//    check if any thread suspend flags are set
+//      call into JVM and possible unlock the JNI critical
+//      if a GC was suppressed while in the critical native.
+//    transition back to thread_in_Java
+//    return to caller
+//
 nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 methodHandle method,
                                                 int compile_id,
@@ -1951,6 +2113,29 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
+  if (method->is_method_handle_intrinsic()) {
+    vmIntrinsics::ID iid = method->intrinsic_id();
+    intptr_t start = (intptr_t)__ pc();
+    int vep_offset = ((intptr_t)__ pc()) - start;
+    gen_special_dispatch(masm,
+                         total_in_args,
+                         comp_args_on_stack,
+                         method->intrinsic_id(),
+                         in_sig_bt,
+                         in_regs);
+    int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
+    __ flush();
+    int stack_slots = SharedRuntime::out_preserve_stack_slots();  // no out slots at all, actually
+    return nmethod::new_native_nmethod(method,
+                                       compile_id,
+                                       masm->code(),
+                                       vep_offset,
+                                       frame_complete,
+                                       stack_slots / VMRegImpl::slots_per_word,
+                                       in_ByteSize(-1),
+                                       in_ByteSize(-1),
+                                       (OopMapSet*)NULL);
+  }
   bool is_critical_native = true;
   address native_func = method->critical_native_function();
   if (native_func == NULL) {

hotspot/src/cpu/sparc/vm/stubGenerator_sparc.cpp

@@ -3404,14 +3404,6 @@ class StubGenerator: public StubCodeGenerator {
     StubRoutines::_atomic_add_ptr_entry      = StubRoutines::_atomic_add_entry;
 #endif  // COMPILER2 !=> _LP64
 
-    // Build this early so it's available for the interpreter.  The
-    // stub expects the required and actual type to already be in O1
-    // and O2 respectively.
-    StubRoutines::_throw_WrongMethodTypeException_entry =
-      generate_throw_exception("WrongMethodTypeException throw_exception",
-                               CAST_FROM_FN_PTR(address, SharedRuntime::throw_WrongMethodTypeException),
-                               G5_method_type, G3_method_handle);
-
     // Build this early so it's available for the interpreter.
     StubRoutines::_throw_StackOverflowError_entry          = generate_throw_exception("StackOverflowError throw_exception",           CAST_FROM_FN_PTR(address, SharedRuntime::throw_StackOverflowError));
   }

hotspot/src/cpu/sparc/vm/templateInterpreter_sparc.cpp

@@ -694,9 +694,9 @@ address InterpreterGenerator::generate_accessor_entry(void) {
     // Need to differentiate between igetfield, agetfield, bgetfield etc.
     // because they are different sizes.
     // Get the type from the constant pool cache
-    __ srl(G1_scratch, ConstantPoolCacheEntry::tosBits, G1_scratch);
+    __ srl(G1_scratch, ConstantPoolCacheEntry::tos_state_shift, G1_scratch);
-    // Make sure we don't need to mask G1_scratch for tosBits after the above shift
+    // Make sure we don't need to mask G1_scratch after the above shift
-    ConstantPoolCacheEntry::verify_tosBits();
+    ConstantPoolCacheEntry::verify_tos_state_shift();
     __ cmp(G1_scratch, atos );
     __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
     __ delayed()->ld_ptr(Otos_i, G3_scratch, Otos_i);
@@ -1662,7 +1662,7 @@ int AbstractInterpreter::layout_activation(methodOop method,
       int computed_sp_adjustment = (delta > 0) ? round_to(delta, WordsPerLong) : 0;
       *interpreter_frame->register_addr(I5_savedSP)    = (intptr_t) (fp + computed_sp_adjustment) - STACK_BIAS;
     } else {
-      assert(caller->is_compiled_frame() || caller->is_entry_frame() || caller->is_ricochet_frame(), "only possible cases");
+      assert(caller->is_compiled_frame() || caller->is_entry_frame(), "only possible cases");
       // Don't have Lesp available; lay out locals block in the caller
       // adjacent to the register window save area.
       //

hotspot/src/cpu/sparc/vm/templateTable_sparc.cpp

@@ -378,7 +378,7 @@ void TemplateTable::fast_aldc(bool wide) {
   Register Rcache = G3_scratch;
   Register Rscratch = G4_scratch;
 
-  resolve_cache_and_index(f1_oop, Otos_i, Rcache, Rscratch, wide ? sizeof(u2) : sizeof(u1));
+  resolve_cache_and_index(f12_oop, Otos_i, Rcache, Rscratch, wide ? sizeof(u2) : sizeof(u1));
 
   __ verify_oop(Otos_i);
 
@@ -2093,10 +2093,12 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
   // Depends on cpCacheOop layout!
   Label resolved;
 
-  if (byte_no == f1_oop) {
+  if (byte_no == f12_oop) {
-    // We are resolved if the f1 field contains a non-null object (CallSite, etc.)
+    // We are resolved if the f1 field contains a non-null object (CallSite, MethodType, etc.)
-    // This kind of CP cache entry does not need to match the flags byte, because
+    // This kind of CP cache entry does not need to match bytecode_1 or bytecode_2, because
     // there is a 1-1 relation between bytecode type and CP entry type.
+    // The caller will also load a methodOop from f2.
+    assert(result != noreg, "");
     assert_different_registers(result, Rcache);
     __ get_cache_and_index_at_bcp(Rcache, index, 1, index_size);
     __ ld_ptr(Rcache, constantPoolCacheOopDesc::base_offset() +
@@ -2123,10 +2125,13 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
     case Bytecodes::_invokespecial  : // fall through
     case Bytecodes::_invokestatic   : // fall through
     case Bytecodes::_invokeinterface: entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invoke);  break;
+    case Bytecodes::_invokehandle   : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokehandle);  break;
     case Bytecodes::_invokedynamic  : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokedynamic);  break;
     case Bytecodes::_fast_aldc      : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);     break;
     case Bytecodes::_fast_aldc_w    : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);     break;
-    default                         : ShouldNotReachHere();                                 break;
+    default:
+      fatal(err_msg("unexpected bytecode: %s", Bytecodes::name(bytecode())));
+      break;
   }
   // first time invocation - must resolve first
   __ call_VM(noreg, entry, O1);
@@ -2139,48 +2144,54 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
 }
 
 void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
-                                               Register Rmethod,
+                                               Register method,
-                                               Register Ritable_index,
+                                               Register itable_index,
-                                               Register Rflags,
+                                               Register flags,
                                                bool is_invokevirtual,
                                                bool is_invokevfinal,
                                                bool is_invokedynamic) {
   // Uses both G3_scratch and G4_scratch
-  Register Rcache = G3_scratch;
+  Register cache = G3_scratch;
-  Register Rscratch = G4_scratch;
+  Register index = G4_scratch;
-  assert_different_registers(Rcache, Rmethod, Ritable_index);
+  assert_different_registers(cache, method, itable_index);
-
-  ByteSize cp_base_offset = constantPoolCacheOopDesc::base_offset();
 
   // determine constant pool cache field offsets
+  assert(is_invokevirtual == (byte_no == f2_byte), "is_invokevirtual flag redundant");
   const int method_offset = in_bytes(
-    cp_base_offset +
+      constantPoolCacheOopDesc::base_offset() +
-      (is_invokevirtual
+      ((byte_no == f2_byte)
        ? ConstantPoolCacheEntry::f2_offset()
        : ConstantPoolCacheEntry::f1_offset()
       )
     );
-  const int flags_offset = in_bytes(cp_base_offset +
+  const int flags_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
                                     ConstantPoolCacheEntry::flags_offset());
   // access constant pool cache fields
-  const int index_offset = in_bytes(cp_base_offset +
+  const int index_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
                                     ConstantPoolCacheEntry::f2_offset());
 
   if (is_invokevfinal) {
-    __ get_cache_and_index_at_bcp(Rcache, Rscratch, 1);
+    __ get_cache_and_index_at_bcp(cache, index, 1);
-    __ ld_ptr(Rcache, method_offset, Rmethod);
+    __ ld_ptr(Address(cache, method_offset), method);
-  } else if (byte_no == f1_oop) {
+  } else if (byte_no == f12_oop) {
-    // Resolved f1_oop goes directly into 'method' register.
+    // Resolved f1_oop (CallSite, MethodType, etc.) goes into 'itable_index'.
-    resolve_cache_and_index(byte_no, Rmethod, Rcache, Rscratch, sizeof(u4));
+    // Resolved f2_oop (methodOop invoker) will go into 'method' (at index_offset).
+    // See ConstantPoolCacheEntry::set_dynamic_call and set_method_handle.
+    size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
+    resolve_cache_and_index(byte_no, itable_index, cache, index, index_size);
+    __ ld_ptr(Address(cache, index_offset), method);
+    itable_index = noreg;  // hack to disable load below
   } else {
-    resolve_cache_and_index(byte_no, noreg, Rcache, Rscratch, sizeof(u2));
+    resolve_cache_and_index(byte_no, noreg, cache, index, sizeof(u2));
-    __ ld_ptr(Rcache, method_offset, Rmethod);
+    __ ld_ptr(Address(cache, method_offset), method);
   }
 
-  if (Ritable_index != noreg) {
+  if (itable_index != noreg) {
-    __ ld_ptr(Rcache, index_offset, Ritable_index);
+    // pick up itable index from f2 also:
+    assert(byte_no == f1_byte, "already picked up f1");
+    __ ld_ptr(Address(cache, index_offset), itable_index);
   }
-  __ ld_ptr(Rcache, flags_offset, Rflags);
+  __ ld_ptr(Address(cache, flags_offset), flags);
 }
 
 // The Rcache register must be set before call
@@ -2272,7 +2283,7 @@ void TemplateTable::getfield_or_static(int byte_no, bool is_static) {
 
   if (__ membar_has_effect(membar_bits)) {
     // Get volatile flag
-    __ set((1 << ConstantPoolCacheEntry::volatileField), Lscratch);
+    __ set((1 << ConstantPoolCacheEntry::is_volatile_shift), Lscratch);
     __ and3(Rflags, Lscratch, Lscratch);
   }
 
@@ -2280,9 +2291,9 @@ void TemplateTable::getfield_or_static(int byte_no, bool is_static) {
 
   // compute field type
   Label notByte, notInt, notShort, notChar, notLong, notFloat, notObj;
-  __ srl(Rflags, ConstantPoolCacheEntry::tosBits, Rflags);
+  __ srl(Rflags, ConstantPoolCacheEntry::tos_state_shift, Rflags);
-  // Make sure we don't need to mask Rflags for tosBits after the above shift
+  // Make sure we don't need to mask Rflags after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
 
   // Check atos before itos for getstatic, more likely (in Queens at least)
   __ cmp(Rflags, atos);
@@ -2445,7 +2456,7 @@ void TemplateTable::fast_accessfield(TosState state) {
   if (__ membar_has_effect(membar_bits)) {
     // Get volatile flag
     __ ld_ptr(Rcache, cp_base_offset + ConstantPoolCacheEntry::f2_offset(), Rflags);
-    __ set((1 << ConstantPoolCacheEntry::volatileField), Lscratch);
+    __ set((1 << ConstantPoolCacheEntry::is_volatile_shift), Lscratch);
   }
 
   switch (bytecode()) {
@@ -2569,9 +2580,9 @@ void TemplateTable::jvmti_post_field_mod(Register Rcache, Register index, bool i
       Label two_word, valsizeknown;
       __ ld_ptr(G1_scratch, cp_base_offset + ConstantPoolCacheEntry::flags_offset(), Rflags);
       __ mov(Lesp, G4_scratch);
-      __ srl(Rflags, ConstantPoolCacheEntry::tosBits, Rflags);
+      __ srl(Rflags, ConstantPoolCacheEntry::tos_state_shift, Rflags);
-      // Make sure we don't need to mask Rflags for tosBits after the above shift
+      // Make sure we don't need to mask Rflags after the above shift
-      ConstantPoolCacheEntry::verify_tosBits();
+      ConstantPoolCacheEntry::verify_tos_state_shift();
       __ cmp(Rflags, ltos);
       __ br(Assembler::equal, false, Assembler::pt, two_word);
       __ delayed()->cmp(Rflags, dtos);
@@ -2625,7 +2636,7 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
 
   Label notVolatile, checkVolatile, exit;
   if (__ membar_has_effect(read_bits) || __ membar_has_effect(write_bits)) {
-    __ set((1 << ConstantPoolCacheEntry::volatileField), Lscratch);
+    __ set((1 << ConstantPoolCacheEntry::is_volatile_shift), Lscratch);
     __ and3(Rflags, Lscratch, Lscratch);
 
     if (__ membar_has_effect(read_bits)) {
@@ -2635,9 +2646,9 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
     }
   }
 
-  __ srl(Rflags, ConstantPoolCacheEntry::tosBits, Rflags);
+  __ srl(Rflags, ConstantPoolCacheEntry::tos_state_shift, Rflags);
-  // Make sure we don't need to mask Rflags for tosBits after the above shift
+  // Make sure we don't need to mask Rflags after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
 
   // compute field type
   Label notInt, notShort, notChar, notObj, notByte, notLong, notFloat;
@@ -2833,7 +2844,7 @@ void TemplateTable::fast_storefield(TosState state) {
   Label notVolatile, checkVolatile, exit;
   if (__ membar_has_effect(read_bits) || __ membar_has_effect(write_bits)) {
     __ ld_ptr(Rcache, cp_base_offset + ConstantPoolCacheEntry::flags_offset(), Rflags);
-    __ set((1 << ConstantPoolCacheEntry::volatileField), Lscratch);
+    __ set((1 << ConstantPoolCacheEntry::is_volatile_shift), Lscratch);
     __ and3(Rflags, Lscratch, Lscratch);
     if (__ membar_has_effect(read_bits)) {
       __ cmp_and_br_short(Lscratch, 0, Assembler::equal, Assembler::pt, notVolatile);
@@ -2916,7 +2927,7 @@ void TemplateTable::fast_xaccess(TosState state) {
 
     // Test volatile
     Label notVolatile;
-    __ set((1 << ConstantPoolCacheEntry::volatileField), Lscratch);
+    __ set((1 << ConstantPoolCacheEntry::is_volatile_shift), Lscratch);
     __ btst(Rflags, Lscratch);
     __ br(Assembler::zero, false, Assembler::pt, notVolatile);
     __ delayed()->nop();
@@ -2936,27 +2947,82 @@ void TemplateTable::count_calls(Register method, Register temp) {
   ShouldNotReachHere();
 }
 
+
+void TemplateTable::prepare_invoke(int byte_no,
+                                   Register method,  // linked method (or i-klass)
+                                   Register ra,      // return address
+                                   Register index,   // itable index, MethodType, etc.
+                                   Register recv,    // if caller wants to see it
+                                   Register flags    // if caller wants to test it
+                                   ) {
+  // determine flags
+  const Bytecodes::Code code = bytecode();
+  const bool is_invokeinterface  = code == Bytecodes::_invokeinterface;
+  const bool is_invokedynamic    = code == Bytecodes::_invokedynamic;
+  const bool is_invokehandle     = code == Bytecodes::_invokehandle;
+  const bool is_invokevirtual    = code == Bytecodes::_invokevirtual;
+  const bool is_invokespecial    = code == Bytecodes::_invokespecial;
+  const bool load_receiver       = (recv != noreg);
+  assert(load_receiver == (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic), "");
+  assert(recv  == noreg || recv  == O0, "");
+  assert(flags == noreg || flags == O1, "");
+
+  // setup registers & access constant pool cache
+  if (recv  == noreg)  recv  = O0;
+  if (flags == noreg)  flags = O1;
+  const Register temp = O2;
+  assert_different_registers(method, ra, index, recv, flags, temp);
+
+  load_invoke_cp_cache_entry(byte_no, method, index, flags, is_invokevirtual, false, is_invokedynamic);
+
+  __ mov(SP, O5_savedSP);  // record SP that we wanted the callee to restore
+
+  // maybe push appendix to arguments
+  if (is_invokedynamic || is_invokehandle) {
+    Label L_no_push;
+    __ verify_oop(index);
+    __ set((1 << ConstantPoolCacheEntry::has_appendix_shift), temp);
+    __ btst(flags, temp);
+    __ br(Assembler::zero, false, Assembler::pt, L_no_push);
+    __ delayed()->nop();
+    // Push the appendix as a trailing parameter.
+    // This must be done before we get the receiver,
+    // since the parameter_size includes it.
+    __ push_ptr(index);  // push appendix (MethodType, CallSite, etc.)
+    __ bind(L_no_push);
+  }
+
+  // load receiver if needed (after appendix is pushed so parameter size is correct)
+  if (load_receiver) {
+    __ and3(flags, ConstantPoolCacheEntry::parameter_size_mask, temp);  // get parameter size
+    __ load_receiver(temp, recv);  //  __ argument_address uses Gargs but we need Lesp
+    __ verify_oop(recv);
+  }
+
+  // compute return type
+  __ srl(flags, ConstantPoolCacheEntry::tos_state_shift, ra);
+  // Make sure we don't need to mask flags after the above shift
+  ConstantPoolCacheEntry::verify_tos_state_shift();
+  // load return address
+  {
+    const address table_addr = (is_invokeinterface || is_invokedynamic) ?
+        (address)Interpreter::return_5_addrs_by_index_table() :
+        (address)Interpreter::return_3_addrs_by_index_table();
+    AddressLiteral table(table_addr);
+    __ set(table, temp);
+    __ sll(ra, LogBytesPerWord, ra);
+    __ ld_ptr(Address(temp, ra), ra);
+  }
+}
+
+
 void TemplateTable::generate_vtable_call(Register Rrecv, Register Rindex, Register Rret) {
   Register Rtemp = G4_scratch;
   Register Rcall = Rindex;
   assert_different_registers(Rcall, G5_method, Gargs, Rret);
 
   // get target methodOop & entry point
-  const int base = instanceKlass::vtable_start_offset() * wordSize;
+  __ lookup_virtual_method(Rrecv, Rindex, G5_method);
-  if (vtableEntry::size() % 3 == 0) {
-    // scale the vtable index by 12:
-    int one_third = vtableEntry::size() / 3;
-    __ sll(Rindex, exact_log2(one_third * 1 * wordSize), Rtemp);
-    __ sll(Rindex, exact_log2(one_third * 2 * wordSize), Rindex);
-    __ add(Rindex, Rtemp, Rindex);
-  } else {
-    // scale the vtable index by 8:
-    __ sll(Rindex, exact_log2(vtableEntry::size() * wordSize), Rindex);
-  }
-
-  __ add(Rrecv, Rindex, Rrecv);
-  __ ld_ptr(Rrecv, base + vtableEntry::method_offset_in_bytes(), G5_method);
-
   __ call_from_interpreter(Rcall, Gargs, Rret);
 }
 
@@ -2967,14 +3033,14 @@ void TemplateTable::invokevirtual(int byte_no) {
   Register Rscratch = G3_scratch;
   Register Rtemp    = G4_scratch;
   Register Rret     = Lscratch;
-  Register Rrecv = G5_method;
+  Register O0_recv  = O0;
   Label notFinal;
 
   load_invoke_cp_cache_entry(byte_no, G5_method, noreg, Rret, true, false, false);
   __ mov(SP, O5_savedSP); // record SP that we wanted the callee to restore
 
   // Check for vfinal
-  __ set((1 << ConstantPoolCacheEntry::vfinalMethod), G4_scratch);
+  __ set((1 << ConstantPoolCacheEntry::is_vfinal_shift), G4_scratch);
   __ btst(Rret, G4_scratch);
   __ br(Assembler::zero, false, Assembler::pt, notFinal);
   __ delayed()->and3(Rret, 0xFF, G4_scratch);      // gets number of parameters
@@ -2986,27 +3052,27 @@ void TemplateTable::invokevirtual(int byte_no) {
   __ bind(notFinal);
 
   __ mov(G5_method, Rscratch);  // better scratch register
-  __ load_receiver(G4_scratch, O0);  // gets receiverOop
+  __ load_receiver(G4_scratch, O0_recv);  // gets receiverOop
-  // receiver is in O0
+  // receiver is in O0_recv
-  __ verify_oop(O0);
+  __ verify_oop(O0_recv);
 
   // get return address
   AddressLiteral table(Interpreter::return_3_addrs_by_index_table());
   __ set(table, Rtemp);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);          // get return type
+  __ srl(Rret, ConstantPoolCacheEntry::tos_state_shift, Rret);          // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
+  // Make sure we don't need to mask Rret after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
   __ sll(Rret,  LogBytesPerWord, Rret);
   __ ld_ptr(Rtemp, Rret, Rret);         // get return address
 
   // get receiver klass
-  __ null_check(O0, oopDesc::klass_offset_in_bytes());
+  __ null_check(O0_recv, oopDesc::klass_offset_in_bytes());
-  __ load_klass(O0, Rrecv);
+  __ load_klass(O0_recv, O0_recv);
-  __ verify_oop(Rrecv);
+  __ verify_oop(O0_recv);
 
-  __ profile_virtual_call(Rrecv, O4);
+  __ profile_virtual_call(O0_recv, O4);
 
-  generate_vtable_call(Rrecv, Rscratch, Rret);
+  generate_vtable_call(O0_recv, Rscratch, Rret);
 }
 
 void TemplateTable::fast_invokevfinal(int byte_no) {
@@ -3036,9 +3102,9 @@ void TemplateTable::invokevfinal_helper(Register Rscratch, Register Rret) {
   // get return address
   AddressLiteral table(Interpreter::return_3_addrs_by_index_table());
   __ set(table, Rtemp);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);          // get return type
+  __ srl(Rret, ConstantPoolCacheEntry::tos_state_shift, Rret);          // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
+  // Make sure we don't need to mask Rret after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
   __ sll(Rret,  LogBytesPerWord, Rret);
   __ ld_ptr(Rtemp, Rret, Rret);         // get return address
 
@@ -3047,65 +3113,37 @@ void TemplateTable::invokevfinal_helper(Register Rscratch, Register Rret) {
   __ call_from_interpreter(Rscratch, Gargs, Rret);
 }
 
+
 void TemplateTable::invokespecial(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
 
-  Register Rscratch = G3_scratch;
+  const Register Rret     = Lscratch;
-  Register Rtemp = G4_scratch;
+  const Register O0_recv  = O0;
-  Register Rret = Lscratch;
+  const Register Rscratch = G3_scratch;
 
-  load_invoke_cp_cache_entry(byte_no, G5_method, noreg, Rret, /*virtual*/ false, false, false);
+  prepare_invoke(byte_no, G5_method, Rret, noreg, O0_recv);  // get receiver also for null check
-  __ mov(SP, O5_savedSP); // record SP that we wanted the callee to restore
+  __ null_check(O0_recv);
-
-  __ verify_oop(G5_method);
-
-  __ lduh(G5_method, in_bytes(methodOopDesc::size_of_parameters_offset()), G4_scratch);
-  __ load_receiver(G4_scratch, O0);
-
-  // receiver NULL check
-  __ null_check(O0);
-
-  __ profile_call(O4);
-
-  // get return address
-  AddressLiteral table(Interpreter::return_3_addrs_by_index_table());
-  __ set(table, Rtemp);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);          // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
-  __ sll(Rret,  LogBytesPerWord, Rret);
-  __ ld_ptr(Rtemp, Rret, Rret);         // get return address
 
   // do the call
+  __ verify_oop(G5_method);
+  __ profile_call(O4);
   __ call_from_interpreter(Rscratch, Gargs, Rret);
 }
 
+
 void TemplateTable::invokestatic(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
 
-  Register Rscratch = G3_scratch;
+  const Register Rret     = Lscratch;
-  Register Rtemp = G4_scratch;
+  const Register Rscratch = G3_scratch;
-  Register Rret = Lscratch;
 
-  load_invoke_cp_cache_entry(byte_no, G5_method, noreg, Rret, /*virtual*/ false, false, false);
+  prepare_invoke(byte_no, G5_method, Rret);  // get f1 methodOop
-  __ mov(SP, O5_savedSP); // record SP that we wanted the callee to restore
-
-  __ verify_oop(G5_method);
-
-  __ profile_call(O4);
-
-  // get return address
-  AddressLiteral table(Interpreter::return_3_addrs_by_index_table());
-  __ set(table, Rtemp);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);          // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
-  __ sll(Rret,  LogBytesPerWord, Rret);
-  __ ld_ptr(Rtemp, Rret, Rret);         // get return address
 
   // do the call
+  __ verify_oop(G5_method);
+  __ profile_call(O4);
   __ call_from_interpreter(Rscratch, Gargs, Rret);
 }
 
@@ -3122,7 +3160,7 @@ void TemplateTable::invokeinterface_object_method(Register RklassOop,
   Label notFinal;
 
   // Check for vfinal
-  __ set((1 << ConstantPoolCacheEntry::vfinalMethod), Rscratch);
+  __ set((1 << ConstantPoolCacheEntry::is_vfinal_shift), Rscratch);
   __ btst(Rflags, Rscratch);
   __ br(Assembler::zero, false, Assembler::pt, notFinal);
   __ delayed()->nop();
@@ -3144,53 +3182,37 @@ void TemplateTable::invokeinterface(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
 
-  Register Rscratch = G4_scratch;
+  const Register Rinterface  = G1_scratch;
-  Register Rret = G3_scratch;
+  const Register Rret        = G3_scratch;
-  Register Rindex = Lscratch;
+  const Register Rindex      = Lscratch;
-  Register Rinterface = G1_scratch;
+  const Register O0_recv     = O0;
-  Register RklassOop = G5_method;
+  const Register O1_flags    = O1;
-  Register Rflags = O1;
+  const Register O2_klassOop = O2;
+  const Register Rscratch    = G4_scratch;
   assert_different_registers(Rscratch, G5_method);
 
-  load_invoke_cp_cache_entry(byte_no, Rinterface, Rindex, Rflags, /*virtual*/ false, false, false);
+  prepare_invoke(byte_no, Rinterface, Rret, Rindex, O0_recv, O1_flags);
-  __ mov(SP, O5_savedSP); // record SP that we wanted the callee to restore
-
-  // get receiver
-  __ and3(Rflags, 0xFF, Rscratch);       // gets number of parameters
-  __ load_receiver(Rscratch, O0);
-  __ verify_oop(O0);
-
-  __ mov(Rflags, Rret);
-
-  // get return address
-  AddressLiteral table(Interpreter::return_5_addrs_by_index_table());
-  __ set(table, Rscratch);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);          // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
-  __ sll(Rret,  LogBytesPerWord, Rret);
-  __ ld_ptr(Rscratch, Rret, Rret);      // get return address
 
   // get receiver klass
-  __ null_check(O0, oopDesc::klass_offset_in_bytes());
+  __ null_check(O0_recv, oopDesc::klass_offset_in_bytes());
-  __ load_klass(O0, RklassOop);
+  __ load_klass(O0_recv, O2_klassOop);
-  __ verify_oop(RklassOop);
+  __ verify_oop(O2_klassOop);
 
   // Special case of invokeinterface called for virtual method of
   // java.lang.Object.  See cpCacheOop.cpp for details.
   // This code isn't produced by javac, but could be produced by
   // another compliant java compiler.
   Label notMethod;
-  __ set((1 << ConstantPoolCacheEntry::methodInterface), Rscratch);
+  __ set((1 << ConstantPoolCacheEntry::is_forced_virtual_shift), Rscratch);
-  __ btst(Rflags, Rscratch);
+  __ btst(O1_flags, Rscratch);
   __ br(Assembler::zero, false, Assembler::pt, notMethod);
   __ delayed()->nop();
 
-  invokeinterface_object_method(RklassOop, Rinterface, Rret, Rflags);
+  invokeinterface_object_method(O2_klassOop, Rinterface, Rret, O1_flags);
 
   __ bind(notMethod);
 
-  __ profile_virtual_call(RklassOop, O4);
+  __ profile_virtual_call(O2_klassOop, O4);
 
   //
   // find entry point to call
@@ -3199,9 +3221,9 @@ void TemplateTable::invokeinterface(int byte_no) {
   // compute start of first itableOffsetEntry (which is at end of vtable)
   const int base = instanceKlass::vtable_start_offset() * wordSize;
   Label search;
-  Register Rtemp = Rflags;
+  Register Rtemp = O1_flags;
 
-  __ ld(RklassOop, instanceKlass::vtable_length_offset() * wordSize, Rtemp);
+  __ ld(O2_klassOop, instanceKlass::vtable_length_offset() * wordSize, Rtemp);
   if (align_object_offset(1) > 1) {
     __ round_to(Rtemp, align_object_offset(1));
   }
@@ -3212,7 +3234,7 @@ void TemplateTable::invokeinterface(int byte_no) {
     __ set(base, Rscratch);
     __ add(Rscratch, Rtemp, Rtemp);
   }
-  __ add(RklassOop, Rtemp, Rscratch);
+  __ add(O2_klassOop, Rtemp, Rscratch);
 
   __ bind(search);
 
@@ -3244,7 +3266,7 @@ void TemplateTable::invokeinterface(int byte_no) {
   assert(itableMethodEntry::method_offset_in_bytes() == 0, "adjust instruction below");
   __ sll(Rindex, exact_log2(itableMethodEntry::size() * wordSize), Rindex);       // Rindex *= 8;
   __ add(Rscratch, Rindex, Rscratch);
-  __ ld_ptr(RklassOop, Rscratch, G5_method);
+  __ ld_ptr(O2_klassOop, Rscratch, G5_method);
 
   // Check for abstract method error.
   {
@@ -3260,13 +3282,42 @@ void TemplateTable::invokeinterface(int byte_no) {
 
   __ verify_oop(G5_method);
   __ call_from_interpreter(Rcall, Gargs, Rret);
+}
 
+
+void TemplateTable::invokehandle(int byte_no) {
+  transition(vtos, vtos);
+  assert(byte_no == f12_oop, "use this argument");
+
+  if (!EnableInvokeDynamic) {
+    // rewriter does not generate this bytecode
+    __ should_not_reach_here();
+    return;
+  }
+
+  const Register Rret       = Lscratch;
+  const Register G4_mtype   = G4_scratch;  // f1
+  const Register O0_recv    = O0;
+  const Register Rscratch   = G3_scratch;
+
+  prepare_invoke(byte_no, G5_method, Rret, G4_mtype, O0_recv);
+  __ null_check(O0_recv);
+
+  // G4: MethodType object (from f1)
+  // G5: MH.linkToCallSite method (from f2)
+
+  // Note:  G4_mtype is already pushed (if necessary) by prepare_invoke
+
+  // do the call
+  __ verify_oop(G5_method);
+  __ profile_final_call(O4);  // FIXME: profile the LambdaForm also
+  __ call_from_interpreter(Rscratch, Gargs, Rret);
 }
 
 
 void TemplateTable::invokedynamic(int byte_no) {
   transition(vtos, vtos);
-  assert(byte_no == f1_oop, "use this argument");
+  assert(byte_no == f12_oop, "use this argument");
 
   if (!EnableInvokeDynamic) {
     // We should not encounter this bytecode if !EnableInvokeDynamic.
@@ -3279,42 +3330,24 @@ void TemplateTable::invokedynamic(int byte_no) {
     return;
   }
 
-  // G5: CallSite object (f1)
+  const Register Rret        = Lscratch;
-  // XX: unused (f2)
+  const Register G4_callsite = G4_scratch;
-  // XX: flags (unused)
+  const Register Rscratch    = G3_scratch;
 
-  Register G5_callsite = G5_method;
+  prepare_invoke(byte_no, G5_method, Rret, G4_callsite);
-  Register Rscratch    = G3_scratch;
-  Register Rtemp       = G1_scratch;
-  Register Rret        = Lscratch;
 
-  load_invoke_cp_cache_entry(byte_no, G5_callsite, noreg, Rret,
+  // G4: CallSite object (from f1)
-                             /*virtual*/ false, /*vfinal*/ false, /*indy*/ true);
+  // G5: MH.linkToCallSite method (from f2)
-  __ mov(SP, O5_savedSP);  // record SP that we wanted the callee to restore
 
+  // Note:  G4_callsite is already pushed by prepare_invoke
+
+  // %%% should make a type profile for any invokedynamic that takes a ref argument
   // profile this call
   __ profile_call(O4);
 
-  // get return address
+  // do the call
-  AddressLiteral table(Interpreter::return_5_addrs_by_index_table());
+  __ verify_oop(G5_method);
-  __ set(table, Rtemp);
+  __ call_from_interpreter(Rscratch, Gargs, Rret);
-  __ srl(Rret, ConstantPoolCacheEntry::tosBits, Rret);  // get return type
-  // Make sure we don't need to mask Rret for tosBits after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
-  __ sll(Rret, LogBytesPerWord, Rret);
-  __ ld_ptr(Rtemp, Rret, Rret);  // get return address
-
-  __ verify_oop(G5_callsite);
-  __ load_heap_oop(G5_callsite, __ delayed_value(java_lang_invoke_CallSite::target_offset_in_bytes, Rscratch), G3_method_handle);
-  __ null_check(G3_method_handle);
-  __ verify_oop(G3_method_handle);
-
-  // Adjust Rret first so Llast_SP can be same as Rret
-  __ add(Rret, -frame::pc_return_offset, O7);
-  __ add(Lesp, BytesPerWord, Gargs);  // setup parameter pointer
-  __ jump_to_method_handle_entry(G3_method_handle, Rtemp, /* emit_delayed_nop */ false);
-  // Record SP so we can remove any stack space allocated by adapter transition
-  __ delayed()->mov(SP, Llast_SP);
 }
 
 

hotspot/src/cpu/sparc/vm/templateTable_sparc.hpp

@@ -25,6 +25,13 @@
 #ifndef CPU_SPARC_VM_TEMPLATETABLE_SPARC_HPP
 #define CPU_SPARC_VM_TEMPLATETABLE_SPARC_HPP
 
+  static void prepare_invoke(int byte_no,
+                             Register method,         // linked method (or i-klass)
+                             Register ra,             // return address
+                             Register index = noreg,  // itable index, MethodType, etc.
+                             Register recv  = noreg,  // if caller wants to see it
+                             Register flags = noreg   // if caller wants to test it
+                             );
   // helper function
   static void invokevfinal_helper(Register Rcache, Register Rret);
   static void invokeinterface_object_method(Register RklassOop, Register Rcall,

hotspot/src/cpu/sparc/vm/vm_version_sparc.cpp

@@ -106,10 +106,10 @@ void VM_Version::initialize() {
     if (FLAG_IS_DEFAULT(OptoLoopAlignment)) {
       FLAG_SET_DEFAULT(OptoLoopAlignment, 4);
     }
-    // When using CMS, we cannot use memset() in BOT updates because
+    // When using CMS or G1, we cannot use memset() in BOT updates
-    // the sun4v/CMT version in libc_psr uses BIS which exposes
+    // because the sun4v/CMT version in libc_psr uses BIS which
-    // "phantom zeros" to concurrent readers. See 6948537.
+    // exposes "phantom zeros" to concurrent readers. See 6948537.
-    if (FLAG_IS_DEFAULT(UseMemSetInBOT) && UseConcMarkSweepGC) {
+    if (FLAG_IS_DEFAULT(UseMemSetInBOT) && (UseConcMarkSweepGC || UseG1GC)) {
       FLAG_SET_DEFAULT(UseMemSetInBOT, false);
     }
 #ifdef _LP64

hotspot/src/cpu/sparc/vm/vm_version_sparc.hpp

@@ -44,10 +44,11 @@ protected:
     fmaf_instructions    = 10,
     fmau_instructions    = 11,
     vis3_instructions    = 12,
-    sparc64_family       = 13,
+    cbcond_instructions  = 13,
-    T_family             = 14,
+    sparc64_family       = 14,
-    T1_model             = 15,
+    M_family             = 15,
-    cbcond_instructions  = 16
+    T_family             = 16,
+    T1_model             = 17
   };
 
   enum Feature_Flag_Set {
@@ -67,10 +68,11 @@ protected:
     fmaf_instructions_m     = 1 << fmaf_instructions,
     fmau_instructions_m     = 1 << fmau_instructions,
     vis3_instructions_m     = 1 << vis3_instructions,
+    cbcond_instructions_m   = 1 << cbcond_instructions,
     sparc64_family_m        = 1 << sparc64_family,
+    M_family_m              = 1 << M_family,
     T_family_m              = 1 << T_family,
     T1_model_m              = 1 << T1_model,
-    cbcond_instructions_m   = 1 << cbcond_instructions,
 
     generic_v8_m        = v8_instructions_m | hardware_mul32_m | hardware_div32_m | hardware_fsmuld_m,
     generic_v9_m        = generic_v8_m | v9_instructions_m,
@@ -89,6 +91,7 @@ protected:
   static int  platform_features(int features);
 
   // Returns true if the platform is in the niagara line (T series)
+  static bool is_M_family(int features) { return (features & M_family_m) != 0; }
   static bool is_T_family(int features) { return (features & T_family_m) != 0; }
   static bool is_niagara() { return is_T_family(_features); }
   DEBUG_ONLY( static bool is_niagara(int features)  { return (features & sun4v_m) != 0; } )

hotspot/src/cpu/sparc/vm/vtableStubs_sparc.cpp

@@ -70,7 +70,6 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
   __ load_klass(O0, G3_scratch);
 
   // set methodOop (in case of interpreted method), and destination address
-  int entry_offset = instanceKlass::vtable_start_offset() + vtable_index*vtableEntry::size();
 #ifndef PRODUCT
   if (DebugVtables) {
     Label L;
@@ -82,13 +81,8 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
     __ bind(L);
   }
 #endif
-  int v_off = entry_offset*wordSize + vtableEntry::method_offset_in_bytes();
+
-  if (Assembler::is_simm13(v_off)) {
+  __ lookup_virtual_method(G3_scratch, vtable_index, G5_method);
-    __ ld_ptr(G3, v_off, G5_method);
-  } else {
-    __ set(v_off,G5);
-    __ ld_ptr(G3, G5, G5_method);
-  }
 
 #ifndef PRODUCT
   if (DebugVtables) {

hotspot/src/cpu/x86/vm/assembler_x86.hpp

@@ -617,6 +617,7 @@ private:
                    VexSimdPrefix pre, VexOpcode opc = VEX_OPCODE_0F) {
     simd_prefix(dst, xnoreg, src, pre, opc);
   }
+
   void simd_prefix(Address dst, XMMRegister src, VexSimdPrefix pre) {
     simd_prefix(src, dst, pre);
   }
@@ -626,16 +627,10 @@ private:
     simd_prefix(dst, nds, src, pre, VEX_OPCODE_0F, rex_w);
   }
 
-
   int simd_prefix_and_encode(XMMRegister dst, XMMRegister nds, XMMRegister src,
                              VexSimdPrefix pre, VexOpcode opc = VEX_OPCODE_0F,
                              bool rex_w = false, bool vector256 = false);
 
-  int simd_prefix_and_encode(XMMRegister dst, XMMRegister src,
-                             VexSimdPrefix pre, VexOpcode opc = VEX_OPCODE_0F) {
-    return simd_prefix_and_encode(dst, xnoreg, src, pre, opc);
-  }
-
   // Move/convert 32-bit integer value.
   int simd_prefix_and_encode(XMMRegister dst, XMMRegister nds, Register src,
                              VexSimdPrefix pre) {
@@ -677,6 +672,15 @@ private:
   void emit_arith(int op1, int op2, Register dst, jobject obj);
   void emit_arith(int op1, int op2, Register dst, Register src);
 
+  void emit_simd_arith(int opcode, XMMRegister dst, Address src, VexSimdPrefix pre);
+  void emit_simd_arith(int opcode, XMMRegister dst, XMMRegister src, VexSimdPrefix pre);
+  void emit_simd_arith_nonds(int opcode, XMMRegister dst, Address src, VexSimdPrefix pre);
+  void emit_simd_arith_nonds(int opcode, XMMRegister dst, XMMRegister src, VexSimdPrefix pre);
+  void emit_vex_arith(int opcode, XMMRegister dst, XMMRegister nds,
+                      Address src, VexSimdPrefix pre, bool vector256);
+  void emit_vex_arith(int opcode, XMMRegister dst, XMMRegister nds,
+                      XMMRegister src, VexSimdPrefix pre, bool vector256);
+
   void emit_operand(Register reg,
                     Register base, Register index, Address::ScaleFactor scale,
                     int disp,
@@ -891,12 +895,6 @@ private:
   void andq(Register dst, Address src);
   void andq(Register dst, Register src);
 
-  // Bitwise Logical AND of Packed Double-Precision Floating-Point Values
-  void andpd(XMMRegister dst, XMMRegister src);
-
-  // Bitwise Logical AND of Packed Single-Precision Floating-Point Values
-  void andps(XMMRegister dst, XMMRegister src);
-
   void bsfl(Register dst, Register src);
   void bsrl(Register dst, Register src);
 
@@ -1436,10 +1434,6 @@ private:
   void prefetcht2(Address src);
   void prefetchw(Address src);
 
-  // POR - Bitwise logical OR
-  void por(XMMRegister dst, XMMRegister src);
-  void por(XMMRegister dst, Address src);
-
   // Shuffle Packed Doublewords
   void pshufd(XMMRegister dst, XMMRegister src, int mode);
   void pshufd(XMMRegister dst, Address src,     int mode);
@@ -1448,9 +1442,6 @@ private:
   void pshuflw(XMMRegister dst, XMMRegister src, int mode);
   void pshuflw(XMMRegister dst, Address src,     int mode);
 
-  // Shift Right by bits Logical Quadword Immediate
-  void psrlq(XMMRegister dst, int shift);
-
   // Shift Right by bytes Logical DoubleQuadword Immediate
   void psrldq(XMMRegister dst, int shift);
 
@@ -1475,10 +1466,6 @@ private:
 
   void pushq(Address src);
 
-  // Xor Packed Byte Integer Values
-  void pxor(XMMRegister dst, Address src);
-  void pxor(XMMRegister dst, XMMRegister src);
-
   void rcll(Register dst, int imm8);
 
   void rclq(Register dst, int imm8);
@@ -1601,15 +1588,10 @@ private:
   void xorq(Register dst, Address src);
   void xorq(Register dst, Register src);
 
-  // Bitwise Logical XOR of Packed Double-Precision Floating-Point Values
-  void xorpd(XMMRegister dst, XMMRegister src);
-
-  // Bitwise Logical XOR of Packed Single-Precision Floating-Point Values
-  void xorps(XMMRegister dst, XMMRegister src);
-
   void set_byte_if_not_zero(Register dst); // sets reg to 1 if not zero, otherwise 0
 
   // AVX 3-operands scalar instructions (encoded with VEX prefix)
+
   void vaddsd(XMMRegister dst, XMMRegister nds, Address src);
   void vaddsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
   void vaddss(XMMRegister dst, XMMRegister nds, Address src);
@@ -1627,14 +1609,147 @@ private:
   void vsubss(XMMRegister dst, XMMRegister nds, Address src);
   void vsubss(XMMRegister dst, XMMRegister nds, XMMRegister src);
 
-  // AVX Vector instrucitons.
+
-  void vandpd(XMMRegister dst, XMMRegister nds, Address src);
+  //====================VECTOR ARITHMETIC=====================================
-  void vandps(XMMRegister dst, XMMRegister nds, Address src);
+
-  void vxorpd(XMMRegister dst, XMMRegister nds, Address src);
+  // Add Packed Floating-Point Values
-  void vxorps(XMMRegister dst, XMMRegister nds, Address src);
+  void addpd(XMMRegister dst, XMMRegister src);
+  void addps(XMMRegister dst, XMMRegister src);
+  void vaddpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vaddps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vaddpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vaddps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Subtract Packed Floating-Point Values
+  void subpd(XMMRegister dst, XMMRegister src);
+  void subps(XMMRegister dst, XMMRegister src);
+  void vsubpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vsubps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vsubpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vsubps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Multiply Packed Floating-Point Values
+  void mulpd(XMMRegister dst, XMMRegister src);
+  void mulps(XMMRegister dst, XMMRegister src);
+  void vmulpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vmulps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vmulpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vmulps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Divide Packed Floating-Point Values
+  void divpd(XMMRegister dst, XMMRegister src);
+  void divps(XMMRegister dst, XMMRegister src);
+  void vdivpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vdivps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vdivpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vdivps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Bitwise Logical AND of Packed Floating-Point Values
+  void andpd(XMMRegister dst, XMMRegister src);
+  void andps(XMMRegister dst, XMMRegister src);
+  void vandpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vandps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vandpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vandps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Bitwise Logical XOR of Packed Floating-Point Values
+  void xorpd(XMMRegister dst, XMMRegister src);
+  void xorps(XMMRegister dst, XMMRegister src);
   void vxorpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
   void vxorps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vxorpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vxorps(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Add packed integers
+  void paddb(XMMRegister dst, XMMRegister src);
+  void paddw(XMMRegister dst, XMMRegister src);
+  void paddd(XMMRegister dst, XMMRegister src);
+  void paddq(XMMRegister dst, XMMRegister src);
+  void vpaddb(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpaddw(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpaddd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpaddq(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpaddb(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpaddw(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpaddd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpaddq(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Sub packed integers
+  void psubb(XMMRegister dst, XMMRegister src);
+  void psubw(XMMRegister dst, XMMRegister src);
+  void psubd(XMMRegister dst, XMMRegister src);
+  void psubq(XMMRegister dst, XMMRegister src);
+  void vpsubb(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpsubw(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpsubd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpsubq(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpsubb(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpsubw(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpsubd(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpsubq(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Multiply packed integers (only shorts and ints)
+  void pmullw(XMMRegister dst, XMMRegister src);
+  void pmulld(XMMRegister dst, XMMRegister src);
+  void vpmullw(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpmulld(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpmullw(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+  void vpmulld(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Shift left packed integers
+  void psllw(XMMRegister dst, int shift);
+  void pslld(XMMRegister dst, int shift);
+  void psllq(XMMRegister dst, int shift);
+  void psllw(XMMRegister dst, XMMRegister shift);
+  void pslld(XMMRegister dst, XMMRegister shift);
+  void psllq(XMMRegister dst, XMMRegister shift);
+  void vpsllw(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpslld(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsllq(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsllw(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+  void vpslld(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+  void vpsllq(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+
+  // Logical shift right packed integers
+  void psrlw(XMMRegister dst, int shift);
+  void psrld(XMMRegister dst, int shift);
+  void psrlq(XMMRegister dst, int shift);
+  void psrlw(XMMRegister dst, XMMRegister shift);
+  void psrld(XMMRegister dst, XMMRegister shift);
+  void psrlq(XMMRegister dst, XMMRegister shift);
+  void vpsrlw(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsrld(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsrlq(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsrlw(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+  void vpsrld(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+  void vpsrlq(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+
+  // Arithmetic shift right packed integers (only shorts and ints, no instructions for longs)
+  void psraw(XMMRegister dst, int shift);
+  void psrad(XMMRegister dst, int shift);
+  void psraw(XMMRegister dst, XMMRegister shift);
+  void psrad(XMMRegister dst, XMMRegister shift);
+  void vpsraw(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsrad(XMMRegister dst, XMMRegister src, int shift, bool vector256);
+  void vpsraw(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+  void vpsrad(XMMRegister dst, XMMRegister src, XMMRegister shift, bool vector256);
+
+  // And packed integers
+  void pand(XMMRegister dst, XMMRegister src);
+  void vpand(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpand(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Or packed integers
+  void por(XMMRegister dst, XMMRegister src);
+  void vpor(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpor(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Xor packed integers
+  void pxor(XMMRegister dst, XMMRegister src);
   void vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256);
+  void vpxor(XMMRegister dst, XMMRegister nds, Address src, bool vector256);
+
+  // Copy low 128bit into high 128bit of YMM registers.
   void vinsertf128h(XMMRegister dst, XMMRegister nds, XMMRegister src);
   void vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src);
 
@@ -1940,6 +2055,7 @@ class MacroAssembler: public Assembler {
   void load_heap_oop(Register dst, Address src);
   void load_heap_oop_not_null(Register dst, Address src);
   void store_heap_oop(Address dst, Register src);
+  void cmp_heap_oop(Register src1, Address src2, Register tmp = noreg);
 
   // Used for storing NULL. All other oop constants should be
   // stored using routines that take a jobject.
@@ -2117,6 +2233,11 @@ class MacroAssembler: public Assembler {
                                Register scan_temp,
                                Label& no_such_interface);
 
+  // virtual method calling
+  void lookup_virtual_method(Register recv_klass,
+                             RegisterOrConstant vtable_index,
+                             Register method_result);
+
   // Test sub_klass against super_klass, with fast and slow paths.
 
   // The fast path produces a tri-state answer: yes / no / maybe-slow.
@@ -2152,15 +2273,8 @@ class MacroAssembler: public Assembler {
                            Label& L_success);
 
   // method handles (JSR 292)
-  void check_method_handle_type(Register mtype_reg, Register mh_reg,
-                                Register temp_reg,
-                                Label& wrong_method_type);
-  void load_method_handle_vmslots(Register vmslots_reg, Register mh_reg,
-                                  Register temp_reg);
-  void jump_to_method_handle_entry(Register mh_reg, Register temp_reg);
   Address argument_address(RegisterOrConstant arg_slot, int extra_slot_offset = 0);
 
-
   //----
   void set_word_if_not_zero(Register reg); // sets reg to 1 if not zero, otherwise 0
 
@@ -2179,8 +2293,13 @@ class MacroAssembler: public Assembler {
   // prints msg and continues
   void warn(const char* msg);
 
+  // dumps registers and other state
+  void print_state();
+
   static void debug32(int rdi, int rsi, int rbp, int rsp, int rbx, int rdx, int rcx, int rax, int eip, char* msg);
   static void debug64(char* msg, int64_t pc, int64_t regs[]);
+  static void print_state32(int rdi, int rsi, int rbp, int rsp, int rbx, int rdx, int rcx, int rax, int eip);
+  static void print_state64(int64_t pc, int64_t regs[]);
 
   void os_breakpoint();
 
@@ -2528,11 +2647,13 @@ public:
   void vaddss(XMMRegister dst, XMMRegister nds, Address src)     { Assembler::vaddss(dst, nds, src); }
   void vaddss(XMMRegister dst, XMMRegister nds, AddressLiteral src);
 
-  void vandpd(XMMRegister dst, XMMRegister nds, Address src)     { Assembler::vandpd(dst, nds, src); }
+  void vandpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256) { Assembler::vandpd(dst, nds, src, vector256); }
-  void vandpd(XMMRegister dst, XMMRegister nds, AddressLiteral src);
+  void vandpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256)     { Assembler::vandpd(dst, nds, src, vector256); }
+  void vandpd(XMMRegister dst, XMMRegister nds, AddressLiteral src, bool vector256);
 
-  void vandps(XMMRegister dst, XMMRegister nds, Address src)     { Assembler::vandps(dst, nds, src); }
+  void vandps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256) { Assembler::vandps(dst, nds, src, vector256); }
-  void vandps(XMMRegister dst, XMMRegister nds, AddressLiteral src);
+  void vandps(XMMRegister dst, XMMRegister nds, Address src, bool vector256)     { Assembler::vandps(dst, nds, src, vector256); }
+  void vandps(XMMRegister dst, XMMRegister nds, AddressLiteral src, bool vector256);
 
   void vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src) { Assembler::vdivsd(dst, nds, src); }
   void vdivsd(XMMRegister dst, XMMRegister nds, Address src)     { Assembler::vdivsd(dst, nds, src); }
@@ -2561,12 +2682,12 @@ public:
   // AVX Vector instructions
 
   void vxorpd(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256) { Assembler::vxorpd(dst, nds, src, vector256); }
-  void vxorpd(XMMRegister dst, XMMRegister nds, Address src) { Assembler::vxorpd(dst, nds, src); }
+  void vxorpd(XMMRegister dst, XMMRegister nds, Address src, bool vector256) { Assembler::vxorpd(dst, nds, src, vector256); }
-  void vxorpd(XMMRegister dst, XMMRegister nds, AddressLiteral src);
+  void vxorpd(XMMRegister dst, XMMRegister nds, AddressLiteral src, bool vector256);
 
   void vxorps(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256) { Assembler::vxorps(dst, nds, src, vector256); }
-  void vxorps(XMMRegister dst, XMMRegister nds, Address src) { Assembler::vxorps(dst, nds, src); }
+  void vxorps(XMMRegister dst, XMMRegister nds, Address src, bool vector256) { Assembler::vxorps(dst, nds, src, vector256); }
-  void vxorps(XMMRegister dst, XMMRegister nds, AddressLiteral src);
+  void vxorps(XMMRegister dst, XMMRegister nds, AddressLiteral src, bool vector256);
 
   void vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, bool vector256) {
     if (UseAVX > 1 || !vector256) // vpxor 256 bit is available only in AVX2
@@ -2574,6 +2695,12 @@ public:
     else
       Assembler::vxorpd(dst, nds, src, vector256);
   }
+  void vpxor(XMMRegister dst, XMMRegister nds, Address src, bool vector256) {
+    if (UseAVX > 1 || !vector256) // vpxor 256 bit is available only in AVX2
+      Assembler::vpxor(dst, nds, src, vector256);
+    else
+      Assembler::vxorpd(dst, nds, src, vector256);
+  }
 
   // Move packed integer values from low 128 bit to hign 128 bit in 256 bit vector.
   void vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src) {

hotspot/src/cpu/x86/vm/c1_CodeStubs_x86.cpp

@@ -488,68 +488,6 @@ void G1PreBarrierStub::emit_code(LIR_Assembler* ce) {
 
 }
 
-void G1UnsafeGetObjSATBBarrierStub::emit_code(LIR_Assembler* ce) {
-  // At this point we know that offset == referent_offset.
-  //
-  // So we might have to emit:
-  //   if (src == null) goto continuation.
-  //
-  // and we definitely have to emit:
-  //   if (klass(src).reference_type == REF_NONE) goto continuation
-  //   if (!marking_active) goto continuation
-  //   if (pre_val == null) goto continuation
-  //   call pre_barrier(pre_val)
-  //   goto continuation
-  //
-  __ bind(_entry);
-
-  assert(src()->is_register(), "sanity");
-  Register src_reg = src()->as_register();
-
-  if (gen_src_check()) {
-    // The original src operand was not a constant.
-    // Generate src == null?
-    __ cmpptr(src_reg, (int32_t) NULL_WORD);
-    __ jcc(Assembler::equal, _continuation);
-  }
-
-  // Generate src->_klass->_reference_type == REF_NONE)?
-  assert(tmp()->is_register(), "sanity");
-  Register tmp_reg = tmp()->as_register();
-
-  __ load_klass(tmp_reg, src_reg);
-
-  Address ref_type_adr(tmp_reg, instanceKlass::reference_type_offset());
-  __ cmpb(ref_type_adr, REF_NONE);
-  __ jcc(Assembler::equal, _continuation);
-
-  // Is marking active?
-  assert(thread()->is_register(), "precondition");
-  Register thread_reg = thread()->as_pointer_register();
-
-  Address in_progress(thread_reg, in_bytes(JavaThread::satb_mark_queue_offset() +
-                                       PtrQueue::byte_offset_of_active()));
-
-  if (in_bytes(PtrQueue::byte_width_of_active()) == 4) {
-    __ cmpl(in_progress, 0);
-  } else {
-    assert(in_bytes(PtrQueue::byte_width_of_active()) == 1, "Assumption");
-    __ cmpb(in_progress, 0);
-  }
-  __ jcc(Assembler::equal, _continuation);
-
-  // val == null?
-  assert(val()->is_register(), "Precondition.");
-  Register val_reg = val()->as_register();
-
-  __ cmpptr(val_reg, (int32_t) NULL_WORD);
-  __ jcc(Assembler::equal, _continuation);
-
-  ce->store_parameter(val()->as_register(), 0);
-  __ call(RuntimeAddress(Runtime1::entry_for(Runtime1::g1_pre_barrier_slow_id)));
-  __ jmp(_continuation);
-}
-
 jbyte* G1PostBarrierStub::_byte_map_base = NULL;
 
 jbyte* G1PostBarrierStub::byte_map_base_slow() {

hotspot/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp

@@ -3508,6 +3508,7 @@ void LIR_Assembler::emit_lock(LIR_OpLock* op) {
 void LIR_Assembler::emit_profile_call(LIR_OpProfileCall* op) {
   ciMethod* method = op->profiled_method();
   int bci          = op->profiled_bci();
+  ciMethod* callee = op->profiled_callee();
 
   // Update counter for all call types
   ciMethodData* md = method->method_data_or_null();
@@ -3519,9 +3520,11 @@ void LIR_Assembler::emit_profile_call(LIR_OpProfileCall* op) {
   __ movoop(mdo, md->constant_encoding());
   Address counter_addr(mdo, md->byte_offset_of_slot(data, CounterData::count_offset()));
   Bytecodes::Code bc = method->java_code_at_bci(bci);
+  const bool callee_is_static = callee->is_loaded() && callee->is_static();
   // Perform additional virtual call profiling for invokevirtual and
   // invokeinterface bytecodes
   if ((bc == Bytecodes::_invokevirtual || bc == Bytecodes::_invokeinterface) &&
+      !callee_is_static &&  // required for optimized MH invokes
       C1ProfileVirtualCalls) {
     assert(op->recv()->is_single_cpu(), "recv must be allocated");
     Register recv = op->recv()->as_register();

hotspot/src/cpu/x86/vm/cppInterpreter_x86.cpp

@@ -871,9 +871,9 @@ address InterpreterGenerator::generate_accessor_entry(void) {
     // Need to differentiate between igetfield, agetfield, bgetfield etc.
     // because they are different sizes.
     // Use the type from the constant pool cache
-    __ shrl(rdx, ConstantPoolCacheEntry::tosBits);
+    __ shrl(rdx, ConstantPoolCacheEntry::tos_state_shift);
-    // Make sure we don't need to mask rdx for tosBits after the above shift
+    // Make sure we don't need to mask rdx after the above shift
-    ConstantPoolCacheEntry::verify_tosBits();
+    ConstantPoolCacheEntry::verify_tos_state_shift();
 #ifdef _LP64
     Label notObj;
     __ cmpl(rdx, atos);

hotspot/src/cpu/x86/vm/frame_x86.cpp

@@ -439,7 +439,6 @@ frame frame::sender_for_interpreter_frame(RegisterMap* map) const {
 // frame::sender_for_compiled_frame
 frame frame::sender_for_compiled_frame(RegisterMap* map) const {
   assert(map != NULL, "map must be set");
-  assert(!is_ricochet_frame(), "caller must handle this");
 
   // frame owned by optimizing compiler
   assert(_cb->frame_size() >= 0, "must have non-zero frame size");
@@ -483,7 +482,6 @@ frame frame::sender(RegisterMap* map) const {
   if (is_entry_frame())       return sender_for_entry_frame(map);
   if (is_interpreted_frame()) return sender_for_interpreter_frame(map);
   assert(_cb == CodeCache::find_blob(pc()),"Must be the same");
-  if (is_ricochet_frame())    return sender_for_ricochet_frame(map);
 
   if (_cb != NULL) {
     return sender_for_compiled_frame(map);
@@ -658,9 +656,7 @@ intptr_t* frame::interpreter_frame_tos_at(jint offset) const {
   values.describe(frame_no, fp() + frame::name##_offset, #name)
 
 void frame::describe_pd(FrameValues& values, int frame_no) {
-  if (is_ricochet_frame()) {
+  if (is_interpreted_frame()) {
-    MethodHandles::RicochetFrame::describe(this, values, frame_no);
-  } else if (is_interpreted_frame()) {
     DESCRIBE_FP_OFFSET(interpreter_frame_sender_sp);
     DESCRIBE_FP_OFFSET(interpreter_frame_last_sp);
     DESCRIBE_FP_OFFSET(interpreter_frame_method);
@@ -682,12 +678,7 @@ intptr_t* frame::real_fp() const {
   if (_cb != NULL) {
     // use the frame size if valid
     int size = _cb->frame_size();
-    if ((size > 0) &&
+    if (size > 0) {
-        (! is_ricochet_frame())) {
-      // Work-around: ricochet explicitly excluded because frame size is not
-      // constant for the ricochet blob but its frame_size could not, for
-      // some reasons, be declared as <= 0. This potentially confusing
-      // size declaration should be fixed as another CR.
       return unextended_sp() + size;
     }
   }

hotspot/src/cpu/x86/vm/globals_x86.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2000, 2012, 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
@@ -78,4 +78,53 @@ define_pd_global(bool, UseMembar,            false);
 
 // GC Ergo Flags
 define_pd_global(intx, CMSYoungGenPerWorker, 64*M);  // default max size of CMS young gen, per GC worker thread
+
+#define ARCH_FLAGS(develop, product, diagnostic, experimental, notproduct) \
+                                                                            \
+  develop(bool, IEEEPrecision, true,                                        \
+          "Enables IEEE precision (for INTEL only)")                        \
+                                                                            \
+  product(intx, FenceInstruction, 0,                                        \
+          "(Unsafe,Unstable) Experimental")                                 \
+                                                                            \
+  product(intx,  ReadPrefetchInstr, 0,                                      \
+          "Prefetch instruction to prefetch ahead")                         \
+                                                                            \
+  product(bool, UseStoreImmI16, true,                                       \
+          "Use store immediate 16-bits value instruction on x86")           \
+                                                                            \
+  product(intx, UseAVX, 99,                                                 \
+          "Highest supported AVX instructions set on x86/x64")              \
+                                                                            \
+  diagnostic(bool, UseIncDec, true,                                         \
+          "Use INC, DEC instructions on x86")                               \
+                                                                            \
+  product(bool, UseNewLongLShift, false,                                    \
+          "Use optimized bitwise shift left")                               \
+                                                                            \
+  product(bool, UseAddressNop, false,                                       \
+          "Use '0F 1F [addr]' NOP instructions on x86 cpus")                \
+                                                                            \
+  product(bool, UseXmmLoadAndClearUpper, true,                              \
+          "Load low part of XMM register and clear upper part")             \
+                                                                            \
+  product(bool, UseXmmRegToRegMoveAll, false,                               \
+          "Copy all XMM register bits when moving value between registers") \
+                                                                            \
+  product(bool, UseXmmI2D, false,                                           \
+          "Use SSE2 CVTDQ2PD instruction to convert Integer to Double")     \
+                                                                            \
+  product(bool, UseXmmI2F, false,                                           \
+          "Use SSE2 CVTDQ2PS instruction to convert Integer to Float")      \
+                                                                            \
+  product(bool, UseUnalignedLoadStores, false,                              \
+          "Use SSE2 MOVDQU instruction for Arraycopy")                      \
+                                                                            \
+  /* assembler */                                                           \
+  product(bool, Use486InstrsOnly, false,                                    \
+          "Use 80486 Compliant instruction subset")                         \
+                                                                            \
+  product(bool, UseCountLeadingZerosInstruction, false,                     \
+          "Use count leading zeros instruction")                            \
+
 #endif // CPU_X86_VM_GLOBALS_X86_HPP

hotspot/src/cpu/x86/vm/interp_masm_x86_32.cpp

@@ -253,8 +253,12 @@ void InterpreterMacroAssembler::get_cache_and_index_and_bytecode_at_bcp(Register
   get_cache_and_index_at_bcp(cache, index, bcp_offset, index_size);
   movptr(bytecode, Address(cache, index, Address::times_ptr, constantPoolCacheOopDesc::base_offset() + ConstantPoolCacheEntry::indices_offset()));
   const int shift_count = (1 + byte_no) * BitsPerByte;
+  assert((byte_no == TemplateTable::f1_byte && shift_count == ConstantPoolCacheEntry::bytecode_1_shift) ||
+         (byte_no == TemplateTable::f2_byte && shift_count == ConstantPoolCacheEntry::bytecode_2_shift),
+         "correct shift count");
   shrptr(bytecode, shift_count);
-  andptr(bytecode, 0xFF);
+  assert(ConstantPoolCacheEntry::bytecode_1_mask == ConstantPoolCacheEntry::bytecode_2_mask, "common mask");
+  andptr(bytecode, ConstantPoolCacheEntry::bytecode_1_mask);
 }
 
 

hotspot/src/cpu/x86/vm/interp_masm_x86_64.cpp

@@ -256,8 +256,12 @@ void InterpreterMacroAssembler::get_cache_and_index_and_bytecode_at_bcp(Register
   // little-endian machines allow us that.
   movl(bytecode, Address(cache, index, Address::times_ptr, constantPoolCacheOopDesc::base_offset() + ConstantPoolCacheEntry::indices_offset()));
   const int shift_count = (1 + byte_no) * BitsPerByte;
+  assert((byte_no == TemplateTable::f1_byte && shift_count == ConstantPoolCacheEntry::bytecode_1_shift) ||
+         (byte_no == TemplateTable::f2_byte && shift_count == ConstantPoolCacheEntry::bytecode_2_shift),
+         "correct shift count");
   shrl(bytecode, shift_count);
-  andl(bytecode, 0xFF);
+  assert(ConstantPoolCacheEntry::bytecode_1_mask == ConstantPoolCacheEntry::bytecode_2_mask, "common mask");
+  andl(bytecode, ConstantPoolCacheEntry::bytecode_1_mask);
 }
 
 

hotspot/src/cpu/x86/vm/interpreterGenerator_x86.hpp

@@ -35,7 +35,6 @@
   address generate_normal_entry(bool synchronized);
   address generate_native_entry(bool synchronized);
   address generate_abstract_entry(void);
-  address generate_method_handle_entry(void);
   address generate_math_entry(AbstractInterpreter::MethodKind kind);
   address generate_empty_entry(void);
   address generate_accessor_entry(void);

hotspot/src/cpu/x86/vm/interpreter_x86_32.cpp

@@ -243,18 +243,6 @@ address InterpreterGenerator::generate_abstract_entry(void) {
 }
 
 
-// Method handle invoker
-// Dispatch a method of the form java.lang.invoke.MethodHandles::invoke(...)
-address InterpreterGenerator::generate_method_handle_entry(void) {
-  if (!EnableInvokeDynamic) {
-    return generate_abstract_entry();
-  }
-
-  address entry_point = MethodHandles::generate_method_handle_interpreter_entry(_masm);
-
-  return entry_point;
-}
-
 void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
 
   // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in

hotspot/src/cpu/x86/vm/interpreter_x86_64.cpp

@@ -325,19 +325,6 @@ address InterpreterGenerator::generate_abstract_entry(void) {
 }
 
 
-// Method handle invoker
-// Dispatch a method of the form java.lang.invoke.MethodHandles::invoke(...)
-address InterpreterGenerator::generate_method_handle_entry(void) {
-  if (!EnableInvokeDynamic) {
-    return generate_abstract_entry();
-  }
-
-  address entry_point = MethodHandles::generate_method_handle_interpreter_entry(_masm);
-
-  return entry_point;
-}
-
-
 // Empty method, generate a very fast return.
 
 address InterpreterGenerator::generate_empty_entry(void) {

hotspot/src/cpu/x86/vm/methodHandles_x86.hpp

@@ -27,266 +27,12 @@
 
 // Adapters
 enum /* platform_dependent_constants */ {
-  adapter_code_size = NOT_LP64(16000 DEBUG_ONLY(+ 15000)) LP64_ONLY(32000 DEBUG_ONLY(+ 120000))
+  adapter_code_size = NOT_LP64(16000 DEBUG_ONLY(+ 25000)) LP64_ONLY(32000 DEBUG_ONLY(+ 150000))
-};
-
-public:
-
-// The stack just after the recursive call from a ricochet frame
-// looks something like this.  Offsets are marked in words, not bytes.
-// rsi (r13 on LP64) is part of the interpreter calling sequence
-// which tells the callee where is my real rsp (for frame walking).
-// (...lower memory addresses)
-// rsp:     [ return pc                 ]   always the global RicochetBlob::bounce_addr
-// rsp+1:   [ recursive arg N           ]
-// rsp+2:   [ recursive arg N-1         ]
-// ...
-// rsp+N:   [ recursive arg 1           ]
-// rsp+N+1: [ recursive method handle   ]
-// ...
-// rbp-6:   [ cleanup continuation pc   ]   <-- (struct RicochetFrame)
-// rbp-5:   [ saved target MH           ]   the MH we will call on the saved args
-// rbp-4:   [ saved args layout oop     ]   an int[] array which describes argument layout
-// rbp-3:   [ saved args pointer        ]   address of transformed adapter arg M (slot 0)
-// rbp-2:   [ conversion                ]   information about how the return value is used
-// rbp-1:   [ exact sender sp           ]   exact TOS (rsi/r13) of original sender frame
-// rbp+0:   [ saved sender fp           ]   (for original sender of AMH)
-// rbp+1:   [ saved sender pc           ]   (back to original sender of AMH)
-// rbp+2:   [ transformed adapter arg M ]   <-- (extended TOS of original sender)
-// rbp+3:   [ transformed adapter arg M-1]
-// ...
-// rbp+M+1: [ transformed adapter arg 1 ]
-// rbp+M+2: [ padding                   ] <-- (rbp + saved args base offset)
-// ...      [ optional padding]
-// (higher memory addresses...)
-//
-// The arguments originally passed by the original sender
-// are lost, and arbitrary amounts of stack motion might have
-// happened due to argument transformation.
-// (This is done by C2I/I2C adapters and non-direct method handles.)
-// This is why there is an unpredictable amount of memory between
-// the extended and exact TOS of the sender.
-// The ricochet adapter itself will also (in general) perform
-// transformations before the recursive call.
-//
-// The transformed and saved arguments, immediately above the saved
-// return PC, are a well-formed method handle invocation ready to execute.
-// When the GC needs to walk the stack, these arguments are described
-// via the saved arg types oop, an int[] array with a private format.
-// This array is derived from the type of the transformed adapter
-// method handle, which also sits at the base of the saved argument
-// bundle.  Since the GC may not be able to fish out the int[]
-// array, so it is pushed explicitly on the stack.  This may be
-// an unnecessary expense.
-//
-// The following register conventions are significant at this point:
-// rsp       the thread stack, as always; preserved by caller
-// rsi/r13   exact TOS of recursive frame (contents of [rbp-2])
-// rcx       recursive method handle (contents of [rsp+N+1])
-// rbp       preserved by caller (not used by caller)
-// Unless otherwise specified, all registers can be blown by the call.
-//
-// If this frame must be walked, the transformed adapter arguments
-// will be found with the help of the saved arguments descriptor.
-//
-// Therefore, the descriptor must match the referenced arguments.
-// The arguments must be followed by at least one word of padding,
-// which will be necessary to complete the final method handle call.
-// That word is not treated as holding an oop.  Neither is the word
-//
-// The word pointed to by the return argument pointer is not
-// treated as an oop, even if points to a saved argument.
-// This allows the saved argument list to have a "hole" in it
-// to receive an oop from the recursive call.
-// (The hole might temporarily contain RETURN_VALUE_PLACEHOLDER.)
-//
-// When the recursive callee returns, RicochetBlob::bounce_addr will
-// immediately jump to the continuation stored in the RF.
-// This continuation will merge the recursive return value
-// into the saved argument list.  At that point, the original
-// rsi, rbp, and rsp will be reloaded, the ricochet frame will
-// disappear, and the final target of the adapter method handle
-// will be invoked on the transformed argument list.
-
-class RicochetFrame {
-  friend class MethodHandles;
-  friend class VMStructs;
-
- private:
-  intptr_t* _continuation;          // what to do when control gets back here
-  oopDesc*  _saved_target;          // target method handle to invoke on saved_args
-  oopDesc*  _saved_args_layout;     // caching point for MethodTypeForm.vmlayout cookie
-  intptr_t* _saved_args_base;       // base of pushed arguments (slot 0, arg N) (-3)
-  intptr_t  _conversion;            // misc. information from original AdapterMethodHandle (-2)
-  intptr_t* _exact_sender_sp;       // parallel to interpreter_frame_sender_sp (-1)
-  intptr_t* _sender_link;           // *must* coincide with frame::link_offset (0)
-  address   _sender_pc;             // *must* coincide with frame::return_addr_offset (1)
-
- public:
-  intptr_t* continuation() const        { return _continuation; }
-  oop       saved_target() const        { return _saved_target; }
-  oop       saved_args_layout() const   { return _saved_args_layout; }
-  intptr_t* saved_args_base() const     { return _saved_args_base; }
-  intptr_t  conversion() const          { return _conversion; }
-  intptr_t* exact_sender_sp() const     { return _exact_sender_sp; }
-  intptr_t* sender_link() const         { return _sender_link; }
-  address   sender_pc() const           { return _sender_pc; }
-
-  intptr_t* extended_sender_sp() const {
-    // The extended sender SP is above the current RicochetFrame.
-    return (intptr_t*) (((address) this) + sizeof(RicochetFrame));
-  }
-
-  intptr_t  return_value_slot_number() const {
-    return adapter_conversion_vminfo(conversion());
-  }
-  BasicType return_value_type() const {
-    return adapter_conversion_dest_type(conversion());
-  }
-  bool has_return_value_slot() const {
-    return return_value_type() != T_VOID;
-  }
-  intptr_t* return_value_slot_addr() const {
-    assert(has_return_value_slot(), "");
-    return saved_arg_slot_addr(return_value_slot_number());
-  }
-  intptr_t* saved_target_slot_addr() const {
-    return saved_arg_slot_addr(saved_args_length());
-  }
-  intptr_t* saved_arg_slot_addr(int slot) const {
-    assert(slot >= 0, "");
-    return (intptr_t*)( (address)saved_args_base() + (slot * Interpreter::stackElementSize) );
-  }
-
-  jint      saved_args_length() const;
-  jint      saved_arg_offset(int arg) const;
-
-  // GC interface
-  oop*  saved_target_addr()                     { return (oop*)&_saved_target; }
-  oop*  saved_args_layout_addr()                { return (oop*)&_saved_args_layout; }
-
-  oop  compute_saved_args_layout(bool read_cache, bool write_cache);
-
-  // Compiler/assembler interface.
-  static int continuation_offset_in_bytes()     { return offset_of(RicochetFrame, _continuation); }
-  static int saved_target_offset_in_bytes()     { return offset_of(RicochetFrame, _saved_target); }
-  static int saved_args_layout_offset_in_bytes(){ return offset_of(RicochetFrame, _saved_args_layout); }
-  static int saved_args_base_offset_in_bytes()  { return offset_of(RicochetFrame, _saved_args_base); }
-  static int conversion_offset_in_bytes()       { return offset_of(RicochetFrame, _conversion); }
-  static int exact_sender_sp_offset_in_bytes()  { return offset_of(RicochetFrame, _exact_sender_sp); }
-  static int sender_link_offset_in_bytes()      { return offset_of(RicochetFrame, _sender_link); }
-  static int sender_pc_offset_in_bytes()        { return offset_of(RicochetFrame, _sender_pc); }
-
-  // This value is not used for much, but it apparently must be nonzero.
-  static int frame_size_in_bytes()              { return sender_link_offset_in_bytes(); }
-
-#ifdef ASSERT
-  // The magic number is supposed to help find ricochet frames within the bytes of stack dumps.
-  enum { MAGIC_NUMBER_1 = 0xFEED03E, MAGIC_NUMBER_2 = 0xBEEF03E };
-  static int magic_number_1_offset_in_bytes()   { return -wordSize; }
-  static int magic_number_2_offset_in_bytes()   { return sizeof(RicochetFrame); }
-  intptr_t magic_number_1() const               { return *(intptr_t*)((address)this + magic_number_1_offset_in_bytes()); };
-  intptr_t magic_number_2() const               { return *(intptr_t*)((address)this + magic_number_2_offset_in_bytes()); };
-#endif //ASSERT
-
-  enum { RETURN_VALUE_PLACEHOLDER = (NOT_DEBUG(0) DEBUG_ONLY(42)) };
-
-  static void verify_offsets() NOT_DEBUG_RETURN;
-  void verify() const NOT_DEBUG_RETURN; // check for MAGIC_NUMBER, etc.
-  void zap_arguments() NOT_DEBUG_RETURN;
-
-  static void generate_ricochet_blob(MacroAssembler* _masm,
-                                     // output params:
-                                     int* bounce_offset,
-                                     int* exception_offset,
-                                     int* frame_size_in_words);
-
-  static void enter_ricochet_frame(MacroAssembler* _masm,
-                                   Register rcx_recv,
-                                   Register rax_argv,
-                                   address return_handler,
-                                   Register rbx_temp);
-  static void leave_ricochet_frame(MacroAssembler* _masm,
-                                   Register rcx_recv,
-                                   Register new_sp_reg,
-                                   Register sender_pc_reg);
-
-  static Address frame_address(int offset = 0) {
-    // The RicochetFrame is found by subtracting a constant offset from rbp.
-    return Address(rbp, - sender_link_offset_in_bytes() + offset);
-  }
-
-  static RicochetFrame* from_frame(const frame& fr) {
-    address bp = (address) fr.fp();
-    RicochetFrame* rf = (RicochetFrame*)(bp - sender_link_offset_in_bytes());
-    rf->verify();
-    return rf;
-  }
-
-  static void verify_clean(MacroAssembler* _masm) NOT_DEBUG_RETURN;
-
-  static void describe(const frame* fr, FrameValues& values, int frame_no) PRODUCT_RETURN;
 };
 
 // Additional helper methods for MethodHandles code generation:
 public:
   static void load_klass_from_Class(MacroAssembler* _masm, Register klass_reg);
-  static void load_conversion_vminfo(MacroAssembler* _masm, Register reg, Address conversion_field_addr);
-  static void load_conversion_dest_type(MacroAssembler* _masm, Register reg, Address conversion_field_addr);
-
-  static void load_stack_move(MacroAssembler* _masm,
-                              Register rdi_stack_move,
-                              Register rcx_amh,
-                              bool might_be_negative);
-
-  static void insert_arg_slots(MacroAssembler* _masm,
-                               RegisterOrConstant arg_slots,
-                               Register rax_argslot,
-                               Register rbx_temp, Register rdx_temp);
-
-  static void remove_arg_slots(MacroAssembler* _masm,
-                               RegisterOrConstant arg_slots,
-                               Register rax_argslot,
-                               Register rbx_temp, Register rdx_temp);
-
-  static void push_arg_slots(MacroAssembler* _masm,
-                                   Register rax_argslot,
-                                   RegisterOrConstant slot_count,
-                                   int skip_words_count,
-                                   Register rbx_temp, Register rdx_temp);
-
-  static void move_arg_slots_up(MacroAssembler* _masm,
-                                Register rbx_bottom,  // invariant
-                                Address  top_addr,    // can use rax_temp
-                                RegisterOrConstant positive_distance_in_slots,
-                                Register rax_temp, Register rdx_temp);
-
-  static void move_arg_slots_down(MacroAssembler* _masm,
-                                  Address  bottom_addr,  // can use rax_temp
-                                  Register rbx_top,      // invariant
-                                  RegisterOrConstant negative_distance_in_slots,
-                                  Register rax_temp, Register rdx_temp);
-
-  static void move_typed_arg(MacroAssembler* _masm,
-                             BasicType type, bool is_element,
-                             Address slot_dest, Address value_src,
-                             Register rbx_temp, Register rdx_temp);
-
-  static void move_return_value(MacroAssembler* _masm, BasicType type,
-                                Address return_slot);
-
-  static void verify_argslot(MacroAssembler* _masm, Register argslot_reg,
-                             const char* error_message) NOT_DEBUG_RETURN;
-
-  static void verify_argslots(MacroAssembler* _masm,
-                              RegisterOrConstant argslot_count,
-                              Register argslot_reg,
-                              bool negate_argslot,
-                              const char* error_message) NOT_DEBUG_RETURN;
-
-  static void verify_stack_move(MacroAssembler* _masm,
-                                RegisterOrConstant arg_slots,
-                                int direction) NOT_DEBUG_RETURN;
 
   static void verify_klass(MacroAssembler* _masm,
                            Register obj, KlassHandle klass,
@@ -297,9 +43,17 @@ public:
                  "reference is a MH");
   }
 
+  static void verify_ref_kind(MacroAssembler* _masm, int ref_kind, Register member_reg, Register temp) NOT_DEBUG_RETURN;
+
   // Similar to InterpreterMacroAssembler::jump_from_interpreted.
   // Takes care of special dispatch from single stepping too.
-  static void jump_from_method_handle(MacroAssembler* _masm, Register method, Register temp);
+  static void jump_from_method_handle(MacroAssembler* _masm, Register method, Register temp,
+                                      bool for_compiler_entry);
+
+  static void jump_to_lambda_form(MacroAssembler* _masm,
+                                  Register recv, Register method_temp,
+                                  Register temp2,
+                                  bool for_compiler_entry);
 
   static void trace_method_handle(MacroAssembler* _masm, const char* adaptername) PRODUCT_RETURN;
 

hotspot/src/cpu/x86/vm/sharedRuntime_x86_32.cpp

@@ -643,6 +643,19 @@ static void move_i2c_double(MacroAssembler *masm, XMMRegister r, Register saved_
   __ movdbl(r, Address(saved_sp, next_val_off));
 }
 
+static void range_check(MacroAssembler* masm, Register pc_reg, Register temp_reg,
+                        address code_start, address code_end,
+                        Label& L_ok) {
+  Label L_fail;
+  __ lea(temp_reg, ExternalAddress(code_start));
+  __ cmpptr(pc_reg, temp_reg);
+  __ jcc(Assembler::belowEqual, L_fail);
+  __ lea(temp_reg, ExternalAddress(code_end));
+  __ cmpptr(pc_reg, temp_reg);
+  __ jcc(Assembler::below, L_ok);
+  __ bind(L_fail);
+}
+
 static void gen_i2c_adapter(MacroAssembler *masm,
                             int total_args_passed,
                             int comp_args_on_stack,
@@ -653,9 +666,53 @@ static void gen_i2c_adapter(MacroAssembler *masm,
   // we may do a i2c -> c2i transition if we lose a race where compiled
   // code goes non-entrant while we get args ready.
 
+  // Adapters can be frameless because they do not require the caller
+  // to perform additional cleanup work, such as correcting the stack pointer.
+  // An i2c adapter is frameless because the *caller* frame, which is interpreted,
+  // routinely repairs its own stack pointer (from interpreter_frame_last_sp),
+  // even if a callee has modified the stack pointer.
+  // A c2i adapter is frameless because the *callee* frame, which is interpreted,
+  // routinely repairs its caller's stack pointer (from sender_sp, which is set
+  // up via the senderSP register).
+  // In other words, if *either* the caller or callee is interpreted, we can
+  // get the stack pointer repaired after a call.
+  // This is why c2i and i2c adapters cannot be indefinitely composed.
+  // In particular, if a c2i adapter were to somehow call an i2c adapter,
+  // both caller and callee would be compiled methods, and neither would
+  // clean up the stack pointer changes performed by the two adapters.
+  // If this happens, control eventually transfers back to the compiled
+  // caller, but with an uncorrected stack, causing delayed havoc.
+
   // Pick up the return address
   __ movptr(rax, Address(rsp, 0));
 
+  if (VerifyAdapterCalls &&
+      (Interpreter::code() != NULL || StubRoutines::code1() != NULL)) {
+    // So, let's test for cascading c2i/i2c adapters right now.
+    //  assert(Interpreter::contains($return_addr) ||
+    //         StubRoutines::contains($return_addr),
+    //         "i2c adapter must return to an interpreter frame");
+    __ block_comment("verify_i2c { ");
+    Label L_ok;
+    if (Interpreter::code() != NULL)
+      range_check(masm, rax, rdi,
+                  Interpreter::code()->code_start(), Interpreter::code()->code_end(),
+                  L_ok);
+    if (StubRoutines::code1() != NULL)
+      range_check(masm, rax, rdi,
+                  StubRoutines::code1()->code_begin(), StubRoutines::code1()->code_end(),
+                  L_ok);
+    if (StubRoutines::code2() != NULL)
+      range_check(masm, rax, rdi,
+                  StubRoutines::code2()->code_begin(), StubRoutines::code2()->code_end(),
+                  L_ok);
+    const char* msg = "i2c adapter must return to an interpreter frame";
+    __ block_comment(msg);
+    __ stop(msg);
+    __ bind(L_ok);
+    __ block_comment("} verify_i2ce ");
+  }
+
   // Must preserve original SP for loading incoming arguments because
   // we need to align the outgoing SP for compiled code.
   __ movptr(rdi, rsp);
@@ -1293,6 +1350,89 @@ static void unpack_array_argument(MacroAssembler* masm, VMRegPair reg, BasicType
   __ bind(done);
 }
 
+static void verify_oop_args(MacroAssembler* masm,
+                            int total_args_passed,
+                            const BasicType* sig_bt,
+                            const VMRegPair* regs) {
+  Register temp_reg = rbx;  // not part of any compiled calling seq
+  if (VerifyOops) {
+    for (int i = 0; i < total_args_passed; i++) {
+      if (sig_bt[i] == T_OBJECT ||
+          sig_bt[i] == T_ARRAY) {
+        VMReg r = regs[i].first();
+        assert(r->is_valid(), "bad oop arg");
+        if (r->is_stack()) {
+          __ movptr(temp_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+          __ verify_oop(temp_reg);
+        } else {
+          __ verify_oop(r->as_Register());
+        }
+      }
+    }
+  }
+}
+
+static void gen_special_dispatch(MacroAssembler* masm,
+                                 int total_args_passed,
+                                 int comp_args_on_stack,
+                                 vmIntrinsics::ID special_dispatch,
+                                 const BasicType* sig_bt,
+                                 const VMRegPair* regs) {
+  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+
+  // Now write the args into the outgoing interpreter space
+  bool     has_receiver   = false;
+  Register receiver_reg   = noreg;
+  int      member_arg_pos = -1;
+  Register member_reg     = noreg;
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  if (ref_kind != 0) {
+    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_reg = rbx;  // known to be free at this point
+    has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
+  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+    has_receiver = true;
+  } else {
+    guarantee(false, err_msg("special_dispatch=%d", special_dispatch));
+  }
+
+  if (member_reg != noreg) {
+    // Load the member_arg into register, if necessary.
+    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
+    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    VMReg r = regs[member_arg_pos].first();
+    assert(r->is_valid(), "bad member arg");
+    if (r->is_stack()) {
+      __ movptr(member_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+    } else {
+      // no data motion is needed
+      member_reg = r->as_Register();
+    }
+  }
+
+  if (has_receiver) {
+    // Make sure the receiver is loaded into a register.
+    assert(total_args_passed > 0, "oob");
+    assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
+    VMReg r = regs[0].first();
+    assert(r->is_valid(), "bad receiver arg");
+    if (r->is_stack()) {
+      // Porting note:  This assumes that compiled calling conventions always
+      // pass the receiver oop in a register.  If this is not true on some
+      // platform, pick a temp and load the receiver from stack.
+      assert(false, "receiver always in a register");
+      receiver_reg = rcx;  // known to be free at this point
+      __ movptr(receiver_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+    } else {
+      // no data motion is needed
+      receiver_reg = r->as_Register();
+    }
+  }
+
+  // Figure out which address we are really jumping to:
+  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+                                                 receiver_reg, member_reg, /*for_compiler_entry:*/ true);
+}
 
 // ---------------------------------------------------------------------------
 // Generate a native wrapper for a given method.  The method takes arguments
@@ -1331,6 +1471,29 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler *masm,
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
+  if (method->is_method_handle_intrinsic()) {
+    vmIntrinsics::ID iid = method->intrinsic_id();
+    intptr_t start = (intptr_t)__ pc();
+    int vep_offset = ((intptr_t)__ pc()) - start;
+    gen_special_dispatch(masm,
+                         total_in_args,
+                         comp_args_on_stack,
+                         method->intrinsic_id(),
+                         in_sig_bt,
+                         in_regs);
+    int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
+    __ flush();
+    int stack_slots = SharedRuntime::out_preserve_stack_slots();  // no out slots at all, actually
+    return nmethod::new_native_nmethod(method,
+                                       compile_id,
+                                       masm->code(),
+                                       vep_offset,
+                                       frame_complete,
+                                       stack_slots / VMRegImpl::slots_per_word,
+                                       in_ByteSize(-1),
+                                       in_ByteSize(-1),
+                                       (OopMapSet*)NULL);
+  }
   bool is_critical_native = true;
   address native_func = method->critical_native_function();
   if (native_func == NULL) {
@@ -1436,7 +1599,7 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler *masm,
       if (in_regs[i].first()->is_Register()) {
         const Register reg = in_regs[i].first()->as_Register();
         switch (in_sig_bt[i]) {
-          case T_ARRAY:
+          case T_ARRAY:  // critical array (uses 2 slots on LP64)
           case T_BOOLEAN:
           case T_BYTE:
           case T_SHORT:

hotspot/src/cpu/x86/vm/sharedRuntime_x86_64.cpp

@@ -590,6 +590,19 @@ static void gen_c2i_adapter(MacroAssembler *masm,
   __ jmp(rcx);
 }
 
+static void range_check(MacroAssembler* masm, Register pc_reg, Register temp_reg,
+                        address code_start, address code_end,
+                        Label& L_ok) {
+  Label L_fail;
+  __ lea(temp_reg, ExternalAddress(code_start));
+  __ cmpptr(pc_reg, temp_reg);
+  __ jcc(Assembler::belowEqual, L_fail);
+  __ lea(temp_reg, ExternalAddress(code_end));
+  __ cmpptr(pc_reg, temp_reg);
+  __ jcc(Assembler::below, L_ok);
+  __ bind(L_fail);
+}
+
 static void gen_i2c_adapter(MacroAssembler *masm,
                             int total_args_passed,
                             int comp_args_on_stack,
@@ -605,9 +618,53 @@ static void gen_i2c_adapter(MacroAssembler *masm,
   // save code can segv when fxsave instructions find improperly
   // aligned stack pointer.
 
+  // Adapters can be frameless because they do not require the caller
+  // to perform additional cleanup work, such as correcting the stack pointer.
+  // An i2c adapter is frameless because the *caller* frame, which is interpreted,
+  // routinely repairs its own stack pointer (from interpreter_frame_last_sp),
+  // even if a callee has modified the stack pointer.
+  // A c2i adapter is frameless because the *callee* frame, which is interpreted,
+  // routinely repairs its caller's stack pointer (from sender_sp, which is set
+  // up via the senderSP register).
+  // In other words, if *either* the caller or callee is interpreted, we can
+  // get the stack pointer repaired after a call.
+  // This is why c2i and i2c adapters cannot be indefinitely composed.
+  // In particular, if a c2i adapter were to somehow call an i2c adapter,
+  // both caller and callee would be compiled methods, and neither would
+  // clean up the stack pointer changes performed by the two adapters.
+  // If this happens, control eventually transfers back to the compiled
+  // caller, but with an uncorrected stack, causing delayed havoc.
+
   // Pick up the return address
   __ movptr(rax, Address(rsp, 0));
 
+  if (VerifyAdapterCalls &&
+      (Interpreter::code() != NULL || StubRoutines::code1() != NULL)) {
+    // So, let's test for cascading c2i/i2c adapters right now.
+    //  assert(Interpreter::contains($return_addr) ||
+    //         StubRoutines::contains($return_addr),
+    //         "i2c adapter must return to an interpreter frame");
+    __ block_comment("verify_i2c { ");
+    Label L_ok;
+    if (Interpreter::code() != NULL)
+      range_check(masm, rax, r11,
+                  Interpreter::code()->code_start(), Interpreter::code()->code_end(),
+                  L_ok);
+    if (StubRoutines::code1() != NULL)
+      range_check(masm, rax, r11,
+                  StubRoutines::code1()->code_begin(), StubRoutines::code1()->code_end(),
+                  L_ok);
+    if (StubRoutines::code2() != NULL)
+      range_check(masm, rax, r11,
+                  StubRoutines::code2()->code_begin(), StubRoutines::code2()->code_end(),
+                  L_ok);
+    const char* msg = "i2c adapter must return to an interpreter frame";
+    __ block_comment(msg);
+    __ stop(msg);
+    __ bind(L_ok);
+    __ block_comment("} verify_i2ce ");
+  }
+
   // Must preserve original SP for loading incoming arguments because
   // we need to align the outgoing SP for compiled code.
   __ movptr(r11, rsp);
@@ -1366,6 +1423,14 @@ static void unpack_array_argument(MacroAssembler* masm, VMRegPair reg, BasicType
 }
 
 
+// Different signatures may require very different orders for the move
+// to avoid clobbering other arguments.  There's no simple way to
+// order them safely.  Compute a safe order for issuing stores and
+// break any cycles in those stores.  This code is fairly general but
+// it's not necessary on the other platforms so we keep it in the
+// platform dependent code instead of moving it into a shared file.
+// (See bugs 7013347 & 7145024.)
+// Note that this code is specific to LP64.
 class ComputeMoveOrder: public StackObj {
   class MoveOperation: public ResourceObj {
     friend class ComputeMoveOrder;
@@ -1532,6 +1597,89 @@ class ComputeMoveOrder: public StackObj {
   }
 };
 
+static void verify_oop_args(MacroAssembler* masm,
+                            int total_args_passed,
+                            const BasicType* sig_bt,
+                            const VMRegPair* regs) {
+  Register temp_reg = rbx;  // not part of any compiled calling seq
+  if (VerifyOops) {
+    for (int i = 0; i < total_args_passed; i++) {
+      if (sig_bt[i] == T_OBJECT ||
+          sig_bt[i] == T_ARRAY) {
+        VMReg r = regs[i].first();
+        assert(r->is_valid(), "bad oop arg");
+        if (r->is_stack()) {
+          __ movptr(temp_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+          __ verify_oop(temp_reg);
+        } else {
+          __ verify_oop(r->as_Register());
+        }
+      }
+    }
+  }
+}
+
+static void gen_special_dispatch(MacroAssembler* masm,
+                                 int total_args_passed,
+                                 int comp_args_on_stack,
+                                 vmIntrinsics::ID special_dispatch,
+                                 const BasicType* sig_bt,
+                                 const VMRegPair* regs) {
+  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+
+  // Now write the args into the outgoing interpreter space
+  bool     has_receiver   = false;
+  Register receiver_reg   = noreg;
+  int      member_arg_pos = -1;
+  Register member_reg     = noreg;
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  if (ref_kind != 0) {
+    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_reg = rbx;  // known to be free at this point
+    has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
+  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+    has_receiver = true;
+  } else {
+    guarantee(false, err_msg("special_dispatch=%d", special_dispatch));
+  }
+
+  if (member_reg != noreg) {
+    // Load the member_arg into register, if necessary.
+    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
+    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    VMReg r = regs[member_arg_pos].first();
+    assert(r->is_valid(), "bad member arg");
+    if (r->is_stack()) {
+      __ movptr(member_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+    } else {
+      // no data motion is needed
+      member_reg = r->as_Register();
+    }
+  }
+
+  if (has_receiver) {
+    // Make sure the receiver is loaded into a register.
+    assert(total_args_passed > 0, "oob");
+    assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
+    VMReg r = regs[0].first();
+    assert(r->is_valid(), "bad receiver arg");
+    if (r->is_stack()) {
+      // Porting note:  This assumes that compiled calling conventions always
+      // pass the receiver oop in a register.  If this is not true on some
+      // platform, pick a temp and load the receiver from stack.
+      assert(false, "receiver always in a register");
+      receiver_reg = j_rarg0;  // known to be free at this point
+      __ movptr(receiver_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
+    } else {
+      // no data motion is needed
+      receiver_reg = r->as_Register();
+    }
+  }
+
+  // Figure out which address we are really jumping to:
+  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+                                                 receiver_reg, member_reg, /*for_compiler_entry:*/ true);
+}
 
 // ---------------------------------------------------------------------------
 // Generate a native wrapper for a given method.  The method takes arguments
@@ -1539,6 +1687,29 @@ class ComputeMoveOrder: public StackObj {
 // convention (handlizes oops, etc), transitions to native, makes the call,
 // returns to java state (possibly blocking), unhandlizes any result and
 // returns.
+//
+// Critical native functions are a shorthand for the use of
+// GetPrimtiveArrayCritical and disallow the use of any other JNI
+// functions.  The wrapper is expected to unpack the arguments before
+// passing them to the callee and perform checks before and after the
+// native call to ensure that they GC_locker
+// lock_critical/unlock_critical semantics are followed.  Some other
+// parts of JNI setup are skipped like the tear down of the JNI handle
+// block and the check for pending exceptions it's impossible for them
+// to be thrown.
+//
+// They are roughly structured like this:
+//    if (GC_locker::needs_gc())
+//      SharedRuntime::block_for_jni_critical();
+//    tranistion to thread_in_native
+//    unpack arrray arguments and call native entry point
+//    check for safepoint in progress
+//    check if any thread suspend flags are set
+//      call into JVM and possible unlock the JNI critical
+//      if a GC was suppressed while in the critical native.
+//    transition back to thread_in_Java
+//    return to caller
+//
 nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 methodHandle method,
                                                 int compile_id,
@@ -1547,6 +1718,29 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler *masm,
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
+  if (method->is_method_handle_intrinsic()) {
+    vmIntrinsics::ID iid = method->intrinsic_id();
+    intptr_t start = (intptr_t)__ pc();
+    int vep_offset = ((intptr_t)__ pc()) - start;
+    gen_special_dispatch(masm,
+                         total_in_args,
+                         comp_args_on_stack,
+                         method->intrinsic_id(),
+                         in_sig_bt,
+                         in_regs);
+    int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
+    __ flush();
+    int stack_slots = SharedRuntime::out_preserve_stack_slots();  // no out slots at all, actually
+    return nmethod::new_native_nmethod(method,
+                                       compile_id,
+                                       masm->code(),
+                                       vep_offset,
+                                       frame_complete,
+                                       stack_slots / VMRegImpl::slots_per_word,
+                                       in_ByteSize(-1),
+                                       in_ByteSize(-1),
+                                       (OopMapSet*)NULL);
+  }
   bool is_critical_native = true;
   address native_func = method->critical_native_function();
   if (native_func == NULL) {
@@ -1658,7 +1852,7 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler *masm,
           case T_SHORT:
           case T_CHAR:
           case T_INT:  single_slots++; break;
-          case T_ARRAY:
+          case T_ARRAY:  // specific to LP64 (7145024)
           case T_LONG: double_slots++; break;
           default:  ShouldNotReachHere();
         }

hotspot/src/cpu/x86/vm/stubGenerator_x86_32.cpp

@@ -2326,12 +2326,6 @@ class StubGenerator: public StubCodeGenerator {
     StubRoutines::_d2l_wrapper                              = generate_d2i_wrapper(T_LONG,
                                                                                    CAST_FROM_FN_PTR(address, SharedRuntime::d2l));
 
-    // Build this early so it's available for the interpreter
-    StubRoutines::_throw_WrongMethodTypeException_entry =
-      generate_throw_exception("WrongMethodTypeException throw_exception",
-                               CAST_FROM_FN_PTR(address, SharedRuntime::throw_WrongMethodTypeException),
-                               rax, rcx);
-
     // Build this early so it's available for the interpreter
     StubRoutines::_throw_StackOverflowError_entry          = generate_throw_exception("StackOverflowError throw_exception",           CAST_FROM_FN_PTR(address, SharedRuntime::throw_StackOverflowError));
   }

hotspot/src/cpu/x86/vm/stubGenerator_x86_64.cpp

@@ -3102,14 +3102,6 @@ class StubGenerator: public StubCodeGenerator {
 
     StubRoutines::x86::_verify_mxcsr_entry    = generate_verify_mxcsr();
 
-    // Build this early so it's available for the interpreter.  Stub
-    // expects the required and actual types as register arguments in
-    // j_rarg0 and j_rarg1 respectively.
-    StubRoutines::_throw_WrongMethodTypeException_entry =
-      generate_throw_exception("WrongMethodTypeException throw_exception",
-                               CAST_FROM_FN_PTR(address, SharedRuntime::throw_WrongMethodTypeException),
-                               rax, rcx);
-
     // Build this early so it's available for the interpreter.
     StubRoutines::_throw_StackOverflowError_entry =
       generate_throw_exception("StackOverflowError throw_exception",

hotspot/src/cpu/x86/vm/templateInterpreter_x86_32.cpp

@@ -710,9 +710,9 @@ address InterpreterGenerator::generate_accessor_entry(void) {
     // Need to differentiate between igetfield, agetfield, bgetfield etc.
     // because they are different sizes.
     // Use the type from the constant pool cache
-    __ shrl(rdx, ConstantPoolCacheEntry::tosBits);
+    __ shrl(rdx, ConstantPoolCacheEntry::tos_state_shift);
-    // Make sure we don't need to mask rdx for tosBits after the above shift
+    // Make sure we don't need to mask rdx after the above shift
-    ConstantPoolCacheEntry::verify_tosBits();
+    ConstantPoolCacheEntry::verify_tos_state_shift();
     __ cmpl(rdx, btos);
     __ jcc(Assembler::notEqual, notByte);
     __ load_signed_byte(rax, field_address);
@@ -1513,7 +1513,6 @@ address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter:
     case Interpreter::empty                  : entry_point = ((InterpreterGenerator*)this)->generate_empty_entry();        break;
     case Interpreter::accessor               : entry_point = ((InterpreterGenerator*)this)->generate_accessor_entry();     break;
     case Interpreter::abstract               : entry_point = ((InterpreterGenerator*)this)->generate_abstract_entry();     break;
-    case Interpreter::method_handle          : entry_point = ((InterpreterGenerator*)this)->generate_method_handle_entry(); break;
 
     case Interpreter::java_lang_math_sin     : // fall thru
     case Interpreter::java_lang_math_cos     : // fall thru
@@ -1526,7 +1525,9 @@ address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter:
     case Interpreter::java_lang_math_exp     : entry_point = ((InterpreterGenerator*)this)->generate_math_entry(kind);     break;
     case Interpreter::java_lang_ref_reference_get
                                              : entry_point = ((InterpreterGenerator*)this)->generate_Reference_get_entry(); break;
-    default                                  : ShouldNotReachHere();                                                       break;
+    default:
+      fatal(err_msg("unexpected method kind: %d", kind));
+      break;
   }
 
   if (entry_point) return entry_point;

hotspot/src/cpu/x86/vm/templateInterpreter_x86_64.cpp

@@ -683,9 +683,9 @@ address InterpreterGenerator::generate_accessor_entry(void) {
     // Need to differentiate between igetfield, agetfield, bgetfield etc.
     // because they are different sizes.
     // Use the type from the constant pool cache
-    __ shrl(rdx, ConstantPoolCacheEntry::tosBits);
+    __ shrl(rdx, ConstantPoolCacheEntry::tos_state_shift);
-    // Make sure we don't need to mask edx for tosBits after the above shift
+    // Make sure we don't need to mask edx after the above shift
-    ConstantPoolCacheEntry::verify_tosBits();
+    ConstantPoolCacheEntry::verify_tos_state_shift();
 
     __ cmpl(rdx, atos);
     __ jcc(Assembler::notEqual, notObj);
@@ -1529,7 +1529,6 @@ address AbstractInterpreterGenerator::generate_method_entry(
   case Interpreter::empty                  : entry_point = ((InterpreterGenerator*)this)->generate_empty_entry();       break;
   case Interpreter::accessor               : entry_point = ((InterpreterGenerator*)this)->generate_accessor_entry();    break;
   case Interpreter::abstract               : entry_point = ((InterpreterGenerator*)this)->generate_abstract_entry();    break;
-  case Interpreter::method_handle          : entry_point = ((InterpreterGenerator*) this)->generate_method_handle_entry();break;
 
   case Interpreter::java_lang_math_sin     : // fall thru
   case Interpreter::java_lang_math_cos     : // fall thru
@@ -1542,7 +1541,9 @@ address AbstractInterpreterGenerator::generate_method_entry(
   case Interpreter::java_lang_math_exp     : entry_point = ((InterpreterGenerator*)this)->generate_math_entry(kind);    break;
   case Interpreter::java_lang_ref_reference_get
                                            : entry_point = ((InterpreterGenerator*)this)->generate_Reference_get_entry(); break;
-  default                                  : ShouldNotReachHere();                                                       break;
+  default:
+    fatal(err_msg("unexpected method kind: %d", kind));
+    break;
   }
 
   if (entry_point) {

hotspot/src/cpu/x86/vm/templateTable_x86_32.cpp

@@ -446,13 +446,13 @@ void TemplateTable::fast_aldc(bool wide) {
   const Register cache = rcx;
   const Register index = rdx;
 
-  resolve_cache_and_index(f1_oop, rax, cache, index, wide ? sizeof(u2) : sizeof(u1));
+  resolve_cache_and_index(f12_oop, rax, cache, index, wide ? sizeof(u2) : sizeof(u1));
   if (VerifyOops) {
     __ verify_oop(rax);
   }
 
   Label L_done, L_throw_exception;
-  const Register con_klass_temp = rcx;  // same as Rcache
+  const Register con_klass_temp = rcx;  // same as cache
   __ load_klass(con_klass_temp, rax);
   __ cmpptr(con_klass_temp, ExternalAddress((address)Universe::systemObjArrayKlassObj_addr()));
   __ jcc(Assembler::notEqual, L_done);
@@ -2084,15 +2084,15 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
                                             Register Rcache,
                                             Register index,
                                             size_t index_size) {
-  Register temp = rbx;
+  const Register temp = rbx;
-
   assert_different_registers(result, Rcache, index, temp);
 
   Label resolved;
-  if (byte_no == f1_oop) {
+  if (byte_no == f12_oop) {
-    // We are resolved if the f1 field contains a non-null object (CallSite, etc.)
+    // We are resolved if the f1 field contains a non-null object (CallSite, MethodType, etc.)
-    // This kind of CP cache entry does not need to match the flags byte, because
+    // This kind of CP cache entry does not need to match bytecode_1 or bytecode_2, because
     // there is a 1-1 relation between bytecode type and CP entry type.
+    // The caller will also load a methodOop from f2.
     assert(result != noreg, ""); //else do cmpptr(Address(...), (int32_t) NULL_WORD)
     __ get_cache_and_index_at_bcp(Rcache, index, 1, index_size);
     __ movptr(result, Address(Rcache, index, Address::times_ptr, constantPoolCacheOopDesc::base_offset() + ConstantPoolCacheEntry::f1_offset()));
@@ -2117,10 +2117,13 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
     case Bytecodes::_invokespecial  : // fall through
     case Bytecodes::_invokestatic   : // fall through
     case Bytecodes::_invokeinterface: entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invoke);         break;
+    case Bytecodes::_invokehandle   : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokehandle);   break;
     case Bytecodes::_invokedynamic  : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokedynamic);  break;
     case Bytecodes::_fast_aldc      : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);            break;
     case Bytecodes::_fast_aldc_w    : entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);            break;
-    default                         : ShouldNotReachHere();                                 break;
+    default:
+      fatal(err_msg("unexpected bytecode: %s", Bytecodes::name(bytecode())));
+      break;
   }
   __ movl(temp, (int)bytecode());
   __ call_VM(noreg, entry, temp);
@@ -2161,7 +2164,7 @@ void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
                                                Register itable_index,
                                                Register flags,
                                                bool is_invokevirtual,
-                                               bool is_invokevfinal /*unused*/,
+                                               bool is_invokevfinal, /*unused*/
                                                bool is_invokedynamic) {
   // setup registers
   const Register cache = rcx;
@@ -2171,28 +2174,33 @@ void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
   assert_different_registers(itable_index, flags);
   assert_different_registers(itable_index, cache, index);
   // determine constant pool cache field offsets
+  assert(is_invokevirtual == (byte_no == f2_byte), "is_invokevirtual flag redundant");
   const int method_offset = in_bytes(
     constantPoolCacheOopDesc::base_offset() +
-      (is_invokevirtual
+      ((byte_no == f2_byte)
        ? ConstantPoolCacheEntry::f2_offset()
-       : ConstantPoolCacheEntry::f1_offset()
+       : ConstantPoolCacheEntry::f1_offset()));
-      )
-    );
   const int flags_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
                                     ConstantPoolCacheEntry::flags_offset());
   // access constant pool cache fields
   const int index_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
                                     ConstantPoolCacheEntry::f2_offset());
 
-  if (byte_no == f1_oop) {
+  if (byte_no == f12_oop) {
-    // Resolved f1_oop goes directly into 'method' register.
+    // Resolved f1_oop (CallSite, MethodType, etc.) goes into 'itable_index'.
-    assert(is_invokedynamic, "");
+    // Resolved f2_oop (methodOop invoker) will go into 'method' (at index_offset).
-    resolve_cache_and_index(byte_no, method, cache, index, sizeof(u4));
+    // See ConstantPoolCacheEntry::set_dynamic_call and set_method_handle.
+    size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
+    resolve_cache_and_index(byte_no, itable_index, cache, index, index_size);
+    __ movptr(method, Address(cache, index, Address::times_ptr, index_offset));
+    itable_index = noreg;  // hack to disable load below
   } else {
     resolve_cache_and_index(byte_no, noreg, cache, index, sizeof(u2));
     __ movptr(method, Address(cache, index, Address::times_ptr, method_offset));
   }
   if (itable_index != noreg) {
+    // pick up itable index from f2 also:
+    assert(byte_no == f1_byte, "already picked up f1");
     __ movptr(itable_index, Address(cache, index, Address::times_ptr, index_offset));
   }
   __ movl(flags, Address(cache, index, Address::times_ptr, flags_offset));
@@ -2260,10 +2268,10 @@ void TemplateTable::getfield_or_static(int byte_no, bool is_static) {
 
   Label Done, notByte, notInt, notShort, notChar, notLong, notFloat, notObj, notDouble;
 
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
   assert(btos == 0, "change code, btos != 0");
   // btos
-  __ andptr(flags, 0x0f);
+  __ andptr(flags, ConstantPoolCacheEntry::tos_state_mask);
   __ jcc(Assembler::notZero, notByte);
 
   __ load_signed_byte(rax, lo );
@@ -2415,9 +2423,9 @@ void TemplateTable::jvmti_post_field_mod(Register cache, Register index, bool is
       __ movl(rcx, Address(rax, rdx, Address::times_ptr, in_bytes(cp_base_offset +
                                    ConstantPoolCacheEntry::flags_offset())));
       __ mov(rbx, rsp);
-      __ shrl(rcx, ConstantPoolCacheEntry::tosBits);
+      __ shrl(rcx, ConstantPoolCacheEntry::tos_state_shift);
-      // Make sure we don't need to mask rcx for tosBits after the above shift
+      // Make sure we don't need to mask rcx after the above shift
-      ConstantPoolCacheEntry::verify_tosBits();
+      ConstantPoolCacheEntry::verify_tos_state_shift();
       __ cmpl(rcx, ltos);
       __ jccb(Assembler::equal, two_word);
       __ cmpl(rcx, dtos);
@@ -2467,7 +2475,7 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
 
   Label notVolatile, Done;
   __ movl(rdx, flags);
-  __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   __ andl(rdx, 0x1);
 
   // field addresses
@@ -2476,9 +2484,9 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
 
   Label notByte, notInt, notShort, notChar, notLong, notFloat, notObj, notDouble;
 
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
   assert(btos == 0, "change code, btos != 0");
-  __ andl(flags, 0x0f);
+  __ andl(flags, ConstantPoolCacheEntry::tos_state_mask);
   __ jcc(Assembler::notZero, notByte);
 
   // btos
@@ -2719,7 +2727,7 @@ void TemplateTable::fast_storefield(TosState state) {
   // volatile_barrier( );
 
   Label notVolatile, Done;
-  __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   __ andl(rdx, 0x1);
   // Check for volatile store
   __ testl(rdx, rdx);
@@ -2885,19 +2893,29 @@ void TemplateTable::count_calls(Register method, Register temp) {
 }
 
 
-void TemplateTable::prepare_invoke(Register method, Register index, int byte_no) {
+void TemplateTable::prepare_invoke(int byte_no,
+                                   Register method,  // linked method (or i-klass)
+                                   Register index,   // itable index, MethodType, etc.
+                                   Register recv,    // if caller wants to see it
+                                   Register flags    // if caller wants to test it
+                                   ) {
   // determine flags
-  Bytecodes::Code code = bytecode();
+  const Bytecodes::Code code = bytecode();
   const bool is_invokeinterface  = code == Bytecodes::_invokeinterface;
   const bool is_invokedynamic    = code == Bytecodes::_invokedynamic;
+  const bool is_invokehandle     = code == Bytecodes::_invokehandle;
   const bool is_invokevirtual    = code == Bytecodes::_invokevirtual;
   const bool is_invokespecial    = code == Bytecodes::_invokespecial;
-  const bool load_receiver      = (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic);
+  const bool load_receiver       = (recv  != noreg);
-  const bool receiver_null_check = is_invokespecial;
+  const bool save_flags          = (flags != noreg);
-  const bool save_flags = is_invokeinterface || is_invokevirtual;
+  assert(load_receiver == (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic), "");
+  assert(save_flags    == (is_invokeinterface || is_invokevirtual), "need flags for vfinal");
+  assert(flags == noreg || flags == rdx, "");
+  assert(recv  == noreg || recv  == rcx, "");
+
   // setup registers & access constant pool cache
-  const Register recv   = rcx;
+  if (recv  == noreg)  recv  = rcx;
-  const Register flags  = rdx;
+  if (flags == noreg)  flags = rdx;
   assert_different_registers(method, index, recv, flags);
 
   // save 'interpreter return address'
@@ -2905,20 +2923,28 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
 
   load_invoke_cp_cache_entry(byte_no, method, index, flags, is_invokevirtual, false, is_invokedynamic);
 
-  // load receiver if needed (note: no return address pushed yet)
+  // maybe push appendix to arguments (just before return address)
-  if (load_receiver) {
+  if (is_invokedynamic || is_invokehandle) {
-    assert(!is_invokedynamic, "");
+    Label L_no_push;
-    __ movl(recv, flags);
+    __ verify_oop(index);
-    __ andl(recv, 0xFF);
+    __ testl(flags, (1 << ConstantPoolCacheEntry::has_appendix_shift));
-    // recv count is 0 based?
+    __ jccb(Assembler::zero, L_no_push);
-    Address recv_addr(rsp, recv, Interpreter::stackElementScale(), -Interpreter::expr_offset_in_bytes(1));
+    // Push the appendix as a trailing parameter.
-    __ movptr(recv, recv_addr);
+    // This must be done before we get the receiver,
-    __ verify_oop(recv);
+    // since the parameter_size includes it.
+    __ push(index);  // push appendix (MethodType, CallSite, etc.)
+    __ bind(L_no_push);
   }
 
-  // do null check if needed
+  // load receiver if needed (note: no return address pushed yet)
-  if (receiver_null_check) {
+  if (load_receiver) {
-    __ null_check(recv);
+    __ movl(recv, flags);
+    __ andl(recv, ConstantPoolCacheEntry::parameter_size_mask);
+    const int no_return_pc_pushed_yet = -1;  // argument slot correction before we push return address
+    const int receiver_is_at_end      = -1;  // back off one slot to get receiver
+    Address recv_addr = __ argument_address(recv, no_return_pc_pushed_yet + receiver_is_at_end);
+    __ movptr(recv, recv_addr);
+    __ verify_oop(recv);
   }
 
   if (save_flags) {
@@ -2926,16 +2952,14 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
   }
 
   // compute return type
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
-  // Make sure we don't need to mask flags for tosBits after the above shift
+  // Make sure we don't need to mask flags after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
   // load return address
   {
-    address table_addr;
+    const address table_addr = (is_invokeinterface || is_invokedynamic) ?
-    if (is_invokeinterface || is_invokedynamic)
+        (address)Interpreter::return_5_addrs_by_index_table() :
-      table_addr = (address)Interpreter::return_5_addrs_by_index_table();
+        (address)Interpreter::return_3_addrs_by_index_table();
-    else
-      table_addr = (address)Interpreter::return_3_addrs_by_index_table();
     ExternalAddress table(table_addr);
     __ movptr(flags, ArrayAddress(table, Address(noreg, flags, Address::times_ptr)));
   }
@@ -2943,7 +2967,7 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
   // push return address
   __ push(flags);
 
-  // Restore flag value from the constant pool cache, and restore rsi
+  // Restore flags value from the constant pool cache, and restore rsi
   // for later null checks.  rsi is the bytecode pointer
   if (save_flags) {
     __ mov(flags, rsi);
@@ -2952,22 +2976,26 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
 }
 
 
-void TemplateTable::invokevirtual_helper(Register index, Register recv,
+void TemplateTable::invokevirtual_helper(Register index,
+                                         Register recv,
                                          Register flags) {
-
   // Uses temporary registers rax, rdx
   assert_different_registers(index, recv, rax, rdx);
+  assert(index == rbx, "");
+  assert(recv  == rcx, "");
 
   // Test for an invoke of a final method
   Label notFinal;
   __ movl(rax, flags);
-  __ andl(rax, (1 << ConstantPoolCacheEntry::vfinalMethod));
+  __ andl(rax, (1 << ConstantPoolCacheEntry::is_vfinal_shift));
   __ jcc(Assembler::zero, notFinal);
 
-  Register method = index;  // method must be rbx,
+  const Register method = index;  // method must be rbx
-  assert(method == rbx, "methodOop must be rbx, for interpreter calling convention");
+  assert(method == rbx,
+         "methodOop must be rbx for interpreter calling convention");
 
   // do the call - the index is actually the method to call
+  // that is, f2 is a vtable index if !is_vfinal, else f2 is a methodOop
   __ verify_oop(method);
 
   // It's final, need a null check here!
@@ -2982,7 +3010,6 @@ void TemplateTable::invokevirtual_helper(Register index, Register recv,
 
   // get receiver klass
   __ null_check(recv, oopDesc::klass_offset_in_bytes());
-  // Keep recv in rcx for callee expects it there
   __ load_klass(rax, recv);
   __ verify_oop(rax);
 
@@ -2990,9 +3017,7 @@ void TemplateTable::invokevirtual_helper(Register index, Register recv,
   __ profile_virtual_call(rax, rdi, rdx);
 
   // get target methodOop & entry point
-  const int base = instanceKlass::vtable_start_offset() * wordSize;
+  __ lookup_virtual_method(rax, index, method);
-  assert(vtableEntry::size() * wordSize == 4, "adjust the scaling in the code below");
-  __ movptr(method, Address(rax, index, Address::times_ptr, base + vtableEntry::method_offset_in_bytes()));
   __ jump_from_interpreted(method, rdx);
 }
 
@@ -3000,9 +3025,12 @@ void TemplateTable::invokevirtual_helper(Register index, Register recv,
 void TemplateTable::invokevirtual(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f2_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no,
+                 rbx,    // method or vtable index
+                 noreg,  // unused itable index
+                 rcx, rdx); // recv, flags
 
-  // rbx,: index
+  // rbx: index
   // rcx: receiver
   // rdx: flags
 
@@ -3013,7 +3041,10 @@ void TemplateTable::invokevirtual(int byte_no) {
 void TemplateTable::invokespecial(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no, rbx, noreg,  // get f1 methodOop
+                 rcx);  // get receiver also for null check
+  __ verify_oop(rcx);
+  __ null_check(rcx);
   // do the call
   __ verify_oop(rbx);
   __ profile_call(rax);
@@ -3024,7 +3055,7 @@ void TemplateTable::invokespecial(int byte_no) {
 void TemplateTable::invokestatic(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no, rbx);  // get f1 methodOop
   // do the call
   __ verify_oop(rbx);
   __ profile_call(rax);
@@ -3042,10 +3073,11 @@ void TemplateTable::fast_invokevfinal(int byte_no) {
 void TemplateTable::invokeinterface(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rax, rbx, byte_no);
+  prepare_invoke(byte_no, rax, rbx,  // get f1 klassOop, f2 itable index
+                 rcx, rdx); // recv, flags
 
-  // rax,: Interface
+  // rax: interface klass (from f1)
-  // rbx,: index
+  // rbx: itable index (from f2)
   // rcx: receiver
   // rdx: flags
 
@@ -3055,7 +3087,7 @@ void TemplateTable::invokeinterface(int byte_no) {
   // another compliant java compiler.
   Label notMethod;
   __ movl(rdi, rdx);
-  __ andl(rdi, (1 << ConstantPoolCacheEntry::methodInterface));
+  __ andl(rdi, (1 << ConstantPoolCacheEntry::is_forced_virtual_shift));
   __ jcc(Assembler::zero, notMethod);
 
   invokevirtual_helper(rbx, rcx, rdx);
@@ -3063,6 +3095,7 @@ void TemplateTable::invokeinterface(int byte_no) {
 
   // Get receiver klass into rdx - also a null check
   __ restore_locals();  // restore rdi
+  __ null_check(rcx, oopDesc::klass_offset_in_bytes());
   __ load_klass(rdx, rcx);
   __ verify_oop(rdx);
 
@@ -3077,7 +3110,7 @@ void TemplateTable::invokeinterface(int byte_no) {
                              rbx, rsi,
                              no_such_interface);
 
-  // rbx,: methodOop to call
+  // rbx: methodOop to call
   // rcx: receiver
   // Check for abstract method error
   // Note: This should be done more efficiently via a throw_abstract_method_error
@@ -3116,9 +3149,39 @@ void TemplateTable::invokeinterface(int byte_no) {
   __ should_not_reach_here();
 }
 
+void TemplateTable::invokehandle(int byte_no) {
+  transition(vtos, vtos);
+  assert(byte_no == f12_oop, "use this argument");
+  const Register rbx_method = rbx;  // (from f2)
+  const Register rax_mtype  = rax;  // (from f1)
+  const Register rcx_recv   = rcx;
+  const Register rdx_flags  = rdx;
+
+  if (!EnableInvokeDynamic) {
+    // rewriter does not generate this bytecode
+    __ should_not_reach_here();
+    return;
+  }
+
+  prepare_invoke(byte_no,
+                 rbx_method, rax_mtype,  // get f2 methodOop, f1 MethodType
+                 rcx_recv);
+  __ verify_oop(rbx_method);
+  __ verify_oop(rcx_recv);
+  __ null_check(rcx_recv);
+
+  // Note:  rax_mtype is already pushed (if necessary) by prepare_invoke
+
+  // FIXME: profile the LambdaForm also
+  __ profile_final_call(rax);
+
+  __ jump_from_interpreted(rbx_method, rdx);
+}
+
+
 void TemplateTable::invokedynamic(int byte_no) {
   transition(vtos, vtos);
-  assert(byte_no == f1_oop, "use this argument");
+  assert(byte_no == f12_oop, "use this argument");
 
   if (!EnableInvokeDynamic) {
     // We should not encounter this bytecode if !EnableInvokeDynamic.
@@ -3131,26 +3194,23 @@ void TemplateTable::invokedynamic(int byte_no) {
     return;
   }
 
-  prepare_invoke(rax, rbx, byte_no);
+  const Register rbx_method   = rbx;
+  const Register rax_callsite = rax;
 
-  // rax: CallSite object (f1)
+  prepare_invoke(byte_no, rbx_method, rax_callsite);
-  // rbx: unused (f2)
-  // rcx: receiver address
-  // rdx: flags (unused)
 
-  Register rax_callsite      = rax;
+  // rax: CallSite object (from f1)
-  Register rcx_method_handle = rcx;
+  // rbx: MH.linkToCallSite method (from f2)
+
+  // Note:  rax_callsite is already pushed by prepare_invoke
 
   // %%% should make a type profile for any invokedynamic that takes a ref argument
   // profile this call
   __ profile_call(rsi);
 
   __ verify_oop(rax_callsite);
-  __ load_heap_oop(rcx_method_handle, Address(rax_callsite, __ delayed_value(java_lang_invoke_CallSite::target_offset_in_bytes, rdx)));
+
-  __ null_check(rcx_method_handle);
+  __ jump_from_interpreted(rbx_method, rdx);
-  __ verify_oop(rcx_method_handle);
-  __ prepare_to_jump_from_interpreted();
-  __ jump_to_method_handle_entry(rcx_method_handle, rdx);
 }
 
 //----------------------------------------------------------------------------------------------------

hotspot/src/cpu/x86/vm/templateTable_x86_32.hpp

@@ -25,7 +25,12 @@
 #ifndef CPU_X86_VM_TEMPLATETABLE_X86_32_HPP
 #define CPU_X86_VM_TEMPLATETABLE_X86_32_HPP
 
-  static void prepare_invoke(Register method, Register index, int byte_no);
+  static void prepare_invoke(int byte_no,
+                             Register method,         // linked method (or i-klass)
+                             Register index = noreg,  // itable index, MethodType, etc.
+                             Register recv  = noreg,  // if caller wants to see it
+                             Register flags = noreg   // if caller wants to test it
+                             );
   static void invokevirtual_helper(Register index, Register recv,
                                    Register flags);
   static void volatile_barrier(Assembler::Membar_mask_bits order_constraint);

hotspot/src/cpu/x86/vm/templateTable_x86_64.cpp

@@ -458,7 +458,7 @@ void TemplateTable::fast_aldc(bool wide) {
   const Register cache = rcx;
   const Register index = rdx;
 
-  resolve_cache_and_index(f1_oop, rax, cache, index, wide ? sizeof(u2) : sizeof(u1));
+  resolve_cache_and_index(f12_oop, rax, cache, index, wide ? sizeof(u2) : sizeof(u1));
   if (VerifyOops) {
     __ verify_oop(rax);
   }
@@ -2125,10 +2125,11 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
   assert_different_registers(result, Rcache, index, temp);
 
   Label resolved;
-  if (byte_no == f1_oop) {
+  if (byte_no == f12_oop) {
-    // We are resolved if the f1 field contains a non-null object (CallSite, etc.)
+    // We are resolved if the f1 field contains a non-null object (CallSite, MethodType, etc.)
-    // This kind of CP cache entry does not need to match the flags byte, because
+    // This kind of CP cache entry does not need to match bytecode_1 or bytecode_2, because
     // there is a 1-1 relation between bytecode type and CP entry type.
+    // The caller will also load a methodOop from f2.
     assert(result != noreg, ""); //else do cmpptr(Address(...), (int32_t) NULL_WORD)
     __ get_cache_and_index_at_bcp(Rcache, index, 1, index_size);
     __ movptr(result, Address(Rcache, index, Address::times_ptr, constantPoolCacheOopDesc::base_offset() + ConstantPoolCacheEntry::f1_offset()));
@@ -2157,6 +2158,9 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
   case Bytecodes::_invokeinterface:
     entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invoke);
     break;
+  case Bytecodes::_invokehandle:
+    entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokehandle);
+    break;
   case Bytecodes::_invokedynamic:
     entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokedynamic);
     break;
@@ -2167,7 +2171,7 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
     entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);
     break;
   default:
-    ShouldNotReachHere();
+    fatal(err_msg("unexpected bytecode: %s", Bytecodes::name(bytecode())));
     break;
   }
   __ movl(temp, (int) bytecode());
@@ -2180,7 +2184,7 @@ void TemplateTable::resolve_cache_and_index(int byte_no,
   __ bind(resolved);
 }
 
-// The Rcache and index registers must be set before call
+// The cache and index registers must be set before call
 void TemplateTable::load_field_cp_cache_entry(Register obj,
                                               Register cache,
                                               Register index,
@@ -2191,17 +2195,17 @@ void TemplateTable::load_field_cp_cache_entry(Register obj,
 
   ByteSize cp_base_offset = constantPoolCacheOopDesc::base_offset();
   // Field offset
-  __ movptr(off, Address(cache, index, Address::times_8,
+  __ movptr(off, Address(cache, index, Address::times_ptr,
                          in_bytes(cp_base_offset +
                                   ConstantPoolCacheEntry::f2_offset())));
   // Flags
-  __ movl(flags, Address(cache, index, Address::times_8,
+  __ movl(flags, Address(cache, index, Address::times_ptr,
                          in_bytes(cp_base_offset +
                                   ConstantPoolCacheEntry::flags_offset())));
 
   // klass overwrite register
   if (is_static) {
-    __ movptr(obj, Address(cache, index, Address::times_8,
+    __ movptr(obj, Address(cache, index, Address::times_ptr,
                            in_bytes(cp_base_offset +
                                     ConstantPoolCacheEntry::f1_offset())));
   }
@@ -2222,9 +2226,10 @@ void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
   assert_different_registers(itable_index, flags);
   assert_different_registers(itable_index, cache, index);
   // determine constant pool cache field offsets
+  assert(is_invokevirtual == (byte_no == f2_byte), "is_invokevirtual flag redundant");
   const int method_offset = in_bytes(
     constantPoolCacheOopDesc::base_offset() +
-      (is_invokevirtual
+      ((byte_no == f2_byte)
        ? ConstantPoolCacheEntry::f2_offset()
        : ConstantPoolCacheEntry::f1_offset()));
   const int flags_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
@@ -2233,15 +2238,21 @@ void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
   const int index_offset = in_bytes(constantPoolCacheOopDesc::base_offset() +
                                     ConstantPoolCacheEntry::f2_offset());
 
-  if (byte_no == f1_oop) {
+  if (byte_no == f12_oop) {
-    // Resolved f1_oop goes directly into 'method' register.
+    // Resolved f1_oop (CallSite, MethodType, etc.) goes into 'itable_index'.
-    assert(is_invokedynamic, "");
+    // Resolved f2_oop (methodOop invoker) will go into 'method' (at index_offset).
-    resolve_cache_and_index(byte_no, method, cache, index, sizeof(u4));
+    // See ConstantPoolCacheEntry::set_dynamic_call and set_method_handle.
+    size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
+    resolve_cache_and_index(byte_no, itable_index, cache, index, index_size);
+    __ movptr(method, Address(cache, index, Address::times_ptr, index_offset));
+    itable_index = noreg;  // hack to disable load below
   } else {
     resolve_cache_and_index(byte_no, noreg, cache, index, sizeof(u2));
     __ movptr(method, Address(cache, index, Address::times_ptr, method_offset));
   }
   if (itable_index != noreg) {
+    // pick up itable index from f2 also:
+    assert(byte_no == f1_byte, "already picked up f1");
     __ movptr(itable_index, Address(cache, index, Address::times_ptr, index_offset));
   }
   __ movl(flags, Address(cache, index, Address::times_ptr, flags_offset));
@@ -2317,10 +2328,11 @@ void TemplateTable::getfield_or_static(int byte_no, bool is_static) {
   Label Done, notByte, notInt, notShort, notChar,
               notLong, notFloat, notObj, notDouble;
 
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
+  // Make sure we don't need to mask edx after the above shift
   assert(btos == 0, "change code, btos != 0");
 
-  __ andl(flags, 0x0F);
+  __ andl(flags, ConstantPoolCacheEntry::tos_state_mask);
   __ jcc(Assembler::notZero, notByte);
   // btos
   __ load_signed_byte(rax, field);
@@ -2466,10 +2478,9 @@ void TemplateTable::jvmti_post_field_mod(Register cache, Register index, bool is
                            Address::times_8,
                            in_bytes(cp_base_offset +
                                      ConstantPoolCacheEntry::flags_offset())));
-      __ shrl(c_rarg3, ConstantPoolCacheEntry::tosBits);
+      __ shrl(c_rarg3, ConstantPoolCacheEntry::tos_state_shift);
-      // Make sure we don't need to mask rcx for tosBits after the
+      // Make sure we don't need to mask rcx after the above shift
-      // above shift
+      ConstantPoolCacheEntry::verify_tos_state_shift();
-      ConstantPoolCacheEntry::verify_tosBits();
       __ movptr(c_rarg1, at_tos_p1());  // initially assume a one word jvalue
       __ cmpl(c_rarg3, ltos);
       __ cmovptr(Assembler::equal,
@@ -2516,7 +2527,7 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
 
   Label notVolatile, Done;
   __ movl(rdx, flags);
-  __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   __ andl(rdx, 0x1);
 
   // field address
@@ -2525,10 +2536,10 @@ void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
   Label notByte, notInt, notShort, notChar,
         notLong, notFloat, notObj, notDouble;
 
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
 
   assert(btos == 0, "change code, btos != 0");
-  __ andl(flags, 0x0f);
+  __ andl(flags, ConstantPoolCacheEntry::tos_state_mask);
   __ jcc(Assembler::notZero, notByte);
 
   // btos
@@ -2751,7 +2762,7 @@ void TemplateTable::fast_storefield(TosState state) {
   //                                              Assembler::StoreStore));
 
   Label notVolatile;
-  __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   __ andl(rdx, 0x1);
 
   // Get object from stack
@@ -2832,7 +2843,7 @@ void TemplateTable::fast_accessfield(TosState state) {
   //   __ movl(rdx, Address(rcx, rbx, Address::times_8,
   //                        in_bytes(constantPoolCacheOopDesc::base_offset() +
   //                                 ConstantPoolCacheEntry::flags_offset())));
-  //   __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  //   __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   //   __ andl(rdx, 0x1);
   // }
   __ movptr(rbx, Address(rcx, rbx, Address::times_8,
@@ -2920,7 +2931,7 @@ void TemplateTable::fast_xaccess(TosState state) {
   //   __ movl(rdx, Address(rcx, rdx, Address::times_8,
   //                        in_bytes(constantPoolCacheOopDesc::base_offset() +
   //                                 ConstantPoolCacheEntry::flags_offset())));
-  //   __ shrl(rdx, ConstantPoolCacheEntry::volatileField);
+  //   __ shrl(rdx, ConstantPoolCacheEntry::is_volatile_shift);
   //   __ testl(rdx, 0x1);
   //   __ jcc(Assembler::zero, notVolatile);
   //   __ membar(Assembler::LoadLoad);
@@ -2940,19 +2951,29 @@ void TemplateTable::count_calls(Register method, Register temp) {
   ShouldNotReachHere();
 }
 
-void TemplateTable::prepare_invoke(Register method, Register index, int byte_no) {
+void TemplateTable::prepare_invoke(int byte_no,
+                                   Register method,  // linked method (or i-klass)
+                                   Register index,   // itable index, MethodType, etc.
+                                   Register recv,    // if caller wants to see it
+                                   Register flags    // if caller wants to test it
+                                   ) {
   // determine flags
-  Bytecodes::Code code = bytecode();
+  const Bytecodes::Code code = bytecode();
   const bool is_invokeinterface  = code == Bytecodes::_invokeinterface;
   const bool is_invokedynamic    = code == Bytecodes::_invokedynamic;
+  const bool is_invokehandle     = code == Bytecodes::_invokehandle;
   const bool is_invokevirtual    = code == Bytecodes::_invokevirtual;
   const bool is_invokespecial    = code == Bytecodes::_invokespecial;
-  const bool load_receiver      = (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic);
+  const bool load_receiver       = (recv  != noreg);
-  const bool receiver_null_check = is_invokespecial;
+  const bool save_flags          = (flags != noreg);
-  const bool save_flags = is_invokeinterface || is_invokevirtual;
+  assert(load_receiver == (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic), "");
+  assert(save_flags    == (is_invokeinterface || is_invokevirtual), "need flags for vfinal");
+  assert(flags == noreg || flags == rdx, "");
+  assert(recv  == noreg || recv  == rcx, "");
+
   // setup registers & access constant pool cache
-  const Register recv   = rcx;
+  if (recv  == noreg)  recv  = rcx;
-  const Register flags  = rdx;
+  if (flags == noreg)  flags = rdx;
   assert_different_registers(method, index, recv, flags);
 
   // save 'interpreter return address'
@@ -2960,19 +2981,29 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
 
   load_invoke_cp_cache_entry(byte_no, method, index, flags, is_invokevirtual, false, is_invokedynamic);
 
-  // load receiver if needed (note: no return address pushed yet)
+  // maybe push appendix to arguments (just before return address)
-  if (load_receiver) {
+  if (is_invokedynamic || is_invokehandle) {
-    assert(!is_invokedynamic, "");
+    Label L_no_push;
-    __ movl(recv, flags);
+    __ verify_oop(index);
-    __ andl(recv, 0xFF);
+    __ testl(flags, (1 << ConstantPoolCacheEntry::has_appendix_shift));
-    Address recv_addr(rsp, recv, Address::times_8, -Interpreter::expr_offset_in_bytes(1));
+    __ jccb(Assembler::zero, L_no_push);
-    __ movptr(recv, recv_addr);
+    // Push the appendix as a trailing parameter.
-    __ verify_oop(recv);
+    // This must be done before we get the receiver,
+    // since the parameter_size includes it.
+    __ push(index);  // push appendix (MethodType, CallSite, etc.)
+    __ bind(L_no_push);
   }
 
-  // do null check if needed
+  // load receiver if needed (after appendix is pushed so parameter size is correct)
-  if (receiver_null_check) {
+  // Note: no return address pushed yet
-    __ null_check(recv);
+  if (load_receiver) {
+    __ movl(recv, flags);
+    __ andl(recv, ConstantPoolCacheEntry::parameter_size_mask);
+    const int no_return_pc_pushed_yet = -1;  // argument slot correction before we push return address
+    const int receiver_is_at_end      = -1;  // back off one slot to get receiver
+    Address recv_addr = __ argument_address(recv, no_return_pc_pushed_yet + receiver_is_at_end);
+    __ movptr(recv, recv_addr);
+    __ verify_oop(recv);
   }
 
   if (save_flags) {
@@ -2980,16 +3011,14 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
   }
 
   // compute return type
-  __ shrl(flags, ConstantPoolCacheEntry::tosBits);
+  __ shrl(flags, ConstantPoolCacheEntry::tos_state_shift);
-  // Make sure we don't need to mask flags for tosBits after the above shift
+  // Make sure we don't need to mask flags after the above shift
-  ConstantPoolCacheEntry::verify_tosBits();
+  ConstantPoolCacheEntry::verify_tos_state_shift();
   // load return address
   {
-    address table_addr;
+    const address table_addr = (is_invokeinterface || is_invokedynamic) ?
-    if (is_invokeinterface || is_invokedynamic)
+        (address)Interpreter::return_5_addrs_by_index_table() :
-      table_addr = (address)Interpreter::return_5_addrs_by_index_table();
+        (address)Interpreter::return_3_addrs_by_index_table();
-    else
-      table_addr = (address)Interpreter::return_3_addrs_by_index_table();
     ExternalAddress table(table_addr);
     __ lea(rscratch1, table);
     __ movptr(flags, Address(rscratch1, flags, Address::times_ptr));
@@ -2998,7 +3027,7 @@ void TemplateTable::prepare_invoke(Register method, Register index, int byte_no)
   // push return address
   __ push(flags);
 
-  // Restore flag field from the constant pool cache, and restore esi
+  // Restore flags value from the constant pool cache, and restore rsi
   // for later null checks.  r13 is the bytecode pointer
   if (save_flags) {
     __ movl(flags, r13);
@@ -3012,11 +3041,13 @@ void TemplateTable::invokevirtual_helper(Register index,
                                          Register flags) {
   // Uses temporary registers rax, rdx
   assert_different_registers(index, recv, rax, rdx);
+  assert(index == rbx, "");
+  assert(recv  == rcx, "");
 
   // Test for an invoke of a final method
   Label notFinal;
   __ movl(rax, flags);
-  __ andl(rax, (1 << ConstantPoolCacheEntry::vfinalMethod));
+  __ andl(rax, (1 << ConstantPoolCacheEntry::is_vfinal_shift));
   __ jcc(Assembler::zero, notFinal);
 
   const Register method = index;  // method must be rbx
@@ -3024,6 +3055,7 @@ void TemplateTable::invokevirtual_helper(Register index,
          "methodOop must be rbx for interpreter calling convention");
 
   // do the call - the index is actually the method to call
+  // that is, f2 is a vtable index if !is_vfinal, else f2 is a methodOop
   __ verify_oop(method);
 
   // It's final, need a null check here!
@@ -3039,20 +3071,13 @@ void TemplateTable::invokevirtual_helper(Register index,
   // get receiver klass
   __ null_check(recv, oopDesc::klass_offset_in_bytes());
   __ load_klass(rax, recv);
-
   __ verify_oop(rax);
 
   // profile this call
   __ profile_virtual_call(rax, r14, rdx);
 
   // get target methodOop & entry point
-  const int base = instanceKlass::vtable_start_offset() * wordSize;
+  __ lookup_virtual_method(rax, index, method);
-  assert(vtableEntry::size() * wordSize == 8,
-         "adjust the scaling in the code below");
-  __ movptr(method, Address(rax, index,
-                                 Address::times_8,
-                                 base + vtableEntry::method_offset_in_bytes()));
-  __ movptr(rdx, Address(method, methodOopDesc::interpreter_entry_offset()));
   __ jump_from_interpreted(method, rdx);
 }
 
@@ -3060,7 +3085,10 @@ void TemplateTable::invokevirtual_helper(Register index,
 void TemplateTable::invokevirtual(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f2_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no,
+                 rbx,    // method or vtable index
+                 noreg,  // unused itable index
+                 rcx, rdx); // recv, flags
 
   // rbx: index
   // rcx: receiver
@@ -3073,7 +3101,10 @@ void TemplateTable::invokevirtual(int byte_no) {
 void TemplateTable::invokespecial(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no, rbx, noreg,  // get f1 methodOop
+                 rcx);  // get receiver also for null check
+  __ verify_oop(rcx);
+  __ null_check(rcx);
   // do the call
   __ verify_oop(rbx);
   __ profile_call(rax);
@@ -3084,7 +3115,7 @@ void TemplateTable::invokespecial(int byte_no) {
 void TemplateTable::invokestatic(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rbx, noreg, byte_no);
+  prepare_invoke(byte_no, rbx);  // get f1 methodOop
   // do the call
   __ verify_oop(rbx);
   __ profile_call(rax);
@@ -3100,10 +3131,11 @@ void TemplateTable::fast_invokevfinal(int byte_no) {
 void TemplateTable::invokeinterface(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  prepare_invoke(rax, rbx, byte_no);
+  prepare_invoke(byte_no, rax, rbx,  // get f1 klassOop, f2 itable index
+                 rcx, rdx); // recv, flags
 
-  // rax: Interface
+  // rax: interface klass (from f1)
-  // rbx: index
+  // rbx: itable index (from f2)
   // rcx: receiver
   // rdx: flags
 
@@ -3113,7 +3145,7 @@ void TemplateTable::invokeinterface(int byte_no) {
   // another compliant java compiler.
   Label notMethod;
   __ movl(r14, rdx);
-  __ andl(r14, (1 << ConstantPoolCacheEntry::methodInterface));
+  __ andl(r14, (1 << ConstantPoolCacheEntry::is_forced_virtual_shift));
   __ jcc(Assembler::zero, notMethod);
 
   invokevirtual_helper(rbx, rcx, rdx);
@@ -3121,6 +3153,7 @@ void TemplateTable::invokeinterface(int byte_no) {
 
   // Get receiver klass into rdx - also a null check
   __ restore_locals();  // restore r14
+  __ null_check(rcx, oopDesc::klass_offset_in_bytes());
   __ load_klass(rdx, rcx);
   __ verify_oop(rdx);
 
@@ -3135,7 +3168,7 @@ void TemplateTable::invokeinterface(int byte_no) {
                              rbx, r13,
                              no_such_interface);
 
-  // rbx,: methodOop to call
+  // rbx: methodOop to call
   // rcx: receiver
   // Check for abstract method error
   // Note: This should be done more efficiently via a throw_abstract_method_error
@@ -3172,12 +3205,42 @@ void TemplateTable::invokeinterface(int byte_no) {
                    InterpreterRuntime::throw_IncompatibleClassChangeError));
   // the call_VM checks for exception, so we should never return here.
   __ should_not_reach_here();
+}
+
+
+void TemplateTable::invokehandle(int byte_no) {
+  transition(vtos, vtos);
+  assert(byte_no == f12_oop, "use this argument");
+  const Register rbx_method = rbx;  // f2
+  const Register rax_mtype  = rax;  // f1
+  const Register rcx_recv   = rcx;
+  const Register rdx_flags  = rdx;
+
+  if (!EnableInvokeDynamic) {
+    // rewriter does not generate this bytecode
+    __ should_not_reach_here();
     return;
   }
 
+  prepare_invoke(byte_no,
+                 rbx_method, rax_mtype,  // get f2 methodOop, f1 MethodType
+                 rcx_recv);
+  __ verify_oop(rbx_method);
+  __ verify_oop(rcx_recv);
+  __ null_check(rcx_recv);
+
+  // Note:  rax_mtype is already pushed (if necessary) by prepare_invoke
+
+  // FIXME: profile the LambdaForm also
+  __ profile_final_call(rax);
+
+  __ jump_from_interpreted(rbx_method, rdx);
+}
+
+
 void TemplateTable::invokedynamic(int byte_no) {
   transition(vtos, vtos);
-  assert(byte_no == f1_oop, "use this argument");
+  assert(byte_no == f12_oop, "use this argument");
 
   if (!EnableInvokeDynamic) {
     // We should not encounter this bytecode if !EnableInvokeDynamic.
@@ -3190,26 +3253,23 @@ void TemplateTable::invokedynamic(int byte_no) {
     return;
   }
 
-  prepare_invoke(rax, rbx, byte_no);
+  const Register rbx_method   = rbx;
+  const Register rax_callsite = rax;
 
-  // rax: CallSite object (f1)
+  prepare_invoke(byte_no, rbx_method, rax_callsite);
-  // rbx: unused (f2)
-  // rcx: receiver address
-  // rdx: flags (unused)
 
-  Register rax_callsite      = rax;
+  // rax: CallSite object (from f1)
-  Register rcx_method_handle = rcx;
+  // rbx: MH.linkToCallSite method (from f2)
+
+  // Note:  rax_callsite is already pushed by prepare_invoke
 
   // %%% should make a type profile for any invokedynamic that takes a ref argument
   // profile this call
   __ profile_call(r13);
 
   __ verify_oop(rax_callsite);
-  __ load_heap_oop(rcx_method_handle, Address(rax_callsite, __ delayed_value(java_lang_invoke_CallSite::target_offset_in_bytes, rdx)));
+
-  __ null_check(rcx_method_handle);
+  __ jump_from_interpreted(rbx_method, rdx);
-  __ verify_oop(rcx_method_handle);
-  __ prepare_to_jump_from_interpreted();
-  __ jump_to_method_handle_entry(rcx_method_handle, rdx);
 }
 
 

hotspot/src/cpu/x86/vm/templateTable_x86_64.hpp

@@ -25,7 +25,12 @@
 #ifndef CPU_X86_VM_TEMPLATETABLE_X86_64_HPP
 #define CPU_X86_VM_TEMPLATETABLE_X86_64_HPP
 
-  static void prepare_invoke(Register method, Register index, int byte_no);
+  static void prepare_invoke(int byte_no,
+                             Register method,         // linked method (or i-klass)
+                             Register index = noreg,  // itable index, MethodType, etc.
+                             Register recv  = noreg,  // if caller wants to see it
+                             Register flags = noreg   // if caller wants to test it
+                             );
   static void invokevirtual_helper(Register index, Register recv,
                                    Register flags);
   static void volatile_barrier(Assembler::Membar_mask_bits order_constraint);

hotspot/src/cpu/x86/vm/vtableStubs_x86_32.cpp

@@ -76,8 +76,7 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
   // get receiver klass
   address npe_addr = __ pc();
   __ movptr(rax, Address(rcx, oopDesc::klass_offset_in_bytes()));
-  // compute entry offset (in words)
+
-  int entry_offset = instanceKlass::vtable_start_offset() + vtable_index*vtableEntry::size();
 #ifndef PRODUCT
   if (DebugVtables) {
     Label L;
@@ -93,7 +92,8 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
   const Register method = rbx;
 
   // load methodOop and target address
-  __ movptr(method, Address(rax, entry_offset*wordSize + vtableEntry::method_offset_in_bytes()));
+  __ lookup_virtual_method(rax, vtable_index, method);
+
   if (DebugVtables) {
     Label L;
     __ cmpptr(method, (int32_t)NULL_WORD);

hotspot/src/cpu/x86/vm/vtableStubs_x86_64.cpp

@@ -69,10 +69,6 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
   address npe_addr = __ pc();
   __ load_klass(rax, j_rarg0);
 
-  // compute entry offset (in words)
-  int entry_offset =
-    instanceKlass::vtable_start_offset() + vtable_index * vtableEntry::size();
-
 #ifndef PRODUCT
   if (DebugVtables) {
     Label L;
@@ -90,9 +86,8 @@ VtableStub* VtableStubs::create_vtable_stub(int vtable_index) {
   // load methodOop and target address
   const Register method = rbx;
 
-  __ movptr(method, Address(rax,
+  __ lookup_virtual_method(rax, vtable_index, method);
-                            entry_offset * wordSize +
+
-                            vtableEntry::method_offset_in_bytes()));
   if (DebugVtables) {
     Label L;
     __ cmpptr(method, (int32_t)NULL_WORD);

hotspot/src/cpu/x86/vm/x86_32.ad

@@ -1367,22 +1367,6 @@ int emit_deopt_handler(CodeBuffer& cbuf) {
   return offset;
 }
 
-
-const bool Matcher::match_rule_supported(int opcode) {
-  if (!has_match_rule(opcode))
-    return false;
-
-  switch (opcode) {
-    case Op_PopCountI:
-    case Op_PopCountL:
-      if (!UsePopCountInstruction)
-        return false;
-    break;
-  }
-  
-  return true;  // Per default match rules are supported.
-}
-
 int Matcher::regnum_to_fpu_offset(int regnum) {
   return regnum - 32; // The FP registers are in the second chunk
 }

hotspot/src/cpu/x86/vm/x86_64.ad

@@ -1513,22 +1513,6 @@ int emit_deopt_handler(CodeBuffer& cbuf)
   return offset;
 }
 
-
-const bool Matcher::match_rule_supported(int opcode) {
-  if (!has_match_rule(opcode))
-    return false;
-
-  switch (opcode) {
-    case Op_PopCountI:
-    case Op_PopCountL:
-      if (!UsePopCountInstruction)
-        return false;
-    break;
-  }
-
-  return true;  // Per default match rules are supported.
-}
-
 int Matcher::regnum_to_fpu_offset(int regnum)
 {
   return regnum - 32; // The FP registers are in the second chunk
@@ -6427,6 +6411,31 @@ instruct castP2X(rRegL dst, rRegP src)
   ins_pipe(ialu_reg_reg); // XXX
 %}
 
+// Convert oop into int for vectors alignment masking
+instruct convP2I(rRegI dst, rRegP src)
+%{
+  match(Set dst (ConvL2I (CastP2X src)));
+
+  format %{ "movl    $dst, $src\t# ptr -> int" %}
+  ins_encode %{
+    __ movl($dst$$Register, $src$$Register);
+  %}
+  ins_pipe(ialu_reg_reg); // XXX
+%}
+
+// Convert compressed oop into int for vectors alignment masking
+// in case of 32bit oops (heap < 4Gb).
+instruct convN2I(rRegI dst, rRegN src)
+%{
+  predicate(Universe::narrow_oop_shift() == 0);
+  match(Set dst (ConvL2I (CastP2X (DecodeN src))));
+
+  format %{ "movl    $dst, $src\t# compressed ptr -> int" %}
+  ins_encode %{
+    __ movl($dst$$Register, $src$$Register);
+  %}
+  ins_pipe(ialu_reg_reg); // XXX
+%}
 
 // Convert oop pointer into compressed form
 instruct encodeHeapOop(rRegN dst, rRegP src, rFlagsReg cr) %{
@@ -10049,11 +10058,10 @@ instruct MoveD2L_reg_reg(rRegL dst, regD src) %{
   ins_pipe( pipe_slow );
 %}
 
-// The next instructions have long latency and use Int unit. Set high cost.
 instruct MoveI2F_reg_reg(regF dst, rRegI src) %{
   match(Set dst (MoveI2F src));
   effect(DEF dst, USE src);
-  ins_cost(300);
+  ins_cost(100);
   format %{ "movd    $dst,$src\t# MoveI2F" %}
   ins_encode %{
     __ movdl($dst$$XMMRegister, $src$$Register);
@@ -10064,7 +10072,7 @@ instruct MoveI2F_reg_reg(regF dst, rRegI src) %{
 instruct MoveL2D_reg_reg(regD dst, rRegL src) %{
   match(Set dst (MoveL2D src));
   effect(DEF dst, USE src);
-  ins_cost(300);
+  ins_cost(100);
   format %{ "movd    $dst,$src\t# MoveL2D" %}
   ins_encode %{
      __ movdq($dst$$XMMRegister, $src$$Register);

hotspot/src/cpu/zero/vm/cppInterpreter_zero.cpp

@@ -646,16 +646,15 @@ int CppInterpreter::method_handle_entry(methodOop method,
   oop method_type = (oop) p;
 
   // The MethodHandle is in the slot after the arguments
-  oop form = java_lang_invoke_MethodType::form(method_type);
+  int num_vmslots = argument_slots - 1;
-  int num_vmslots = java_lang_invoke_MethodTypeForm::vmslots(form);
-  assert(argument_slots == num_vmslots + 1, "should be");
   oop method_handle = VMSLOTS_OBJECT(num_vmslots);
 
   // InvokeGeneric requires some extra shuffling
   oop mhtype = java_lang_invoke_MethodHandle::type(method_handle);
   bool is_exact = mhtype == method_type;
   if (!is_exact) {
-    if (method->intrinsic_id() == vmIntrinsics::_invokeExact) {
+    if (true || // FIXME
+        method->intrinsic_id() == vmIntrinsics::_invokeExact) {
       CALL_VM_NOCHECK_NOFIX(
         SharedRuntime::throw_WrongMethodTypeException(
           thread, method_type, mhtype));
@@ -670,8 +669,8 @@ int CppInterpreter::method_handle_entry(methodOop method,
     // NB the x86 code for this (in methodHandles_x86.cpp, search for
     // "genericInvoker") is really really odd.  I'm hoping it's trying
     // to accomodate odd VM/class library combinations I can ignore.
-    oop adapter = java_lang_invoke_MethodTypeForm::genericInvoker(form);
+    oop adapter = NULL; //FIXME: load the adapter from the CP cache
-    if (adapter == NULL) {
+    IF (adapter == NULL) {
       CALL_VM_NOCHECK_NOFIX(
         SharedRuntime::throw_WrongMethodTypeException(
           thread, method_type, mhtype));
@@ -761,7 +760,7 @@ void CppInterpreter::process_method_handle(oop method_handle, TRAPS) {
           return;
       }
       if (entry_kind != MethodHandles::_invokespecial_mh) {
-        int index = java_lang_invoke_DirectMethodHandle::vmindex(method_handle);
+        intptr_t index = java_lang_invoke_DirectMethodHandle::vmindex(method_handle);
         instanceKlass* rcvrKlass =
           (instanceKlass *) receiver->klass()->klass_part();
         if (entry_kind == MethodHandles::_invokevirtual_mh) {
@@ -1179,8 +1178,7 @@ BasicType CppInterpreter::result_type_of_handle(oop method_handle) {
 intptr_t* CppInterpreter::calculate_unwind_sp(ZeroStack* stack,
                                               oop method_handle) {
   oop method_type = java_lang_invoke_MethodHandle::type(method_handle);
-  oop form = java_lang_invoke_MethodType::form(method_type);
+  int argument_slots = java_lang_invoke_MethodType::ptype_slot_count(method_type);
-  int argument_slots = java_lang_invoke_MethodTypeForm::vmslots(form);
 
   return stack->sp() + argument_slots;
 }

hotspot/src/cpu/zero/vm/globals_zero.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2000, 2012, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2007, 2008, 2009, 2010, 2011 Red Hat, Inc.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -60,4 +60,7 @@ define_pd_global(bool,  UseMembar,            false);
 
 // GC Ergo Flags
 define_pd_global(intx, CMSYoungGenPerWorker, 16*M);  // default max size of CMS young gen, per GC worker thread
+
+#define ARCH_FLAGS(develop, product, diagnostic, experimental, notproduct)
+
 #endif // CPU_ZERO_VM_GLOBALS_ZERO_HPP

hotspot/src/cpu/zero/vm/interpreterGenerator_zero.hpp

@@ -38,6 +38,5 @@
   address generate_empty_entry();
   address generate_accessor_entry();
   address generate_Reference_get_entry();
-  address generate_method_handle_entry();
 
 #endif // CPU_ZERO_VM_INTERPRETERGENERATOR_ZERO_HPP