--- a/agent/src/os/solaris/proc/saproc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/agent/src/os/solaris/proc/saproc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -314,7 +314,7 @@
     handle = dlopen(name, mode);
   }
   if (_libsaproc_debug) {
-    printf("libsaproc DEBUG: pathmap_dlopen %s return 0x%x\n", name, handle);
+    printf("libsaproc DEBUG: pathmap_dlopen %s return 0x%lx\n", name, (unsigned long) handle);
   }
   return handle;
 }
@@ -661,30 +661,30 @@
   // read FileMapHeader
   size_t n = read(fd, pheader, sizeof(struct FileMapHeader));
   if (n != sizeof(struct FileMapHeader)) {
-    free(pheader);
-    close(fd);
     char errMsg[ERR_MSG_SIZE];
     sprintf(errMsg, "unable to read shared archive file map header from %s", classes_jsa);
+    close(fd);
+    free(pheader);
     THROW_NEW_DEBUGGER_EXCEPTION_(errMsg, 1);
   }
 
   // check file magic
   if (pheader->_magic != 0xf00baba2) {
-    free(pheader);
-    close(fd);
     char errMsg[ERR_MSG_SIZE];
     sprintf(errMsg, "%s has bad shared archive magic 0x%x, expecting 0xf00baba2",
                    classes_jsa, pheader->_magic);
+    close(fd);
+    free(pheader);
     THROW_NEW_DEBUGGER_EXCEPTION_(errMsg, 1);
   }
 
   // check version
   if (pheader->_version != CURRENT_ARCHIVE_VERSION) {
-    free(pheader);
-    close(fd);
     char errMsg[ERR_MSG_SIZE];
     sprintf(errMsg, "%s has wrong shared archive version %d, expecting %d",
                    classes_jsa, pheader->_version, CURRENT_ARCHIVE_VERSION);
+    close(fd);
+    free(pheader);
     THROW_NEW_DEBUGGER_EXCEPTION_(errMsg, 1);
   }
 

--- a/make/excludeSrc.make	Thu Aug 21 14:16:15 2014 -0700
+++ b/make/excludeSrc.make	Thu Aug 28 14:53:43 2014 -0700
@@ -119,8 +119,8 @@
       CFLAGS += -DINCLUDE_NMT=0
 
       Src_Files_EXCLUDE += \
-	 memBaseline.cpp memPtr.cpp memRecorder.cpp memReporter.cpp memSnapshot.cpp memTrackWorker.cpp \
-	 memTracker.cpp nmtDCmd.cpp
+	 memBaseline.cpp memReporter.cpp mallocTracker.cpp virtualMemoryTracker.cpp nmtCommon.cpp \
+	 memTracker.cpp nmtDCmd.cpp mallocSiteTable.cpp
 endif
 
 -include $(HS_ALT_MAKE)/excludeSrc.make

--- a/make/jprt.properties	Thu Aug 21 14:16:15 2014 -0700
+++ b/make/jprt.properties	Thu Aug 28 14:53:43 2014 -0700
@@ -356,14 +356,15 @@
   ${jprt.my.windows.x64}-fastdebug-c2-internalvmtests
 
 jprt.make.rule.test.targets.standard.reg.group = \
-  ${jprt.my.solaris.sparcv9}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.solaris.x64}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.linux.i586}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.linux.x64}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.windows.i586}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.windows.x64}-{product|fastdebug}-c2-GROUP, \
-  ${jprt.my.linux.i586}-{product|fastdebug}-c1-GROUP, \
-  ${jprt.my.windows.i586}-{product|fastdebug}-c1-GROUP
+  ${jprt.my.solaris.sparcv9}-fastdebug-c2-GROUP, \
+  ${jprt.my.solaris.x64}-fastdebug-c2-GROUP, \
+  ${jprt.my.linux.i586}-fastdebug-c2-GROUP, \
+  ${jprt.my.linux.x64}-fastdebug-c2-GROUP, \
+  ${jprt.my.macosx.x64}-fastdebug-c2-GROUP, \
+  ${jprt.my.windows.i586}-fastdebug-c2-GROUP, \
+  ${jprt.my.windows.x64}-fastdebug-c2-GROUP, \
+  ${jprt.my.linux.i586}-fastdebug-c1-GROUP, \
+  ${jprt.my.windows.i586}-fastdebug-c1-GROUP
 
 jprt.make.rule.test.targets.standard = \
   ${jprt.make.rule.test.targets.standard.client}, \

--- a/make/solaris/makefiles/dtrace.make	Thu Aug 21 14:16:15 2014 -0700
+++ b/make/solaris/makefiles/dtrace.make	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2005, 2013, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2005, 2014, 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
@@ -58,6 +58,8 @@
 DTRACE_COMMON_SRCDIR = $(GAMMADIR)/src/os/posix/dtrace
 DTRACE = dtrace
 DTRACE.o = $(DTRACE).o
+DTRACE_JHELPER = dtrace_jhelper
+DTRACE_JHELPER.o = $(DTRACE_JHELPER).o
 
 # to remove '-g' option which causes link problems
 # also '-z nodefs' is used as workaround
@@ -255,7 +257,10 @@
 endif
 
 $(DTRACE).d: $(DTRACE_COMMON_SRCDIR)/hotspot.d $(DTRACE_COMMON_SRCDIR)/hotspot_jni.d \
-             $(DTRACE_COMMON_SRCDIR)/hs_private.d $(DTRACE_SRCDIR)/jhelper.d
+             $(DTRACE_COMMON_SRCDIR)/hs_private.d
+	$(QUIETLY) cat $^ > $@
+
+$(DTRACE_JHELPER).d: $(DTRACE_SRCDIR)/jhelper.d
 	$(QUIETLY) cat $^ > $@
 
 DTraced_Files = ciEnv.o \
@@ -280,7 +285,7 @@
                 vmGCOperations.o \
 
 # Dtrace is available, so we build $(DTRACE.o)  
-$(DTRACE.o): $(DTRACE).d $(JVMOFFS).h $(JVMOFFS)Index.h $(DTraced_Files)
+$(DTRACE.o): $(DTRACE).d $(DTraced_Files)
 	@echo Compiling $(DTRACE).d
 
 	$(QUIETLY) $(DTRACE_PROG) $(DTRACE_OPTS) -C -I. -G -xlazyload -o $@ -s $(DTRACE).d \
@@ -344,6 +349,11 @@
 
 dtrace_gen_headers: $(DtraceOutDir)/hotspot.h $(DtraceOutDir)/hotspot_jni.h $(DtraceOutDir)/hs_private.h
 
+# The jhelper.d and hotspot probes are separated into two different SUNW_dof sections.
+# Now the jhelper.d is built without the -Xlazyload flag.
+$(DTRACE_JHELPER.o) : $(DTRACE_JHELPER).d $(JVMOFFS).h $(JVMOFFS)Index.h
+	@echo Compiling $(DTRACE_JHELPER).d
+	$(QUIETLY) $(DTRACE_PROG) $(DTRACE_OPTS) -C -I. -G -o $@ -s $(DTRACE_JHELPER).d
 
 .PHONY: dtraceCheck
 
@@ -372,7 +382,7 @@
 ifneq ("${DTRACE_PROG}", "")
 ifeq ("${HOTSPOT_DISABLE_DTRACE_PROBES}", "")
 
-DTRACE_OBJS = $(DTRACE.o) $(JVMOFFS.o)
+DTRACE_OBJS = $(DTRACE.o) $(JVMOFFS.o) $(DTRACE_JHELPER.o)
 CFLAGS += $(DTRACE_INCL) -DDTRACE_ENABLED
 MAPFILE_DTRACE_OPT = $(MAPFILE_DTRACE)
 

--- a/src/cpu/ppc/vm/assembler_ppc.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/assembler_ppc.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -298,6 +298,7 @@
     LWZ_OPCODE   = (32u << OPCODE_SHIFT),
     LWZX_OPCODE  = (31u << OPCODE_SHIFT |  23u << 1),
     LWZU_OPCODE  = (33u << OPCODE_SHIFT),
+    LWBRX_OPCODE = (31u << OPCODE_SHIFT |  534 << 1),
 
     LHA_OPCODE   = (42u << OPCODE_SHIFT),
     LHAX_OPCODE  = (31u << OPCODE_SHIFT | 343u << 1),
@@ -306,6 +307,7 @@
     LHZ_OPCODE   = (40u << OPCODE_SHIFT),
     LHZX_OPCODE  = (31u << OPCODE_SHIFT | 279u << 1),
     LHZU_OPCODE  = (41u << OPCODE_SHIFT),
+    LHBRX_OPCODE = (31u << OPCODE_SHIFT |  790 << 1),
 
     LBZ_OPCODE   = (34u << OPCODE_SHIFT),
     LBZX_OPCODE  = (31u << OPCODE_SHIFT |  87u << 1),
@@ -1364,11 +1366,17 @@
   inline void lwax( Register d, Register s1, Register s2);
   inline void lwa(  Register d, int si16,    Register s1);
 
+  // 4 bytes reversed
+  inline void lwbrx( Register d, Register s1, Register s2);
+
   // 2 bytes
   inline void lhzx( Register d, Register s1, Register s2);
   inline void lhz(  Register d, int si16,    Register s1);
   inline void lhzu( Register d, int si16,    Register s1);
 
+  // 2 bytes reversed
+  inline void lhbrx( Register d, Register s1, Register s2);
+
   // 2 bytes
   inline void lhax( Register d, Register s1, Register s2);
   inline void lha(  Register d, int si16,    Register s1);
@@ -1858,10 +1866,12 @@
   inline void lwz(  Register d, int si16);
   inline void lwax( Register d, Register s2);
   inline void lwa(  Register d, int si16);
+  inline void lwbrx(Register d, Register s2);
   inline void lhzx( Register d, Register s2);
   inline void lhz(  Register d, int si16);
   inline void lhax( Register d, Register s2);
   inline void lha(  Register d, int si16);
+  inline void lhbrx(Register d, Register s2);
   inline void lbzx( Register d, Register s2);
   inline void lbz(  Register d, int si16);
   inline void ldx(  Register d, Register s2);

--- a/src/cpu/ppc/vm/assembler_ppc.inline.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/assembler_ppc.inline.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -263,10 +263,14 @@
 inline void Assembler::lwax( Register d, Register s1, Register s2) { emit_int32(LWAX_OPCODE | rt(d) | ra0mem(s1) | rb(s2));}
 inline void Assembler::lwa(  Register d, int si16,    Register s1) { emit_int32(LWA_OPCODE  | rt(d) | ds(si16)   | ra0mem(s1));}
 
+inline void Assembler::lwbrx( Register d, Register s1, Register s2) { emit_int32(LWBRX_OPCODE | rt(d) | ra0mem(s1) | rb(s2));}
+
 inline void Assembler::lhzx( Register d, Register s1, Register s2) { emit_int32(LHZX_OPCODE | rt(d) | ra0mem(s1) | rb(s2));}
 inline void Assembler::lhz(  Register d, int si16,    Register s1) { emit_int32(LHZ_OPCODE  | rt(d) | d1(si16)   | ra0mem(s1));}
 inline void Assembler::lhzu( Register d, int si16,    Register s1) { assert(d != s1, "according to ibm manual"); emit_int32(LHZU_OPCODE | rt(d) | d1(si16) | rta0mem(s1));}
 
+inline void Assembler::lhbrx( Register d, Register s1, Register s2) { emit_int32(LHBRX_OPCODE | rt(d) | ra0mem(s1) | rb(s2));}
+
 inline void Assembler::lhax( Register d, Register s1, Register s2) { emit_int32(LHAX_OPCODE | rt(d) | ra0mem(s1) | rb(s2));}
 inline void Assembler::lha(  Register d, int si16,    Register s1) { emit_int32(LHA_OPCODE  | rt(d) | d1(si16)   | ra0mem(s1));}
 inline void Assembler::lhau( Register d, int si16,    Register s1) { assert(d != s1, "according to ibm manual"); emit_int32(LHAU_OPCODE | rt(d) | d1(si16) | rta0mem(s1));}
@@ -736,10 +740,12 @@
 inline void Assembler::lwz(  Register d, int si16   ) { emit_int32( LWZ_OPCODE  | rt(d) | d1(si16));}
 inline void Assembler::lwax( Register d, Register s2) { emit_int32( LWAX_OPCODE | rt(d) | rb(s2));}
 inline void Assembler::lwa(  Register d, int si16   ) { emit_int32( LWA_OPCODE  | rt(d) | ds(si16));}
+inline void Assembler::lwbrx(Register d, Register s2) { emit_int32( LWBRX_OPCODE| rt(d) | rb(s2));}
 inline void Assembler::lhzx( Register d, Register s2) { emit_int32( LHZX_OPCODE | rt(d) | rb(s2));}
 inline void Assembler::lhz(  Register d, int si16   ) { emit_int32( LHZ_OPCODE  | rt(d) | d1(si16));}
 inline void Assembler::lhax( Register d, Register s2) { emit_int32( LHAX_OPCODE | rt(d) | rb(s2));}
 inline void Assembler::lha(  Register d, int si16   ) { emit_int32( LHA_OPCODE  | rt(d) | d1(si16));}
+inline void Assembler::lhbrx(Register d, Register s2) { emit_int32( LHBRX_OPCODE| rt(d) | rb(s2));}
 inline void Assembler::lbzx( Register d, Register s2) { emit_int32( LBZX_OPCODE | rt(d) | rb(s2));}
 inline void Assembler::lbz(  Register d, int si16   ) { emit_int32( LBZ_OPCODE  | rt(d) | d1(si16));}
 inline void Assembler::ld(   Register d, int si16   ) { emit_int32( LD_OPCODE   | rt(d) | ds(si16));}

--- a/src/cpu/ppc/vm/cppInterpreterGenerator_ppc.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/cppInterpreterGenerator_ppc.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2014, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2012, 2013 SAP AG. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -26,8 +26,9 @@
 #ifndef CPU_PPC_VM_CPPINTERPRETERGENERATOR_PPC_HPP
 #define CPU_PPC_VM_CPPINTERPRETERGENERATOR_PPC_HPP
 
-  address generate_normal_entry(void);
-  address generate_native_entry(void);
+  address generate_normal_entry(bool synchronized);
+  address generate_native_entry(bool synchronized);
+  address generate_math_entry(AbstractInterpreter::MethodKind kind) { return NULL; }
 
   void lock_method(void);
   void unlock_method(void);

--- a/src/cpu/ppc/vm/cppInterpreter_ppc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/cppInterpreter_ppc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -938,8 +938,9 @@
 // Interpreter stub for calling a native method. (C++ interpreter)
 // This sets up a somewhat different looking stack for calling the native method
 // than the typical interpreter frame setup.
+// The synchronized parameter is ignored.
 //
-address CppInterpreterGenerator::generate_native_entry(void) {
+address CppInterpreterGenerator::generate_native_entry(bool synchronized) {
   if (native_entry != NULL) return native_entry;
   address entry = __ pc();
 
@@ -1729,7 +1730,8 @@
   __ std(R0, BasicObjectLock::obj_offset_in_bytes(), stack_base);  // Mark lock as unused
 }
 
-address CppInterpreterGenerator::generate_normal_entry(void) {
+// The synchronized parameter is ignored
+address CppInterpreterGenerator::generate_normal_entry(bool synchronized) {
   if (interpreter_frame_manager != NULL) return interpreter_frame_manager;
 
   address entry = __ pc();
@@ -2789,38 +2791,6 @@
   return interpreter_frame_manager;
 }
 
-// Generate code for various sorts of method entries
-//
-address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter::MethodKind kind) {
-  address entry_point = NULL;
-
-  switch (kind) {
-    case Interpreter::zerolocals                 :                                                                              break;
-    case Interpreter::zerolocals_synchronized    :                                                                              break;
-    case Interpreter::native                     : // Fall thru
-    case Interpreter::native_synchronized        : entry_point = ((CppInterpreterGenerator*)this)->generate_native_entry();     break;
-    case Interpreter::empty                      :                                                                              break;
-    case Interpreter::accessor                   : entry_point = ((InterpreterGenerator*)this)->generate_accessor_entry();      break;
-    case Interpreter::abstract                   : entry_point = ((InterpreterGenerator*)this)->generate_abstract_entry();      break;
-    // These are special interpreter intrinsics which we don't support so far.
-    case Interpreter::java_lang_math_sin         :                                                                              break;
-    case Interpreter::java_lang_math_cos         :                                                                              break;
-    case Interpreter::java_lang_math_tan         :                                                                              break;
-    case Interpreter::java_lang_math_abs         :                                                                              break;
-    case Interpreter::java_lang_math_log         :                                                                              break;
-    case Interpreter::java_lang_math_log10       :                                                                              break;
-    case Interpreter::java_lang_math_sqrt        :                                                                              break;
-    case Interpreter::java_lang_math_pow         :                                                                              break;
-    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;
-  }
-
-  if (entry_point) {
-    return entry_point;
-  }
-  return ((InterpreterGenerator*)this)->generate_normal_entry();
-}
 
 InterpreterGenerator::InterpreterGenerator(StubQueue* code)
  : CppInterpreterGenerator(code) {

--- a/src/cpu/ppc/vm/interp_masm_ppc_64.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/interp_masm_ppc_64.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -119,9 +119,15 @@
     // Call the Interpreter::remove_activation_preserving_args_entry()
     // func to get the address of the same-named entrypoint in the
     // generated interpreter code.
+#if defined(ABI_ELFv2)
+    call_c(CAST_FROM_FN_PTR(address,
+                            Interpreter::remove_activation_preserving_args_entry),
+           relocInfo::none);
+#else
     call_c(CAST_FROM_FN_PTR(FunctionDescriptor*,
                             Interpreter::remove_activation_preserving_args_entry),
            relocInfo::none);
+#endif
 
     // Jump to Interpreter::_remove_activation_preserving_args_entry.
     mtctr(R3_RET);
@@ -331,29 +337,40 @@
 void InterpreterMacroAssembler::get_2_byte_integer_at_bcp(int         bcp_offset,
                                                           Register    Rdst,
                                                           signedOrNot is_signed) {
+#if defined(VM_LITTLE_ENDIAN)
+  if (bcp_offset) {
+    load_const_optimized(Rdst, bcp_offset);
+    lhbrx(Rdst, R14_bcp, Rdst);
+  } else {
+    lhbrx(Rdst, R14_bcp);
+  }
+  if (is_signed == Signed) {
+    extsh(Rdst, Rdst);
+  }
+#else
   // Read Java big endian format.
   if (is_signed == Signed) {
     lha(Rdst, bcp_offset, R14_bcp);
   } else {
     lhz(Rdst, bcp_offset, R14_bcp);
   }
-#if 0
-  assert(Rtmp != Rdst, "need separate temp register");
-  Register Rfirst = Rtmp;
-  lbz(Rfirst, bcp_offset, R14_bcp); // first byte
-  lbz(Rdst, bcp_offset+1, R14_bcp); // second byte
-
-  // Rdst = ((Rfirst<<8) & 0xFF00) | (Rdst &~ 0xFF00)
-  rldimi(/*RA=*/Rdst, /*RS=*/Rfirst, /*sh=*/8, /*mb=*/48);
-  if (is_signed == Signed) {
-    extsh(Rdst, Rdst);
-  }
 #endif
 }
 
 void InterpreterMacroAssembler::get_4_byte_integer_at_bcp(int         bcp_offset,
                                                           Register    Rdst,
                                                           signedOrNot is_signed) {
+#if defined(VM_LITTLE_ENDIAN)
+  if (bcp_offset) {
+    load_const_optimized(Rdst, bcp_offset);
+    lwbrx(Rdst, R14_bcp, Rdst);
+  } else {
+    lwbrx(Rdst, R14_bcp);
+  }
+  if (is_signed == Signed) {
+    extsw(Rdst, Rdst);
+  }
+#else
   // Read Java big endian format.
   if (bcp_offset & 3) { // Offset unaligned?
     load_const_optimized(Rdst, bcp_offset);
@@ -369,18 +386,26 @@
       lwz(Rdst, bcp_offset, R14_bcp);
     }
   }
+#endif
 }
 
+
 // Load the constant pool cache index from the bytecode stream.
 //
 // Kills / writes:
 //   - Rdst, Rscratch
 void InterpreterMacroAssembler::get_cache_index_at_bcp(Register Rdst, int bcp_offset, size_t index_size) {
   assert(bcp_offset > 0, "bcp is still pointing to start of bytecode");
+  // Cache index is always in the native format, courtesy of Rewriter.
   if (index_size == sizeof(u2)) {
-    get_2_byte_integer_at_bcp(bcp_offset, Rdst, Unsigned);
+    lhz(Rdst, bcp_offset, R14_bcp);
   } else if (index_size == sizeof(u4)) {
-    get_4_byte_integer_at_bcp(bcp_offset, Rdst, Signed);
+    if (bcp_offset & 3) {
+      load_const_optimized(Rdst, bcp_offset);
+      lwax(Rdst, R14_bcp, Rdst);
+    } else {
+      lwa(Rdst, bcp_offset, R14_bcp);
+    }
     assert(ConstantPool::decode_invokedynamic_index(~123) == 123, "else change next line");
     nand(Rdst, Rdst, Rdst); // convert to plain index
   } else if (index_size == sizeof(u1)) {
@@ -397,6 +422,29 @@
   add(cache, R27_constPoolCache, cache);
 }
 
+// Load 4-byte signed or unsigned integer in Java format (that is, big-endian format)
+// from (Rsrc)+offset.
+void InterpreterMacroAssembler::get_u4(Register Rdst, Register Rsrc, int offset,
+                                       signedOrNot is_signed) {
+#if defined(VM_LITTLE_ENDIAN)
+  if (offset) {
+    load_const_optimized(Rdst, offset);
+    lwbrx(Rdst, Rdst, Rsrc);
+  } else {
+    lwbrx(Rdst, Rsrc);
+  }
+  if (is_signed == Signed) {
+    extsw(Rdst, Rdst);
+  }
+#else
+  if (is_signed == Signed) {
+    lwa(Rdst, offset, Rsrc);
+  } else {
+    lwz(Rdst, offset, Rsrc);
+  }
+#endif
+}
+
 // Load object from cpool->resolved_references(index).
 void InterpreterMacroAssembler::load_resolved_reference_at_index(Register result, Register index) {
   assert_different_registers(result, index);

--- a/src/cpu/ppc/vm/interp_masm_ppc_64.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/interp_masm_ppc_64.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -130,6 +130,7 @@
 
   void get_cache_and_index_at_bcp(Register cache, int bcp_offset, size_t index_size = sizeof(u2));
 
+  void get_u4(Register Rdst, Register Rsrc, int offset, signedOrNot is_signed);
 
   // common code
 

--- a/src/cpu/ppc/vm/interpreterGenerator_ppc.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/interpreterGenerator_ppc.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2014, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2012, 2013 SAP AG. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -31,7 +31,12 @@
  private:
 
   address generate_abstract_entry(void);
-  address generate_accessor_entry(void);
+  address generate_jump_to_normal_entry(void);
+  address generate_accessor_entry(void) { return generate_jump_to_normal_entry(); }
+  address generate_empty_entry(void) { return generate_jump_to_normal_entry(); }
   address generate_Reference_get_entry(void);
 
+  // Not supported
+  address generate_CRC32_update_entry() { return NULL; }
+  address generate_CRC32_updateBytes_entry(AbstractInterpreter::MethodKind kind) { return NULL; }
 #endif // CPU_PPC_VM_INTERPRETERGENERATOR_PPC_HPP

--- a/src/cpu/ppc/vm/interpreter_ppc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/interpreter_ppc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -428,6 +428,19 @@
   return entry;
 }
 
+
+// Call an accessor method (assuming it is resolved, otherwise drop into
+// vanilla (slow path) entry.
+address InterpreterGenerator::generate_jump_to_normal_entry(void) {
+  address entry = __ pc();
+  address normal_entry = Interpreter::entry_for_kind(Interpreter::zerolocals);
+  assert(normal_entry != NULL, "should already be generated.");
+  __ branch_to_entry(normal_entry, R11_scratch1);
+  __ flush();
+
+  return entry;
+}
+
 // Abstract method entry.
 //
 address InterpreterGenerator::generate_abstract_entry(void) {
@@ -485,203 +498,6 @@
   return entry;
 }
 
-// Call an accessor method (assuming it is resolved, otherwise drop into
-// vanilla (slow path) entry.
-address InterpreterGenerator::generate_accessor_entry(void) {
-  if (!UseFastAccessorMethods && (!FLAG_IS_ERGO(UseFastAccessorMethods))) {
-    return NULL;
-  }
-
-  Label Lslow_path, Lacquire;
-
-  const Register
-         Rclass_or_obj = R3_ARG1,
-         Rconst_method = R4_ARG2,
-         Rcodes        = Rconst_method,
-         Rcpool_cache  = R5_ARG3,
-         Rscratch      = R11_scratch1,
-         Rjvmti_mode   = Rscratch,
-         Roffset       = R12_scratch2,
-         Rflags        = R6_ARG4,
-         Rbtable       = R7_ARG5;
-
-  static address branch_table[number_of_states];
-
-  address entry = __ pc();
-
-  // Check for safepoint:
-  // Ditch this, real man don't need safepoint checks.
-
-  // Also check for JVMTI mode
-  // Check for null obj, take slow path if so.
-  __ ld(Rclass_or_obj, Interpreter::stackElementSize, CC_INTERP_ONLY(R17_tos) NOT_CC_INTERP(R15_esp));
-  __ lwz(Rjvmti_mode, thread_(interp_only_mode));
-  __ cmpdi(CCR1, Rclass_or_obj, 0);
-  __ cmpwi(CCR0, Rjvmti_mode, 0);
-  __ crorc(/*CCR0 eq*/2, /*CCR1 eq*/4+2, /*CCR0 eq*/2);
-  __ beq(CCR0, Lslow_path); // this==null or jvmti_mode!=0
-
-  // Do 2 things in parallel:
-  // 1. Load the index out of the first instruction word, which looks like this:
-  //    <0x2a><0xb4><index (2 byte, native endianess)>.
-  // 2. Load constant pool cache base.
-  __ ld(Rconst_method, in_bytes(Method::const_offset()), R19_method);
-  __ ld(Rcpool_cache, in_bytes(ConstMethod::constants_offset()), Rconst_method);
-
-  __ lhz(Rcodes, in_bytes(ConstMethod::codes_offset()) + 2, Rconst_method); // Lower half of 32 bit field.
-  __ ld(Rcpool_cache, ConstantPool::cache_offset_in_bytes(), Rcpool_cache);
-
-  // Get the const pool entry by means of <index>.
-  const int codes_shift = exact_log2(in_words(ConstantPoolCacheEntry::size()) * BytesPerWord);
-  __ slwi(Rscratch, Rcodes, codes_shift); // (codes&0xFFFF)<<codes_shift
-  __ add(Rcpool_cache, Rscratch, Rcpool_cache);
-
-  // Check if cpool cache entry is resolved.
-  // We are resolved if the indices offset contains the current bytecode.
-  ByteSize cp_base_offset = ConstantPoolCache::base_offset();
-  // Big Endian:
-  __ lbz(Rscratch, in_bytes(cp_base_offset) + in_bytes(ConstantPoolCacheEntry::indices_offset()) + 7 - 2, Rcpool_cache);
-  __ cmpwi(CCR0, Rscratch, Bytecodes::_getfield);
-  __ bne(CCR0, Lslow_path);
-  __ isync(); // Order succeeding loads wrt. load of _indices field from cpool_cache.
-
-  // Finally, start loading the value: Get cp cache entry into regs.
-  __ ld(Rflags, in_bytes(cp_base_offset) + in_bytes(ConstantPoolCacheEntry::flags_offset()), Rcpool_cache);
-  __ ld(Roffset, in_bytes(cp_base_offset) + in_bytes(ConstantPoolCacheEntry::f2_offset()), Rcpool_cache);
-
-  // Following code is from templateTable::getfield_or_static
-  // Load pointer to branch table
-  __ load_const_optimized(Rbtable, (address)branch_table, Rscratch);
-
-  // Get volatile flag
-  __ rldicl(Rscratch, Rflags, 64-ConstantPoolCacheEntry::is_volatile_shift, 63); // extract volatile bit
-  // note: sync is needed before volatile load on PPC64
-
-  // Check field type
-  __ rldicl(Rflags, Rflags, 64-ConstantPoolCacheEntry::tos_state_shift, 64-ConstantPoolCacheEntry::tos_state_bits);
-
-#ifdef ASSERT
-  Label LFlagInvalid;
-  __ cmpldi(CCR0, Rflags, number_of_states);
-  __ bge(CCR0, LFlagInvalid);
-
-  __ ld(R9_ARG7, 0, R1_SP);
-  __ ld(R10_ARG8, 0, R21_sender_SP);
-  __ cmpd(CCR0, R9_ARG7, R10_ARG8);
-  __ asm_assert_eq("backlink", 0x543);
-#endif // ASSERT
-  __ mr(R1_SP, R21_sender_SP); // Cut the stack back to where the caller started.
-
-  // Load from branch table and dispatch (volatile case: one instruction ahead)
-  __ sldi(Rflags, Rflags, LogBytesPerWord);
-  __ cmpwi(CCR6, Rscratch, 1); // volatile?
-  if (support_IRIW_for_not_multiple_copy_atomic_cpu) {
-    __ sldi(Rscratch, Rscratch, exact_log2(BytesPerInstWord)); // volatile ? size of 1 instruction : 0
-  }
-  __ ldx(Rbtable, Rbtable, Rflags);
-
-  if (support_IRIW_for_not_multiple_copy_atomic_cpu) {
-    __ subf(Rbtable, Rscratch, Rbtable); // point to volatile/non-volatile entry point
-  }
-  __ mtctr(Rbtable);
-  __ bctr();
-
-#ifdef ASSERT
-  __ bind(LFlagInvalid);
-  __ stop("got invalid flag", 0x6541);
-
-  bool all_uninitialized = true,
-       all_initialized   = true;
-  for (int i = 0; i<number_of_states; ++i) {
-    all_uninitialized = all_uninitialized && (branch_table[i] == NULL);
-    all_initialized   = all_initialized   && (branch_table[i] != NULL);
-  }
-  assert(all_uninitialized != all_initialized, "consistency"); // either or
-
-  __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-  if (branch_table[vtos] == 0) branch_table[vtos] = __ pc(); // non-volatile_entry point
-  if (branch_table[dtos] == 0) branch_table[dtos] = __ pc(); // non-volatile_entry point
-  if (branch_table[ftos] == 0) branch_table[ftos] = __ pc(); // non-volatile_entry point
-  __ stop("unexpected type", 0x6551);
-#endif
-
-  if (branch_table[itos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[itos] = __ pc(); // non-volatile_entry point
-    __ lwax(R3_RET, Rclass_or_obj, Roffset);
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  if (branch_table[ltos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[ltos] = __ pc(); // non-volatile_entry point
-    __ ldx(R3_RET, Rclass_or_obj, Roffset);
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  if (branch_table[btos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[btos] = __ pc(); // non-volatile_entry point
-    __ lbzx(R3_RET, Rclass_or_obj, Roffset);
-    __ extsb(R3_RET, R3_RET);
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  if (branch_table[ctos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[ctos] = __ pc(); // non-volatile_entry point
-    __ lhzx(R3_RET, Rclass_or_obj, Roffset);
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  if (branch_table[stos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[stos] = __ pc(); // non-volatile_entry point
-    __ lhax(R3_RET, Rclass_or_obj, Roffset);
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  if (branch_table[atos] == 0) { // generate only once
-    __ align(32, 28, 28); // align load
-    __ fence(); // volatile entry point (one instruction before non-volatile_entry point)
-    branch_table[atos] = __ pc(); // non-volatile_entry point
-    __ load_heap_oop(R3_RET, (RegisterOrConstant)Roffset, Rclass_or_obj);
-    __ verify_oop(R3_RET);
-    //__ dcbt(R3_RET); // prefetch
-    __ beq(CCR6, Lacquire);
-    __ blr();
-  }
-
-  __ align(32, 12);
-  __ bind(Lacquire);
-  __ twi_0(R3_RET);
-  __ isync(); // acquire
-  __ blr();
-
-#ifdef ASSERT
-  for (int i = 0; i<number_of_states; ++i) {
-    assert(branch_table[i], "accessor_entry initialization");
-    //tty->print_cr("accessor_entry: branch_table[%d] = 0x%llx (opcode 0x%llx)", i, branch_table[i], *((unsigned int*)branch_table[i]));
-  }
-#endif
-
-  __ bind(Lslow_path);
-  __ branch_to_entry(Interpreter::entry_for_kind(Interpreter::zerolocals), Rscratch);
-  __ flush();
-
-  return entry;
-}
-
 // Interpreter intrinsic for WeakReference.get().
 // 1. Don't push a full blown frame and go on dispatching, but fetch the value
 //    into R8 and return quickly
@@ -713,7 +529,6 @@
   //   and so we don't need to call the G1 pre-barrier. Thus we can use the
   //   regular method entry code to generate the NPE.
   //
-  // This code is based on generate_accessor_enty.
 
   address entry = __ pc();
 
@@ -768,7 +583,7 @@
 
     return entry;
   } else {
-    return generate_accessor_entry();
+    return generate_jump_to_normal_entry();
   }
 }
 

--- a/src/cpu/ppc/vm/ppc.ad	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/ppc.ad	Thu Aug 28 14:53:43 2014 -0700
@@ -1283,8 +1283,6 @@
 
 bool MachConstantBaseNode::requires_postalloc_expand() const { return true; }
 void MachConstantBaseNode::postalloc_expand(GrowableArray <Node *> *nodes, PhaseRegAlloc *ra_) {
-  Compile *C = ra_->C;
-
   iRegPdstOper *op_dst = new iRegPdstOper();
   MachNode *m1 = new loadToc_hiNode();
   MachNode *m2 = new loadToc_loNode();
@@ -2229,7 +2227,7 @@
 }
 /* TODO: PPC port
 // Make a new machine dependent decode node (with its operands).
-MachTypeNode *Matcher::make_decode_node(Compile *C) {
+MachTypeNode *Matcher::make_decode_node() {
   assert(Universe::narrow_oop_base() == NULL && Universe::narrow_oop_shift() == 0,
          "This method is only implemented for unscaled cOops mode so far");
   MachTypeNode *decode = new decodeN_unscaledNode();
@@ -2593,7 +2591,7 @@
   MachNode        *_last;
 } loadConLNodesTuple;
 
-loadConLNodesTuple loadConLNodesTuple_create(Compile *C, PhaseRegAlloc *ra_, Node *toc, immLOper *immSrc,
+loadConLNodesTuple loadConLNodesTuple_create(PhaseRegAlloc *ra_, Node *toc, immLOper *immSrc,
                                              OptoReg::Name reg_second, OptoReg::Name reg_first) {
   loadConLNodesTuple nodes;
 
@@ -2669,7 +2667,7 @@
   enc_class postalloc_expand_load_long_constant(iRegLdst dst, immL src, iRegLdst toc) %{
     // Create new nodes.
     loadConLNodesTuple loadConLNodes =
-      loadConLNodesTuple_create(C, ra_, n_toc, op_src,
+      loadConLNodesTuple_create(ra_, n_toc, op_src,
                                 ra_->get_reg_second(this), ra_->get_reg_first(this));
 
     // Push new nodes.
@@ -3391,7 +3389,7 @@
     immLOper *op_repl = new immLOper((jlong)replicate_immF(op_src->constantF()));
 
     loadConLNodesTuple loadConLNodes =
-      loadConLNodesTuple_create(C, ra_, n_toc, op_repl,
+      loadConLNodesTuple_create(ra_, n_toc, op_repl,
                                 ra_->get_reg_second(this), ra_->get_reg_first(this));
 
     // Push new nodes.
@@ -3611,7 +3609,7 @@
 
     // Create the nodes for loading the IC from the TOC.
     loadConLNodesTuple loadConLNodes_IC =
-      loadConLNodesTuple_create(C, ra_, n_toc, new immLOper((jlong)Universe::non_oop_word()),
+      loadConLNodesTuple_create(ra_, n_toc, new immLOper((jlong)Universe::non_oop_word()),
                                 OptoReg::Name(R19_H_num), OptoReg::Name(R19_num));
 
     // Create the call node.
@@ -3765,7 +3763,7 @@
 #if defined(ABI_ELFv2)
     jlong entry_address = (jlong) this->entry_point();
     assert(entry_address, "need address here");
-    loadConLNodes_Entry = loadConLNodesTuple_create(C, ra_, n_toc, new immLOper(entry_address),
+    loadConLNodes_Entry = loadConLNodesTuple_create(ra_, n_toc, new immLOper(entry_address),
                                                     OptoReg::Name(R12_H_num), OptoReg::Name(R12_num));
 #else
     // Get the struct that describes the function we are about to call.
@@ -3777,13 +3775,13 @@
     loadConLNodesTuple loadConLNodes_Toc;
 
     // Create nodes and operands for loading the entry point.
-    loadConLNodes_Entry = loadConLNodesTuple_create(C, ra_, n_toc, new immLOper(entry_address),
+    loadConLNodes_Entry = loadConLNodesTuple_create(ra_, n_toc, new immLOper(entry_address),
                                                     OptoReg::Name(R12_H_num), OptoReg::Name(R12_num));
 
 
     // Create nodes and operands for loading the env pointer.
     if (fd->env() != NULL) {
-      loadConLNodes_Env = loadConLNodesTuple_create(C, ra_, n_toc, new immLOper((jlong) fd->env()),
+      loadConLNodes_Env = loadConLNodesTuple_create(ra_, n_toc, new immLOper((jlong) fd->env()),
                                                     OptoReg::Name(R11_H_num), OptoReg::Name(R11_num));
     } else {
       loadConLNodes_Env._large_hi = NULL;
@@ -3796,7 +3794,7 @@
     }
 
     // Create nodes and operands for loading the Toc point.
-    loadConLNodes_Toc = loadConLNodesTuple_create(C, ra_, n_toc, new immLOper((jlong) fd->toc()),
+    loadConLNodes_Toc = loadConLNodesTuple_create(ra_, n_toc, new immLOper((jlong) fd->toc()),
                                                   OptoReg::Name(R2_H_num), OptoReg::Name(R2_num));
 #endif // ABI_ELFv2
     // mtctr node

--- a/src/cpu/ppc/vm/templateInterpreterGenerator_ppc.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/templateInterpreterGenerator_ppc.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -30,7 +30,6 @@
   address generate_normal_entry(bool synchronized);
   address generate_native_entry(bool synchronized);
   address generate_math_entry(AbstractInterpreter::MethodKind kind);
-  address generate_empty_entry(void);
 
   void lock_method(Register Rflags, Register Rscratch1, Register Rscratch2, bool flags_preloaded=false);
   void unlock_method(bool check_exceptions = true);

--- a/src/cpu/ppc/vm/templateInterpreter_ppc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/templateInterpreter_ppc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -176,8 +176,12 @@
   const Register size  = R12_scratch2;
   __ get_cache_and_index_at_bcp(cache, 1, index_size);
 
-  // Big Endian (get least significant byte of 64 bit value):
+  // Get least significant byte of 64 bit value:
+#if defined(VM_LITTLE_ENDIAN)
+  __ lbz(size, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::flags_offset()), cache);
+#else
   __ lbz(size, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::flags_offset()) + 7, cache);
+#endif
   __ sldi(size, size, Interpreter::logStackElementSize);
   __ add(R15_esp, R15_esp, size);
   __ dispatch_next(state, step);
@@ -598,48 +602,6 @@
 
 // End of helpers
 
-// ============================================================================
-// Various method entries
-//
-
-// Empty method, generate a very fast return. We must skip this entry if
-// someone's debugging, indicated by the flag
-// "interp_mode" in the Thread obj.
-// Note: empty methods are generated mostly methods that do assertions, which are
-// disabled in the "java opt build".
-address TemplateInterpreterGenerator::generate_empty_entry(void) {
-  if (!UseFastEmptyMethods) {
-    NOT_PRODUCT(__ should_not_reach_here();)
-    return Interpreter::entry_for_kind(Interpreter::zerolocals);
-  }
-
-  Label Lslow_path;
-  const Register Rjvmti_mode = R11_scratch1;
-  address entry = __ pc();
-
-  __ lwz(Rjvmti_mode, thread_(interp_only_mode));
-  __ cmpwi(CCR0, Rjvmti_mode, 0);
-  __ bne(CCR0, Lslow_path); // jvmti_mode!=0
-
-  // Noone's debuggin: Simply return.
-  // Pop c2i arguments (if any) off when we return.
-#ifdef ASSERT
-    __ ld(R9_ARG7, 0, R1_SP);
-    __ ld(R10_ARG8, 0, R21_sender_SP);
-    __ cmpd(CCR0, R9_ARG7, R10_ARG8);
-    __ asm_assert_eq("backlink", 0x545);
-#endif // ASSERT
-  __ mr(R1_SP, R21_sender_SP); // Cut the stack back to where the caller started.
-
-  // And we're done.
-  __ blr();
-
-  __ bind(Lslow_path);
-  __ branch_to_entry(Interpreter::entry_for_kind(Interpreter::zerolocals), R11_scratch1);
-  __ flush();
-
-  return entry;
-}
 
 // Support abs and sqrt like in compiler.
 // For others we can use a normal (native) entry.
@@ -858,7 +820,9 @@
   // Our signature handlers copy required arguments to the C stack
   // (outgoing C args), R3_ARG1 to R10_ARG8, and FARG1 to FARG13.
   __ mr(R3_ARG1, R18_locals);
+#if !defined(ABI_ELFv2)
   __ ld(signature_handler_fd, 0, signature_handler_fd);
+#endif
 
   __ call_stub(signature_handler_fd);
 
@@ -1020,8 +984,13 @@
   // native result across the call. No oop is present.
 
   __ mr(R3_ARG1, R16_thread);
+#if defined(ABI_ELFv2)
+  __ call_c(CAST_FROM_FN_PTR(address, JavaThread::check_special_condition_for_native_trans),
+            relocInfo::none);
+#else
   __ call_c(CAST_FROM_FN_PTR(FunctionDescriptor*, JavaThread::check_special_condition_for_native_trans),
             relocInfo::none);
+#endif
 
   __ bind(sync_check_done);
 
@@ -1278,45 +1247,6 @@
   return entry;
 }
 
-// =============================================================================
-// Entry points
-
-address AbstractInterpreterGenerator::generate_method_entry(
-                                        AbstractInterpreter::MethodKind kind) {
-  // Determine code generation flags.
-  bool synchronized = false;
-  address entry_point = NULL;
-
-  switch (kind) {
-  case Interpreter::zerolocals             :                                                                             break;
-  case Interpreter::zerolocals_synchronized: synchronized = true;                                                        break;
-  case Interpreter::native                 : entry_point = ((InterpreterGenerator*) this)->generate_native_entry(false); break;
-  case Interpreter::native_synchronized    : entry_point = ((InterpreterGenerator*) this)->generate_native_entry(true);  break;
-  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::java_lang_math_sin     : // fall thru
-  case Interpreter::java_lang_math_cos     : // fall thru
-  case Interpreter::java_lang_math_tan     : // fall thru
-  case Interpreter::java_lang_math_abs     : // fall thru
-  case Interpreter::java_lang_math_log     : // fall thru
-  case Interpreter::java_lang_math_log10   : // fall thru
-  case Interpreter::java_lang_math_sqrt    : // fall thru
-  case Interpreter::java_lang_math_pow     : // fall thru
-  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;
-  }
-
-  if (entry_point) {
-    return entry_point;
-  }
-
-  return ((InterpreterGenerator*) this)->generate_normal_entry(synchronized);
-}
-
 // These should never be compiled since the interpreter will prefer
 // the compiled version to the intrinsic version.
 bool AbstractInterpreter::can_be_compiled(methodHandle m) {
@@ -1344,7 +1274,7 @@
                                          int callee_locals,
                                          bool is_top_frame) {
   // Note: This calculation must exactly parallel the frame setup
-  // in AbstractInterpreterGenerator::generate_method_entry.
+  // in InterpreterGenerator::generate_fixed_frame.
   assert(Interpreter::stackElementWords == 1, "sanity");
   const int max_alignment_space = StackAlignmentInBytes / Interpreter::stackElementSize;
   const int abi_scratch = is_top_frame ? (frame::abi_reg_args_size / Interpreter::stackElementSize) :

--- a/src/cpu/ppc/vm/templateTable_ppc_64.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/templateTable_ppc_64.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -189,8 +189,12 @@
       assert(byte_no == f1_byte || byte_no == f2_byte, "byte_no out of range");
       assert(load_bc_into_bc_reg, "we use bc_reg as temp");
       __ get_cache_and_index_at_bcp(Rtemp /* dst = cache */, 1);
-      // Big Endian: ((*(cache+indices))>>((1+byte_no)*8))&0xFF
+      // ((*(cache+indices))>>((1+byte_no)*8))&0xFF:
+#if defined(VM_LITTLE_ENDIAN)
+      __ lbz(Rnew_bc, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()) + 1 + byte_no, Rtemp);
+#else
       __ lbz(Rnew_bc, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()) + 7 - (1 + byte_no), Rtemp);
+#endif
       __ cmpwi(CCR0, Rnew_bc, 0);
       __ li(Rnew_bc, (unsigned int)(unsigned char)new_bc);
       __ beq(CCR0, L_patch_done);
@@ -1839,8 +1843,8 @@
   __ clrrdi(Rdef_offset_addr, Rdef_offset_addr, log2_long((jlong)BytesPerInt));
 
   // Load lo & hi.
-  __ lwz(Rlow_byte, BytesPerInt, Rdef_offset_addr);
-  __ lwz(Rhigh_byte, BytesPerInt * 2, Rdef_offset_addr);
+  __ get_u4(Rlow_byte, Rdef_offset_addr, BytesPerInt, InterpreterMacroAssembler::Unsigned);
+  __ get_u4(Rhigh_byte, Rdef_offset_addr, 2 *BytesPerInt, InterpreterMacroAssembler::Unsigned);
 
   // Check for default case (=index outside [low,high]).
   __ cmpw(CCR0, R17_tos, Rlow_byte);
@@ -1854,12 +1858,17 @@
   __ profile_switch_case(Rindex, Rhigh_byte /* scratch */, Rscratch1, Rscratch2);
   __ sldi(Rindex, Rindex, LogBytesPerInt);
   __ addi(Rindex, Rindex, 3 * BytesPerInt);
+#if defined(VM_LITTLE_ENDIAN)
+  __ lwbrx(Roffset, Rdef_offset_addr, Rindex);
+  __ extsw(Roffset, Roffset);
+#else
   __ lwax(Roffset, Rdef_offset_addr, Rindex);
+#endif
   __ b(Ldispatch);
 
   __ bind(Ldefault_case);
   __ profile_switch_default(Rhigh_byte, Rscratch1);
-  __ lwa(Roffset, 0, Rdef_offset_addr);
+  __ get_u4(Roffset, Rdef_offset_addr, 0, InterpreterMacroAssembler::Signed);
 
   __ bind(Ldispatch);
 
@@ -1875,12 +1884,11 @@
 // Table switch using linear search through cases.
 // Bytecode stream format:
 // Bytecode (1) | 4-byte padding | default offset (4) | count (4) | value/offset pair1 (8) | value/offset pair2 (8) | ...
-// Note: Everything is big-endian format here. So on little endian machines, we have to revers offset and count and cmp value.
+// Note: Everything is big-endian format here.
 void TemplateTable::fast_linearswitch() {
   transition(itos, vtos);
 
-  Label Lloop_entry, Lsearch_loop, Lfound, Lcontinue_execution, Ldefault_case;
-
+  Label Lloop_entry, Lsearch_loop, Lcontinue_execution, Ldefault_case;
   Register Rcount           = R3_ARG1,
            Rcurrent_pair    = R4_ARG2,
            Rdef_offset_addr = R5_ARG3, // Is going to contain address of default offset.
@@ -1894,47 +1902,40 @@
   __ clrrdi(Rdef_offset_addr, Rdef_offset_addr, log2_long((jlong)BytesPerInt));
 
   // Setup loop counter and limit.
-  __ lwz(Rcount, BytesPerInt, Rdef_offset_addr);    // Load count.
+  __ get_u4(Rcount, Rdef_offset_addr, BytesPerInt, InterpreterMacroAssembler::Unsigned);
   __ addi(Rcurrent_pair, Rdef_offset_addr, 2 * BytesPerInt); // Rcurrent_pair now points to first pair.
 
-  // Set up search loop.
-  __ cmpwi(CCR0, Rcount, 0);
-  __ beq(CCR0, Ldefault_case);
-
   __ mtctr(Rcount);
-
-  // linear table search
-  __ bind(Lsearch_loop);
-
-  __ lwz(Rvalue, 0, Rcurrent_pair);
-  __ lwa(Roffset, 1 * BytesPerInt, Rcurrent_pair);
-
-  __ cmpw(CCR0, Rvalue, Rcmp_value);
-  __ beq(CCR0, Lfound);
-
-  __ addi(Rcurrent_pair, Rcurrent_pair, 2 * BytesPerInt);
-  __ bdnz(Lsearch_loop);
-
-  // default case
+  __ cmpwi(CCR0, Rcount, 0);
+  __ bne(CCR0, Lloop_entry);
+
+  // Default case
   __ bind(Ldefault_case);
-
-  __ lwa(Roffset, 0, Rdef_offset_addr);
+  __ get_u4(Roffset, Rdef_offset_addr, 0, InterpreterMacroAssembler::Signed);
   if (ProfileInterpreter) {
     __ profile_switch_default(Rdef_offset_addr, Rcount/* scratch */);
-    __ b(Lcontinue_execution);
   }
-
-  // Entry found, skip Roffset bytecodes and continue.
-  __ bind(Lfound);
+  __ b(Lcontinue_execution);
+
+  // Next iteration
+  __ bind(Lsearch_loop);
+  __ bdz(Ldefault_case);
+  __ addi(Rcurrent_pair, Rcurrent_pair, 2 * BytesPerInt);
+  __ bind(Lloop_entry);
+  __ get_u4(Rvalue, Rcurrent_pair, 0, InterpreterMacroAssembler::Unsigned);
+  __ cmpw(CCR0, Rvalue, Rcmp_value);
+  __ bne(CCR0, Lsearch_loop);
+
+  // Found, load offset.
+  __ get_u4(Roffset, Rcurrent_pair, BytesPerInt, InterpreterMacroAssembler::Signed);
+  // Calculate case index and profile
+  __ mfctr(Rcurrent_pair);
   if (ProfileInterpreter) {
-    // Calc the num of the pair we hit. Careful, Rcurrent_pair points 2 ints
-    // beyond the actual current pair due to the auto update load above!
-    __ sub(Rcurrent_pair, Rcurrent_pair, Rdef_offset_addr);
-    __ addi(Rcurrent_pair, Rcurrent_pair, - 2 * BytesPerInt);
-    __ srdi(Rcurrent_pair, Rcurrent_pair, LogBytesPerInt + 1);
+    __ sub(Rcurrent_pair, Rcount, Rcurrent_pair);
     __ profile_switch_case(Rcurrent_pair, Rcount /*scratch*/, Rdef_offset_addr/*scratch*/, Rscratch);
-    __ bind(Lcontinue_execution);
   }
+
+  __ bind(Lcontinue_execution);
   __ add(R14_bcp, Roffset, R14_bcp);
   __ dispatch_next(vtos);
 }
@@ -1990,7 +1991,7 @@
 
   // initialize i & j
   __ li(Ri,0);
-  __ lwz(Rj, -BytesPerInt, Rarray);
+  __ get_u4(Rj, Rarray, -BytesPerInt, InterpreterMacroAssembler::Unsigned);
 
   // and start.
   Label entry;
@@ -2007,7 +2008,11 @@
     //   i = h;
     // }
     __ sldi(Rscratch, Rh, log_entry_size);
+#if defined(VM_LITTLE_ENDIAN)
+    __ lwbrx(Rscratch, Rscratch, Rarray);
+#else
     __ lwzx(Rscratch, Rscratch, Rarray);
+#endif
 
     // if (key < current value)
     //   Rh = Rj
@@ -2039,20 +2044,20 @@
   // Ri = value offset
   __ sldi(Ri, Ri, log_entry_size);
   __ add(Ri, Ri, Rarray);
-  __ lwz(Rscratch, 0, Ri);
+  __ get_u4(Rscratch, Ri, 0, InterpreterMacroAssembler::Unsigned);
 
   Label not_found;
   // Ri = offset offset
   __ cmpw(CCR0, Rkey, Rscratch);
   __ beq(CCR0, not_found);
   // entry not found -> j = default offset
-  __ lwz(Rj, -2 * BytesPerInt, Rarray);
+  __ get_u4(Rj, Rarray, -2 * BytesPerInt, InterpreterMacroAssembler::Unsigned);
   __ b(default_case);
 
   __ bind(not_found);
   // entry found -> j = offset
   __ profile_switch_case(Rh, Rj, Rscratch, Rkey);
-  __ lwz(Rj, BytesPerInt, Ri);
+  __ get_u4(Rj, Ri, BytesPerInt, InterpreterMacroAssembler::Unsigned);
 
   if (ProfileInterpreter) {
     __ b(continue_execution);
@@ -2147,8 +2152,11 @@
 
   assert(byte_no == f1_byte || byte_no == f2_byte, "byte_no out of range");
   // We are resolved if the indices offset contains the current bytecode.
-  // Big Endian:
+#if defined(VM_LITTLE_ENDIAN)
+  __ lbz(Rscratch, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()) + byte_no + 1, Rcache);
+#else
   __ lbz(Rscratch, in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()) + 7 - (byte_no + 1), Rcache);
+#endif
   // Acquire by cmp-br-isync (see below).
   __ cmpdi(CCR0, Rscratch, (int)bytecode());
   __ beq(CCR0, Lresolved);

--- a/src/cpu/ppc/vm/vm_version_ppc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/ppc/vm/vm_version_ppc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -29,6 +29,7 @@
 #include "compiler/disassembler.hpp"
 #include "memory/resourceArea.hpp"
 #include "runtime/java.hpp"
+#include "runtime/os.hpp"
 #include "runtime/stubCodeGenerator.hpp"
 #include "utilities/defaultStream.hpp"
 #include "vm_version_ppc.hpp"
@@ -108,7 +109,7 @@
                (has_vand()    ? " vand"    : "")
                // Make sure number of %s matches num_features!
               );
-  _features_str = strdup(buf);
+  _features_str = os::strdup(buf);
   NOT_PRODUCT(if (Verbose) print_features(););
 
   // PPC64 supports 8-byte compare-exchange operations (see

--- a/src/cpu/sparc/vm/cppInterpreter_sparc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/cppInterpreter_sparc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2007, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2007, 2014, 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
@@ -29,6 +29,7 @@
 #include "interpreter/interpreter.hpp"
 #include "interpreter/interpreterGenerator.hpp"
 #include "interpreter/interpreterRuntime.hpp"
+#include "interpreter/interp_masm.hpp"
 #include "oops/arrayOop.hpp"
 #include "oops/methodData.hpp"
 #include "oops/method.hpp"
@@ -68,9 +69,7 @@
 #define STATE(field_name) Lstate, in_bytes(byte_offset_of(BytecodeInterpreter, field_name))
 #define __ _masm->
 
-Label frame_manager_entry;
-Label fast_accessor_slow_entry_path;  // fast accessor methods need to be able to jmp to unsynchronized
-                                      // c++ interpreter entry point this holds that entry point label.
+Label frame_manager_entry; // c++ interpreter entry point this holds that entry point label.
 
 static address unctrap_frame_manager_entry  = NULL;
 
@@ -452,110 +451,6 @@
   return NULL;
 }
 
-// Call an accessor method (assuming it is resolved, otherwise drop into
-// vanilla (slow path) entry
-
-// Generates code to elide accessor methods
-// Uses G3_scratch and G1_scratch as scratch
-address InterpreterGenerator::generate_accessor_entry(void) {
-
-  // Code: _aload_0, _(i|a)getfield, _(i|a)return or any rewrites thereof;
-  // parameter size = 1
-  // Note: We can only use this code if the getfield has been resolved
-  //       and if we don't have a null-pointer exception => check for
-  //       these conditions first and use slow path if necessary.
-  address entry = __ pc();
-  Label slow_path;
-
-  if ( UseFastAccessorMethods) {
-    // Check if we need to reach a safepoint and generate full interpreter
-    // frame if so.
-    AddressLiteral sync_state(SafepointSynchronize::address_of_state());
-    __ load_contents(sync_state, G3_scratch);
-    __ cmp(G3_scratch, SafepointSynchronize::_not_synchronized);
-    __ br(Assembler::notEqual, false, Assembler::pn, slow_path);
-    __ delayed()->nop();
-
-    // Check if local 0 != NULL
-    __ ld_ptr(Gargs, G0, Otos_i ); // get local 0
-    __ tst(Otos_i);  // check if local 0 == NULL and go the slow path
-    __ brx(Assembler::zero, false, Assembler::pn, slow_path);
-    __ delayed()->nop();
-
-
-    // read first instruction word and extract bytecode @ 1 and index @ 2
-    // get first 4 bytes of the bytecodes (big endian!)
-    __ ld_ptr(Address(G5_method, in_bytes(Method::const_offset())), G1_scratch);
-    __ ld(Address(G1_scratch, in_bytes(ConstMethod::codes_offset())), G1_scratch);
-
-    // move index @ 2 far left then to the right most two bytes.
-    __ sll(G1_scratch, 2*BitsPerByte, G1_scratch);
-    __ srl(G1_scratch, 2*BitsPerByte - exact_log2(in_words(
-                      ConstantPoolCacheEntry::size()) * BytesPerWord), G1_scratch);
-
-    // get constant pool cache
-    __ ld_ptr(G5_method, in_bytes(Method::const_offset()), G3_scratch);
-    __ ld_ptr(G3_scratch, in_bytes(ConstMethod::constants_offset()), G3_scratch);
-    __ ld_ptr(G3_scratch, ConstantPool::cache_offset_in_bytes(), G3_scratch);
-
-    // get specific constant pool cache entry
-    __ add(G3_scratch, G1_scratch, G3_scratch);
-
-    // Check the constant Pool cache entry to see if it has been resolved.
-    // If not, need the slow path.
-    ByteSize cp_base_offset = ConstantPoolCache::base_offset();
-    __ ld_ptr(G3_scratch, in_bytes(cp_base_offset + ConstantPoolCacheEntry::indices_offset()), G1_scratch);
-    __ srl(G1_scratch, 2*BitsPerByte, G1_scratch);
-    __ and3(G1_scratch, 0xFF, G1_scratch);
-    __ cmp(G1_scratch, Bytecodes::_getfield);
-    __ br(Assembler::notEqual, false, Assembler::pn, slow_path);
-    __ delayed()->nop();
-
-    // Get the type and return field offset from the constant pool cache
-    __ ld_ptr(G3_scratch, in_bytes(cp_base_offset + ConstantPoolCacheEntry::flags_offset()), G1_scratch);
-    __ ld_ptr(G3_scratch, in_bytes(cp_base_offset + ConstantPoolCacheEntry::f2_offset()), G3_scratch);
-
-    Label xreturn_path;
-    // 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::tos_state_shift, G1_scratch);
-    // Make sure we don't need to mask G1_scratch after the above shift
-    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);
-    __ cmp(G1_scratch, itos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ld(Otos_i, G3_scratch, Otos_i);
-    __ cmp(G1_scratch, stos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ldsh(Otos_i, G3_scratch, Otos_i);
-    __ cmp(G1_scratch, ctos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->lduh(Otos_i, G3_scratch, Otos_i);
-#ifdef ASSERT
-    __ cmp(G1_scratch, btos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ldsb(Otos_i, G3_scratch, Otos_i);
-    __ should_not_reach_here();
-#endif
-    __ ldsb(Otos_i, G3_scratch, Otos_i);
-    __ bind(xreturn_path);
-
-    // _ireturn/_areturn
-    __ retl();                      // return from leaf routine
-    __ delayed()->mov(O5_savedSP, SP);
-
-    // Generate regular method entry
-    __ bind(slow_path);
-    __ ba(fast_accessor_slow_entry_path);
-    __ delayed()->nop();
-    return entry;
-  }
-  return NULL;
-}
-
 address InterpreterGenerator::generate_Reference_get_entry(void) {
 #if INCLUDE_ALL_GCS
   if (UseG1GC) {
@@ -573,7 +468,7 @@
 
   // If G1 is not enabled then attempt to go through the accessor entry point
   // Reference.get is an accessor
-  return generate_accessor_entry();
+  return generate_jump_to_normal_entry();
 }
 
 //
@@ -1870,23 +1765,6 @@
   __ ba(call_interpreter_2);
   __ delayed()->st_ptr(O1, STATE(_stack));
 
-
-  // Fast accessor methods share this entry point.
-  // This works because frame manager is in the same codelet
-  // This can either be an entry via call_stub/c1/c2 or a recursive interpreter call
-  // we need to do a little register fixup here once we distinguish the two of them
-  if (UseFastAccessorMethods && !synchronized) {
-  // Call stub_return address still in O7
-    __ bind(fast_accessor_slow_entry_path);
-    __ set((intptr_t)return_from_native_method - 8, Gtmp1);
-    __ cmp(Gtmp1, O7);                                                // returning to interpreter?
-    __ brx(Assembler::equal, true, Assembler::pt, re_dispatch);       // yep
-    __ delayed()->nop();
-    __ ba(re_dispatch);
-    __ delayed()->mov(G0, prevState);                                 // initial entry
-
-  }
-
   // interpreter returning to native code (call_stub/c1/c2)
   // convert result and unwind initial activation
   // L2_scratch - scaled result type index

--- a/src/cpu/sparc/vm/interpreterGenerator_sparc.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/interpreterGenerator_sparc.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2014, 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
@@ -32,9 +32,11 @@
   address generate_normal_entry(bool synchronized);
   address generate_native_entry(bool synchronized);
   address generate_abstract_entry(void);
-  address generate_math_entry(AbstractInterpreter::MethodKind kind);
-  address generate_empty_entry(void);
-  address generate_accessor_entry(void);
+  // there are no math intrinsics on sparc
+  address generate_math_entry(AbstractInterpreter::MethodKind kind) { return NULL; }
+  address generate_jump_to_normal_entry(void);
+  address generate_accessor_entry(void) { return generate_jump_to_normal_entry(); }
+  address generate_empty_entry(void) { return generate_jump_to_normal_entry(); }
   address generate_Reference_get_entry(void);
   void lock_method(void);
   void save_native_result(void);
@@ -43,4 +45,7 @@
   void generate_counter_incr(Label* overflow, Label* profile_method, Label* profile_method_continue);
   void generate_counter_overflow(Label& Lcontinue);
 
+  // Not supported
+  address generate_CRC32_update_entry() { return NULL; }
+  address generate_CRC32_updateBytes_entry(AbstractInterpreter::MethodKind kind) { return NULL; }
 #endif // CPU_SPARC_VM_INTERPRETERGENERATOR_SPARC_HPP

--- a/src/cpu/sparc/vm/interpreter_sparc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/interpreter_sparc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -241,6 +241,15 @@
 
 // Various method entries
 
+address InterpreterGenerator::generate_jump_to_normal_entry(void) {
+  address entry = __ pc();
+  assert(Interpreter::entry_for_kind(Interpreter::zerolocals) != NULL, "should already be generated");
+  AddressLiteral al(Interpreter::entry_for_kind(Interpreter::zerolocals));
+  __ jump_to(al, G3_scratch);
+  __ delayed()->nop();
+  return entry;
+}
+
 // Abstract method entry
 // Attempt to execute abstract method. Throw exception
 //
@@ -255,159 +264,6 @@
 
 }
 
-
-//----------------------------------------------------------------------------------------------------
-// Entry points & stack frame layout
-//
-// Here we generate the various kind of entries into the interpreter.
-// The two main entry type are generic bytecode methods and native call method.
-// These both come in synchronized and non-synchronized versions but the
-// frame layout they create is very similar. The other method entry
-// types are really just special purpose entries that are really entry
-// and interpretation all in one. These are for trivial methods like
-// accessor, empty, or special math methods.
-//
-// When control flow reaches any of the entry types for the interpreter
-// the following holds ->
-//
-// C2 Calling Conventions:
-//
-// The entry code below assumes that the following registers are set
-// when coming in:
-//    G5_method: holds the Method* of the method to call
-//    Lesp:    points to the TOS of the callers expression stack
-//             after having pushed all the parameters
-//
-// The entry code does the following to setup an interpreter frame
-//   pop parameters from the callers stack by adjusting Lesp
-//   set O0 to Lesp
-//   compute X = (max_locals - num_parameters)
-//   bump SP up by X to accomadate the extra locals
-//   compute X = max_expression_stack
-//               + vm_local_words
-//               + 16 words of register save area
-//   save frame doing a save sp, -X, sp growing towards lower addresses
-//   set Lbcp, Lmethod, LcpoolCache
-//   set Llocals to i0
-//   set Lmonitors to FP - rounded_vm_local_words
-//   set Lesp to Lmonitors - 4
-//
-//  The frame has now been setup to do the rest of the entry code
-
-// Try this optimization:  Most method entries could live in a
-// "one size fits all" stack frame without all the dynamic size
-// calculations.  It might be profitable to do all this calculation
-// statically and approximately for "small enough" methods.
-
-//-----------------------------------------------------------------------------------------------
-
-// C1 Calling conventions
-//
-// Upon method entry, the following registers are setup:
-//
-// g2 G2_thread: current thread
-// g5 G5_method: method to activate
-// g4 Gargs  : pointer to last argument
-//
-//
-// Stack:
-//
-// +---------------+ <--- sp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- sp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- sp + 0x5c
-// |               |
-// :     free      :
-// |               |
-// +---------------+ <--- Gargs
-// |               |
-// :   arguments   :
-// |               |
-// +---------------+
-// |               |
-//
-//
-//
-// AFTER FRAME HAS BEEN SETUP for method interpretation the stack looks like:
-//
-// +---------------+ <--- sp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- sp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- sp + 0x5c
-// |               |
-// :               :
-// |               | <--- Lesp
-// +---------------+ <--- Lmonitors (fp - 0x18)
-// |   VM locals   |
-// +---------------+ <--- fp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- fp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- fp + 0x5c
-// |               |
-// :     free      :
-// |               |
-// +---------------+
-// |               |
-// : nonarg locals :
-// |               |
-// +---------------+
-// |               |
-// :   arguments   :
-// |               | <--- Llocals
-// +---------------+ <--- Gargs
-// |               |
-
-address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter::MethodKind kind) {
-  // determine code generation flags
-  bool synchronized = false;
-  address entry_point = NULL;
-
-  switch (kind) {
-    case Interpreter::zerolocals             :                                                                             break;
-    case Interpreter::zerolocals_synchronized: synchronized = true;                                                        break;
-    case Interpreter::native                 : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(false);  break;
-    case Interpreter::native_synchronized    : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(true);   break;
-    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::java_lang_math_sin     :                                                                             break;
-    case Interpreter::java_lang_math_cos     :                                                                             break;
-    case Interpreter::java_lang_math_tan     :                                                                             break;
-    case Interpreter::java_lang_math_sqrt    :                                                                             break;
-    case Interpreter::java_lang_math_abs     :                                                                             break;
-    case Interpreter::java_lang_math_log     :                                                                             break;
-    case Interpreter::java_lang_math_log10   :                                                                             break;
-    case Interpreter::java_lang_math_pow     :                                                                             break;
-    case Interpreter::java_lang_math_exp     :                                                                             break;
-    case Interpreter::java_lang_ref_reference_get
-                                             : entry_point = ((InterpreterGenerator*)this)->generate_Reference_get_entry(); break;
-    default:
-      fatal(err_msg("unexpected method kind: %d", kind));
-      break;
-  }
-
-  if (entry_point) return entry_point;
-
-  return ((InterpreterGenerator*)this)->generate_normal_entry(synchronized);
-}
-
-
 bool AbstractInterpreter::can_be_compiled(methodHandle m) {
   // No special entry points that preclude compilation
   return true;

--- a/src/cpu/sparc/vm/sparc.ad	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/sparc.ad	Thu Aug 28 14:53:43 2014 -0700
@@ -6184,7 +6184,11 @@
   ins_cost(DEFAULT_COST * 3/2);
   format %{ "SET    $con,$dst\t! non-oop ptr" %}
   ins_encode %{
-    __ set($con$$constant, $dst$$Register);
+    if (_opnds[1]->constant_reloc() == relocInfo::metadata_type) {
+      __ set_metadata_constant((Metadata*)$con$$constant, $dst$$Register);
+    } else {
+      __ set($con$$constant, $dst$$Register);
+    }
   %}
   ins_pipe(loadConP);
 %}

--- a/src/cpu/sparc/vm/templateInterpreter_sparc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/templateInterpreter_sparc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -456,6 +456,115 @@
 // Generate a fixed interpreter frame. This is identical setup for interpreted
 // methods and for native methods hence the shared code.
 
+
+//----------------------------------------------------------------------------------------------------
+// Stack frame layout
+//
+// When control flow reaches any of the entry types for the interpreter
+// the following holds ->
+//
+// C2 Calling Conventions:
+//
+// The entry code below assumes that the following registers are set
+// when coming in:
+//    G5_method: holds the Method* of the method to call
+//    Lesp:    points to the TOS of the callers expression stack
+//             after having pushed all the parameters
+//
+// The entry code does the following to setup an interpreter frame
+//   pop parameters from the callers stack by adjusting Lesp
+//   set O0 to Lesp
+//   compute X = (max_locals - num_parameters)
+//   bump SP up by X to accomadate the extra locals
+//   compute X = max_expression_stack
+//               + vm_local_words
+//               + 16 words of register save area
+//   save frame doing a save sp, -X, sp growing towards lower addresses
+//   set Lbcp, Lmethod, LcpoolCache
+//   set Llocals to i0
+//   set Lmonitors to FP - rounded_vm_local_words
+//   set Lesp to Lmonitors - 4
+//
+//  The frame has now been setup to do the rest of the entry code
+
+// Try this optimization:  Most method entries could live in a
+// "one size fits all" stack frame without all the dynamic size
+// calculations.  It might be profitable to do all this calculation
+// statically and approximately for "small enough" methods.
+
+//-----------------------------------------------------------------------------------------------
+
+// C1 Calling conventions
+//
+// Upon method entry, the following registers are setup:
+//
+// g2 G2_thread: current thread
+// g5 G5_method: method to activate
+// g4 Gargs  : pointer to last argument
+//
+//
+// Stack:
+//
+// +---------------+ <--- sp
+// |               |
+// : reg save area :
+// |               |
+// +---------------+ <--- sp + 0x40
+// |               |
+// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
+// |               |
+// +---------------+ <--- sp + 0x5c
+// |               |
+// :     free      :
+// |               |
+// +---------------+ <--- Gargs
+// |               |
+// :   arguments   :
+// |               |
+// +---------------+
+// |               |
+//
+//
+//
+// AFTER FRAME HAS BEEN SETUP for method interpretation the stack looks like:
+//
+// +---------------+ <--- sp
+// |               |
+// : reg save area :
+// |               |
+// +---------------+ <--- sp + 0x40
+// |               |
+// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
+// |               |
+// +---------------+ <--- sp + 0x5c
+// |               |
+// :               :
+// |               | <--- Lesp
+// +---------------+ <--- Lmonitors (fp - 0x18)
+// |   VM locals   |
+// +---------------+ <--- fp
+// |               |
+// : reg save area :
+// |               |
+// +---------------+ <--- fp + 0x40
+// |               |
+// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
+// |               |
+// +---------------+ <--- fp + 0x5c
+// |               |
+// :     free      :
+// |               |
+// +---------------+
+// |               |
+// : nonarg locals :
+// |               |
+// +---------------+
+// |               |
+// :   arguments   :
+// |               | <--- Llocals
+// +---------------+ <--- Gargs
+// |               |
+
 void TemplateInterpreterGenerator::generate_fixed_frame(bool native_call) {
   //
   //
@@ -599,136 +708,6 @@
 
 }
 
-// Empty method, generate a very fast return.
-
-address InterpreterGenerator::generate_empty_entry(void) {
-
-  // A method that does nother but return...
-
-  address entry = __ pc();
-  Label slow_path;
-
-  // do nothing for empty methods (do not even increment invocation counter)
-  if ( UseFastEmptyMethods) {
-    // If we need a safepoint check, generate full interpreter entry.
-    AddressLiteral sync_state(SafepointSynchronize::address_of_state());
-    __ set(sync_state, G3_scratch);
-    __ cmp_and_br_short(G3_scratch, SafepointSynchronize::_not_synchronized, Assembler::notEqual, Assembler::pn, slow_path);
-
-    // Code: _return
-    __ retl();
-    __ delayed()->mov(O5_savedSP, SP);
-
-    __ bind(slow_path);
-    (void) generate_normal_entry(false);
-
-    return entry;
-  }
-  return NULL;
-}
-
-// Call an accessor method (assuming it is resolved, otherwise drop into
-// vanilla (slow path) entry
-
-// Generates code to elide accessor methods
-// Uses G3_scratch and G1_scratch as scratch
-address InterpreterGenerator::generate_accessor_entry(void) {
-
-  // Code: _aload_0, _(i|a)getfield, _(i|a)return or any rewrites thereof;
-  // parameter size = 1
-  // Note: We can only use this code if the getfield has been resolved
-  //       and if we don't have a null-pointer exception => check for
-  //       these conditions first and use slow path if necessary.
-  address entry = __ pc();
-  Label slow_path;
-
-
-  // XXX: for compressed oops pointer loading and decoding doesn't fit in
-  // delay slot and damages G1
-  if ( UseFastAccessorMethods && !UseCompressedOops ) {
-    // Check if we need to reach a safepoint and generate full interpreter
-    // frame if so.
-    AddressLiteral sync_state(SafepointSynchronize::address_of_state());
-    __ load_contents(sync_state, G3_scratch);
-    __ cmp(G3_scratch, SafepointSynchronize::_not_synchronized);
-    __ cmp_and_br_short(G3_scratch, SafepointSynchronize::_not_synchronized, Assembler::notEqual, Assembler::pn, slow_path);
-
-    // Check if local 0 != NULL
-    __ ld_ptr(Gargs, G0, Otos_i ); // get local 0
-    // check if local 0 == NULL and go the slow path
-    __ br_null_short(Otos_i, Assembler::pn, slow_path);
-
-
-    // read first instruction word and extract bytecode @ 1 and index @ 2
-    // get first 4 bytes of the bytecodes (big endian!)
-    __ ld_ptr(G5_method, Method::const_offset(), G1_scratch);
-    __ ld(G1_scratch, ConstMethod::codes_offset(), G1_scratch);
-
-    // move index @ 2 far left then to the right most two bytes.
-    __ sll(G1_scratch, 2*BitsPerByte, G1_scratch);
-    __ srl(G1_scratch, 2*BitsPerByte - exact_log2(in_words(
-                      ConstantPoolCacheEntry::size()) * BytesPerWord), G1_scratch);
-
-    // get constant pool cache
-    __ ld_ptr(G5_method, Method::const_offset(), G3_scratch);
-    __ ld_ptr(G3_scratch, ConstMethod::constants_offset(), G3_scratch);
-    __ ld_ptr(G3_scratch, ConstantPool::cache_offset_in_bytes(), G3_scratch);
-
-    // get specific constant pool cache entry
-    __ add(G3_scratch, G1_scratch, G3_scratch);
-
-    // Check the constant Pool cache entry to see if it has been resolved.
-    // If not, need the slow path.
-    ByteSize cp_base_offset = ConstantPoolCache::base_offset();
-    __ ld_ptr(G3_scratch, cp_base_offset + ConstantPoolCacheEntry::indices_offset(), G1_scratch);
-    __ srl(G1_scratch, 2*BitsPerByte, G1_scratch);
-    __ and3(G1_scratch, 0xFF, G1_scratch);
-    __ cmp_and_br_short(G1_scratch, Bytecodes::_getfield, Assembler::notEqual, Assembler::pn, slow_path);
-
-    // Get the type and return field offset from the constant pool cache
-    __ ld_ptr(G3_scratch, cp_base_offset + ConstantPoolCacheEntry::flags_offset(), G1_scratch);
-    __ ld_ptr(G3_scratch, cp_base_offset + ConstantPoolCacheEntry::f2_offset(), G3_scratch);
-
-    Label xreturn_path;
-    // 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::tos_state_shift, G1_scratch);
-    // Make sure we don't need to mask G1_scratch after the above shift
-    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);
-    __ cmp(G1_scratch, itos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ld(Otos_i, G3_scratch, Otos_i);
-    __ cmp(G1_scratch, stos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ldsh(Otos_i, G3_scratch, Otos_i);
-    __ cmp(G1_scratch, ctos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->lduh(Otos_i, G3_scratch, Otos_i);
-#ifdef ASSERT
-    __ cmp(G1_scratch, btos);
-    __ br(Assembler::equal, true, Assembler::pt, xreturn_path);
-    __ delayed()->ldsb(Otos_i, G3_scratch, Otos_i);
-    __ should_not_reach_here();
-#endif
-    __ ldsb(Otos_i, G3_scratch, Otos_i);
-    __ bind(xreturn_path);
-
-    // _ireturn/_areturn
-    __ retl();                      // return from leaf routine
-    __ delayed()->mov(O5_savedSP, SP);
-
-    // Generate regular method entry
-    __ bind(slow_path);
-    (void) generate_normal_entry(false);
-    return entry;
-  }
-  return NULL;
-}
-
 // Method entry for java.lang.ref.Reference.get.
 address InterpreterGenerator::generate_Reference_get_entry(void) {
 #if INCLUDE_ALL_GCS
@@ -806,7 +785,7 @@
 
   // If G1 is not enabled then attempt to go through the accessor entry point
   // Reference.get is an accessor
-  return generate_accessor_entry();
+  return generate_jump_to_normal_entry();
 }
 
 //
@@ -1242,8 +1221,6 @@
 
 
 // Generic method entry to (asm) interpreter
-//------------------------------------------------------------------------------------------------------------------------
-//
 address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   address entry = __ pc();
 
@@ -1410,123 +1387,6 @@
   return entry;
 }
 
-
-//----------------------------------------------------------------------------------------------------
-// Entry points & stack frame layout
-//
-// Here we generate the various kind of entries into the interpreter.
-// The two main entry type are generic bytecode methods and native call method.
-// These both come in synchronized and non-synchronized versions but the
-// frame layout they create is very similar. The other method entry
-// types are really just special purpose entries that are really entry
-// and interpretation all in one. These are for trivial methods like
-// accessor, empty, or special math methods.
-//
-// When control flow reaches any of the entry types for the interpreter
-// the following holds ->
-//
-// C2 Calling Conventions:
-//
-// The entry code below assumes that the following registers are set
-// when coming in:
-//    G5_method: holds the Method* of the method to call
-//    Lesp:    points to the TOS of the callers expression stack
-//             after having pushed all the parameters
-//
-// The entry code does the following to setup an interpreter frame
-//   pop parameters from the callers stack by adjusting Lesp
-//   set O0 to Lesp
-//   compute X = (max_locals - num_parameters)
-//   bump SP up by X to accomadate the extra locals
-//   compute X = max_expression_stack
-//               + vm_local_words
-//               + 16 words of register save area
-//   save frame doing a save sp, -X, sp growing towards lower addresses
-//   set Lbcp, Lmethod, LcpoolCache
-//   set Llocals to i0
-//   set Lmonitors to FP - rounded_vm_local_words
-//   set Lesp to Lmonitors - 4
-//
-//  The frame has now been setup to do the rest of the entry code
-
-// Try this optimization:  Most method entries could live in a
-// "one size fits all" stack frame without all the dynamic size
-// calculations.  It might be profitable to do all this calculation
-// statically and approximately for "small enough" methods.
-
-//-----------------------------------------------------------------------------------------------
-
-// C1 Calling conventions
-//
-// Upon method entry, the following registers are setup:
-//
-// g2 G2_thread: current thread
-// g5 G5_method: method to activate
-// g4 Gargs  : pointer to last argument
-//
-//
-// Stack:
-//
-// +---------------+ <--- sp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- sp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- sp + 0x5c
-// |               |
-// :     free      :
-// |               |
-// +---------------+ <--- Gargs
-// |               |
-// :   arguments   :
-// |               |
-// +---------------+
-// |               |
-//
-//
-//
-// AFTER FRAME HAS BEEN SETUP for method interpretation the stack looks like:
-//
-// +---------------+ <--- sp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- sp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- sp + 0x5c
-// |               |
-// :               :
-// |               | <--- Lesp
-// +---------------+ <--- Lmonitors (fp - 0x18)
-// |   VM locals   |
-// +---------------+ <--- fp
-// |               |
-// : reg save area :
-// |               |
-// +---------------+ <--- fp + 0x40
-// |               |
-// : extra 7 slots :      note: these slots are not really needed for the interpreter (fix later)
-// |               |
-// +---------------+ <--- fp + 0x5c
-// |               |
-// :     free      :
-// |               |
-// +---------------+
-// |               |
-// : nonarg locals :
-// |               |
-// +---------------+
-// |               |
-// :   arguments   :
-// |               | <--- Llocals
-// +---------------+ <--- Gargs
-// |               |
-
 static int size_activation_helper(int callee_extra_locals, int max_stack, int monitor_size) {
 
   // Figure out the size of an interpreter frame (in words) given that we have a fully allocated

--- a/src/cpu/sparc/vm/vm_version_sparc.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/sparc/vm/vm_version_sparc.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -26,6 +26,7 @@
 #include "asm/macroAssembler.inline.hpp"
 #include "memory/resourceArea.hpp"
 #include "runtime/java.hpp"
+#include "runtime/os.hpp"
 #include "runtime/stubCodeGenerator.hpp"
 #include "vm_version_sparc.hpp"
 
@@ -249,7 +250,7 @@
                (!has_hardware_fsmuld() ? ", no-fsmuld" : ""));
 
   // buf is started with ", " or is empty
-  _features_str = strdup(strlen(buf) > 2 ? buf + 2 : buf);
+  _features_str = os::strdup(strlen(buf) > 2 ? buf + 2 : buf);
 
   // There are three 64-bit SPARC families that do not overlap, e.g.,
   // both is_ultra3() and is_sparc64() cannot be true at the same time.

--- a/src/cpu/x86/vm/assembler_x86.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/assembler_x86.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -3854,6 +3854,15 @@
 }
 
 // Carry-Less Multiplication Quadword
+void Assembler::pclmulqdq(XMMRegister dst, XMMRegister src, int mask) {
+  assert(VM_Version::supports_clmul(), "");
+  int encode = simd_prefix_and_encode(dst, dst, src, VEX_SIMD_66, VEX_OPCODE_0F_3A);
+  emit_int8(0x44);
+  emit_int8((unsigned char)(0xC0 | encode));
+  emit_int8((unsigned char)mask);
+}
+
+// Carry-Less Multiplication Quadword
 void Assembler::vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask) {
   assert(VM_Version::supports_avx() && VM_Version::supports_clmul(), "");
   bool vector256 = false;

--- a/src/cpu/x86/vm/assembler_x86.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/assembler_x86.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1837,6 +1837,7 @@
   void vpbroadcastd(XMMRegister dst, XMMRegister src);
 
   // Carry-Less Multiplication Quadword
+  void pclmulqdq(XMMRegister dst, XMMRegister src, int mask);
   void vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask);
 
   // AVX instruction which is used to clear upper 128 bits of YMM registers and

--- a/src/cpu/x86/vm/cppInterpreterGenerator_x86.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/cppInterpreterGenerator_x86.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2014, 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
@@ -27,21 +27,6 @@
 
  protected:
 
-#if 0
-  address generate_asm_interpreter_entry(bool synchronized);
-  address generate_native_entry(bool synchronized);
-  address generate_abstract_entry(void);
-  address generate_math_entry(AbstractInterpreter::MethodKind kind);
-  address generate_empty_entry(void);
-  address generate_accessor_entry(void);
-  address generate_Reference_get_entry(void);
-  void lock_method(void);
-  void generate_stack_overflow_check(void);
-
-  void generate_counter_incr(Label* overflow, Label* profile_method, Label* profile_method_continue);
-  void generate_counter_overflow(Label* do_continue);
-#endif
-
   void generate_more_monitors();
   void generate_deopt_handling();
   address generate_interpreter_frame_manager(bool synchronized); // C++ interpreter only

--- a/src/cpu/x86/vm/cppInterpreter_x86.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/cppInterpreter_x86.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2007, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2007, 2014, 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
@@ -66,9 +66,6 @@
 #define __ _masm->
 #define STATE(field_name) (Address(state, byte_offset_of(BytecodeInterpreter, field_name)))
 
-Label fast_accessor_slow_entry_path;  // fast accessor methods need to be able to jmp to unsynchronized
-                                      // c++ interpreter entry point this holds that entry point label.
-
 // default registers for state and sender_sp
 // state and sender_sp are the same on 32bit because we have no choice.
 // state could be rsi on 64bit but it is an arg reg and not callee save
@@ -660,7 +657,6 @@
   // generate_method_entry) so the guard should work for them too.
   //
 
-  // monitor entry size: see picture of stack set (generate_method_entry) and frame_i486.hpp
   const int entry_size    = frame::interpreter_frame_monitor_size() * wordSize;
 
   // total overhead size: entry_size + (saved rbp, thru expr stack bottom).
@@ -794,156 +790,6 @@
   __ lock_object(monitor);
 }
 
-// Call an accessor method (assuming it is resolved, otherwise drop into vanilla (slow path) entry
-
-address InterpreterGenerator::generate_accessor_entry(void) {
-
-  // rbx: Method*
-
-  // rsi/r13: senderSP must preserved for slow path, set SP to it on fast path
-
-  Label xreturn_path;
-
-  // do fastpath for resolved accessor methods
-  if (UseFastAccessorMethods) {
-
-    address entry_point = __ pc();
-
-    Label slow_path;
-    // If we need a safepoint check, generate full interpreter entry.
-    ExternalAddress state(SafepointSynchronize::address_of_state());
-    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
-             SafepointSynchronize::_not_synchronized);
-
-    __ jcc(Assembler::notEqual, slow_path);
-    // ASM/C++ Interpreter
-    // Code: _aload_0, _(i|a)getfield, _(i|a)return or any rewrites thereof; parameter size = 1
-    // Note: We can only use this code if the getfield has been resolved
-    //       and if we don't have a null-pointer exception => check for
-    //       these conditions first and use slow path if necessary.
-    // rbx,: method
-    // rcx: receiver
-    __ movptr(rax, Address(rsp, wordSize));
-
-    // check if local 0 != NULL and read field
-    __ testptr(rax, rax);
-    __ jcc(Assembler::zero, slow_path);
-
-    // read first instruction word and extract bytecode @ 1 and index @ 2
-    __ movptr(rdx, Address(rbx, Method::const_offset()));
-    __ movptr(rdi, Address(rdx, ConstMethod::constants_offset()));
-    __ movl(rdx, Address(rdx, ConstMethod::codes_offset()));
-    // Shift codes right to get the index on the right.
-    // The bytecode fetched looks like <index><0xb4><0x2a>
-    __ shrl(rdx, 2*BitsPerByte);
-    __ shll(rdx, exact_log2(in_words(ConstantPoolCacheEntry::size())));
-    __ movptr(rdi, Address(rdi, ConstantPool::cache_offset_in_bytes()));
-
-    // rax,: local 0
-    // rbx,: method
-    // rcx: receiver - do not destroy since it is needed for slow path!
-    // rcx: scratch
-    // rdx: constant pool cache index
-    // rdi: constant pool cache
-    // rsi/r13: sender sp
-
-    // check if getfield has been resolved and read constant pool cache entry
-    // check the validity of the cache entry by testing whether _indices field
-    // contains Bytecode::_getfield in b1 byte.
-    assert(in_words(ConstantPoolCacheEntry::size()) == 4, "adjust shift below");
-    __ movl(rcx,
-            Address(rdi,
-                    rdx,
-                    Address::times_ptr, ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()));
-    __ shrl(rcx, 2*BitsPerByte);
-    __ andl(rcx, 0xFF);
-    __ cmpl(rcx, Bytecodes::_getfield);
-    __ jcc(Assembler::notEqual, slow_path);
-
-    // Note: constant pool entry is not valid before bytecode is resolved
-    __ movptr(rcx,
-            Address(rdi,
-                    rdx,
-                    Address::times_ptr, ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::f2_offset()));
-    __ movl(rdx,
-            Address(rdi,
-                    rdx,
-                    Address::times_ptr, ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::flags_offset()));
-
-    Label notByte, notShort, notChar;
-    const Address field_address (rax, rcx, Address::times_1);
-
-    // Need to differentiate between igetfield, agetfield, bgetfield etc.
-    // because they are different sizes.
-    // Use the type from the constant pool cache
-    __ shrl(rdx, ConstantPoolCacheEntry::tos_state_shift);
-    // Make sure we don't need to mask rdx after the above shift
-    ConstantPoolCacheEntry::verify_tos_state_shift();
-#ifdef _LP64
-    Label notObj;
-    __ cmpl(rdx, atos);
-    __ jcc(Assembler::notEqual, notObj);
-    // atos
-    __ movptr(rax, field_address);
-    __ jmp(xreturn_path);
-
-    __ bind(notObj);
-#endif // _LP64
-    __ cmpl(rdx, btos);
-    __ jcc(Assembler::notEqual, notByte);
-    __ load_signed_byte(rax, field_address);
-    __ jmp(xreturn_path);
-
-    __ bind(notByte);
-    __ cmpl(rdx, stos);
-    __ jcc(Assembler::notEqual, notShort);
-    __ load_signed_short(rax, field_address);
-    __ jmp(xreturn_path);
-
-    __ bind(notShort);
-    __ cmpl(rdx, ctos);
-    __ jcc(Assembler::notEqual, notChar);
-    __ load_unsigned_short(rax, field_address);
-    __ jmp(xreturn_path);
-
-    __ bind(notChar);
-#ifdef ASSERT
-    Label okay;
-#ifndef _LP64
-    __ cmpl(rdx, atos);
-    __ jcc(Assembler::equal, okay);
-#endif // _LP64
-    __ cmpl(rdx, itos);
-    __ jcc(Assembler::equal, okay);
-    __ stop("what type is this?");
-    __ bind(okay);
-#endif // ASSERT
-    // All the rest are a 32 bit wordsize
-    __ movl(rax, field_address);
-
-    __ bind(xreturn_path);
-
-    // _ireturn/_areturn
-    __ pop(rdi);                               // get return address
-    __ mov(rsp, sender_sp_on_entry);           // set sp to sender sp
-    __ jmp(rdi);
-
-    // generate a vanilla interpreter entry as the slow path
-    __ bind(slow_path);
-    // We will enter c++ interpreter looking like it was
-    // called by the call_stub this will cause it to return
-    // a tosca result to the invoker which might have been
-    // the c++ interpreter itself.
-
-    __ jmp(fast_accessor_slow_entry_path);
-    return entry_point;
-
-  } else {
-    return NULL;
-  }
-
-}
-
 address InterpreterGenerator::generate_Reference_get_entry(void) {
 #if INCLUDE_ALL_GCS
   if (UseG1GC) {
@@ -961,7 +807,7 @@
 
   // If G1 is not enabled then attempt to go through the accessor entry point
   // Reference.get is an accessor
-  return generate_accessor_entry();
+  return generate_jump_to_normal_entry();
 }
 
 //
@@ -1670,10 +1516,6 @@
 
   address entry_point = __ pc();
 
-  // Fast accessor methods share this entry point.
-  // This works because frame manager is in the same codelet
-  if (UseFastAccessorMethods && !synchronized) __ bind(fast_accessor_slow_entry_path);
-
   Label dispatch_entry_2;
   __ movptr(rcx, sender_sp_on_entry);
   __ movptr(state, (int32_t)NULL_WORD);                              // no current activation
@@ -2212,40 +2054,6 @@
   return entry_point;
 }
 
-address AbstractInterpreterGenerator::generate_method_entry(AbstractInterpreter::MethodKind kind) {
-  // determine code generation flags
-  bool synchronized = false;
-  address entry_point = NULL;
-
-  switch (kind) {
-    case Interpreter::zerolocals             :                                                                             break;
-    case Interpreter::zerolocals_synchronized: synchronized = true;                                                        break;
-    case Interpreter::native                 : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(false);  break;
-    case Interpreter::native_synchronized    : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(true);   break;
-    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::java_lang_math_sin     : // fall thru
-    case Interpreter::java_lang_math_cos     : // fall thru
-    case Interpreter::java_lang_math_tan     : // fall thru
-    case Interpreter::java_lang_math_abs     : // fall thru
-    case Interpreter::java_lang_math_log     : // fall thru
-    case Interpreter::java_lang_math_log10   : // fall thru
-    case Interpreter::java_lang_math_sqrt    : // fall thru
-    case Interpreter::java_lang_math_pow     : // fall thru
-    case Interpreter::java_lang_math_exp     : // fall thru
-      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;
-  }
-
-  if (entry_point) return entry_point;
-
-  return ((InterpreterGenerator*)this)->generate_normal_entry(synchronized);
-
-}
 
 InterpreterGenerator::InterpreterGenerator(StubQueue* code)
  : CppInterpreterGenerator(code) {

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/src/cpu/x86/vm/interpreterGenerator_x86.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2014, 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.
+ *
+ */
+
+#include "precompiled.hpp"
+#include "asm/macroAssembler.hpp"
+#include "interpreter/interpreter.hpp"
+#include "interpreter/interpreterGenerator.hpp"
+#include "interpreter/interpreterRuntime.hpp"
+#include "interpreter/interp_masm.hpp"
+
+#define __ _masm->
+
+// Jump into normal path for accessor and empty entry to jump to normal entry
+// The "fast" optimization don't update compilation count therefore can disable inlining
+// for these functions that should be inlined.
+address InterpreterGenerator::generate_jump_to_normal_entry(void) {
+  address entry_point = __ pc();
+
+  assert(Interpreter::entry_for_kind(Interpreter::zerolocals) != NULL, "should already be generated");
+  __ jump(RuntimeAddress(Interpreter::entry_for_kind(Interpreter::zerolocals)));
+  return entry_point;
+}
+
+// Abstract method entry
+// Attempt to execute abstract method. Throw exception
+address InterpreterGenerator::generate_abstract_entry(void) {
+
+  address entry_point = __ pc();
+
+  // abstract method entry
+
+#ifndef CC_INTERP
+  //  pop return address, reset last_sp to NULL
+  __ empty_expression_stack();
+  __ restore_bcp();      // rsi must be correct for exception handler   (was destroyed)
+  __ restore_locals();   // make sure locals pointer is correct as well (was destroyed)
+#endif
+
+  // throw exception
+  __ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_AbstractMethodError));
+  // the call_VM checks for exception, so we should never return here.
+  __ should_not_reach_here();
+
+  return entry_point;
+}

--- a/src/cpu/x86/vm/interpreterGenerator_x86.hpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/interpreterGenerator_x86.hpp	Thu Aug 28 14:53:43 2014 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2014, 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
@@ -36,8 +36,9 @@
   address generate_native_entry(bool synchronized);
   address generate_abstract_entry(void);
   address generate_math_entry(AbstractInterpreter::MethodKind kind);
-  address generate_empty_entry(void);
-  address generate_accessor_entry(void);
+  address generate_jump_to_normal_entry(void);
+  address generate_accessor_entry(void) { return generate_jump_to_normal_entry(); }
+  address generate_empty_entry(void) { return generate_jump_to_normal_entry(); }
   address generate_Reference_get_entry();
   address generate_CRC32_update_entry();
   address generate_CRC32_updateBytes_entry(AbstractInterpreter::MethodKind kind);

--- a/src/cpu/x86/vm/interpreter_x86_32.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/interpreter_x86_32.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -67,45 +67,6 @@
 }
 
 
-//
-// Various method entries (that c++ and asm interpreter agree upon)
-//------------------------------------------------------------------------------------------------------------------------
-//
-//
-
-// Empty method, generate a very fast return.
-
-address InterpreterGenerator::generate_empty_entry(void) {
-
-  // rbx,: Method*
-  // rcx: receiver (unused)
-  // rsi: previous interpreter state (C++ interpreter) must preserve
-  // rsi: sender sp must set sp to this value on return
-
-  if (!UseFastEmptyMethods) return NULL;
-
-  address entry_point = __ pc();
-
-  // If we need a safepoint check, generate full interpreter entry.
-  Label slow_path;
-  ExternalAddress state(SafepointSynchronize::address_of_state());
-  __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
-           SafepointSynchronize::_not_synchronized);
-  __ jcc(Assembler::notEqual, slow_path);
-
-  // do nothing for empty methods (do not even increment invocation counter)
-  // Code: _return
-  // _return
-  // return w/o popping parameters
-  __ pop(rax);
-  __ mov(rsp, rsi);
-  __ jmp(rax);
-
-  __ bind(slow_path);
-  (void) generate_normal_entry(false);
-  return entry_point;
-}
-
 address InterpreterGenerator::generate_math_entry(AbstractInterpreter::MethodKind kind) {
 
   // rbx,: Method*
@@ -216,36 +177,6 @@
 }
 
 
-// Abstract method entry
-// Attempt to execute abstract method. Throw exception
-address InterpreterGenerator::generate_abstract_entry(void) {
-
-  // rbx,: Method*
-  // rcx: receiver (unused)
-  // rsi: previous interpreter state (C++ interpreter) must preserve
-
-  // rsi: sender SP
-
-  address entry_point = __ pc();
-
-  // abstract method entry
-
-#ifndef CC_INTERP
-  //  pop return address, reset last_sp to NULL
-  __ empty_expression_stack();
-  __ restore_bcp();      // rsi must be correct for exception handler   (was destroyed)
-  __ restore_locals();   // make sure locals pointer is correct as well (was destroyed)
-#endif
-
-  // throw exception
-  __ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_AbstractMethodError));
-  // the call_VM checks for exception, so we should never return here.
-  __ should_not_reach_here();
-
-  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

--- a/src/cpu/x86/vm/interpreter_x86_64.cpp	Thu Aug 21 14:16:15 2014 -0700
+++ b/src/cpu/x86/vm/interpreter_x86_64.cpp	Thu Aug 28 14:53:43 2014 -0700
@@ -301,66 +301,6 @@
   return entry_point;
 }
 
-
-// Abstract method entry
-// Attempt to execute abstract method. Throw exception
-address InterpreterGenerator::generate_abstract_entry(void) {
-  // rbx: Method*
-  // r13: sender SP
-
-  address entry_point = __ pc();
-
-  // abstract method entry
-
-#ifndef CC_INTERP
-  //  pop return address, reset last_sp to NULL
-  __ empty_expression_stack();
-  __ restore_bcp();      // rsi must be correct for exception handler   (was destroyed)
-  __ restore_locals();   // make sure locals pointer is correct as well (was destroyed)
-#endif
-
-  // throw exception
-  __ call_VM(noreg, CAST_FROM_FN_PTR(address,
-                             InterpreterRuntime::throw_AbstractMethodError));
-  // the call_VM checks for exception, so we should never return here.
-  __ should_not_reach_here();
-
-  return entry_point;
-}
-
-
-// Empty method, generate a very fast return.
-
-address InterpreterGenerator::generate_empty_entry(void) {
-  // rbx: Method*
-  // r13: sender sp must set sp to this value on return
-
-  if (!UseFastEmptyMethods) {
-    return NULL;
-  }
-
-  address entry_point = __ pc();
-
-  // If we need a safepoint check, generate full interpreter entry.
-  Label slow_path;
-  __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
-           SafepointSynchronize::_not_synchronized);
-  __ jcc(Assembler::notEqual, slow_path);
-
-  // do nothing for empty methods (do not even increment invocation counter)
-  // Code: _return
-  // _return
-  // return w/o popping parameters
-  __ pop(rax);
-  __ mov(rsp, r13);
-  __ jmp(rax);
-
-  __ bind(slow_path);
-  (void) generate_normal_entry(false);
-  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