Mergehs25-b03

author: kvn <none@none> 2012-09-28 10:16:29 -0700
committer: kvn <none@none> 2012-09-28 10:16:29 -0700
commit: 7ce94fd32972533edc4ecf0d2a270005abab7c10 (patch)
tree: add83032c0821b8b7210ba3d7e7b58b27a16fe96
parent: 1bac024e917c46f2b5884889b8faffb60546d7c1 (diff)
parent: 00e5c6d91a3db80e48359621f9b85332477b6e1b (diff)
138 files changed, 5464 insertions, 4475 deletions
diff --git a/make/windows/makefiles/projectcreator.make b/make/windows/makefiles/projectcreator.make
index ebc5a0a84..2af47d48f 100644
--- a/make/windows/makefiles/projectcreator.make
+++ b/make/windows/makefiles/projectcreator.make
@@ -29,12 +29,11 @@
 # HOTSPOTRELEASEBINDEST, or HOTSPOTDEBUGBINDEST environment variables.
 
 ProjectCreatorSources=\
-        $(WorkSpace)\src\share\tools\ProjectCreator\DirectoryTree.java \
-        $(WorkSpace)\src\share\tools\ProjectCreator\DirectoryTreeNode.java \
-        $(WorkSpace)\src\share\tools\ProjectCreator\FileFormatException.java \
         $(WorkSpace)\src\share\tools\ProjectCreator\ProjectCreator.java \
+        $(WorkSpace)\src\share\tools\ProjectCreator\FileTreeCreator.java \
+        $(WorkSpace)\src\share\tools\ProjectCreator\FileTreeCreatorVC7.java \
+        $(WorkSpace)\src\share\tools\ProjectCreator\FileTreeCreatorVC10.java \
         $(WorkSpace)\src\share\tools\ProjectCreator\WinGammaPlatform.java \
-        $(WorkSpace)\src\share\tools\ProjectCreator\WinGammaPlatformVC6.java \
         $(WorkSpace)\src\share\tools\ProjectCreator\WinGammaPlatformVC7.java \
         $(WorkSpace)\src\share\tools\ProjectCreator\WinGammaPlatformVC8.java \
         $(WorkSpace)\src\share\tools\ProjectCreator\WinGammaPlatformVC9.java \
@@ -57,10 +56,24 @@ ProjectCreatorIncludesPRIVATE=\
         -relativeInclude src\os_cpu\windows_$(Platform_arch)\vm \
         -relativeInclude src\cpu\$(Platform_arch)\vm \
         -absoluteInclude $(HOTSPOTBUILDSPACE)/%f/generated \
-        -ignorePath $(HOTSPOTBUILDSPACE)/%f/generated \
-        -ignorePath src\share\vm\adlc \
-        -ignorePath src\share\vm\shark \
-        -ignorePath posix
+        -relativeSrcInclude src \
+        -absoluteSrcInclude $(HOTSPOTBUILDSPACE) \
+        -ignorePath $(HOTSPOTBUILDSPACE) \
+        -ignorePath launcher \
+        -ignorePath share\vm\adlc \
+        -ignorePath share\vm\shark \
+        -ignorePath share\tools \
+        -ignorePath solaris \
+        -ignorePath posix \
+        -ignorePath sparc \
+        -ignorePath linux \
+        -ignorePath bsd \
+        -ignorePath osx \
+        -ignorePath arm \
+        -ignorePath ppc \
+        -ignorePath zero \
+        -hidePath .hg
+	
 
 # This is referenced externally by both the IDE and batch builds
 ProjectCreatorOptions=
@@ -84,6 +97,7 @@ ProjectCreatorIDEOptions=\
         $(ProjectCreatorIDEOptions) \
         -sourceBase $(HOTSPOTWORKSPACE) \
         -buildBase $(HOTSPOTBUILDSPACE)\%f\%b \
+        -buildSpace $(HOTSPOTBUILDSPACE) \
         -startAt src \
         -compiler $(VcVersion) \
         -projectFileName $(HOTSPOTBUILDSPACE)\$(ProjectFile) \
@@ -103,6 +117,7 @@ ProjectCreatorIDEOptions=\
         -define TARGET_OS_ARCH_windows_x86 \
         -define TARGET_OS_FAMILY_windows \
         -define TARGET_COMPILER_visCPP \
+        -define INCLUDE_TRACE \
        $(ProjectCreatorIncludesPRIVATE)
 
 # Add in build-specific options
@@ -125,9 +140,13 @@ ProjectCreatorIDEOptions=$(ProjectCreatorIDEOptions) \
 !endif
 
 ProjectCreatorIDEOptionsIgnoreCompiler1=\
+ -ignorePath_TARGET compiler1 \
+ -ignorePath_TARGET tiered \
  -ignorePath_TARGET c1_
 
 ProjectCreatorIDEOptionsIgnoreCompiler2=\
+ -ignorePath_TARGET compiler2 \
+ -ignorePath_TARGET tiered \
  -ignorePath_TARGET src/share/vm/opto \
  -ignorePath_TARGET src/share/vm/libadt \
  -ignorePath_TARGET adfiles \
@@ -209,6 +228,7 @@ $(ProjectCreatorIDEOptionsIgnoreCompiler2:TARGET=kernel) \
 ##################################################
 ProjectCreatorIDEOptions=$(ProjectCreatorIDEOptions) \
  -define_compiler1 COMPILER1 \
+ -ignorePath_compiler1 core \
 $(ProjectCreatorIDEOptionsIgnoreCompiler2:TARGET=compiler1)
 
 ##################################################
@@ -217,18 +237,19 @@ $(ProjectCreatorIDEOptionsIgnoreCompiler2:TARGET=compiler1)
 #NOTE! This list must be kept in sync with GENERATED_NAMES in adlc.make.
 ProjectCreatorIDEOptions=$(ProjectCreatorIDEOptions) \
  -define_compiler2 COMPILER2 \
+ -ignorePath_compiler2 core \
  -additionalFile_compiler2 $(Platform_arch_model).ad \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model).cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model).hpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_clone.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_expand.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_format.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_gen.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_misc.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_peephole.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles ad_$(Platform_arch_model)_pipeline.cpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles adGlobals_$(Platform_arch_model).hpp \
- -additionalGeneratedFile_compiler2 $(HOTSPOTBUILDSPACE)/%f/generated/adfiles dfa_$(Platform_arch_model).cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model).cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model).hpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_clone.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_expand.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_format.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_gen.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_misc.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_peephole.cpp \
+ -additionalFile_compiler2 ad_$(Platform_arch_model)_pipeline.cpp \
+ -additionalFile_compiler2 adGlobals_$(Platform_arch_model).hpp \
+ -additionalFile_compiler2 dfa_$(Platform_arch_model).cpp \
  $(ProjectCreatorIDEOptionsIgnoreCompiler1:TARGET=compiler2)
 
 # Add in the jvmti (JSR-163) options
@@ -237,8 +258,8 @@ ProjectCreatorIDEOptions=$(ProjectCreatorIDEOptions) \
 #       code merge was done correctly (@see jvmti.make and jvmtiEnvFill.java).
 #       If so, they would then check it in as a new version of jvmtiEnv.cpp.
 ProjectCreatorIDEOptions=$(ProjectCreatorIDEOptions) \
- -additionalGeneratedFile $(HOTSPOTBUILDSPACE)/%f/generated/jvmtifiles jvmtiEnv.hpp \
- -additionalGeneratedFile $(HOTSPOTBUILDSPACE)/%f/generated/jvmtifiles jvmtiEnter.cpp \
- -additionalGeneratedFile $(HOTSPOTBUILDSPACE)/%f/generated/jvmtifiles jvmtiEnterTrace.cpp \
- -additionalGeneratedFile $(HOTSPOTBUILDSPACE)/%f/generated/jvmtifiles jvmti.h \
- -additionalGeneratedFile $(HOTSPOTBUILDSPACE)/%f/generated/jvmtifiles bytecodeInterpreterWithChecks.cpp
+ -additionalFile jvmtiEnv.hpp \
+ -additionalFile jvmtiEnter.cpp \
+ -additionalFile jvmtiEnterTrace.cpp \
+ -additionalFile jvmti.h \
+ -additionalFile bytecodeInterpreterWithChecks.cpp
diff --git a/src/cpu/sparc/vm/assembler_sparc.cpp b/src/cpu/sparc/vm/assembler_sparc.cpp
index 22cf4ced7..c4e319d0a 100644
--- a/src/cpu/sparc/vm/assembler_sparc.cpp
+++ b/src/cpu/sparc/vm/assembler_sparc.cpp
@@ -725,24 +725,6 @@ void MacroAssembler::jump(const AddressLiteral& addrlit, Register temp, int offs
 }
 
 
-// Convert to C varargs format
-void MacroAssembler::set_varargs( Argument inArg, Register d ) {
-  // spill register-resident args to their memory slots
-  // (SPARC calling convention requires callers to have already preallocated these)
-  // Note that the inArg might in fact be an outgoing argument,
-  // if a leaf routine or stub does some tricky argument shuffling.
-  // This routine must work even though one of the saved arguments
-  // is in the d register (e.g., set_varargs(Argument(0, false), O0)).
-  for (Argument savePtr = inArg;
-       savePtr.is_register();
-       savePtr = savePtr.successor()) {
-    st_ptr(savePtr.as_register(), savePtr.address_in_frame());
-  }
-  // return the address of the first memory slot
-  Address a = inArg.address_in_frame();
-  add(a.base(), a.disp(), d);
-}
-
 // Conditional breakpoint (for assertion checks in assembly code)
 void MacroAssembler::breakpoint_trap(Condition c, CC cc) {
   trap(c, cc, G0, ST_RESERVED_FOR_USER_0);
@@ -2943,6 +2925,20 @@ void MacroAssembler::lookup_interface_method(Register recv_klass,
   assert(itable_index.is_constant() || itable_index.as_register() == method_result,
          "caller must use same register for non-constant itable index as for method");
 
+  Label L_no_such_interface_restore;
+  bool did_save = false;
+  if (scan_temp == noreg || sethi_temp == noreg) {
+    Register recv_2 = recv_klass->is_global() ? recv_klass : L0;
+    Register intf_2 = intf_klass->is_global() ? intf_klass : L1;
+    assert(method_result->is_global(), "must be able to return value");
+    scan_temp  = L2;
+    sethi_temp = L3;
+    save_frame_and_mov(0, recv_klass, recv_2, intf_klass, intf_2);
+    recv_klass = recv_2;
+    intf_klass = intf_2;
+    did_save = true;
+  }
+
   // Compute start of first itableOffsetEntry (which is at the end of the vtable)
   int vtable_base = InstanceKlass::vtable_start_offset() * wordSize;
   int scan_step   = itableOffsetEntry::size() * wordSize;
@@ -2981,7 +2977,7 @@ void MacroAssembler::lookup_interface_method(Register recv_klass,
   //     result = (klass + scan->offset() + itable_index);
   //   }
   // }
-  Label search, found_method;
+  Label L_search, L_found_method;
 
   for (int peel = 1; peel >= 0; peel--) {
     // %%%% Could load both offset and interface in one ldx, if they were
@@ -2991,23 +2987,23 @@ void MacroAssembler::lookup_interface_method(Register recv_klass,
     // Check that this entry is non-null.  A null entry means that
     // the receiver class doesn't implement the interface, and wasn't the
     // same as when the caller was compiled.
-    bpr(Assembler::rc_z, false, Assembler::pn, method_result, L_no_such_interface);
+    bpr(Assembler::rc_z, false, Assembler::pn, method_result, did_save ? L_no_such_interface_restore : L_no_such_interface);
     delayed()->cmp(method_result, intf_klass);
 
     if (peel) {
-      brx(Assembler::equal,    false, Assembler::pt, found_method);
+      brx(Assembler::equal,    false, Assembler::pt, L_found_method);
     } else {
-      brx(Assembler::notEqual, false, Assembler::pn, search);
+      brx(Assembler::notEqual, false, Assembler::pn, L_search);
       // (invert the test to fall through to found_method...)
     }
     delayed()->add(scan_temp, scan_step, scan_temp);
 
     if (!peel)  break;
 
-    bind(search);
+    bind(L_search);
   }
 
-  bind(found_method);
+  bind(L_found_method);
 
   // Got a hit.
   int ito_offset = itableOffsetEntry::offset_offset_in_bytes();
@@ -3015,6 +3011,18 @@ void MacroAssembler::lookup_interface_method(Register recv_klass,
   ito_offset -= scan_step;
   lduw(scan_temp, ito_offset, scan_temp);
   ld_ptr(recv_klass, scan_temp, method_result);
+
+  if (did_save) {
+    Label L_done;
+    ba(L_done);
+    delayed()->restore();
+
+    bind(L_no_such_interface_restore);
+    ba(L_no_such_interface);
+    delayed()->restore();
+
+    bind(L_done);
+  }
 }
 
 
diff --git a/src/cpu/sparc/vm/assembler_sparc.hpp b/src/cpu/sparc/vm/assembler_sparc.hpp
index 99153f33b..48f76a85d 100644
--- a/src/cpu/sparc/vm/assembler_sparc.hpp
+++ b/src/cpu/sparc/vm/assembler_sparc.hpp
@@ -2428,9 +2428,6 @@ public:
   static void test();
 #endif
 
-  // convert an incoming arglist to varargs format; put the pointer in d
-  void set_varargs( Argument a, Register d );
-
   int total_frame_size_in_bytes(int extraWords);
 
   // used when extraWords known statically
diff --git a/src/cpu/sparc/vm/assembler_sparc.inline.hpp b/src/cpu/sparc/vm/assembler_sparc.inline.hpp
index bccbe13d4..63359d511 100644
--- a/src/cpu/sparc/vm/assembler_sparc.inline.hpp
+++ b/src/cpu/sparc/vm/assembler_sparc.inline.hpp
@@ -347,7 +347,11 @@ inline void Assembler::sub(Register s1, RegisterOrConstant s2, Register d, int o
 inline void Assembler::swap(    Register s1, Register s2, Register d) { v9_dep();  emit_long( op(ldst_op) | rd(d) | op3(swap_op3) | rs1(s1) | rs2(s2) ); }
 inline void Assembler::swap(    Register s1, int simm13a, Register d) { v9_dep();  emit_data( op(ldst_op) | rd(d) | op3(swap_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
 
-inline void Assembler::swap(    Address& a, Register d, int offset ) { relocate(a.rspec(offset)); swap(  a.base(), a.disp() + offset, d ); }
+inline void Assembler::swap(    Address& a, Register d, int offset ) {
+  relocate(a.rspec(offset));
+  if (a.has_index()) { assert(offset == 0, ""); swap( a.base(), a.index(), d         ); }
+  else               {                          swap( a.base(), a.disp() + offset, d ); }
+}
 
 
 // Use the right loads/stores for the platform
diff --git a/src/cpu/sparc/vm/c1_LIRAssembler_sparc.cpp b/src/cpu/sparc/vm/c1_LIRAssembler_sparc.cpp
index 2890c56d8..61ec673c5 100644
--- a/src/cpu/sparc/vm/c1_LIRAssembler_sparc.cpp
+++ b/src/cpu/sparc/vm/c1_LIRAssembler_sparc.cpp
@@ -1315,7 +1315,13 @@ void LIR_Assembler::const2reg(LIR_Opr src, LIR_Opr dest, LIR_PatchCode patch_cod
 
 Address LIR_Assembler::as_Address(LIR_Address* addr) {
   Register reg = addr->base()->as_register();
-  return Address(reg, addr->disp());
+  LIR_Opr index = addr->index();
+  if (index->is_illegal()) {
+    return Address(reg, addr->disp());
+  } else {
+    assert (addr->disp() == 0, "unsupported address mode");
+    return Address(reg, index->as_pointer_register());
+  }
 }
 
 
@@ -3438,7 +3444,28 @@ void LIR_Assembler::peephole(LIR_List* lir) {
   }
 }
 
+void LIR_Assembler::atomic_op(LIR_Code code, LIR_Opr src, LIR_Opr data, LIR_Opr dest, LIR_Opr tmp) {
+  LIR_Address* addr = src->as_address_ptr();
 
+  assert(data == dest, "swap uses only 2 operands");
+  assert (code == lir_xchg, "no xadd on sparc");
 
+  if (data->type() == T_INT) {
+    __ swap(as_Address(addr), data->as_register());
+  } else if (data->is_oop()) {
+    Register obj = data->as_register();
+    Register narrow = tmp->as_register();
+#ifdef _LP64
+    assert(UseCompressedOops, "swap is 32bit only");
+    __ encode_heap_oop(obj, narrow);
+    __ swap(as_Address(addr), narrow);
+    __ decode_heap_oop(narrow, obj);
+#else
+    __ swap(as_Address(addr), obj);
+#endif
+  } else {
+    ShouldNotReachHere();
+  }
+}
 
 #undef __
diff --git a/src/cpu/sparc/vm/c1_LIRGenerator_sparc.cpp b/src/cpu/sparc/vm/c1_LIRGenerator_sparc.cpp
index 75b7b6cdf..2d4b3a2f1 100644
--- a/src/cpu/sparc/vm/c1_LIRGenerator_sparc.cpp
+++ b/src/cpu/sparc/vm/c1_LIRGenerator_sparc.cpp
@@ -1204,3 +1204,58 @@ void LIRGenerator::get_Object_unsafe(LIR_Opr dst, LIR_Opr src, LIR_Opr offset,
     __ load(addr, dst);
   }
 }
+
+void LIRGenerator::do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) {
+  BasicType type = x->basic_type();
+  LIRItem src(x->object(), this);
+  LIRItem off(x->offset(), this);
+  LIRItem value(x->value(), this);
+
+  src.load_item();
+  value.load_item();
+  off.load_nonconstant();
+
+  LIR_Opr dst = rlock_result(x, type);
+  LIR_Opr data = value.result();
+  bool is_obj = (type == T_ARRAY || type == T_OBJECT);
+  LIR_Opr offset = off.result();
+
+  if (data != dst) {
+    __ move(data, dst);
+    data = dst;
+  }
+
+  assert (!x->is_add() && (type == T_INT || (is_obj LP64_ONLY(&& UseCompressedOops))), "unexpected type");
+  LIR_Address* addr;
+  if (offset->is_constant()) {
+
+#ifdef _LP64
+    jlong l = offset->as_jlong();
+    assert((jlong)((jint)l) == l, "offset too large for constant");
+    jint c = (jint)l;
+#else
+    jint c = offset->as_jint();
+#endif
+    addr = new LIR_Address(src.result(), c, type);
+  } else {
+    addr = new LIR_Address(src.result(), offset, type);
+  }
+
+  LIR_Opr tmp = LIR_OprFact::illegalOpr;
+  LIR_Opr ptr = LIR_OprFact::illegalOpr;
+
+  if (is_obj) {
+    // Do the pre-write barrier, if any.
+    // barriers on sparc don't work with a base + index address
+    tmp = FrameMap::G3_opr;
+    ptr = new_pointer_register();
+    __ add(src.result(), off.result(), ptr);
+    pre_barrier(ptr, LIR_OprFact::illegalOpr /* pre_val */,
+                true /* do_load */, false /* patch */, NULL);
+  }
+  __ xchg(LIR_OprFact::address(addr), data, dst, tmp);
+  if (is_obj) {
+    // Seems to be a precise address
+    post_barrier(ptr, data);
+  }
+}
diff --git a/src/cpu/sparc/vm/interp_masm_sparc.cpp b/src/cpu/sparc/vm/interp_masm_sparc.cpp
index d92291fb3..2e2040efc 100644
--- a/src/cpu/sparc/vm/interp_masm_sparc.cpp
+++ b/src/cpu/sparc/vm/interp_masm_sparc.cpp
@@ -1395,12 +1395,17 @@ void InterpreterMacroAssembler::test_invocation_counter_for_mdp(Register invocat
   AddressLiteral profile_limit((address) &InvocationCounter::InterpreterProfileLimit);
   sethi(profile_limit, Rtmp);
   ld(Rtmp, profile_limit.low10(), Rtmp);
-  cmp_and_br_short(invocation_count, Rtmp, Assembler::lessUnsigned, Assembler::pn, profile_continue);
+  cmp(invocation_count, Rtmp);
+  // Use long branches because call_VM() code and following code generated by
+  // test_backedge_count_for_osr() is large in debug VM.
+  br(Assembler::lessUnsigned, false, Assembler::pn, profile_continue);
+  delayed()->nop();
 
   // Build it now.
   call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::profile_method));
   set_method_data_pointer_for_bcp();
-  ba_short(profile_continue);
+  ba(profile_continue);
+  delayed()->nop();
   bind(done);
 }
 
diff --git a/src/cpu/sparc/vm/methodHandles_sparc.cpp b/src/cpu/sparc/vm/methodHandles_sparc.cpp
index 3e6accae3..f8b5f24d5 100644
--- a/src/cpu/sparc/vm/methodHandles_sparc.cpp
+++ b/src/cpu/sparc/vm/methodHandles_sparc.cpp
@@ -121,6 +121,7 @@ void MethodHandles::verify_ref_kind(MacroAssembler* _masm, int ref_kind, Registe
 void MethodHandles::jump_from_method_handle(MacroAssembler* _masm, Register method, Register target, Register temp,
                                             bool for_compiler_entry) {
   assert(method == G5_method, "interpreter calling convention");
+  assert_different_registers(method, target, temp);
 
   if (!for_compiler_entry && JvmtiExport::can_post_interpreter_events()) {
     Label run_compiled_code;
@@ -153,19 +154,19 @@ void MethodHandles::jump_to_lambda_form(MacroAssembler* _masm,
   BLOCK_COMMENT("jump_to_lambda_form {");
   // This is the initial entry point of a lazy method handle.
   // After type checking, it picks up the invoker from the LambdaForm.
-  assert_different_registers(recv, method_temp, temp2, temp3);
+  assert_different_registers(recv, method_temp, temp2);  // temp3 is only passed on
   assert(method_temp == G5_method, "required register for loading method");
 
   //NOT_PRODUCT({ FlagSetting fs(TraceMethodHandles, true); trace_method_handle(_masm, "LZMH"); });
 
   // Load the invoker, as MH -> MH.form -> LF.vmentry
   __ verify_oop(recv);
-  __ load_heap_oop(Address(recv,        NONZERO(java_lang_invoke_MethodHandle::form_offset_in_bytes())),       method_temp);
+  __ load_heap_oop(Address(recv,        NONZERO(java_lang_invoke_MethodHandle::form_offset_in_bytes())),   method_temp);
   __ verify_oop(method_temp);
-  __ load_heap_oop(Address(method_temp, NONZERO(java_lang_invoke_LambdaForm::vmentry_offset_in_bytes())), method_temp);
+  __ load_heap_oop(Address(method_temp, NONZERO(java_lang_invoke_LambdaForm::vmentry_offset_in_bytes())),  method_temp);
   __ verify_oop(method_temp);
   // the following assumes that a Method* is normally compressed in the vmtarget field:
-  __ ld_ptr(Address(method_temp, NONZERO(java_lang_invoke_MemberName::vmtarget_offset_in_bytes())),     method_temp);
+  __ ld_ptr(       Address(method_temp, NONZERO(java_lang_invoke_MemberName::vmtarget_offset_in_bytes())), method_temp);
 
   if (VerifyMethodHandles && !for_compiler_entry) {
     // make sure recv is already on stack
@@ -303,25 +304,25 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
                                                     Register member_reg,
                                                     bool for_compiler_entry) {
   assert(is_signature_polymorphic(iid), "expected invoke iid");
-  // temps used in this code are not used in *either* compiled or interpreted calling sequences
   Register temp1 = (for_compiler_entry ? G1_scratch : O1);
-  Register temp2 = (for_compiler_entry ? G4_scratch : O4);
-  Register temp3 = G3_scratch;
-  Register temp4 = (for_compiler_entry ? noreg      : O2);
+  Register temp2 = (for_compiler_entry ? G3_scratch : O2);
+  Register temp3 = (for_compiler_entry ? G4_scratch : O3);
+  Register temp4 = (for_compiler_entry ? noreg      : O4);
   if (for_compiler_entry) {
     assert(receiver_reg == (iid == vmIntrinsics::_linkToStatic ? noreg : O0), "only valid assignment");
-    assert_different_registers(temp1,      O0, O1, O2, O3, O4, O5);
-    assert_different_registers(temp2,      O0, O1, O2, O3, O4, O5);
-    assert_different_registers(temp3,      O0, O1, O2, O3, O4, O5);
-    assert_different_registers(temp4,      O0, O1, O2, O3, O4, O5);
+    assert_different_registers(temp1, O0, O1, O2, O3, O4, O5);
+    assert_different_registers(temp2, O0, O1, O2, O3, O4, O5);
+    assert_different_registers(temp3, O0, O1, O2, O3, O4, O5);
+    assert_different_registers(temp4, O0, O1, O2, O3, O4, O5);
+  } else {
+    assert_different_registers(temp1, temp2, temp3, temp4, O5_savedSP);  // don't trash lastSP
   }
   if (receiver_reg != noreg)  assert_different_registers(temp1, temp2, temp3, temp4, receiver_reg);
   if (member_reg   != noreg)  assert_different_registers(temp1, temp2, temp3, temp4, member_reg);
-  if (!for_compiler_entry)    assert_different_registers(temp1, temp2, temp3, temp4, O5_savedSP);  // don't trash lastSP
 
   if (iid == vmIntrinsics::_invokeBasic) {
     // indirect through MH.form.vmentry.vmtarget
-    jump_to_lambda_form(_masm, receiver_reg, G5_method, temp2, temp3, for_compiler_entry);
+    jump_to_lambda_form(_masm, receiver_reg, G5_method, temp1, temp2, for_compiler_entry);
 
   } else {
     // The method is a member invoker used by direct method handles.
@@ -378,24 +379,22 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
     //  member_reg - MemberName that was the trailing argument
     //  temp1_recv_klass - klass of stacked receiver, if needed
     //  O5_savedSP - interpreter linkage (if interpreted)
-    //  O0..O7,G1,G4 - compiler arguments (if compiled)
+    //  O0..O5 - compiler arguments (if compiled)
 
-    bool method_is_live = false;
+    Label L_incompatible_class_change_error;
     switch (iid) {
     case vmIntrinsics::_linkToSpecial:
       if (VerifyMethodHandles) {
-        verify_ref_kind(_masm, JVM_REF_invokeSpecial, member_reg, temp3);
+        verify_ref_kind(_masm, JVM_REF_invokeSpecial, member_reg, temp2);
       }
       __ ld_ptr(member_vmtarget, G5_method);
-      method_is_live = true;
       break;
 
     case vmIntrinsics::_linkToStatic:
       if (VerifyMethodHandles) {
-        verify_ref_kind(_masm, JVM_REF_invokeStatic, member_reg, temp3);
+        verify_ref_kind(_masm, JVM_REF_invokeStatic, member_reg, temp2);
       }
       __ ld_ptr(member_vmtarget, G5_method);
-      method_is_live = true;
       break;
 
     case vmIntrinsics::_linkToVirtual:
@@ -404,7 +403,7 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
       // minus the CP setup and profiling:
 
       if (VerifyMethodHandles) {
-        verify_ref_kind(_masm, JVM_REF_invokeVirtual, member_reg, temp3);
+        verify_ref_kind(_masm, JVM_REF_invokeVirtual, member_reg, temp2);
       }
 
       // pick out the vtable index from the MemberName, and then we can discard it:
@@ -423,7 +422,6 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
 
       // get target Method* & entry point
       __ lookup_virtual_method(temp1_recv_klass, temp2_index, G5_method);
-      method_is_live = true;
       break;
     }
 
@@ -432,13 +430,13 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
       // same as TemplateTable::invokeinterface
       // (minus the CP setup and profiling, with different argument motion)
       if (VerifyMethodHandles) {
-        verify_ref_kind(_masm, JVM_REF_invokeInterface, member_reg, temp3);
+        verify_ref_kind(_masm, JVM_REF_invokeInterface, member_reg, temp2);
       }
 
-      Register temp3_intf = temp3;
-      __ load_heap_oop(member_clazz, temp3_intf);
-      load_klass_from_Class(_masm, temp3_intf, temp2, temp4);
-      __ verify_klass_ptr(temp3_intf);
+      Register temp2_intf = temp2;
+      __ load_heap_oop(member_clazz, temp2_intf);
+      load_klass_from_Class(_masm, temp2_intf, temp3, temp4);
+      __ verify_klass_ptr(temp2_intf);
 
       Register G5_index = G5_method;
       __ ld_ptr(member_vmindex, G5_index);
@@ -450,37 +448,34 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
       }
 
       // given intf, index, and recv klass, dispatch to the implementation method
-      Label L_no_such_interface;
-      Register no_sethi_temp = noreg;
-      __ lookup_interface_method(temp1_recv_klass, temp3_intf,
+      __ lookup_interface_method(temp1_recv_klass, temp2_intf,
                                  // note: next two args must be the same:
                                  G5_index, G5_method,
-                                 temp2, no_sethi_temp,
-                                 L_no_such_interface);
-
-      __ verify_method_ptr(G5_method);
-      jump_from_method_handle(_masm, G5_method, temp2, temp3, for_compiler_entry);
-
-      __ bind(L_no_such_interface);
-      AddressLiteral icce(StubRoutines::throw_IncompatibleClassChangeError_entry());
-      __ jump_to(icce, temp3);
-      __ delayed()->nop();
+                                 temp3, temp4,
+                                 L_incompatible_class_change_error);
       break;
     }
 
     default:
-      fatal(err_msg("unexpected intrinsic %d: %s", iid, vmIntrinsics::name_at(iid)));
+      fatal(err_msg_res("unexpected intrinsic %d: %s", iid, vmIntrinsics::name_at(iid)));
       break;
     }
 
-    if (method_is_live) {
-      // live at this point:  G5_method, O5_savedSP (if interpreted)
+    // Live at this point:
+    //   G5_method
+    //   O5_savedSP (if interpreted)
+
+    // After figuring out which concrete method to call, jump into it.
+    // Note that this works in the interpreter with no data motion.
+    // But the compiled version will require that rcx_recv be shifted out.
+    __ verify_method_ptr(G5_method);
+    jump_from_method_handle(_masm, G5_method, temp1, temp2, for_compiler_entry);
 
-      // After figuring out which concrete method to call, jump into it.
-      // Note that this works in the interpreter with no data motion.
-      // But the compiled version will require that rcx_recv be shifted out.
-      __ verify_method_ptr(G5_method);
-      jump_from_method_handle(_masm, G5_method, temp1, temp3, for_compiler_entry);
+    if (iid == vmIntrinsics::_linkToInterface) {
+      __ BIND(L_incompatible_class_change_error);
+      AddressLiteral icce(StubRoutines::throw_IncompatibleClassChangeError_entry());
+      __ jump_to(icce, temp1);
+      __ delayed()->nop();
     }
   }
 }
diff --git a/src/cpu/sparc/vm/sharedRuntime_sparc.cpp b/src/cpu/sparc/vm/sharedRuntime_sparc.cpp
index 2968bc8df..bc35d196d 100644
--- a/src/cpu/sparc/vm/sharedRuntime_sparc.cpp
+++ b/src/cpu/sparc/vm/sharedRuntime_sparc.cpp
@@ -313,6 +313,14 @@ void RegisterSaver::restore_result_registers(MacroAssembler* masm) {
 
 }
 
+// Is vector's size (in bytes) bigger than a size saved by default?
+// 8 bytes FP registers are saved by default on SPARC.
+bool SharedRuntime::is_wide_vector(int size) {
+  // Note, MaxVectorSize == 8 on SPARC.
+  assert(size <= 8, err_msg_res("%d bytes vectors are not supported", size));
+  return size > 8;
+}
+
 // The java_calling_convention describes stack locations as ideal slots on
 // a frame with no abi restrictions. Since we must observe abi restrictions
 // (like the placement of the register window) the slots must be biased by
@@ -364,9 +372,9 @@ static VMRegPair reg64_to_VMRegPair(Register r) {
 // ---------------------------------------------------------------------------
 // The compiled Java calling convention.  The Java convention always passes
 // 64-bit values in adjacent aligned locations (either registers or stack),
-// floats in float registers and doubles in aligned float pairs.  Values are
-// packed in the registers.  There is no backing varargs store for values in
-// registers.  In the 32-bit build, longs are passed in G1 and G4 (cannot be
+// floats in float registers and doubles in aligned float pairs.  There is
+// no backing varargs store for values in registers.
+// In the 32-bit build, longs are passed on the stack (cannot be
 // passed in I's, because longs in I's get their heads chopped off at
 // interrupt).
 int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
@@ -375,76 +383,13 @@ int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
                                            int is_outgoing) {
   assert(F31->as_VMReg()->is_reg(), "overlapping stack/register numbers");
 
-  // Convention is to pack the first 6 int/oop args into the first 6 registers
-  // (I0-I5), extras spill to the stack.  Then pack the first 8 float args
-  // into F0-F7, extras spill to the stack.  Then pad all register sets to
-  // align.  Then put longs and doubles into the same registers as they fit,
-  // else spill to the stack.
   const int int_reg_max = SPARC_ARGS_IN_REGS_NUM;
   const int flt_reg_max = 8;
-  //
-  // Where 32-bit 1-reg longs start being passed
-  // In tiered we must pass on stack because c1 can't use a "pair" in a single reg.
-  // So make it look like we've filled all the G regs that c2 wants to use.
-  Register g_reg = TieredCompilation ? noreg : G1;
-
-  // Count int/oop and float args.  See how many stack slots we'll need and
-  // where the longs & doubles will go.
-  int int_reg_cnt   = 0;
-  int flt_reg_cnt   = 0;
-  // int stk_reg_pairs = frame::register_save_words*(wordSize>>2);
-  // int stk_reg_pairs = SharedRuntime::out_preserve_stack_slots();
-  int stk_reg_pairs = 0;
-  for (int i = 0; i < total_args_passed; i++) {
-    switch (sig_bt[i]) {
-    case T_LONG:                // LP64, longs compete with int args
-      assert(sig_bt[i+1] == T_VOID, "");
-#ifdef _LP64
-      if (int_reg_cnt < int_reg_max)  int_reg_cnt++;
-#endif
-      break;
-    case T_OBJECT:
-    case T_ARRAY:
-    case T_ADDRESS: // Used, e.g., in slow-path locking for the lock's stack address
-      if (int_reg_cnt < int_reg_max)  int_reg_cnt++;
-#ifndef _LP64
-      else                            stk_reg_pairs++;
-#endif
-      break;
-    case T_INT:
-    case T_SHORT:
-    case T_CHAR:
-    case T_BYTE:
-    case T_BOOLEAN:
-      if (int_reg_cnt < int_reg_max)  int_reg_cnt++;
-      else                            stk_reg_pairs++;
-      break;
-    case T_FLOAT:
-      if (flt_reg_cnt < flt_reg_max)  flt_reg_cnt++;
-      else                            stk_reg_pairs++;
-      break;
-    case T_DOUBLE:
-      assert(sig_bt[i+1] == T_VOID, "");
-      break;
-    case T_VOID:
-      break;
-    default:
-      ShouldNotReachHere();
-    }
-  }
-
-  // This is where the longs/doubles start on the stack.
-  stk_reg_pairs = (stk_reg_pairs+1) & ~1; // Round
 
-  int flt_reg_pairs = (flt_reg_cnt+1) & ~1;
-
-  // int stk_reg = frame::register_save_words*(wordSize>>2);
-  // int stk_reg = SharedRuntime::out_preserve_stack_slots();
-  int stk_reg = 0;
   int int_reg = 0;
   int flt_reg = 0;
+  int slot = 0;
 
-  // Now do the signature layout
   for (int i = 0; i < total_args_passed; i++) {
     switch (sig_bt[i]) {
     case T_INT:
@@ -461,11 +406,14 @@ int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
         Register r = is_outgoing ? as_oRegister(int_reg++) : as_iRegister(int_reg++);
         regs[i].set1(r->as_VMReg());
       } else {
-        regs[i].set1(VMRegImpl::stack2reg(stk_reg++));
+        regs[i].set1(VMRegImpl::stack2reg(slot++));
       }
       break;
 
 #ifdef _LP64
+    case T_LONG:
+      assert(sig_bt[i+1] == T_VOID, "expecting VOID in other half");
+      // fall-through
     case T_OBJECT:
     case T_ARRAY:
     case T_ADDRESS: // Used, e.g., in slow-path locking for the lock's stack address
@@ -473,78 +421,57 @@ int SharedRuntime::java_calling_convention(const BasicType *sig_bt,
         Register r = is_outgoing ? as_oRegister(int_reg++) : as_iRegister(int_reg++);
         regs[i].set2(r->as_VMReg());
       } else {
-        regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
-        stk_reg_pairs += 2;
+        slot = round_to(slot, 2);  // align
+        regs[i].set2(VMRegImpl::stack2reg(slot));
+        slot += 2;
       }
       break;
-#endif // _LP64
-
+#else
     case T_LONG:
       assert(sig_bt[i+1] == T_VOID, "expecting VOID in other half");
-#ifdef _LP64
-        if (int_reg < int_reg_max) {
-          Register r = is_outgoing ? as_oRegister(int_reg++) : as_iRegister(int_reg++);
-          regs[i].set2(r->as_VMReg());
-        } else {
-          regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
-          stk_reg_pairs += 2;
-        }
-#else
-#ifdef COMPILER2
-        // For 32-bit build, can't pass longs in O-regs because they become
-        // I-regs and get trashed.  Use G-regs instead.  G1 and G4 are almost
-        // spare and available.  This convention isn't used by the Sparc ABI or
-        // anywhere else. If we're tiered then we don't use G-regs because c1
-        // can't deal with them as a "pair". (Tiered makes this code think g's are filled)
-        // G0: zero
-        // G1: 1st Long arg
-        // G2: global allocated to TLS
-        // G3: used in inline cache check
-        // G4: 2nd Long arg
-        // G5: used in inline cache check
-        // G6: used by OS
-        // G7: used by OS
-
-        if (g_reg == G1) {
-          regs[i].set2(G1->as_VMReg()); // This long arg in G1
-          g_reg = G4;                  // Where the next arg goes
-        } else if (g_reg == G4) {
-          regs[i].set2(G4->as_VMReg()); // The 2nd long arg in G4
-          g_reg = noreg;               // No more longs in registers
-        } else {
-          regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
-          stk_reg_pairs += 2;
-        }
-#else // COMPILER2
-          regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
-          stk_reg_pairs += 2;
-#endif // COMPILER2
-#endif // _LP64
+      // On 32-bit SPARC put longs always on the stack to keep the pressure off
+      // integer argument registers.  They should be used for oops.
+      slot = round_to(slot, 2);  // align
+      regs[i].set2(VMRegImpl::stack2reg(slot));
+      slot += 2;
+#endif
       break;
 
     case T_FLOAT:
-      if (flt_reg < flt_reg_max) regs[i].set1(as_FloatRegister(flt_reg++)->as_VMReg());
-      else                       regs[i].set1(VMRegImpl::stack2reg(stk_reg++));
+      if (flt_reg < flt_reg_max) {
+        FloatRegister r = as_FloatRegister(flt_reg++);
+        regs[i].set1(r->as_VMReg());
+      } else {
+        regs[i].set1(VMRegImpl::stack2reg(slot++));
+      }
       break;
+
     case T_DOUBLE:
       assert(sig_bt[i+1] == T_VOID, "expecting half");
-      if (flt_reg_pairs + 1 < flt_reg_max) {
-        regs[i].set2(as_FloatRegister(flt_reg_pairs)->as_VMReg());
-        flt_reg_pairs += 2;
+      if (round_to(flt_reg, 2) + 1 < flt_reg_max) {
+        flt_reg = round_to(flt_reg, 2);  // align
+        FloatRegister r = as_FloatRegister(flt_reg);
+        regs[i].set2(r->as_VMReg());
+        flt_reg += 2;
       } else {
-        regs[i].set2(VMRegImpl::stack2reg(stk_reg_pairs));
-        stk_reg_pairs += 2;
+        slot = round_to(slot, 2);  // align
+        regs[i].set2(VMRegImpl::stack2reg(slot));
+        slot += 2;
       }
       break;
-    case T_VOID: regs[i].set_bad();  break; // Halves of longs & doubles
+
+    case T_VOID:
+      regs[i].set_bad();   // Halves of longs & doubles
+      break;
+
     default:
-      ShouldNotReachHere();
+      fatal(err_msg_res("unknown basic type %d", sig_bt[i]));
+      break;
     }
   }
 
   // retun the amount of stack space these arguments will need.
-  return stk_reg_pairs;
-
+  return slot;
 }
 
 // Helper class mostly to avoid passing masm everywhere, and handle
@@ -601,8 +528,7 @@ void AdapterGenerator::patch_callers_callsite() {
   Label L;
   __ ld_ptr(G5_method, in_bytes(Method::code_offset()), G3_scratch);
   __ br_null(G3_scratch, false, Assembler::pt, L);
-  // Schedule the branch target address early.
-  __ delayed()->ld_ptr(G5_method, in_bytes(Method::interpreter_entry_offset()), G3_scratch);
+  __ delayed()->nop();
   // Call into the VM to patch the caller, then jump to compiled callee
   __ save_frame(4);     // Args in compiled layout; do not blow them
 
@@ -645,7 +571,6 @@ void AdapterGenerator::patch_callers_callsite() {
   __ ldx(FP, -8 + STACK_BIAS, G1);
   __ ldx(FP, -16 + STACK_BIAS, G4);
   __ mov(L5, G5_method);
-  __ ld_ptr(G5_method, in_bytes(Method::interpreter_entry_offset()), G3_scratch);
 #endif /* _LP64 */
 
   __ restore();      // Restore args
@@ -726,7 +651,7 @@ void AdapterGenerator::gen_c2i_adapter(
                             int comp_args_on_stack, // VMRegStackSlots
                             const BasicType *sig_bt,
                             const VMRegPair *regs,
-                            Label& skip_fixup) {
+                            Label& L_skip_fixup) {
 
   // Before we get into the guts of the C2I adapter, see if we should be here
   // at all.  We've come from compiled code and are attempting to jump to the
@@ -747,7 +672,7 @@ void AdapterGenerator::gen_c2i_adapter(
 
   patch_callers_callsite();
 
-  __ bind(skip_fixup);
+  __ bind(L_skip_fixup);
 
   // Since all args are passed on the stack, total_args_passed*wordSize is the
   // space we need.  Add in varargs area needed by the interpreter. Round up
@@ -757,46 +682,18 @@ void AdapterGenerator::gen_c2i_adapter(
                  (frame::varargs_offset - frame::register_save_words)*wordSize;
   const int extraspace = round_to(arg_size + varargs_area, 2*wordSize);
 
-  int bias = STACK_BIAS;
+  const int bias = STACK_BIAS;
   const int interp_arg_offset = frame::varargs_offset*wordSize +
                         (total_args_passed-1)*Interpreter::stackElementSize;
 
-  Register base = SP;
+  const Register base = SP;
 
-#ifdef _LP64
-  // In the 64bit build because of wider slots and STACKBIAS we can run
-  // out of bits in the displacement to do loads and stores.  Use g3 as
-  // temporary displacement.
-  if (!Assembler::is_simm13(extraspace)) {
-    __ set(extraspace, G3_scratch);
-    __ sub(SP, G3_scratch, SP);
-  } else {
-    __ sub(SP, extraspace, SP);
-  }
+  // Make some extra space on the stack.
+  __ sub(SP, __ ensure_simm13_or_reg(extraspace, G3_scratch), SP);
   set_Rdisp(G3_scratch);
-#else
-  __ sub(SP, extraspace, SP);
-#endif // _LP64
 
-  // First write G1 (if used) to where ever it must go
-  for (int i=0; i<total_args_passed; i++) {
-    const int st_off = interp_arg_offset - (i*Interpreter::stackElementSize) + bias;
-    VMReg r_1 = regs[i].first();
-    VMReg r_2 = regs[i].second();
-    if (r_1 == G1_scratch->as_VMReg()) {
-      if (sig_bt[i] == T_OBJECT || sig_bt[i] == T_ARRAY) {
-        store_c2i_object(G1_scratch, base, st_off);
-      } else if (sig_bt[i] == T_LONG) {
-        assert(!TieredCompilation, "should not use register args for longs");
-        store_c2i_long(G1_scratch, base, st_off, false);
-      } else {
-        store_c2i_int(G1_scratch, base, st_off);
-      }
-    }
-  }
-
-  // Now write the args into the outgoing interpreter space
-  for (int i=0; i<total_args_passed; i++) {
+  // Write the args into the outgoing interpreter space.
+  for (int i = 0; i < total_args_passed; i++) {
     const int st_off = interp_arg_offset - (i*Interpreter::stackElementSize) + bias;
     VMReg r_1 = regs[i].first();
     VMReg r_2 = regs[i].second();
@@ -804,23 +701,9 @@ void AdapterGenerator::gen_c2i_adapter(
       assert(!r_2->is_valid(), "");
       continue;
     }
-    // Skip G1 if found as we did it first in order to free it up
-    if (r_1 == G1_scratch->as_VMReg()) {
-      continue;
-    }
-#ifdef ASSERT
-    bool G1_forced = false;
-#endif // ASSERT
     if (r_1->is_stack()) {        // Pretend stack targets are loaded into G1
-#ifdef _LP64
-      Register ld_off = Rdisp;
-      __ set(reg2offset(r_1) + extraspace + bias, ld_off);
-#else
-      int ld_off = reg2offset(r_1) + extraspace + bias;
-#endif // _LP64
-#ifdef ASSERT
-      G1_forced = true;
-#endif // ASSERT
+      RegisterOrConstant ld_off = reg2offset(r_1) + extraspace + bias;
+      ld_off = __ ensure_simm13_or_reg(ld_off, Rdisp);
       r_1 = G1_scratch->as_VMReg();// as part of the load/store shuffle
       if (!r_2->is_valid()) __ ld (base, ld_off, G1_scratch);
       else                  __ ldx(base, ld_off, G1_scratch);
@@ -831,11 +714,6 @@ void AdapterGenerator::gen_c2i_adapter(
       if (sig_bt[i] == T_OBJECT || sig_bt[i] == T_ARRAY) {
         store_c2i_object(r, base, st_off);
       } else if (sig_bt[i] == T_LONG || sig_bt[i] == T_DOUBLE) {
-#ifndef _LP64
-        if (TieredCompilation) {
-          assert(G1_forced || sig_bt[i] != T_LONG, "should not use register args for longs");
-        }
-#endif // _LP64
         store_c2i_long(r, base, st_off, r_2->is_stack());
       } else {
         store_c2i_int(r, base, st_off);
@@ -851,19 +729,12 @@ void AdapterGenerator::gen_c2i_adapter(
     }
   }
 
-#ifdef _LP64
-  // Need to reload G3_scratch, used for temporary displacements.
+  // Load the interpreter entry point.
   __ ld_ptr(G5_method, in_bytes(Method::interpreter_entry_offset()), G3_scratch);
 
   // Pass O5_savedSP as an argument to the interpreter.
   // The interpreter will restore SP to this value before returning.
-  __ set(extraspace, G1);
-  __ add(SP, G1, O5_savedSP);
-#else
-  // Pass O5_savedSP as an argument to the interpreter.
-  // The interpreter will restore SP to this value before returning.
-  __ add(SP, extraspace, O5_savedSP);
-#endif // _LP64
+  __ add(SP, __ ensure_simm13_or_reg(extraspace, G1), O5_savedSP);
 
   __ mov((frame::varargs_offset)*wordSize -
          1*Interpreter::stackElementSize+bias+BytesPerWord, G1);
@@ -971,7 +842,6 @@ void AdapterGenerator::gen_i2c_adapter(
 
   // Outputs:
   // G2_thread      - TLS
-  // G1, G4         - Outgoing long args in 32-bit build
   // O0-O5          - Outgoing args in compiled layout
   // O6             - Adjusted or restored SP
   // O7             - Valid return address
@@ -1016,10 +886,10 @@ void AdapterGenerator::gen_i2c_adapter(
   // +--------------+ <--- start of outgoing args
   // |  pad, align  |   |
   // +--------------+   |
-  // | ints, floats |   |---Outgoing stack args, packed low.
-  // +--------------+   |   First few args in registers.
-  // :   doubles    :   |
-  // |   longs      |   |
+  // | ints, longs, |   |
+  // |    floats,   |   |---Outgoing stack args.
+  // :    doubles   :   |   First few args in registers.
+  // |              |   |
   // +--------------+ <--- SP' + 16*wordsize
   // |              |
   // :    window    :
@@ -1033,7 +903,6 @@ void AdapterGenerator::gen_i2c_adapter(
   // Cut-out for having no stack args.  Since up to 6 args are passed
   // in registers, we will commonly have no stack args.
   if (comp_args_on_stack > 0) {
-
     // Convert VMReg stack slots to words.
     int comp_words_on_stack = round_to(comp_args_on_stack*VMRegImpl::stack_slot_size, wordSize)>>LogBytesPerWord;
     // Round up to miminum stack alignment, in wordSize
@@ -1044,13 +913,9 @@ void AdapterGenerator::gen_i2c_adapter(
     __ sub(SP, (comp_words_on_stack)*wordSize, SP);
   }
 
-  // Will jump to the compiled code just as if compiled code was doing it.
-  // Pre-load the register-jump target early, to schedule it better.
-  __ ld_ptr(G5_method, in_bytes(Method::from_compiled_offset()), G3);
-
   // Now generate the shuffle code.  Pick up all register args and move the
   // rest through G1_scratch.
-  for (int i=0; i<total_args_passed; i++) {
+  for (int i = 0; i < total_args_passed; i++) {
     if (sig_bt[i] == T_VOID) {
       // Longs and doubles are passed in native word order, but misaligned
       // in the 32-bit build.
@@ -1088,14 +953,13 @@ void AdapterGenerator::gen_i2c_adapter(
               next_arg_slot(ld_off) : arg_slot(ld_off);
         __ ldx(Gargs, slot, r);
 #else
-        // Need to load a 64-bit value into G1/G4, but G1/G4 is being used in the
-        // stack shuffle.  Load the first 2 longs into G1/G4 later.
+        fatal("longs should be on stack");
 #endif
       }
     } else {
       assert(r_1->is_FloatRegister(), "");
       if (!r_2->is_valid()) {
-        __ ldf(FloatRegisterImpl::S, Gargs, arg_slot(ld_off), r_1->as_FloatRegister());
+        __ ldf(FloatRegisterImpl::S, Gargs,      arg_slot(ld_off), r_1->as_FloatRegister());
       } else {
 #ifdef _LP64
         // In V9, doubles are given 2 64-bit slots in the interpreter, but the
@@ -1104,11 +968,11 @@ void AdapterGenerator::gen_i2c_adapter(
         // spare float register.
         RegisterOrConstant slot = (sig_bt[i] == T_LONG || sig_bt[i] == T_DOUBLE) ?
               next_arg_slot(ld_off) : arg_slot(ld_off);
-        __ ldf(FloatRegisterImpl::D, Gargs, slot, r_1->as_FloatRegister());
+        __ ldf(FloatRegisterImpl::D, Gargs,                  slot, r_1->as_FloatRegister());
 #else
         // Need to marshal 64-bit value from misaligned Lesp loads
         __ ldf(FloatRegisterImpl::S, Gargs, next_arg_slot(ld_off), r_1->as_FloatRegister());
-        __ ldf(FloatRegisterImpl::S, Gargs, arg_slot(ld_off), r_2->as_FloatRegister());
+        __ ldf(FloatRegisterImpl::S, Gargs,      arg_slot(ld_off), r_2->as_FloatRegister());
 #endif
       }
     }
@@ -1124,76 +988,35 @@ void AdapterGenerator::gen_i2c_adapter(
       else                  __ stf(FloatRegisterImpl::D, r_1->as_FloatRegister(), SP, slot);
     }
   }
-  bool made_space = false;
-#ifndef _LP64
-  // May need to pick up a few long args in G1/G4
-  bool g4_crushed = false;
-  bool g3_crushed = false;
-  for (int i=0; i<total_args_passed; i++) {
-    if (regs[i].first()->is_Register() && regs[i].second()->is_valid()) {
-      // Load in argument order going down
-      int ld_off = (total_args_passed-i)*Interpreter::stackElementSize;
-      // Need to marshal 64-bit value from misaligned Lesp loads
-      Register r = regs[i].first()->as_Register()->after_restore();
-      if (r == G1 || r == G4) {
-        assert(!g4_crushed, "ordering problem");
-        if (r == G4){
-          g4_crushed = true;
-          __ lduw(Gargs, arg_slot(ld_off)     , G3_scratch); // Load lo bits
-          __ ld  (Gargs, next_arg_slot(ld_off), r);          // Load hi bits
-        } else {
-          // better schedule this way
-          __ ld  (Gargs, next_arg_slot(ld_off), r);          // Load hi bits
-          __ lduw(Gargs, arg_slot(ld_off)     , G3_scratch); // Load lo bits
-        }
-        g3_crushed = true;
-        __ sllx(r, 32, r);
-        __ or3(G3_scratch, r, r);
-      } else {
-        assert(r->is_out(), "longs passed in two O registers");
-        __ ld  (Gargs, arg_slot(ld_off)     , r->successor()); // Load lo bits
-        __ ld  (Gargs, next_arg_slot(ld_off), r);              // Load hi bits
-      }
-    }
-  }
-#endif
 
   // Jump to the compiled code just as if compiled code was doing it.
-  //
-#ifndef _LP64
-    if (g3_crushed) {
-      // Rats load was wasted, at least it is in cache...
-      __ ld_ptr(G5_method, Method::from_compiled_offset(), G3);
-    }
-#endif /* _LP64 */
-
-    // 6243940 We might end up in handle_wrong_method if
-    // the callee is deoptimized as we race thru here. If that
-    // happens we don't want to take a safepoint because the
-    // caller frame will look interpreted and arguments are now
-    // "compiled" so it is much better to make this transition
-    // invisible to the stack walking code. Unfortunately if
-    // we try and find the callee by normal means a safepoint
-    // is possible. So we stash the desired callee in the thread
-    // and the vm will find there should this case occur.
-    Address callee_target_addr(G2_thread, JavaThread::callee_target_offset());
-    __ st_ptr(G5_method, callee_target_addr);
-
-    if (StressNonEntrant) {
-      // Open a big window for deopt failure
-      __ save_frame(0);
-      __ mov(G0, L0);
-      Label loop;
-      __ bind(loop);
-      __ sub(L0, 1, L0);
-      __ br_null_short(L0, Assembler::pt, loop);
-
-      __ restore();
-    }
+  __ ld_ptr(G5_method, in_bytes(Method::from_compiled_offset()), G3);
 
+  // 6243940 We might end up in handle_wrong_method if
+  // the callee is deoptimized as we race thru here. If that
+  // happens we don't want to take a safepoint because the
+  // caller frame will look interpreted and arguments are now
+  // "compiled" so it is much better to make this transition
+  // invisible to the stack walking code. Unfortunately if
+  // we try and find the callee by normal means a safepoint
+  // is possible. So we stash the desired callee in the thread
+  // and the vm will find there should this case occur.
+  Address callee_target_addr(G2_thread, JavaThread::callee_target_offset());
+  __ st_ptr(G5_method, callee_target_addr);
+
+  if (StressNonEntrant) {
+    // Open a big window for deopt failure
+    __ save_frame(0);
+    __ mov(G0, L0);
+    Label loop;
+    __ bind(loop);
+    __ sub(L0, 1, L0);
+    __ br_null_short(L0, Assembler::pt, loop);
+    __ restore();
+  }
 
-    __ jmpl(G3, 0, G0);
-    __ delayed()->nop();
+  __ jmpl(G3, 0, G0);
+  __ delayed()->nop();
 }
 
 // ---------------------------------------------------------------
@@ -1221,28 +1044,17 @@ AdapterHandlerEntry* SharedRuntime::generate_i2c2i_adapters(MacroAssembler *masm
   // compiled code, which relys solely on SP and not FP, get sick).
 
   address c2i_unverified_entry = __ pc();
-  Label skip_fixup;
+  Label L_skip_fixup;
   {
-#if !defined(_LP64) && defined(COMPILER2)
-    Register R_temp   = L0;   // another scratch register
-#else
-    Register R_temp   = G1;   // another scratch register
-#endif
+    Register R_temp = G1;  // another scratch register
 
     AddressLiteral ic_miss(SharedRuntime::get_ic_miss_stub());
 
     __ verify_oop(O0);
     __ load_klass(O0, G3_scratch);
 
-#if !defined(_LP64) && defined(COMPILER2)
-    __ save(SP, -frame::register_save_words*wordSize, SP);
-    __ ld_ptr(G5_method, CompiledICHolder::holder_klass_offset(), R_temp);
-    __ cmp(G3_scratch, R_temp);
-    __ restore();
-#else
     __ ld_ptr(G5_method, CompiledICHolder::holder_klass_offset(), R_temp);
     __ cmp(G3_scratch, R_temp);
-#endif
 
     Label ok, ok2;
     __ brx(Assembler::equal, false, Assembler::pt, ok);
@@ -1256,8 +1068,8 @@ AdapterHandlerEntry* SharedRuntime::generate_i2c2i_adapters(MacroAssembler *masm
     // the call site corrected.
     __ ld_ptr(G5_method, in_bytes(Method::code_offset()), G3_scratch);
     __ bind(ok2);
-    __ br_null(G3_scratch, false, Assembler::pt, skip_fixup);
-    __ delayed()->ld_ptr(G5_method, in_bytes(Method::interpreter_entry_offset()), G3_scratch);
+    __ br_null(G3_scratch, false, Assembler::pt, L_skip_fixup);
+    __ delayed()->nop();
     __ jump_to(ic_miss, G3_scratch);
     __ delayed()->nop();
 
@@ -1265,7 +1077,7 @@ AdapterHandlerEntry* SharedRuntime::generate_i2c2i_adapters(MacroAssembler *masm
 
   address c2i_entry = __ pc();
 
-  agen.gen_c2i_adapter(total_args_passed, comp_args_on_stack, sig_bt, regs, skip_fixup);
+  agen.gen_c2i_adapter(total_args_passed, comp_args_on_stack, sig_bt, regs, L_skip_fixup);
 
   __ flush();
   return AdapterHandlerLibrary::new_entry(fingerprint, i2c_entry, c2i_entry, c2i_unverified_entry);
@@ -1985,12 +1797,12 @@ static void unpack_array_argument(MacroAssembler* masm, VMRegPair reg, BasicType
 }
 
 static void verify_oop_args(MacroAssembler* masm,
-                            int total_args_passed,
+                            methodHandle method,
                             const BasicType* sig_bt,
                             const VMRegPair* regs) {
   Register temp_reg = G5_method;  // not part of any compiled calling seq
   if (VerifyOops) {
-    for (int i = 0; i < total_args_passed; i++) {
+    for (int i = 0; i < method->size_of_parameters(); i++) {
       if (sig_bt[i] == T_OBJECT ||
           sig_bt[i] == T_ARRAY) {
         VMReg r = regs[i].first();
@@ -2009,35 +1821,32 @@ static void verify_oop_args(MacroAssembler* masm,
 }
 
 static void gen_special_dispatch(MacroAssembler* masm,
-                                 int total_args_passed,
-                                 int comp_args_on_stack,
-                                 vmIntrinsics::ID special_dispatch,
+                                 methodHandle method,
                                  const BasicType* sig_bt,
                                  const VMRegPair* regs) {
-  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+  verify_oop_args(masm, method, sig_bt, regs);
+  vmIntrinsics::ID iid = method->intrinsic_id();
 
   // Now write the args into the outgoing interpreter space
   bool     has_receiver   = false;
   Register receiver_reg   = noreg;
   int      member_arg_pos = -1;
   Register member_reg     = noreg;
-  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(iid);
   if (ref_kind != 0) {
-    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_arg_pos = method->size_of_parameters() - 1;  // trailing MemberName argument
     member_reg = G5_method;  // known to be free at this point
     has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
-  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+  } else if (iid == vmIntrinsics::_invokeBasic) {
     has_receiver = true;
   } else {
-    fatal(err_msg("special_dispatch=%d", special_dispatch));
+    fatal(err_msg_res("unexpected intrinsic id %d", iid));
   }
 
   if (member_reg != noreg) {
     // Load the member_arg into register, if necessary.
-    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
-    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    SharedRuntime::check_member_name_argument_is_last_argument(method, sig_bt, regs);
     VMReg r = regs[member_arg_pos].first();
-    assert(r->is_valid(), "bad member arg");
     if (r->is_stack()) {
       RegisterOrConstant ld_off = reg2offset(r) + STACK_BIAS;
       ld_off = __ ensure_simm13_or_reg(ld_off, member_reg);
@@ -2050,7 +1859,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
 
   if (has_receiver) {
     // Make sure the receiver is loaded into a register.
-    assert(total_args_passed > 0, "oob");
+    assert(method->size_of_parameters() > 0, "oob");
     assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
     VMReg r = regs[0].first();
     assert(r->is_valid(), "bad receiver arg");
@@ -2058,7 +1867,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
       // Porting note:  This assumes that compiled calling conventions always
       // pass the receiver oop in a register.  If this is not true on some
       // platform, pick a temp and load the receiver from stack.
-      assert(false, "receiver always in a register");
+      fatal("receiver always in a register");
       receiver_reg = G3_scratch;  // known to be free at this point
       RegisterOrConstant ld_off = reg2offset(r) + STACK_BIAS;
       ld_off = __ ensure_simm13_or_reg(ld_off, member_reg);
@@ -2070,7 +1879,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
   }
 
   // Figure out which address we are really jumping to:
-  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+  MethodHandles::generate_method_handle_dispatch(masm, iid,
                                                  receiver_reg, member_reg, /*for_compiler_entry:*/ true);
 }
 
@@ -2103,11 +1912,9 @@ static void gen_special_dispatch(MacroAssembler* masm,
 //    transition back to thread_in_Java
 //    return to caller
 //
-nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
+nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 methodHandle method,
                                                 int compile_id,
-                                                int total_in_args,
-                                                int comp_args_on_stack, // in VMRegStackSlots
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
@@ -2116,9 +1923,7 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
     intptr_t start = (intptr_t)__ pc();
     int vep_offset = ((intptr_t)__ pc()) - start;
     gen_special_dispatch(masm,
-                         total_in_args,
-                         comp_args_on_stack,
-                         method->intrinsic_id(),
+                         method,
                          in_sig_bt,
                          in_regs);
     int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
@@ -2220,6 +2025,7 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
   // we convert the java signature to a C signature by inserting
   // the hidden arguments as arg[0] and possibly arg[1] (static method)
 
+  const int total_in_args = method->size_of_parameters();
   int total_c_args = total_in_args;
   int total_save_slots = 6 * VMRegImpl::slots_per_word;
   if (!is_critical_native) {
@@ -3936,7 +3742,7 @@ void SharedRuntime::generate_uncommon_trap_blob() {
 // the 64-bit %o's, then do a save, then fixup the caller's SP (our FP).
 // Tricky, tricky, tricky...
 
-SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause_return) {
+SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, int poll_type) {
   assert (StubRoutines::forward_exception_entry() != NULL, "must be generated before");
 
   // allocate space for the code
@@ -3954,6 +3760,7 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
 
   int start = __ offset();
 
+  bool cause_return = (poll_type == POLL_AT_RETURN);
   // If this causes a return before the processing, then do a "restore"
   if (cause_return) {
     __ restore();
diff --git a/src/cpu/sparc/vm/sparc.ad b/src/cpu/sparc/vm/sparc.ad
index 42b4cf81b..f7040f54e 100644
--- a/src/cpu/sparc/vm/sparc.ad
+++ b/src/cpu/sparc/vm/sparc.ad
@@ -1838,6 +1838,12 @@ const bool Matcher::match_rule_supported(int opcode) {
   case Op_PopCountL:
     if (!UsePopCountInstruction)
       return false;
+  case Op_CompareAndSwapL:
+#ifdef _LP64
+  case Op_CompareAndSwapP:
+#endif
+    if (!VM_Version::supports_cx8())
+      return false;
     break;
   }
 
@@ -7199,6 +7205,7 @@ instruct storeLConditional( iRegP mem_ptr, iRegL oldval, g3RegL newval, flagsReg
 // No flag versions for CompareAndSwap{P,I,L} because matcher can't match them
 
 instruct compareAndSwapL_bool(iRegP mem_ptr, iRegL oldval, iRegL newval, iRegI res, o7RegI tmp1, flagsReg ccr ) %{
+  predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapL mem_ptr (Binary oldval newval)));
   effect( USE mem_ptr, KILL ccr, KILL tmp1);
   format %{
@@ -7230,6 +7237,9 @@ instruct compareAndSwapI_bool(iRegP mem_ptr, iRegI oldval, iRegI newval, iRegI r
 %}
 
 instruct compareAndSwapP_bool(iRegP mem_ptr, iRegP oldval, iRegP newval, iRegI res, o7RegI tmp1, flagsReg ccr ) %{
+#ifdef _LP64
+  predicate(VM_Version::supports_cx8());
+#endif
   match(Set res (CompareAndSwapP mem_ptr (Binary oldval newval)));
   effect( USE mem_ptr, KILL ccr, KILL tmp1);
   format %{
@@ -7264,6 +7274,38 @@ instruct compareAndSwapN_bool(iRegP mem_ptr, iRegN oldval, iRegN newval, iRegI r
   ins_pipe( long_memory_op );
 %}
 
+instruct xchgI( memory mem, iRegI newval) %{
+  match(Set newval (GetAndSetI mem newval));
+  format %{ "SWAP  [$mem],$newval" %}
+  size(4);
+  ins_encode %{
+    __ swap($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( long_memory_op );
+%}
+
+#ifndef _LP64
+instruct xchgP( memory mem, iRegP newval) %{
+  match(Set newval (GetAndSetP mem newval));
+  format %{ "SWAP  [$mem],$newval" %}
+  size(4);
+  ins_encode %{
+    __ swap($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( long_memory_op );
+%}
+#endif
+
+instruct xchgN( memory mem, iRegN newval) %{
+  match(Set newval (GetAndSetN mem newval));
+  format %{ "SWAP  [$mem],$newval" %}
+  size(4);
+  ins_encode %{
+    __ swap($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( long_memory_op );
+%}
+
 //---------------------
 // Subtraction Instructions
 // Register Subtraction
diff --git a/src/cpu/sparc/vm/vm_version_sparc.cpp b/src/cpu/sparc/vm/vm_version_sparc.cpp
index ca93deb0e..07eadd5bf 100644
--- a/src/cpu/sparc/vm/vm_version_sparc.cpp
+++ b/src/cpu/sparc/vm/vm_version_sparc.cpp
@@ -96,6 +96,7 @@ void VM_Version::initialize() {
   UseSSE = 0; // Only on x86 and x64
 
   _supports_cx8 = has_v9();
+  _supports_atomic_getset4 = true; // swap instruction
 
   if (is_niagara()) {
     // Indirect branch is the same cost as direct
@@ -338,7 +339,11 @@ void VM_Version::revert() {
 
 unsigned int VM_Version::calc_parallel_worker_threads() {
   unsigned int result;
-  if (is_niagara_plus()) {
+  if (is_M_series()) {
+    // for now, use same gc thread calculation for M-series as for niagara-plus
+    // in future, we may want to tweak parameters for nof_parallel_worker_thread
+    result = nof_parallel_worker_threads(5, 16, 8);
+  } else if (is_niagara_plus()) {
     result = nof_parallel_worker_threads(5, 16, 8);
   } else {
     result = nof_parallel_worker_threads(5, 8, 8);
diff --git a/src/cpu/sparc/vm/vm_version_sparc.hpp b/src/cpu/sparc/vm/vm_version_sparc.hpp
index d7f21c328..d602fd08c 100644
--- a/src/cpu/sparc/vm/vm_version_sparc.hpp
+++ b/src/cpu/sparc/vm/vm_version_sparc.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -124,6 +124,8 @@ public:
   // Returns true if the platform is in the niagara line (T series)
   // and newer than the niagara1.
   static bool is_niagara_plus()         { return is_T_family(_features) && !is_T1_model(_features); }
+
+  static bool is_M_series()             { return is_M_family(_features); }
   static bool is_T4()                   { return is_T_family(_features) && has_cbcond(); }
 
   // Fujitsu SPARC64
diff --git a/src/cpu/x86/vm/assembler_x86.cpp b/src/cpu/x86/vm/assembler_x86.cpp
index 611d7ab50..0f6b99dd5 100644
--- a/src/cpu/x86/vm/assembler_x86.cpp
+++ b/src/cpu/x86/vm/assembler_x86.cpp
@@ -3496,6 +3496,33 @@ void Assembler::vinsertf128h(XMMRegister dst, XMMRegister nds, XMMRegister src)
   emit_byte(0x01);
 }
 
+void Assembler::vinsertf128h(XMMRegister dst, Address src) {
+  assert(VM_Version::supports_avx(), "");
+  InstructionMark im(this);
+  bool vector256 = true;
+  assert(dst != xnoreg, "sanity");
+  int dst_enc = dst->encoding();
+  // swap src<->dst for encoding
+  vex_prefix(src, dst_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F_3A, false, vector256);
+  emit_byte(0x18);
+  emit_operand(dst, src);
+  // 0x01 - insert into upper 128 bits
+  emit_byte(0x01);
+}
+
+void Assembler::vextractf128h(Address dst, XMMRegister src) {
+  assert(VM_Version::supports_avx(), "");
+  InstructionMark im(this);
+  bool vector256 = true;
+  assert(src != xnoreg, "sanity");
+  int src_enc = src->encoding();
+  vex_prefix(dst, 0, src_enc, VEX_SIMD_66, VEX_OPCODE_0F_3A, false, vector256);
+  emit_byte(0x19);
+  emit_operand(src, dst);
+  // 0x01 - extract from upper 128 bits
+  emit_byte(0x01);
+}
+
 void Assembler::vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src) {
   assert(VM_Version::supports_avx2(), "");
   bool vector256 = true;
@@ -3507,6 +3534,33 @@ void Assembler::vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src)
   emit_byte(0x01);
 }
 
+void Assembler::vinserti128h(XMMRegister dst, Address src) {
+  assert(VM_Version::supports_avx2(), "");
+  InstructionMark im(this);
+  bool vector256 = true;
+  assert(dst != xnoreg, "sanity");
+  int dst_enc = dst->encoding();
+  // swap src<->dst for encoding
+  vex_prefix(src, dst_enc, dst_enc, VEX_SIMD_66, VEX_OPCODE_0F_3A, false, vector256);
+  emit_byte(0x38);
+  emit_operand(dst, src);
+  // 0x01 - insert into upper 128 bits
+  emit_byte(0x01);
+}
+
+void Assembler::vextracti128h(Address dst, XMMRegister src) {
+  assert(VM_Version::supports_avx2(), "");
+  InstructionMark im(this);
+  bool vector256 = true;
+  assert(src != xnoreg, "sanity");
+  int src_enc = src->encoding();
+  vex_prefix(dst, 0, src_enc, VEX_SIMD_66, VEX_OPCODE_0F_3A, false, vector256);
+  emit_byte(0x39);
+  emit_operand(src, dst);
+  // 0x01 - extract from upper 128 bits
+  emit_byte(0x01);
+}
+
 void Assembler::vzeroupper() {
   assert(VM_Version::supports_avx(), "");
   (void)vex_prefix_and_encode(xmm0, xmm0, xmm0, VEX_SIMD_NONE);
@@ -8907,11 +8961,9 @@ void MacroAssembler::fp_runtime_fallback(address runtime_entry, int nb_args, int
   pusha();
 
   // if we are coming from c1, xmm registers may be live
-  if (UseSSE >= 1) {
-    subptr(rsp, sizeof(jdouble)* LP64_ONLY(16) NOT_LP64(8));
-  }
   int off = 0;
   if (UseSSE == 1)  {
+    subptr(rsp, sizeof(jdouble)*8);
     movflt(Address(rsp,off++*sizeof(jdouble)),xmm0);
     movflt(Address(rsp,off++*sizeof(jdouble)),xmm1);
     movflt(Address(rsp,off++*sizeof(jdouble)),xmm2);
@@ -8921,23 +8973,50 @@ void MacroAssembler::fp_runtime_fallback(address runtime_entry, int nb_args, int
     movflt(Address(rsp,off++*sizeof(jdouble)),xmm6);
     movflt(Address(rsp,off++*sizeof(jdouble)),xmm7);
   } else if (UseSSE >= 2)  {
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm0);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm1);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm2);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm3);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm4);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm5);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm6);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm7);
+#ifdef COMPILER2
+    if (MaxVectorSize > 16) {
+      assert(UseAVX > 0, "256bit vectors are supported only with AVX");
+      // Save upper half of YMM registes
+      subptr(rsp, 16 * LP64_ONLY(16) NOT_LP64(8));
+      vextractf128h(Address(rsp,  0),xmm0);
+      vextractf128h(Address(rsp, 16),xmm1);
+      vextractf128h(Address(rsp, 32),xmm2);
+      vextractf128h(Address(rsp, 48),xmm3);
+      vextractf128h(Address(rsp, 64),xmm4);
+      vextractf128h(Address(rsp, 80),xmm5);
+      vextractf128h(Address(rsp, 96),xmm6);
+      vextractf128h(Address(rsp,112),xmm7);
+#ifdef _LP64
+      vextractf128h(Address(rsp,128),xmm8);
+      vextractf128h(Address(rsp,144),xmm9);
+      vextractf128h(Address(rsp,160),xmm10);
+      vextractf128h(Address(rsp,176),xmm11);
+      vextractf128h(Address(rsp,192),xmm12);
+      vextractf128h(Address(rsp,208),xmm13);
+      vextractf128h(Address(rsp,224),xmm14);
+      vextractf128h(Address(rsp,240),xmm15);
+#endif
+    }
+#endif
+    // Save whole 128bit (16 bytes) XMM regiters
+    subptr(rsp, 16 * LP64_ONLY(16) NOT_LP64(8));
+    movdqu(Address(rsp,off++*16),xmm0);
+    movdqu(Address(rsp,off++*16),xmm1);
+    movdqu(Address(rsp,off++*16),xmm2);
+    movdqu(Address(rsp,off++*16),xmm3);
+    movdqu(Address(rsp,off++*16),xmm4);
+    movdqu(Address(rsp,off++*16),xmm5);
+    movdqu(Address(rsp,off++*16),xmm6);
+    movdqu(Address(rsp,off++*16),xmm7);
 #ifdef _LP64
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm8);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm9);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm10);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm11);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm12);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm13);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm14);
-    movdbl(Address(rsp,off++*sizeof(jdouble)),xmm15);
+    movdqu(Address(rsp,off++*16),xmm8);
+    movdqu(Address(rsp,off++*16),xmm9);
+    movdqu(Address(rsp,off++*16),xmm10);
+    movdqu(Address(rsp,off++*16),xmm11);
+    movdqu(Address(rsp,off++*16),xmm12);
+    movdqu(Address(rsp,off++*16),xmm13);
+    movdqu(Address(rsp,off++*16),xmm14);
+    movdqu(Address(rsp,off++*16),xmm15);
 #endif
   }
 
@@ -9015,28 +9094,52 @@ void MacroAssembler::fp_runtime_fallback(address runtime_entry, int nb_args, int
     movflt(xmm5, Address(rsp,off++*sizeof(jdouble)));
     movflt(xmm6, Address(rsp,off++*sizeof(jdouble)));
     movflt(xmm7, Address(rsp,off++*sizeof(jdouble)));
+    addptr(rsp, sizeof(jdouble)*8);
   } else if (UseSSE >= 2)  {
-    movdbl(xmm0, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm1, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm2, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm3, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm4, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm5, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm6, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm7, Address(rsp,off++*sizeof(jdouble)));
+    // Restore whole 128bit (16 bytes) XMM regiters
+    movdqu(xmm0, Address(rsp,off++*16));
+    movdqu(xmm1, Address(rsp,off++*16));
+    movdqu(xmm2, Address(rsp,off++*16));
+    movdqu(xmm3, Address(rsp,off++*16));
+    movdqu(xmm4, Address(rsp,off++*16));
+    movdqu(xmm5, Address(rsp,off++*16));
+    movdqu(xmm6, Address(rsp,off++*16));
+    movdqu(xmm7, Address(rsp,off++*16));
 #ifdef _LP64
-    movdbl(xmm8, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm9, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm10, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm11, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm12, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm13, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm14, Address(rsp,off++*sizeof(jdouble)));
-    movdbl(xmm15, Address(rsp,off++*sizeof(jdouble)));
+    movdqu(xmm8, Address(rsp,off++*16));
+    movdqu(xmm9, Address(rsp,off++*16));
+    movdqu(xmm10, Address(rsp,off++*16));
+    movdqu(xmm11, Address(rsp,off++*16));
+    movdqu(xmm12, Address(rsp,off++*16));
+    movdqu(xmm13, Address(rsp,off++*16));
+    movdqu(xmm14, Address(rsp,off++*16));
+    movdqu(xmm15, Address(rsp,off++*16));
+#endif
+    addptr(rsp, 16 * LP64_ONLY(16) NOT_LP64(8));
+#ifdef COMPILER2
+    if (MaxVectorSize > 16) {
+      // Restore upper half of YMM registes.
+      vinsertf128h(xmm0, Address(rsp,  0));
+      vinsertf128h(xmm1, Address(rsp, 16));
+      vinsertf128h(xmm2, Address(rsp, 32));
+      vinsertf128h(xmm3, Address(rsp, 48));
+      vinsertf128h(xmm4, Address(rsp, 64));
+      vinsertf128h(xmm5, Address(rsp, 80));
+      vinsertf128h(xmm6, Address(rsp, 96));
+      vinsertf128h(xmm7, Address(rsp,112));
+#ifdef _LP64
+      vinsertf128h(xmm8, Address(rsp,128));
+      vinsertf128h(xmm9, Address(rsp,144));
+      vinsertf128h(xmm10, Address(rsp,160));
+      vinsertf128h(xmm11, Address(rsp,176));
+      vinsertf128h(xmm12, Address(rsp,192));
+      vinsertf128h(xmm13, Address(rsp,208));
+      vinsertf128h(xmm14, Address(rsp,224));
+      vinsertf128h(xmm15, Address(rsp,240));
+#endif
+      addptr(rsp, 16 * LP64_ONLY(16) NOT_LP64(8));
+    }
 #endif
-  }
-  if (UseSSE >= 1) {
-    addptr(rsp, sizeof(jdouble)* LP64_ONLY(16) NOT_LP64(8));
   }
   popa();
 }
diff --git a/src/cpu/x86/vm/assembler_x86.hpp b/src/cpu/x86/vm/assembler_x86.hpp
index 368f5c438..dc98fa8f6 100644
--- a/src/cpu/x86/vm/assembler_x86.hpp
+++ b/src/cpu/x86/vm/assembler_x86.hpp
@@ -1743,6 +1743,12 @@ private:
   void vinsertf128h(XMMRegister dst, XMMRegister nds, XMMRegister src);
   void vinserti128h(XMMRegister dst, XMMRegister nds, XMMRegister src);
 
+  // Load/store high 128bit of YMM registers which does not destroy other half.
+  void vinsertf128h(XMMRegister dst, Address src);
+  void vinserti128h(XMMRegister dst, Address src);
+  void vextractf128h(Address dst, XMMRegister src);
+  void vextracti128h(Address dst, XMMRegister src);
+
   // AVX instruction which is used to clear upper 128 bits of YMM registers and
   // to avoid transaction penalty between AVX and SSE states. There is no
   // penalty if legacy SSE instructions are encoded using VEX prefix because
diff --git a/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp b/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp
index 36bd4d764..60eab0920 100644
--- a/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp
+++ b/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp
@@ -3794,5 +3794,49 @@ void LIR_Assembler::peephole(LIR_List*) {
   // do nothing for now
 }
 
+void LIR_Assembler::atomic_op(LIR_Code code, LIR_Opr src, LIR_Opr data, LIR_Opr dest, LIR_Opr tmp) {
+  assert(data == dest, "xchg/xadd uses only 2 operands");
+
+  if (data->type() == T_INT) {
+    if (code == lir_xadd) {
+      if (os::is_MP()) {
+        __ lock();
+      }
+      __ xaddl(as_Address(src->as_address_ptr()), data->as_register());
+    } else {
+      __ xchgl(data->as_register(), as_Address(src->as_address_ptr()));
+    }
+  } else if (data->is_oop()) {
+    assert (code == lir_xchg, "xadd for oops");
+    Register obj = data->as_register();
+#ifdef _LP64
+    if (UseCompressedOops) {
+      __ encode_heap_oop(obj);
+      __ xchgl(obj, as_Address(src->as_address_ptr()));
+      __ decode_heap_oop(obj);
+    } else {
+      __ xchgptr(obj, as_Address(src->as_address_ptr()));
+    }
+#else
+    __ xchgl(obj, as_Address(src->as_address_ptr()));
+#endif
+  } else if (data->type() == T_LONG) {
+#ifdef _LP64
+    assert(data->as_register_lo() == data->as_register_hi(), "should be a single register");
+    if (code == lir_xadd) {
+      if (os::is_MP()) {
+        __ lock();
+      }
+      __ xaddq(as_Address(src->as_address_ptr()), data->as_register_lo());
+    } else {
+      __ xchgq(data->as_register_lo(), as_Address(src->as_address_ptr()));
+    }
+#else
+    ShouldNotReachHere();
+#endif
+  } else {
+    ShouldNotReachHere();
+  }
+}
 
 #undef __
diff --git a/src/cpu/x86/vm/c1_LIRGenerator_x86.cpp b/src/cpu/x86/vm/c1_LIRGenerator_x86.cpp
index d363c0eec..d185613c2 100644
--- a/src/cpu/x86/vm/c1_LIRGenerator_x86.cpp
+++ b/src/cpu/x86/vm/c1_LIRGenerator_x86.cpp
@@ -753,9 +753,24 @@ void LIRGenerator::do_CompareAndSwap(Intrinsic* x, ValueType* type) {
   LIR_Opr addr = new_pointer_register();
   LIR_Address* a;
   if(offset.result()->is_constant()) {
+#ifdef _LP64
+    jlong c = offset.result()->as_jlong();
+    if ((jlong)((jint)c) == c) {
+      a = new LIR_Address(obj.result(),
+                          (jint)c,
+                          as_BasicType(type));
+    } else {
+      LIR_Opr tmp = new_register(T_LONG);
+      __ move(offset.result(), tmp);
+      a = new LIR_Address(obj.result(),
+                          tmp,
+                          as_BasicType(type));
+    }
+#else
     a = new LIR_Address(obj.result(),
-                        NOT_LP64(offset.result()->as_constant_ptr()->as_jint()) LP64_ONLY((int)offset.result()->as_constant_ptr()->as_jlong()),
+                        offset.result()->as_jint(),
                         as_BasicType(type));
+#endif
   } else {
     a = new LIR_Address(obj.result(),
                         offset.result(),
@@ -1345,3 +1360,57 @@ void LIRGenerator::put_Object_unsafe(LIR_Opr src, LIR_Opr offset, LIR_Opr data,
     }
   }
 }
+
+void LIRGenerator::do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) {
+  BasicType type = x->basic_type();
+  LIRItem src(x->object(), this);
+  LIRItem off(x->offset(), this);
+  LIRItem value(x->value(), this);
+
+  src.load_item();
+  value.load_item();
+  off.load_nonconstant();
+
+  LIR_Opr dst = rlock_result(x, type);
+  LIR_Opr data = value.result();
+  bool is_obj = (type == T_ARRAY || type == T_OBJECT);
+  LIR_Opr offset = off.result();
+
+  assert (type == T_INT || (!x->is_add() && is_obj) LP64_ONLY( || type == T_LONG ), "unexpected type");
+  LIR_Address* addr;
+  if (offset->is_constant()) {
+#ifdef _LP64
+    jlong c = offset->as_jlong();
+    if ((jlong)((jint)c) == c) {
+      addr = new LIR_Address(src.result(), (jint)c, type);
+    } else {
+      LIR_Opr tmp = new_register(T_LONG);
+      __ move(offset, tmp);
+      addr = new LIR_Address(src.result(), tmp, type);
+    }
+#else
+    addr = new LIR_Address(src.result(), offset->as_jint(), type);
+#endif
+  } else {
+    addr = new LIR_Address(src.result(), offset, type);
+  }
+
+  if (data != dst) {
+    __ move(data, dst);
+    data = dst;
+  }
+  if (x->is_add()) {
+    __ xadd(LIR_OprFact::address(addr), data, dst, LIR_OprFact::illegalOpr);
+  } else {
+    if (is_obj) {
+      // Do the pre-write barrier, if any.
+      pre_barrier(LIR_OprFact::address(addr), LIR_OprFact::illegalOpr /* pre_val */,
+                  true /* do_load */, false /* patch */, NULL);
+    }
+    __ xchg(LIR_OprFact::address(addr), data, dst, LIR_OprFact::illegalOpr);
+    if (is_obj) {
+      // Seems to be a precise address
+      post_barrier(LIR_OprFact::address(addr), data);
+    }
+  }
+}
diff --git a/src/cpu/x86/vm/methodHandles_x86.cpp b/src/cpu/x86/vm/methodHandles_x86.cpp
index eb4e20fb2..fd01176d9 100644
--- a/src/cpu/x86/vm/methodHandles_x86.cpp
+++ b/src/cpu/x86/vm/methodHandles_x86.cpp
@@ -327,10 +327,11 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
     assert_different_registers(temp3,        rcx, rdx);
   }
 #endif
+  else {
+    assert_different_registers(temp1, temp2, temp3, saved_last_sp_register());  // don't trash lastSP
+  }
   assert_different_registers(temp1, temp2, temp3, receiver_reg);
   assert_different_registers(temp1, temp2, temp3, member_reg);
-  if (!for_compiler_entry)
-    assert_different_registers(temp1, temp2, temp3, saved_last_sp_register());  // don't trash lastSP
 
   if (iid == vmIntrinsics::_invokeBasic) {
     // indirect through MH.form.vmentry.vmtarget
@@ -392,14 +393,13 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
     //  rsi/r13 - interpreter linkage (if interpreted)
     //  rcx, rdx, rsi, rdi, r8, r8 - compiler arguments (if compiled)
 
-    bool method_is_live = false;
+    Label L_incompatible_class_change_error;
     switch (iid) {
     case vmIntrinsics::_linkToSpecial:
       if (VerifyMethodHandles) {
         verify_ref_kind(_masm, JVM_REF_invokeSpecial, member_reg, temp3);
       }
       __ movptr(rbx_method, member_vmtarget);
-      method_is_live = true;
       break;
 
     case vmIntrinsics::_linkToStatic:
@@ -407,7 +407,6 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
         verify_ref_kind(_masm, JVM_REF_invokeStatic, member_reg, temp3);
       }
       __ movptr(rbx_method, member_vmtarget);
-      method_is_live = true;
       break;
 
     case vmIntrinsics::_linkToVirtual:
@@ -436,7 +435,6 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
 
       // get target Method* & entry point
       __ lookup_virtual_method(temp1_recv_klass, temp2_index, rbx_method);
-      method_is_live = true;
       break;
     }
 
@@ -464,35 +462,32 @@ void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm,
       }
 
       // given intf, index, and recv klass, dispatch to the implementation method
-      Label L_no_such_interface;
       __ lookup_interface_method(temp1_recv_klass, temp3_intf,
                                  // note: next two args must be the same:
                                  rbx_index, rbx_method,
                                  temp2,
-                                 L_no_such_interface);
-
-      __ verify_method_ptr(rbx_method);
-      jump_from_method_handle(_masm, rbx_method, temp2, for_compiler_entry);
-      __ hlt();
-
-      __ bind(L_no_such_interface);
-      __ jump(RuntimeAddress(StubRoutines::throw_IncompatibleClassChangeError_entry()));
+                                 L_incompatible_class_change_error);
       break;
     }
 
     default:
-      fatal(err_msg("unexpected intrinsic %d: %s", iid, vmIntrinsics::name_at(iid)));
+      fatal(err_msg_res("unexpected intrinsic %d: %s", iid, vmIntrinsics::name_at(iid)));
       break;
     }
 
-    if (method_is_live) {
-      // live at this point:  rbx_method, rsi/r13 (if interpreted)
+    // Live at this point:
+    //   rbx_method
+    //   rsi/r13 (if interpreted)
 
-      // After figuring out which concrete method to call, jump into it.
-      // Note that this works in the interpreter with no data motion.
-      // But the compiled version will require that rcx_recv be shifted out.
-      __ verify_method_ptr(rbx_method);
-      jump_from_method_handle(_masm, rbx_method, temp1, for_compiler_entry);
+    // After figuring out which concrete method to call, jump into it.
+    // Note that this works in the interpreter with no data motion.
+    // But the compiled version will require that rcx_recv be shifted out.
+    __ verify_method_ptr(rbx_method);
+    jump_from_method_handle(_masm, rbx_method, temp1, for_compiler_entry);
+
+    if (iid == vmIntrinsics::_linkToInterface) {
+      __ bind(L_incompatible_class_change_error);
+      __ jump(RuntimeAddress(StubRoutines::throw_IncompatibleClassChangeError_entry()));
     }
   }
 }
diff --git a/src/cpu/x86/vm/sharedRuntime_x86_32.cpp b/src/cpu/x86/vm/sharedRuntime_x86_32.cpp
index 30b5cfca9..7cb2bb81e 100644
--- a/src/cpu/x86/vm/sharedRuntime_x86_32.cpp
+++ b/src/cpu/x86/vm/sharedRuntime_x86_32.cpp
@@ -46,11 +46,11 @@
 const int StackAlignmentInSlots = StackAlignmentInBytes / VMRegImpl::stack_slot_size;
 
 class RegisterSaver {
-  enum { FPU_regs_live = 8 /*for the FPU stack*/+8/*eight more for XMM registers*/ };
   // Capture info about frame layout
+#define DEF_XMM_OFFS(regnum) xmm ## regnum ## _off = xmm_off + (regnum)*16/BytesPerInt, xmm ## regnum ## H_off
   enum layout {
                 fpu_state_off = 0,
-                fpu_state_end = fpu_state_off+FPUStateSizeInWords-1,
+                fpu_state_end = fpu_state_off+FPUStateSizeInWords,
                 st0_off, st0H_off,
                 st1_off, st1H_off,
                 st2_off, st2H_off,
@@ -59,16 +59,16 @@ class RegisterSaver {
                 st5_off, st5H_off,
                 st6_off, st6H_off,
                 st7_off, st7H_off,
-
-                xmm0_off, xmm0H_off,
-                xmm1_off, xmm1H_off,
-                xmm2_off, xmm2H_off,
-                xmm3_off, xmm3H_off,
-                xmm4_off, xmm4H_off,
-                xmm5_off, xmm5H_off,
-                xmm6_off, xmm6H_off,
-                xmm7_off, xmm7H_off,
-                flags_off,
+                xmm_off,
+                DEF_XMM_OFFS(0),
+                DEF_XMM_OFFS(1),
+                DEF_XMM_OFFS(2),
+                DEF_XMM_OFFS(3),
+                DEF_XMM_OFFS(4),
+                DEF_XMM_OFFS(5),
+                DEF_XMM_OFFS(6),
+                DEF_XMM_OFFS(7),
+                flags_off = xmm7_off + 16/BytesPerInt + 1, // 16-byte stack alignment fill word
                 rdi_off,
                 rsi_off,
                 ignore_off,  // extra copy of rbp,
@@ -83,13 +83,13 @@ class RegisterSaver {
                 rbp_off,
                 return_off,      // slot for return address
                 reg_save_size };
-
+  enum { FPU_regs_live = flags_off - fpu_state_end };
 
   public:
 
   static OopMap* save_live_registers(MacroAssembler* masm, int additional_frame_words,
-                                     int* total_frame_words, bool verify_fpu = true);
-  static void restore_live_registers(MacroAssembler* masm);
+                                     int* total_frame_words, bool verify_fpu = true, bool save_vectors = false);
+  static void restore_live_registers(MacroAssembler* masm, bool restore_vectors = false);
 
   static int rax_offset() { return rax_off; }
   static int rbx_offset() { return rbx_off; }
@@ -113,9 +113,20 @@ class RegisterSaver {
 };
 
 OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_frame_words,
-                                           int* total_frame_words, bool verify_fpu) {
-
-  int frame_size_in_bytes =  (reg_save_size + additional_frame_words) * wordSize;
+                                           int* total_frame_words, bool verify_fpu, bool save_vectors) {
+  int vect_words = 0;
+#ifdef COMPILER2
+  if (save_vectors) {
+    assert(UseAVX > 0, "256bit vectors are supported only with AVX");
+    assert(MaxVectorSize == 32, "only 256bit vectors are supported now");
+    // Save upper half of YMM registes
+    vect_words = 8 * 16 / wordSize;
+    additional_frame_words += vect_words;
+  }
+#else
+  assert(!save_vectors, "vectors are generated only by C2");
+#endif
+  int frame_size_in_bytes = (reg_save_size + additional_frame_words) * wordSize;
   int frame_words = frame_size_in_bytes / wordSize;
   *total_frame_words = frame_words;
 
@@ -129,7 +140,7 @@ OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_
   __ enter();
   __ pusha();
   __ pushf();
-  __ subptr(rsp,FPU_regs_live*sizeof(jdouble)); // Push FPU registers space
+  __ subptr(rsp,FPU_regs_live*wordSize); // Push FPU registers space
   __ push_FPU_state();          // Save FPU state & init
 
   if (verify_fpu) {
@@ -183,14 +194,28 @@ OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_
     __ movflt(Address(rsp,xmm6_off*wordSize),xmm6);
     __ movflt(Address(rsp,xmm7_off*wordSize),xmm7);
   } else if( UseSSE >= 2 ) {
-    __ movdbl(Address(rsp,xmm0_off*wordSize),xmm0);
-    __ movdbl(Address(rsp,xmm1_off*wordSize),xmm1);
-    __ movdbl(Address(rsp,xmm2_off*wordSize),xmm2);
-    __ movdbl(Address(rsp,xmm3_off*wordSize),xmm3);
-    __ movdbl(Address(rsp,xmm4_off*wordSize),xmm4);
-    __ movdbl(Address(rsp,xmm5_off*wordSize),xmm5);
-    __ movdbl(Address(rsp,xmm6_off*wordSize),xmm6);
-    __ movdbl(Address(rsp,xmm7_off*wordSize),xmm7);
+    // Save whole 128bit (16 bytes) XMM regiters
+    __ movdqu(Address(rsp,xmm0_off*wordSize),xmm0);
+    __ movdqu(Address(rsp,xmm1_off*wordSize),xmm1);
+    __ movdqu(Address(rsp,xmm2_off*wordSize),xmm2);
+    __ movdqu(Address(rsp,xmm3_off*wordSize),xmm3);
+    __ movdqu(Address(rsp,xmm4_off*wordSize),xmm4);
+    __ movdqu(Address(rsp,xmm5_off*wordSize),xmm5);
+    __ movdqu(Address(rsp,xmm6_off*wordSize),xmm6);
+    __ movdqu(Address(rsp,xmm7_off*wordSize),xmm7);
+  }
+
+  if (vect_words > 0) {
+    assert(vect_words*wordSize == 128, "");
+    __ subptr(rsp, 128); // Save upper half of YMM registes
+    __ vextractf128h(Address(rsp,  0),xmm0);
+    __ vextractf128h(Address(rsp, 16),xmm1);
+    __ vextractf128h(Address(rsp, 32),xmm2);
+    __ vextractf128h(Address(rsp, 48),xmm3);
+    __ vextractf128h(Address(rsp, 64),xmm4);
+    __ vextractf128h(Address(rsp, 80),xmm5);
+    __ vextractf128h(Address(rsp, 96),xmm6);
+    __ vextractf128h(Address(rsp,112),xmm7);
   }
 
   // Set an oopmap for the call site.  This oopmap will map all
@@ -253,10 +278,20 @@ OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_
 
 }
 
-void RegisterSaver::restore_live_registers(MacroAssembler* masm) {
-
+void RegisterSaver::restore_live_registers(MacroAssembler* masm, bool restore_vectors) {
   // Recover XMM & FPU state
-  if( UseSSE == 1 ) {
+  int additional_frame_bytes = 0;
+#ifdef COMPILER2
+  if (restore_vectors) {
+    assert(UseAVX > 0, "256bit vectors are supported only with AVX");
+    assert(MaxVectorSize == 32, "only 256bit vectors are supported now");
+    additional_frame_bytes = 128;
+  }
+#else
+  assert(!restore_vectors, "vectors are generated only by C2");
+#endif
+  if (UseSSE == 1) {
+    assert(additional_frame_bytes == 0, "");
     __ movflt(xmm0,Address(rsp,xmm0_off*wordSize));
     __ movflt(xmm1,Address(rsp,xmm1_off*wordSize));
     __ movflt(xmm2,Address(rsp,xmm2_off*wordSize));
@@ -265,18 +300,33 @@ void RegisterSaver::restore_live_registers(MacroAssembler* masm) {
     __ movflt(xmm5,Address(rsp,xmm5_off*wordSize));
     __ movflt(xmm6,Address(rsp,xmm6_off*wordSize));
     __ movflt(xmm7,Address(rsp,xmm7_off*wordSize));
-  } else if( UseSSE >= 2 ) {
-    __ movdbl(xmm0,Address(rsp,xmm0_off*wordSize));
-    __ movdbl(xmm1,Address(rsp,xmm1_off*wordSize));
-    __ movdbl(xmm2,Address(rsp,xmm2_off*wordSize));
-    __ movdbl(xmm3,Address(rsp,xmm3_off*wordSize));
-    __ movdbl(xmm4,Address(rsp,xmm4_off*wordSize));
-    __ movdbl(xmm5,Address(rsp,xmm5_off*wordSize));
-    __ movdbl(xmm6,Address(rsp,xmm6_off*wordSize));
-    __ movdbl(xmm7,Address(rsp,xmm7_off*wordSize));
+  } else if (UseSSE >= 2) {
+#define STACK_ADDRESS(x) Address(rsp,(x)*wordSize + additional_frame_bytes)
+    __ movdqu(xmm0,STACK_ADDRESS(xmm0_off));
+    __ movdqu(xmm1,STACK_ADDRESS(xmm1_off));
+    __ movdqu(xmm2,STACK_ADDRESS(xmm2_off));
+    __ movdqu(xmm3,STACK_ADDRESS(xmm3_off));
+    __ movdqu(xmm4,STACK_ADDRESS(xmm4_off));
+    __ movdqu(xmm5,STACK_ADDRESS(xmm5_off));
+    __ movdqu(xmm6,STACK_ADDRESS(xmm6_off));
+    __ movdqu(xmm7,STACK_ADDRESS(xmm7_off));
+#undef STACK_ADDRESS
+  }
+  if (restore_vectors) {
+    // Restore upper half of YMM registes.
+    assert(additional_frame_bytes == 128, "");
+    __ vinsertf128h(xmm0, Address(rsp,  0));
+    __ vinsertf128h(xmm1, Address(rsp, 16));
+    __ vinsertf128h(xmm2, Address(rsp, 32));
+    __ vinsertf128h(xmm3, Address(rsp, 48));
+    __ vinsertf128h(xmm4, Address(rsp, 64));
+    __ vinsertf128h(xmm5, Address(rsp, 80));
+    __ vinsertf128h(xmm6, Address(rsp, 96));
+    __ vinsertf128h(xmm7, Address(rsp,112));
+    __ addptr(rsp, additional_frame_bytes);
   }
   __ pop_FPU_state();
-  __ addptr(rsp, FPU_regs_live*sizeof(jdouble)); // Pop FPU registers
+  __ addptr(rsp, FPU_regs_live*wordSize); // Pop FPU registers
 
   __ popf();
   __ popa();
@@ -308,6 +358,13 @@ void RegisterSaver::restore_result_registers(MacroAssembler* masm) {
   __ addptr(rsp, return_off * wordSize);
 }
 
+// Is vector's size (in bytes) bigger than a size saved by default?
+// 16 bytes XMM registers are saved by default using SSE2 movdqu instructions.
+// Note, MaxVectorSize == 0 with UseSSE < 2 and vectors are not generated.
+bool SharedRuntime::is_wide_vector(int size) {
+  return size > 16;
+}
+
 // The java_calling_convention describes stack locations as ideal slots on
 // a frame with no abi restrictions. Since we must observe abi restrictions
 // (like the placement of the register window) the slots must be biased by
@@ -1346,12 +1403,12 @@ static void unpack_array_argument(MacroAssembler* masm, VMRegPair reg, BasicType
 }
 
 static void verify_oop_args(MacroAssembler* masm,
-                            int total_args_passed,
+                            methodHandle method,
                             const BasicType* sig_bt,
                             const VMRegPair* regs) {
   Register temp_reg = rbx;  // not part of any compiled calling seq
   if (VerifyOops) {
-    for (int i = 0; i < total_args_passed; i++) {
+    for (int i = 0; i < method->size_of_parameters(); i++) {
       if (sig_bt[i] == T_OBJECT ||
           sig_bt[i] == T_ARRAY) {
         VMReg r = regs[i].first();
@@ -1368,35 +1425,32 @@ static void verify_oop_args(MacroAssembler* masm,
 }
 
 static void gen_special_dispatch(MacroAssembler* masm,
-                                 int total_args_passed,
-                                 int comp_args_on_stack,
-                                 vmIntrinsics::ID special_dispatch,
+                                 methodHandle method,
                                  const BasicType* sig_bt,
                                  const VMRegPair* regs) {
-  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+  verify_oop_args(masm, method, sig_bt, regs);
+  vmIntrinsics::ID iid = method->intrinsic_id();
 
   // Now write the args into the outgoing interpreter space
   bool     has_receiver   = false;
   Register receiver_reg   = noreg;
   int      member_arg_pos = -1;
   Register member_reg     = noreg;
-  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(iid);
   if (ref_kind != 0) {
-    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_arg_pos = method->size_of_parameters() - 1;  // trailing MemberName argument
     member_reg = rbx;  // known to be free at this point
     has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
-  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+  } else if (iid == vmIntrinsics::_invokeBasic) {
     has_receiver = true;
   } else {
-    guarantee(false, err_msg("special_dispatch=%d", special_dispatch));
+    fatal(err_msg_res("unexpected intrinsic id %d", iid));
   }
 
   if (member_reg != noreg) {
     // Load the member_arg into register, if necessary.
-    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
-    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    SharedRuntime::check_member_name_argument_is_last_argument(method, sig_bt, regs);
     VMReg r = regs[member_arg_pos].first();
-    assert(r->is_valid(), "bad member arg");
     if (r->is_stack()) {
       __ movptr(member_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
     } else {
@@ -1407,7 +1461,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
 
   if (has_receiver) {
     // Make sure the receiver is loaded into a register.
-    assert(total_args_passed > 0, "oob");
+    assert(method->size_of_parameters() > 0, "oob");
     assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
     VMReg r = regs[0].first();
     assert(r->is_valid(), "bad receiver arg");
@@ -1415,7 +1469,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
       // Porting note:  This assumes that compiled calling conventions always
       // pass the receiver oop in a register.  If this is not true on some
       // platform, pick a temp and load the receiver from stack.
-      assert(false, "receiver always in a register");
+      fatal("receiver always in a register");
       receiver_reg = rcx;  // known to be free at this point
       __ movptr(receiver_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
     } else {
@@ -1425,7 +1479,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
   }
 
   // Figure out which address we are really jumping to:
-  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+  MethodHandles::generate_method_handle_dispatch(masm, iid,
                                                  receiver_reg, member_reg, /*for_compiler_entry:*/ true);
 }
 
@@ -1461,8 +1515,6 @@ static void gen_special_dispatch(MacroAssembler* masm,
 nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 methodHandle method,
                                                 int compile_id,
-                                                int total_in_args,
-                                                int comp_args_on_stack,
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
@@ -1471,9 +1523,7 @@ nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
     intptr_t start = (intptr_t)__ pc();
     int vep_offset = ((intptr_t)__ pc()) - start;
     gen_special_dispatch(masm,
-                         total_in_args,
-                         comp_args_on_stack,
-                         method->intrinsic_id(),
+                         method,
                          in_sig_bt,
                          in_regs);
     int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
@@ -1506,6 +1556,7 @@ nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
   // we convert the java signature to a C signature by inserting
   // the hidden arguments as arg[0] and possibly arg[1] (static method)
 
+  const int total_in_args = method->size_of_parameters();
   int total_c_args = total_in_args;
   if (!is_critical_native) {
     total_c_args += 1;
@@ -2738,7 +2789,6 @@ uint SharedRuntime::out_preserve_stack_slots() {
   return 0;
 }
 
-
 //------------------------------generate_deopt_blob----------------------------
 void SharedRuntime::generate_deopt_blob() {
   // allocate space for the code
@@ -3276,7 +3326,7 @@ void SharedRuntime::generate_uncommon_trap_blob() {
 // setup oopmap, and calls safepoint code to stop the compiled code for
 // a safepoint.
 //
-SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause_return) {
+SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, int poll_type) {
 
   // Account for thread arg in our frame
   const int additional_words = 1;
@@ -3296,17 +3346,18 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   const Register java_thread = rdi; // callee-saved for VC++
   address start   = __ pc();
   address call_pc = NULL;
-
+  bool cause_return = (poll_type == POLL_AT_RETURN);
+  bool save_vectors = (poll_type == POLL_AT_VECTOR_LOOP);
   // If cause_return is true we are at a poll_return and there is
   // the return address on the stack to the caller on the nmethod
   // that is safepoint. We can leave this return on the stack and
   // effectively complete the return and safepoint in the caller.
   // Otherwise we push space for a return address that the safepoint
   // handler will install later to make the stack walking sensible.
-  if( !cause_return )
-    __ push(rbx);                // Make room for return address (or push it again)
+  if (!cause_return)
+    __ push(rbx);  // Make room for return address (or push it again)
 
-  map = RegisterSaver::save_live_registers(masm, additional_words, &frame_size_in_words, false);
+  map = RegisterSaver::save_live_registers(masm, additional_words, &frame_size_in_words, false, save_vectors);
 
   // The following is basically a call_VM. However, we need the precise
   // address of the call in order to generate an oopmap. Hence, we do all the
@@ -3318,7 +3369,7 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   __ set_last_Java_frame(java_thread, noreg, noreg, NULL);
 
   // if this was not a poll_return then we need to correct the return address now.
-  if( !cause_return ) {
+  if (!cause_return) {
     __ movptr(rax, Address(java_thread, JavaThread::saved_exception_pc_offset()));
     __ movptr(Address(rbp, wordSize), rax);
   }
@@ -3346,15 +3397,14 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   __ jcc(Assembler::equal, noException);
 
   // Exception pending
-
-  RegisterSaver::restore_live_registers(masm);
+  RegisterSaver::restore_live_registers(masm, save_vectors);
 
   __ jump(RuntimeAddress(StubRoutines::forward_exception_entry()));
 
   __ bind(noException);
 
   // Normal exit, register restoring and exit
-  RegisterSaver::restore_live_registers(masm);
+  RegisterSaver::restore_live_registers(masm, save_vectors);
 
   __ ret(0);
 
diff --git a/src/cpu/x86/vm/sharedRuntime_x86_64.cpp b/src/cpu/x86/vm/sharedRuntime_x86_64.cpp
index 4059ca6b3..49ddcb8f2 100644
--- a/src/cpu/x86/vm/sharedRuntime_x86_64.cpp
+++ b/src/cpu/x86/vm/sharedRuntime_x86_64.cpp
@@ -116,8 +116,8 @@ class RegisterSaver {
   };
 
  public:
-  static OopMap* save_live_registers(MacroAssembler* masm, int additional_frame_words, int* total_frame_words);
-  static void restore_live_registers(MacroAssembler* masm);
+  static OopMap* save_live_registers(MacroAssembler* masm, int additional_frame_words, int* total_frame_words, bool save_vectors = false);
+  static void restore_live_registers(MacroAssembler* masm, bool restore_vectors = false);
 
   // Offsets into the register save area
   // Used by deoptimization when it is managing result register
@@ -134,7 +134,19 @@ class RegisterSaver {
   static void restore_result_registers(MacroAssembler* masm);
 };
 
-OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_frame_words, int* total_frame_words) {
+OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_frame_words, int* total_frame_words, bool save_vectors) {
+  int vect_words = 0;
+#ifdef COMPILER2
+  if (save_vectors) {
+    assert(UseAVX > 0, "256bit vectors are supported only with AVX");
+    assert(MaxVectorSize == 32, "only 256bit vectors are supported now");
+    // Save upper half of YMM registes
+    vect_words = 16 * 16 / wordSize;
+    additional_frame_words += vect_words;
+  }
+#else
+  assert(!save_vectors, "vectors are generated only by C2");
+#endif
 
   // Always make the frame size 16-byte aligned
   int frame_size_in_bytes = round_to(additional_frame_words*wordSize +
@@ -155,6 +167,27 @@ OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_
 
   __ enter();          // rsp becomes 16-byte aligned here
   __ push_CPU_state(); // Push a multiple of 16 bytes
+
+  if (vect_words > 0) {
+    assert(vect_words*wordSize == 256, "");
+    __ subptr(rsp, 256); // Save upper half of YMM registes
+    __ vextractf128h(Address(rsp,  0),xmm0);
+    __ vextractf128h(Address(rsp, 16),xmm1);
+    __ vextractf128h(Address(rsp, 32),xmm2);
+    __ vextractf128h(Address(rsp, 48),xmm3);
+    __ vextractf128h(Address(rsp, 64),xmm4);
+    __ vextractf128h(Address(rsp, 80),xmm5);
+    __ vextractf128h(Address(rsp, 96),xmm6);
+    __ vextractf128h(Address(rsp,112),xmm7);
+    __ vextractf128h(Address(rsp,128),xmm8);
+    __ vextractf128h(Address(rsp,144),xmm9);
+    __ vextractf128h(Address(rsp,160),xmm10);
+    __ vextractf128h(Address(rsp,176),xmm11);
+    __ vextractf128h(Address(rsp,192),xmm12);
+    __ vextractf128h(Address(rsp,208),xmm13);
+    __ vextractf128h(Address(rsp,224),xmm14);
+    __ vextractf128h(Address(rsp,240),xmm15);
+  }
   if (frame::arg_reg_save_area_bytes != 0) {
     // Allocate argument register save area
     __ subptr(rsp, frame::arg_reg_save_area_bytes);
@@ -167,112 +200,111 @@ OopMap* RegisterSaver::save_live_registers(MacroAssembler* masm, int additional_
 
   OopMapSet *oop_maps = new OopMapSet();
   OopMap* map = new OopMap(frame_size_in_slots, 0);
-  map->set_callee_saved(VMRegImpl::stack2reg( rax_off  + additional_frame_slots), rax->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( rcx_off  + additional_frame_slots), rcx->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( rdx_off  + additional_frame_slots), rdx->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( rbx_off  + additional_frame_slots), rbx->as_VMReg());
+
+#define STACK_OFFSET(x) VMRegImpl::stack2reg((x) + additional_frame_slots)
+
+  map->set_callee_saved(STACK_OFFSET( rax_off ), rax->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( rcx_off ), rcx->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( rdx_off ), rdx->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( rbx_off ), rbx->as_VMReg());
   // rbp location is known implicitly by the frame sender code, needs no oopmap
   // and the location where rbp was saved by is ignored
-  map->set_callee_saved(VMRegImpl::stack2reg( rsi_off  + additional_frame_slots), rsi->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( rdi_off  + additional_frame_slots), rdi->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r8_off   + additional_frame_slots), r8->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r9_off   + additional_frame_slots), r9->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r10_off  + additional_frame_slots), r10->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r11_off  + additional_frame_slots), r11->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r12_off  + additional_frame_slots), r12->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r13_off  + additional_frame_slots), r13->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r14_off  + additional_frame_slots), r14->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg( r15_off  + additional_frame_slots), r15->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm0_off  + additional_frame_slots), xmm0->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm1_off  + additional_frame_slots), xmm1->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm2_off  + additional_frame_slots), xmm2->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm3_off  + additional_frame_slots), xmm3->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm4_off  + additional_frame_slots), xmm4->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm5_off  + additional_frame_slots), xmm5->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm6_off  + additional_frame_slots), xmm6->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm7_off  + additional_frame_slots), xmm7->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm8_off  + additional_frame_slots), xmm8->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm9_off  + additional_frame_slots), xmm9->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm10_off + additional_frame_slots), xmm10->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm11_off + additional_frame_slots), xmm11->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm12_off + additional_frame_slots), xmm12->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm13_off + additional_frame_slots), xmm13->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm14_off + additional_frame_slots), xmm14->as_VMReg());
-  map->set_callee_saved(VMRegImpl::stack2reg(xmm15_off + additional_frame_slots), xmm15->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( rsi_off ), rsi->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( rdi_off ), rdi->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r8_off  ), r8->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r9_off  ), r9->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r10_off ), r10->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r11_off ), r11->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r12_off ), r12->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r13_off ), r13->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r14_off ), r14->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET( r15_off ), r15->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm0_off ), xmm0->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm1_off ), xmm1->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm2_off ), xmm2->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm3_off ), xmm3->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm4_off ), xmm4->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm5_off ), xmm5->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm6_off ), xmm6->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm7_off ), xmm7->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm8_off ), xmm8->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm9_off ), xmm9->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm10_off), xmm10->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm11_off), xmm11->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm12_off), xmm12->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm13_off), xmm13->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm14_off), xmm14->as_VMReg());
+  map->set_callee_saved(STACK_OFFSET(xmm15_off), xmm15->as_VMReg());
 
   // %%% These should all be a waste but we'll keep things as they were for now
   if (true) {
-    map->set_callee_saved(VMRegImpl::stack2reg( raxH_off  + additional_frame_slots),
-                          rax->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( rcxH_off  + additional_frame_slots),
-                          rcx->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( rdxH_off  + additional_frame_slots),
-                          rdx->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( rbxH_off  + additional_frame_slots),
-                          rbx->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( raxH_off ), rax->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( rcxH_off ), rcx->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( rdxH_off ), rdx->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( rbxH_off ), rbx->as_VMReg()->next());
     // rbp location is known implicitly by the frame sender code, needs no oopmap
-    map->set_callee_saved(VMRegImpl::stack2reg( rsiH_off  + additional_frame_slots),
-                          rsi->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( rdiH_off  + additional_frame_slots),
-                          rdi->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r8H_off   + additional_frame_slots),
-                          r8->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r9H_off   + additional_frame_slots),
-                          r9->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r10H_off  + additional_frame_slots),
-                          r10->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r11H_off  + additional_frame_slots),
-                          r11->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r12H_off  + additional_frame_slots),
-                          r12->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r13H_off  + additional_frame_slots),
-                          r13->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r14H_off  + additional_frame_slots),
-                          r14->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg( r15H_off  + additional_frame_slots),
-                          r15->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm0H_off  + additional_frame_slots),
-                          xmm0->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm1H_off  + additional_frame_slots),
-                          xmm1->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm2H_off  + additional_frame_slots),
-                          xmm2->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm3H_off  + additional_frame_slots),
-                          xmm3->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm4H_off  + additional_frame_slots),
-                          xmm4->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm5H_off  + additional_frame_slots),
-                          xmm5->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm6H_off  + additional_frame_slots),
-                          xmm6->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm7H_off  + additional_frame_slots),
-                          xmm7->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm8H_off  + additional_frame_slots),
-                          xmm8->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm9H_off  + additional_frame_slots),
-                          xmm9->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm10H_off + additional_frame_slots),
-                          xmm10->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm11H_off + additional_frame_slots),
-                          xmm11->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm12H_off + additional_frame_slots),
-                          xmm12->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm13H_off + additional_frame_slots),
-                          xmm13->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm14H_off + additional_frame_slots),
-                          xmm14->as_VMReg()->next());
-    map->set_callee_saved(VMRegImpl::stack2reg(xmm15H_off + additional_frame_slots),
-                          xmm15->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( rsiH_off ), rsi->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( rdiH_off ), rdi->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r8H_off  ), r8->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r9H_off  ), r9->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r10H_off ), r10->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r11H_off ), r11->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r12H_off ), r12->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r13H_off ), r13->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r14H_off ), r14->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET( r15H_off ), r15->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm0H_off ), xmm0->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm1H_off ), xmm1->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm2H_off ), xmm2->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm3H_off ), xmm3->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm4H_off ), xmm4->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm5H_off ), xmm5->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm6H_off ), xmm6->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm7H_off ), xmm7->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm8H_off ), xmm8->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm9H_off ), xmm9->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm10H_off), xmm10->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm11H_off), xmm11->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm12H_off), xmm12->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm13H_off), xmm13->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm14H_off), xmm14->as_VMReg()->next());
+    map->set_callee_saved(STACK_OFFSET(xmm15H_off), xmm15->as_VMReg()->next());
   }
 
   return map;
 }
 
-void RegisterSaver::restore_live_registers(MacroAssembler* masm) {
+void RegisterSaver::restore_live_registers(MacroAssembler* masm, bool restore_vectors) {
   if (frame::arg_reg_save_area_bytes != 0) {
     // Pop arg register save area
     __ addptr(rsp, frame::arg_reg_save_area_bytes);
   }
+#ifdef COMPILER2
+  if (restore_vectors) {
+    // Restore upper half of YMM registes.
+    assert(UseAVX > 0, "256bit vectors are supported only with AVX");
+    assert(MaxVectorSize == 32, "only 256bit vectors are supported now");
+    __ vinsertf128h(xmm0, Address(rsp,  0));
+    __ vinsertf128h(xmm1, Address(rsp, 16));
+    __ vinsertf128h(xmm2, Address(rsp, 32));
+    __ vinsertf128h(xmm3, Address(rsp, 48));
+    __ vinsertf128h(xmm4, Address(rsp, 64));
+    __ vinsertf128h(xmm5, Address(rsp, 80));
+    __ vinsertf128h(xmm6, Address(rsp, 96));
+    __ vinsertf128h(xmm7, Address(rsp,112));
+    __ vinsertf128h(xmm8, Address(rsp,128));
+    __ vinsertf128h(xmm9, Address(rsp,144));
+    __ vinsertf128h(xmm10, Address(rsp,160));
+    __ vinsertf128h(xmm11, Address(rsp,176));
+    __ vinsertf128h(xmm12, Address(rsp,192));
+    __ vinsertf128h(xmm13, Address(rsp,208));
+    __ vinsertf128h(xmm14, Address(rsp,224));
+    __ vinsertf128h(xmm15, Address(rsp,240));
+    __ addptr(rsp, 256);
+  }
+#else
+  assert(!restore_vectors, "vectors are generated only by C2");
+#endif
   // Recover CPU state
   __ pop_CPU_state();
   // Get the rbp described implicitly by the calling convention (no oopMap)
@@ -297,6 +329,12 @@ void RegisterSaver::restore_result_registers(MacroAssembler* masm) {
   __ addptr(rsp, return_offset_in_bytes());
 }
 
+// Is vector's size (in bytes) bigger than a size saved by default?
+// 16 bytes XMM registers are saved by default using fxsave/fxrstor instructions.
+bool SharedRuntime::is_wide_vector(int size) {
+  return size > 16;
+}
+
 // The java_calling_convention describes stack locations as ideal slots on
 // a frame with no abi restrictions. Since we must observe abi restrictions
 // (like the placement of the register window) the slots must be biased by
@@ -1593,12 +1631,12 @@ class ComputeMoveOrder: public StackObj {
 };
 
 static void verify_oop_args(MacroAssembler* masm,
-                            int total_args_passed,
+                            methodHandle method,
                             const BasicType* sig_bt,
                             const VMRegPair* regs) {
   Register temp_reg = rbx;  // not part of any compiled calling seq
   if (VerifyOops) {
-    for (int i = 0; i < total_args_passed; i++) {
+    for (int i = 0; i < method->size_of_parameters(); i++) {
       if (sig_bt[i] == T_OBJECT ||
           sig_bt[i] == T_ARRAY) {
         VMReg r = regs[i].first();
@@ -1615,35 +1653,32 @@ static void verify_oop_args(MacroAssembler* masm,
 }
 
 static void gen_special_dispatch(MacroAssembler* masm,
-                                 int total_args_passed,
-                                 int comp_args_on_stack,
-                                 vmIntrinsics::ID special_dispatch,
+                                 methodHandle method,
                                  const BasicType* sig_bt,
                                  const VMRegPair* regs) {
-  verify_oop_args(masm, total_args_passed, sig_bt, regs);
+  verify_oop_args(masm, method, sig_bt, regs);
+  vmIntrinsics::ID iid = method->intrinsic_id();
 
   // Now write the args into the outgoing interpreter space
   bool     has_receiver   = false;
   Register receiver_reg   = noreg;
   int      member_arg_pos = -1;
   Register member_reg     = noreg;
-  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(special_dispatch);
+  int      ref_kind       = MethodHandles::signature_polymorphic_intrinsic_ref_kind(iid);
   if (ref_kind != 0) {
-    member_arg_pos = total_args_passed - 1;  // trailing MemberName argument
+    member_arg_pos = method->size_of_parameters() - 1;  // trailing MemberName argument
     member_reg = rbx;  // known to be free at this point
     has_receiver = MethodHandles::ref_kind_has_receiver(ref_kind);
-  } else if (special_dispatch == vmIntrinsics::_invokeBasic) {
+  } else if (iid == vmIntrinsics::_invokeBasic) {
     has_receiver = true;
   } else {
-    guarantee(false, err_msg("special_dispatch=%d", special_dispatch));
+    fatal(err_msg_res("unexpected intrinsic id %d", iid));
   }
 
   if (member_reg != noreg) {
     // Load the member_arg into register, if necessary.
-    assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
-    assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+    SharedRuntime::check_member_name_argument_is_last_argument(method, sig_bt, regs);
     VMReg r = regs[member_arg_pos].first();
-    assert(r->is_valid(), "bad member arg");
     if (r->is_stack()) {
       __ movptr(member_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
     } else {
@@ -1654,7 +1689,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
 
   if (has_receiver) {
     // Make sure the receiver is loaded into a register.
-    assert(total_args_passed > 0, "oob");
+    assert(method->size_of_parameters() > 0, "oob");
     assert(sig_bt[0] == T_OBJECT, "receiver argument must be an object");
     VMReg r = regs[0].first();
     assert(r->is_valid(), "bad receiver arg");
@@ -1662,7 +1697,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
       // Porting note:  This assumes that compiled calling conventions always
       // pass the receiver oop in a register.  If this is not true on some
       // platform, pick a temp and load the receiver from stack.
-      assert(false, "receiver always in a register");
+      fatal("receiver always in a register");
       receiver_reg = j_rarg0;  // known to be free at this point
       __ movptr(receiver_reg, Address(rsp, r->reg2stack() * VMRegImpl::stack_slot_size + wordSize));
     } else {
@@ -1672,7 +1707,7 @@ static void gen_special_dispatch(MacroAssembler* masm,
   }
 
   // Figure out which address we are really jumping to:
-  MethodHandles::generate_method_handle_dispatch(masm, special_dispatch,
+  MethodHandles::generate_method_handle_dispatch(masm, iid,
                                                  receiver_reg, member_reg, /*for_compiler_entry:*/ true);
 }
 
@@ -1708,8 +1743,6 @@ static void gen_special_dispatch(MacroAssembler* masm,
 nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
                                                 methodHandle method,
                                                 int compile_id,
-                                                int total_in_args,
-                                                int comp_args_on_stack,
                                                 BasicType* in_sig_bt,
                                                 VMRegPair* in_regs,
                                                 BasicType ret_type) {
@@ -1718,9 +1751,7 @@ nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
     intptr_t start = (intptr_t)__ pc();
     int vep_offset = ((intptr_t)__ pc()) - start;
     gen_special_dispatch(masm,
-                         total_in_args,
-                         comp_args_on_stack,
-                         method->intrinsic_id(),
+                         method,
                          in_sig_bt,
                          in_regs);
     int frame_complete = ((intptr_t)__ pc()) - start;  // not complete, period
@@ -1754,6 +1785,7 @@ nmethod* SharedRuntime::generate_native_wrapper(MacroAssembler* masm,
   // we convert the java signature to a C signature by inserting
   // the hidden arguments as arg[0] and possibly arg[1] (static method)
 
+  const int total_in_args = method->size_of_parameters();
   int total_c_args = total_in_args;
   if (!is_critical_native) {
     total_c_args += 1;
@@ -3241,7 +3273,6 @@ uint SharedRuntime::out_preserve_stack_slots() {
   return 0;
 }
 
-
 //------------------------------generate_deopt_blob----------------------------
 void SharedRuntime::generate_deopt_blob() {
   // Allocate space for the code
@@ -3746,7 +3777,7 @@ void SharedRuntime::generate_uncommon_trap_blob() {
 // Generate a special Compile2Runtime blob that saves all registers,
 // and setup oopmap.
 //
-SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause_return) {
+SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, int poll_type) {
   assert(StubRoutines::forward_exception_entry() != NULL,
          "must be generated before");
 
@@ -3761,6 +3792,8 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   address start   = __ pc();
   address call_pc = NULL;
   int frame_size_in_words;
+  bool cause_return = (poll_type == POLL_AT_RETURN);
+  bool save_vectors = (poll_type == POLL_AT_VECTOR_LOOP);
 
   // Make room for return address (or push it again)
   if (!cause_return) {
@@ -3768,7 +3801,7 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   }
 
   // Save registers, fpu state, and flags
-  map = RegisterSaver::save_live_registers(masm, 0, &frame_size_in_words);
+  map = RegisterSaver::save_live_registers(masm, 0, &frame_size_in_words, save_vectors);
 
   // The following is basically a call_VM.  However, we need the precise
   // address of the call in order to generate an oopmap. Hence, we do all the
@@ -3805,7 +3838,7 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
 
   // Exception pending
 
-  RegisterSaver::restore_live_registers(masm);
+  RegisterSaver::restore_live_registers(masm, save_vectors);
 
   __ jump(RuntimeAddress(StubRoutines::forward_exception_entry()));
 
@@ -3813,7 +3846,7 @@ SafepointBlob* SharedRuntime::generate_handler_blob(address call_ptr, bool cause
   __ bind(noException);
 
   // Normal exit, restore registers and exit.
-  RegisterSaver::restore_live_registers(masm);
+  RegisterSaver::restore_live_registers(masm, save_vectors);
 
   __ ret(0);
 
diff --git a/src/cpu/x86/vm/vm_version_x86.cpp b/src/cpu/x86/vm/vm_version_x86.cpp
index 12f3f2584..bf7b3c213 100644
--- a/src/cpu/x86/vm/vm_version_x86.cpp
+++ b/src/cpu/x86/vm/vm_version_x86.cpp
@@ -363,6 +363,11 @@ void VM_Version::get_processor_features() {
   }
 
   _supports_cx8 = supports_cmpxchg8();
+  // xchg and xadd instructions
+  _supports_atomic_getset4 = true;
+  _supports_atomic_getadd4 = true;
+  LP64_ONLY(_supports_atomic_getset8 = true);
+  LP64_ONLY(_supports_atomic_getadd8 = true);
 
 #ifdef _LP64
   // OS should support SSE for x64 and hardware should support at least SSE2.
@@ -562,10 +567,10 @@ void VM_Version::get_processor_features() {
         AllocatePrefetchInstr = 3;
       }
       // On family 15h processors use XMM and UnalignedLoadStores for Array Copy
-      if( supports_sse2() && FLAG_IS_DEFAULT(UseXMMForArrayCopy) ) {
+      if (supports_sse2() && FLAG_IS_DEFAULT(UseXMMForArrayCopy)) {
         UseXMMForArrayCopy = true;
       }
-      if( FLAG_IS_DEFAULT(UseUnalignedLoadStores) && UseXMMForArrayCopy ) {
+      if (supports_sse2() && FLAG_IS_DEFAULT(UseUnalignedLoadStores)) {
         UseUnalignedLoadStores = true;
       }
     }
@@ -612,16 +617,16 @@ void VM_Version::get_processor_features() {
         MaxLoopPad = 11;
       }
 #endif // COMPILER2
-      if( FLAG_IS_DEFAULT(UseXMMForArrayCopy) ) {
+      if (FLAG_IS_DEFAULT(UseXMMForArrayCopy)) {
         UseXMMForArrayCopy = true; // use SSE2 movq on new Intel cpus
       }
-      if( supports_sse4_2() && supports_ht() ) { // Newest Intel cpus
-        if( FLAG_IS_DEFAULT(UseUnalignedLoadStores) && UseXMMForArrayCopy ) {
+      if (supports_sse4_2() && supports_ht()) { // Newest Intel cpus
+        if (FLAG_IS_DEFAULT(UseUnalignedLoadStores)) {
           UseUnalignedLoadStores = true; // use movdqu on newest Intel cpus
         }
       }
-      if( supports_sse4_2() && UseSSE >= 4 ) {
-        if( FLAG_IS_DEFAULT(UseSSE42Intrinsics)) {
+      if (supports_sse4_2() && UseSSE >= 4) {
+        if (FLAG_IS_DEFAULT(UseSSE42Intrinsics)) {
           UseSSE42Intrinsics = true;
         }
       }
@@ -638,6 +643,13 @@ void VM_Version::get_processor_features() {
     FLAG_SET_DEFAULT(UsePopCountInstruction, false);
   }
 
+#ifdef COMPILER2
+  if (FLAG_IS_DEFAULT(AlignVector)) {
+    // Modern processors allow misaligned memory operations for vectors.
+    AlignVector = !UseUnalignedLoadStores;
+  }
+#endif // COMPILER2
+
   assert(0 <= ReadPrefetchInstr && ReadPrefetchInstr <= 3, "invalid value");
   assert(0 <= AllocatePrefetchInstr && AllocatePrefetchInstr <= 3, "invalid value");
 
diff --git a/src/cpu/x86/vm/x86.ad b/src/cpu/x86/vm/x86.ad
index 71e26fb20..c3831006b 100644
--- a/src/cpu/x86/vm/x86.ad
+++ b/src/cpu/x86/vm/x86.ad
@@ -498,10 +498,18 @@ const bool Matcher::match_rule_supported(int opcode) {
     case Op_PopCountL:
       if (!UsePopCountInstruction)
         return false;
+    break;
     case Op_MulVI:
       if ((UseSSE < 4) && (UseAVX < 1)) // only with SSE4_1 or AVX
         return false;
     break;
+    case Op_CompareAndSwapL:
+#ifdef _LP64
+    case Op_CompareAndSwapP:
+#endif
+      if (!VM_Version::supports_cx8())
+        return false;
+    break;
   }
 
   return true;  // Per default match rules are supported.
diff --git a/src/cpu/x86/vm/x86_32.ad b/src/cpu/x86/vm/x86_32.ad
index e82d799ec..f5019cb59 100644
--- a/src/cpu/x86/vm/x86_32.ad
+++ b/src/cpu/x86/vm/x86_32.ad
@@ -7762,6 +7762,7 @@ instruct storeLConditional( memory mem, eADXRegL oldval, eBCXRegL newval, eFlags
 // No flag versions for CompareAndSwap{P,I,L} because matcher can't match them
 
 instruct compareAndSwapL( rRegI res, eSIRegP mem_ptr, eADXRegL oldval, eBCXRegL newval, eFlagsReg cr ) %{
+  predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapL mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
   format %{ "CMPXCHG8 [$mem_ptr],$newval\t# If EDX:EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t"
@@ -7798,6 +7799,47 @@ instruct compareAndSwapI( rRegI res, pRegP mem_ptr, eAXRegI oldval, eCXRegI newv
   ins_pipe( pipe_cmpxchg );
 %}
 
+instruct xaddI_no_res( memory mem, Universe dummy, immI add, eFlagsReg cr) %{
+  predicate(n->as_LoadStore()->result_not_used());
+  match(Set dummy (GetAndAddI mem add));
+  effect(KILL cr);
+  format %{ "ADDL  [$mem],$add" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ addl($mem$$Address, $add$$constant);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xaddI( memory mem, rRegI newval, eFlagsReg cr) %{
+  match(Set newval (GetAndAddI mem newval));
+  effect(KILL cr);
+  format %{ "XADDL  [$mem],$newval" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ xaddl($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgI( memory mem, rRegI newval) %{
+  match(Set newval (GetAndSetI mem newval));
+  format %{ "XCHGL  $newval,[$mem]" %}
+  ins_encode %{
+    __ xchgl($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgP( memory mem, pRegP newval) %{
+  match(Set newval (GetAndSetP mem newval));
+  format %{ "XCHGL  $newval,[$mem]" %}
+  ins_encode %{
+    __ xchgl($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
 //----------Subtraction Instructions-------------------------------------------
 // Integer Subtraction Instructions
 instruct subI_eReg(rRegI dst, rRegI src, eFlagsReg cr) %{
diff --git a/src/cpu/x86/vm/x86_64.ad b/src/cpu/x86/vm/x86_64.ad
index 8c0fb376f..dc2d4d89d 100644
--- a/src/cpu/x86/vm/x86_64.ad
+++ b/src/cpu/x86/vm/x86_64.ad
@@ -7242,6 +7242,7 @@ instruct compareAndSwapP(rRegI res,
                          rax_RegP oldval, rRegP newval,
                          rFlagsReg cr)
 %{
+  predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapP mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
@@ -7265,6 +7266,7 @@ instruct compareAndSwapL(rRegI res,
                          rax_RegL oldval, rRegL newval,
                          rFlagsReg cr)
 %{
+  predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapL mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
@@ -7329,6 +7331,88 @@ instruct compareAndSwapN(rRegI res,
   ins_pipe( pipe_cmpxchg );
 %}
 
+instruct xaddI_no_res( memory mem, Universe dummy, immI add, rFlagsReg cr) %{
+  predicate(n->as_LoadStore()->result_not_used());
+  match(Set dummy (GetAndAddI mem add));
+  effect(KILL cr);
+  format %{ "ADDL  [$mem],$add" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ addl($mem$$Address, $add$$constant);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xaddI( memory mem, rRegI newval, rFlagsReg cr) %{
+  match(Set newval (GetAndAddI mem newval));
+  effect(KILL cr);
+  format %{ "XADDL  [$mem],$newval" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ xaddl($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xaddL_no_res( memory mem, Universe dummy, immL add, rFlagsReg cr) %{
+  predicate(n->as_LoadStore()->result_not_used());
+  match(Set dummy (GetAndAddL mem add));
+  effect(KILL cr);
+  format %{ "ADDQ  [$mem],$add" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ addq($mem$$Address, $add$$constant);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xaddL( memory mem, rRegL newval, rFlagsReg cr) %{
+  match(Set newval (GetAndAddL mem newval));
+  effect(KILL cr);
+  format %{ "XADDQ  [$mem],$newval" %}
+  ins_encode %{
+    if (os::is_MP()) { __ lock(); }
+    __ xaddq($mem$$Address, $newval$$Register);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgI( memory mem, rRegI newval) %{
+  match(Set newval (GetAndSetI mem newval));
+  format %{ "XCHGL  $newval,[$mem]" %}
+  ins_encode %{
+    __ xchgl($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgL( memory mem, rRegL newval) %{
+  match(Set newval (GetAndSetL mem newval));
+  format %{ "XCHGL  $newval,[$mem]" %}
+  ins_encode %{
+    __ xchgq($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgP( memory mem, rRegP newval) %{
+  match(Set newval (GetAndSetP mem newval));
+  format %{ "XCHGQ  $newval,[$mem]" %}
+  ins_encode %{
+    __ xchgq($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct xchgN( memory mem, rRegN newval) %{
+  match(Set newval (GetAndSetN mem newval));
+  format %{ "XCHGL  $newval,$mem]" %}
+  ins_encode %{
+    __ xchgl($newval$$Register, $mem$$Address);
+  %}
+  ins_pipe( pipe_cmpxchg );
+%}
+
 //----------Subtraction Instructions-------------------------------------------
 
 // Integer Subtraction Instructions
diff --git a/src/share/tools/ProjectCreator/BuildConfig.java b/src/share/tools/ProjectCreator/BuildConfig.java
index 92fbf7746..00441dfb2 100644
--- a/src/share/tools/ProjectCreator/BuildConfig.java
+++ b/src/share/tools/ProjectCreator/BuildConfig.java
@@ -22,15 +22,14 @@
  *
  */
 
-import java.io.File;
 import java.util.Enumeration;
 import java.util.Hashtable;
-import java.util.Iterator;
 import java.util.Vector;
 
 class BuildConfig {
+    @SuppressWarnings("rawtypes")
     Hashtable vars;
-    Vector basicNames, basicPaths;
+    Vector<String> basicNames, basicPaths;
     String[] context;
 
     static CompilerInterface ci;
@@ -47,6 +46,7 @@ class BuildConfig {
         return ci;
     }
 
+    @SuppressWarnings("rawtypes")
     protected void initNames(String flavour, String build, String outDll) {
         if (vars == null) vars = new Hashtable();
 
@@ -63,26 +63,28 @@ class BuildConfig {
         // ones mentioned above were needed to expand format
         String buildBase = expandFormat(getFieldString(null, "BuildBase"));
         String sourceBase = getFieldString(null, "SourceBase");
+        String buildSpace = getFieldString(null, "BuildSpace");
         String outDir = buildBase;
 
         put("Id", flavourBuild);
         put("OutputDir", outDir);
         put("SourceBase", sourceBase);
         put("BuildBase", buildBase);
+        put("BuildSpace", buildSpace);
         put("OutputDll", outDir + Util.sep + outDll);
 
         context = new String [] {flavourBuild, flavour, build, null};
     }
 
-    protected void init(Vector includes, Vector defines) {
+    protected void init(Vector<String> includes, Vector<String> defines) {
         initDefaultDefines(defines);
         initDefaultCompilerFlags(includes);
         initDefaultLinkerFlags();
-        handleDB();
+        //handleDB();
     }
 
 
-    protected void initDefaultCompilerFlags(Vector includes) {
+    protected void initDefaultCompilerFlags(Vector<String> includes) {
         Vector compilerFlags = new Vector();
 
         compilerFlags.addAll(getCI().getBaseCompilerFlags(getV("Define"),
@@ -100,143 +102,48 @@ class BuildConfig {
         put("LinkerFlags", linkerFlags);
     }
 
-    DirectoryTree getSourceTree(String sourceBase, String startAt) {
-        DirectoryTree tree = new DirectoryTree();
-
-        tree.addSubdirToIgnore("Codemgr_wsdata");
-        tree.addSubdirToIgnore("deleted_files");
-        tree.addSubdirToIgnore("SCCS");
-        tree.setVerbose(true);
-        if (startAt != null) {
-            tree.readDirectory(sourceBase + File.separator + startAt);
-        } else {
-            tree.readDirectory(sourceBase);
-        }
-
-        return tree;
-    }
-
-
-    Vector getPreferredPaths() {
-        Vector preferredPaths = new Vector();
-
-        // In the case of multiple files with the same name in
-        // different subdirectories, prefer these versions
-        preferredPaths.add("windows");
-        preferredPaths.add("x86");
-        preferredPaths.add("closed");
-
-        // Also prefer "opto" over "adlc" for adlcVMDeps.hpp
-        preferredPaths.add("opto");
-
-        return preferredPaths;
-    }
-
-
-    void handleDB() {
-        WinGammaPlatform platform = (WinGammaPlatform)getField(null, "PlatformObject");
-
-        putSpecificField("AllFilesHash", computeAllFiles(platform));
-    }
-
-
-    private boolean matchesIgnoredPath(String prefixedName) {
-        Vector rv = new Vector();
+    public boolean matchesIgnoredPath(String path) {
+        Vector<String> rv = new Vector<String>();
         collectRelevantVectors(rv, "IgnorePath");
-        for (Iterator i = rv.iterator(); i.hasNext(); ) {
-            String pathPart = (String) i.next();
-            if (prefixedName.contains(Util.normalize(pathPart)))  {
+        for (String pathPart : rv) {
+            if (path.contains(pathPart))  {
                 return true;
             }
         }
         return false;
     }
 
-    void addAll(Iterator i, Hashtable hash,
-                WinGammaPlatform platform, DirectoryTree tree,
-                Vector preferredPaths, Vector filesNotFound, Vector filesDuplicate) {
-        for (; i.hasNext(); ) {
-            String fileName = (String) i.next();
-            if (lookupHashFieldInContext("IgnoreFile", fileName) == null) {
-                String prefixedName = platform.envVarPrefixedFileName(fileName,
-                                                                      0, /* ignored */
-                                                                      tree,
-                                                                      preferredPaths,
-                                                                      filesNotFound,
-                                                                      filesDuplicate);
-                if (prefixedName != null) {
-                    prefixedName = Util.normalize(prefixedName);
-                    if (!matchesIgnoredPath(prefixedName)) {
-                        addTo(hash, prefixedName, fileName);
-                    }
-                }
+    public boolean matchesHidePath(String path) {
+        Vector<String> rv = new Vector<String>();
+        collectRelevantVectors(rv, "HidePath");
+        for (String pathPart : rv) {
+            if (path.contains(Util.normalize(pathPart)))  {
+                return true;
             }
         }
+        return false;
     }
 
-    void addTo(Hashtable ht, String key, String value) {
-        ht.put(expandFormat(key), expandFormat(value));
-    }
-
-    Hashtable computeAllFiles(WinGammaPlatform platform) {
-        Hashtable rv = new Hashtable();
-        DirectoryTree tree = getSourceTree(get("SourceBase"), getFieldString(null, "StartAt"));
-        Vector preferredPaths = getPreferredPaths();
-
-        // Hold errors until end
-        Vector filesNotFound = new Vector();
-        Vector filesDuplicate = new Vector();
+   public Vector<String> matchesAdditionalGeneratedPath(String fullPath) {
+        Vector<String> rv = new Vector<String>();
+        Hashtable<String, String> v = (Hashtable<String, String>)BuildConfig.getField(this.toString(), "AdditionalGeneratedFile");
+        if (v != null) {
+            for (Enumeration<String> e=v.keys(); e.hasMoreElements(); ) {
+                String key = e.nextElement();
+                String val = v.get(key);
 
-        Vector includedFiles = new Vector();
-
-        // find all files
-        Vector dirs = getSourceIncludes();
-        for (Iterator i = dirs.iterator(); i.hasNext(); ) {
-            String dir = (String)i.next();
-            DirectoryTree subtree = getSourceTree(dir, null);
-            for (Iterator fi = subtree.getFileIterator(); fi.hasNext(); ) {
-                String name = ((File)fi.next()).getName();
-                includedFiles.add(name);
-            }
-        }
-        addAll(includedFiles.iterator(), rv,
-               platform, tree,
-               preferredPaths, filesNotFound, filesDuplicate);
-
-        Vector addFiles = new Vector();
-        collectRelevantVectors(addFiles, "AdditionalFile");
-        addAll(addFiles.iterator(), rv,
-               platform, tree,
-               preferredPaths, filesNotFound, filesDuplicate);
-
-        collectRelevantHashes(rv, "AdditionalGeneratedFile");
-
-        if ((filesNotFound.size() != 0) ||
-            (filesDuplicate.size() != 0)) {
-            System.err.println("Error: some files were not found or " +
-                               "appeared in multiple subdirectories of " +
-                               "directory " + get("SourceBase") + " and could not " +
-                               "be resolved with os_family and arch.");
-            if (filesNotFound.size() != 0) {
-                System.err.println("Files not found:");
-                for (Iterator iter = filesNotFound.iterator();
-                     iter.hasNext(); ) {
-                    System.err.println("  " + (String) iter.next());
-                }
-            }
-            if (filesDuplicate.size() != 0) {
-                System.err.println("Duplicate files:");
-                for (Iterator iter = filesDuplicate.iterator();
-                     iter.hasNext(); ) {
-                    System.err.println("  " + (String) iter.next());
+                if (fullPath.endsWith(expandFormat(key))) {
+                    rv.add(expandFormat(val));
                 }
             }
-            throw new RuntimeException();
         }
-
         return rv;
     }
 
+    void addTo(Hashtable ht, String key, String value) {
+        ht.put(expandFormat(key), expandFormat(value));
+    }
+
     void initDefaultDefines(Vector defines) {
         Vector sysDefines = new Vector();
         sysDefines.add("WIN32");
@@ -324,20 +231,19 @@ class BuildConfig {
     }
 
     void collectRelevantVectors(Vector rv, String field) {
-        for (int i = 0; i < context.length; i++) {
-            Vector v = getFieldVector(context[i], field);
+        for (String ctx : context) {
+            Vector<String> v = getFieldVector(ctx, field);
             if (v != null) {
-                for (Iterator j=v.iterator(); j.hasNext(); ) {
-                    String val = (String)j.next();
-                    rv.add(expandFormat(val));
+                for (String val : v) {
+                    rv.add(expandFormat(val).replace('/', '\\'));
                 }
             }
         }
     }
 
     void collectRelevantHashes(Hashtable rv, String field) {
-        for (int i = 0; i < context.length; i++) {
-            Hashtable v = (Hashtable)getField(context[i], field);
+        for (String ctx : context) {
+            Hashtable v = (Hashtable)getField(ctx, field);
             if (v != null) {
                 for (Enumeration e=v.keys(); e.hasMoreElements(); ) {
                     String key = (String)e.nextElement();
@@ -357,21 +263,17 @@ class BuildConfig {
 
     Vector getIncludes() {
         Vector rv = new Vector();
-
         collectRelevantVectors(rv, "AbsoluteInclude");
-
         rv.addAll(getSourceIncludes());
-
         return rv;
     }
 
     private Vector getSourceIncludes() {
-        Vector rv = new Vector();
-        Vector ri = new Vector();
+        Vector<String> rv = new Vector<String>();
+        Vector<String> ri = new Vector<String>();
         String sourceBase = getFieldString(null, "SourceBase");
         collectRelevantVectors(ri, "RelativeInclude");
-        for (Iterator i = ri.iterator(); i.hasNext(); ) {
-            String f = (String)i.next();
+        for (String f : ri) {
             rv.add(sourceBase + Util.sep + f);
         }
         return rv;
@@ -604,7 +506,6 @@ class TieredFastDebugConfig extends GenericDebugNonKernelConfig {
     }
 }
 
-
 abstract class ProductConfig extends BuildConfig {
     protected void init(Vector includes, Vector defines) {
         defines.add("NDEBUG");
@@ -638,7 +539,6 @@ class TieredProductConfig extends ProductConfig {
     }
 }
 
-
 class CoreDebugConfig extends GenericDebugNonKernelConfig {
     String getOptFlag() {
         return getCI().getNoOptFlag();
@@ -650,7 +550,6 @@ class CoreDebugConfig extends GenericDebugNonKernelConfig {
     }
 }
 
-
 class CoreFastDebugConfig extends GenericDebugNonKernelConfig {
     String getOptFlag() {
         return getCI().getOptFlag();
@@ -662,7 +561,6 @@ class CoreFastDebugConfig extends GenericDebugNonKernelConfig {
     }
 }
 
-
 class CoreProductConfig extends ProductConfig {
     CoreProductConfig() {
         initNames("core", "product", "jvm.dll");
@@ -700,6 +598,7 @@ class KernelProductConfig extends ProductConfig {
         init(getIncludes(), getDefines());
     }
 }
+
 abstract class CompilerInterface {
     abstract Vector getBaseCompilerFlags(Vector defines, Vector includes, String outDir);
     abstract Vector getBaseLinkerFlags(String outDir, String outDll, String platformName);
diff --git a/src/share/tools/ProjectCreator/DirectoryTree.java b/src/share/tools/ProjectCreator/DirectoryTree.java
deleted file mode 100644
index 76a5e4668..000000000
--- a/src/share/tools/ProjectCreator/DirectoryTree.java
+++ /dev/null
@@ -1,287 +0,0 @@
-/*
- * Copyright (c) 1999, 2011, Oracle and/or its affiliates. All rights reserved.
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This code is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-/** Encapsulates a notion of a directory tree. Designed to allow fast
-    querying of full paths for unique filenames in the hierarchy. */
-
-import java.io.*;
-import java.util.*;
-
-public class DirectoryTree {
-
-    /** The root of the read directoryTree */
-    private Node rootNode;
-
-    /** Subdirs to ignore; Vector of Strings */
-    private Vector subdirsToIgnore;
-
-    /** This maps file names to Lists of nodes. */
-    private Hashtable nameToNodeListTable;
-
-    /** Output "."'s as directories are read. Defaults to false. */
-    private boolean verbose;
-
-    public DirectoryTree() {
-        subdirsToIgnore = new Vector();
-        verbose = false;
-    }
-
-    public void addSubdirToIgnore(String subdir) {
-        subdirsToIgnore.add(subdir);
-    }
-
-    private class FileIterator implements Iterator {
-        private Vector nodes = new Vector();
-
-        public FileIterator(Node rootNode) {
-            if(rootNode == null) {
-                return;
-            }
-            nodes.add(rootNode);
-            prune();
-        }
-        public boolean hasNext() {
-            return nodes.size() > 0;
-        }
-        public Object next() {
-            Node last = (Node)nodes.remove(nodes.size() - 1);
-            prune();
-            return new File(last.getName());
-        }
-
-        public void remove() {
-            throw new RuntimeException();
-        }
-
-        private void prune() {
-            while (nodes.size() > 0) {
-                Node last = (Node)nodes.get(nodes.size() - 1);
-
-                if (last.isDirectory()) {
-                    nodes.remove(nodes.size() - 1);
-                    nodes.addAll(last.children);
-                } else {
-                    // Is at file
-                    return;
-                }
-            }
-        }
-    }
-
-    public Iterator getFileIterator() {
-        return new FileIterator(rootNode);
-    }
-
-    /** Output "."'s to System.out as directories are read. Defaults
-        to false. */
-    public void setVerbose(boolean newValue) {
-        verbose = newValue;
-    }
-
-    public boolean getVerbose() {
-        return verbose;
-    }
-
-    public String getRootNodeName() {
-        return rootNode.getName();
-    }
-
-    /** Takes an absolute path to the root directory of this
-        DirectoryTree. Throws IllegalArgumentException if the given
-        string represents a plain file or nonexistent directory. */
-
-    public void readDirectory(String baseDirectory)
-        throws IllegalArgumentException {
-        File root = new File(Util.normalize(baseDirectory));
-        if (!root.isDirectory()) {
-            return;
-        }
-        try {
-            root = root.getCanonicalFile();
-        }
-        catch (IOException e) {
-            throw new RuntimeException(e.toString());
-        }
-        rootNode = new Node(root);
-        readDirectory(rootNode, root);
-    }
-
-    /** Queries the DirectoryTree for a file or directory name. Takes
-        only the name of the file or directory itself (i.e., no parent
-        directory information should be in the passed name). Returns a
-        List of DirectoryTreeNodes specifying the full paths of all of
-        the files or directories of this name in the DirectoryTree.
-        Returns null if the directory tree has not been read from disk
-        yet or if the file was not found in the tree. */
-    public List findFile(String name) {
-        if (rootNode == null) {
-            return null;
-        }
-
-        if (nameToNodeListTable == null) {
-            nameToNodeListTable = new Hashtable();
-            try {
-                buildNameToNodeListTable(rootNode);
-            } catch (IOException e) {
-                e.printStackTrace();
-                return null;
-            }
-        }
-
-        return (List) nameToNodeListTable.get(name);
-    }
-
-    private void buildNameToNodeListTable(Node curNode)
-      throws IOException {
-        String fullName = curNode.getName();
-        String parent = curNode.getParent();
-        String separator = System.getProperty("file.separator");
-
-        if (parent != null) {
-          if (!fullName.startsWith(parent)) {
-            throw new RuntimeException(
-                "Internal error: parent of file name \"" + fullName +
-                "\" does not match file name \"" + parent + "\""
-            );
-          }
-
-          int len = parent.length();
-          if (!parent.endsWith(separator)) {
-            len += separator.length();
-          }
-
-          String fileName = fullName.substring(len);
-
-          if (fileName == null) {
-            throw new RuntimeException(
-                "Internal error: file name was empty"
-            );
-          }
-
-          List nodeList = (List) nameToNodeListTable.get(fileName);
-          if (nodeList == null) {
-            nodeList = new Vector();
-            nameToNodeListTable.put(fileName, nodeList);
-          }
-
-          nodeList.add(curNode);
-        } else {
-          if (curNode != rootNode) {
-            throw new RuntimeException(
-                "Internal error: parent of file + \"" + fullName + "\"" +
-                " was null"
-            );
-          }
-        }
-
-        if (curNode.isDirectory()) {
-          Iterator iter = curNode.getChildren();
-          if (iter != null) {
-            while (iter.hasNext()) {
-              buildNameToNodeListTable((Node) iter.next());
-            }
-          }
-        }
-    }
-
-    /** Reads all of the files in the given directory and adds them as
-        children of the directory tree node. Requires that the passed
-        node represents a directory. */
-
-    private void readDirectory(Node parentNode, File parentDir) {
-        File[] children = parentDir.listFiles();
-        if (children == null)
-            return;
-        if (verbose) {
-            System.out.print(".");
-            System.out.flush();
-        }
-        for (int i = 0; i < children.length; i++) {
-            File child = children[i];
-            children[i] = null;
-            boolean isDir = child.isDirectory();
-            boolean mustSkip = false;
-            if (isDir) {
-                for (Iterator iter = subdirsToIgnore.iterator();
-                     iter.hasNext(); ) {
-                    if (child.getName().equals((String) iter.next())) {
-                        mustSkip = true;
-                        break;
-                    }
-                }
-            }
-            if (!mustSkip) {
-                Node childNode = new Node(child);
-                parentNode.addChild(childNode);
-                if (isDir) {
-                    readDirectory(childNode, child);
-                }
-            }
-        }
-    }
-
-    private class Node implements DirectoryTreeNode {
-        private File file;
-        private Vector children;
-
-        /** file must be a canonical file */
-        Node(File file) {
-            this.file = file;
-            children = new Vector();
-        }
-
-        public boolean isFile() {
-            return file.isFile();
-        }
-
-        public boolean isDirectory() {
-            return file.isDirectory();
-        }
-
-        public String getName() {
-            return file.getPath();
-        }
-
-        public String getParent() {
-            return file.getParent();
-        }
-
-        public void addChild(Node n) {
-            children.add(n);
-        }
-
-        public Iterator getChildren() throws IllegalArgumentException {
-            return children.iterator();
-        }
-
-        public int getNumChildren() throws IllegalArgumentException {
-            return children.size();
-        }
-
-        public DirectoryTreeNode getChild(int i)
-            throws IllegalArgumentException, ArrayIndexOutOfBoundsException {
-            return (DirectoryTreeNode) children.get(i);
-        }
-    }
-}
diff --git a/src/share/tools/ProjectCreator/DirectoryTreeNode.java b/src/share/tools/ProjectCreator/DirectoryTreeNode.java
deleted file mode 100644
index f198dc24b..000000000
--- a/src/share/tools/ProjectCreator/DirectoryTreeNode.java
+++ /dev/null
@@ -1,36 +0,0 @@
-/*
- * Copyright (c) 1999, 2010, 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.
- *
- */
-
-import java.util.*;
-
-public interface DirectoryTreeNode {
-    public boolean isFile();
-    public boolean isDirectory();
-    public String getName();
-    public String getParent();
-    public Iterator getChildren() throws IllegalArgumentException;
-    public int getNumChildren() throws IllegalArgumentException;
-    public DirectoryTreeNode getChild(int i)
-        throws IllegalArgumentException, ArrayIndexOutOfBoundsException;
-}
diff --git a/src/share/tools/ProjectCreator/FileFormatException.java b/src/share/tools/ProjectCreator/FileFormatException.java
deleted file mode 100644
index 077886f85..000000000
--- a/src/share/tools/ProjectCreator/FileFormatException.java
+++ /dev/null
@@ -1,35 +0,0 @@
-/*
- * Copyright (c) 1999, 2011, Oracle and/or its affiliates. All rights reserved.
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This code is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-@SuppressWarnings("serial")
-public class FileFormatException extends Exception {
-
-    public FileFormatException() {
-        super();
-    }
-
-    public FileFormatException(String s) {
-        super(s);
-    }
-}
diff --git a/src/share/tools/ProjectCreator/FileTreeCreator.java b/src/share/tools/ProjectCreator/FileTreeCreator.java
new file mode 100644
index 000000000..3643c572e
--- /dev/null
+++ b/src/share/tools/ProjectCreator/FileTreeCreator.java
@@ -0,0 +1,72 @@
+import java.nio.file.FileSystems;
+import java.nio.file.Path;
+import java.nio.file.SimpleFileVisitor;
+import java.util.HashSet;
+import java.util.Stack;
+import java.util.Vector;
+
+public class FileTreeCreator extends SimpleFileVisitor<Path>
+{
+   Path vcProjLocation;
+   Path startDir;
+   final int startDirLength;
+   Stack<DirAttributes> attributes = new Stack<DirAttributes>();
+   Vector<BuildConfig> allConfigs;
+   WinGammaPlatformVC10 wg;
+
+   public FileTreeCreator(Path startDir, Vector<BuildConfig> allConfigs, WinGammaPlatformVC10 wg) {
+      super();
+      this.wg = wg;
+      this.allConfigs = allConfigs;
+      this.startDir = startDir;
+      startDirLength = startDir.toAbsolutePath().toString().length();
+      vcProjLocation = FileSystems.getDefault().getPath(allConfigs.firstElement().get("BuildSpace"));
+      attributes.push(new DirAttributes());
+   }
+
+   public class DirAttributes {
+
+      private HashSet<BuildConfig> ignores;
+      private HashSet<BuildConfig> disablePch;
+
+      public DirAttributes() {
+         ignores = new HashSet<BuildConfig>();
+         disablePch = new HashSet<BuildConfig>();
+      }
+
+      public DirAttributes(HashSet<BuildConfig> excludes2, HashSet<BuildConfig> disablePch2) {
+         ignores = excludes2;
+         disablePch = disablePch2;
+      }
+
+      @SuppressWarnings("unchecked")
+      public DirAttributes clone() {
+         return new DirAttributes((HashSet<BuildConfig>)this.ignores.clone(), (HashSet<BuildConfig>)this.disablePch.clone());
+      }
+
+      public void setIgnore(BuildConfig conf) {
+         ignores.add(conf);
+      }
+
+      public boolean hasIgnore(BuildConfig cfg) {
+         return ignores.contains(cfg);
+      }
+
+      public void removeFromIgnored(BuildConfig cfg) {
+         ignores.remove(cfg);
+      }
+
+      public void setDisablePch(BuildConfig conf) {
+         disablePch.add(conf);
+      }
+
+      public boolean hasDisablePch(BuildConfig cfg) {
+         return disablePch.contains(cfg);
+      }
+
+      public void removeFromDisablePch(BuildConfig cfg) {
+         disablePch.remove(cfg);
+      }
+
+   }
+}
diff --git a/src/share/tools/ProjectCreator/FileTreeCreatorVC10.java b/src/share/tools/ProjectCreator/FileTreeCreatorVC10.java
new file mode 100644
index 000000000..837eef1a5
--- /dev/null
+++ b/src/share/tools/ProjectCreator/FileTreeCreatorVC10.java
@@ -0,0 +1,142 @@
+import static java.nio.file.FileVisitResult.CONTINUE;
+
+import java.io.IOException;
+import java.nio.file.FileSystems;
+import java.nio.file.FileVisitResult;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.attribute.BasicFileAttributes;
+import java.util.Stack;
+import java.util.Vector;
+
+public class FileTreeCreatorVC10 extends FileTreeCreator {
+
+      public FileTreeCreatorVC10(Path startDir, Vector<BuildConfig> allConfigs, WinGammaPlatformVC10 wg) {
+         super(startDir, allConfigs, wg);
+      }
+
+      @Override
+      public FileVisitResult visitFile(Path file, BasicFileAttributes attr) {
+         DirAttributes currentFileAttr = attributes.peek().clone();
+         boolean usePch = false;
+         boolean disablePch = false;
+         boolean useIgnore = false;
+         String fileName = file.getFileName().toString();
+
+         // TODO hideFile
+
+         // usePch applies to all configs for a file.
+         if (fileName.equals(BuildConfig.getFieldString(null, "UseToGeneratePch"))) {
+            usePch = true;
+         }
+
+         for (BuildConfig cfg : allConfigs) {
+            if (cfg.lookupHashFieldInContext("IgnoreFile", fileName) != null) {
+               useIgnore = true;
+               currentFileAttr.setIgnore(cfg);
+            } else if (cfg.matchesIgnoredPath(file.toAbsolutePath().toString())) {
+               useIgnore = true;
+               currentFileAttr.setIgnore(cfg);
+            }
+
+            if (cfg.lookupHashFieldInContext("DisablePch", fileName) != null) {
+               disablePch = true;
+               currentFileAttr.setDisablePch(cfg);
+            }
+
+            Vector<String> rv = new Vector<String>();
+            cfg.collectRelevantVectors(rv, "AdditionalFile");
+            for(String addFile : rv) {
+               if (addFile.equals(fileName)) {
+                  // supress any ignore
+                  // TODO - may need some adjustments
+                  if (file.toAbsolutePath().toString().contains(cfg.get("Flavour"))) {
+                     currentFileAttr.removeFromIgnored(cfg);
+                  }
+               }
+            }
+         }
+
+         String tagName = wg.getFileTagFromSuffix(fileName);
+         String fileLoc = vcProjLocation.relativize(file).toString();
+
+         if (!useIgnore && !disablePch && !usePch) {
+            wg.tag(tagName, new String[] { "Include", fileLoc});
+         } else {
+            wg.startTag(
+                  tagName,
+                  new String[] { "Include", fileLoc});
+
+            for (BuildConfig cfg : allConfigs) {
+               boolean ignore = currentFileAttr.hasIgnore(cfg);
+               if (ignore) {
+                  wg.tagData("ExcludedFromBuild", "true", "Condition", "'$(Configuration)|$(Platform)'=='" + cfg.get("Name") + "'");
+               }
+               if (usePch) {
+                  wg.tagData("PrecompiledHeader", "Create", "Condition", "'$(Configuration)|$(Platform)'=='" + cfg.get("Name") + "'");
+               }
+               if (disablePch) {
+                  wg.tag("PrecompiledHeader", "Condition", "'$(Configuration)|$(Platform)'=='" + cfg.get("Name") + "'");
+               }
+            }
+            wg.endTag();
+         }
+
+         String filter = startDir.relativize(file.getParent().toAbsolutePath()).toString();
+         wg.addFilterDependency(fileLoc, filter);
+
+         return CONTINUE;
+      }
+
+      @Override
+      public FileVisitResult preVisitDirectory(Path path, BasicFileAttributes attrs)
+            throws IOException {
+         Boolean hide = false;
+         // TODO remove attrs, if path is matched in this dir, then it is too in every subdir.
+         // And we will check anyway
+         DirAttributes newAttr = attributes.peek().clone();
+
+         // check per config ignorePaths!
+         for (BuildConfig cfg : allConfigs) {
+            if (cfg.matchesIgnoredPath(path.toAbsolutePath().toString())) {
+               newAttr.setIgnore(cfg);
+            }
+
+            // Hide is always on all configs. And additional files are never hiddden
+            if (cfg.matchesHidePath(path.toAbsolutePath().toString())) {
+               hide = true;
+               break;
+            }
+         }
+
+         if (!hide) {
+            String name = startDir.relativize(path.toAbsolutePath()).toString();
+            if (!"".equals(name)) {
+               wg.addFilter(name);
+            }
+
+            attributes.push(newAttr);
+            return super.preVisitDirectory(path, attrs);
+         } else {
+            return FileVisitResult.SKIP_SUBTREE;
+         }
+      }
+
+      @Override
+      public FileVisitResult postVisitDirectory(Path dir, IOException exc) {
+         //end matching attributes set by ignorepath
+         attributes.pop();
+         return CONTINUE;
+      }
+
+      @Override
+      public FileVisitResult visitFileFailed(Path file, IOException exc) {
+         return CONTINUE;
+      }
+
+      public void writeFileTree() throws IOException {
+         Files.walkFileTree(this.startDir, this);
+      }
+
+
+   }
+\ No newline at end of file
diff --git a/src/share/tools/ProjectCreator/FileTreeCreatorVC7.java b/src/share/tools/ProjectCreator/FileTreeCreatorVC7.java
new file mode 100644
index 000000000..b36e0121f
--- /dev/null
+++ b/src/share/tools/ProjectCreator/FileTreeCreatorVC7.java
@@ -0,0 +1,156 @@
+import static java.nio.file.FileVisitResult.CONTINUE;
+
+import java.io.IOException;
+import java.nio.file.FileSystems;
+import java.nio.file.FileVisitResult;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.attribute.BasicFileAttributes;
+import java.util.Stack;
+import java.util.Vector;
+
+public class FileTreeCreatorVC7 extends FileTreeCreator {
+
+      public FileTreeCreatorVC7(Path startDir, Vector<BuildConfig> allConfigs, WinGammaPlatform wg) {
+         super(startDir, allConfigs, null);
+      }
+
+      @Override
+      public FileVisitResult visitFile(Path file, BasicFileAttributes attr) {
+         DirAttributes currentFileAttr = attributes.peek().clone();
+         boolean usePch = false;
+         boolean disablePch = false;
+         boolean useIgnore = false;
+         String fileName = file.getFileName().toString();
+
+         // usePch applies to all configs for a file.
+         if (fileName.equals(BuildConfig.getFieldString(null, "UseToGeneratePch"))) {
+            usePch = true;
+         }
+
+         for (BuildConfig cfg : allConfigs) {
+            if (cfg.lookupHashFieldInContext("IgnoreFile", fileName) != null) {
+               useIgnore = true;
+               currentFileAttr.setIgnore(cfg);
+            } else if (cfg.matchesIgnoredPath(file.toAbsolutePath().toString())) {
+               useIgnore = true;
+               currentFileAttr.setIgnore(cfg);
+            }
+
+            if (cfg.lookupHashFieldInContext("DisablePch", fileName) != null) {
+               disablePch = true;
+               currentFileAttr.setDisablePch(cfg);
+            }
+
+            Vector<String> rv = new Vector<String>();
+            cfg.collectRelevantVectors(rv, "AdditionalFile");
+            for(String addFile : rv) {
+               if (addFile.equals(fileName)) {
+                  // supress any ignore
+                  currentFileAttr.removeFromIgnored(cfg);
+               }
+            }
+         }
+
+         if (!useIgnore && !disablePch && !usePch) {
+            wg.tag("File", new String[] { "RelativePath", vcProjLocation.relativize(file).toString()});
+         } else {
+            wg.startTag(
+                  "File",
+                  new String[] { "RelativePath", vcProjLocation.relativize(file).toString()});
+
+            for (BuildConfig cfg : allConfigs) {
+               boolean ignore = currentFileAttr.hasIgnore(cfg);
+               String [] fileConfAttr;
+
+               if (ignore) {
+                  fileConfAttr = new String[] {"Name", cfg.get("Name"), "ExcludedFromBuild", "TRUE" };
+               } else {
+                  fileConfAttr = new String[] {"Name", cfg.get("Name")};
+               }
+
+               if (!disablePch && !usePch && !ignore) {
+                  continue;
+               } else if (!disablePch && !usePch) {
+                  wg.tag("FileConfiguration", fileConfAttr);
+               } else {
+                  wg.startTag("FileConfiguration", fileConfAttr);
+                  if (usePch) {
+                     // usePch always applies to all configs, might not always be so.
+                     wg.tag("Tool", new String[] {
+                           "Name", "VCCLCompilerTool", "UsePrecompiledHeader",
+                     "1" });
+                     assert(!disablePch);
+                  }
+                  if (disablePch) {
+                     if (currentFileAttr.hasDisablePch(cfg)) {
+                        wg.tag("Tool", new String[] {
+                              "Name", "VCCLCompilerTool", "UsePrecompiledHeader",
+                        "0" });
+                     }
+                     assert(!usePch);
+                  }
+                  wg.endTag();
+               }
+            }
+            wg.endTag();
+         }
+
+         return CONTINUE;
+      }
+
+      @Override
+      public FileVisitResult preVisitDirectory(Path path, BasicFileAttributes attrs)
+            throws IOException {
+         Boolean hide = false;
+         DirAttributes newAttr = attributes.peek().clone();
+
+         String rPath;
+         if (path.toAbsolutePath().toString().equals(this.startDir.toAbsolutePath().toString())){
+            rPath = startDir.toString();
+         } else {
+            rPath = path.getFileName().toString();
+         }
+
+         // check per config ignorePaths!
+         for (BuildConfig cfg : allConfigs) {
+            if (cfg.matchesIgnoredPath(path.toAbsolutePath().toString())) {
+               newAttr.setIgnore(cfg);
+            }
+
+            // Hide is always on all configs. And additional files are never hiddden
+            if (cfg.matchesHidePath(path.toAbsolutePath().toString())) {
+               hide = true;
+               break;
+            }
+         }
+
+         if (!hide) {
+            wg.startTag("Filter", new String[] {
+                  "Name", rPath});
+
+            attributes.push(newAttr);
+            return super.preVisitDirectory(path, attrs);
+         } else {
+            return FileVisitResult.SKIP_SUBTREE;
+         }
+      }
+
+      @Override
+      public FileVisitResult postVisitDirectory(Path dir, IOException exc) {
+         //end matching attributes set by ignorepath
+         wg.endTag();
+         attributes.pop();
+
+         return CONTINUE;
+      }
+
+      @Override
+      public FileVisitResult visitFileFailed(Path file, IOException exc) {
+         return CONTINUE;
+      }
+
+      public void writeFileTree() throws IOException {
+         Files.walkFileTree(this.startDir, this);
+      }
+   }
+\ No newline at end of file
diff --git a/src/share/tools/ProjectCreator/ProjectCreator.java b/src/share/tools/ProjectCreator/ProjectCreator.java
index 5203c5a36..93a7b1dca 100644
--- a/src/share/tools/ProjectCreator/ProjectCreator.java
+++ b/src/share/tools/ProjectCreator/ProjectCreator.java
@@ -24,75 +24,76 @@
 
 public class ProjectCreator {
 
-    public static void usage() {
-        System.out.println("ProjectCreator options:");
-        System.err.println("WinGammaPlatform platform-specific options:");
-        System.err.println("  -sourceBase <path to directory (workspace) " +
-                           "containing source files; no trailing slash>");
-        System.err.println("  -dspFileName <full pathname to which .dsp file " +
-                           "will be written; all parent directories must " +
-                           "already exist>");
-        System.err.println("  -envVar <environment variable to be inserted " +
-                           "into .dsp file, substituting for path given in " +
-                           "-sourceBase. Example: HotSpotWorkSpace>");
-        System.err.println("  -dllLoc <path to directory in which to put " +
-                           "jvm.dll and jvm_g.dll; no trailing slash>");
-        System.err.println("  If any of the above are specified, "+
-                           "they must all be.");
-        System.err.println("  Additional, optional arguments, which can be " +
-                           "specified multiple times:");
-        System.err.println("    -absoluteInclude <string containing absolute " +
-                           "path to include directory>");
-        System.err.println("    -relativeInclude <string containing include " +
-                           "directory relative to -envVar>");
-        System.err.println("    -define <preprocessor flag to be #defined " +
-                           "(note: doesn't yet support " +
-                           "#define (flag) (value))>");
-        System.err.println("    -perFileLine <file> <line>");
-        System.err.println("    -conditionalPerFileLine <file> <line for " +
-                           "release build> <line for debug build>");
-        System.err.println("  (NOTE: To work around a bug in nmake, where " +
-                           "you can't have a '#' character in a quoted " +
-                           "string, all of the lines outputted have \"#\"" +
-                           "prepended)");
-        System.err.println("    -startAt <subdir of sourceBase>");
-        System.err.println("    -ignoreFile <file which won't be able to be " +
-                           "found in the sourceBase because it's generated " +
-                           "later>");
-        System.err.println("    -additionalFile <file not in database but " +
-                           "which should show up in .dsp file>");
-        System.err.println("    -additionalGeneratedFile <environment variable of " +
-                           "generated file's location> <relative path to " +
-                           "directory containing file; no trailing slash> " +
-                           "<name of file generated later in the build process>");
-        System.err.println("    -prelink <build> <desc> <cmds>:");
-        System.err.println(" Generate a set of prelink commands for the given BUILD");
-        System.err.println(" (\"Debug\" or \"Release\"). The prelink description and commands");
-        System.err.println(" are both quoted strings.");
-        System.err.println("    Default includes: \".\"");
-        System.err.println("    Default defines: WIN32, _WINDOWS, \"HOTSPOT_BUILD_USER=$(USERNAME)\"");
-    }
+   public static void usage() {
+      System.out.println("ProjectCreator options:");
+      System.err.println("WinGammaPlatform platform-specific options:");
+      System.err.println("  -sourceBase <path to directory (workspace) "
+            + "containing source files; no trailing slash>");
+      System.err.println("  -dspFileName <full pathname to which .dsp file "
+            + "will be written; all parent directories must "
+            + "already exist>");
+      System.err.println("  -envVar <environment variable to be inserted "
+            + "into .dsp file, substituting for path given in "
+            + "-sourceBase. Example: HotSpotWorkSpace>");
+      System.err.println("  -dllLoc <path to directory in which to put "
+            + "jvm.dll and jvm_g.dll; no trailing slash>");
+      System.err.println("  If any of the above are specified, "
+            + "they must all be.");
+      System.err.println("  Additional, optional arguments, which can be "
+            + "specified multiple times:");
+      System.err.println("    -absoluteInclude <string containing absolute "
+            + "path to include directory>");
+      System.err.println("    -relativeInclude <string containing include "
+            + "directory relative to -envVar>");
+      System.err.println("    -define <preprocessor flag to be #defined "
+            + "(note: doesn't yet support " + "#define (flag) (value))>");
+      System.err.println("    -perFileLine <file> <line>");
+      System.err.println("    -conditionalPerFileLine <file> <line for "
+            + "release build> <line for debug build>");
+      System.err.println("  (NOTE: To work around a bug in nmake, where "
+            + "you can't have a '#' character in a quoted "
+            + "string, all of the lines outputted have \"#\"" + "prepended)");
+      System.err.println("    -startAt <subdir of sourceBase>");
+      System.err.println("    -ignoreFile <file which won't be able to be "
+            + "found in the sourceBase because it's generated " + "later>");
+      System.err.println("    -additionalFile <file not in database but "
+            + "which should show up in .dsp file>");
+      System.err
+            .println("    -additionalGeneratedFile <environment variable of "
+                  + "generated file's location> <relative path to "
+                  + "directory containing file; no trailing slash> "
+                  + "<name of file generated later in the build process>");
+      System.err.println("    -prelink <build> <desc> <cmds>:");
+      System.err
+            .println(" Generate a set of prelink commands for the given BUILD");
+      System.err
+            .println(" (\"Debug\" or \"Release\"). The prelink description and commands");
+      System.err.println(" are both quoted strings.");
+      System.err.println("    Default includes: \".\"");
+      System.err
+            .println("    Default defines: WIN32, _WINDOWS, \"HOTSPOT_BUILD_USER=$(USERNAME)\"");
+   }
 
-    public static void main(String[] args) {
-        try {
-            if (args.length < 3) {
-                usage();
-                System.exit(1);
-            }
+   public static void main(String[] args) {
+      try {
+         if (args.length < 3) {
+            usage();
+            System.exit(1);
+         }
 
-            String platformName = args[0];
-            Class platformClass = Class.forName(platformName);
-            WinGammaPlatform platform = (WinGammaPlatform) platformClass.newInstance();
+         String platformName = args[0];
+         Class platformClass = Class.forName(platformName);
+         WinGammaPlatform platform = (WinGammaPlatform) platformClass
+               .newInstance();
 
-            String[] platformArgs = new String[args.length - 1];
-            System.arraycopy(args, 1, platformArgs, 0, platformArgs.length);
+         String[] platformArgs = new String[args.length - 1];
+         System.arraycopy(args, 1, platformArgs, 0, platformArgs.length);
 
-            // Allow the platform to write platform-specific files
-            platform.createVcproj(platformArgs);
-        }
-        catch (Exception e) {
-            e.printStackTrace();
-              System.exit(1);
-        }
-    }
+         // Allow the platform to write platform-specific files
+         platform.createVcproj(platformArgs);
+      } catch (Exception e) {
+         e.printStackTrace();
+         System.exit(1);
+      }
+   }
 }
diff --git a/src/share/tools/ProjectCreator/Util.java b/src/share/tools/ProjectCreator/Util.java
index 466a09974..b8fe156e7 100644
--- a/src/share/tools/ProjectCreator/Util.java
+++ b/src/share/tools/ProjectCreator/Util.java
@@ -26,18 +26,19 @@ import java.util.*;
 import java.io.File;
 
 public class Util {
-    static String join(String padder, Vector v) {
+
+    static String join(String padder, Vector<String> v) {
         return join(padder, v, false);
     }
 
-    static String join(String padder, Vector v, boolean quoted) {
+    static String join(String padder, Vector<String> v, boolean quoted) {
         StringBuffer sb = new StringBuffer();
 
-        for (Iterator iter = v.iterator(); iter.hasNext(); ) {
+        for (Iterator<String> iter = v.iterator(); iter.hasNext(); ) {
             if (quoted) {
                 sb.append('"');
             }
-            sb.append((String)iter.next());
+            sb.append(iter.next());
             if (quoted) {
                 sb.append('"');
             }
@@ -48,10 +49,10 @@ public class Util {
     }
 
 
-    static String prefixed_join(String padder, Vector v, boolean quoted) {
+    static String prefixed_join(String padder, Vector<String> v, boolean quoted) {
         StringBuffer sb = new StringBuffer();
 
-        for (Iterator iter = v.iterator(); iter.hasNext(); ) {
+        for (Iterator<String> iter = v.iterator(); iter.hasNext(); ) {
             sb.append(padder);
 
             if (quoted) {
diff --git a/src/share/tools/ProjectCreator/WinGammaPlatform.java b/src/share/tools/ProjectCreator/WinGammaPlatform.java
index 7c4dc53e2..41deb5238 100644
--- a/src/share/tools/ProjectCreator/WinGammaPlatform.java
+++ b/src/share/tools/ProjectCreator/WinGammaPlatform.java
@@ -29,6 +29,7 @@ import java.util.Enumeration;
 import java.util.Hashtable;
 import java.util.Iterator;
 import java.util.List;
+import java.util.Stack;
 import java.util.TreeSet;
 import java.util.Vector;
 
@@ -218,69 +219,6 @@ public abstract class WinGammaPlatform {
         return false;
     }
 
-    /* This returns a String containing the full path to the passed
-       file name, or null if an error occurred. If the file was not
-       found or was a duplicate and couldn't be resolved using the
-       preferred paths, the file name is added to the appropriate
-       Vector of Strings. */
-    private String findFileInDirectory(String fileName,
-                                       DirectoryTree directory,
-                                       Vector preferredPaths,
-                                       Vector filesNotFound,
-                                       Vector filesDuplicate) {
-        List locationsInTree = directory.findFile(fileName);
-        int  rootNameLength = directory.getRootNodeName().length();
-        String name = null;
-        if ((locationsInTree == null) ||
-            (locationsInTree.size() == 0)) {
-            filesNotFound.add(fileName);
-        } else if (locationsInTree.size() > 1) {
-            // Iterate through them, trying to find one with a
-            // preferred path
-        search:
-            {
-                for (Iterator locIter = locationsInTree.iterator();
-                     locIter.hasNext(); ) {
-                    DirectoryTreeNode node =
-                        (DirectoryTreeNode) locIter.next();
-                    String tmpName = node.getName();
-                    for (Iterator prefIter = preferredPaths.iterator();
-                         prefIter.hasNext(); ) {
-                        // We need to make sure the preferred path is
-                        // found from the file path not including the root node name.
-                        if (tmpName.indexOf((String)prefIter.next(),
-                                            rootNameLength) != -1) {
-                            name = tmpName;
-                            break search;
-                        }
-                    }
-                }
-            }
-
-            if (name == null) {
-                filesDuplicate.add(fileName);
-            }
-        } else {
-            name = ((DirectoryTreeNode) locationsInTree.get(0)).getName();
-        }
-
-        return name;
-    }
-
-    protected String envVarPrefixedFileName(String fileName,
-                                            int sourceBaseLen,
-                                            DirectoryTree tree,
-                                            Vector preferredPaths,
-                                            Vector filesNotFound,
-                                            Vector filesDuplicate) {
-        String fullName = findFileInDirectory(fileName,
-                                              tree,
-                                              preferredPaths,
-                                              filesNotFound,
-                                              filesDuplicate);
-        return fullName;
-    }
-
      String getProjectName(String fullPath, String extension)
         throws IllegalArgumentException, IOException {
         File file = new File(fullPath).getCanonicalFile();
@@ -369,6 +307,12 @@ public abstract class WinGammaPlatform {
                               HsArgHandler.STRING
                               ),
 
+               new HsArgRule("-buildSpace",
+                              "BuildSpace",
+                              null,
+                              HsArgHandler.STRING
+                              ),
+
               new HsArgRule("-platformName",
                               "PlatformName",
                               null,
@@ -405,6 +349,18 @@ public abstract class WinGammaPlatform {
                               HsArgHandler.VECTOR
                               ),
 
+                new HsArgRule("-absoluteSrcInclude",
+                              "AbsoluteSrcInclude",
+                              null,
+                              HsArgHandler.VECTOR
+                              ),
+
+                new HsArgRule("-relativeSrcInclude",
+                              "RelativeSrcInclude",
+                              null,
+                              HsArgHandler.VECTOR
+                              ),
+
                 new HsArgRule("-define",
                               "Define",
                               null,
@@ -494,6 +450,12 @@ public abstract class WinGammaPlatform {
                               HsArgHandler.VECTOR
                               ),
 
+                new HsArgRule("-hidePath",
+                      "HidePath",
+                      null,
+                      HsArgHandler.VECTOR
+                      ),
+
                 new HsArgRule("-additionalFile",
                               "AdditionalFile",
                               null,
@@ -611,107 +573,101 @@ public abstract class WinGammaPlatform {
         return allConfigs;
     }
 
-    class FileAttribute {
-        int     numConfigs;
-        Vector  configs;
-        String  shortName;
-        boolean noPch, pchRoot;
-
-        FileAttribute(String shortName, BuildConfig cfg, int numConfigs) {
-            this.shortName = shortName;
-            this.noPch =  (cfg.lookupHashFieldInContext("DisablePch", shortName) != null);
-            this.pchRoot = shortName.equals(BuildConfig.getFieldString(null, "UseToGeneratePch"));
-            this.numConfigs = numConfigs;
-
-            configs = new Vector();
-            add(cfg.get("Name"));
-        }
-
-        void add(String confName) {
-            configs.add(confName);
+    PrintWriter printWriter;
 
-            // if presented in all configs
-            if (configs.size() == numConfigs) {
-                configs = null;
-            }
-        }
+    public void writeProjectFile(String projectFileName, String projectName,
+                                 Vector<BuildConfig> allConfigs) throws IOException {
+        throw new RuntimeException("use compiler version specific version");
     }
 
-    class FileInfo implements Comparable {
-        String        full;
-        FileAttribute attr;
-
-        FileInfo(String full, FileAttribute  attr) {
-            this.full = full;
-            this.attr = attr;
-        }
-
-        public int compareTo(Object o) {
-            FileInfo oo = (FileInfo)o;
-            return full.compareTo(oo.full);
-        }
-
-        boolean isHeader() {
-            return attr.shortName.endsWith(".h") || attr.shortName.endsWith(".hpp");
-        }
+    int indent;
+    private Stack<String> tagStack = new Stack<String>();
 
-        boolean isCpp() {
-            return attr.shortName.endsWith(".cpp");
-        }
+    private void startTagPrim(String name, String[] attrs, boolean close) {
+       startTagPrim(name, attrs, close, true);
     }
 
+    private void startTagPrim(String name, String[] attrs, boolean close,
+          boolean newline) {
+       doIndent();
+       printWriter.print("<" + name);
+       indent++;
+
+       if (attrs != null && attrs.length > 0) {
+          for (int i = 0; i < attrs.length; i += 2) {
+             printWriter.print(" " + attrs[i] + "=\"" + attrs[i + 1] + "\"");
+             if (i < attrs.length - 2) {
+             }
+          }
+       }
+
+       if (close) {
+          indent--;
+          printWriter.print(" />");
+       } else {
+          // TODO push tag name, and change endTag to pop and print.
+          tagStack.push(name);
+          printWriter.print(">");
+       }
+       if (newline) {
+          printWriter.println();
+       }
+    }
 
-    TreeSet sortFiles(Hashtable allFiles) {
-        TreeSet rv = new TreeSet();
-        Enumeration e = allFiles.keys();
-        while (e.hasMoreElements()) {
-            String fullPath = (String)e.nextElement();
-            rv.add(new FileInfo(fullPath, (FileAttribute)allFiles.get(fullPath)));
-        }
-        return rv;
+    void startTag(String name, String... attrs) {
+       startTagPrim(name, attrs, false);
     }
 
-    Hashtable computeAttributedFiles(Vector allConfigs) {
-        Hashtable ht = new Hashtable();
-        int numConfigs = allConfigs.size();
+    void startTagV(String name, Vector attrs) {
+       String s[] = new String[attrs.size()];
+       for (int i = 0; i < attrs.size(); i++) {
+          s[i] = (String) attrs.elementAt(i);
+       }
+       startTagPrim(name, s, false);
+    }
 
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            BuildConfig bc = (BuildConfig)i.next();
-            Hashtable  confFiles = (Hashtable)bc.getSpecificField("AllFilesHash");
-            String confName = bc.get("Name");
+    void endTag() {
+       String name = tagStack.pop();
+       indent--;
+       doIndent();
+       printWriter.println("</" + name + ">");
+    }
 
-            for (Enumeration e=confFiles.keys(); e.hasMoreElements(); ) {
-                String filePath = (String)e.nextElement();
-                FileAttribute fa = (FileAttribute)ht.get(filePath);
+    private void endTagNoIndent() {
+       String name = tagStack.pop();
+       indent--;
+       printWriter.println("</" + name + ">");
+    }
 
-                if (fa == null) {
-                    fa = new FileAttribute((String)confFiles.get(filePath), bc, numConfigs);
-                    ht.put(filePath, fa);
-                } else {
-                    fa.add(confName);
-                }
-            }
-        }
+    void tag(String name, String... attrs) {
+       startTagPrim(name, attrs, true);
+    }
 
-        return ht;
+    void tagData(String name, String data) {
+       startTagPrim(name, null, false, false);
+       printWriter.print(data);
+       endTagNoIndent();
     }
 
-     Hashtable computeAttributedFiles(BuildConfig bc) {
-        Hashtable ht = new Hashtable();
-        Hashtable confFiles = (Hashtable)bc.getSpecificField("AllFilesHash");
+    void tagData(String name, String data, String... attrs) {
+       startTagPrim(name, attrs, false, false);
+       printWriter.print(data);
+       endTagNoIndent();
+    }
 
-        for (Enumeration e = confFiles.keys(); e.hasMoreElements(); ) {
-            String filePath = (String)e.nextElement();
-            ht.put(filePath,  new FileAttribute((String)confFiles.get(filePath), bc, 1));
-        }
+    void tagV(String name, Vector attrs) {
+       String s[] = new String[attrs.size()];
+       for (int i = 0; i < attrs.size(); i++) {
+          s[i] = (String) attrs.elementAt(i);
+       }
+       startTagPrim(name, s, true);
+    }
 
-        return ht;
+    void doIndent() {
+       for (int i = 0; i < indent; i++) {
+          printWriter.print("  ");
+       }
     }
 
-    PrintWriter printWriter;
 
-    public void writeProjectFile(String projectFileName, String projectName,
-                                 Vector<BuildConfig> allConfigs) throws IOException {
-        throw new RuntimeException("use compiler version specific version");
-    }
 }
diff --git a/src/share/tools/ProjectCreator/WinGammaPlatformVC10.java b/src/share/tools/ProjectCreator/WinGammaPlatformVC10.java
index 4a96981e7..137abdc30 100644
--- a/src/share/tools/ProjectCreator/WinGammaPlatformVC10.java
+++ b/src/share/tools/ProjectCreator/WinGammaPlatformVC10.java
@@ -3,14 +3,18 @@ import java.io.FileNotFoundException;
 import java.io.IOException;
 import java.io.PrintWriter;
 import java.io.UnsupportedEncodingException;
-import java.util.Hashtable;
+import java.nio.file.FileSystems;
 import java.util.Iterator;
-import java.util.TreeSet;
+import java.util.LinkedList;
 import java.util.UUID;
 import java.util.Vector;
 
 public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
 
+
+   LinkedList <String>filters = new LinkedList<String>();
+   LinkedList <String[]>filterDeps = new LinkedList<String[]>();
+
     @Override
     protected String getProjectExt() {
         return ".vcxproj";
@@ -37,15 +41,15 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
                     "Include", cfg.get("Name"));
             tagData("Configuration", cfg.get("Id"));
             tagData("Platform", cfg.get("PlatformName"));
-            endTag("ProjectConfiguration");
+            endTag();
         }
-        endTag("ItemGroup");
+        endTag();
 
         startTag("PropertyGroup", "Label", "Globals");
         tagData("ProjectGuid", "{8822CB5C-1C41-41C2-8493-9F6E1994338B}");
         tag("SccProjectName");
         tag("SccLocalPath");
-        endTag("PropertyGroup");
+        endTag();
 
         tag("Import", "Project", "$(VCTargetsPath)\\Microsoft.Cpp.Default.props");
 
@@ -53,19 +57,19 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
             startTag(cfg, "PropertyGroup", "Label", "Configuration");
             tagData("ConfigurationType", "DynamicLibrary");
             tagData("UseOfMfc", "false");
-            endTag("PropertyGroup");
+            endTag();
         }
 
         tag("Import", "Project", "$(VCTargetsPath)\\Microsoft.Cpp.props");
         startTag("ImportGroup", "Label", "ExtensionSettings");
-        endTag("ImportGroup");
+        endTag();
         for (BuildConfig cfg : allConfigs) {
             startTag(cfg, "ImportGroup", "Label", "PropertySheets");
             tag("Import",
                     "Project", "$(UserRootDir)\\Microsoft.Cpp.$(Platform).user.props",
                     "Condition", "exists('$(UserRootDir)\\Microsoft.Cpp.$(Platform).user.props')",
                     "Label", "LocalAppDataPlatform");
-            endTag("ImportGroup");
+            endTag();
         }
 
         tag("PropertyGroup", "Label", "UserMacros");
@@ -82,38 +86,38 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
             tag(cfg, "CodeAnalysisRules");
             tag(cfg, "CodeAnalysisRuleAssemblies");
         }
-        endTag("PropertyGroup");
+        endTag();
 
         for (BuildConfig cfg : allConfigs) {
             startTag(cfg, "ItemDefinitionGroup");
             startTag("ClCompile");
             tagV(cfg.getV("CompilerFlags"));
-            endTag("ClCompile");
+            endTag();
 
             startTag("Link");
             tagV(cfg.getV("LinkerFlags"));
-            endTag("Link");
+            endTag();
 
             startTag("PostBuildEvent");
             tagData("Message", BuildConfig.getFieldString(null, "PostbuildDescription"));
             tagData("Command", cfg.expandFormat(BuildConfig.getFieldString(null, "PostbuildCommand").replace("\t", "\r\n")));
-            endTag("PostBuildEvent");
+            endTag();
 
             startTag("PreLinkEvent");
             tagData("Message", BuildConfig.getFieldString(null, "PrelinkDescription"));
             tagData("Command", cfg.expandFormat(BuildConfig.getFieldString(null, "PrelinkCommand").replace("\t", "\r\n")));
-            endTag("PreLinkEvent");
+            endTag();
 
-            endTag("ItemDefinitionGroup");
+            endTag();
         }
 
         writeFiles(allConfigs, projDir);
 
         tag("Import", "Project", "$(VCTargetsPath)\\Microsoft.Cpp.targets");
         startTag("ImportGroup", "Label", "ExtensionTargets");
-        endTag("ImportGroup");
+        endTag();
 
-        endTag("Project");
+        endTag();
         printWriter.close();
         System.out.println("    Done.");
 
@@ -138,14 +142,22 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
         for (BuildConfig cfg : allConfigs) {
             startTag(cfg, "PropertyGroup");
             tagData("LocalDebuggerCommand", "$(TargetDir)/hotspot.exe");
-            endTag("PropertyGroup");
+            endTag();
         }
 
-        endTag("Project");
+        endTag();
         printWriter.close();
         System.out.println("    Done.");
     }
 
+    public void addFilter(String rPath) {
+       filters.add(rPath);
+    }
+
+    public void addFilterDependency(String fileLoc, String filter) {
+      filterDeps.add(new String[] {fileLoc, filter});
+    }
+
     private void writeFilterFile(String projectFileName, String projectName,
             Vector<BuildConfig> allConfigs, String base) throws FileNotFoundException, UnsupportedEncodingException {
         String filterFileName = projectFileName + ".filters";
@@ -157,210 +169,92 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
                 "ToolsVersion", "4.0",
                 "xmlns", "http://schemas.microsoft.com/developer/msbuild/2003");
 
-        Hashtable<String, FileAttribute> allFiles = computeAttributedFiles(allConfigs);
-        TreeSet<FileInfo> sortedFiles = sortFiles(allFiles);
-        Vector<NameFilter> filters = makeFilters(sortedFiles);
-
-        // first all filters
         startTag("ItemGroup");
-        for (NameFilter filter : filters) {
-            doWriteFilter(filter, "");
+        for (String filter : filters) {
+           startTag("Filter", "Include",filter);
+           UUID uuid = UUID.randomUUID();
+           tagData("UniqueIdentifier", "{" + uuid.toString() + "}");
+           endTag();
         }
         startTag("Filter", "Include", "Resource Files");
         UUID uuid = UUID.randomUUID();
         tagData("UniqueIdentifier", "{" + uuid.toString() + "}");
         tagData("Extensions", "ico;cur;bmp;dlg;rc2;rct;bin;cnt;rtf;gif;jpg;jpeg;jpe");
-        endTag("Filter");
-        endTag("ItemGroup");
+        endTag();
+        endTag();
 
-        // then all cpp files
-        startTag("ItemGroup");
-        for (NameFilter filter : filters) {
-            doWriteFiles(sortedFiles, filter, "", "ClCompile", new Evaluator() {
-                public boolean pick(FileInfo fi) {
-                    return fi.isCpp();
-                }
-            }, base);
-        }
-        endTag("ItemGroup");
+        //TODO - do I need to split cpp and hpp files?
 
-        // then all header files
+        // then all files
         startTag("ItemGroup");
-        for (NameFilter filter : filters) {
-            doWriteFiles(sortedFiles, filter, "", "ClInclude", new Evaluator() {
-                public boolean pick(FileInfo fi) {
-                    return fi.isHeader();
-                }
-            }, base);
-        }
-        endTag("ItemGroup");
+        for (String[] dep : filterDeps) {
+           String tagName = getFileTagFromSuffix(dep[0]);
 
-        // then all other files
-        startTag("ItemGroup");
-        for (NameFilter filter : filters) {
-            doWriteFiles(sortedFiles, filter, "", "None", new Evaluator() {
-                public boolean pick(FileInfo fi) {
-                    return true;
-                }
-            }, base);
+           startTag(tagName, "Include", dep[0]);
+           tagData("Filter", dep[1]);
+           endTag();
         }
-        endTag("ItemGroup");
+        endTag();
 
-        endTag("Project");
+        endTag();
         printWriter.close();
         System.out.println("    Done.");
     }
 
-
-    private void doWriteFilter(NameFilter filter, String start) {
-        startTag("Filter", "Include", start + filter.fname);
-        UUID uuid = UUID.randomUUID();
-        tagData("UniqueIdentifier", "{" + uuid.toString() + "}");
-        endTag("Filter");
-        if (filter instanceof ContainerFilter) {
-            Iterator i = ((ContainerFilter)filter).babies();
-            while (i.hasNext()) {
-                doWriteFilter((NameFilter)i.next(), start + filter.fname + "\\");
-            }
-        }
+    public String getFileTagFromSuffix(String fileName) {
+       if (fileName.endsWith(".cpp")) {
+          return"ClCompile";
+       } else if (fileName.endsWith(".c")) {
+          return "ClCompile";
+       } else if (fileName.endsWith(".hpp")) {
+          return"ClInclude";
+       } else if (fileName.endsWith(".h")) {
+          return "ClInclude";
+       } else {
+          return"None";
+       }
     }
 
-    interface Evaluator {
-        boolean pick(FileInfo fi);
-    }
-
-    private void doWriteFiles(TreeSet<FileInfo> allFiles, NameFilter filter, String start, String tool, Evaluator eval, String base) {
-        if (filter instanceof ContainerFilter) {
-            Iterator i = ((ContainerFilter)filter).babies();
-            while (i.hasNext()) {
-                doWriteFiles(allFiles, (NameFilter)i.next(), start + filter.fname + "\\", tool, eval, base);
-            }
-        }
-        else {
-            Iterator i = allFiles.iterator();
-            while (i.hasNext()) {
-                FileInfo fi = (FileInfo)i.next();
-
-                if (!filter.match(fi)) {
-                    continue;
-                }
-                if (eval.pick(fi)) {
-                    startTag(tool, "Include", rel(fi.full, base));
-                    tagData("Filter", start + filter.fname);
-                    endTag(tool);
-
-                    // we not gonna look at this file anymore (sic!)
-                    i.remove();
-                }
-            }
-        }
-    }
-
-
     void writeFiles(Vector<BuildConfig> allConfigs, String projDir) {
-        Hashtable<String, FileAttribute> allFiles = computeAttributedFiles(allConfigs);
-        TreeSet<FileInfo> sortedFiles = sortFiles(allFiles);
+       // This code assummes there are no config specific includes.
+       startTag("ItemGroup");
 
-        // first cpp-files
-        startTag("ItemGroup");
-        for (FileInfo fi : sortedFiles) {
-            if (!fi.isCpp()) {
-                continue;
-            }
-            writeFile("ClCompile", allConfigs, fi, projDir);
-        }
-        endTag("ItemGroup");
+       String sourceBase = BuildConfig.getFieldString(null, "SourceBase");
 
-        // then header-files
-        startTag("ItemGroup");
-        for (FileInfo fi : sortedFiles) {
-            if (!fi.isHeader()) {
-                continue;
-            }
-            writeFile("ClInclude", allConfigs, fi, projDir);
-        }
-        endTag("ItemGroup");
+       // Use first config for all global absolute includes.
+       BuildConfig baseConfig = allConfigs.firstElement();
+       Vector<String> rv = new Vector<String>();
 
-        // then others
-        startTag("ItemGroup");
-        for (FileInfo fi : sortedFiles) {
-            if (fi.isHeader() || fi.isCpp()) {
-                continue;
-            }
-            writeFile("None", allConfigs, fi, projDir);
-        }
-        endTag("ItemGroup");
-    }
+       // Then use first config for all relative includes
+       Vector<String> ri = new Vector<String>();
+       baseConfig.collectRelevantVectors(ri, "RelativeSrcInclude");
+       for (String f : ri) {
+          rv.add(sourceBase + Util.sep + f);
+       }
 
-    /**
-     * Make "path" into a relative path using "base" as the base.
-     *
-     * path and base are assumed to be normalized with / as the file separator.
-     * returned path uses "\\" as file separator
-     */
-    private String rel(String path, String base)
-    {
-        if(!base.endsWith("/")) {
-                base += "/";
-        }
-        String[] pathTok = path.split("/");
-        String[] baseTok = base.split("/");
-        int pi = 0;
-        int bi = 0;
-        StringBuilder newPath = new StringBuilder();
-
-        // first step past all path components that are the same
-        while (pi < pathTok.length &&
-                bi < baseTok.length &&
-                pathTok[pi].equals(baseTok[bi])) {
-            pi++;
-            bi++;
-        }
+       baseConfig.collectRelevantVectors(rv, "AbsoluteSrcInclude");
 
-        // for each path component left in base, add "../"
-        while (bi < baseTok.length) {
-            bi++;
-                newPath.append("..\\");
-        }
+       handleIncludes(rv, allConfigs);
 
-        // now add everything left in path
-        while (pi < pathTok.length) {
-                newPath.append(pathTok[pi]);
-                pi++;
-            if (pi != pathTok.length) {
-                newPath.append("\\");
-            }
-        }
-        return newPath.toString();
+       endTag();
     }
 
-    private void writeFile(String tool, Vector<BuildConfig> allConfigs, FileInfo fi, String base) {
-        if (fi.attr.configs == null && fi.attr.pchRoot == false && fi.attr.noPch == false) {
-            tag(tool, "Include", rel(fi.full, base));
-        }
-        else {
-            startTag(tool, "Include", rel(fi.full, base));
-            for (BuildConfig cfg : allConfigs) {
-                if (fi.attr.configs != null && !fi.attr.configs.contains(cfg.get("Name"))) {
-                    tagData(cfg, "ExcludedFromBuild", "true");
-                }
-                if (fi.attr.pchRoot) {
-                        tagData(cfg, "PrecompiledHeader", "Create");
-                }
-                if (fi.attr.noPch) {
-                        startTag(cfg, "PrecompiledHeader");
-                        endTag("PrecompiledHeader");
-                }
-            }
-            endTag(tool);
-        }
+    // Will visit file tree for each include
+    private void handleIncludes(Vector<String> includes, Vector<BuildConfig> allConfigs) {
+       for (String path : includes)  {
+          FileTreeCreatorVC10 ftc = new FileTreeCreatorVC10(FileSystems.getDefault().getPath(path) , allConfigs, this);
+          try {
+             ftc.writeFileTree();
+          } catch (IOException e) {
+             e.printStackTrace();
+          }
+       }
     }
 
     String buildCond(BuildConfig cfg) {
         return "'$(Configuration)|$(Platform)'=='"+cfg.get("Name")+"'";
     }
 
-
     void tagV(Vector<String> v) {
         Iterator<String> i = v.iterator();
         while(i.hasNext()) {
@@ -391,6 +285,7 @@ public class WinGammaPlatformVC10 extends WinGammaPlatformVC7 {
 
         startTag(name, ss);
     }
+
 }
 
 class CompilerInterfaceVC10 extends CompilerInterface {
diff --git a/src/share/tools/ProjectCreator/WinGammaPlatformVC6.java b/src/share/tools/ProjectCreator/WinGammaPlatformVC6.java
deleted file mode 100644
index f29abc051..000000000
--- a/src/share/tools/ProjectCreator/WinGammaPlatformVC6.java
+++ /dev/null
@@ -1,297 +0,0 @@
-/*
- * Copyright (c) 2005, 2011, Oracle and/or its affiliates. All rights reserved.
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This code is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- *
- */
-
-import java.io.*;
-import java.util.*;
-
-public class WinGammaPlatformVC6 extends WinGammaPlatform {
-    public void writeProjectFile(String projectFileName, String projectName,
-                                 Vector allConfigs) throws IOException {
-        Vector allConfigNames = new Vector();
-
-        printWriter = new PrintWriter(new FileWriter(projectFileName));
-        String cfg = ((BuildConfig)allConfigs.get(0)).get("Name");
-
-        printWriter.println("# Microsoft Developer Studio Project File - Name=\"" + projectName + "\" - Package Owner=<4>");
-        printWriter.println("# Microsoft Developer Studio Generated Build File, Format Version 6.00");
-        printWriter.println("# ** DO NOT EDIT **");
-        printWriter.println("");
-        printWriter.println("# TARGTYPE \"Win32 (x86) Dynamic-Link Library\" 0x0102");
-        printWriter.println("CFG=" + cfg);
-        printWriter.println("");
-
-        printWriter.println("!MESSAGE This is not a valid makefile. To build this project using NMAKE,");
-        printWriter.println("!MESSAGE use the Export Makefile command and run");
-        printWriter.println("!MESSAGE ");
-        printWriter.println("!MESSAGE NMAKE /f \"" + projectName + ".mak\".");
-        printWriter.println("!MESSAGE ");
-        printWriter.println("!MESSAGE You can specify a configuration when running NMAKE");
-        printWriter.println("!MESSAGE by defining the macro CFG on the command line. For example:");
-        printWriter.println("!MESSAGE ");
-        printWriter.println("!MESSAGE NMAKE /f \"" + projectName + ".mak\" CFG=\"" + cfg + "\"");
-        printWriter.println("!MESSAGE ");
-        printWriter.println("!MESSAGE Possible choices for configuration are:");
-        printWriter.println("!MESSAGE ");
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            String name = ((BuildConfig)i.next()).get("Name");
-            printWriter.println("!MESSAGE \""+ name + "\" (based on \"Win32 (x86) Dynamic-Link Library\")");
-            allConfigNames.add(name);
-        }
-        printWriter.println("!MESSAGE ");
-        printWriter.println("");
-
-        printWriter.println("# Begin Project");
-        printWriter.println("# PROP AllowPerConfigDependencies 0");
-        printWriter.println("# PROP Scc_ProjName \"\"");
-        printWriter.println("# PROP Scc_LocalPath \"\"");
-        printWriter.println("CPP=cl.exe");
-        printWriter.println("MTL=midl.exe");
-        printWriter.println("RSC=rc.exe");
-
-
-        String keyword = "!IF";
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            BuildConfig bcfg = (BuildConfig)i.next();
-            printWriter.println(keyword + "  \"$(CFG)\" == \"" + bcfg.get("Name") + "\"");
-            writeConfigHeader(bcfg);
-            keyword = "!ELSEIF";
-            if (!i.hasNext()) printWriter.println("!ENDIF");
-        }
-
-
-        TreeSet sortedFiles = sortFiles(computeAttributedFiles(allConfigs));
-
-        printWriter.println("# Begin Target");
-
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            printWriter.println("# Name \"" + ((BuildConfig)i.next()).get("Name") + "\"");
-        }
-        printWriter.println("# Begin Group \"Header Files\"");
-        printWriter.println("# PROP Default_Filter \"h;hpp;hxx;hm;inl;fi;fd\"");
-
-        Iterator i = sortedFiles.iterator();
-
-        while (i.hasNext()) {
-            FileInfo fi = (FileInfo)i.next();
-
-            // skip sources
-            if (!fi.isHeader()) {
-                continue;
-            }
-
-            printFile(fi, allConfigNames);
-        }
-        printWriter.println("# End Group");
-        printWriter.println("");
-
-        printWriter.println("# Begin Group \"Source Files\"");
-        printWriter.println("# PROP Default_Filter \"cpp;c;cxx;rc;def;r;odl;hpj;bat;for;f90\"");
-
-        i = sortedFiles.iterator();
-        while (i.hasNext()) {
-            FileInfo fi = (FileInfo)i.next();
-
-            // skip headers
-            if (fi.isHeader()) {
-                continue;
-            }
-
-            printFile(fi, allConfigNames);
-        }
-        printWriter.println("# End Group");
-        printWriter.println("");
-
-
-        printWriter.println("# Begin Group \"Resource Files\"");
-        printWriter.println("# PROP Default_Filter \"ico;cur;bmp;dlg;rc2;rct;bin;cnt;rtf;gif;jpg;jpeg;jpe\"");
-        printWriter.println("# End Group");
-        printWriter.println("");
-        printWriter.println("# End Target");
-
-        printWriter.println("# End Project");
-
-        printWriter.close();
-    }
-
-
-    void printFile(FileInfo fi, Vector allConfigNames) {
-        printWriter.println("# Begin Source File");
-        printWriter.println("");
-        printWriter.println("SOURCE=\"" + fi.full + "\"");
-        FileAttribute attr = fi.attr;
-
-        if (attr.noPch) {
-            printWriter.println("# SUBTRACT CPP /YX /Yc /Yu");
-        }
-
-        if (attr.pchRoot) {
-            printWriter.println("# ADD CPP /Yc\"incls/_precompiled.incl\"");
-        }
-        if (attr.configs != null) {
-            String keyword = "!IF";
-            for (Iterator j=allConfigNames.iterator(); j.hasNext();) {
-                String cfg = (String)j.next();
-                if (!attr.configs.contains(cfg)) {
-                    printWriter.println(keyword+" \"$(CFG)\" == \"" + cfg +"\"");
-                    printWriter.println("# PROP BASE Exclude_From_Build 1");
-                    printWriter.println("# PROP Exclude_From_Build 1");
-                    keyword = "!ELSEIF";
-                }
-            }
-            printWriter.println("!ENDIF");
-        }
-
-        printWriter.println("# End Source File");
-    }
-
-    void writeConfigHeader(BuildConfig cfg) {
-        printWriter.println("# Begin Special Build Tool");
-        printWriter.println("SOURCE=\"$(InputPath)\"");
-        printWriter.println("PreLink_Desc=" +  BuildConfig.getFieldString(null, "PrelinkDescription"));
-        printWriter.println("PreLink_Cmds=" +
-                            cfg.expandFormat(BuildConfig.getFieldString(null, "PrelinkCommand")));
-        printWriter.println("# End Special Build Tool");
-        printWriter.println("");
-
-        for (Iterator i = cfg.getV("CompilerFlags").iterator(); i.hasNext(); ) {
-            printWriter.println("# "+(String)i.next());
-        }
-
-
-        printWriter.println("LINK32=link.exe");
-
-        for (Iterator i = cfg.getV("LinkerFlags").iterator(); i.hasNext(); ) {
-            printWriter.println("# "+(String)i.next());
-        }
-
-        printWriter.println("ADD BASE MTL /nologo /D \"_DEBUG\" /mktyplib203 /win32");
-        printWriter.println("ADD MTL /nologo /D \"_DEBUG\" /mktyplib203 /win32");
-        printWriter.println("ADD BASE RSC /l 0x409 /d \"_DEBUG\"");
-        printWriter.println("ADD RSC /l 0x409 /d \"_DEBUG\"");
-        printWriter.println("BSC32=bscmake.exe");
-        printWriter.println("ADD BASE BSC32 /nologo");
-        printWriter.println("ADD BSC32 /nologo");
-        printWriter.println("");
-    }
-
-    protected String getProjectExt() {
-        return ".dsp";
-    }
-}
-
-
-class CompilerInterfaceVC6  extends CompilerInterface {
-    Vector getBaseCompilerFlags(Vector defines, Vector includes, String outDir) {
-        Vector rv = new Vector();
-
-        rv.add("PROP BASE Use_MFC 0");
-        rv.add("PROP Use_MFC 0");
-        rv.add("ADD CPP /nologo /MT /W3 /WX /GX /YX /Fr /FD /c");
-        rv.add("PROP BASE Output_Dir \""+outDir+"\"");
-        rv.add("PROP Output_Dir \""+outDir+"\"");
-        rv.add("PROP BASE Intermediate_Dir \""+outDir+"\"");
-        rv.add("PROP Intermediate_Dir \""+outDir+"\"");
-        rv.add("PROP BASE Target_Dir \"\"");
-        rv.add("PROP Target_Dir \"\"");
-        rv.add("ADD BASE CPP "+Util.prefixed_join(" /I ", includes, true));
-        rv.add("ADD CPP "+Util.prefixed_join(" /I ", includes, true));
-        rv.add("ADD BASE CPP "+Util.prefixed_join(" /D ", defines, true));
-        rv.add("ADD CPP "+Util.prefixed_join(" /D ", defines, true));
-        rv.add("ADD CPP /Yu\"incls/_precompiled.incl\"");
-
-        return rv;
-    }
-
-    Vector getBaseLinkerFlags(String outDir, String outDll, String platformName) {
-        Vector rv = new Vector();
-
-        rv.add("PROP Ignore_Export_Lib 0");
-        rv.add("ADD BASE CPP /MD");
-        rv.add("ADD CPP /MD");
-        rv.add("ADD LINK32 kernel32.lib user32.lib gdi32.lib winspool.lib comdlg32.lib " +
-               "           advapi32.lib shell32.lib ole32.lib oleaut32.lib winmm.lib");
-        String machine = "/machine:I386";
-        if (platformName.equals("x64")) {
-                machine = "/machine:X64";
-        }
-        rv.add("ADD LINK32      /out:\""+outDll+"\" "+
-               "                /nologo /subsystem:windows /machine:" + machine +
-               "                /nologo /base:\"0x8000000\" /subsystem:windows /dll" +
-               "                /export:JNI_GetDefaultJavaVMInitArgs /export:JNI_CreateJavaVM /export:JNI_GetCreatedJavaVMs "+
-               "                /export:jio_snprintf /export:jio_printf /export:jio_fprintf /export:jio_vfprintf "+
-               "                /export:jio_vsnprintf ");
-        rv.add("SUBTRACT LINK32 /pdb:none /map");
-
-        return rv;
-    }
-
-    Vector getDebugCompilerFlags(String opt) {
-        Vector rv = new Vector();
-
-        rv.add("ADD BASE CPP /Gm /Zi /O"+opt);
-
-        return rv;
-    }
-
-    Vector getDebugLinkerFlags() {
-        Vector rv = new Vector();
-
-        rv.add("PROP BASE Use_Debug_Libraries 1");
-        rv.add("PROP Use_Debug_Libraries 1");
-        rv.add("ADD LINK32 /debug");
-
-        return rv;
-    }
-
-    void getAdditionalNonKernelLinkerFlags(Vector rv) {}
-
-    Vector getProductCompilerFlags() {
-        Vector rv = new Vector();
-
-        rv.add("ADD CPP /O"+getOptFlag());
-
-        return rv;
-    }
-
-    Vector getProductLinkerFlags() {
-        Vector rv = new Vector();
-
-        rv.add("PROP BASE Use_Debug_Libraries 0");
-        rv.add("PROP Use_Debug_Libraries 0");
-
-        return rv;
-    }
-
-    String getOptFlag() {
-        return "2";
-    }
-
-    String getNoOptFlag() {
-        return "d";
-    }
-
-    String makeCfgName(String flavourBuild, String platform) {
-        return "vm - "+ platform + " " + flavourBuild;
-    }
-}
diff --git a/src/share/tools/ProjectCreator/WinGammaPlatformVC7.java b/src/share/tools/ProjectCreator/WinGammaPlatformVC7.java
index d1d63c847..735b1a813 100644
--- a/src/share/tools/ProjectCreator/WinGammaPlatformVC7.java
+++ b/src/share/tools/ProjectCreator/WinGammaPlatformVC7.java
@@ -25,758 +25,326 @@
 import java.io.FileWriter;
 import java.io.IOException;
 import java.io.PrintWriter;
-import java.util.Hashtable;
-import java.util.Iterator;
-import java.util.TreeSet;
+import java.nio.file.FileSystems;
 import java.util.Vector;
 
 public class WinGammaPlatformVC7 extends WinGammaPlatform {
 
-    String projectVersion() {return "7.10";};
+   // TODO How about moving all globals configs to its own BuildConfig?
 
-    public void writeProjectFile(String projectFileName, String projectName,
-                                 Vector<BuildConfig> allConfigs) throws IOException {
-        System.out.println();
-        System.out.println("    Writing .vcproj file: "+projectFileName);
-        // If we got this far without an error, we're safe to actually
-        // write the .vcproj file
-        printWriter = new PrintWriter(new FileWriter(projectFileName));
+   String projectVersion() {
+      return "7.10";
+   };
 
-        printWriter.println("<?xml version=\"1.0\" encoding=\"windows-1251\"?>");
-        startTag(
-            "VisualStudioProject",
-            new String[] {
-                "ProjectType", "Visual C++",
-                "Version", projectVersion(),
-                "Name", projectName,
-                "ProjectGUID", "{8822CB5C-1C41-41C2-8493-9F6E1994338B}",
-                "SccProjectName", "",
-                "SccLocalPath", ""
-            }
-            );
-        startTag("Platforms");
-        tag("Platform", new String[] {"Name", (String) BuildConfig.getField(null, "PlatformName")});
-        endTag("Platforms");
-
-        startTag("Configurations");
-
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            writeConfiguration((BuildConfig)i.next());
-        }
-
-        endTag("Configurations");
-
-        tag("References");
-
-        writeFiles(allConfigs);
-
-        tag("Globals");
-
-        endTag("VisualStudioProject");
-        printWriter.close();
-
-        System.out.println("    Done.");
-    }
-
-
-    abstract class NameFilter {
-                protected String fname;
-
-        abstract boolean match(FileInfo fi);
-
-        String  filterString() { return ""; }
-        String name() { return this.fname;}
-
-        @Override
-        // eclipse auto-generated
-        public int hashCode() {
-            final int prime = 31;
-            int result = 1;
-            result = prime * result + getOuterType().hashCode();
-            result = prime * result + ((fname == null) ? 0 : fname.hashCode());
-            return result;
-        }
-
-        @Override
-        // eclipse auto-generated
-        public boolean equals(Object obj) {
-            if (this == obj)
-                return true;
-            if (obj == null)
-                return false;
-            if (getClass() != obj.getClass())
-                return false;
-            NameFilter other = (NameFilter) obj;
-            if (!getOuterType().equals(other.getOuterType()))
-                return false;
-            if (fname == null) {
-                if (other.fname != null)
-                    return false;
-            } else if (!fname.equals(other.fname))
-                return false;
-            return true;
-        }
-
-        // eclipse auto-generated
-        private WinGammaPlatformVC7 getOuterType() {
-            return WinGammaPlatformVC7.this;
-        }
-    }
-
-    class DirectoryFilter extends NameFilter {
-        String dir;
-        int baseLen, dirLen;
-
-        DirectoryFilter(String dir, String sbase) {
-            this.dir = dir;
-            this.baseLen = sbase.length();
-            this.dirLen = dir.length();
-            this.fname = dir;
-        }
-
-        DirectoryFilter(String fname, String dir, String sbase) {
-            this.dir = dir;
-            this.baseLen = sbase.length();
-            this.dirLen = dir.length();
-            this.fname = fname;
-        }
-
-
-        boolean match(FileInfo fi) {
-            int lastSlashIndex = fi.full.lastIndexOf('/');
-            String fullDir = fi.full.substring(0, lastSlashIndex);
-            return fullDir.endsWith(dir);
-        }
-
-        @Override
-        // eclipse auto-generated
-        public int hashCode() {
-            final int prime = 31;
-            int result = super.hashCode();
-            result = prime * result + getOuterType().hashCode();
-            result = prime * result + baseLen;
-            result = prime * result + ((dir == null) ? 0 : dir.hashCode());
-            result = prime * result + dirLen;
-            return result;
-        }
-
-        @Override
-        // eclipse auto-generated
-        public boolean equals(Object obj) {
-            if (this == obj)
-                return true;
-            if (!super.equals(obj))
-                return false;
-            if (getClass() != obj.getClass())
-                return false;
-            DirectoryFilter other = (DirectoryFilter) obj;
-            if (!getOuterType().equals(other.getOuterType()))
-                return false;
-            if (baseLen != other.baseLen)
-                return false;
-            if (dir == null) {
-                if (other.dir != null)
-                    return false;
-            } else if (!dir.equals(other.dir))
-                return false;
-            if (dirLen != other.dirLen)
-                return false;
-            return true;
-        }
-
-        // eclipse auto-generated
-        private WinGammaPlatformVC7 getOuterType() {
-            return WinGammaPlatformVC7.this;
-        }
-    }
-
-    class TerminatorFilter extends NameFilter {
-        TerminatorFilter(String fname) {
-            this.fname = fname;
-
-        }
-        boolean match(FileInfo fi) {
-            return true;
-        }
-
-    }
-
-    class SpecificNameFilter extends NameFilter {
-        String pats[];
-
-        SpecificNameFilter(String fname, String[] pats) {
-            this.fname = fname;
-            this.pats = pats;
-        }
-
-        boolean match(FileInfo fi) {
-            for (int i=0; i<pats.length; i++) {
-                if (fi.attr.shortName.matches(pats[i])) {
-                    return true;
-                }
-            }
-            return false;
-        }
-
-    }
-
-    class SpecificPathFilter extends NameFilter {
-        String pats[];
-
-        SpecificPathFilter(String fname, String[] pats) {
-            this.fname = fname;
-            this.pats = pats;
-        }
-
-        boolean match(FileInfo fi) {
-            for (int i=0; i<pats.length; i++) {
-                if (fi.full.matches(pats[i])) {
-                    return true;
-                }
-            }
-            return false;
-        }
-
-    }
-
-    class ContainerFilter extends NameFilter {
-        Vector children;
-
-        ContainerFilter(String fname) {
-            this.fname = fname;
-            children = new Vector();
-
-        }
-        boolean match(FileInfo fi) {
-            return false;
-        }
-
-        Iterator babies() { return children.iterator(); }
-
-        void add(NameFilter f) {
-            children.add(f);
-        }
-    }
-
-
-    void writeCustomToolConfig(Vector configs, String[] customToolAttrs) {
-        for (Iterator i = configs.iterator(); i.hasNext(); ) {
-            startTag("FileConfiguration",
-                     new String[] {
-                         "Name",  (String)i.next()
-                     }
-                     );
-            tag("Tool", customToolAttrs);
-
-            endTag("FileConfiguration");
-        }
-    }
-
-    // here we define filters, which define layout of what can be seen in 'Solution View' of MSVC
-    // Basically there are two types of entities - container filters and real filters
-    //   - container filter just provides a container to group together real filters
-    //   - real filter can select elements from the set according to some rule, put it into XML
-    //     and remove from the list
-    Vector<NameFilter> makeFilters(TreeSet<FileInfo> files) {
-        Vector<NameFilter> rv = new Vector<NameFilter>();
-        String sbase = Util.normalize(BuildConfig.getFieldString(null, "SourceBase")+"/src/");
-
-        String currentDir = "";
-        DirectoryFilter container = null;
-        for(FileInfo fileInfo : files) {
-
-            if (!fileInfo.full.startsWith(sbase)) {
-                continue;
-            }
-
-            int lastSlash = fileInfo.full.lastIndexOf('/');
-            String dir = fileInfo.full.substring(sbase.length(), lastSlash);
-            if(dir.equals("share/vm")) {
-                // skip files directly in share/vm - should only be precompiled.hpp which is handled below
-                continue;
-            }
-            if (!dir.equals(currentDir)) {
-                currentDir = dir;
-                if (container != null && !rv.contains(container)) {
-                    rv.add(container);
-                }
-
-                // remove "share/vm/" from names
-                String name = dir;
-                if (dir.startsWith("share/vm/")) {
-                    name = dir.substring("share/vm/".length(), dir.length());
-                }
-                DirectoryFilter newfilter = new DirectoryFilter(name, dir, sbase);
-                int i = rv.indexOf(newfilter);
-                if(i == -1) {
-                    container = newfilter;
-                } else {
-                    // if the filter already exists, reuse it
-                    container = (DirectoryFilter) rv.get(i);
-                }
-            }
-        }
-        if (container != null && !rv.contains(container)) {
-            rv.add(container);
-        }
-
-        ContainerFilter generated = new ContainerFilter("Generated");
-        ContainerFilter c1Generated = new ContainerFilter("C1");
-        c1Generated.add(new SpecificPathFilter("C++ Interpreter Generated", new String[] {".*compiler1/generated/jvmtifiles/bytecodeInterpreterWithChecks.+"}));
-        c1Generated.add(new SpecificPathFilter("jvmtifiles", new String[] {".*compiler1/generated/jvmtifiles/.*"}));
-        generated.add(c1Generated);
-        ContainerFilter c2Generated = new ContainerFilter("C2");
-        c2Generated.add(new SpecificPathFilter("C++ Interpreter Generated", new String[] {".*compiler2/generated/jvmtifiles/bytecodeInterpreterWithChecks.+"}));
-        c2Generated.add(new SpecificPathFilter("adfiles", new String[] {".*compiler2/generated/adfiles/.*"}));
-        c2Generated.add(new SpecificPathFilter("jvmtifiles", new String[] {".*compiler2/generated/jvmtifiles/.*"}));
-        generated.add(c2Generated);
-        ContainerFilter coreGenerated = new ContainerFilter("Core");
-        coreGenerated.add(new SpecificPathFilter("C++ Interpreter Generated", new String[] {".*core/generated/jvmtifiles/bytecodeInterpreterWithChecks.+"}));
-        coreGenerated.add(new SpecificPathFilter("jvmtifiles", new String[] {".*core/generated/jvmtifiles/.*"}));
-        generated.add(coreGenerated);
-        ContainerFilter tieredGenerated = new ContainerFilter("Tiered");
-        tieredGenerated.add(new SpecificPathFilter("C++ Interpreter Generated", new String[] {".*tiered/generated/jvmtifiles/bytecodeInterpreterWithChecks.+"}));
-        tieredGenerated.add(new SpecificPathFilter("adfiles", new String[] {".*tiered/generated/adfiles/.*"}));
-        tieredGenerated.add(new SpecificPathFilter("jvmtifiles", new String[] {".*tiered/generated/jvmtifiles/.*"}));
-        generated.add(tieredGenerated);
-        ContainerFilter kernelGenerated = new ContainerFilter("Kernel");
-        kernelGenerated.add(new SpecificPathFilter("C++ Interpreter Generated", new String[] {".*kernel/generated/jvmtifiles/bytecodeInterpreterWithChecks.+"}));
-        kernelGenerated.add(new SpecificPathFilter("jvmtifiles", new String[] {".*kernel/generated/jvmtifiles/.*"}));
-        generated.add(kernelGenerated);
-        rv.add(generated);
-
-        rv.add(new SpecificNameFilter("Precompiled Header", new String[] {"precompiled.hpp"}));
-
-        // this one is to catch files not caught by other filters
-        rv.add(new TerminatorFilter("Source Files"));
-
-        return rv;
-    }
-
-    void writeFiles(Vector<BuildConfig> allConfigs) {
-
-        Hashtable allFiles = computeAttributedFiles(allConfigs);
-
-        Vector allConfigNames = new Vector();
-        for (Iterator i = allConfigs.iterator(); i.hasNext(); ) {
-            allConfigNames.add(((BuildConfig)i.next()).get("Name"));
-        }
-
-        TreeSet sortedFiles = sortFiles(allFiles);
-
-        startTag("Files");
-
-        for (Iterator i = makeFilters(sortedFiles).iterator(); i.hasNext(); ) {
-            doWriteFiles(sortedFiles, allConfigNames, (NameFilter)i.next());
-        }
-
-
-        startTag("Filter",
-                 new String[] {
-                     "Name", "Resource Files",
-                     "Filter", "ico;cur;bmp;dlg;rc2;rct;bin;cnt;rtf;gif;jpg;jpeg;jpe"
-                 }
-                 );
-        endTag("Filter");
-
-        endTag("Files");
-    }
-
-    void doWriteFiles(TreeSet allFiles, Vector allConfigNames, NameFilter filter) {
-        startTag("Filter",
-                 new String[] {
-                     "Name",   filter.name(),
-                     "Filter", filter.filterString()
-                 }
-                 );
-
-        if (filter instanceof ContainerFilter) {
-
-            Iterator i = ((ContainerFilter)filter).babies();
-            while (i.hasNext()) {
-                doWriteFiles(allFiles, allConfigNames, (NameFilter)i.next());
-            }
-
-        } else {
-
-            Iterator i = allFiles.iterator();
-            while (i.hasNext()) {
-                FileInfo fi = (FileInfo)i.next();
-
-                if (!filter.match(fi)) {
-                    continue;
-                }
-
-                startTag("File",
-                         new String[] {
-                             "RelativePath", fi.full.replace('/', '\\')
-                         }
-                         );
-
-                FileAttribute a = fi.attr;
-                if (a.pchRoot) {
-                    writeCustomToolConfig(allConfigNames,
-                                          new String[] {
-                                              "Name", "VCCLCompilerTool",
-                                              "UsePrecompiledHeader", "1"
-                                          });
-                }
-
-                if (a.noPch) {
-                    writeCustomToolConfig(allConfigNames,
-                                          new String[] {
-                                              "Name", "VCCLCompilerTool",
-                                              "UsePrecompiledHeader", "0"
-                                          });
-                }
-
-                if (a.configs != null) {
-                    for (Iterator j=allConfigNames.iterator(); j.hasNext();) {
-                        String cfg = (String)j.next();
-                        if (!a.configs.contains(cfg)) {
-                            startTag("FileConfiguration",
-                                     new String[] {
-                                         "Name", cfg,
-                                         "ExcludedFromBuild", "TRUE"
-                                     });
-                            endTag("FileConfiguration");
-
-                        }
-                    }
-                }
+   public void writeProjectFile(String projectFileName, String projectName,
+         Vector<BuildConfig> allConfigs) throws IOException {
+      System.out.println();
+      System.out.println("    Writing .vcproj file: " + projectFileName);
+      // If we got this far without an error, we're safe to actually
+      // write the .vcproj file
+      printWriter = new PrintWriter(new FileWriter(projectFileName));
 
-                endTag("File");
+      printWriter
+      .println("<?xml version=\"1.0\" encoding=\"windows-1251\"?>");
+      startTag("VisualStudioProject", new String[] { "ProjectType",
+            "Visual C++", "Version", projectVersion(), "Name", projectName,
+            "ProjectGUID", "{8822CB5C-1C41-41C2-8493-9F6E1994338B}",
+            "SccProjectName", "", "SccLocalPath", "" });
+      startTag("Platforms");
+      tag("Platform",
+            new String[] { "Name",
+            (String) BuildConfig.getField(null, "PlatformName") });
+      endTag();
 
-                // we not gonna look at this file anymore
-                i.remove();
-            }
-        }
-
-        endTag("Filter");
-    }
-
-
-    void writeConfiguration(BuildConfig cfg) {
-        startTag("Configuration",
-                 new String[] {
-                     "Name", cfg.get("Name"),
-                     "OutputDirectory",  cfg.get("OutputDir"),
-                     "IntermediateDirectory",  cfg.get("OutputDir"),
-                     "ConfigurationType", "2",
-                     "UseOfMFC", "0",
-                     "ATLMinimizesCRunTimeLibraryUsage", "FALSE"
-                 }
-                 );
-
-
-
-        tagV("Tool", cfg.getV("CompilerFlags"));
-
-        tag("Tool",
-            new String[] {
-                "Name", "VCCustomBuildTool"
-            }
-            );
+      startTag("Configurations");
 
-        tagV("Tool", cfg.getV("LinkerFlags"));
+      for (BuildConfig cfg : allConfigs) {
+         writeConfiguration(cfg);
+      }
 
-        tag("Tool",
-            new String[] {
-               "Name", "VCPostBuildEventTool",
-                "Description", BuildConfig.getFieldString(null, "PostbuildDescription"),
-                //Caution: String.replace(String,String) is available from JDK5 onwards only
-                "CommandLine", cfg.expandFormat(BuildConfig.getFieldString(null, "PostbuildCommand").replace
-                   ("\t", "&#x0D;&#x0A;"))
-            }
-            );
-
-        tag("Tool",
-            new String[] {
-                "Name", "VCPreBuildEventTool"
-            }
-            );
+      endTag();
 
-        tag("Tool",
-            new String[] {
-                "Name", "VCPreLinkEventTool",
-                "Description", BuildConfig.getFieldString(null, "PrelinkDescription"),
-                //Caution: String.replace(String,String) is available from JDK5 onwards only
-                "CommandLine", cfg.expandFormat(BuildConfig.getFieldString(null, "PrelinkCommand").replace
-                   ("\t", "&#x0D;&#x0A;"))
-            }
-            );
-
-        tag("Tool",
-            new String[] {
-                "Name", "VCResourceCompilerTool",
-                // XXX???
-                "PreprocessorDefinitions", "NDEBUG",
-                "Culture", "1033"
-            }
-            );
-
-        tag("Tool",
-            new String[] {
-                "Name", "VCMIDLTool",
-                "PreprocessorDefinitions", "NDEBUG",
-                "MkTypLibCompatible", "TRUE",
-                "SuppressStartupBanner", "TRUE",
-                "TargetEnvironment", "1",
-                "TypeLibraryName", cfg.get("OutputDir") + Util.sep + "vm.tlb",
-                "HeaderFileName", ""
-            }
-            );
-
-        endTag("Configuration");
-    }
-
-    int indent;
-
-    private void startTagPrim(String name,
-            String[] attrs,
-            boolean close) {
-        startTagPrim(name, attrs, close, true);
-    }
-
-    private void startTagPrim(String name,
-                              String[] attrs,
-                              boolean close,
-                              boolean newline) {
-        doIndent();
-        printWriter.print("<"+name);
-        indent++;
-
-        if (attrs != null && attrs.length > 0) {
-            for (int i=0; i<attrs.length; i+=2) {
-                printWriter.print(" " + attrs[i]+"=\""+attrs[i+1]+"\"");
-                if (i < attrs.length - 2) {
-                }
-            }
-        }
-
-        if (close) {
-            indent--;
-            printWriter.print(" />");
-        } else {
-                printWriter.print(">");
-        }
-        if(newline) {
-                printWriter.println();
-        }
-    }
-
-    void startTag(String name, String... attrs) {
-        startTagPrim(name, attrs, false);
-    }
-
-    void startTagV(String name, Vector attrs) {
-        String s[] = new String [attrs.size()];
-         for (int i=0; i<attrs.size(); i++) {
-             s[i] = (String)attrs.elementAt(i);
-         }
-        startTagPrim(name, s, false);
-    }
-
-    void endTag(String name) {
-        indent--;
-        doIndent();
-        printWriter.println("</"+name+">");
-    }
-
-    void tag(String name, String... attrs) {
-        startTagPrim(name, attrs, true);
-    }
-
-    void tagData(String name, String data) {
-        doIndent();
-        printWriter.print("<"+name+">");
-        printWriter.print(data);
-        printWriter.println("</"+name+">");
-    }
-
-    void tagData(String name, String data, String... attrs) {
-        startTagPrim(name, attrs, false, false);
-        printWriter.print(data);
-        printWriter.println("</"+name+">");
-        indent--;
-    }
-
-    void tagV(String name, Vector attrs) {
-         String s[] = new String [attrs.size()];
-         for (int i=0; i<attrs.size(); i++) {
-             s[i] = (String)attrs.elementAt(i);
+      tag("References");
+
+      writeFiles(allConfigs);
+
+      tag("Globals");
+
+      endTag();
+      printWriter.close();
+
+      System.out.println("    Done.");
+   }
+
+   void writeCustomToolConfig(Vector<BuildConfig> configs, String[] customToolAttrs) {
+      for (BuildConfig cfg : configs) {
+         startTag("FileConfiguration",
+               new String[] { "Name", (String) cfg.get("Name") });
+         tag("Tool", customToolAttrs);
+
+         endTag();
+      }
+   }
+
+   void writeFiles(Vector<BuildConfig> allConfigs) {
+
+      // This code assummes there are no config specific includes.
+      startTag("Files");
+      String sourceBase = BuildConfig.getFieldString(null, "SourceBase");
+
+      // Use first config for all global absolute includes.
+      BuildConfig baseConfig = allConfigs.firstElement();
+      Vector<String> rv = new Vector<String>();
+
+      // Then use first config for all relative includes
+      Vector<String> ri = new Vector<String>();
+      baseConfig.collectRelevantVectors(ri, "RelativeSrcInclude");
+      for (String f : ri) {
+         rv.add(sourceBase + Util.sep + f);
+      }
+
+      baseConfig.collectRelevantVectors(rv, "AbsoluteSrcInclude");
+
+      handleIncludes(rv, allConfigs);
+
+      startTag("Filter", new String[] { "Name", "Resource Files", "Filter",
+      "ico;cur;bmp;dlg;rc2;rct;bin;cnt;rtf;gif;jpg;jpeg;jpe" });
+      endTag();
+
+      endTag();
+   }
+
+   // Will visit file tree for each include
+   private void handleIncludes(Vector<String> includes, Vector<BuildConfig> allConfigs) {
+      for (String path : includes)  {
+         FileTreeCreatorVC7 ftc = new FileTreeCreatorVC7(FileSystems.getDefault().getPath(path) , allConfigs, this);
+         try {
+            ftc.writeFileTree();
+         } catch (IOException e) {
+            e.printStackTrace();
          }
-         startTagPrim(name, s, true);
-    }
+      }
+   }
+
+   void writeConfiguration(BuildConfig cfg) {
+      startTag("Configuration", new String[] { "Name", cfg.get("Name"),
+            "OutputDirectory", cfg.get("OutputDir"),
+            "IntermediateDirectory", cfg.get("OutputDir"),
+            "ConfigurationType", "2", "UseOfMFC", "0",
+            "ATLMinimizesCRunTimeLibraryUsage", "FALSE" });
+
+      tagV("Tool", cfg.getV("CompilerFlags"));
 
+      tag("Tool", new String[] { "Name", "VCCustomBuildTool" });
 
-    void doIndent() {
-        for (int i=0; i<indent; i++) {
-            printWriter.print("  ");
-        }
-    }
+      tagV("Tool", cfg.getV("LinkerFlags"));
 
-    protected String getProjectExt() {
-        return ".vcproj";
-    }
+      tag("Tool",
+            new String[] {
+            "Name",
+            "VCPostBuildEventTool",
+            "Description",
+            BuildConfig
+            .getFieldString(null, "PostbuildDescription"),
+            // Caution: String.replace(String,String) is available
+            // from JDK5 onwards only
+            "CommandLine",
+            cfg.expandFormat(BuildConfig.getFieldString(null,
+                  "PostbuildCommand").replace("\t",
+                        "&#x0D;&#x0A;")) });
+
+      tag("Tool", new String[] { "Name", "VCPreBuildEventTool" });
+
+      tag("Tool",
+            new String[] {
+            "Name",
+            "VCPreLinkEventTool",
+            "Description",
+            BuildConfig.getFieldString(null, "PrelinkDescription"),
+            // Caution: String.replace(String,String) is available
+            // from JDK5 onwards only
+            "CommandLine",
+            cfg.expandFormat(BuildConfig.getFieldString(null,
+                  "PrelinkCommand").replace("\t", "&#x0D;&#x0A;")) });
+
+      tag("Tool", new String[] { "Name", "VCResourceCompilerTool",
+            "PreprocessorDefinitions", "NDEBUG", "Culture", "1033" });
+
+      tag("Tool", new String[] { "Name", "VCMIDLTool",
+            "PreprocessorDefinitions", "NDEBUG", "MkTypLibCompatible",
+            "TRUE", "SuppressStartupBanner", "TRUE", "TargetEnvironment",
+            "1", "TypeLibraryName",
+            cfg.get("OutputDir") + Util.sep + "vm.tlb", "HeaderFileName",
+      "" });
+
+      endTag();
+   }
+
+
+
+   protected String getProjectExt() {
+      return ".vcproj";
+   }
 }
 
 class CompilerInterfaceVC7 extends CompilerInterface {
-    void getBaseCompilerFlags_common(Vector defines, Vector includes, String outDir,Vector rv) {
-
-        // advanced M$ IDE (2003) can only recognize name if it's first or
-        // second attribute in the tag - go guess
-        addAttr(rv, "Name", "VCCLCompilerTool");
-        addAttr(rv, "AdditionalIncludeDirectories", Util.join(",", includes));
-        addAttr(rv, "PreprocessorDefinitions",
-                                Util.join(";", defines).replace("\"","&quot;"));
-        addAttr(rv, "PrecompiledHeaderThrough", "precompiled.hpp");
-        addAttr(rv, "PrecompiledHeaderFile", outDir+Util.sep+"vm.pch");
-        addAttr(rv, "AssemblerListingLocation", outDir);
-        addAttr(rv, "ObjectFile", outDir+Util.sep);
-        addAttr(rv, "ProgramDataBaseFileName", outDir+Util.sep+"jvm.pdb");
-        // Set /nologo optin
-        addAttr(rv, "SuppressStartupBanner", "TRUE");
-        // Surpass the default /Tc or /Tp. 0 is compileAsDefault
-        addAttr(rv, "CompileAs", "0");
-        // Set /W3 option. 3 is warningLevel_3
-        addAttr(rv, "WarningLevel", "3");
-        // Set /WX option,
-        addAttr(rv, "WarnAsError", "TRUE");
-        // Set /GS option
-        addAttr(rv, "BufferSecurityCheck", "FALSE");
-        // Set /Zi option. 3 is debugEnabled
-        addAttr(rv, "DebugInformationFormat", "3");
-    }
-    Vector getBaseCompilerFlags(Vector defines, Vector includes, String outDir) {
-        Vector rv = new Vector();
-
-        getBaseCompilerFlags_common(defines,includes, outDir, rv);
-        // Set /Yu option. 3 is pchUseUsingSpecific
-        // Note: Starting VC8 pchUseUsingSpecific is 2 !!!
-        addAttr(rv, "UsePrecompiledHeader", "3");
-        // Set /EHsc- option
-        addAttr(rv, "ExceptionHandling", "FALSE");
-
-        return rv;
-    }
-
-    Vector getBaseLinkerFlags(String outDir, String outDll, String platformName) {
-        Vector rv = new Vector();
-
-        addAttr(rv, "Name", "VCLinkerTool");
-        addAttr(rv, "AdditionalOptions",
-                "/export:JNI_GetDefaultJavaVMInitArgs " +
-                "/export:JNI_CreateJavaVM " +
-                "/export:JVM_FindClassFromBootLoader "+
-                "/export:JNI_GetCreatedJavaVMs "+
-                "/export:jio_snprintf /export:jio_printf "+
-                "/export:jio_fprintf /export:jio_vfprintf "+
-                "/export:jio_vsnprintf "+
-                "/export:JVM_GetVersionInfo "+
-                "/export:JVM_GetThreadStateNames "+
-                "/export:JVM_GetThreadStateValues "+
-                "/export:JVM_InitAgentProperties ");
-        addAttr(rv, "AdditionalDependencies", "Wsock32.lib winmm.lib");
-        addAttr(rv, "OutputFile", outDll);
-        // Set /INCREMENTAL option. 1 is linkIncrementalNo
-        addAttr(rv, "LinkIncremental", "1");
-        addAttr(rv, "SuppressStartupBanner", "TRUE");
-        addAttr(rv, "ModuleDefinitionFile", outDir+Util.sep+"vm.def");
-        addAttr(rv, "ProgramDatabaseFile", outDir+Util.sep+"jvm.pdb");
-        // Set /SUBSYSTEM option. 2 is subSystemWindows
-        addAttr(rv, "SubSystem", "2");
-        addAttr(rv, "BaseAddress", "0x8000000");
-        addAttr(rv, "ImportLibrary", outDir+Util.sep+"jvm.lib");
-        if(platformName.equals("Win32")) {
-            // Set /MACHINE option. 1 is X86
-            addAttr(rv, "TargetMachine", "1");
-        } else {
-            // Set /MACHINE option. 17 is X64
-            addAttr(rv, "TargetMachine", "17");
-        }
-
-        return rv;
-    }
-
-    void  getDebugCompilerFlags_common(String opt,Vector rv) {
-
-        // Set /On option
-        addAttr(rv, "Optimization", opt);
-        // Set /FR option. 1 is brAllInfo
-        addAttr(rv, "BrowseInformation", "1");
-        addAttr(rv, "BrowseInformationFile", "$(IntDir)" + Util.sep);
-        // Set /MD option. 2 is rtMultiThreadedDLL
-        addAttr(rv, "RuntimeLibrary", "2");
-        // Set /Oy- option
-        addAttr(rv, "OmitFramePointers", "FALSE");
-
-    }
-
-    Vector getDebugCompilerFlags(String opt) {
-        Vector rv = new Vector();
-
-        getDebugCompilerFlags_common(opt,rv);
-
-        return rv;
-    }
-
-    Vector getDebugLinkerFlags() {
-        Vector rv = new Vector();
-
-        addAttr(rv, "GenerateDebugInformation", "TRUE"); // == /DEBUG option
-
-        return rv;
-    }
-
-    void getAdditionalNonKernelLinkerFlags(Vector rv) {
-        extAttr(rv, "AdditionalOptions",
-                "/export:AsyncGetCallTrace ");
-    }
-
-    void getProductCompilerFlags_common(Vector rv) {
-        // Set /O2 option. 2 is optimizeMaxSpeed
-        addAttr(rv, "Optimization", "2");
-        // Set /Oy- option
-        addAttr(rv, "OmitFramePointers", "FALSE");
-        // Set /Ob option.  1 is expandOnlyInline
-        addAttr(rv, "InlineFunctionExpansion", "1");
-        // Set /GF option.
-        addAttr(rv, "StringPooling", "TRUE");
-        // Set /MD option. 2 is rtMultiThreadedDLL
-        addAttr(rv, "RuntimeLibrary", "2");
-        // Set /Gy option
-        addAttr(rv, "EnableFunctionLevelLinking", "TRUE");
-    }
-
-    Vector getProductCompilerFlags() {
-        Vector rv = new Vector();
-
-        getProductCompilerFlags_common(rv);
-
-        return rv;
-    }
-
-    Vector getProductLinkerFlags() {
-        Vector rv = new Vector();
-
-        // Set /OPT:REF option. 2 is optReferences
-        addAttr(rv, "OptimizeReferences", "2");
-        // Set /OPT:optFolding option. 2 is optFolding
-        addAttr(rv, "EnableCOMDATFolding", "2");
-
-        return rv;
-    }
-
-    String getOptFlag() {
-        return "2";
-    }
-
-    String getNoOptFlag() {
-        return "0";
-    }
-
-    String makeCfgName(String flavourBuild, String platform) {
-        return  flavourBuild + "|" + platform;
-    }
+   void getBaseCompilerFlags_common(Vector defines, Vector includes,
+         String outDir, Vector rv) {
+
+      // advanced M$ IDE (2003) can only recognize name if it's first or
+      // second attribute in the tag - go guess
+      addAttr(rv, "Name", "VCCLCompilerTool");
+      addAttr(rv, "AdditionalIncludeDirectories", Util.join(",", includes));
+      addAttr(rv, "PreprocessorDefinitions",
+            Util.join(";", defines).replace("\"", "&quot;"));
+      addAttr(rv, "PrecompiledHeaderThrough", "precompiled.hpp");
+      addAttr(rv, "PrecompiledHeaderFile", outDir + Util.sep + "vm.pch");
+      addAttr(rv, "AssemblerListingLocation", outDir);
+      addAttr(rv, "ObjectFile", outDir + Util.sep);
+      addAttr(rv, "ProgramDataBaseFileName", outDir + Util.sep + "jvm.pdb");
+      // Set /nologo optin
+      addAttr(rv, "SuppressStartupBanner", "TRUE");
+      // Surpass the default /Tc or /Tp. 0 is compileAsDefault
+      addAttr(rv, "CompileAs", "0");
+      // Set /W3 option. 3 is warningLevel_3
+      addAttr(rv, "WarningLevel", "3");
+      // Set /WX option,
+      addAttr(rv, "WarnAsError", "TRUE");
+      // Set /GS option
+      addAttr(rv, "BufferSecurityCheck", "FALSE");
+      // Set /Zi option. 3 is debugEnabled
+      addAttr(rv, "DebugInformationFormat", "3");
+   }
+
+   Vector getBaseCompilerFlags(Vector defines, Vector includes, String outDir) {
+      Vector rv = new Vector();
+
+      getBaseCompilerFlags_common(defines, includes, outDir, rv);
+      // Set /Yu option. 3 is pchUseUsingSpecific
+      // Note: Starting VC8 pchUseUsingSpecific is 2 !!!
+      addAttr(rv, "UsePrecompiledHeader", "3");
+      // Set /EHsc- option
+      addAttr(rv, "ExceptionHandling", "FALSE");
+
+      return rv;
+   }
+
+   Vector getBaseLinkerFlags(String outDir, String outDll, String platformName) {
+      Vector rv = new Vector();
+
+      addAttr(rv, "Name", "VCLinkerTool");
+      addAttr(rv, "AdditionalOptions",
+            "/export:JNI_GetDefaultJavaVMInitArgs "
+                  + "/export:JNI_CreateJavaVM "
+                  + "/export:JVM_FindClassFromBootLoader "
+                  + "/export:JNI_GetCreatedJavaVMs "
+                  + "/export:jio_snprintf /export:jio_printf "
+                  + "/export:jio_fprintf /export:jio_vfprintf "
+                  + "/export:jio_vsnprintf "
+                  + "/export:JVM_GetVersionInfo "
+                  + "/export:JVM_GetThreadStateNames "
+                  + "/export:JVM_GetThreadStateValues "
+                  + "/export:JVM_InitAgentProperties ");
+      addAttr(rv, "AdditionalDependencies", "Wsock32.lib winmm.lib");
+      addAttr(rv, "OutputFile", outDll);
+      // Set /INCREMENTAL option. 1 is linkIncrementalNo
+      addAttr(rv, "LinkIncremental", "1");
+      addAttr(rv, "SuppressStartupBanner", "TRUE");
+      addAttr(rv, "ModuleDefinitionFile", outDir + Util.sep + "vm.def");
+      addAttr(rv, "ProgramDatabaseFile", outDir + Util.sep + "jvm.pdb");
+      // Set /SUBSYSTEM option. 2 is subSystemWindows
+      addAttr(rv, "SubSystem", "2");
+      addAttr(rv, "BaseAddress", "0x8000000");
+      addAttr(rv, "ImportLibrary", outDir + Util.sep + "jvm.lib");
+      if (platformName.equals("Win32")) {
+         // Set /MACHINE option. 1 is X86
+         addAttr(rv, "TargetMachine", "1");
+      } else {
+         // Set /MACHINE option. 17 is X64
+         addAttr(rv, "TargetMachine", "17");
+      }
+
+      return rv;
+   }
+
+   void getDebugCompilerFlags_common(String opt, Vector rv) {
+
+      // Set /On option
+      addAttr(rv, "Optimization", opt);
+      // Set /FR option. 1 is brAllInfo
+      addAttr(rv, "BrowseInformation", "1");
+      addAttr(rv, "BrowseInformationFile", "$(IntDir)" + Util.sep);
+      // Set /MD option. 2 is rtMultiThreadedDLL
+      addAttr(rv, "RuntimeLibrary", "2");
+      // Set /Oy- option
+      addAttr(rv, "OmitFramePointers", "FALSE");
+
+   }
+
+   Vector getDebugCompilerFlags(String opt) {
+      Vector rv = new Vector();
+
+      getDebugCompilerFlags_common(opt, rv);
+
+      return rv;
+   }
+
+   Vector getDebugLinkerFlags() {
+      Vector rv = new Vector();
+
+      addAttr(rv, "GenerateDebugInformation", "TRUE"); // == /DEBUG option
+
+      return rv;
+   }
+
+   void getAdditionalNonKernelLinkerFlags(Vector rv) {
+      extAttr(rv, "AdditionalOptions", "/export:AsyncGetCallTrace ");
+   }
+
+   void getProductCompilerFlags_common(Vector rv) {
+      // Set /O2 option. 2 is optimizeMaxSpeed
+      addAttr(rv, "Optimization", "2");
+      // Set /Oy- option
+      addAttr(rv, "OmitFramePointers", "FALSE");
+      // Set /Ob option. 1 is expandOnlyInline
+      addAttr(rv, "InlineFunctionExpansion", "1");
+      // Set /GF option.
+      addAttr(rv, "StringPooling", "TRUE");
+      // Set /MD option. 2 is rtMultiThreadedDLL
+      addAttr(rv, "RuntimeLibrary", "2");
+      // Set /Gy option
+      addAttr(rv, "EnableFunctionLevelLinking", "TRUE");
+   }
+
+   Vector getProductCompilerFlags() {
+      Vector rv = new Vector();
+
+      getProductCompilerFlags_common(rv);
+
+      return rv;
+   }
+
+   Vector getProductLinkerFlags() {
+      Vector rv = new Vector();
+
+      // Set /OPT:REF option. 2 is optReferences
+      addAttr(rv, "OptimizeReferences", "2");
+      // Set /OPT:optFolding option. 2 is optFolding
+      addAttr(rv, "EnableCOMDATFolding", "2");
+
+      return rv;
+   }
+
+   String getOptFlag() {
+      return "2";
+   }
+
+   String getNoOptFlag() {
+      return "0";
+   }
+
+   String makeCfgName(String flavourBuild, String platform) {
+      return flavourBuild + "|" + platform;
+   }
+
 }
diff --git a/src/share/vm/adlc/formssel.cpp b/src/share/vm/adlc/formssel.cpp
index cccce31c2..fa5f91ca8 100644
--- a/src/share/vm/adlc/formssel.cpp
+++ b/src/share/vm/adlc/formssel.cpp
@@ -751,6 +751,7 @@ bool InstructForm::captures_bottom_type(FormDict &globals) const {
         !strcmp(_matrule->_rChild->_opType,"DecodeN")    ||
         !strcmp(_matrule->_rChild->_opType,"EncodeP")    ||
         !strcmp(_matrule->_rChild->_opType,"LoadN")      ||
+        !strcmp(_matrule->_rChild->_opType,"GetAndSetN") ||
         !strcmp(_matrule->_rChild->_opType,"LoadNKlass") ||
         !strcmp(_matrule->_rChild->_opType,"CreateEx")   ||  // type of exception
         !strcmp(_matrule->_rChild->_opType,"CheckCastPP")) ) return true;
@@ -3399,7 +3400,9 @@ int MatchNode::needs_ideal_memory_edge(FormDict &globals) const {
     "StorePConditional", "StoreIConditional", "StoreLConditional",
     "CompareAndSwapI", "CompareAndSwapL", "CompareAndSwapP", "CompareAndSwapN",
     "StoreCM",
-    "ClearArray"
+    "ClearArray",
+    "GetAndAddI", "GetAndSetI", "GetAndSetP",
+    "GetAndAddL", "GetAndSetL", "GetAndSetN",
   };
   int cnt = sizeof(needs_ideal_memory_list)/sizeof(char*);
   if( strcmp(_opType,"PrefetchRead")==0 ||
@@ -4040,18 +4043,27 @@ Form::DataType MatchRule::is_ideal_load() const {
 }
 
 bool MatchRule::is_vector() const {
-  if( _rChild ) {
+  static const char *vector_list[] = {
+    "AddVB","AddVS","AddVI","AddVL","AddVF","AddVD",
+    "SubVB","SubVS","SubVI","SubVL","SubVF","SubVD",
+    "MulVS","MulVI","MulVF","MulVD",
+    "DivVF","DivVD",
+    "AndV" ,"XorV" ,"OrV",
+    "LShiftVB","LShiftVS","LShiftVI","LShiftVL",
+    "RShiftVB","RShiftVS","RShiftVI","RShiftVL",
+    "URShiftVB","URShiftVS","URShiftVI","URShiftVL",
+    "ReplicateB","ReplicateS","ReplicateI","ReplicateL","ReplicateF","ReplicateD",
+    "LoadVector","StoreVector",
+    // Next are not supported currently.
+    "PackB","PackS","PackI","PackL","PackF","PackD","Pack2L","Pack2D",
+    "ExtractB","ExtractUB","ExtractC","ExtractS","ExtractI","ExtractL","ExtractF","ExtractD"
+  };
+  int cnt = sizeof(vector_list)/sizeof(char*);
+  if (_rChild) {
     const char  *opType = _rChild->_opType;
-    if( strcmp(opType,"ReplicateB")==0 ||
-        strcmp(opType,"ReplicateS")==0 ||
-        strcmp(opType,"ReplicateI")==0 ||
-        strcmp(opType,"ReplicateL")==0 ||
-        strcmp(opType,"ReplicateF")==0 ||
-        strcmp(opType,"ReplicateD")==0 ||
-        strcmp(opType,"LoadVector")==0 ||
-        strcmp(opType,"StoreVector")==0 ||
-        0 /* 0 to line up columns nicely */ )
-      return true;
+    for (int i=0; i<cnt; i++)
+      if (strcmp(opType,vector_list[i]) == 0)
+        return true;
   }
   return false;
 }
diff --git a/src/share/vm/adlc/output_c.cpp b/src/share/vm/adlc/output_c.cpp
index 68a0ff818..fe6ba16c8 100644
--- a/src/share/vm/adlc/output_c.cpp
+++ b/src/share/vm/adlc/output_c.cpp
@@ -1606,6 +1606,12 @@ void ArchDesc::defineExpand(FILE *fp, InstructForm *node) {
         fprintf(fp, "  ((MachFastLockNode*)n%d)->_counters = _counters;\n",cnt);
       }
 
+      // Fill in the bottom_type where requested
+      if (node->captures_bottom_type(_globalNames) &&
+          new_inst->captures_bottom_type(_globalNames)) {
+        fprintf(fp, "  ((MachTypeNode*)n%d)->_bottom_type = bottom_type();\n", cnt);
+      }
+
       const char *resultOper = new_inst->reduce_result();
       fprintf(fp,"  n%d->set_opnd_array(0, state->MachOperGenerator( %s, C ));\n",
               cnt, machOperEnum(resultOper));
@@ -1767,7 +1773,7 @@ void ArchDesc::defineExpand(FILE *fp, InstructForm *node) {
         }
 
         fprintf(fp,"  kill = ");
-        fprintf(fp,"new (C, 1) MachProjNode( %s, %d, (%s), Op_%s );\n",
+        fprintf(fp,"new (C) MachProjNode( %s, %d, (%s), Op_%s );\n",
                 machNode, proj_no++, regmask, ideal_type);
         fprintf(fp,"  proj_list.push(kill);\n");
       }
diff --git a/src/share/vm/asm/codeBuffer.cpp b/src/share/vm/asm/codeBuffer.cpp
index 64994a9f4..bb1ae18fc 100644
--- a/src/share/vm/asm/codeBuffer.cpp
+++ b/src/share/vm/asm/codeBuffer.cpp
@@ -1026,25 +1026,30 @@ class CodeComment: public CHeapObj<mtCode> {
     }
     return a;
   }
+
+  // Convenience for add_comment.
+  CodeComment* find_last(intptr_t offset) {
+    CodeComment* a = find(offset);
+    if (a != NULL) {
+      while ((a->_next != NULL) && (a->_next->_offset == offset)) {
+        a = a->_next;
+      }
+    }
+    return a;
+  }
 };
 
 
 void CodeComments::add_comment(intptr_t offset, const char * comment) {
-  CodeComment* c = new CodeComment(offset, comment);
-  CodeComment* insert = NULL;
-  if (_comments != NULL) {
-    CodeComment* c = _comments->find(offset);
-    insert = c;
-    while (c && c->offset() == offset) {
-      insert = c;
-      c = c->next();
-    }
-  }
-  if (insert) {
-    // insert after comments with same offset
-    c->set_next(insert->next());
-    insert->set_next(c);
+  CodeComment* c      = new CodeComment(offset, comment);
+  CodeComment* inspos = (_comments == NULL) ? NULL : _comments->find_last(offset);
+
+  if (inspos) {
+    // insert after already existing comments with same offset
+    c->set_next(inspos->next());
+    inspos->set_next(c);
   } else {
+    // no comments with such offset, yet. Insert before anything else.
     c->set_next(_comments);
     _comments = c;
   }
@@ -1052,12 +1057,11 @@ void CodeComments::add_comment(intptr_t offset, const char * comment) {
 
 
 void CodeComments::assign(CodeComments& other) {
-  assert(_comments == NULL, "don't overwrite old value");
   _comments = other._comments;
 }
 
 
-void CodeComments::print_block_comment(outputStream* stream, intptr_t offset) {
+void CodeComments::print_block_comment(outputStream* stream, intptr_t offset) const {
   if (_comments != NULL) {
     CodeComment* c = _comments->find(offset);
     while (c && c->offset() == offset) {
@@ -1085,6 +1089,7 @@ void CodeComments::free() {
 
 
 void CodeBuffer::decode() {
+  ttyLocker ttyl;
   Disassembler::decode(decode_begin(), insts_end());
   _decode_begin = insts_end();
 }
@@ -1096,6 +1101,7 @@ void CodeBuffer::skip_decode() {
 
 
 void CodeBuffer::decode_all() {
+  ttyLocker ttyl;
   for (int n = 0; n < (int)SECT_LIMIT; n++) {
     // dump contents of each section
     CodeSection* cs = code_section(n);
diff --git a/src/share/vm/asm/codeBuffer.hpp b/src/share/vm/asm/codeBuffer.hpp
index 0e01a9fbf..63dba2dbb 100644
--- a/src/share/vm/asm/codeBuffer.hpp
+++ b/src/share/vm/asm/codeBuffer.hpp
@@ -253,7 +253,7 @@ public:
   }
 
   void add_comment(intptr_t offset, const char * comment) PRODUCT_RETURN;
-  void print_block_comment(outputStream* stream, intptr_t offset)  PRODUCT_RETURN;
+  void print_block_comment(outputStream* stream, intptr_t offset) const PRODUCT_RETURN;
   void assign(CodeComments& other)  PRODUCT_RETURN;
   void free() PRODUCT_RETURN;
 };
diff --git a/src/share/vm/asm/register.hpp b/src/share/vm/asm/register.hpp
index c0b247de2..5a94a0ac0 100644
--- a/src/share/vm/asm/register.hpp
+++ b/src/share/vm/asm/register.hpp
@@ -103,8 +103,8 @@ inline void assert_different_registers(
 ) {
   assert(
     a != b,
-    err_msg("registers must be different: a=%d, b=%d",
-            a, b)
+    err_msg_res("registers must be different: a=%d, b=%d",
+                a, b)
   );
 }
 
@@ -117,8 +117,8 @@ inline void assert_different_registers(
   assert(
     a != b && a != c
            && b != c,
-    err_msg("registers must be different: a=%d, b=%d, c=%d",
-            a, b, c)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d",
+                a, b, c)
   );
 }
 
@@ -133,8 +133,8 @@ inline void assert_different_registers(
     a != b && a != c && a != d
            && b != c && b != d
                      && c != d,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d",
-            a, b, c, d)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d",
+                a, b, c, d)
   );
 }
 
@@ -151,8 +151,8 @@ inline void assert_different_registers(
            && b != c && b != d && b != e
                      && c != d && c != e
                                && d != e,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d",
-            a, b, c, d, e)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d",
+                a, b, c, d, e)
   );
 }
 
@@ -171,8 +171,8 @@ inline void assert_different_registers(
                      && c != d && c != e && c != f
                                && d != e && d != f
                                          && e != f,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d",
-            a, b, c, d, e, f)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d",
+                a, b, c, d, e, f)
   );
 }
 
@@ -193,8 +193,8 @@ inline void assert_different_registers(
                                && d != e && d != f && d != g
                                          && e != f && e != g
                                                    && f != g,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d",
-            a, b, c, d, e, f, g)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d",
+                a, b, c, d, e, f, g)
   );
 }
 
@@ -217,8 +217,8 @@ inline void assert_different_registers(
                                          && e != f && e != g && e != h
                                                    && f != g && f != h
                                                              && g != h,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d, h=%d",
-            a, b, c, d, e, f, g, h)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d, h=%d",
+                a, b, c, d, e, f, g, h)
   );
 }
 
@@ -243,8 +243,8 @@ inline void assert_different_registers(
                                                    && f != g && f != h && f != i
                                                              && g != h && g != i
                                                                        && h != i,
-    err_msg("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d, h=%d, i=%d",
-            a, b, c, d, e, f, g, h, i)
+    err_msg_res("registers must be different: a=%d, b=%d, c=%d, d=%d, e=%d, f=%d, g=%d, h=%d, i=%d",
+                a, b, c, d, e, f, g, h, i)
   );
 }
 
diff --git a/src/share/vm/c1/c1_Canonicalizer.cpp b/src/share/vm/c1/c1_Canonicalizer.cpp
index 7d9ddf852..1fcdd64e6 100644
--- a/src/share/vm/c1/c1_Canonicalizer.cpp
+++ b/src/share/vm/c1/c1_Canonicalizer.cpp
@@ -931,6 +931,7 @@ void Canonicalizer::do_UnsafeGetRaw(UnsafeGetRaw* x) { if (OptimizeUnsafes) do_U
 void Canonicalizer::do_UnsafePutRaw(UnsafePutRaw* x) { if (OptimizeUnsafes) do_UnsafeRawOp(x); }
 void Canonicalizer::do_UnsafeGetObject(UnsafeGetObject* x) {}
 void Canonicalizer::do_UnsafePutObject(UnsafePutObject* x) {}
+void Canonicalizer::do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) {}
 void Canonicalizer::do_UnsafePrefetchRead (UnsafePrefetchRead*  x) {}
 void Canonicalizer::do_UnsafePrefetchWrite(UnsafePrefetchWrite* x) {}
 void Canonicalizer::do_ProfileCall(ProfileCall* x) {}
diff --git a/src/share/vm/c1/c1_Canonicalizer.hpp b/src/share/vm/c1/c1_Canonicalizer.hpp
index b7ab8f5ae..b7c9f7e87 100644
--- a/src/share/vm/c1/c1_Canonicalizer.hpp
+++ b/src/share/vm/c1/c1_Canonicalizer.hpp
@@ -100,6 +100,7 @@ class Canonicalizer: InstructionVisitor {
   virtual void do_UnsafePutRaw   (UnsafePutRaw*    x);
   virtual void do_UnsafeGetObject(UnsafeGetObject* x);
   virtual void do_UnsafePutObject(UnsafePutObject* x);
+  virtual void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x);
   virtual void do_UnsafePrefetchRead (UnsafePrefetchRead*  x);
   virtual void do_UnsafePrefetchWrite(UnsafePrefetchWrite* x);
   virtual void do_ProfileCall    (ProfileCall*     x);
diff --git a/src/share/vm/c1/c1_Compilation.cpp b/src/share/vm/c1/c1_Compilation.cpp
index d4fe226ea..b0ef8dc04 100644
--- a/src/share/vm/c1/c1_Compilation.cpp
+++ b/src/share/vm/c1/c1_Compilation.cpp
@@ -346,7 +346,8 @@ void Compilation::install_code(int frame_size) {
     implicit_exception_table(),
     compiler(),
     _env->comp_level(),
-    has_unsafe_access()
+    has_unsafe_access(),
+    SharedRuntime::is_wide_vector(max_vector_size())
   );
 }
 
diff --git a/src/share/vm/c1/c1_Compilation.hpp b/src/share/vm/c1/c1_Compilation.hpp
index 376b6f337..d5c224afd 100644
--- a/src/share/vm/c1/c1_Compilation.hpp
+++ b/src/share/vm/c1/c1_Compilation.hpp
@@ -127,6 +127,7 @@ class Compilation: public StackObj {
   bool has_exception_handlers() const            { return _has_exception_handlers; }
   bool has_fpu_code() const                      { return _has_fpu_code; }
   bool has_unsafe_access() const                 { return _has_unsafe_access; }
+  int max_vector_size() const                    { return 0; }
   ciMethod* method() const                       { return _method; }
   int osr_bci() const                            { return _osr_bci; }
   bool is_osr_compile() const                    { return osr_bci() >= 0; }
diff --git a/src/share/vm/c1/c1_GraphBuilder.cpp b/src/share/vm/c1/c1_GraphBuilder.cpp
index 636fed2f2..95d99025b 100644
--- a/src/share/vm/c1/c1_GraphBuilder.cpp
+++ b/src/share/vm/c1/c1_GraphBuilder.cpp
@@ -3383,6 +3383,41 @@ bool GraphBuilder::try_inline_intrinsics(ciMethod* callee) {
       append_unsafe_CAS(callee);
       return true;
 
+    case vmIntrinsics::_getAndAddInt:
+      if (!VM_Version::supports_atomic_getadd4()) {
+        return false;
+      }
+      return append_unsafe_get_and_set_obj(callee, true);
+    case vmIntrinsics::_getAndAddLong:
+      if (!VM_Version::supports_atomic_getadd8()) {
+        return false;
+      }
+      return append_unsafe_get_and_set_obj(callee, true);
+    case vmIntrinsics::_getAndSetInt:
+      if (!VM_Version::supports_atomic_getset4()) {
+        return false;
+      }
+      return append_unsafe_get_and_set_obj(callee, false);
+    case vmIntrinsics::_getAndSetLong:
+      if (!VM_Version::supports_atomic_getset8()) {
+        return false;
+      }
+      return append_unsafe_get_and_set_obj(callee, false);
+    case vmIntrinsics::_getAndSetObject:
+#ifdef _LP64
+      if (!UseCompressedOops && !VM_Version::supports_atomic_getset8()) {
+        return false;
+      }
+      if (UseCompressedOops && !VM_Version::supports_atomic_getset4()) {
+        return false;
+      }
+#else
+      if (!VM_Version::supports_atomic_getset4()) {
+        return false;
+      }
+#endif
+      return append_unsafe_get_and_set_obj(callee, false);
+
     case vmIntrinsics::_Reference_get:
       // Use the intrinsic version of Reference.get() so that the value in
       // the referent field can be registered by the G1 pre-barrier code.
@@ -4106,6 +4141,22 @@ void GraphBuilder::print_inlining(ciMethod* callee, const char* msg, bool succes
   }
 }
 
+bool GraphBuilder::append_unsafe_get_and_set_obj(ciMethod* callee, bool is_add) {
+  if (InlineUnsafeOps) {
+    Values* args = state()->pop_arguments(callee->arg_size());
+    BasicType t = callee->return_type()->basic_type();
+    null_check(args->at(0));
+    Instruction* offset = args->at(2);
+#ifndef _LP64
+    offset = append(new Convert(Bytecodes::_l2i, offset, as_ValueType(T_INT)));
+#endif
+    Instruction* op = append(new UnsafeGetAndSetObject(t, args->at(1), offset, args->at(3), is_add));
+    compilation()->set_has_unsafe_access(true);
+    kill_all();
+    push(op->type(), op);
+  }
+  return InlineUnsafeOps;
+}
 
 #ifndef PRODUCT
 void GraphBuilder::print_stats() {
diff --git a/src/share/vm/c1/c1_GraphBuilder.hpp b/src/share/vm/c1/c1_GraphBuilder.hpp
index 8d7e7de34..09970583c 100644
--- a/src/share/vm/c1/c1_GraphBuilder.hpp
+++ b/src/share/vm/c1/c1_GraphBuilder.hpp
@@ -367,6 +367,7 @@ class GraphBuilder VALUE_OBJ_CLASS_SPEC {
   bool append_unsafe_put_raw(ciMethod* callee, BasicType t);
   bool append_unsafe_prefetch(ciMethod* callee, bool is_store, bool is_static);
   void append_unsafe_CAS(ciMethod* callee);
+  bool append_unsafe_get_and_set_obj(ciMethod* callee, bool is_add);
 
   void print_inlining(ciMethod* callee, const char* msg, bool success = true);
 
diff --git a/src/share/vm/c1/c1_Instruction.hpp b/src/share/vm/c1/c1_Instruction.hpp
index 930a599aa..4fff026e0 100644
--- a/src/share/vm/c1/c1_Instruction.hpp
+++ b/src/share/vm/c1/c1_Instruction.hpp
@@ -102,6 +102,7 @@ class       UnsafePutRaw;
 class     UnsafeObjectOp;
 class       UnsafeGetObject;
 class       UnsafePutObject;
+class         UnsafeGetAndSetObject;
 class       UnsafePrefetch;
 class         UnsafePrefetchRead;
 class         UnsafePrefetchWrite;
@@ -202,6 +203,7 @@ class InstructionVisitor: public StackObj {
   virtual void do_UnsafePutRaw   (UnsafePutRaw*    x) = 0;
   virtual void do_UnsafeGetObject(UnsafeGetObject* x) = 0;
   virtual void do_UnsafePutObject(UnsafePutObject* x) = 0;
+  virtual void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) = 0;
   virtual void do_UnsafePrefetchRead (UnsafePrefetchRead*  x) = 0;
   virtual void do_UnsafePrefetchWrite(UnsafePrefetchWrite* x) = 0;
   virtual void do_ProfileCall    (ProfileCall*     x) = 0;
@@ -2273,6 +2275,27 @@ LEAF(UnsafePutObject, UnsafeObjectOp)
                                                    f->visit(&_value); }
 };
 
+LEAF(UnsafeGetAndSetObject, UnsafeObjectOp)
+ private:
+  Value _value;                                  // Value to be stored
+  bool  _is_add;
+ public:
+  UnsafeGetAndSetObject(BasicType basic_type, Value object, Value offset, Value value, bool is_add)
+  : UnsafeObjectOp(basic_type, object, offset, false, false)
+    , _value(value)
+    , _is_add(is_add)
+  {
+    ASSERT_VALUES
+  }
+
+  // accessors
+  bool is_add() const                            { return _is_add; }
+  Value value()                                  { return _value; }
+
+  // generic
+  virtual void input_values_do(ValueVisitor* f)   { UnsafeObjectOp::input_values_do(f);
+                                                   f->visit(&_value); }
+};
 
 BASE(UnsafePrefetch, UnsafeObjectOp)
  public:
diff --git a/src/share/vm/c1/c1_InstructionPrinter.cpp b/src/share/vm/c1/c1_InstructionPrinter.cpp
index 383ffbb61..a2f6f86fd 100644
--- a/src/share/vm/c1/c1_InstructionPrinter.cpp
+++ b/src/share/vm/c1/c1_InstructionPrinter.cpp
@@ -831,6 +831,12 @@ void InstructionPrinter::do_UnsafePutObject(UnsafePutObject* x) {
   output()->put(')');
 }
 
+void InstructionPrinter::do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) {
+  print_unsafe_object_op(x, x->is_add()?"UnsafeGetAndSetObject (add)":"UnsafeGetAndSetObject");
+  output()->print(", value ");
+  print_value(x->value());
+  output()->put(')');
+}
 
 void InstructionPrinter::do_UnsafePrefetchRead(UnsafePrefetchRead* x) {
   print_unsafe_object_op(x, "UnsafePrefetchRead");
diff --git a/src/share/vm/c1/c1_InstructionPrinter.hpp b/src/share/vm/c1/c1_InstructionPrinter.hpp
index 3d09ef02b..a89908c83 100644
--- a/src/share/vm/c1/c1_InstructionPrinter.hpp
+++ b/src/share/vm/c1/c1_InstructionPrinter.hpp
@@ -128,6 +128,7 @@ class InstructionPrinter: public InstructionVisitor {
   virtual void do_UnsafePutRaw   (UnsafePutRaw*    x);
   virtual void do_UnsafeGetObject(UnsafeGetObject* x);
   virtual void do_UnsafePutObject(UnsafePutObject* x);
+  virtual void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x);
   virtual void do_UnsafePrefetchRead (UnsafePrefetchRead*  x);
   virtual void do_UnsafePrefetchWrite(UnsafePrefetchWrite* x);
   virtual void do_ProfileCall    (ProfileCall*     x);
diff --git a/src/share/vm/c1/c1_LIR.cpp b/src/share/vm/c1/c1_LIR.cpp
index 9bc7ab8ed..178d18671 100644
--- a/src/share/vm/c1/c1_LIR.cpp
+++ b/src/share/vm/c1/c1_LIR.cpp
@@ -264,6 +264,7 @@ void LIR_Op2::verify() const {
 #ifdef ASSERT
   switch (code()) {
     case lir_cmove:
+    case lir_xchg:
       break;
 
     default:
@@ -630,6 +631,8 @@ void LIR_OpVisitState::visit(LIR_Op* op) {
     case lir_shl:
     case lir_shr:
     case lir_ushr:
+    case lir_xadd:
+    case lir_xchg:
     {
       assert(op->as_Op2() != NULL, "must be");
       LIR_Op2* op2 = (LIR_Op2*)op;
@@ -641,6 +644,13 @@ void LIR_OpVisitState::visit(LIR_Op* op) {
       if (op2->_opr2->is_valid())         do_input(op2->_opr2);
       if (op2->_tmp1->is_valid())         do_temp(op2->_tmp1);
       if (op2->_result->is_valid())       do_output(op2->_result);
+      if (op->code() == lir_xchg || op->code() == lir_xadd) {
+        // on ARM and PPC, return value is loaded first so could
+        // destroy inputs. On other platforms that implement those
+        // (x86, sparc), the extra constrainsts are harmless.
+        if (op2->_opr1->is_valid())       do_temp(op2->_opr1);
+        if (op2->_opr2->is_valid())       do_temp(op2->_opr2);
+      }
 
       break;
     }
@@ -1733,6 +1743,8 @@ const char * LIR_Op::name() const {
      case lir_shr:                   s = "shift_right";   break;
      case lir_ushr:                  s = "ushift_right";  break;
      case lir_alloc_array:           s = "alloc_array";   break;
+     case lir_xadd:                  s = "xadd";          break;
+     case lir_xchg:                  s = "xchg";          break;
      // LIR_Op3
      case lir_idiv:                  s = "idiv";          break;
      case lir_irem:                  s = "irem";          break;
diff --git a/src/share/vm/c1/c1_LIR.hpp b/src/share/vm/c1/c1_LIR.hpp
index a79a8ace4..122ebe8ce 100644
--- a/src/share/vm/c1/c1_LIR.hpp
+++ b/src/share/vm/c1/c1_LIR.hpp
@@ -963,6 +963,8 @@ enum LIR_Code {
       , lir_alloc_array
       , lir_throw
       , lir_compare_to
+      , lir_xadd
+      , lir_xchg
   , end_op2
   , begin_op3
       , lir_idiv
@@ -2191,6 +2193,9 @@ class LIR_List: public CompilationResourceObj {
   void profile_call(ciMethod* method, int bci, ciMethod* callee, LIR_Opr mdo, LIR_Opr recv, LIR_Opr t1, ciKlass* cha_klass) {
     append(new LIR_OpProfileCall(lir_profile_call, method, bci, callee, mdo, recv, t1, cha_klass));
   }
+
+  void xadd(LIR_Opr src, LIR_Opr add, LIR_Opr res, LIR_Opr tmp) { append(new LIR_Op2(lir_xadd, src, add, res, tmp)); }
+  void xchg(LIR_Opr src, LIR_Opr set, LIR_Opr res, LIR_Opr tmp) { append(new LIR_Op2(lir_xchg, src, set, res, tmp)); }
 };
 
 void print_LIR(BlockList* blocks);
@@ -2287,16 +2292,21 @@ class LIR_OpVisitState: public StackObj {
       LIR_Address* address = opr->as_address_ptr();
       if (address != NULL) {
         // special handling for addresses: add base and index register of the address
-        // both are always input operands!
+        // both are always input operands or temp if we want to extend
+        // their liveness!
+        if (mode == outputMode) {
+          mode = inputMode;
+        }
+        assert (mode == inputMode || mode == tempMode, "input or temp only for addresses");
         if (address->_base->is_valid()) {
           assert(address->_base->is_register(), "must be");
-          assert(_oprs_len[inputMode] < maxNumberOfOperands, "array overflow");
-          _oprs_new[inputMode][_oprs_len[inputMode]++] = &address->_base;
+          assert(_oprs_len[mode] < maxNumberOfOperands, "array overflow");
+          _oprs_new[mode][_oprs_len[mode]++] = &address->_base;
         }
         if (address->_index->is_valid()) {
           assert(address->_index->is_register(), "must be");
-          assert(_oprs_len[inputMode] < maxNumberOfOperands, "array overflow");
-          _oprs_new[inputMode][_oprs_len[inputMode]++] = &address->_index;
+          assert(_oprs_len[mode] < maxNumberOfOperands, "array overflow");
+          _oprs_new[mode][_oprs_len[mode]++] = &address->_index;
         }
 
       } else {
diff --git a/src/share/vm/c1/c1_LIRAssembler.cpp b/src/share/vm/c1/c1_LIRAssembler.cpp
index e67987d78..a76f5bb2e 100644
--- a/src/share/vm/c1/c1_LIRAssembler.cpp
+++ b/src/share/vm/c1/c1_LIRAssembler.cpp
@@ -773,6 +773,11 @@ void LIR_Assembler::emit_op2(LIR_Op2* op) {
       throw_op(op->in_opr1(), op->in_opr2(), op->info());
       break;
 
+    case lir_xadd:
+    case lir_xchg:
+      atomic_op(op->code(), op->in_opr1(), op->in_opr2(), op->result_opr(), op->tmp1_opr());
+      break;
+
     default:
       Unimplemented();
       break;
diff --git a/src/share/vm/c1/c1_LIRAssembler.hpp b/src/share/vm/c1/c1_LIRAssembler.hpp
index 226e4f83f..5cce9d0b7 100644
--- a/src/share/vm/c1/c1_LIRAssembler.hpp
+++ b/src/share/vm/c1/c1_LIRAssembler.hpp
@@ -252,6 +252,8 @@ class LIR_Assembler: public CompilationResourceObj {
 
   void verify_oop_map(CodeEmitInfo* info);
 
+  void atomic_op(LIR_Code code, LIR_Opr src, LIR_Opr data, LIR_Opr dest, LIR_Opr tmp);
+
 #ifdef TARGET_ARCH_x86
 # include "c1_LIRAssembler_x86.hpp"
 #endif
diff --git a/src/share/vm/c1/c1_LIRGenerator.hpp b/src/share/vm/c1/c1_LIRGenerator.hpp
index ac20e7a33..aedd6a69c 100644
--- a/src/share/vm/c1/c1_LIRGenerator.hpp
+++ b/src/share/vm/c1/c1_LIRGenerator.hpp
@@ -527,6 +527,7 @@ class LIRGenerator: public InstructionVisitor, public BlockClosure {
   virtual void do_UnsafePutRaw   (UnsafePutRaw*    x);
   virtual void do_UnsafeGetObject(UnsafeGetObject* x);
   virtual void do_UnsafePutObject(UnsafePutObject* x);
+  virtual void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x);
   virtual void do_UnsafePrefetchRead (UnsafePrefetchRead*  x);
   virtual void do_UnsafePrefetchWrite(UnsafePrefetchWrite* x);
   virtual void do_ProfileCall    (ProfileCall*     x);
diff --git a/src/share/vm/c1/c1_Optimizer.cpp b/src/share/vm/c1/c1_Optimizer.cpp
index 487a7a7ef..3ba17653d 100644
--- a/src/share/vm/c1/c1_Optimizer.cpp
+++ b/src/share/vm/c1/c1_Optimizer.cpp
@@ -505,6 +505,7 @@ public:
   void do_UnsafePutRaw   (UnsafePutRaw*    x);
   void do_UnsafeGetObject(UnsafeGetObject* x);
   void do_UnsafePutObject(UnsafePutObject* x);
+  void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x);
   void do_UnsafePrefetchRead (UnsafePrefetchRead*  x);
   void do_UnsafePrefetchWrite(UnsafePrefetchWrite* x);
   void do_ProfileCall    (ProfileCall*     x);
@@ -676,6 +677,7 @@ void NullCheckVisitor::do_UnsafeGetRaw   (UnsafeGetRaw*    x) {}
 void NullCheckVisitor::do_UnsafePutRaw   (UnsafePutRaw*    x) {}
 void NullCheckVisitor::do_UnsafeGetObject(UnsafeGetObject* x) {}
 void NullCheckVisitor::do_UnsafePutObject(UnsafePutObject* x) {}
+void NullCheckVisitor::do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) {}
 void NullCheckVisitor::do_UnsafePrefetchRead (UnsafePrefetchRead*  x) {}
 void NullCheckVisitor::do_UnsafePrefetchWrite(UnsafePrefetchWrite* x) {}
 void NullCheckVisitor::do_ProfileCall    (ProfileCall*     x) { nce()->clear_last_explicit_null_check(); }
diff --git a/src/share/vm/c1/c1_ValueMap.hpp b/src/share/vm/c1/c1_ValueMap.hpp
index 6f89738d4..95fa73281 100644
--- a/src/share/vm/c1/c1_ValueMap.hpp
+++ b/src/share/vm/c1/c1_ValueMap.hpp
@@ -157,6 +157,7 @@ class ValueNumberingVisitor: public InstructionVisitor {
   void do_Invoke         (Invoke*          x) { kill_memory(); }
   void do_UnsafePutRaw   (UnsafePutRaw*    x) { kill_memory(); }
   void do_UnsafePutObject(UnsafePutObject* x) { kill_memory(); }
+  void do_UnsafeGetAndSetObject(UnsafeGetAndSetObject* x) { kill_memory(); }
   void do_Intrinsic      (Intrinsic*       x) { if (!x->preserves_state()) kill_memory(); }
 
   void do_Phi            (Phi*             x) { /* nothing to do */ }
diff --git a/src/share/vm/ci/ciEnv.cpp b/src/share/vm/ci/ciEnv.cpp
index ef4826b4f..a895e97d0 100644
--- a/src/share/vm/ci/ciEnv.cpp
+++ b/src/share/vm/ci/ciEnv.cpp
@@ -921,7 +921,8 @@ void ciEnv::register_method(ciMethod* target,
                             ImplicitExceptionTable* inc_table,
                             AbstractCompiler* compiler,
                             int comp_level,
-                            bool has_unsafe_access) {
+                            bool has_unsafe_access,
+                            bool has_wide_vectors) {
   VM_ENTRY_MARK;
   nmethod* nm = NULL;
   {
@@ -1016,6 +1017,7 @@ void ciEnv::register_method(ciMethod* target,
       }
     } else {
       nm->set_has_unsafe_access(has_unsafe_access);
+      nm->set_has_wide_vectors(has_wide_vectors);
 
       // Record successful registration.
       // (Put nm into the task handle *before* publishing to the Java heap.)
diff --git a/src/share/vm/ci/ciEnv.hpp b/src/share/vm/ci/ciEnv.hpp
index 63b316e61..f252dc7c0 100644
--- a/src/share/vm/ci/ciEnv.hpp
+++ b/src/share/vm/ci/ciEnv.hpp
@@ -362,7 +362,8 @@ public:
                        ImplicitExceptionTable*   inc_table,
                        AbstractCompiler*         compiler,
                        int                       comp_level,
-                       bool                      has_unsafe_access);
+                       bool                      has_unsafe_access,
+                       bool                      has_wide_vectors);
 
 
   // Access to certain well known ciObjects.
diff --git a/src/share/vm/classfile/vmSymbols.hpp b/src/share/vm/classfile/vmSymbols.hpp
index 417c25528..10cf48e7b 100644
--- a/src/share/vm/classfile/vmSymbols.hpp
+++ b/src/share/vm/classfile/vmSymbols.hpp
@@ -873,6 +873,20 @@
    do_name(     putOrderedInt_name,                              "putOrderedInt")                                       \
    do_alias(    putOrderedInt_signature,                        /*(Ljava/lang/Object;JI)V*/ putInt_signature)           \
                                                                                                                         \
+  do_intrinsic(_getAndAddInt,             sun_misc_Unsafe,        getAndAddInt_name, getAndAddInt_signature, F_R)       \
+   do_name(     getAndAddInt_name,                                "getAndAddInt")                                       \
+   do_signature(getAndAddInt_signature,                           "(Ljava/lang/Object;JI)I" )                           \
+  do_intrinsic(_getAndAddLong,            sun_misc_Unsafe,        getAndAddLong_name, getAndAddLong_signature, F_R)     \
+   do_name(     getAndAddLong_name,                               "getAndAddLong")                                      \
+   do_signature(getAndAddLong_signature,                          "(Ljava/lang/Object;JJ)J" )                           \
+  do_intrinsic(_getAndSetInt,             sun_misc_Unsafe,        getAndSet_name, getAndSetInt_signature, F_R)          \
+   do_name(     getAndSet_name,                                   "getAndSet")                                          \
+   do_alias(    getAndSetInt_signature,                         /*"(Ljava/lang/Object;JI)I"*/ getAndAddInt_signature)   \
+  do_intrinsic(_getAndSetLong,            sun_misc_Unsafe,        getAndSet_name, getAndSetLong_signature, F_R)         \
+   do_alias(    getAndSetLong_signature,                        /*"(Ljava/lang/Object;JJ)J"*/ getAndAddLong_signature)  \
+  do_intrinsic(_getAndSetObject,          sun_misc_Unsafe,        getAndSet_name, getAndSetObject_signature,  F_R)      \
+   do_signature(getAndSetObject_signature,                        "(Ljava/lang/Object;JLjava/lang/Object;)Ljava/lang/Object;" ) \
+                                                                                                                        \
   /* prefetch_signature is shared by all prefetch variants */                                                           \
   do_signature( prefetch_signature,        "(Ljava/lang/Object;J)V")                                                    \
                                                                                                                         \
diff --git a/src/share/vm/code/codeBlob.cpp b/src/share/vm/code/codeBlob.cpp
index 425ad7f31..779235f25 100644
--- a/src/share/vm/code/codeBlob.cpp
+++ b/src/share/vm/code/codeBlob.cpp
@@ -162,8 +162,10 @@ void CodeBlob::trace_new_stub(CodeBlob* stub, const char* name1, const char* nam
     assert(strlen(name1) + strlen(name2) < sizeof(stub_id), "");
     jio_snprintf(stub_id, sizeof(stub_id), "%s%s", name1, name2);
     if (PrintStubCode) {
+      ttyLocker ttyl;
       tty->print_cr("Decoding %s " INTPTR_FORMAT, stub_id, (intptr_t) stub);
       Disassembler::decode(stub->code_begin(), stub->code_end());
+      tty->cr();
     }
     Forte::register_stub(stub_id, stub->code_begin(), stub->code_end());
 
@@ -548,6 +550,7 @@ void RuntimeStub::verify() {
 }
 
 void RuntimeStub::print_on(outputStream* st) const {
+  ttyLocker ttyl;
   CodeBlob::print_on(st);
   st->print("Runtime Stub (" INTPTR_FORMAT "): ", this);
   st->print_cr(name());
@@ -563,6 +566,7 @@ void SingletonBlob::verify() {
 }
 
 void SingletonBlob::print_on(outputStream* st) const {
+  ttyLocker ttyl;
   CodeBlob::print_on(st);
   st->print_cr(name());
   Disassembler::decode((CodeBlob*)this, st);
diff --git a/src/share/vm/code/codeBlob.hpp b/src/share/vm/code/codeBlob.hpp
index a8b73ae76..1d47bd972 100644
--- a/src/share/vm/code/codeBlob.hpp
+++ b/src/share/vm/code/codeBlob.hpp
@@ -184,7 +184,7 @@ class CodeBlob VALUE_OBJ_CLASS_SPEC {
   static void trace_new_stub(CodeBlob* blob, const char* name1, const char* name2 = "");
 
   // Print the comment associated with offset on stream, if there is one
-  virtual void print_block_comment(outputStream* stream, address block_begin) {
+  virtual void print_block_comment(outputStream* stream, address block_begin) const {
     intptr_t offset = (intptr_t)(block_begin - code_begin());
     _comments.print_block_comment(stream, offset);
   }
diff --git a/src/share/vm/code/icBuffer.hpp b/src/share/vm/code/icBuffer.hpp
index 340dbb893..36d0c7616 100644
--- a/src/share/vm/code/icBuffer.hpp
+++ b/src/share/vm/code/icBuffer.hpp
@@ -25,6 +25,7 @@
 #ifndef SHARE_VM_CODE_ICBUFFER_HPP
 #define SHARE_VM_CODE_ICBUFFER_HPP
 
+#include "asm/codeBuffer.hpp"
 #include "code/stubs.hpp"
 #include "interpreter/bytecodes.hpp"
 #include "memory/allocation.hpp"
@@ -48,7 +49,8 @@ class ICStub: public Stub {
  protected:
   friend class ICStubInterface;
   // This will be called only by ICStubInterface
-  void    initialize(int size) { _size = size; _ic_site = NULL; }
+  void    initialize(int size,
+                     CodeComments comments)      { _size = size; _ic_site = NULL; }
   void    finalize(); // called when a method is removed
 
   // General info
diff --git a/src/share/vm/code/nmethod.cpp b/src/share/vm/code/nmethod.cpp
index 8d206ea6a..d935845fd 100644
--- a/src/share/vm/code/nmethod.cpp
+++ b/src/share/vm/code/nmethod.cpp
@@ -463,6 +463,7 @@ void nmethod::init_defaults() {
   _has_unsafe_access          = 0;
   _has_method_handle_invokes  = 0;
   _lazy_critical_native       = 0;
+  _has_wide_vectors           = 0;
   _marked_for_deoptimization  = 0;
   _lock_count                 = 0;
   _stack_traversal_mark       = 0;
@@ -700,7 +701,9 @@ nmethod::nmethod(
     // then print the requested information
     if (PrintNativeNMethods) {
       print_code();
-      oop_maps->print();
+      if (oop_maps != NULL) {
+        oop_maps->print();
+      }
     }
     if (PrintRelocations) {
       print_relocations();
@@ -2666,7 +2669,7 @@ ScopeDesc* nmethod::scope_desc_in(address begin, address end) {
   return NULL;
 }
 
-void nmethod::print_nmethod_labels(outputStream* stream, address block_begin) {
+void nmethod::print_nmethod_labels(outputStream* stream, address block_begin) const {
   if (block_begin == entry_point())             stream->print_cr("[Entry Point]");
   if (block_begin == verified_entry_point())    stream->print_cr("[Verified Entry Point]");
   if (block_begin == exception_begin())         stream->print_cr("[Exception Handler]");
diff --git a/src/share/vm/code/nmethod.hpp b/src/share/vm/code/nmethod.hpp
index fbd0b2cdb..477ceb6dc 100644
--- a/src/share/vm/code/nmethod.hpp
+++ b/src/share/vm/code/nmethod.hpp
@@ -177,6 +177,7 @@ class nmethod : public CodeBlob {
   unsigned int _has_unsafe_access:1;         // May fault due to unsafe access.
   unsigned int _has_method_handle_invokes:1; // Has this method MethodHandle invokes?
   unsigned int _lazy_critical_native:1;      // Lazy JNI critical native
+  unsigned int _has_wide_vectors:1;          // Preserve wide vectors at safepoints
 
   // Protected by Patching_lock
   unsigned char _state;                      // {alive, not_entrant, zombie, unloaded}
@@ -442,6 +443,9 @@ class nmethod : public CodeBlob {
   bool  is_lazy_critical_native() const           { return _lazy_critical_native; }
   void  set_lazy_critical_native(bool z)          { _lazy_critical_native = z; }
 
+  bool  has_wide_vectors() const                  { return _has_wide_vectors; }
+  void  set_has_wide_vectors(bool z)              { _has_wide_vectors = z; }
+
   int   comp_level() const                        { return _comp_level; }
 
   // Support for oops in scopes and relocs:
@@ -647,11 +651,11 @@ public:
   void log_state_change() const;
 
   // Prints block-level comments, including nmethod specific block labels:
-  virtual void print_block_comment(outputStream* stream, address block_begin) {
+  virtual void print_block_comment(outputStream* stream, address block_begin) const {
     print_nmethod_labels(stream, block_begin);
     CodeBlob::print_block_comment(stream, block_begin);
   }
-  void print_nmethod_labels(outputStream* stream, address block_begin);
+  void print_nmethod_labels(outputStream* stream, address block_begin) const;
 
   // Prints a comment for one native instruction (reloc info, pc desc)
   void print_code_comment_on(outputStream* st, int column, address begin, address end);
diff --git a/src/share/vm/code/stubs.cpp b/src/share/vm/code/stubs.cpp
index 124c0a640..0d5e61c8e 100644
--- a/src/share/vm/code/stubs.cpp
+++ b/src/share/vm/code/stubs.cpp
@@ -101,7 +101,8 @@ Stub* StubQueue::stub_containing(address pc) const {
 
 Stub* StubQueue::request_committed(int code_size) {
   Stub* s = request(code_size);
-  if (s != NULL) commit(code_size);
+  CodeComments comments;
+  if (s != NULL) commit(code_size, comments);
   return s;
 }
 
@@ -118,7 +119,8 @@ Stub* StubQueue::request(int requested_code_size) {
       assert(_buffer_limit == _buffer_size, "buffer must be fully usable");
       if (_queue_end + requested_size <= _buffer_size) {
         // code fits in at the end => nothing to do
-        stub_initialize(s, requested_size);
+        CodeComments comments;
+        stub_initialize(s, requested_size, comments);
         return s;
       } else {
         // stub doesn't fit in at the queue end
@@ -135,7 +137,8 @@ Stub* StubQueue::request(int requested_code_size) {
     // Queue: |XXX|.......|XXXXXXX|.......|
     //        ^0  ^end    ^begin  ^limit  ^size
     s = current_stub();
-    stub_initialize(s, requested_size);
+    CodeComments comments;
+    stub_initialize(s, requested_size, comments);
     return s;
   }
   // Not enough space left
@@ -144,12 +147,12 @@ Stub* StubQueue::request(int requested_code_size) {
 }
 
 
-void StubQueue::commit(int committed_code_size) {
+void StubQueue::commit(int committed_code_size, CodeComments& comments) {
   assert(committed_code_size > 0, "committed_code_size must be > 0");
   int committed_size = round_to(stub_code_size_to_size(committed_code_size), CodeEntryAlignment);
   Stub* s = current_stub();
   assert(committed_size <= stub_size(s), "committed size must not exceed requested size");
-  stub_initialize(s, committed_size);
+  stub_initialize(s, committed_size, comments);
   _queue_end += committed_size;
   _number_of_stubs++;
   if (_mutex != NULL) _mutex->unlock();
diff --git a/src/share/vm/code/stubs.hpp b/src/share/vm/code/stubs.hpp
index afc8f04d4..5701a45a5 100644
--- a/src/share/vm/code/stubs.hpp
+++ b/src/share/vm/code/stubs.hpp
@@ -25,6 +25,7 @@
 #ifndef SHARE_VM_CODE_STUBS_HPP
 #define SHARE_VM_CODE_STUBS_HPP
 
+#include "asm/codeBuffer.hpp"
 #include "memory/allocation.hpp"
 #ifdef TARGET_OS_FAMILY_linux
 # include "os_linux.inline.hpp"
@@ -71,7 +72,8 @@
 class Stub VALUE_OBJ_CLASS_SPEC {
  public:
   // Initialization/finalization
-  void    initialize(int size)                   { ShouldNotCallThis(); }                // called to initialize/specify the stub's size
+  void    initialize(int size,
+                     CodeComments& comments)     { ShouldNotCallThis(); }                // called to initialize/specify the stub's size
   void    finalize()                             { ShouldNotCallThis(); }                // called before the stub is deallocated
 
   // General info/converters
@@ -104,7 +106,8 @@ class Stub VALUE_OBJ_CLASS_SPEC {
 class StubInterface: public CHeapObj<mtCode> {
  public:
   // Initialization/finalization
-  virtual void    initialize(Stub* self, int size)         = 0; // called after creation (called twice if allocated via (request, commit))
+  virtual void    initialize(Stub* self, int size,
+                             CodeComments& comments)       = 0; // called after creation (called twice if allocated via (request, commit))
   virtual void    finalize(Stub* self)                     = 0; // called before deallocation
 
   // General info/converters
@@ -132,7 +135,8 @@ class StubInterface: public CHeapObj<mtCode> {
                                                            \
    public:                                                 \
     /* Initialization/finalization */                      \
-    virtual void    initialize(Stub* self, int size)       { cast(self)->initialize(size); }       \
+    virtual void    initialize(Stub* self, int size,       \
+                               CodeComments& comments)     { cast(self)->initialize(size, comments); } \
     virtual void    finalize(Stub* self)                   { cast(self)->finalize(); }             \
                                                            \
     /* General info */                                     \
@@ -171,7 +175,8 @@ class StubQueue: public CHeapObj<mtCode> {
   Stub* current_stub() const                     { return stub_at(_queue_end); }
 
   // Stub functionality accessed via interface
-  void  stub_initialize(Stub* s, int size)       { assert(size % CodeEntryAlignment == 0, "size not aligned"); _stub_interface->initialize(s, size); }
+  void  stub_initialize(Stub* s, int size,
+                        CodeComments& comments)  { assert(size % CodeEntryAlignment == 0, "size not aligned"); _stub_interface->initialize(s, size, comments); }
   void  stub_finalize(Stub* s)                   { _stub_interface->finalize(s); }
   int   stub_size(Stub* s) const                 { return _stub_interface->size(s); }
   bool  stub_contains(Stub* s, address pc) const { return _stub_interface->code_begin(s) <= pc && pc < _stub_interface->code_end(s); }
@@ -200,7 +205,8 @@ class StubQueue: public CHeapObj<mtCode> {
   // Stub allocation (atomic transactions)
   Stub* request_committed(int code_size);        // request a stub that provides exactly code_size space for code
   Stub* request(int requested_code_size);        // request a stub with a (maximum) code space - locks the queue
-  void  commit (int committed_code_size);        // commit the previously requested stub - unlocks the queue
+  void  commit (int committed_code_size,
+                CodeComments& comments);         // commit the previously requested stub - unlocks the queue
 
   // Stub deallocation
   void  remove_first();                          // remove the first stub in the queue
diff --git a/src/share/vm/compiler/compileBroker.cpp b/src/share/vm/compiler/compileBroker.cpp
index 6fe00d8ae..02086da4a 100644
--- a/src/share/vm/compiler/compileBroker.cpp
+++ b/src/share/vm/compiler/compileBroker.cpp
@@ -197,9 +197,9 @@ class CompilationLog : public StringEventLog {
 
   void log_compile(JavaThread* thread, CompileTask* task) {
     StringLogMessage lm;
-    stringStream msg = lm.stream();
+    stringStream sstr = lm.stream();
     // msg.time_stamp().update_to(tty->time_stamp().ticks());
-    task->print_compilation(&msg, true);
+    task->print_compilation(&sstr, NULL, true);
     log(thread, "%s", (const char*)lm);
   }
 
@@ -491,9 +491,9 @@ void CompileTask::print_inline_indent(int inline_level, outputStream* st) {
 
 // ------------------------------------------------------------------
 // CompileTask::print_compilation
-void CompileTask::print_compilation(outputStream* st, bool short_form) {
+void CompileTask::print_compilation(outputStream* st, const char* msg, bool short_form) {
   bool is_osr_method = osr_bci() != InvocationEntryBci;
-  print_compilation_impl(st, method(), compile_id(), comp_level(), is_osr_method, osr_bci(), is_blocking(), NULL, short_form);
+  print_compilation_impl(st, method(), compile_id(), comp_level(), is_osr_method, osr_bci(), is_blocking(), msg, short_form);
 }
 
 // ------------------------------------------------------------------
@@ -1249,7 +1249,7 @@ nmethod* CompileBroker::compile_method(methodHandle method, int osr_bci,
     // We accept a higher level osr method
     nmethod* nm = method->lookup_osr_nmethod_for(osr_bci, comp_level, false);
     if (nm != NULL) return nm;
-    if (method->is_not_osr_compilable()) return NULL;
+    if (method->is_not_osr_compilable(comp_level)) return NULL;
   }
 
   assert(!HAS_PENDING_EXCEPTION, "No exception should be present");
@@ -1330,7 +1330,7 @@ bool CompileBroker::compilation_is_complete(methodHandle method,
                                             int          comp_level) {
   bool is_osr = (osr_bci != standard_entry_bci);
   if (is_osr) {
-    if (method->is_not_osr_compilable()) {
+    if (method->is_not_osr_compilable(comp_level)) {
       return true;
     } else {
       nmethod* result = method->lookup_osr_nmethod_for(osr_bci, comp_level, true);
@@ -1381,7 +1381,7 @@ bool CompileBroker::compilation_is_prohibited(methodHandle method, int osr_bci,
   // Some compilers may not support on stack replacement.
   if (is_osr &&
       (!CICompileOSR || !compiler(comp_level)->supports_osr())) {
-    method->set_not_osr_compilable();
+    method->set_not_osr_compilable(comp_level);
     return true;
   }
 
@@ -1807,11 +1807,10 @@ void CompileBroker::invoke_compiler_on_method(CompileTask* task) {
         _compilation_log->log_failure(thread, task, ci_env.failure_reason(), retry_message);
       }
       if (PrintCompilation) {
-        tty->print("%4d   COMPILE SKIPPED: %s", compile_id, ci_env.failure_reason());
-        if (retry_message != NULL) {
-          tty->print(" (%s)", retry_message);
-        }
-        tty->cr();
+        FormatBufferResource msg = retry_message != NULL ?
+            err_msg_res("COMPILE SKIPPED: %s (%s)", ci_env.failure_reason(), retry_message) :
+            err_msg_res("COMPILE SKIPPED: %s",      ci_env.failure_reason());
+        task->print_compilation(tty, msg);
       }
     } else {
       task->mark_success();
@@ -1840,14 +1839,20 @@ void CompileBroker::invoke_compiler_on_method(CompileTask* task) {
     tty->print_cr("size: %d time: %d inlined: %d bytes", code_size, (int)time.milliseconds(), task->num_inlined_bytecodes());
   }
 
-  if (compilable == ciEnv::MethodCompilable_never) {
-    if (is_osr) {
-      method->set_not_osr_compilable();
-    } else {
+  // Disable compilation, if required.
+  switch (compilable) {
+  case ciEnv::MethodCompilable_never:
+    if (is_osr)
+      method->set_not_osr_compilable_quietly();
+    else
       method->set_not_compilable_quietly();
-    }
-  } else if (compilable == ciEnv::MethodCompilable_not_at_tier) {
-    method->set_not_compilable_quietly(task->comp_level());
+    break;
+  case ciEnv::MethodCompilable_not_at_tier:
+    if (is_osr)
+      method->set_not_osr_compilable_quietly(task->comp_level());
+    else
+      method->set_not_compilable_quietly(task->comp_level());
+    break;
   }
 
   // Note that the queued_for_compilation bits are cleared without
diff --git a/src/share/vm/compiler/compileBroker.hpp b/src/share/vm/compiler/compileBroker.hpp
index 485726e41..5169cbe35 100644
--- a/src/share/vm/compiler/compileBroker.hpp
+++ b/src/share/vm/compiler/compileBroker.hpp
@@ -105,7 +105,7 @@ private:
                                       const char* msg = NULL, bool short_form = false);
 
 public:
-  void         print_compilation(outputStream* st = tty, bool short_form = false);
+  void         print_compilation(outputStream* st = tty, const char* msg = NULL, bool short_form = false);
   static void  print_compilation(outputStream* st, const nmethod* nm, const char* msg = NULL, bool short_form = false) {
     print_compilation_impl(st, nm->method(), nm->compile_id(), nm->comp_level(),
                            nm->is_osr_method(), nm->is_osr_method() ? nm->osr_entry_bci() : -1, /*is_blocking*/ false,
diff --git a/src/share/vm/compiler/compileLog.cpp b/src/share/vm/compiler/compileLog.cpp
index 9c63c1840..9d9c76060 100644
--- a/src/share/vm/compiler/compileLog.cpp
+++ b/src/share/vm/compiler/compileLog.cpp
@@ -125,46 +125,46 @@ int CompileLog::identify(ciBaseObject* obj) {
     ciMetadata* mobj = obj->as_metadata();
     if (mobj->is_klass()) {
       ciKlass* klass = mobj->as_klass();
-    begin_elem("klass id='%d'", id);
-    name(klass->name());
-    if (!klass->is_loaded()) {
-      print(" unloaded='1'");
-    } else {
-      print(" flags='%d'", klass->modifier_flags());
-    }
-    end_elem();
+      begin_elem("klass id='%d'", id);
+      name(klass->name());
+      if (!klass->is_loaded()) {
+        print(" unloaded='1'");
+      } else {
+        print(" flags='%d'", klass->modifier_flags());
+      }
+      end_elem();
     } else if (mobj->is_method()) {
       ciMethod* method = mobj->as_method();
-    ciSignature* sig = method->signature();
-    // Pre-identify items that we will need!
-    identify(sig->return_type());
-    for (int i = 0; i < sig->count(); i++) {
-      identify(sig->type_at(i));
-    }
-    begin_elem("method id='%d' holder='%d'",
-               id, identify(method->holder()));
-    name(method->name());
-    print(" return='%d'", identify(sig->return_type()));
-    if (sig->count() > 0) {
-      print(" arguments='");
+      ciSignature* sig = method->signature();
+      // Pre-identify items that we will need!
+      identify(sig->return_type());
       for (int i = 0; i < sig->count(); i++) {
-        print((i == 0) ? "%d" : " %d", identify(sig->type_at(i)));
+        identify(sig->type_at(i));
       }
-      print("'");
-    }
-    if (!method->is_loaded()) {
-      print(" unloaded='1'");
-    } else {
-      print(" flags='%d'", (jchar) method->flags().as_int());
-      // output a few metrics
-      print(" bytes='%d'", method->code_size());
-      method->log_nmethod_identity(this);
-      //print(" count='%d'", method->invocation_count());
-      //int bec = method->backedge_count();
-      //if (bec != 0)  print(" backedge_count='%d'", bec);
-      print(" iicount='%d'", method->interpreter_invocation_count());
-    }
-    end_elem();
+      begin_elem("method id='%d' holder='%d'",
+          id, identify(method->holder()));
+      name(method->name());
+      print(" return='%d'", identify(sig->return_type()));
+      if (sig->count() > 0) {
+        print(" arguments='");
+        for (int i = 0; i < sig->count(); i++) {
+          print((i == 0) ? "%d" : " %d", identify(sig->type_at(i)));
+        }
+        print("'");
+      }
+      if (!method->is_loaded()) {
+        print(" unloaded='1'");
+      } else {
+        print(" flags='%d'", (jchar) method->flags().as_int());
+        // output a few metrics
+        print(" bytes='%d'", method->code_size());
+        method->log_nmethod_identity(this);
+        //print(" count='%d'", method->invocation_count());
+        //int bec = method->backedge_count();
+        //if (bec != 0)  print(" backedge_count='%d'", bec);
+        print(" iicount='%d'", method->interpreter_invocation_count());
+      }
+      end_elem();
     } else if (mobj->is_type()) {
       BasicType type = mobj->as_type()->basic_type();
       elem("type id='%d' name='%s'", id, type2name(type));
diff --git a/src/share/vm/compiler/disassembler.cpp b/src/share/vm/compiler/disassembler.cpp
index e089ab7b8..e1fed2e4d 100644
--- a/src/share/vm/compiler/disassembler.cpp
+++ b/src/share/vm/compiler/disassembler.cpp
@@ -148,6 +148,7 @@ class decode_env {
  private:
   nmethod*      _nm;
   CodeBlob*     _code;
+  CodeComments  _comments;
   outputStream* _output;
   address       _start, _end;
 
@@ -187,7 +188,7 @@ class decode_env {
   void print_address(address value);
 
  public:
-  decode_env(CodeBlob* code, outputStream* output);
+  decode_env(CodeBlob* code, outputStream* output, CodeComments c = CodeComments());
 
   address decode_instructions(address start, address end);
 
@@ -231,12 +232,13 @@ class decode_env {
   const char* options() { return _option_buf; }
 };
 
-decode_env::decode_env(CodeBlob* code, outputStream* output) {
+decode_env::decode_env(CodeBlob* code, outputStream* output, CodeComments c) {
   memset(this, 0, sizeof(*this));
   _output = output ? output : tty;
   _code = code;
   if (code != NULL && code->is_nmethod())
     _nm = (nmethod*) code;
+  _comments.assign(c);
 
   // by default, output pc but not bytes:
   _print_pc       = true;
@@ -358,6 +360,7 @@ void decode_env::print_insn_labels() {
   if (cb != NULL) {
     cb->print_block_comment(st, p);
   }
+  _comments.print_block_comment(st, (intptr_t)(p - _start));
   if (_print_pc) {
     st->print("  " PTR_FORMAT ": ", p);
   }
@@ -471,10 +474,9 @@ void Disassembler::decode(CodeBlob* cb, outputStream* st) {
   env.decode_instructions(cb->code_begin(), cb->code_end());
 }
 
-
-void Disassembler::decode(address start, address end, outputStream* st) {
+void Disassembler::decode(address start, address end, outputStream* st, CodeComments c) {
   if (!load_library())  return;
-  decode_env env(CodeCache::find_blob_unsafe(start), st);
+  decode_env env(CodeCache::find_blob_unsafe(start), st, c);
   env.decode_instructions(start, end);
 }
 
diff --git a/src/share/vm/compiler/disassembler.hpp b/src/share/vm/compiler/disassembler.hpp
index 2507e7bf8..053b4689f 100644
--- a/src/share/vm/compiler/disassembler.hpp
+++ b/src/share/vm/compiler/disassembler.hpp
@@ -25,6 +25,7 @@
 #ifndef SHARE_VM_COMPILER_DISASSEMBLER_HPP
 #define SHARE_VM_COMPILER_DISASSEMBLER_HPP
 
+#include "asm/codeBuffer.hpp"
 #include "runtime/globals.hpp"
 #ifdef TARGET_OS_FAMILY_linux
 # include "os_linux.inline.hpp"
@@ -88,7 +89,7 @@ class Disassembler {
   }
   static void decode(CodeBlob *cb,               outputStream* st = NULL);
   static void decode(nmethod* nm,                outputStream* st = NULL);
-  static void decode(address begin, address end, outputStream* st = NULL);
+  static void decode(address begin, address end, outputStream* st = NULL, CodeComments c = CodeComments());
 };
 
 #endif // SHARE_VM_COMPILER_DISASSEMBLER_HPP
diff --git a/src/share/vm/interpreter/interpreter.cpp b/src/share/vm/interpreter/interpreter.cpp
index 9e329ebf8..dad8f9ec7 100644
--- a/src/share/vm/interpreter/interpreter.cpp
+++ b/src/share/vm/interpreter/interpreter.cpp
@@ -60,6 +60,8 @@ void InterpreterCodelet::verify() {
 
 
 void InterpreterCodelet::print_on(outputStream* st) const {
+  ttyLocker ttyl;
+
   if (PrintInterpreter) {
     st->cr();
     st->print_cr("----------------------------------------------------------------------");
@@ -72,7 +74,7 @@ void InterpreterCodelet::print_on(outputStream* st) const {
 
   if (PrintInterpreter) {
     st->cr();
-    Disassembler::decode(code_begin(), code_end(), st);
+    Disassembler::decode(code_begin(), code_end(), st, DEBUG_ONLY(_comments) NOT_DEBUG(CodeComments()));
   }
 }
 
diff --git a/src/share/vm/interpreter/interpreter.hpp b/src/share/vm/interpreter/interpreter.hpp
index 0ab0be747..478edb1b3 100644
--- a/src/share/vm/interpreter/interpreter.hpp
+++ b/src/share/vm/interpreter/interpreter.hpp
@@ -48,10 +48,12 @@ class InterpreterCodelet: public Stub {
   int         _size;                             // the size in bytes
   const char* _description;                      // a description of the codelet, for debugging & printing
   Bytecodes::Code _bytecode;                     // associated bytecode if any
+  DEBUG_ONLY(CodeComments _comments;)            // Comments for annotating assembler output.
 
  public:
   // Initialization/finalization
-  void    initialize(int size)                   { _size = size; }
+  void    initialize(int size,
+                     CodeComments& comments)     { _size = size; DEBUG_ONLY(_comments.assign(comments);) }
   void    finalize()                             { ShouldNotCallThis(); }
 
   // General info/converters
@@ -129,7 +131,7 @@ class CodeletMark: ResourceMark {
 
 
     // commit Codelet
-    AbstractInterpreter::code()->commit((*_masm)->code()->pure_insts_size());
+    AbstractInterpreter::code()->commit((*_masm)->code()->pure_insts_size(), (*_masm)->code()->comments());
     // make sure nobody can use _masm outside a CodeletMark lifespan
     *_masm = NULL;
   }
diff --git a/src/share/vm/oops/method.cpp b/src/share/vm/oops/method.cpp
index e6f3a4a7e..d57f5df26 100644
--- a/src/share/vm/oops/method.cpp
+++ b/src/share/vm/oops/method.cpp
@@ -251,8 +251,12 @@ void Method::mask_for(int bci, InterpreterOopMap* mask) {
 
 
 int Method::bci_from(address bcp) const {
+#ifdef ASSERT
+  { ResourceMark rm;
   assert(is_native() && bcp == code_base() || contains(bcp) || is_error_reported(),
          err_msg("bcp doesn't belong to this method: bcp: " INTPTR_FORMAT ", method: %s", bcp, name_and_sig_as_C_string()));
+  }
+#endif
   return bcp - code_base();
 }
 
@@ -688,30 +692,18 @@ void Method::set_signature_handler(address handler) {
 }
 
 
-bool Method::is_not_compilable(int comp_level) const {
-  if (number_of_breakpoints() > 0) {
-    return true;
-  }
-  if (is_method_handle_intrinsic()) {
-    return !is_synthetic();  // the generated adapters must be compiled
-  }
-  if (comp_level == CompLevel_any) {
-    return is_not_c1_compilable() || is_not_c2_compilable();
-  }
-  if (is_c1_compile(comp_level)) {
-    return is_not_c1_compilable();
-  }
-  if (is_c2_compile(comp_level)) {
-    return is_not_c2_compilable();
-  }
-  return false;
-}
-
-// call this when compiler finds that this method is not compilable
-void Method::set_not_compilable(int comp_level, bool report) {
+void Method::print_made_not_compilable(int comp_level, bool is_osr, bool report) {
   if (PrintCompilation && report) {
     ttyLocker ttyl;
-    tty->print("made not compilable ");
+    tty->print("made not %scompilable on ", is_osr ? "OSR " : "");
+    if (comp_level == CompLevel_all) {
+      tty->print("all levels ");
+    } else {
+      tty->print("levels ");
+      for (int i = (int)CompLevel_none; i <= comp_level; i++) {
+        tty->print("%d ", i);
+      }
+    }
     this->print_short_name(tty);
     int size = this->code_size();
     if (size > 0)
@@ -720,21 +712,64 @@ void Method::set_not_compilable(int comp_level, bool report) {
   }
   if ((TraceDeoptimization || LogCompilation) && (xtty != NULL)) {
     ttyLocker ttyl;
-    xtty->begin_elem("make_not_compilable thread='%d'", (int) os::current_thread_id());
+    xtty->begin_elem("make_not_%scompilable thread='%d'", is_osr ? "osr_" : "", (int) os::current_thread_id());
     xtty->method(this);
     xtty->stamp();
     xtty->end_elem();
   }
+}
+
+bool Method::is_not_compilable(int comp_level) const {
+  if (number_of_breakpoints() > 0)
+    return true;
+  if (is_method_handle_intrinsic())
+    return !is_synthetic();  // the generated adapters must be compiled
+  if (comp_level == CompLevel_any)
+    return is_not_c1_compilable() || is_not_c2_compilable();
+  if (is_c1_compile(comp_level))
+    return is_not_c1_compilable();
+  if (is_c2_compile(comp_level))
+    return is_not_c2_compilable();
+  return false;
+}
+
+// call this when compiler finds that this method is not compilable
+void Method::set_not_compilable(int comp_level, bool report) {
+  print_made_not_compilable(comp_level, /*is_osr*/ false, report);
   if (comp_level == CompLevel_all) {
     set_not_c1_compilable();
     set_not_c2_compilable();
   } else {
-    if (is_c1_compile(comp_level)) {
+    if (is_c1_compile(comp_level))
       set_not_c1_compilable();
-    } else
-      if (is_c2_compile(comp_level)) {
-        set_not_c2_compilable();
-      }
+    if (is_c2_compile(comp_level))
+      set_not_c2_compilable();
+  }
+  CompilationPolicy::policy()->disable_compilation(this);
+}
+
+bool Method::is_not_osr_compilable(int comp_level) const {
+  if (is_not_compilable(comp_level))
+    return true;
+  if (comp_level == CompLevel_any)
+    return is_not_c1_osr_compilable() || is_not_c2_osr_compilable();
+  if (is_c1_compile(comp_level))
+    return is_not_c1_osr_compilable();
+  if (is_c2_compile(comp_level))
+    return is_not_c2_osr_compilable();
+  return false;
+}
+
+void Method::set_not_osr_compilable(int comp_level, bool report) {
+  print_made_not_compilable(comp_level, /*is_osr*/ true, report);
+  if (comp_level == CompLevel_all) {
+    set_not_c1_osr_compilable();
+    set_not_c2_osr_compilable();
+  } else {
+    if (is_c1_compile(comp_level))
+      set_not_c1_osr_compilable();
+    if (is_c2_compile(comp_level))
+      set_not_c2_osr_compilable();
   }
   CompilationPolicy::policy()->disable_compilation(this);
 }
diff --git a/src/share/vm/oops/method.hpp b/src/share/vm/oops/method.hpp
index 66dfbae3a..5e779a57c 100644
--- a/src/share/vm/oops/method.hpp
+++ b/src/share/vm/oops/method.hpp
@@ -745,19 +745,30 @@ class Method : public Metadata {
   // Indicates whether compilation failed earlier for this method, or
   // whether it is not compilable for another reason like having a
   // breakpoint set in it.
-  bool is_not_compilable(int comp_level = CompLevel_any) const;
+  bool  is_not_compilable(int comp_level = CompLevel_any) const;
   void set_not_compilable(int comp_level = CompLevel_all, bool report = true);
   void set_not_compilable_quietly(int comp_level = CompLevel_all) {
     set_not_compilable(comp_level, false);
   }
-  bool is_not_osr_compilable(int comp_level = CompLevel_any) const {
-    return is_not_compilable(comp_level) || access_flags().is_not_osr_compilable();
+  bool  is_not_osr_compilable(int comp_level = CompLevel_any) const;
+  void set_not_osr_compilable(int comp_level = CompLevel_all, bool report = true);
+  void set_not_osr_compilable_quietly(int comp_level = CompLevel_all) {
+    set_not_osr_compilable(comp_level, false);
   }
-  void set_not_osr_compilable()               { _access_flags.set_not_osr_compilable();       }
-  bool is_not_c1_compilable() const           { return access_flags().is_not_c1_compilable(); }
-  void set_not_c1_compilable()                { _access_flags.set_not_c1_compilable();        }
-  bool is_not_c2_compilable() const           { return access_flags().is_not_c2_compilable(); }
-  void set_not_c2_compilable()                { _access_flags.set_not_c2_compilable();        }
+
+ private:
+  void print_made_not_compilable(int comp_level, bool is_osr, bool report);
+
+ public:
+  bool  is_not_c1_compilable() const          { return access_flags().is_not_c1_compilable(); }
+  void set_not_c1_compilable()                {       _access_flags.set_not_c1_compilable();  }
+  bool  is_not_c2_compilable() const          { return access_flags().is_not_c2_compilable(); }
+  void set_not_c2_compilable()                {       _access_flags.set_not_c2_compilable();  }
+
+  bool  is_not_c1_osr_compilable() const      { return is_not_c1_compilable(); }  // don't waste an accessFlags bit
+  void set_not_c1_osr_compilable()            {       set_not_c1_compilable(); }  // don't waste an accessFlags bit
+  bool  is_not_c2_osr_compilable() const      { return access_flags().is_not_c2_osr_compilable(); }
+  void set_not_c2_osr_compilable()            {       _access_flags.set_not_c2_osr_compilable();  }
 
   // Background compilation support
   bool queued_for_compilation() const  { return access_flags().queued_for_compilation(); }
diff --git a/src/share/vm/opto/addnode.cpp b/src/share/vm/opto/addnode.cpp
index ee6f6bc84..b3b92ec2c 100644
--- a/src/share/vm/opto/addnode.cpp
+++ b/src/share/vm/opto/addnode.cpp
@@ -248,47 +248,47 @@ Node *AddINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     const Type *t_sub1 = phase->type( in1->in(1) );
     const Type *t_2    = phase->type( in2        );
     if( t_sub1->singleton() && t_2->singleton() && t_sub1 != Type::TOP && t_2 != Type::TOP )
-      return new (phase->C, 3) SubINode(phase->makecon( add_ring( t_sub1, t_2 ) ),
+      return new (phase->C) SubINode(phase->makecon( add_ring( t_sub1, t_2 ) ),
                               in1->in(2) );
     // Convert "(a-b)+(c-d)" into "(a+c)-(b+d)"
     if( op2 == Op_SubI ) {
       // Check for dead cycle: d = (a-b)+(c-d)
       assert( in1->in(2) != this && in2->in(2) != this,
               "dead loop in AddINode::Ideal" );
-      Node *sub  = new (phase->C, 3) SubINode(NULL, NULL);
-      sub->init_req(1, phase->transform(new (phase->C, 3) AddINode(in1->in(1), in2->in(1) ) ));
-      sub->init_req(2, phase->transform(new (phase->C, 3) AddINode(in1->in(2), in2->in(2) ) ));
+      Node *sub  = new (phase->C) SubINode(NULL, NULL);
+      sub->init_req(1, phase->transform(new (phase->C) AddINode(in1->in(1), in2->in(1) ) ));
+      sub->init_req(2, phase->transform(new (phase->C) AddINode(in1->in(2), in2->in(2) ) ));
       return sub;
     }
     // Convert "(a-b)+(b+c)" into "(a+c)"
     if( op2 == Op_AddI && in1->in(2) == in2->in(1) ) {
       assert(in1->in(1) != this && in2->in(2) != this,"dead loop in AddINode::Ideal");
-      return new (phase->C, 3) AddINode(in1->in(1), in2->in(2));
+      return new (phase->C) AddINode(in1->in(1), in2->in(2));
     }
     // Convert "(a-b)+(c+b)" into "(a+c)"
     if( op2 == Op_AddI && in1->in(2) == in2->in(2) ) {
       assert(in1->in(1) != this && in2->in(1) != this,"dead loop in AddINode::Ideal");
-      return new (phase->C, 3) AddINode(in1->in(1), in2->in(1));
+      return new (phase->C) AddINode(in1->in(1), in2->in(1));
     }
     // Convert "(a-b)+(b-c)" into "(a-c)"
     if( op2 == Op_SubI && in1->in(2) == in2->in(1) ) {
       assert(in1->in(1) != this && in2->in(2) != this,"dead loop in AddINode::Ideal");
-      return new (phase->C, 3) SubINode(in1->in(1), in2->in(2));
+      return new (phase->C) SubINode(in1->in(1), in2->in(2));
     }
     // Convert "(a-b)+(c-a)" into "(c-b)"
     if( op2 == Op_SubI && in1->in(1) == in2->in(2) ) {
       assert(in1->in(2) != this && in2->in(1) != this,"dead loop in AddINode::Ideal");
-      return new (phase->C, 3) SubINode(in2->in(1), in1->in(2));
+      return new (phase->C) SubINode(in2->in(1), in1->in(2));
     }
   }
 
   // Convert "x+(0-y)" into "(x-y)"
   if( op2 == Op_SubI && phase->type(in2->in(1)) == TypeInt::ZERO )
-    return new (phase->C, 3) SubINode(in1, in2->in(2) );
+    return new (phase->C) SubINode(in1, in2->in(2) );
 
   // Convert "(0-y)+x" into "(x-y)"
   if( op1 == Op_SubI && phase->type(in1->in(1)) == TypeInt::ZERO )
-    return new (phase->C, 3) SubINode( in2, in1->in(2) );
+    return new (phase->C) SubINode( in2, in1->in(2) );
 
   // Convert (x>>>z)+y into (x+(y<<z))>>>z for small constant z and y.
   // Helps with array allocation math constant folding
@@ -309,8 +309,8 @@ Node *AddINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     if( z < 5 && -5 < y && y < 0 ) {
       const Type *t_in11 = phase->type(in1->in(1));
       if( t_in11 != Type::TOP && (t_in11->is_int()->_lo >= -(y << z)) ) {
-        Node *a = phase->transform( new (phase->C, 3) AddINode( in1->in(1), phase->intcon(y<<z) ) );
-        return new (phase->C, 3) URShiftINode( a, in1->in(2) );
+        Node *a = phase->transform( new (phase->C) AddINode( in1->in(1), phase->intcon(y<<z) ) );
+        return new (phase->C) URShiftINode( a, in1->in(2) );
       }
     }
   }
@@ -381,47 +381,47 @@ Node *AddLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     const Type *t_sub1 = phase->type( in1->in(1) );
     const Type *t_2    = phase->type( in2        );
     if( t_sub1->singleton() && t_2->singleton() && t_sub1 != Type::TOP && t_2 != Type::TOP )
-      return new (phase->C, 3) SubLNode(phase->makecon( add_ring( t_sub1, t_2 ) ),
+      return new (phase->C) SubLNode(phase->makecon( add_ring( t_sub1, t_2 ) ),
                               in1->in(2) );
     // Convert "(a-b)+(c-d)" into "(a+c)-(b+d)"
     if( op2 == Op_SubL ) {
       // Check for dead cycle: d = (a-b)+(c-d)
       assert( in1->in(2) != this && in2->in(2) != this,
               "dead loop in AddLNode::Ideal" );
-      Node *sub  = new (phase->C, 3) SubLNode(NULL, NULL);
-      sub->init_req(1, phase->transform(new (phase->C, 3) AddLNode(in1->in(1), in2->in(1) ) ));
-      sub->init_req(2, phase->transform(new (phase->C, 3) AddLNode(in1->in(2), in2->in(2) ) ));
+      Node *sub  = new (phase->C) SubLNode(NULL, NULL);
+      sub->init_req(1, phase->transform(new (phase->C) AddLNode(in1->in(1), in2->in(1) ) ));
+      sub->init_req(2, phase->transform(new (phase->C) AddLNode(in1->in(2), in2->in(2) ) ));
       return sub;
     }
     // Convert "(a-b)+(b+c)" into "(a+c)"
     if( op2 == Op_AddL && in1->in(2) == in2->in(1) ) {
       assert(in1->in(1) != this && in2->in(2) != this,"dead loop in AddLNode::Ideal");
-      return new (phase->C, 3) AddLNode(in1->in(1), in2->in(2));
+      return new (phase->C) AddLNode(in1->in(1), in2->in(2));
     }
     // Convert "(a-b)+(c+b)" into "(a+c)"
     if( op2 == Op_AddL && in1->in(2) == in2->in(2) ) {
       assert(in1->in(1) != this && in2->in(1) != this,"dead loop in AddLNode::Ideal");
-      return new (phase->C, 3) AddLNode(in1->in(1), in2->in(1));
+      return new (phase->C) AddLNode(in1->in(1), in2->in(1));
     }
     // Convert "(a-b)+(b-c)" into "(a-c)"
     if( op2 == Op_SubL && in1->in(2) == in2->in(1) ) {
       assert(in1->in(1) != this && in2->in(2) != this,"dead loop in AddLNode::Ideal");
-      return new (phase->C, 3) SubLNode(in1->in(1), in2->in(2));
+      return new (phase->C) SubLNode(in1->in(1), in2->in(2));
     }
     // Convert "(a-b)+(c-a)" into "(c-b)"
     if( op2 == Op_SubL && in1->in(1) == in1->in(2) ) {
       assert(in1->in(2) != this && in2->in(1) != this,"dead loop in AddLNode::Ideal");
-      return new (phase->C, 3) SubLNode(in2->in(1), in1->in(2));
+      return new (phase->C) SubLNode(in2->in(1), in1->in(2));
     }
   }
 
   // Convert "x+(0-y)" into "(x-y)"
   if( op2 == Op_SubL && phase->type(in2->in(1)) == TypeLong::ZERO )
-    return new (phase->C, 3) SubLNode( in1, in2->in(2) );
+    return new (phase->C) SubLNode( in1, in2->in(2) );
 
   // Convert "(0-y)+x" into "(x-y)"
   if( op1 == Op_SubL && phase->type(in1->in(1)) == TypeInt::ZERO )
-    return new (phase->C, 3) SubLNode( in2, in1->in(2) );
+    return new (phase->C) SubLNode( in2, in1->in(2) );
 
   // Convert "X+X+X+X+X...+X+Y" into "k*X+Y" or really convert "X+(X+Y)"
   // into "(X<<1)+Y" and let shift-folding happen.
@@ -429,8 +429,8 @@ Node *AddLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       in2->in(1) == in1 &&
       op1 != Op_ConL &&
       0 ) {
-    Node *shift = phase->transform(new (phase->C, 3) LShiftLNode(in1,phase->intcon(1)));
-    return new (phase->C, 3) AddLNode(shift,in2->in(2));
+    Node *shift = phase->transform(new (phase->C) LShiftLNode(in1,phase->intcon(1)));
+    return new (phase->C) AddLNode(shift,in2->in(2));
   }
 
   return AddNode::Ideal(phase, can_reshape);
@@ -590,7 +590,7 @@ Node *AddPNode::Ideal(PhaseGVN *phase, bool can_reshape) {
         offset  = phase->MakeConX(t2->get_con() + t12->get_con());
       } else {
         // Else move the constant to the right.  ((A+con)+B) into ((A+B)+con)
-        address = phase->transform(new (phase->C, 4) AddPNode(in(Base),addp->in(Address),in(Offset)));
+        address = phase->transform(new (phase->C) AddPNode(in(Base),addp->in(Address),in(Offset)));
         offset  = addp->in(Offset);
       }
       PhaseIterGVN *igvn = phase->is_IterGVN();
@@ -610,7 +610,7 @@ Node *AddPNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     // If this is a NULL+long form (from unsafe accesses), switch to a rawptr.
     if (phase->type(in(Address)) == TypePtr::NULL_PTR) {
       Node* offset = in(Offset);
-      return new (phase->C, 2) CastX2PNode(offset);
+      return new (phase->C) CastX2PNode(offset);
     }
   }
 
@@ -622,7 +622,7 @@ Node *AddPNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if( add->Opcode() == Op_AddX && add->in(1) != add ) {
     const Type *t22 = phase->type( add->in(2) );
     if( t22->singleton() && (t22 != Type::TOP) ) {  // Right input is an add of a constant?
-      set_req(Address, phase->transform(new (phase->C, 4) AddPNode(in(Base),in(Address),add->in(1))));
+      set_req(Address, phase->transform(new (phase->C) AddPNode(in(Base),in(Address),add->in(1))));
       set_req(Offset, add->in(2));
       return this;              // Made progress
     }
@@ -847,7 +847,7 @@ Node *MinINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // to force a right-spline graph for the rest of MinINode::Ideal().
   if( l->Opcode() == Op_MinI ) {
     assert( l != l->in(1), "dead loop in MinINode::Ideal" );
-    r = phase->transform(new (phase->C, 3) MinINode(l->in(2),r));
+    r = phase->transform(new (phase->C) MinINode(l->in(2),r));
     l = l->in(1);
     set_req(1, l);
     set_req(2, r);
@@ -895,18 +895,18 @@ Node *MinINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     }
 
     if( x->_idx > y->_idx )
-      return new (phase->C, 3) MinINode(r->in(1),phase->transform(new (phase->C, 3) MinINode(l,r->in(2))));
+      return new (phase->C) MinINode(r->in(1),phase->transform(new (phase->C) MinINode(l,r->in(2))));
 
     // See if covers: MIN2(x+c0,MIN2(y+c1,z))
     if( !phase->eqv(x,y) ) return NULL;
     // If (y == x) transform MIN2(x+c0, MIN2(x+c1,z)) into
     // MIN2(x+c0 or x+c1 which less, z).
-    return new (phase->C, 3) MinINode(phase->transform(new (phase->C, 3) AddINode(x,phase->intcon(MIN2(x_off,y_off)))),r->in(2));
+    return new (phase->C) MinINode(phase->transform(new (phase->C) AddINode(x,phase->intcon(MIN2(x_off,y_off)))),r->in(2));
   } else {
     // See if covers: MIN2(x+c0,y+c1)
     if( !phase->eqv(x,y) ) return NULL;
     // If (y == x) transform MIN2(x+c0,x+c1) into x+c0 or x+c1 which less.
-    return new (phase->C, 3) AddINode(x,phase->intcon(MIN2(x_off,y_off)));
+    return new (phase->C) AddINode(x,phase->intcon(MIN2(x_off,y_off)));
   }
 
 }
diff --git a/src/share/vm/opto/block.cpp b/src/share/vm/opto/block.cpp
index 3e6b7bf1e..60b0aa9da 100644
--- a/src/share/vm/opto/block.cpp
+++ b/src/share/vm/opto/block.cpp
@@ -378,7 +378,7 @@ PhaseCFG::PhaseCFG( Arena *a, RootNode *r, Matcher &m ) :
   // I'll need a few machine-specific GotoNodes.  Make an Ideal GotoNode,
   // then Match it into a machine-specific Node.  Then clone the machine
   // Node on demand.
-  Node *x = new (C, 1) GotoNode(NULL);
+  Node *x = new (C) GotoNode(NULL);
   x->init_req(0, x);
   _goto = m.match_tree(x);
   assert(_goto != NULL, "");
@@ -432,7 +432,7 @@ uint PhaseCFG::build_cfg() {
                !p->is_block_start() );
       // Make the block begin with one of Region or StartNode.
       if( !p->is_block_start() ) {
-        RegionNode *r = new (C, 2) RegionNode( 2 );
+        RegionNode *r = new (C) RegionNode( 2 );
         r->init_req(1, p);         // Insert RegionNode in the way
         proj->set_req(0, r);        // Insert RegionNode in the way
         p = r;
@@ -508,7 +508,7 @@ void PhaseCFG::insert_goto_at(uint block_no, uint succ_no) {
   // get ProjNode corresponding to the succ_no'th successor of the in block
   ProjNode* proj = in->_nodes[in->_nodes.size() - in->_num_succs + succ_no]->as_Proj();
   // create region for basic block
-  RegionNode* region = new (C, 2) RegionNode(2);
+  RegionNode* region = new (C) RegionNode(2);
   region->init_req(1, proj);
   // setup corresponding basic block
   Block* block = new (_bbs._arena) Block(_bbs._arena, region);
diff --git a/src/share/vm/opto/c2_globals.hpp b/src/share/vm/opto/c2_globals.hpp
index c5c7768dc..80996a5d7 100644
--- a/src/share/vm/opto/c2_globals.hpp
+++ b/src/share/vm/opto/c2_globals.hpp
@@ -85,7 +85,7 @@
           "Max vector size in bytes, "                                      \
           "actual size could be less depending on elements type")           \
                                                                             \
-  product(bool, AlignVector, false,                                         \
+  product(bool, AlignVector, true,                                          \
           "Perform vector store/load alignment in loop")                    \
                                                                             \
   product(intx, NumberOfLoopInstrToAlign, 4,                                \
@@ -535,7 +535,7 @@
   notproduct(bool, TraceSpilling, false,                                    \
           "Trace spilling")                                                 \
                                                                             \
-  notproduct(bool, TraceTypeProfile, false,                                 \
+  diagnostic(bool, TraceTypeProfile, false,                                 \
           "Trace type profile")                                             \
                                                                             \
   develop(bool, PoisonOSREntry, true,                                       \
diff --git a/src/share/vm/opto/callGenerator.cpp b/src/share/vm/opto/callGenerator.cpp
index 1c7b9ffee..547096b3d 100644
--- a/src/share/vm/opto/callGenerator.cpp
+++ b/src/share/vm/opto/callGenerator.cpp
@@ -134,7 +134,7 @@ JVMState* DirectCallGenerator::generate(JVMState* jvms) {
     kit.C->log()->elem("direct_call bci='%d'", jvms->bci());
   }
 
-  CallStaticJavaNode *call = new (kit.C, tf()->domain()->cnt()) CallStaticJavaNode(tf(), target, method(), kit.bci());
+  CallStaticJavaNode *call = new (kit.C) CallStaticJavaNode(tf(), target, method(), kit.bci());
   _call_node = call;  // Save the call node in case we need it later
   if (!is_static) {
     // Make an explicit receiver null_check as part of this call.
@@ -221,7 +221,7 @@ JVMState* VirtualCallGenerator::generate(JVMState* jvms) {
          "no vtable calls if +UseInlineCaches ");
   address target = SharedRuntime::get_resolve_virtual_call_stub();
   // Normal inline cache used for call
-  CallDynamicJavaNode *call = new (kit.C, tf()->domain()->cnt()) CallDynamicJavaNode(tf(), target, method(), _vtable_index, kit.bci());
+  CallDynamicJavaNode *call = new (kit.C) CallDynamicJavaNode(tf(), target, method(), _vtable_index, kit.bci());
   kit.set_arguments_for_java_call(call);
   kit.set_edges_for_java_call(call);
   Node* ret = kit.set_results_for_java_call(call);
@@ -300,7 +300,7 @@ void LateInlineCallGenerator::do_late_inline() {
   Compile* C = Compile::current();
   JVMState* jvms     = call->jvms()->clone_shallow(C);
   uint size = call->req();
-  SafePointNode* map = new (C, size) SafePointNode(size, jvms);
+  SafePointNode* map = new (C) SafePointNode(size, jvms);
   for (uint i1 = 0; i1 < size; i1++) {
     map->init_req(i1, call->in(i1));
   }
@@ -551,7 +551,7 @@ JVMState* PredictedCallGenerator::generate(JVMState* jvms) {
 
   // Finish the diamond.
   kit.C->set_has_split_ifs(true); // Has chance for split-if optimization
-  RegionNode* region = new (kit.C, 3) RegionNode(3);
+  RegionNode* region = new (kit.C) RegionNode(3);
   region->init_req(1, kit.control());
   region->init_req(2, slow_map->control());
   kit.set_control(gvn.transform(region));
@@ -636,7 +636,7 @@ CallGenerator* CallGenerator::for_method_handle_inline(JVMState* jvms, ciMethod*
           const TypeOopPtr* arg_type = arg->bottom_type()->isa_oopptr();
           const Type*       sig_type = TypeOopPtr::make_from_klass(signature->accessing_klass());
           if (arg_type != NULL && !arg_type->higher_equal(sig_type)) {
-            Node* cast_obj = gvn.transform(new (C, 2) CheckCastPPNode(kit.control(), arg, sig_type));
+            Node* cast_obj = gvn.transform(new (C) CheckCastPPNode(kit.control(), arg, sig_type));
             kit.set_argument(0, cast_obj);
           }
         }
@@ -648,7 +648,7 @@ CallGenerator* CallGenerator::for_method_handle_inline(JVMState* jvms, ciMethod*
             const TypeOopPtr* arg_type = arg->bottom_type()->isa_oopptr();
             const Type*       sig_type = TypeOopPtr::make_from_klass(t->as_klass());
             if (arg_type != NULL && !arg_type->higher_equal(sig_type)) {
-              Node* cast_obj = gvn.transform(new (C, 2) CheckCastPPNode(kit.control(), arg, sig_type));
+              Node* cast_obj = gvn.transform(new (C) CheckCastPPNode(kit.control(), arg, sig_type));
               kit.set_argument(receiver_skip + i, cast_obj);
             }
           }
diff --git a/src/share/vm/opto/callnode.cpp b/src/share/vm/opto/callnode.cpp
index 398497f9b..067e9197d 100644
--- a/src/share/vm/opto/callnode.cpp
+++ b/src/share/vm/opto/callnode.cpp
@@ -72,20 +72,20 @@ Node *StartNode::match( const ProjNode *proj, const Matcher *match ) {
   case TypeFunc::Control:
   case TypeFunc::I_O:
   case TypeFunc::Memory:
-    return new (match->C, 1) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
+    return new (match->C) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
   case TypeFunc::FramePtr:
-    return new (match->C, 1) MachProjNode(this,proj->_con,Matcher::c_frame_ptr_mask, Op_RegP);
+    return new (match->C) MachProjNode(this,proj->_con,Matcher::c_frame_ptr_mask, Op_RegP);
   case TypeFunc::ReturnAdr:
-    return new (match->C, 1) MachProjNode(this,proj->_con,match->_return_addr_mask,Op_RegP);
+    return new (match->C) MachProjNode(this,proj->_con,match->_return_addr_mask,Op_RegP);
   case TypeFunc::Parms:
   default: {
       uint parm_num = proj->_con - TypeFunc::Parms;
       const Type *t = _domain->field_at(proj->_con);
       if (t->base() == Type::Half)  // 2nd half of Longs and Doubles
-        return new (match->C, 1) ConNode(Type::TOP);
+        return new (match->C) ConNode(Type::TOP);
       uint ideal_reg = t->ideal_reg();
       RegMask &rm = match->_calling_convention_mask[parm_num];
-      return new (match->C, 1) MachProjNode(this,proj->_con,rm,ideal_reg);
+      return new (match->C) MachProjNode(this,proj->_con,rm,ideal_reg);
     }
   }
   return NULL;
@@ -625,12 +625,12 @@ Node *CallNode::match( const ProjNode *proj, const Matcher *match ) {
   case TypeFunc::Control:
   case TypeFunc::I_O:
   case TypeFunc::Memory:
-    return new (match->C, 1) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
+    return new (match->C) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
 
   case TypeFunc::Parms+1:       // For LONG & DOUBLE returns
     assert(tf()->_range->field_at(TypeFunc::Parms+1) == Type::HALF, "");
     // 2nd half of doubles and longs
-    return new (match->C, 1) MachProjNode(this,proj->_con, RegMask::Empty, (uint)OptoReg::Bad);
+    return new (match->C) MachProjNode(this,proj->_con, RegMask::Empty, (uint)OptoReg::Bad);
 
   case TypeFunc::Parms: {       // Normal returns
     uint ideal_reg = tf()->range()->field_at(TypeFunc::Parms)->ideal_reg();
@@ -640,7 +640,7 @@ Node *CallNode::match( const ProjNode *proj, const Matcher *match ) {
     RegMask rm = RegMask(regs.first());
     if( OptoReg::is_valid(regs.second()) )
       rm.Insert( regs.second() );
-    return new (match->C, 1) MachProjNode(this,proj->_con,rm,ideal_reg);
+    return new (match->C) MachProjNode(this,proj->_con,rm,ideal_reg);
   }
 
   case TypeFunc::ReturnAdr:
@@ -1175,10 +1175,10 @@ Node* AllocateArrayNode::Ideal(PhaseGVN *phase, bool can_reshape) {
         Node* nproj = catchproj->clone();
         igvn->register_new_node_with_optimizer(nproj);
 
-        Node *frame = new (phase->C, 1) ParmNode( phase->C->start(), TypeFunc::FramePtr );
+        Node *frame = new (phase->C) ParmNode( phase->C->start(), TypeFunc::FramePtr );
         frame = phase->transform(frame);
         // Halt & Catch Fire
-        Node *halt = new (phase->C, TypeFunc::Parms) HaltNode( nproj, frame );
+        Node *halt = new (phase->C) HaltNode( nproj, frame );
         phase->C->root()->add_req(halt);
         phase->transform(halt);
 
@@ -1218,7 +1218,7 @@ Node *AllocateArrayNode::make_ideal_length(const TypeOopPtr* oop_type, PhaseTran
       if (!allow_new_nodes) return NULL;
       // Create a cast which is control dependent on the initialization to
       // propagate the fact that the array length must be positive.
-      length = new (phase->C, 2) CastIINode(length, narrow_length_type);
+      length = new (phase->C) CastIINode(length, narrow_length_type);
       length->set_req(0, initialization()->proj_out(0));
     }
   }
diff --git a/src/share/vm/opto/cfgnode.cpp b/src/share/vm/opto/cfgnode.cpp
index 795673da2..a85cd34c7 100644
--- a/src/share/vm/opto/cfgnode.cpp
+++ b/src/share/vm/opto/cfgnode.cpp
@@ -612,17 +612,17 @@ Node *RegionNode::Ideal(PhaseGVN *phase, bool can_reshape) {
                 convf2i->in(1) == bot_in ) {
                 // Matched pattern, including LShiftI; RShiftI, replace with integer compares
                 // max test
-                Node *cmp   = gvn->register_new_node_with_optimizer(new (phase->C, 3) CmpINode( convf2i, min ));
-                Node *boo   = gvn->register_new_node_with_optimizer(new (phase->C, 2) BoolNode( cmp, BoolTest::lt ));
-                IfNode *iff = (IfNode*)gvn->register_new_node_with_optimizer(new (phase->C, 2) IfNode( top_if->in(0), boo, PROB_UNLIKELY_MAG(5), top_if->_fcnt ));
-                Node *if_min= gvn->register_new_node_with_optimizer(new (phase->C, 1) IfTrueNode (iff));
-                Node *ifF   = gvn->register_new_node_with_optimizer(new (phase->C, 1) IfFalseNode(iff));
+                Node *cmp   = gvn->register_new_node_with_optimizer(new (phase->C) CmpINode( convf2i, min ));
+                Node *boo   = gvn->register_new_node_with_optimizer(new (phase->C) BoolNode( cmp, BoolTest::lt ));
+                IfNode *iff = (IfNode*)gvn->register_new_node_with_optimizer(new (phase->C) IfNode( top_if->in(0), boo, PROB_UNLIKELY_MAG(5), top_if->_fcnt ));
+                Node *if_min= gvn->register_new_node_with_optimizer(new (phase->C) IfTrueNode (iff));
+                Node *ifF   = gvn->register_new_node_with_optimizer(new (phase->C) IfFalseNode(iff));
                 // min test
-                cmp         = gvn->register_new_node_with_optimizer(new (phase->C, 3) CmpINode( convf2i, max ));
-                boo         = gvn->register_new_node_with_optimizer(new (phase->C, 2) BoolNode( cmp, BoolTest::gt ));
-                iff         = (IfNode*)gvn->register_new_node_with_optimizer(new (phase->C, 2) IfNode( ifF, boo, PROB_UNLIKELY_MAG(5), bot_if->_fcnt ));
-                Node *if_max= gvn->register_new_node_with_optimizer(new (phase->C, 1) IfTrueNode (iff));
-                ifF         = gvn->register_new_node_with_optimizer(new (phase->C, 1) IfFalseNode(iff));
+                cmp         = gvn->register_new_node_with_optimizer(new (phase->C) CmpINode( convf2i, max ));
+                boo         = gvn->register_new_node_with_optimizer(new (phase->C) BoolNode( cmp, BoolTest::gt ));
+                iff         = (IfNode*)gvn->register_new_node_with_optimizer(new (phase->C) IfNode( ifF, boo, PROB_UNLIKELY_MAG(5), bot_if->_fcnt ));
+                Node *if_max= gvn->register_new_node_with_optimizer(new (phase->C) IfTrueNode (iff));
+                ifF         = gvn->register_new_node_with_optimizer(new (phase->C) IfFalseNode(iff));
                 // update input edges to region node
                 set_req_X( min_idx, if_min, gvn );
                 set_req_X( max_idx, if_max, gvn );
@@ -681,7 +681,7 @@ const TypePtr* flatten_phi_adr_type(const TypePtr* at) {
 PhiNode* PhiNode::make(Node* r, Node* x, const Type *t, const TypePtr* at) {
   uint preds = r->req();   // Number of predecessor paths
   assert(t != Type::MEMORY || at == flatten_phi_adr_type(at), "flatten at");
-  PhiNode* p = new (Compile::current(), preds) PhiNode(r, t, at);
+  PhiNode* p = new (Compile::current()) PhiNode(r, t, at);
   for (uint j = 1; j < preds; j++) {
     // Fill in all inputs, except those which the region does not yet have
     if (r->in(j) != NULL)
@@ -699,7 +699,7 @@ PhiNode* PhiNode::make_blank(Node* r, Node* x) {
   const Type*    t  = x->bottom_type();
   const TypePtr* at = NULL;
   if (t == Type::MEMORY)  at = flatten_phi_adr_type(x->adr_type());
-  return new (Compile::current(), r->req()) PhiNode(r, t, at);
+  return new (Compile::current()) PhiNode(r, t, at);
 }
 
 
@@ -1205,9 +1205,9 @@ static Node *is_x2logic( PhaseGVN *phase, PhiNode *phi, int true_path ) {
   } else return NULL;
 
   // Build int->bool conversion
-  Node *n = new (phase->C, 2) Conv2BNode( cmp->in(1) );
+  Node *n = new (phase->C) Conv2BNode( cmp->in(1) );
   if( flipped )
-    n = new (phase->C, 3) XorINode( phase->transform(n), phase->intcon(1) );
+    n = new (phase->C) XorINode( phase->transform(n), phase->intcon(1) );
 
   return n;
 }
@@ -1266,9 +1266,9 @@ static Node* is_cond_add(PhaseGVN *phase, PhiNode *phi, int true_path) {
   if( q->is_Con() && phase->type(q) != TypeInt::ZERO && y->is_Con() )
     return NULL;
 
-  Node *cmplt = phase->transform( new (phase->C, 3) CmpLTMaskNode(p,q) );
-  Node *j_and   = phase->transform( new (phase->C, 3) AndINode(cmplt,y) );
-  return new (phase->C, 3) AddINode(j_and,x);
+  Node *cmplt = phase->transform( new (phase->C) CmpLTMaskNode(p,q) );
+  Node *j_and   = phase->transform( new (phase->C) AndINode(cmplt,y) );
+  return new (phase->C) AddINode(j_and,x);
 }
 
 //------------------------------is_absolute------------------------------------
@@ -1330,17 +1330,17 @@ static Node* is_absolute( PhaseGVN *phase, PhiNode *phi_root, int true_path) {
     if( sub->Opcode() != Op_SubF ||
         sub->in(2) != x ||
         phase->type(sub->in(1)) != tzero ) return NULL;
-    x = new (phase->C, 2) AbsFNode(x);
+    x = new (phase->C) AbsFNode(x);
     if (flip) {
-      x = new (phase->C, 3) SubFNode(sub->in(1), phase->transform(x));
+      x = new (phase->C) SubFNode(sub->in(1), phase->transform(x));
     }
   } else {
     if( sub->Opcode() != Op_SubD ||
         sub->in(2) != x ||
         phase->type(sub->in(1)) != tzero ) return NULL;
-    x = new (phase->C, 2) AbsDNode(x);
+    x = new (phase->C) AbsDNode(x);
     if (flip) {
-      x = new (phase->C, 3) SubDNode(sub->in(1), phase->transform(x));
+      x = new (phase->C) SubDNode(sub->in(1), phase->transform(x));
     }
   }
 
@@ -1415,7 +1415,7 @@ static Node* split_flow_path(PhaseGVN *phase, PhiNode *phi) {
   // Now start splitting out the flow paths that merge the same value.
   // Split first the RegionNode.
   PhaseIterGVN *igvn = phase->is_IterGVN();
-  RegionNode *newr = new (phase->C, hit+1) RegionNode(hit+1);
+  RegionNode *newr = new (phase->C) RegionNode(hit+1);
   split_once(igvn, phi, val, r, newr);
 
   // Now split all other Phis than this one
@@ -1723,13 +1723,13 @@ Node *PhiNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       }
       if (doit) {
         if (base == NULL) {
-          base = new (phase->C, in(0)->req()) PhiNode(in(0), type, NULL);
+          base = new (phase->C) PhiNode(in(0), type, NULL);
           for (uint i = 1; i < req(); i++) {
             base->init_req(i, in(i)->in(AddPNode::Base));
           }
           phase->is_IterGVN()->register_new_node_with_optimizer(base);
         }
-        return new (phase->C, 4) AddPNode(base, base, y);
+        return new (phase->C) AddPNode(base, base, y);
       }
     }
   }
@@ -1806,7 +1806,7 @@ Node *PhiNode::Ideal(PhaseGVN *phase, bool can_reshape) {
         // Phi(...MergeMem(m0, m1:AT1, m2:AT2)...) into
         //     MergeMem(Phi(...m0...), Phi:AT1(...m1...), Phi:AT2(...m2...))
         PhaseIterGVN *igvn = phase->is_IterGVN();
-        Node* hook = new (phase->C, 1) Node(1);
+        Node* hook = new (phase->C) Node(1);
         PhiNode* new_base = (PhiNode*) clone();
         // Must eagerly register phis, since they participate in loops.
         if (igvn) {
@@ -1896,7 +1896,7 @@ Node *PhiNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       PhaseIterGVN *igvn = phase->is_IterGVN();
       // Make narrow type for new phi.
       const Type* narrow_t = TypeNarrowOop::make(this->bottom_type()->is_ptr());
-      PhiNode* new_phi = new (phase->C, r->req()) PhiNode(r, narrow_t);
+      PhiNode* new_phi = new (phase->C) PhiNode(r, narrow_t);
       uint orig_cnt = req();
       for (uint i=1; i<req(); ++i) {// For all paths in
         Node *ii = in(i);
@@ -1909,14 +1909,14 @@ Node *PhiNode::Ideal(PhaseGVN *phase, bool can_reshape) {
           if (ii->as_Phi() == this) {
             new_ii = new_phi;
           } else {
-            new_ii = new (phase->C, 2) EncodePNode(ii, narrow_t);
+            new_ii = new (phase->C) EncodePNode(ii, narrow_t);
             igvn->register_new_node_with_optimizer(new_ii);
           }
         }
         new_phi->set_req(i, new_ii);
       }
       igvn->register_new_node_with_optimizer(new_phi, this);
-      progress = new (phase->C, 2) DecodeNNode(new_phi, bottom_type());
+      progress = new (phase->C) DecodeNNode(new_phi, bottom_type());
     }
   }
 #endif
diff --git a/src/share/vm/opto/chaitin.cpp b/src/share/vm/opto/chaitin.cpp
index 9e5234de6..e626a10d8 100644
--- a/src/share/vm/opto/chaitin.cpp
+++ b/src/share/vm/opto/chaitin.cpp
@@ -1588,7 +1588,7 @@ Node *PhaseChaitin::find_base_for_derived( Node **derived_base_map, Node *derive
 
   // Now we see we need a base-Phi here to merge the bases
   const Type *t = base->bottom_type();
-  base = new (C, derived->req()) PhiNode( derived->in(0), t );
+  base = new (C) PhiNode( derived->in(0), t );
   for( i = 1; i < derived->req(); i++ ) {
     base->init_req(i, find_base_for_derived(derived_base_map, derived->in(i), maxlrg));
     t = t->meet(base->in(i)->bottom_type());
diff --git a/src/share/vm/opto/classes.hpp b/src/share/vm/opto/classes.hpp
index 02c62f68a..131658157 100644
--- a/src/share/vm/opto/classes.hpp
+++ b/src/share/vm/opto/classes.hpp
@@ -83,6 +83,12 @@ macro(CompareAndSwapI)
 macro(CompareAndSwapL)
 macro(CompareAndSwapP)
 macro(CompareAndSwapN)
+macro(GetAndAddI)
+macro(GetAndAddL)
+macro(GetAndSetI)
+macro(GetAndSetL)
+macro(GetAndSetP)
+macro(GetAndSetN)
 macro(Con)
 macro(ConN)
 macro(ConD)
diff --git a/src/share/vm/opto/compile.cpp b/src/share/vm/opto/compile.cpp
index 292c52412..36933f976 100644
--- a/src/share/vm/opto/compile.cpp
+++ b/src/share/vm/opto/compile.cpp
@@ -654,14 +654,14 @@ Compile::Compile( ciEnv* ci_env, C2Compiler* compiler, ciMethod* target, int osr
       const TypeTuple *domain = StartOSRNode::osr_domain();
       const TypeTuple *range = TypeTuple::make_range(method()->signature());
       init_tf(TypeFunc::make(domain, range));
-      StartNode* s = new (this, 2) StartOSRNode(root(), domain);
+      StartNode* s = new (this) StartOSRNode(root(), domain);
       initial_gvn()->set_type_bottom(s);
       init_start(s);
       cg = CallGenerator::for_osr(method(), entry_bci());
     } else {
       // Normal case.
       init_tf(TypeFunc::make(method()));
-      StartNode* s = new (this, 2) StartNode(root(), tf()->domain());
+      StartNode* s = new (this) StartNode(root(), tf()->domain());
       initial_gvn()->set_type_bottom(s);
       init_start(s);
       if (method()->intrinsic_id() == vmIntrinsics::_Reference_get && UseG1GC) {
@@ -825,7 +825,8 @@ Compile::Compile( ciEnv* ci_env, C2Compiler* compiler, ciMethod* target, int osr
                            &_handler_table, &_inc_table,
                            compiler,
                            env()->comp_level(),
-                           has_unsafe_access()
+                           has_unsafe_access(),
+                           SharedRuntime::is_wide_vector(max_vector_size())
                            );
   }
 }
@@ -946,9 +947,9 @@ void Compile::Init(int aliaslevel) {
   // Globally visible Nodes
   // First set TOP to NULL to give safe behavior during creation of RootNode
   set_cached_top_node(NULL);
-  set_root(new (this, 3) RootNode());
+  set_root(new (this) RootNode());
   // Now that you have a Root to point to, create the real TOP
-  set_cached_top_node( new (this, 1) ConNode(Type::TOP) );
+  set_cached_top_node( new (this) ConNode(Type::TOP) );
   set_recent_alloc(NULL, NULL);
 
   // Create Debug Information Recorder to record scopes, oopmaps, etc.
@@ -963,6 +964,7 @@ void Compile::Init(int aliaslevel) {
   _trap_can_recompile = false;  // no traps emitted yet
   _major_progress = true; // start out assuming good things will happen
   set_has_unsafe_access(false);
+  set_max_vector_size(0);
   Copy::zero_to_bytes(_trap_hist, sizeof(_trap_hist));
   set_decompile_count(0);
 
@@ -2274,6 +2276,12 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
   case Op_CompareAndSwapL:
   case Op_CompareAndSwapP:
   case Op_CompareAndSwapN:
+  case Op_GetAndAddI:
+  case Op_GetAndAddL:
+  case Op_GetAndSetI:
+  case Op_GetAndSetL:
+  case Op_GetAndSetP:
+  case Op_GetAndSetN:
   case Op_StoreP:
   case Op_StoreN:
   case Op_LoadB:
@@ -2337,7 +2345,7 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
         if (nn != NULL) {
           // Decode a narrow oop to match address
           // [R12 + narrow_oop_reg<<3 + offset]
-          nn = new (C,  2) DecodeNNode(nn, t);
+          nn = new (C) DecodeNNode(nn, t);
           n->set_req(AddPNode::Base, nn);
           n->set_req(AddPNode::Address, nn);
           if (addp->outcnt() == 0) {
@@ -2455,7 +2463,7 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
         }
       }
       if (new_in2 != NULL) {
-        Node* cmpN = new (C, 3) CmpNNode(in1->in(1), new_in2);
+        Node* cmpN = new (C) CmpNNode(in1->in(1), new_in2);
         n->subsume_by( cmpN );
         if (in1->outcnt() == 0) {
           in1->disconnect_inputs(NULL);
@@ -2551,8 +2559,8 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
           n->subsume_by(divmod->mod_proj());
         } else {
           // replace a%b with a-((a/b)*b)
-          Node* mult = new (C, 3) MulINode(d, d->in(2));
-          Node* sub  = new (C, 3) SubINode(d->in(1), mult);
+          Node* mult = new (C) MulINode(d, d->in(2));
+          Node* sub  = new (C) SubINode(d->in(1), mult);
           n->subsume_by( sub );
         }
       }
@@ -2572,8 +2580,8 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
           n->subsume_by(divmod->mod_proj());
         } else {
           // replace a%b with a-((a/b)*b)
-          Node* mult = new (C, 3) MulLNode(d, d->in(2));
-          Node* sub  = new (C, 3) SubLNode(d->in(1), mult);
+          Node* mult = new (C) MulLNode(d, d->in(2));
+          Node* sub  = new (C) SubLNode(d->in(1), mult);
           n->subsume_by( sub );
         }
       }
@@ -2624,7 +2632,7 @@ static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) {
       } else {
         if (t == NULL || t->_lo < 0 || t->_hi > (int)mask) {
           Compile* C = Compile::current();
-          Node* shift = new (C, 3) AndINode(in2, ConNode::make(C, TypeInt::make(mask)));
+          Node* shift = new (C) AndINode(in2, ConNode::make(C, TypeInt::make(mask)));
           n->set_req(2, shift);
         }
       }
diff --git a/src/share/vm/opto/compile.hpp b/src/share/vm/opto/compile.hpp
index a594b0913..6bcf3d3da 100644
--- a/src/share/vm/opto/compile.hpp
+++ b/src/share/vm/opto/compile.hpp
@@ -279,6 +279,7 @@ class Compile : public Phase {
   bool                  _has_split_ifs;         // True if the method _may_ have some split-if
   bool                  _has_unsafe_access;     // True if the method _may_ produce faults in unsafe loads or stores.
   bool                  _has_stringbuilder;     // True StringBuffers or StringBuilders are allocated
+  int                   _max_vector_size;       // Maximum size of generated vectors
   uint                  _trap_hist[trapHistLength];  // Cumulative traps
   bool                  _trap_can_recompile;    // Have we emitted a recompiling trap?
   uint                  _decompile_count;       // Cumulative decompilation counts.
@@ -443,6 +444,8 @@ class Compile : public Phase {
   void          set_has_unsafe_access(bool z)   { _has_unsafe_access = z; }
   bool              has_stringbuilder() const   { return _has_stringbuilder; }
   void          set_has_stringbuilder(bool z)   { _has_stringbuilder = z; }
+  int               max_vector_size() const     { return _max_vector_size; }
+  void          set_max_vector_size(int s)      { _max_vector_size = s; }
   void          set_trap_count(uint r, uint c)  { assert(r < trapHistLength, "oob");        _trap_hist[r] = c; }
   uint              trap_count(uint r) const    { assert(r < trapHistLength, "oob"); return _trap_hist[r]; }
   bool              trap_can_recompile() const  { return _trap_can_recompile; }
diff --git a/src/share/vm/opto/connode.cpp b/src/share/vm/opto/connode.cpp
index e038ccbc2..a2da3ee46 100644
--- a/src/share/vm/opto/connode.cpp
+++ b/src/share/vm/opto/connode.cpp
@@ -45,16 +45,16 @@ uint ConNode::hash() const {
 //------------------------------make-------------------------------------------
 ConNode *ConNode::make( Compile* C, const Type *t ) {
   switch( t->basic_type() ) {
-  case T_INT:       return new (C, 1) ConINode( t->is_int() );
-  case T_LONG:      return new (C, 1) ConLNode( t->is_long() );
-  case T_FLOAT:     return new (C, 1) ConFNode( t->is_float_constant() );
-  case T_DOUBLE:    return new (C, 1) ConDNode( t->is_double_constant() );
-  case T_VOID:      return new (C, 1) ConNode ( Type::TOP );
-  case T_OBJECT:    return new (C, 1) ConPNode( t->is_ptr() );
-  case T_ARRAY:     return new (C, 1) ConPNode( t->is_aryptr() );
-  case T_ADDRESS:   return new (C, 1) ConPNode( t->is_ptr() );
-  case T_NARROWOOP: return new (C, 1) ConNNode( t->is_narrowoop() );
-  case T_METADATA:  return new (C, 1) ConPNode( t->is_ptr() );
+  case T_INT:       return new (C) ConINode( t->is_int() );
+  case T_LONG:      return new (C) ConLNode( t->is_long() );
+  case T_FLOAT:     return new (C) ConFNode( t->is_float_constant() );
+  case T_DOUBLE:    return new (C) ConDNode( t->is_double_constant() );
+  case T_VOID:      return new (C) ConNode ( Type::TOP );
+  case T_OBJECT:    return new (C) ConPNode( t->is_ptr() );
+  case T_ARRAY:     return new (C) ConPNode( t->is_aryptr() );
+  case T_ADDRESS:   return new (C) ConPNode( t->is_ptr() );
+  case T_NARROWOOP: return new (C) ConNNode( t->is_narrowoop() );
+  case T_METADATA:  return new (C) ConPNode( t->is_ptr() );
     // Expected cases:  TypePtr::NULL_PTR, any is_rawptr()
     // Also seen: AnyPtr(TopPTR *+top); from command line:
     //   r -XX:+PrintOpto -XX:CIStart=285 -XX:+CompileTheWorld -XX:CompileTheWorldStartAt=660
@@ -195,13 +195,13 @@ const Type *CMoveNode::Value( PhaseTransform *phase ) const {
 // from the inputs we do not need to specify it here.
 CMoveNode *CMoveNode::make( Compile *C, Node *c, Node *bol, Node *left, Node *right, const Type *t ) {
   switch( t->basic_type() ) {
-  case T_INT:     return new (C, 4) CMoveINode( bol, left, right, t->is_int() );
-  case T_FLOAT:   return new (C, 4) CMoveFNode( bol, left, right, t );
-  case T_DOUBLE:  return new (C, 4) CMoveDNode( bol, left, right, t );
-  case T_LONG:    return new (C, 4) CMoveLNode( bol, left, right, t->is_long() );
-  case T_OBJECT:  return new (C, 4) CMovePNode( c, bol, left, right, t->is_oopptr() );
-  case T_ADDRESS: return new (C, 4) CMovePNode( c, bol, left, right, t->is_ptr() );
-  case T_NARROWOOP: return new (C, 4) CMoveNNode( c, bol, left, right, t );
+  case T_INT:     return new (C) CMoveINode( bol, left, right, t->is_int() );
+  case T_FLOAT:   return new (C) CMoveFNode( bol, left, right, t );
+  case T_DOUBLE:  return new (C) CMoveDNode( bol, left, right, t );
+  case T_LONG:    return new (C) CMoveLNode( bol, left, right, t->is_long() );
+  case T_OBJECT:  return new (C) CMovePNode( c, bol, left, right, t->is_oopptr() );
+  case T_ADDRESS: return new (C) CMovePNode( c, bol, left, right, t->is_ptr() );
+  case T_NARROWOOP: return new (C) CMoveNNode( c, bol, left, right, t );
   default:
     ShouldNotReachHere();
     return NULL;
@@ -268,9 +268,9 @@ Node *CMoveINode::Ideal(PhaseGVN *phase, bool can_reshape) {
 #ifndef PRODUCT
   if( PrintOpto ) tty->print_cr("CMOV to I2B");
 #endif
-  Node *n = new (phase->C, 2) Conv2BNode( cmp->in(1) );
+  Node *n = new (phase->C) Conv2BNode( cmp->in(1) );
   if( flip )
-    n = new (phase->C, 3) XorINode( phase->transform(n), phase->intcon(1) );
+    n = new (phase->C) XorINode( phase->transform(n), phase->intcon(1) );
 
   return n;
 }
@@ -324,9 +324,9 @@ Node *CMoveFNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       sub->in(2) != X ||
       phase->type(sub->in(1)) != TypeF::ZERO ) return NULL;
 
-  Node *abs = new (phase->C, 2) AbsFNode( X );
+  Node *abs = new (phase->C) AbsFNode( X );
   if( flip )
-    abs = new (phase->C, 3) SubFNode(sub->in(1), phase->transform(abs));
+    abs = new (phase->C) SubFNode(sub->in(1), phase->transform(abs));
 
   return abs;
 }
@@ -380,9 +380,9 @@ Node *CMoveDNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       sub->in(2) != X ||
       phase->type(sub->in(1)) != TypeD::ZERO ) return NULL;
 
-  Node *abs = new (phase->C, 2) AbsDNode( X );
+  Node *abs = new (phase->C) AbsDNode( X );
   if( flip )
-    abs = new (phase->C, 3) SubDNode(sub->in(1), phase->transform(abs));
+    abs = new (phase->C) SubDNode(sub->in(1), phase->transform(abs));
 
   return abs;
 }
@@ -480,7 +480,9 @@ static bool can_cause_alias(Node *n, PhaseTransform *phase) {
         opc == Op_CheckCastPP ||
         opc == Op_StorePConditional ||
         opc == Op_CompareAndSwapP ||
-        opc == Op_CompareAndSwapN;
+        opc == Op_CompareAndSwapN ||
+        opc == Op_GetAndSetP ||
+        opc == Op_GetAndSetN;
   }
   return possible_alias;
 }
@@ -959,11 +961,11 @@ Node *ConvI2LNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       ryhi = -rylo0;
     }
 
-    Node* cx = phase->transform( new (phase->C, 2) ConvI2LNode(x, TypeLong::make(rxlo, rxhi, widen)) );
-    Node* cy = phase->transform( new (phase->C, 2) ConvI2LNode(y, TypeLong::make(rylo, ryhi, widen)) );
+    Node* cx = phase->transform( new (phase->C) ConvI2LNode(x, TypeLong::make(rxlo, rxhi, widen)) );
+    Node* cy = phase->transform( new (phase->C) ConvI2LNode(y, TypeLong::make(rylo, ryhi, widen)) );
     switch (op) {
-    case Op_AddI:  return new (phase->C, 3) AddLNode(cx, cy);
-    case Op_SubI:  return new (phase->C, 3) SubLNode(cx, cy);
+    case Op_AddI:  return new (phase->C) AddLNode(cx, cy);
+    case Op_SubI:  return new (phase->C) SubLNode(cx, cy);
     default:       ShouldNotReachHere();
     }
   }
@@ -1037,9 +1039,9 @@ Node *ConvL2INode::Ideal(PhaseGVN *phase, bool can_reshape) {
     assert( x != andl && y != andl, "dead loop in ConvL2INode::Ideal" );
     if (phase->type(x) == Type::TOP)  return NULL;
     if (phase->type(y) == Type::TOP)  return NULL;
-    Node *add1 = phase->transform(new (phase->C, 2) ConvL2INode(x));
-    Node *add2 = phase->transform(new (phase->C, 2) ConvL2INode(y));
-    return new (phase->C, 3) AddINode(add1,add2);
+    Node *add1 = phase->transform(new (phase->C) ConvL2INode(x));
+    Node *add2 = phase->transform(new (phase->C) ConvL2INode(y));
+    return new (phase->C) AddINode(add1,add2);
   }
 
   // Disable optimization: LoadL->ConvL2I ==> LoadI.
@@ -1076,10 +1078,10 @@ static inline Node* addP_of_X2P(PhaseGVN *phase,
                                 Node* dispX,
                                 bool negate = false) {
   if (negate) {
-    dispX = new (phase->C, 3) SubXNode(phase->MakeConX(0), phase->transform(dispX));
+    dispX = new (phase->C) SubXNode(phase->MakeConX(0), phase->transform(dispX));
   }
-  return new (phase->C, 4) AddPNode(phase->C->top(),
-                          phase->transform(new (phase->C, 2) CastX2PNode(base)),
+  return new (phase->C) AddPNode(phase->C->top(),
+                          phase->transform(new (phase->C) CastX2PNode(base)),
                           phase->transform(dispX));
 }
 
diff --git a/src/share/vm/opto/connode.hpp b/src/share/vm/opto/connode.hpp
index 78a4427f2..2f8c79aac 100644
--- a/src/share/vm/opto/connode.hpp
+++ b/src/share/vm/opto/connode.hpp
@@ -58,7 +58,7 @@ public:
 
   // Factory method:
   static ConINode* make( Compile* C, int con ) {
-    return new (C, 1) ConINode( TypeInt::make(con) );
+    return new (C) ConINode( TypeInt::make(con) );
   }
 
 };
@@ -73,9 +73,9 @@ public:
   // Factory methods:
   static ConPNode* make( Compile *C ,address con ) {
     if (con == NULL)
-      return new (C, 1) ConPNode( TypePtr::NULL_PTR ) ;
+      return new (C) ConPNode( TypePtr::NULL_PTR ) ;
     else
-      return new (C, 1) ConPNode( TypeRawPtr::make(con) );
+      return new (C) ConPNode( TypeRawPtr::make(con) );
   }
 };
 
@@ -98,7 +98,7 @@ public:
 
   // Factory method:
   static ConLNode* make( Compile *C ,jlong con ) {
-    return new (C, 1) ConLNode( TypeLong::make(con) );
+    return new (C) ConLNode( TypeLong::make(con) );
   }
 
 };
@@ -112,7 +112,7 @@ public:
 
   // Factory method:
   static ConFNode* make( Compile *C, float con  ) {
-    return new (C, 1) ConFNode( TypeF::make(con) );
+    return new (C) ConFNode( TypeF::make(con) );
   }
 
 };
@@ -126,7 +126,7 @@ public:
 
   // Factory method:
   static ConDNode* make( Compile *C, double con ) {
-    return new (C, 1) ConDNode( TypeD::make(con) );
+    return new (C) ConDNode( TypeD::make(con) );
   }
 
 };
diff --git a/src/share/vm/opto/divnode.cpp b/src/share/vm/opto/divnode.cpp
index 14fa0790b..c22241242 100644
--- a/src/share/vm/opto/divnode.cpp
+++ b/src/share/vm/opto/divnode.cpp
@@ -104,7 +104,7 @@ static Node *transform_int_divide( PhaseGVN *phase, Node *dividend, jint divisor
     // division by +/- 1
     if (!d_pos) {
       // Just negate the value
-      q = new (phase->C, 3) SubINode(phase->intcon(0), dividend);
+      q = new (phase->C) SubINode(phase->intcon(0), dividend);
     }
   } else if ( is_power_of_2(d) ) {
     // division by +/- a power of 2
@@ -141,18 +141,18 @@ static Node *transform_int_divide( PhaseGVN *phase, Node *dividend, jint divisor
       // (-2+3)>>2 becomes 0, etc.
 
       // Compute 0 or -1, based on sign bit
-      Node *sign = phase->transform(new (phase->C, 3) RShiftINode(dividend, phase->intcon(N - 1)));
+      Node *sign = phase->transform(new (phase->C) RShiftINode(dividend, phase->intcon(N - 1)));
       // Mask sign bit to the low sign bits
-      Node *round = phase->transform(new (phase->C, 3) URShiftINode(sign, phase->intcon(N - l)));
+      Node *round = phase->transform(new (phase->C) URShiftINode(sign, phase->intcon(N - l)));
       // Round up before shifting
-      dividend = phase->transform(new (phase->C, 3) AddINode(dividend, round));
+      dividend = phase->transform(new (phase->C) AddINode(dividend, round));
     }
 
     // Shift for division
-    q = new (phase->C, 3) RShiftINode(dividend, phase->intcon(l));
+    q = new (phase->C) RShiftINode(dividend, phase->intcon(l));
 
     if (!d_pos) {
-      q = new (phase->C, 3) SubINode(phase->intcon(0), phase->transform(q));
+      q = new (phase->C) SubINode(phase->intcon(0), phase->transform(q));
     }
   } else {
     // Attempt the jint constant divide -> multiply transform found in
@@ -164,33 +164,33 @@ static Node *transform_int_divide( PhaseGVN *phase, Node *dividend, jint divisor
     jint shift_const;
     if (magic_int_divide_constants(d, magic_const, shift_const)) {
       Node *magic = phase->longcon(magic_const);
-      Node *dividend_long = phase->transform(new (phase->C, 2) ConvI2LNode(dividend));
+      Node *dividend_long = phase->transform(new (phase->C) ConvI2LNode(dividend));
 
       // Compute the high half of the dividend x magic multiplication
-      Node *mul_hi = phase->transform(new (phase->C, 3) MulLNode(dividend_long, magic));
+      Node *mul_hi = phase->transform(new (phase->C) MulLNode(dividend_long, magic));
 
       if (magic_const < 0) {
-        mul_hi = phase->transform(new (phase->C, 3) RShiftLNode(mul_hi, phase->intcon(N)));
-        mul_hi = phase->transform(new (phase->C, 2) ConvL2INode(mul_hi));
+        mul_hi = phase->transform(new (phase->C) RShiftLNode(mul_hi, phase->intcon(N)));
+        mul_hi = phase->transform(new (phase->C) ConvL2INode(mul_hi));
 
         // The magic multiplier is too large for a 32 bit constant. We've adjusted
         // it down by 2^32, but have to add 1 dividend back in after the multiplication.
         // This handles the "overflow" case described by Granlund and Montgomery.
-        mul_hi = phase->transform(new (phase->C, 3) AddINode(dividend, mul_hi));
+        mul_hi = phase->transform(new (phase->C) AddINode(dividend, mul_hi));
 
         // Shift over the (adjusted) mulhi
         if (shift_const != 0) {
-          mul_hi = phase->transform(new (phase->C, 3) RShiftINode(mul_hi, phase->intcon(shift_const)));
+          mul_hi = phase->transform(new (phase->C) RShiftINode(mul_hi, phase->intcon(shift_const)));
         }
       } else {
         // No add is required, we can merge the shifts together.
-        mul_hi = phase->transform(new (phase->C, 3) RShiftLNode(mul_hi, phase->intcon(N + shift_const)));
-        mul_hi = phase->transform(new (phase->C, 2) ConvL2INode(mul_hi));
+        mul_hi = phase->transform(new (phase->C) RShiftLNode(mul_hi, phase->intcon(N + shift_const)));
+        mul_hi = phase->transform(new (phase->C) ConvL2INode(mul_hi));
       }
 
       // Get a 0 or -1 from the sign of the dividend.
       Node *addend0 = mul_hi;
-      Node *addend1 = phase->transform(new (phase->C, 3) RShiftINode(dividend, phase->intcon(N-1)));
+      Node *addend1 = phase->transform(new (phase->C) RShiftINode(dividend, phase->intcon(N-1)));
 
       // If the divisor is negative, swap the order of the input addends;
       // this has the effect of negating the quotient.
@@ -200,7 +200,7 @@ static Node *transform_int_divide( PhaseGVN *phase, Node *dividend, jint divisor
 
       // Adjust the final quotient by subtracting -1 (adding 1)
       // from the mul_hi.
-      q = new (phase->C, 3) SubINode(addend0, addend1);
+      q = new (phase->C) SubINode(addend0, addend1);
     }
   }
 
@@ -259,7 +259,7 @@ static Node* long_by_long_mulhi(PhaseGVN* phase, Node* dividend, jlong magic_con
   // no need to synthesize it in ideal nodes.
   if (Matcher::has_match_rule(Op_MulHiL)) {
     Node* v = phase->longcon(magic_const);
-    return new (phase->C, 3) MulHiLNode(dividend, v);
+    return new (phase->C) MulHiLNode(dividend, v);
   }
 
   // Taken from Hacker's Delight, Fig. 8-2. Multiply high signed.
@@ -285,11 +285,11 @@ static Node* long_by_long_mulhi(PhaseGVN* phase, Node* dividend, jlong magic_con
   const int N = 64;
 
   // Dummy node to keep intermediate nodes alive during construction
-  Node* hook = new (phase->C, 4) Node(4);
+  Node* hook = new (phase->C) Node(4);
 
   // u0 = u & 0xFFFFFFFF;  u1 = u >> 32;
-  Node* u0 = phase->transform(new (phase->C, 3) AndLNode(dividend, phase->longcon(0xFFFFFFFF)));
-  Node* u1 = phase->transform(new (phase->C, 3) RShiftLNode(dividend, phase->intcon(N / 2)));
+  Node* u0 = phase->transform(new (phase->C) AndLNode(dividend, phase->longcon(0xFFFFFFFF)));
+  Node* u1 = phase->transform(new (phase->C) RShiftLNode(dividend, phase->intcon(N / 2)));
   hook->init_req(0, u0);
   hook->init_req(1, u1);
 
@@ -298,29 +298,29 @@ static Node* long_by_long_mulhi(PhaseGVN* phase, Node* dividend, jlong magic_con
   Node* v1 = phase->longcon(magic_const >> (N / 2));
 
   // w0 = u0*v0;
-  Node* w0 = phase->transform(new (phase->C, 3) MulLNode(u0, v0));
+  Node* w0 = phase->transform(new (phase->C) MulLNode(u0, v0));
 
   // t = u1*v0 + (w0 >> 32);
-  Node* u1v0 = phase->transform(new (phase->C, 3) MulLNode(u1, v0));
-  Node* temp = phase->transform(new (phase->C, 3) URShiftLNode(w0, phase->intcon(N / 2)));
-  Node* t    = phase->transform(new (phase->C, 3) AddLNode(u1v0, temp));
+  Node* u1v0 = phase->transform(new (phase->C) MulLNode(u1, v0));
+  Node* temp = phase->transform(new (phase->C) URShiftLNode(w0, phase->intcon(N / 2)));
+  Node* t    = phase->transform(new (phase->C) AddLNode(u1v0, temp));
   hook->init_req(2, t);
 
   // w1 = t & 0xFFFFFFFF;
-  Node* w1 = phase->transform(new (phase->C, 3) AndLNode(t, phase->longcon(0xFFFFFFFF)));
+  Node* w1 = phase->transform(new (phase->C) AndLNode(t, phase->longcon(0xFFFFFFFF)));
   hook->init_req(3, w1);
 
   // w2 = t >> 32;
-  Node* w2 = phase->transform(new (phase->C, 3) RShiftLNode(t, phase->intcon(N / 2)));
+  Node* w2 = phase->transform(new (phase->C) RShiftLNode(t, phase->intcon(N / 2)));
 
   // w1 = u0*v1 + w1;
-  Node* u0v1 = phase->transform(new (phase->C, 3) MulLNode(u0, v1));
-  w1         = phase->transform(new (phase->C, 3) AddLNode(u0v1, w1));
+  Node* u0v1 = phase->transform(new (phase->C) MulLNode(u0, v1));
+  w1         = phase->transform(new (phase->C) AddLNode(u0v1, w1));
 
   // return u1*v1 + w2 + (w1 >> 32);
-  Node* u1v1  = phase->transform(new (phase->C, 3) MulLNode(u1, v1));
-  Node* temp1 = phase->transform(new (phase->C, 3) AddLNode(u1v1, w2));
-  Node* temp2 = phase->transform(new (phase->C, 3) RShiftLNode(w1, phase->intcon(N / 2)));
+  Node* u1v1  = phase->transform(new (phase->C) MulLNode(u1, v1));
+  Node* temp1 = phase->transform(new (phase->C) AddLNode(u1v1, w2));
+  Node* temp2 = phase->transform(new (phase->C) RShiftLNode(w1, phase->intcon(N / 2)));
 
   // Remove the bogus extra edges used to keep things alive
   PhaseIterGVN* igvn = phase->is_IterGVN();
@@ -332,7 +332,7 @@ static Node* long_by_long_mulhi(PhaseGVN* phase, Node* dividend, jlong magic_con
     }
   }
 
-  return new (phase->C, 3) AddLNode(temp1, temp2);
+  return new (phase->C) AddLNode(temp1, temp2);
 }
 
 
@@ -355,7 +355,7 @@ static Node *transform_long_divide( PhaseGVN *phase, Node *dividend, jlong divis
     // division by +/- 1
     if (!d_pos) {
       // Just negate the value
-      q = new (phase->C, 3) SubLNode(phase->longcon(0), dividend);
+      q = new (phase->C) SubLNode(phase->longcon(0), dividend);
     }
   } else if ( is_power_of_2_long(d) ) {
 
@@ -394,18 +394,18 @@ static Node *transform_long_divide( PhaseGVN *phase, Node *dividend, jlong divis
       // (-2+3)>>2 becomes 0, etc.
 
       // Compute 0 or -1, based on sign bit
-      Node *sign = phase->transform(new (phase->C, 3) RShiftLNode(dividend, phase->intcon(N - 1)));
+      Node *sign = phase->transform(new (phase->C) RShiftLNode(dividend, phase->intcon(N - 1)));
       // Mask sign bit to the low sign bits
-      Node *round = phase->transform(new (phase->C, 3) URShiftLNode(sign, phase->intcon(N - l)));
+      Node *round = phase->transform(new (phase->C) URShiftLNode(sign, phase->intcon(N - l)));
       // Round up before shifting
-      dividend = phase->transform(new (phase->C, 3) AddLNode(dividend, round));
+      dividend = phase->transform(new (phase->C) AddLNode(dividend, round));
     }
 
     // Shift for division
-    q = new (phase->C, 3) RShiftLNode(dividend, phase->intcon(l));
+    q = new (phase->C) RShiftLNode(dividend, phase->intcon(l));
 
     if (!d_pos) {
-      q = new (phase->C, 3) SubLNode(phase->longcon(0), phase->transform(q));
+      q = new (phase->C) SubLNode(phase->longcon(0), phase->transform(q));
     }
   } else if ( !Matcher::use_asm_for_ldiv_by_con(d) ) { // Use hardware DIV instruction when
                                                        // it is faster than code generated below.
@@ -425,17 +425,17 @@ static Node *transform_long_divide( PhaseGVN *phase, Node *dividend, jlong divis
         // The magic multiplier is too large for a 64 bit constant. We've adjusted
         // it down by 2^64, but have to add 1 dividend back in after the multiplication.
         // This handles the "overflow" case described by Granlund and Montgomery.
-        mul_hi = phase->transform(new (phase->C, 3) AddLNode(dividend, mul_hi));
+        mul_hi = phase->transform(new (phase->C) AddLNode(dividend, mul_hi));
       }
 
       // Shift over the (adjusted) mulhi
       if (shift_const != 0) {
-        mul_hi = phase->transform(new (phase->C, 3) RShiftLNode(mul_hi, phase->intcon(shift_const)));
+        mul_hi = phase->transform(new (phase->C) RShiftLNode(mul_hi, phase->intcon(shift_const)));
       }
 
       // Get a 0 or -1 from the sign of the dividend.
       Node *addend0 = mul_hi;
-      Node *addend1 = phase->transform(new (phase->C, 3) RShiftLNode(dividend, phase->intcon(N-1)));
+      Node *addend1 = phase->transform(new (phase->C) RShiftLNode(dividend, phase->intcon(N-1)));
 
       // If the divisor is negative, swap the order of the input addends;
       // this has the effect of negating the quotient.
@@ -445,7 +445,7 @@ static Node *transform_long_divide( PhaseGVN *phase, Node *dividend, jlong divis
 
       // Adjust the final quotient by subtracting -1 (adding 1)
       // from the mul_hi.
-      q = new (phase->C, 3) SubLNode(addend0, addend1);
+      q = new (phase->C) SubLNode(addend0, addend1);
     }
   }
 
@@ -735,7 +735,7 @@ Node *DivFNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   assert( frexp((double)reciprocal, &exp) == 0.5, "reciprocal should be power of 2" );
 
   // return multiplication by the reciprocal
-  return (new (phase->C, 3) MulFNode(in(1), phase->makecon(TypeF::make(reciprocal))));
+  return (new (phase->C) MulFNode(in(1), phase->makecon(TypeF::make(reciprocal))));
 }
 
 //=============================================================================
@@ -829,7 +829,7 @@ Node *DivDNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   assert( frexp(reciprocal, &exp) == 0.5, "reciprocal should be power of 2" );
 
   // return multiplication by the reciprocal
-  return (new (phase->C, 3) MulDNode(in(1), phase->makecon(TypeD::make(reciprocal))));
+  return (new (phase->C) MulDNode(in(1), phase->makecon(TypeD::make(reciprocal))));
 }
 
 //=============================================================================
@@ -856,7 +856,7 @@ Node *ModINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if( !ti->is_con() ) return NULL;
   jint con = ti->get_con();
 
-  Node *hook = new (phase->C, 1) Node(1);
+  Node *hook = new (phase->C) Node(1);
 
   // First, special check for modulo 2^k-1
   if( con >= 0 && con < max_jint && is_power_of_2(con+1) ) {
@@ -876,24 +876,24 @@ Node *ModINode::Ideal(PhaseGVN *phase, bool can_reshape) {
       hook->init_req(0, x);       // Add a use to x to prevent him from dying
       // Generate code to reduce X rapidly to nearly 2^k-1.
       for( int i = 0; i < trip_count; i++ ) {
-        Node *xl = phase->transform( new (phase->C, 3) AndINode(x,divisor) );
-        Node *xh = phase->transform( new (phase->C, 3) RShiftINode(x,phase->intcon(k)) ); // Must be signed
-        x = phase->transform( new (phase->C, 3) AddINode(xh,xl) );
+        Node *xl = phase->transform( new (phase->C) AndINode(x,divisor) );
+        Node *xh = phase->transform( new (phase->C) RShiftINode(x,phase->intcon(k)) ); // Must be signed
+        x = phase->transform( new (phase->C) AddINode(xh,xl) );
         hook->set_req(0, x);
       }
 
       // Generate sign-fixup code.  Was original value positive?
       // int hack_res = (i >= 0) ? divisor : 1;
-      Node *cmp1 = phase->transform( new (phase->C, 3) CmpINode( in(1), phase->intcon(0) ) );
-      Node *bol1 = phase->transform( new (phase->C, 2) BoolNode( cmp1, BoolTest::ge ) );
-      Node *cmov1= phase->transform( new (phase->C, 4) CMoveINode(bol1, phase->intcon(1), divisor, TypeInt::POS) );
+      Node *cmp1 = phase->transform( new (phase->C) CmpINode( in(1), phase->intcon(0) ) );
+      Node *bol1 = phase->transform( new (phase->C) BoolNode( cmp1, BoolTest::ge ) );
+      Node *cmov1= phase->transform( new (phase->C) CMoveINode(bol1, phase->intcon(1), divisor, TypeInt::POS) );
       // if( x >= hack_res ) x -= divisor;
-      Node *sub  = phase->transform( new (phase->C, 3) SubINode( x, divisor ) );
-      Node *cmp2 = phase->transform( new (phase->C, 3) CmpINode( x, cmov1 ) );
-      Node *bol2 = phase->transform( new (phase->C, 2) BoolNode( cmp2, BoolTest::ge ) );
+      Node *sub  = phase->transform( new (phase->C) SubINode( x, divisor ) );
+      Node *cmp2 = phase->transform( new (phase->C) CmpINode( x, cmov1 ) );
+      Node *bol2 = phase->transform( new (phase->C) BoolNode( cmp2, BoolTest::ge ) );
       // Convention is to not transform the return value of an Ideal
       // since Ideal is expected to return a modified 'this' or a new node.
-      Node *cmov2= new (phase->C, 4) CMoveINode(bol2, x, sub, TypeInt::INT);
+      Node *cmov2= new (phase->C) CMoveINode(bol2, x, sub, TypeInt::INT);
       // cmov2 is now the mod
 
       // Now remove the bogus extra edges used to keep things alive
@@ -916,7 +916,7 @@ Node *ModINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   jint pos_con = (con >= 0) ? con : -con;
 
   // integer Mod 1 is always 0
-  if( pos_con == 1 ) return new (phase->C, 1) ConINode(TypeInt::ZERO);
+  if( pos_con == 1 ) return new (phase->C) ConINode(TypeInt::ZERO);
 
   int log2_con = -1;
 
@@ -929,7 +929,7 @@ Node *ModINode::Ideal(PhaseGVN *phase, bool can_reshape) {
 
     // See if this can be masked, if the dividend is non-negative
     if( dti && dti->_lo >= 0 )
-      return ( new (phase->C, 3) AndINode( in(1), phase->intcon( pos_con-1 ) ) );
+      return ( new (phase->C) AndINode( in(1), phase->intcon( pos_con-1 ) ) );
   }
 
   // Save in(1) so that it cannot be changed or deleted
@@ -944,12 +944,12 @@ Node *ModINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *mult = NULL;
 
     if( log2_con >= 0 )
-      mult = phase->transform( new (phase->C, 3) LShiftINode( divide, phase->intcon( log2_con ) ) );
+      mult = phase->transform( new (phase->C) LShiftINode( divide, phase->intcon( log2_con ) ) );
     else
-      mult = phase->transform( new (phase->C, 3) MulINode( divide, phase->intcon( pos_con ) ) );
+      mult = phase->transform( new (phase->C) MulINode( divide, phase->intcon( pos_con ) ) );
 
     // Finally, subtract the multiplied divided value from the original
-    result = new (phase->C, 3) SubINode( in(1), mult );
+    result = new (phase->C) SubINode( in(1), mult );
   }
 
   // Now remove the bogus extra edges used to keep things alive
@@ -1027,7 +1027,7 @@ Node *ModLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if( !tl->is_con() ) return NULL;
   jlong con = tl->get_con();
 
-  Node *hook = new (phase->C, 1) Node(1);
+  Node *hook = new (phase->C) Node(1);
 
   // Expand mod
   if( con >= 0 && con < max_jlong && is_power_of_2_long(con+1) ) {
@@ -1049,24 +1049,24 @@ Node *ModLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       hook->init_req(0, x);       // Add a use to x to prevent him from dying
       // Generate code to reduce X rapidly to nearly 2^k-1.
       for( int i = 0; i < trip_count; i++ ) {
-        Node *xl = phase->transform( new (phase->C, 3) AndLNode(x,divisor) );
-        Node *xh = phase->transform( new (phase->C, 3) RShiftLNode(x,phase->intcon(k)) ); // Must be signed
-        x = phase->transform( new (phase->C, 3) AddLNode(xh,xl) );
+        Node *xl = phase->transform( new (phase->C) AndLNode(x,divisor) );
+        Node *xh = phase->transform( new (phase->C) RShiftLNode(x,phase->intcon(k)) ); // Must be signed
+        x = phase->transform( new (phase->C) AddLNode(xh,xl) );
         hook->set_req(0, x);    // Add a use to x to prevent him from dying
       }
 
       // Generate sign-fixup code.  Was original value positive?
       // long hack_res = (i >= 0) ? divisor : CONST64(1);
-      Node *cmp1 = phase->transform( new (phase->C, 3) CmpLNode( in(1), phase->longcon(0) ) );
-      Node *bol1 = phase->transform( new (phase->C, 2) BoolNode( cmp1, BoolTest::ge ) );
-      Node *cmov1= phase->transform( new (phase->C, 4) CMoveLNode(bol1, phase->longcon(1), divisor, TypeLong::LONG) );
+      Node *cmp1 = phase->transform( new (phase->C) CmpLNode( in(1), phase->longcon(0) ) );
+      Node *bol1 = phase->transform( new (phase->C) BoolNode( cmp1, BoolTest::ge ) );
+      Node *cmov1= phase->transform( new (phase->C) CMoveLNode(bol1, phase->longcon(1), divisor, TypeLong::LONG) );
       // if( x >= hack_res ) x -= divisor;
-      Node *sub  = phase->transform( new (phase->C, 3) SubLNode( x, divisor ) );
-      Node *cmp2 = phase->transform( new (phase->C, 3) CmpLNode( x, cmov1 ) );
-      Node *bol2 = phase->transform( new (phase->C, 2) BoolNode( cmp2, BoolTest::ge ) );
+      Node *sub  = phase->transform( new (phase->C) SubLNode( x, divisor ) );
+      Node *cmp2 = phase->transform( new (phase->C) CmpLNode( x, cmov1 ) );
+      Node *bol2 = phase->transform( new (phase->C) BoolNode( cmp2, BoolTest::ge ) );
       // Convention is to not transform the return value of an Ideal
       // since Ideal is expected to return a modified 'this' or a new node.
-      Node *cmov2= new (phase->C, 4) CMoveLNode(bol2, x, sub, TypeLong::LONG);
+      Node *cmov2= new (phase->C) CMoveLNode(bol2, x, sub, TypeLong::LONG);
       // cmov2 is now the mod
 
       // Now remove the bogus extra edges used to keep things alive
@@ -1089,7 +1089,7 @@ Node *ModLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   jlong pos_con = (con >= 0) ? con : -con;
 
   // integer Mod 1 is always 0
-  if( pos_con == 1 ) return new (phase->C, 1) ConLNode(TypeLong::ZERO);
+  if( pos_con == 1 ) return new (phase->C) ConLNode(TypeLong::ZERO);
 
   int log2_con = -1;
 
@@ -1102,7 +1102,7 @@ Node *ModLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
 
     // See if this can be masked, if the dividend is non-negative
     if( dtl && dtl->_lo >= 0 )
-      return ( new (phase->C, 3) AndLNode( in(1), phase->longcon( pos_con-1 ) ) );
+      return ( new (phase->C) AndLNode( in(1), phase->longcon( pos_con-1 ) ) );
   }
 
   // Save in(1) so that it cannot be changed or deleted
@@ -1117,12 +1117,12 @@ Node *ModLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *mult = NULL;
 
     if( log2_con >= 0 )
-      mult = phase->transform( new (phase->C, 3) LShiftLNode( divide, phase->intcon( log2_con ) ) );
+      mult = phase->transform( new (phase->C) LShiftLNode( divide, phase->intcon( log2_con ) ) );
     else
-      mult = phase->transform( new (phase->C, 3) MulLNode( divide, phase->longcon( pos_con ) ) );
+      mult = phase->transform( new (phase->C) MulLNode( divide, phase->longcon( pos_con ) ) );
 
     // Finally, subtract the multiplied divided value from the original
-    result = new (phase->C, 3) SubLNode( in(1), mult );
+    result = new (phase->C) SubLNode( in(1), mult );
   }
 
   // Now remove the bogus extra edges used to keep things alive
@@ -1277,9 +1277,9 @@ DivModINode* DivModINode::make(Compile* C, Node* div_or_mod) {
   assert(n->Opcode() == Op_DivI || n->Opcode() == Op_ModI,
          "only div or mod input pattern accepted");
 
-  DivModINode* divmod = new (C, 3) DivModINode(n->in(0), n->in(1), n->in(2));
-  Node*        dproj  = new (C, 1) ProjNode(divmod, DivModNode::div_proj_num);
-  Node*        mproj  = new (C, 1) ProjNode(divmod, DivModNode::mod_proj_num);
+  DivModINode* divmod = new (C) DivModINode(n->in(0), n->in(1), n->in(2));
+  Node*        dproj  = new (C) ProjNode(divmod, DivModNode::div_proj_num);
+  Node*        mproj  = new (C) ProjNode(divmod, DivModNode::mod_proj_num);
   return divmod;
 }
 
@@ -1289,9 +1289,9 @@ DivModLNode* DivModLNode::make(Compile* C, Node* div_or_mod) {
   assert(n->Opcode() == Op_DivL || n->Opcode() == Op_ModL,
          "only div or mod input pattern accepted");
 
-  DivModLNode* divmod = new (C, 3) DivModLNode(n->in(0), n->in(1), n->in(2));
-  Node*        dproj  = new (C, 1) ProjNode(divmod, DivModNode::div_proj_num);
-  Node*        mproj  = new (C, 1) ProjNode(divmod, DivModNode::mod_proj_num);
+  DivModLNode* divmod = new (C) DivModLNode(n->in(0), n->in(1), n->in(2));
+  Node*        dproj  = new (C) ProjNode(divmod, DivModNode::div_proj_num);
+  Node*        mproj  = new (C) ProjNode(divmod, DivModNode::mod_proj_num);
   return divmod;
 }
 
@@ -1306,7 +1306,7 @@ Node *DivModINode::match( const ProjNode *proj, const Matcher *match ) {
     assert(proj->_con == mod_proj_num, "must be div or mod projection");
     rm = match->modI_proj_mask();
   }
-  return new (match->C, 1)MachProjNode(this, proj->_con, rm, ideal_reg);
+  return new (match->C)MachProjNode(this, proj->_con, rm, ideal_reg);
 }
 
 
@@ -1321,5 +1321,5 @@ Node *DivModLNode::match( const ProjNode *proj, const Matcher *match ) {
     assert(proj->_con == mod_proj_num, "must be div or mod projection");
     rm = match->modL_proj_mask();
   }
-  return new (match->C, 1)MachProjNode(this, proj->_con, rm, ideal_reg);
+  return new (match->C)MachProjNode(this, proj->_con, rm, ideal_reg);
 }
diff --git a/src/share/vm/opto/doCall.cpp b/src/share/vm/opto/doCall.cpp
index c8576dc50..30a01f34b 100644
--- a/src/share/vm/opto/doCall.cpp
+++ b/src/share/vm/opto/doCall.cpp
@@ -40,11 +40,10 @@
 #include "prims/nativeLookup.hpp"
 #include "runtime/sharedRuntime.hpp"
 
-#ifndef PRODUCT
 void trace_type_profile(ciMethod *method, int depth, int bci, ciMethod *prof_method, ciKlass *prof_klass, int site_count, int receiver_count) {
-  if (TraceTypeProfile || PrintInlining || PrintOptoInlining) {
+  if (TraceTypeProfile || PrintInlining NOT_PRODUCT(|| PrintOptoInlining)) {
     if (!PrintInlining) {
-      if (!PrintOpto && !PrintCompilation) {
+      if (NOT_PRODUCT(!PrintOpto &&) !PrintCompilation) {
         method->print_short_name();
         tty->cr();
       }
@@ -56,7 +55,6 @@ void trace_type_profile(ciMethod *method, int depth, int bci, ciMethod *prof_met
     tty->cr();
   }
 }
-#endif
 
 CallGenerator* Compile::call_generator(ciMethod* callee, int vtable_index, bool call_is_virtual,
                                        JVMState* jvms, bool allow_inline,
@@ -225,13 +223,13 @@ CallGenerator* Compile::call_generator(ciMethod* callee, int vtable_index, bool
           }
           if (miss_cg != NULL) {
             if (next_hit_cg != NULL) {
-              NOT_PRODUCT(trace_type_profile(jvms->method(), jvms->depth() - 1, jvms->bci(), next_receiver_method, profile.receiver(1), site_count, profile.receiver_count(1)));
+              trace_type_profile(jvms->method(), jvms->depth() - 1, jvms->bci(), next_receiver_method, profile.receiver(1), site_count, profile.receiver_count(1));
               // We don't need to record dependency on a receiver here and below.
               // Whenever we inline, the dependency is added by Parse::Parse().
               miss_cg = CallGenerator::for_predicted_call(profile.receiver(1), miss_cg, next_hit_cg, PROB_MAX);
             }
             if (miss_cg != NULL) {
-              NOT_PRODUCT(trace_type_profile(jvms->method(), jvms->depth() - 1, jvms->bci(), receiver_method, profile.receiver(0), site_count, receiver_count));
+              trace_type_profile(jvms->method(), jvms->depth() - 1, jvms->bci(), receiver_method, profile.receiver(0), site_count, receiver_count);
               CallGenerator* cg = CallGenerator::for_predicted_call(profile.receiver(0), miss_cg, hit_cg, profile.receiver_prob(0));
               if (cg != NULL)  return cg;
             }
@@ -514,15 +512,15 @@ void Parse::do_call() {
         } else if (rt == T_INT || is_subword_type(rt)) {
           // FIXME: This logic should be factored out.
           if (ct == T_BOOLEAN) {
-            retnode = _gvn.transform( new (C, 3) AndINode(retnode, intcon(0x1)) );
+            retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0x1)) );
           } else if (ct == T_CHAR) {
-            retnode = _gvn.transform( new (C, 3) AndINode(retnode, intcon(0xFFFF)) );
+            retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0xFFFF)) );
           } else if (ct == T_BYTE) {
-            retnode = _gvn.transform( new (C, 3) LShiftINode(retnode, intcon(24)) );
-            retnode = _gvn.transform( new (C, 3) RShiftINode(retnode, intcon(24)) );
+            retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(24)) );
+            retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(24)) );
           } else if (ct == T_SHORT) {
-            retnode = _gvn.transform( new (C, 3) LShiftINode(retnode, intcon(16)) );
-            retnode = _gvn.transform( new (C, 3) RShiftINode(retnode, intcon(16)) );
+            retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(16)) );
+            retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(16)) );
           } else {
             assert(ct == T_INT, err_msg_res("rt=%s, ct=%s", type2name(rt), type2name(ct)));
           }
@@ -532,7 +530,7 @@ void Parse::do_call() {
             const TypeOopPtr* arg_type = TypeOopPtr::make_from_klass(rtype->as_klass());
             const Type*       sig_type = TypeOopPtr::make_from_klass(ctype->as_klass());
             if (arg_type != NULL && !arg_type->higher_equal(sig_type)) {
-              Node* cast_obj = _gvn.transform(new (C, 2) CheckCastPPNode(control(), retnode, sig_type));
+              Node* cast_obj = _gvn.transform(new (C) CheckCastPPNode(control(), retnode, sig_type));
               pop();
               push(cast_obj);
             }
@@ -614,7 +612,7 @@ void Parse::catch_call_exceptions(ciExceptionHandlerStream& handlers) {
   }
 
   int len = bcis->length();
-  CatchNode *cn = new (C, 2) CatchNode(control(), i_o, len+1);
+  CatchNode *cn = new (C) CatchNode(control(), i_o, len+1);
   Node *catch_ = _gvn.transform(cn);
 
   // now branch with the exception state to each of the (potential)
@@ -625,14 +623,14 @@ void Parse::catch_call_exceptions(ciExceptionHandlerStream& handlers) {
     // Locals are just copied from before the call.
     // Get control from the CatchNode.
     int handler_bci = bcis->at(i);
-    Node* ctrl = _gvn.transform( new (C, 1) CatchProjNode(catch_, i+1,handler_bci));
+    Node* ctrl = _gvn.transform( new (C) CatchProjNode(catch_, i+1,handler_bci));
     // This handler cannot happen?
     if (ctrl == top())  continue;
     set_control(ctrl);
 
     // Create exception oop
     const TypeInstPtr* extype = extypes->at(i)->is_instptr();
-    Node *ex_oop = _gvn.transform(new (C, 2) CreateExNode(extypes->at(i), ctrl, i_o));
+    Node *ex_oop = _gvn.transform(new (C) CreateExNode(extypes->at(i), ctrl, i_o));
 
     // Handle unloaded exception classes.
     if (saw_unloaded->contains(handler_bci)) {
@@ -671,7 +669,7 @@ void Parse::catch_call_exceptions(ciExceptionHandlerStream& handlers) {
 
   // The first CatchProj is for the normal return.
   // (Note:  If this is a call to rethrow_Java, this node goes dead.)
-  set_control(_gvn.transform( new (C, 1) CatchProjNode(catch_, CatchProjNode::fall_through_index, CatchProjNode::no_handler_bci)));
+  set_control(_gvn.transform( new (C) CatchProjNode(catch_, CatchProjNode::fall_through_index, CatchProjNode::no_handler_bci)));
 }
 
 
@@ -722,7 +720,7 @@ void Parse::catch_inline_exceptions(SafePointNode* ex_map) {
     // I'm loading the class from, I can replace the LoadKlass with the
     // klass constant for the exception oop.
     if( ex_node->is_Phi() ) {
-      ex_klass_node = new (C, ex_node->req()) PhiNode( ex_node->in(0), TypeKlassPtr::OBJECT );
+      ex_klass_node = new (C) PhiNode( ex_node->in(0), TypeKlassPtr::OBJECT );
       for( uint i = 1; i < ex_node->req(); i++ ) {
         Node* p = basic_plus_adr( ex_node->in(i), ex_node->in(i), oopDesc::klass_offset_in_bytes() );
         Node* k = _gvn.transform( LoadKlassNode::make(_gvn, immutable_memory(), p, TypeInstPtr::KLASS, TypeKlassPtr::OBJECT) );
@@ -788,7 +786,7 @@ void Parse::catch_inline_exceptions(SafePointNode* ex_map) {
       PreserveJVMState pjvms(this);
       const TypeInstPtr* tinst = TypeOopPtr::make_from_klass_unique(klass)->cast_to_ptr_type(TypePtr::NotNull)->is_instptr();
       assert(klass->has_subklass() || tinst->klass_is_exact(), "lost exactness");
-      Node* ex_oop = _gvn.transform(new (C, 2) CheckCastPPNode(control(), ex_node, tinst));
+      Node* ex_oop = _gvn.transform(new (C) CheckCastPPNode(control(), ex_node, tinst));
       push_ex_oop(ex_oop);      // Push exception oop for handler
 #ifndef PRODUCT
       if (PrintOpto && WizardMode) {
diff --git a/src/share/vm/opto/escape.cpp b/src/share/vm/opto/escape.cpp
index 49206f7ab..85d40b0d3 100644
--- a/src/share/vm/opto/escape.cpp
+++ b/src/share/vm/opto/escape.cpp
@@ -282,6 +282,26 @@ bool ConnectionGraph::compute_escape() {
   return has_non_escaping_obj;
 }
 
+// Utility function for nodes that load an object
+void ConnectionGraph::add_objload_to_connection_graph(Node *n, Unique_Node_List *delayed_worklist) {
+  // Using isa_ptr() instead of isa_oopptr() for LoadP and Phi because
+  // ThreadLocal has RawPtr type.
+  const Type* t = _igvn->type(n);
+  if (t->make_ptr() != NULL) {
+    Node* adr = n->in(MemNode::Address);
+#ifdef ASSERT
+    if (!adr->is_AddP()) {
+      assert(_igvn->type(adr)->isa_rawptr(), "sanity");
+    } else {
+      assert((ptnode_adr(adr->_idx) == NULL ||
+              ptnode_adr(adr->_idx)->as_Field()->is_oop()), "sanity");
+    }
+#endif
+    add_local_var_and_edge(n, PointsToNode::NoEscape,
+                           adr, delayed_worklist);
+  }
+}
+
 // Populate Connection Graph with PointsTo nodes and create simple
 // connection graph edges.
 void ConnectionGraph::add_node_to_connection_graph(Node *n, Unique_Node_List *delayed_worklist) {
@@ -387,22 +407,7 @@ void ConnectionGraph::add_node_to_connection_graph(Node *n, Unique_Node_List *de
     case Op_LoadP:
     case Op_LoadN:
     case Op_LoadPLocked: {
-      // Using isa_ptr() instead of isa_oopptr() for LoadP and Phi because
-      // ThreadLocal has RawPrt type.
-      const Type* t = igvn->type(n);
-      if (t->make_ptr() != NULL) {
-        Node* adr = n->in(MemNode::Address);
-#ifdef ASSERT
-        if (!adr->is_AddP()) {
-          assert(igvn->type(adr)->isa_rawptr(), "sanity");
-        } else {
-          assert((ptnode_adr(adr->_idx) == NULL ||
-                  ptnode_adr(adr->_idx)->as_Field()->is_oop()), "sanity");
-        }
-#endif
-        add_local_var_and_edge(n, PointsToNode::NoEscape,
-                               adr, delayed_worklist);
-      }
+      add_objload_to_connection_graph(n, delayed_worklist);
       break;
     }
     case Op_Parm: {
@@ -417,7 +422,7 @@ void ConnectionGraph::add_node_to_connection_graph(Node *n, Unique_Node_List *de
     }
     case Op_Phi: {
       // Using isa_ptr() instead of isa_oopptr() for LoadP and Phi because
-      // ThreadLocal has RawPrt type.
+      // ThreadLocal has RawPtr type.
       const Type* t = n->as_Phi()->type();
       if (t->make_ptr() != NULL) {
         add_local_var(n, PointsToNode::NoEscape);
@@ -446,6 +451,11 @@ void ConnectionGraph::add_node_to_connection_graph(Node *n, Unique_Node_List *de
       }
       break;
     }
+    case Op_GetAndSetP:
+    case Op_GetAndSetN: {
+      add_objload_to_connection_graph(n, delayed_worklist);
+      // fallthrough
+    }
     case Op_StoreP:
     case Op_StoreN:
     case Op_StorePConditional:
@@ -585,7 +595,7 @@ void ConnectionGraph::add_final_edges(Node *n) {
     case Op_LoadN:
     case Op_LoadPLocked: {
       // Using isa_ptr() instead of isa_oopptr() for LoadP and Phi because
-      // ThreadLocal has RawPrt type.
+      // ThreadLocal has RawPtr type.
       const Type* t = _igvn->type(n);
       if (t->make_ptr() != NULL) {
         Node* adr = n->in(MemNode::Address);
@@ -596,7 +606,7 @@ void ConnectionGraph::add_final_edges(Node *n) {
     }
     case Op_Phi: {
       // Using isa_ptr() instead of isa_oopptr() for LoadP and Phi because
-      // ThreadLocal has RawPrt type.
+      // ThreadLocal has RawPtr type.
       const Type* t = n->as_Phi()->type();
       if (t->make_ptr() != NULL) {
         for (uint i = 1; i < n->req(); i++) {
@@ -638,8 +648,16 @@ void ConnectionGraph::add_final_edges(Node *n) {
     case Op_StoreN:
     case Op_StorePConditional:
     case Op_CompareAndSwapP:
-    case Op_CompareAndSwapN: {
+    case Op_CompareAndSwapN:
+    case Op_GetAndSetP:
+    case Op_GetAndSetN: {
       Node* adr = n->in(MemNode::Address);
+      if (opcode == Op_GetAndSetP || opcode == Op_GetAndSetN) {
+        const Type* t = _igvn->type(n);
+        if (t->make_ptr() != NULL) {
+          add_local_var_and_edge(n, PointsToNode::NoEscape, adr, NULL);
+        }
+      }
       const Type *adr_type = _igvn->type(adr);
       adr_type = adr_type->make_ptr();
       if (adr_type->isa_oopptr() ||
diff --git a/src/share/vm/opto/escape.hpp b/src/share/vm/opto/escape.hpp
index 54e5f5853..78447f030 100644
--- a/src/share/vm/opto/escape.hpp
+++ b/src/share/vm/opto/escape.hpp
@@ -371,6 +371,8 @@ private:
     _nodes.at_put(n->_idx, ptn);
   }
 
+  // Utility function for nodes that load an object
+  void add_objload_to_connection_graph(Node *n, Unique_Node_List *delayed_worklist);
   // Create PointsToNode node and add it to Connection Graph.
   void add_node_to_connection_graph(Node *n, Unique_Node_List *delayed_worklist);
 
diff --git a/src/share/vm/opto/generateOptoStub.cpp b/src/share/vm/opto/generateOptoStub.cpp
index 31600a749..032f0081d 100644
--- a/src/share/vm/opto/generateOptoStub.cpp
+++ b/src/share/vm/opto/generateOptoStub.cpp
@@ -50,7 +50,7 @@ void GraphKit::gen_stub(address C_function,
   const TypeTuple *jrange  = C->tf()->range();
 
   // The procedure start
-  StartNode* start = new (C, 2) StartNode(root(), jdomain);
+  StartNode* start = new (C) StartNode(root(), jdomain);
   _gvn.set_type_bottom(start);
 
   // Make a map, with JVM state
@@ -63,7 +63,7 @@ void GraphKit::gen_stub(address C_function,
   jvms->set_monoff(max_map);
   jvms->set_endoff(max_map);
   {
-    SafePointNode *map = new (C, max_map) SafePointNode( max_map, jvms );
+    SafePointNode *map = new (C) SafePointNode( max_map, jvms );
     jvms->set_map(map);
     set_jvms(jvms);
     assert(map == this->map(), "kit.map is set");
@@ -72,7 +72,7 @@ void GraphKit::gen_stub(address C_function,
   // Make up the parameters
   uint i;
   for( i = 0; i < parm_cnt; i++ )
-    map()->init_req(i, _gvn.transform(new (C, 1) ParmNode(start, i)));
+    map()->init_req(i, _gvn.transform(new (C) ParmNode(start, i)));
   for( ; i<map()->req(); i++ )
     map()->init_req(i, top());      // For nicer debugging
 
@@ -80,7 +80,7 @@ void GraphKit::gen_stub(address C_function,
   set_all_memory(map()->memory());
 
   // Get base of thread-local storage area
-  Node* thread = _gvn.transform( new (C, 1) ThreadLocalNode() );
+  Node* thread = _gvn.transform( new (C) ThreadLocalNode() );
 
   const int NoAlias = Compile::AliasIdxBot;
 
@@ -160,7 +160,7 @@ void GraphKit::gen_stub(address C_function,
 
   //-----------------------------
   // Make the call node
-  CallRuntimeNode *call = new (C, c_sig->domain()->cnt())
+  CallRuntimeNode *call = new (C)
     CallRuntimeNode(c_sig, C_function, name, TypePtr::BOTTOM);
   //-----------------------------
 
@@ -186,23 +186,23 @@ void GraphKit::gen_stub(address C_function,
 
   //-----------------------------
   // Now set up the return results
-  set_control( _gvn.transform( new (C, 1) ProjNode(call,TypeFunc::Control)) );
-  set_i_o(     _gvn.transform( new (C, 1) ProjNode(call,TypeFunc::I_O    )) );
+  set_control( _gvn.transform( new (C) ProjNode(call,TypeFunc::Control)) );
+  set_i_o(     _gvn.transform( new (C) ProjNode(call,TypeFunc::I_O    )) );
   set_all_memory_call(call);
   if (range->cnt() > TypeFunc::Parms) {
-    Node* retnode = _gvn.transform( new (C, 1) ProjNode(call,TypeFunc::Parms) );
+    Node* retnode = _gvn.transform( new (C) ProjNode(call,TypeFunc::Parms) );
     // C-land is allowed to return sub-word values.  Convert to integer type.
     assert( retval != Type::TOP, "" );
     if (retval == TypeInt::BOOL) {
-      retnode = _gvn.transform( new (C, 3) AndINode(retnode, intcon(0xFF)) );
+      retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0xFF)) );
     } else if (retval == TypeInt::CHAR) {
-      retnode = _gvn.transform( new (C, 3) AndINode(retnode, intcon(0xFFFF)) );
+      retnode = _gvn.transform( new (C) AndINode(retnode, intcon(0xFFFF)) );
     } else if (retval == TypeInt::BYTE) {
-      retnode = _gvn.transform( new (C, 3) LShiftINode(retnode, intcon(24)) );
-      retnode = _gvn.transform( new (C, 3) RShiftINode(retnode, intcon(24)) );
+      retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(24)) );
+      retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(24)) );
     } else if (retval == TypeInt::SHORT) {
-      retnode = _gvn.transform( new (C, 3) LShiftINode(retnode, intcon(16)) );
-      retnode = _gvn.transform( new (C, 3) RShiftINode(retnode, intcon(16)) );
+      retnode = _gvn.transform( new (C) LShiftINode(retnode, intcon(16)) );
+      retnode = _gvn.transform( new (C) RShiftINode(retnode, intcon(16)) );
     }
     map()->set_req( TypeFunc::Parms, retnode );
   }
@@ -247,21 +247,21 @@ void GraphKit::gen_stub(address C_function,
 
   Node* exit_memory = reset_memory();
 
-  Node* cmp = _gvn.transform( new (C, 3) CmpPNode(pending, null()) );
-  Node* bo  = _gvn.transform( new (C, 2) BoolNode(cmp, BoolTest::ne) );
+  Node* cmp = _gvn.transform( new (C) CmpPNode(pending, null()) );
+  Node* bo  = _gvn.transform( new (C) BoolNode(cmp, BoolTest::ne) );
   IfNode   *iff = create_and_map_if(control(), bo, PROB_MIN, COUNT_UNKNOWN);
 
-  Node* if_null     = _gvn.transform( new (C, 1) IfFalseNode(iff) );
-  Node* if_not_null = _gvn.transform( new (C, 1) IfTrueNode(iff)  );
+  Node* if_null     = _gvn.transform( new (C) IfFalseNode(iff) );
+  Node* if_not_null = _gvn.transform( new (C) IfTrueNode(iff)  );
 
   assert (StubRoutines::forward_exception_entry() != NULL, "must be generated before");
   Node *exc_target = makecon(TypeRawPtr::make( StubRoutines::forward_exception_entry() ));
-  Node *to_exc = new (C, TypeFunc::Parms+2) TailCallNode(if_not_null,
-                                  i_o(),
-                                  exit_memory,
-                                  frameptr(),
-                                  returnadr(),
-                                  exc_target, null());
+  Node *to_exc = new (C) TailCallNode(if_not_null,
+                                      i_o(),
+                                      exit_memory,
+                                      frameptr(),
+                                      returnadr(),
+                                      exc_target, null());
   root()->add_req(_gvn.transform(to_exc));  // bind to root to keep live
   C->init_start(start);
 
@@ -271,31 +271,31 @@ void GraphKit::gen_stub(address C_function,
   switch( is_fancy_jump ) {
   case 0:                       // Make a return instruction
     // Return to caller, free any space for return address
-    ret = new (C, TypeFunc::Parms) ReturnNode(TypeFunc::Parms, if_null,
-                         i_o(),
-                         exit_memory,
-                         frameptr(),
-                         returnadr());
+    ret = new (C) ReturnNode(TypeFunc::Parms, if_null,
+                             i_o(),
+                             exit_memory,
+                             frameptr(),
+                             returnadr());
     if (C->tf()->range()->cnt() > TypeFunc::Parms)
       ret->add_req( map()->in(TypeFunc::Parms) );
     break;
   case 1:    // This is a fancy tail-call jump.  Jump to computed address.
     // Jump to new callee; leave old return address alone.
-    ret = new (C, TypeFunc::Parms+2) TailCallNode(if_null,
-                           i_o(),
-                           exit_memory,
-                           frameptr(),
-                           returnadr(),
-                           target, map()->in(TypeFunc::Parms));
+    ret = new (C) TailCallNode(if_null,
+                               i_o(),
+                               exit_memory,
+                               frameptr(),
+                               returnadr(),
+                               target, map()->in(TypeFunc::Parms));
     break;
   case 2:                       // Pop return address & jump
     // Throw away old return address; jump to new computed address
     //assert(C_function == CAST_FROM_FN_PTR(address, OptoRuntime::rethrow_C), "fancy_jump==2 only for rethrow");
-    ret = new (C, TypeFunc::Parms+2) TailJumpNode(if_null,
-                           i_o(),
-                           exit_memory,
-                           frameptr(),
-                           target, map()->in(TypeFunc::Parms));
+    ret = new (C) TailJumpNode(if_null,
+                               i_o(),
+                               exit_memory,
+                               frameptr(),
+                               target, map()->in(TypeFunc::Parms));
     break;
   default:
     ShouldNotReachHere();
diff --git a/src/share/vm/opto/graphKit.cpp b/src/share/vm/opto/graphKit.cpp
index 200122620..53130fd45 100644
--- a/src/share/vm/opto/graphKit.cpp
+++ b/src/share/vm/opto/graphKit.cpp
@@ -280,7 +280,7 @@ JVMState* GraphKit::transfer_exceptions_into_jvms() {
       JVMState* jvms = new (C) JVMState(_method, NULL);
       jvms->set_bci(_bci);
       jvms->set_sp(_sp);
-      jvms->set_map(new (C, TypeFunc::Parms) SafePointNode(TypeFunc::Parms, jvms));
+      jvms->set_map(new (C) SafePointNode(TypeFunc::Parms, jvms));
       set_jvms(jvms);
       for (uint i = 0; i < map()->req(); i++)  map()->init_req(i, top());
       set_all_memory(top());
@@ -332,7 +332,7 @@ void GraphKit::combine_exception_states(SafePointNode* ex_map, SafePointNode* ph
   if (region->in(0) != hidden_merge_mark) {
     // The control input is not (yet) a specially-marked region in phi_map.
     // Make it so, and build some phis.
-    region = new (C, 2) RegionNode(2);
+    region = new (C) RegionNode(2);
     _gvn.set_type(region, Type::CONTROL);
     region->set_req(0, hidden_merge_mark);  // marks an internal ex-state
     region->init_req(1, phi_map->control());
@@ -481,13 +481,13 @@ void GraphKit::uncommon_trap_if_should_post_on_exceptions(Deoptimization::DeoptR
     // take the uncommon_trap in the BuildCutout below.
 
     // first must access the should_post_on_exceptions_flag in this thread's JavaThread
-    Node* jthread = _gvn.transform(new (C, 1) ThreadLocalNode());
+    Node* jthread = _gvn.transform(new (C) ThreadLocalNode());
     Node* adr = basic_plus_adr(top(), jthread, in_bytes(JavaThread::should_post_on_exceptions_flag_offset()));
     Node* should_post_flag = make_load(control(), adr, TypeInt::INT, T_INT, Compile::AliasIdxRaw, false);
 
     // Test the should_post_on_exceptions_flag vs. 0
-    Node* chk = _gvn.transform( new (C, 3) CmpINode(should_post_flag, intcon(0)) );
-    Node* tst = _gvn.transform( new (C, 2) BoolNode(chk, BoolTest::eq) );
+    Node* chk = _gvn.transform( new (C) CmpINode(should_post_flag, intcon(0)) );
+    Node* tst = _gvn.transform( new (C) BoolNode(chk, BoolTest::eq) );
 
     // Branch to slow_path if should_post_on_exceptions_flag was true
     { BuildCutout unless(this, tst, PROB_MAX);
@@ -656,8 +656,8 @@ BuildCutout::BuildCutout(GraphKit* kit, Node* p, float prob, float cnt)
   SafePointNode* outer_map = _map;   // preserved map is caller's
   SafePointNode* inner_map = kit->map();
   IfNode* iff = kit->create_and_map_if(outer_map->control(), p, prob, cnt);
-  outer_map->set_control(kit->gvn().transform( new (kit->C, 1) IfTrueNode(iff) ));
-  inner_map->set_control(kit->gvn().transform( new (kit->C, 1) IfFalseNode(iff) ));
+  outer_map->set_control(kit->gvn().transform( new (kit->C) IfTrueNode(iff) ));
+  inner_map->set_control(kit->gvn().transform( new (kit->C) IfFalseNode(iff) ));
 }
 BuildCutout::~BuildCutout() {
   GraphKit* kit = _kit;
@@ -1108,7 +1108,7 @@ bool GraphKit::compute_stack_effects(int& inputs, int& depth, bool for_parse) {
 Node* GraphKit::basic_plus_adr(Node* base, Node* ptr, Node* offset) {
   // short-circuit a common case
   if (offset == intcon(0))  return ptr;
-  return _gvn.transform( new (C, 4) AddPNode(base, ptr, offset) );
+  return _gvn.transform( new (C) AddPNode(base, ptr, offset) );
 }
 
 Node* GraphKit::ConvI2L(Node* offset) {
@@ -1117,7 +1117,7 @@ Node* GraphKit::ConvI2L(Node* offset) {
   if (offset_con != Type::OffsetBot) {
     return longcon((long) offset_con);
   }
-  return _gvn.transform( new (C, 2) ConvI2LNode(offset));
+  return _gvn.transform( new (C) ConvI2LNode(offset));
 }
 Node* GraphKit::ConvL2I(Node* offset) {
   // short-circuit a common case
@@ -1125,7 +1125,7 @@ Node* GraphKit::ConvL2I(Node* offset) {
   if (offset_con != (jlong)Type::OffsetBot) {
     return intcon((int) offset_con);
   }
-  return _gvn.transform( new (C, 2) ConvL2INode(offset));
+  return _gvn.transform( new (C) ConvL2INode(offset));
 }
 
 //-------------------------load_object_klass-----------------------------------
@@ -1144,7 +1144,7 @@ Node* GraphKit::load_array_length(Node* array) {
   Node *alen;
   if (alloc == NULL) {
     Node *r_adr = basic_plus_adr(array, arrayOopDesc::length_offset_in_bytes());
-    alen = _gvn.transform( new (C, 3) LoadRangeNode(0, immutable_memory(), r_adr, TypeInt::POS));
+    alen = _gvn.transform( new (C) LoadRangeNode(0, immutable_memory(), r_adr, TypeInt::POS));
   } else {
     alen = alloc->Ideal_length();
     Node* ccast = alloc->make_ideal_length(_gvn.type(array)->is_oopptr(), &_gvn);
@@ -1177,8 +1177,8 @@ Node* GraphKit::null_check_common(Node* value, BasicType type,
   // Construct NULL check
   Node *chk = NULL;
   switch(type) {
-    case T_LONG   : chk = new (C, 3) CmpLNode(value, _gvn.zerocon(T_LONG)); break;
-    case T_INT    : chk = new (C, 3) CmpINode( value, _gvn.intcon(0)); break;
+    case T_LONG   : chk = new (C) CmpLNode(value, _gvn.zerocon(T_LONG)); break;
+    case T_INT    : chk = new (C) CmpINode( value, _gvn.intcon(0)); break;
     case T_ARRAY  : // fall through
       type = T_OBJECT;  // simplify further tests
     case T_OBJECT : {
@@ -1225,7 +1225,7 @@ Node* GraphKit::null_check_common(Node* value, BasicType type,
           return value;           // Elided null check quickly!
         }
       }
-      chk = new (C, 3) CmpPNode( value, null() );
+      chk = new (C) CmpPNode( value, null() );
       break;
     }
 
@@ -1235,7 +1235,7 @@ Node* GraphKit::null_check_common(Node* value, BasicType type,
   chk = _gvn.transform(chk);
 
   BoolTest::mask btest = assert_null ? BoolTest::eq : BoolTest::ne;
-  BoolNode *btst = new (C, 2) BoolNode( chk, btest);
+  BoolNode *btst = new (C) BoolNode( chk, btest);
   Node   *tst = _gvn.transform( btst );
 
   //-----------
@@ -1302,8 +1302,8 @@ Node* GraphKit::null_check_common(Node* value, BasicType type,
 
   if (null_control != NULL) {
     IfNode* iff = create_and_map_if(control(), tst, ok_prob, COUNT_UNKNOWN);
-    Node* null_true = _gvn.transform( new (C, 1) IfFalseNode(iff));
-    set_control(      _gvn.transform( new (C, 1) IfTrueNode(iff)));
+    Node* null_true = _gvn.transform( new (C) IfFalseNode(iff));
+    set_control(      _gvn.transform( new (C) IfTrueNode(iff)));
     if (null_true == top())
       explicit_null_checks_elided++;
     (*null_control) = null_true;
@@ -1355,7 +1355,7 @@ Node* GraphKit::cast_not_null(Node* obj, bool do_replace_in_map) {
   // Object is already not-null?
   if( t == t_not_null ) return obj;
 
-  Node *cast = new (C, 2) CastPPNode(obj,t_not_null);
+  Node *cast = new (C) CastPPNode(obj,t_not_null);
   cast->init_req(0, control());
   cast = _gvn.transform( cast );
 
@@ -1410,7 +1410,7 @@ void GraphKit::set_all_memory(Node* newmem) {
 
 //------------------------------set_all_memory_call----------------------------
 void GraphKit::set_all_memory_call(Node* call, bool separate_io_proj) {
-  Node* newmem = _gvn.transform( new (C, 1) ProjNode(call, TypeFunc::Memory, separate_io_proj) );
+  Node* newmem = _gvn.transform( new (C) ProjNode(call, TypeFunc::Memory, separate_io_proj) );
   set_all_memory(newmem);
 }
 
@@ -1614,9 +1614,9 @@ Node* GraphKit::array_element_address(Node* ary, Node* idx, BasicType elembt,
   int index_max = max_jint - 1;  // array size is max_jint, index is one less
   if (sizetype != NULL)  index_max = sizetype->_hi - 1;
   const TypeLong* lidxtype = TypeLong::make(CONST64(0), index_max, Type::WidenMax);
-  idx = _gvn.transform( new (C, 2) ConvI2LNode(idx, lidxtype) );
+  idx = _gvn.transform( new (C) ConvI2LNode(idx, lidxtype) );
 #endif
-  Node* scale = _gvn.transform( new (C, 3) LShiftXNode(idx, intcon(shift)) );
+  Node* scale = _gvn.transform( new (C) LShiftXNode(idx, intcon(shift)) );
   return basic_plus_adr(ary, base, scale);
 }
 
@@ -1664,8 +1664,8 @@ void GraphKit::set_edges_for_java_call(CallJavaNode* call, bool must_throw, bool
 
   // Re-use the current map to produce the result.
 
-  set_control(_gvn.transform(new (C, 1) ProjNode(call, TypeFunc::Control)));
-  set_i_o(    _gvn.transform(new (C, 1) ProjNode(call, TypeFunc::I_O    , separate_io_proj)));
+  set_control(_gvn.transform(new (C) ProjNode(call, TypeFunc::Control)));
+  set_i_o(    _gvn.transform(new (C) ProjNode(call, TypeFunc::I_O    , separate_io_proj)));
   set_all_memory_call(xcall, separate_io_proj);
 
   //return xcall;   // no need, caller already has it
@@ -1679,7 +1679,7 @@ Node* GraphKit::set_results_for_java_call(CallJavaNode* call, bool separate_io_p
   if (call->method() == NULL ||
       call->method()->return_type()->basic_type() == T_VOID)
         ret = top();
-  else  ret = _gvn.transform(new (C, 1) ProjNode(call, TypeFunc::Parms));
+  else  ret = _gvn.transform(new (C) ProjNode(call, TypeFunc::Parms));
 
   // Note:  Since any out-of-line call can produce an exception,
   // we always insert an I_O projection from the call into the result.
@@ -1690,8 +1690,8 @@ Node* GraphKit::set_results_for_java_call(CallJavaNode* call, bool separate_io_p
     // The caller requested separate projections be used by the fall
     // through and exceptional paths, so replace the projections for
     // the fall through path.
-    set_i_o(_gvn.transform( new (C, 1) ProjNode(call, TypeFunc::I_O) ));
-    set_all_memory(_gvn.transform( new (C, 1) ProjNode(call, TypeFunc::Memory) ));
+    set_i_o(_gvn.transform( new (C) ProjNode(call, TypeFunc::I_O) ));
+    set_all_memory(_gvn.transform( new (C) ProjNode(call, TypeFunc::Memory) ));
   }
   return ret;
 }
@@ -1731,13 +1731,13 @@ void GraphKit::set_predefined_output_for_runtime_call(Node* call,
                                                       Node* keep_mem,
                                                       const TypePtr* hook_mem) {
   // no i/o
-  set_control(_gvn.transform( new (C, 1) ProjNode(call,TypeFunc::Control) ));
+  set_control(_gvn.transform( new (C) ProjNode(call,TypeFunc::Control) ));
   if (keep_mem) {
     // First clone the existing memory state
     set_all_memory(keep_mem);
     if (hook_mem != NULL) {
       // Make memory for the call
-      Node* mem = _gvn.transform( new (C, 1) ProjNode(call, TypeFunc::Memory) );
+      Node* mem = _gvn.transform( new (C) ProjNode(call, TypeFunc::Memory) );
       // Set the RawPtr memory state only.  This covers all the heap top/GC stuff
       // We also use hook_mem to extract specific effects from arraycopy stubs.
       set_memory(mem, hook_mem);
@@ -1841,7 +1841,7 @@ void GraphKit::increment_counter(Node* counter_addr) {
   int adr_type = Compile::AliasIdxRaw;
   Node* ctrl = control();
   Node* cnt  = make_load(ctrl, counter_addr, TypeInt::INT, T_INT, adr_type);
-  Node* incr = _gvn.transform(new (C, 3) AddINode(cnt, _gvn.intcon(1)));
+  Node* incr = _gvn.transform(new (C) AddINode(cnt, _gvn.intcon(1)));
   store_to_memory( ctrl, counter_addr, incr, T_INT, adr_type );
 }
 
@@ -1957,7 +1957,7 @@ void GraphKit::uncommon_trap(int trap_request,
   // The debug info is the only real input to this call.
 
   // Halt-and-catch fire here.  The above call should never return!
-  HaltNode* halt = new(C, TypeFunc::Parms) HaltNode(control(), frameptr());
+  HaltNode* halt = new(C) HaltNode(control(), frameptr());
   _gvn.set_type_bottom(halt);
   root()->add_req(halt);
 
@@ -2013,7 +2013,7 @@ void GraphKit::round_double_result(ciMethod* dest_method) {
 Node* GraphKit::precision_rounding(Node* n) {
   return UseStrictFP && _method->flags().is_strict()
     && UseSSE == 0 && Matcher::strict_fp_requires_explicit_rounding
-    ? _gvn.transform( new (C, 2) RoundFloatNode(0, n) )
+    ? _gvn.transform( new (C) RoundFloatNode(0, n) )
     : n;
 }
 
@@ -2021,7 +2021,7 @@ Node* GraphKit::precision_rounding(Node* n) {
 Node* GraphKit::dprecision_rounding(Node *n) {
   return UseStrictFP && _method->flags().is_strict()
     && UseSSE <= 1 && Matcher::strict_fp_requires_explicit_rounding
-    ? _gvn.transform( new (C, 2) RoundDoubleNode(0, n) )
+    ? _gvn.transform( new (C) RoundDoubleNode(0, n) )
     : n;
 }
 
@@ -2029,7 +2029,7 @@ Node* GraphKit::dprecision_rounding(Node *n) {
 Node* GraphKit::dstore_rounding(Node* n) {
   return Matcher::strict_fp_requires_explicit_rounding
     && UseSSE <= 1
-    ? _gvn.transform( new (C, 2) RoundDoubleNode(0, n) )
+    ? _gvn.transform( new (C) RoundDoubleNode(0, n) )
     : n;
 }
 
@@ -2102,11 +2102,11 @@ Node* GraphKit::opt_iff(Node* region, Node* iff) {
   IfNode *opt_iff = _gvn.transform(iff)->as_If();
 
   // Fast path taken; set region slot 2
-  Node *fast_taken = _gvn.transform( new (C, 1) IfFalseNode(opt_iff) );
+  Node *fast_taken = _gvn.transform( new (C) IfFalseNode(opt_iff) );
   region->init_req(2,fast_taken); // Capture fast-control
 
   // Fast path not-taken, i.e. slow path
-  Node *slow_taken = _gvn.transform( new (C, 1) IfTrueNode(opt_iff) );
+  Node *slow_taken = _gvn.transform( new (C) IfTrueNode(opt_iff) );
   return slow_taken;
 }
 
@@ -2122,7 +2122,6 @@ Node* GraphKit::make_runtime_call(int flags,
                                   Node* parm4, Node* parm5,
                                   Node* parm6, Node* parm7) {
   // Slow-path call
-  int size = call_type->domain()->cnt();
   bool is_leaf = !(flags & RC_NO_LEAF);
   bool has_io  = (!is_leaf && !(flags & RC_NO_IO));
   if (call_name == NULL) {
@@ -2131,12 +2130,12 @@ Node* GraphKit::make_runtime_call(int flags,
   }
   CallNode* call;
   if (!is_leaf) {
-    call = new(C, size) CallStaticJavaNode(call_type, call_addr, call_name,
+    call = new(C) CallStaticJavaNode(call_type, call_addr, call_name,
                                            bci(), adr_type);
   } else if (flags & RC_NO_FP) {
-    call = new(C, size) CallLeafNoFPNode(call_type, call_addr, call_name, adr_type);
+    call = new(C) CallLeafNoFPNode(call_type, call_addr, call_name, adr_type);
   } else {
-    call = new(C, size) CallLeafNode(call_type, call_addr, call_name, adr_type);
+    call = new(C) CallLeafNode(call_type, call_addr, call_name, adr_type);
   }
 
   // The following is similar to set_edges_for_java_call,
@@ -2197,7 +2196,7 @@ Node* GraphKit::make_runtime_call(int flags,
   }
 
   if (has_io) {
-    set_i_o(_gvn.transform(new (C, 1) ProjNode(call, TypeFunc::I_O)));
+    set_i_o(_gvn.transform(new (C) ProjNode(call, TypeFunc::I_O)));
   }
   return call;
 
@@ -2238,10 +2237,10 @@ void GraphKit::make_slow_call_ex(Node* call, ciInstanceKlass* ex_klass, bool sep
   if (stopped())  return;
 
   // Make a catch node with just two handlers:  fall-through and catch-all
-  Node* i_o  = _gvn.transform( new (C, 1) ProjNode(call, TypeFunc::I_O, separate_io_proj) );
-  Node* catc = _gvn.transform( new (C, 2) CatchNode(control(), i_o, 2) );
-  Node* norm = _gvn.transform( new (C, 1) CatchProjNode(catc, CatchProjNode::fall_through_index, CatchProjNode::no_handler_bci) );
-  Node* excp = _gvn.transform( new (C, 1) CatchProjNode(catc, CatchProjNode::catch_all_index,    CatchProjNode::no_handler_bci) );
+  Node* i_o  = _gvn.transform( new (C) ProjNode(call, TypeFunc::I_O, separate_io_proj) );
+  Node* catc = _gvn.transform( new (C) CatchNode(control(), i_o, 2) );
+  Node* norm = _gvn.transform( new (C) CatchProjNode(catc, CatchProjNode::fall_through_index, CatchProjNode::no_handler_bci) );
+  Node* excp = _gvn.transform( new (C) CatchProjNode(catc, CatchProjNode::catch_all_index,    CatchProjNode::no_handler_bci) );
 
   { PreserveJVMState pjvms(this);
     set_control(excp);
@@ -2251,7 +2250,7 @@ void GraphKit::make_slow_call_ex(Node* call, ciInstanceKlass* ex_klass, bool sep
       // Create an exception state also.
       // Use an exact type if the caller has specified a specific exception.
       const Type* ex_type = TypeOopPtr::make_from_klass_unique(ex_klass)->cast_to_ptr_type(TypePtr::NotNull);
-      Node*       ex_oop  = new (C, 2) CreateExNode(ex_type, control(), i_o);
+      Node*       ex_oop  = new (C) CreateExNode(ex_type, control(), i_o);
       add_exception_state(make_exception_state(_gvn.transform(ex_oop)));
     }
   }
@@ -2301,11 +2300,11 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
     case SSC_easy_test:
       {
         // Just do a direct pointer compare and be done.
-        Node* cmp = _gvn.transform( new(C, 3) CmpPNode(subklass, superklass) );
-        Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::eq) );
+        Node* cmp = _gvn.transform( new(C) CmpPNode(subklass, superklass) );
+        Node* bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::eq) );
         IfNode* iff = create_and_xform_if(control(), bol, PROB_STATIC_FREQUENT, COUNT_UNKNOWN);
-        set_control( _gvn.transform( new(C, 1) IfTrueNode (iff) ) );
-        return       _gvn.transform( new(C, 1) IfFalseNode(iff) );
+        set_control( _gvn.transform( new(C) IfTrueNode (iff) ) );
+        return       _gvn.transform( new(C) IfFalseNode(iff) );
       }
     case SSC_full_test:
       break;
@@ -2320,7 +2319,7 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
 
   // First load the super-klass's check-offset
   Node *p1 = basic_plus_adr( superklass, superklass, in_bytes(Klass::super_check_offset_offset()) );
-  Node *chk_off = _gvn.transform( new (C, 3) LoadINode( NULL, memory(p1), p1, _gvn.type(p1)->is_ptr() ) );
+  Node *chk_off = _gvn.transform( new (C) LoadINode( NULL, memory(p1), p1, _gvn.type(p1)->is_ptr() ) );
   int cacheoff_con = in_bytes(Klass::secondary_super_cache_offset());
   bool might_be_cache = (find_int_con(chk_off, cacheoff_con) == cacheoff_con);
 
@@ -2331,7 +2330,7 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
   // Worst-case type is a little odd: NULL is allowed as a result (usually
   // klass loads can never produce a NULL).
   Node *chk_off_X = ConvI2X(chk_off);
-  Node *p2 = _gvn.transform( new (C, 4) AddPNode(subklass,subklass,chk_off_X) );
+  Node *p2 = _gvn.transform( new (C) AddPNode(subklass,subklass,chk_off_X) );
   // For some types like interfaces the following loadKlass is from a 1-word
   // cache which is mutable so can't use immutable memory.  Other
   // types load from the super-class display table which is immutable.
@@ -2345,11 +2344,11 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
   // See if we get an immediate positive hit.  Happens roughly 83% of the
   // time.  Test to see if the value loaded just previously from the subklass
   // is exactly the superklass.
-  Node *cmp1 = _gvn.transform( new (C, 3) CmpPNode( superklass, nkls ) );
-  Node *bol1 = _gvn.transform( new (C, 2) BoolNode( cmp1, BoolTest::eq ) );
+  Node *cmp1 = _gvn.transform( new (C) CmpPNode( superklass, nkls ) );
+  Node *bol1 = _gvn.transform( new (C) BoolNode( cmp1, BoolTest::eq ) );
   IfNode *iff1 = create_and_xform_if( control(), bol1, PROB_LIKELY(0.83f), COUNT_UNKNOWN );
-  Node *iftrue1 = _gvn.transform( new (C, 1) IfTrueNode ( iff1 ) );
-  set_control(    _gvn.transform( new (C, 1) IfFalseNode( iff1 ) ) );
+  Node *iftrue1 = _gvn.transform( new (C) IfTrueNode ( iff1 ) );
+  set_control(    _gvn.transform( new (C) IfFalseNode( iff1 ) ) );
 
   // Compile speed common case: Check for being deterministic right now.  If
   // chk_off is a constant and not equal to cacheoff then we are NOT a
@@ -2362,9 +2361,9 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
   }
 
   // Gather the various success & failures here
-  RegionNode *r_ok_subtype = new (C, 4) RegionNode(4);
+  RegionNode *r_ok_subtype = new (C) RegionNode(4);
   record_for_igvn(r_ok_subtype);
-  RegionNode *r_not_subtype = new (C, 3) RegionNode(3);
+  RegionNode *r_not_subtype = new (C) RegionNode(3);
   record_for_igvn(r_not_subtype);
 
   r_ok_subtype->init_req(1, iftrue1);
@@ -2375,20 +2374,20 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
   // cache.  If it points to the display (and NOT the cache) and the display
   // missed then it's not a subtype.
   Node *cacheoff = _gvn.intcon(cacheoff_con);
-  Node *cmp2 = _gvn.transform( new (C, 3) CmpINode( chk_off, cacheoff ) );
-  Node *bol2 = _gvn.transform( new (C, 2) BoolNode( cmp2, BoolTest::ne ) );
+  Node *cmp2 = _gvn.transform( new (C) CmpINode( chk_off, cacheoff ) );
+  Node *bol2 = _gvn.transform( new (C) BoolNode( cmp2, BoolTest::ne ) );
   IfNode *iff2 = create_and_xform_if( control(), bol2, PROB_LIKELY(0.63f), COUNT_UNKNOWN );
-  r_not_subtype->init_req(1, _gvn.transform( new (C, 1) IfTrueNode (iff2) ) );
-  set_control(                _gvn.transform( new (C, 1) IfFalseNode(iff2) ) );
+  r_not_subtype->init_req(1, _gvn.transform( new (C) IfTrueNode (iff2) ) );
+  set_control(                _gvn.transform( new (C) IfFalseNode(iff2) ) );
 
   // Check for self.  Very rare to get here, but it is taken 1/3 the time.
   // No performance impact (too rare) but allows sharing of secondary arrays
   // which has some footprint reduction.
-  Node *cmp3 = _gvn.transform( new (C, 3) CmpPNode( subklass, superklass ) );
-  Node *bol3 = _gvn.transform( new (C, 2) BoolNode( cmp3, BoolTest::eq ) );
+  Node *cmp3 = _gvn.transform( new (C) CmpPNode( subklass, superklass ) );
+  Node *bol3 = _gvn.transform( new (C) BoolNode( cmp3, BoolTest::eq ) );
   IfNode *iff3 = create_and_xform_if( control(), bol3, PROB_LIKELY(0.36f), COUNT_UNKNOWN );
-  r_ok_subtype->init_req(2, _gvn.transform( new (C, 1) IfTrueNode ( iff3 ) ) );
-  set_control(               _gvn.transform( new (C, 1) IfFalseNode( iff3 ) ) );
+  r_ok_subtype->init_req(2, _gvn.transform( new (C) IfTrueNode ( iff3 ) ) );
+  set_control(               _gvn.transform( new (C) IfFalseNode( iff3 ) ) );
 
   // -- Roads not taken here: --
   // We could also have chosen to perform the self-check at the beginning
@@ -2412,13 +2411,13 @@ Node* GraphKit::gen_subtype_check(Node* subklass, Node* superklass) {
   // The decision to inline or out-of-line this final check is platform
   // dependent, and is found in the AD file definition of PartialSubtypeCheck.
   Node* psc = _gvn.transform(
-    new (C, 3) PartialSubtypeCheckNode(control(), subklass, superklass) );
+    new (C) PartialSubtypeCheckNode(control(), subklass, superklass) );
 
-  Node *cmp4 = _gvn.transform( new (C, 3) CmpPNode( psc, null() ) );
-  Node *bol4 = _gvn.transform( new (C, 2) BoolNode( cmp4, BoolTest::ne ) );
+  Node *cmp4 = _gvn.transform( new (C) CmpPNode( psc, null() ) );
+  Node *bol4 = _gvn.transform( new (C) BoolNode( cmp4, BoolTest::ne ) );
   IfNode *iff4 = create_and_xform_if( control(), bol4, PROB_FAIR, COUNT_UNKNOWN );
-  r_not_subtype->init_req(2, _gvn.transform( new (C, 1) IfTrueNode (iff4) ) );
-  r_ok_subtype ->init_req(3, _gvn.transform( new (C, 1) IfFalseNode(iff4) ) );
+  r_not_subtype->init_req(2, _gvn.transform( new (C) IfTrueNode (iff4) ) );
+  r_ok_subtype ->init_req(3, _gvn.transform( new (C) IfFalseNode(iff4) ) );
 
   // Return false path; set default control to true path.
   set_control( _gvn.transform(r_ok_subtype) );
@@ -2482,18 +2481,18 @@ Node* GraphKit::type_check_receiver(Node* receiver, ciKlass* klass,
   const TypeKlassPtr* tklass = TypeKlassPtr::make(klass);
   Node* recv_klass = load_object_klass(receiver);
   Node* want_klass = makecon(tklass);
-  Node* cmp = _gvn.transform( new(C, 3) CmpPNode(recv_klass, want_klass) );
-  Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::eq) );
+  Node* cmp = _gvn.transform( new(C) CmpPNode(recv_klass, want_klass) );
+  Node* bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::eq) );
   IfNode* iff = create_and_xform_if(control(), bol, prob, COUNT_UNKNOWN);
-  set_control( _gvn.transform( new(C, 1) IfTrueNode (iff) ));
-  Node* fail = _gvn.transform( new(C, 1) IfFalseNode(iff) );
+  set_control( _gvn.transform( new(C) IfTrueNode (iff) ));
+  Node* fail = _gvn.transform( new(C) IfFalseNode(iff) );
 
   const TypeOopPtr* recv_xtype = tklass->as_instance_type();
   assert(recv_xtype->klass_is_exact(), "");
 
   // Subsume downstream occurrences of receiver with a cast to
   // recv_xtype, since now we know what the type will be.
-  Node* cast = new(C, 2) CheckCastPPNode(control(), receiver, recv_xtype);
+  Node* cast = new(C) CheckCastPPNode(control(), receiver, recv_xtype);
   (*casted_receiver) = _gvn.transform(cast);
   // (User must make the replace_in_map call.)
 
@@ -2580,8 +2579,8 @@ Node* GraphKit::gen_instanceof(Node* obj, Node* superklass) {
 
   // Make the merge point
   enum { _obj_path = 1, _fail_path, _null_path, PATH_LIMIT };
-  RegionNode* region = new(C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  Node*       phi    = new(C, PATH_LIMIT) PhiNode(region, TypeInt::BOOL);
+  RegionNode* region = new(C) RegionNode(PATH_LIMIT);
+  Node*       phi    = new(C) PhiNode(region, TypeInt::BOOL);
   C->set_has_split_ifs(true); // Has chance for split-if optimization
 
   ciProfileData* data = NULL;
@@ -2683,8 +2682,8 @@ Node* GraphKit::gen_checkcast(Node *obj, Node* superklass,
 
   // Make the merge point
   enum { _obj_path = 1, _null_path, PATH_LIMIT };
-  RegionNode* region = new (C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  Node*       phi    = new (C, PATH_LIMIT) PhiNode(region, toop);
+  RegionNode* region = new (C) RegionNode(PATH_LIMIT);
+  Node*       phi    = new (C) PhiNode(region, toop);
   C->set_has_split_ifs(true); // Has chance for split-if optimization
 
   // Use null-cast information if it is available
@@ -2733,7 +2732,7 @@ Node* GraphKit::gen_checkcast(Node *obj, Node* superklass,
     Node* not_subtype_ctrl = gen_subtype_check( obj_klass, superklass );
 
     // Plug in success path into the merge
-    cast_obj = _gvn.transform(new (C, 2) CheckCastPPNode(control(),
+    cast_obj = _gvn.transform(new (C) CheckCastPPNode(control(),
                                                          not_null_obj, toop));
     // Failure path ends in uncommon trap (or may be dead - failure impossible)
     if (failure_control == NULL) {
@@ -2787,7 +2786,7 @@ Node* GraphKit::insert_mem_bar(int opcode, Node* precedent) {
   mb->init_req(TypeFunc::Control, control());
   mb->init_req(TypeFunc::Memory,  reset_memory());
   Node* membar = _gvn.transform(mb);
-  set_control(_gvn.transform(new (C, 1) ProjNode(membar,TypeFunc::Control) ));
+  set_control(_gvn.transform(new (C) ProjNode(membar,TypeFunc::Control) ));
   set_all_memory_call(membar);
   return membar;
 }
@@ -2816,11 +2815,11 @@ Node* GraphKit::insert_mem_bar_volatile(int opcode, int alias_idx, Node* precede
     mb->set_req(TypeFunc::Memory, memory(alias_idx));
   }
   Node* membar = _gvn.transform(mb);
-  set_control(_gvn.transform(new (C, 1) ProjNode(membar, TypeFunc::Control)));
+  set_control(_gvn.transform(new (C) ProjNode(membar, TypeFunc::Control)));
   if (alias_idx == Compile::AliasIdxBot) {
-    merged_memory()->set_base_memory(_gvn.transform(new (C, 1) ProjNode(membar, TypeFunc::Memory)));
+    merged_memory()->set_base_memory(_gvn.transform(new (C) ProjNode(membar, TypeFunc::Memory)));
   } else {
-    set_memory(_gvn.transform(new (C, 1) ProjNode(membar, TypeFunc::Memory)),alias_idx);
+    set_memory(_gvn.transform(new (C) ProjNode(membar, TypeFunc::Memory)),alias_idx);
   }
   return membar;
 }
@@ -2840,10 +2839,10 @@ FastLockNode* GraphKit::shared_lock(Node* obj) {
   assert(dead_locals_are_killed(), "should kill locals before sync. point");
 
   // Box the stack location
-  Node* box = _gvn.transform(new (C, 1) BoxLockNode(next_monitor()));
+  Node* box = _gvn.transform(new (C) BoxLockNode(next_monitor()));
   Node* mem = reset_memory();
 
-  FastLockNode * flock = _gvn.transform(new (C, 3) FastLockNode(0, obj, box) )->as_FastLock();
+  FastLockNode * flock = _gvn.transform(new (C) FastLockNode(0, obj, box) )->as_FastLock();
   if (PrintPreciseBiasedLockingStatistics) {
     // Create the counters for this fast lock.
     flock->create_lock_counter(sync_jvms()); // sync_jvms used to get current bci
@@ -2853,7 +2852,7 @@ FastLockNode* GraphKit::shared_lock(Node* obj) {
   map()->push_monitor( flock );
 
   const TypeFunc *tf = LockNode::lock_type();
-  LockNode *lock = new (C, tf->domain()->cnt()) LockNode(C, tf);
+  LockNode *lock = new (C) LockNode(C, tf);
 
   lock->init_req( TypeFunc::Control, control() );
   lock->init_req( TypeFunc::Memory , mem );
@@ -2907,7 +2906,7 @@ void GraphKit::shared_unlock(Node* box, Node* obj) {
   insert_mem_bar(Op_MemBarReleaseLock);
 
   const TypeFunc *tf = OptoRuntime::complete_monitor_exit_Type();
-  UnlockNode *unlock = new (C, tf->domain()->cnt()) UnlockNode(C, tf);
+  UnlockNode *unlock = new (C) UnlockNode(C, tf);
   uint raw_idx = Compile::AliasIdxRaw;
   unlock->init_req( TypeFunc::Control, control() );
   unlock->init_req( TypeFunc::Memory , memory(raw_idx) );
@@ -2973,19 +2972,19 @@ Node* GraphKit::set_output_for_allocation(AllocateNode* alloc,
   alloc->set_req( TypeFunc::FramePtr, frameptr() );
   add_safepoint_edges(alloc);
   Node* allocx = _gvn.transform(alloc);
-  set_control( _gvn.transform(new (C, 1) ProjNode(allocx, TypeFunc::Control) ) );
+  set_control( _gvn.transform(new (C) ProjNode(allocx, TypeFunc::Control) ) );
   // create memory projection for i_o
-  set_memory ( _gvn.transform( new (C, 1) ProjNode(allocx, TypeFunc::Memory, true) ), rawidx );
+  set_memory ( _gvn.transform( new (C) ProjNode(allocx, TypeFunc::Memory, true) ), rawidx );
   make_slow_call_ex(allocx, env()->OutOfMemoryError_klass(), true);
 
   // create a memory projection as for the normal control path
-  Node* malloc = _gvn.transform(new (C, 1) ProjNode(allocx, TypeFunc::Memory));
+  Node* malloc = _gvn.transform(new (C) ProjNode(allocx, TypeFunc::Memory));
   set_memory(malloc, rawidx);
 
   // a normal slow-call doesn't change i_o, but an allocation does
   // we create a separate i_o projection for the normal control path
-  set_i_o(_gvn.transform( new (C, 1) ProjNode(allocx, TypeFunc::I_O, false) ) );
-  Node* rawoop = _gvn.transform( new (C, 1) ProjNode(allocx, TypeFunc::Parms) );
+  set_i_o(_gvn.transform( new (C) ProjNode(allocx, TypeFunc::I_O, false) ) );
+  Node* rawoop = _gvn.transform( new (C) ProjNode(allocx, TypeFunc::Parms) );
 
   // put in an initialization barrier
   InitializeNode* init = insert_mem_bar_volatile(Op_Initialize, rawidx,
@@ -3021,7 +3020,7 @@ Node* GraphKit::set_output_for_allocation(AllocateNode* alloc,
   }
 
   // Cast raw oop to the real thing...
-  Node* javaoop = new (C, 2) CheckCastPPNode(control(), rawoop, oop_type);
+  Node* javaoop = new (C) CheckCastPPNode(control(), rawoop, oop_type);
   javaoop = _gvn.transform(javaoop);
   C->set_recent_alloc(control(), javaoop);
   assert(just_allocated_object(control()) == javaoop, "just allocated");
@@ -3080,9 +3079,9 @@ Node* GraphKit::new_instance(Node* klass_node,
     // (It may be stress-tested by specifying StressReflectiveCode.)
     // Basically, we want to get into the VM is there's an illegal argument.
     Node* bit = intcon(Klass::_lh_instance_slow_path_bit);
-    initial_slow_test = _gvn.transform( new (C, 3) AndINode(layout_val, bit) );
+    initial_slow_test = _gvn.transform( new (C) AndINode(layout_val, bit) );
     if (extra_slow_test != intcon(0)) {
-      initial_slow_test = _gvn.transform( new (C, 3) OrINode(initial_slow_test, extra_slow_test) );
+      initial_slow_test = _gvn.transform( new (C) OrINode(initial_slow_test, extra_slow_test) );
     }
     // (Macro-expander will further convert this to a Bool, if necessary.)
   }
@@ -3099,7 +3098,7 @@ Node* GraphKit::new_instance(Node* klass_node,
     // Clear the low bits to extract layout_helper_size_in_bytes:
     assert((int)Klass::_lh_instance_slow_path_bit < BytesPerLong, "clear bit");
     Node* mask = MakeConX(~ (intptr_t)right_n_bits(LogBytesPerLong));
-    size = _gvn.transform( new (C, 3) AndXNode(size, mask) );
+    size = _gvn.transform( new (C) AndXNode(size, mask) );
   }
   if (return_size_val != NULL) {
     (*return_size_val) = size;
@@ -3120,11 +3119,10 @@ Node* GraphKit::new_instance(Node* klass_node,
   set_all_memory(mem); // Create new memory state
 
   AllocateNode* alloc
-    = new (C, AllocateNode::ParmLimit)
-        AllocateNode(C, AllocateNode::alloc_type(),
-                     control(), mem, i_o(),
-                     size, klass_node,
-                     initial_slow_test);
+    = new (C) AllocateNode(C, AllocateNode::alloc_type(),
+                           control(), mem, i_o(),
+                           size, klass_node,
+                           initial_slow_test);
 
   return set_output_for_allocation(alloc, oop_type);
 }
@@ -3147,8 +3145,8 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
     // Optimistically assume that it is a subtype of Object[],
     // so that we can fold up all the address arithmetic.
     layout_con = Klass::array_layout_helper(T_OBJECT);
-    Node* cmp_lh = _gvn.transform( new(C, 3) CmpINode(layout_val, intcon(layout_con)) );
-    Node* bol_lh = _gvn.transform( new(C, 2) BoolNode(cmp_lh, BoolTest::eq) );
+    Node* cmp_lh = _gvn.transform( new(C) CmpINode(layout_val, intcon(layout_con)) );
+    Node* bol_lh = _gvn.transform( new(C) BoolNode(cmp_lh, BoolTest::eq) );
     { BuildCutout unless(this, bol_lh, PROB_MAX);
       _sp += nargs;
       uncommon_trap(Deoptimization::Reason_class_check,
@@ -3172,8 +3170,8 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
     fast_size_limit <<= (LogBytesPerLong - log2_esize);
   }
 
-  Node* initial_slow_cmp  = _gvn.transform( new (C, 3) CmpUNode( length, intcon( fast_size_limit ) ) );
-  Node* initial_slow_test = _gvn.transform( new (C, 2) BoolNode( initial_slow_cmp, BoolTest::gt ) );
+  Node* initial_slow_cmp  = _gvn.transform( new (C) CmpUNode( length, intcon( fast_size_limit ) ) );
+  Node* initial_slow_test = _gvn.transform( new (C) BoolNode( initial_slow_cmp, BoolTest::gt ) );
   if (initial_slow_test->is_Bool()) {
     // Hide it behind a CMoveI, or else PhaseIdealLoop::split_up will get sick.
     initial_slow_test = initial_slow_test->as_Bool()->as_int_value(&_gvn);
@@ -3201,10 +3199,10 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
   } else {
     Node* hss   = intcon(Klass::_lh_header_size_shift);
     Node* hsm   = intcon(Klass::_lh_header_size_mask);
-    Node* hsize = _gvn.transform( new(C, 3) URShiftINode(layout_val, hss) );
-    hsize       = _gvn.transform( new(C, 3) AndINode(hsize, hsm) );
+    Node* hsize = _gvn.transform( new(C) URShiftINode(layout_val, hss) );
+    hsize       = _gvn.transform( new(C) AndINode(hsize, hsm) );
     Node* mask  = intcon(round_mask);
-    header_size = _gvn.transform( new(C, 3) AddINode(hsize, mask) );
+    header_size = _gvn.transform( new(C) AddINode(hsize, mask) );
   }
 
   Node* elem_shift = NULL;
@@ -3229,7 +3227,7 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
       jlong size_max = arrayOopDesc::max_array_length(T_BYTE);
       if (size_max > tllen->_hi)  size_max = tllen->_hi;
       const TypeLong* tlcon = TypeLong::make(CONST64(0), size_max, Type::WidenMin);
-      lengthx = _gvn.transform( new (C, 2) ConvI2LNode(length, tlcon));
+      lengthx = _gvn.transform( new (C) ConvI2LNode(length, tlcon));
     }
   }
 #endif
@@ -3240,11 +3238,11 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
   // after a successful allocation.
   Node* abody = lengthx;
   if (elem_shift != NULL)
-    abody     = _gvn.transform( new(C, 3) LShiftXNode(lengthx, elem_shift) );
-  Node* size  = _gvn.transform( new(C, 3) AddXNode(headerx, abody) );
+    abody     = _gvn.transform( new(C) LShiftXNode(lengthx, elem_shift) );
+  Node* size  = _gvn.transform( new(C) AddXNode(headerx, abody) );
   if (round_mask != 0) {
     Node* mask = MakeConX(~round_mask);
-    size       = _gvn.transform( new(C, 3) AndXNode(size, mask) );
+    size       = _gvn.transform( new(C) AndXNode(size, mask) );
   }
   // else if round_mask == 0, the size computation is self-rounding
 
@@ -3262,12 +3260,11 @@ Node* GraphKit::new_array(Node* klass_node,     // array klass (maybe variable)
 
   // Create the AllocateArrayNode and its result projections
   AllocateArrayNode* alloc
-    = new (C, AllocateArrayNode::ParmLimit)
-        AllocateArrayNode(C, AllocateArrayNode::alloc_type(),
-                          control(), mem, i_o(),
-                          size, klass_node,
-                          initial_slow_test,
-                          length);
+    = new (C) AllocateArrayNode(C, AllocateArrayNode::alloc_type(),
+                                control(), mem, i_o(),
+                                size, klass_node,
+                                initial_slow_test,
+                                length);
 
   // Cast to correct type.  Note that the klass_node may be constant or not,
   // and in the latter case the actual array type will be inexact also.
@@ -3386,10 +3383,10 @@ void GraphKit::add_predicate_impl(Deoptimization::DeoptReason reason, int nargs)
   }
 
   Node *cont    = _gvn.intcon(1);
-  Node* opq     = _gvn.transform(new (C, 2) Opaque1Node(C, cont));
-  Node *bol     = _gvn.transform(new (C, 2) Conv2BNode(opq));
+  Node* opq     = _gvn.transform(new (C) Opaque1Node(C, cont));
+  Node *bol     = _gvn.transform(new (C) Conv2BNode(opq));
   IfNode* iff   = create_and_map_if(control(), bol, PROB_MAX, COUNT_UNKNOWN);
-  Node* iffalse = _gvn.transform(new (C, 1) IfFalseNode(iff));
+  Node* iffalse = _gvn.transform(new (C) IfFalseNode(iff));
   C->add_predicate_opaq(opq);
   {
     PreserveJVMState pjvms(this);
@@ -3397,7 +3394,7 @@ void GraphKit::add_predicate_impl(Deoptimization::DeoptReason reason, int nargs)
     _sp += nargs;
     uncommon_trap(reason, Deoptimization::Action_maybe_recompile);
   }
-  Node* iftrue = _gvn.transform(new (C, 1) IfTrueNode(iff));
+  Node* iftrue = _gvn.transform(new (C) IfTrueNode(iff));
   set_control(iftrue);
 }
 
@@ -3590,7 +3587,7 @@ void GraphKit::g1_write_barrier_pre(bool do_load,
 #ifdef _LP64
         // We could refine the type for what it's worth
         // const TypeLong* lidxtype = TypeLong::make(CONST64(0), get_size_from_queue);
-        next_indexX = _gvn.transform( new (C, 2) ConvI2LNode(next_index, TypeLong::make(0, max_jlong, Type::WidenMax)) );
+        next_indexX = _gvn.transform( new (C) ConvI2LNode(next_index, TypeLong::make(0, max_jlong, Type::WidenMax)) );
 #endif
 
         // Now get the buffer location we will log the previous value into and store it
@@ -3638,7 +3635,7 @@ void GraphKit::g1_mark_card(IdealKit& ideal,
 #ifdef _LP64
     // We could refine the type for what it's worth
     // const TypeLong* lidxtype = TypeLong::make(CONST64(0), get_size_from_queue);
-    next_indexX = _gvn.transform( new (C, 2) ConvI2LNode(next_index, TypeLong::make(0, max_jlong, Type::WidenMax)) );
+    next_indexX = _gvn.transform( new (C) ConvI2LNode(next_index, TypeLong::make(0, max_jlong, Type::WidenMax)) );
 #endif // _LP64
     Node* log_addr = __ AddP(no_base, buffer, next_indexX);
 
diff --git a/src/share/vm/opto/graphKit.hpp b/src/share/vm/opto/graphKit.hpp
index 2c09486c3..ede5b92e6 100644
--- a/src/share/vm/opto/graphKit.hpp
+++ b/src/share/vm/opto/graphKit.hpp
@@ -303,31 +303,31 @@ class GraphKit : public Phase {
 
 
   // Some convenient shortcuts for common nodes
-  Node* IfTrue(IfNode* iff)                   { return _gvn.transform(new (C,1) IfTrueNode(iff));      }
-  Node* IfFalse(IfNode* iff)                  { return _gvn.transform(new (C,1) IfFalseNode(iff));     }
+  Node* IfTrue(IfNode* iff)                   { return _gvn.transform(new (C) IfTrueNode(iff));      }
+  Node* IfFalse(IfNode* iff)                  { return _gvn.transform(new (C) IfFalseNode(iff));     }
 
-  Node* AddI(Node* l, Node* r)                { return _gvn.transform(new (C,3) AddINode(l, r));       }
-  Node* SubI(Node* l, Node* r)                { return _gvn.transform(new (C,3) SubINode(l, r));       }
-  Node* MulI(Node* l, Node* r)                { return _gvn.transform(new (C,3) MulINode(l, r));       }
-  Node* DivI(Node* ctl, Node* l, Node* r)     { return _gvn.transform(new (C,3) DivINode(ctl, l, r));  }
+  Node* AddI(Node* l, Node* r)                { return _gvn.transform(new (C) AddINode(l, r));       }
+  Node* SubI(Node* l, Node* r)                { return _gvn.transform(new (C) SubINode(l, r));       }
+  Node* MulI(Node* l, Node* r)                { return _gvn.transform(new (C) MulINode(l, r));       }
+  Node* DivI(Node* ctl, Node* l, Node* r)     { return _gvn.transform(new (C) DivINode(ctl, l, r));  }
 
-  Node* AndI(Node* l, Node* r)                { return _gvn.transform(new (C,3) AndINode(l, r));       }
-  Node* OrI(Node* l, Node* r)                 { return _gvn.transform(new (C,3) OrINode(l, r));        }
-  Node* XorI(Node* l, Node* r)                { return _gvn.transform(new (C,3) XorINode(l, r));       }
+  Node* AndI(Node* l, Node* r)                { return _gvn.transform(new (C) AndINode(l, r));       }
+  Node* OrI(Node* l, Node* r)                 { return _gvn.transform(new (C) OrINode(l, r));        }
+  Node* XorI(Node* l, Node* r)                { return _gvn.transform(new (C) XorINode(l, r));       }
 
-  Node* MaxI(Node* l, Node* r)                { return _gvn.transform(new (C,3) MaxINode(l, r));       }
-  Node* MinI(Node* l, Node* r)                { return _gvn.transform(new (C,3) MinINode(l, r));       }
+  Node* MaxI(Node* l, Node* r)                { return _gvn.transform(new (C) MaxINode(l, r));       }
+  Node* MinI(Node* l, Node* r)                { return _gvn.transform(new (C) MinINode(l, r));       }
 
-  Node* LShiftI(Node* l, Node* r)             { return _gvn.transform(new (C,3) LShiftINode(l, r));    }
-  Node* RShiftI(Node* l, Node* r)             { return _gvn.transform(new (C,3) RShiftINode(l, r));    }
-  Node* URShiftI(Node* l, Node* r)            { return _gvn.transform(new (C,3) URShiftINode(l, r));   }
+  Node* LShiftI(Node* l, Node* r)             { return _gvn.transform(new (C) LShiftINode(l, r));    }
+  Node* RShiftI(Node* l, Node* r)             { return _gvn.transform(new (C) RShiftINode(l, r));    }
+  Node* URShiftI(Node* l, Node* r)            { return _gvn.transform(new (C) URShiftINode(l, r));   }
 
-  Node* CmpI(Node* l, Node* r)                { return _gvn.transform(new (C,3) CmpINode(l, r));       }
-  Node* CmpL(Node* l, Node* r)                { return _gvn.transform(new (C,3) CmpLNode(l, r));       }
-  Node* CmpP(Node* l, Node* r)                { return _gvn.transform(new (C,3) CmpPNode(l, r));       }
-  Node* Bool(Node* cmp, BoolTest::mask relop) { return _gvn.transform(new (C,2) BoolNode(cmp, relop)); }
+  Node* CmpI(Node* l, Node* r)                { return _gvn.transform(new (C) CmpINode(l, r));       }
+  Node* CmpL(Node* l, Node* r)                { return _gvn.transform(new (C) CmpLNode(l, r));       }
+  Node* CmpP(Node* l, Node* r)                { return _gvn.transform(new (C) CmpPNode(l, r));       }
+  Node* Bool(Node* cmp, BoolTest::mask relop) { return _gvn.transform(new (C) BoolNode(cmp, relop)); }
 
-  Node* AddP(Node* b, Node* a, Node* o)       { return _gvn.transform(new (C,4) AddPNode(b, a, o));    }
+  Node* AddP(Node* b, Node* a, Node* o)       { return _gvn.transform(new (C) AddPNode(b, a, o));    }
 
   // Convert between int and long, and size_t.
   // (See macros ConvI2X, etc., in type.hpp for ConvI2X, etc.)
@@ -792,7 +792,7 @@ class GraphKit : public Phase {
 
   // Handy for making control flow
   IfNode* create_and_map_if(Node* ctrl, Node* tst, float prob, float cnt) {
-    IfNode* iff = new (C, 2) IfNode(ctrl, tst, prob, cnt);// New IfNode's
+    IfNode* iff = new (C) IfNode(ctrl, tst, prob, cnt);// New IfNode's
     _gvn.set_type(iff, iff->Value(&_gvn)); // Value may be known at parse-time
     // Place 'if' on worklist if it will be in graph
     if (!tst->is_Con())  record_for_igvn(iff);     // Range-check and Null-check removal is later
@@ -800,7 +800,7 @@ class GraphKit : public Phase {
   }
 
   IfNode* create_and_xform_if(Node* ctrl, Node* tst, float prob, float cnt) {
-    IfNode* iff = new (C, 2) IfNode(ctrl, tst, prob, cnt);// New IfNode's
+    IfNode* iff = new (C) IfNode(ctrl, tst, prob, cnt);// New IfNode's
     _gvn.transform(iff);                           // Value may be known at parse-time
     // Place 'if' on worklist if it will be in graph
     if (!tst->is_Con())  record_for_igvn(iff);     // Range-check and Null-check removal is later
diff --git a/src/share/vm/opto/idealKit.cpp b/src/share/vm/opto/idealKit.cpp
index 1cc4b2bd1..9083c6206 100644
--- a/src/share/vm/opto/idealKit.cpp
+++ b/src/share/vm/opto/idealKit.cpp
@@ -86,7 +86,7 @@ void IdealKit::if_then(Node* left, BoolTest::mask relop,
   }
   // Delay gvn.tranform on if-nodes until construction is finished
   // to prevent a constant bool input from discarding a control output.
-  IfNode* iff = delay_transform(new (C, 2) IfNode(ctrl(), bol, prob, cnt))->as_If();
+  IfNode* iff = delay_transform(new (C) IfNode(ctrl(), bol, prob, cnt))->as_If();
   Node* then  = IfTrue(iff);
   Node* elsen = IfFalse(iff);
   Node* else_cvstate = copy_cvstate();
@@ -205,7 +205,7 @@ Node* IdealKit::make_label(int goto_ct) {
   assert(_cvstate != NULL, "must declare variables before labels");
   Node* lab = new_cvstate();
   int sz = 1 + goto_ct + 1 /* fall thru */;
-  Node* reg = delay_transform(new (C, sz) RegionNode(sz));
+  Node* reg = delay_transform(new (C) RegionNode(sz));
   lab->init_req(TypeFunc::Control, reg);
   return lab;
 }
@@ -315,7 +315,7 @@ Node* IdealKit::delay_transform(Node* n) {
 //-----------------------------new_cvstate-----------------------------------
 Node* IdealKit::new_cvstate() {
   uint sz = _var_ct + first_var;
-  return new (C, sz) Node(sz);
+  return new (C) Node(sz);
 }
 
 //-----------------------------copy_cvstate-----------------------------------
@@ -413,7 +413,7 @@ Node* IdealKit::storeCM(Node* ctl, Node* adr, Node *val, Node* oop_store, int oo
 
   // Add required edge to oop_store, optimizer does not support precedence edges.
   // Convert required edge to precedence edge before allocation.
-  Node* st = new (C, 5) StoreCMNode(ctl, mem, adr, adr_type, val, oop_store, oop_adr_idx);
+  Node* st = new (C) StoreCMNode(ctl, mem, adr, adr_type, val, oop_store, oop_adr_idx);
 
   st = transform(st);
   set_memory(st, adr_idx);
@@ -513,8 +513,7 @@ void IdealKit::make_leaf_call(const TypeFunc *slow_call_type,
   uint adr_idx = C->get_alias_index(adr_type);
 
   // Slow-path leaf call
-  int size = slow_call_type->domain()->cnt();
-  CallNode *call =  (CallNode*)new (C, size) CallLeafNode( slow_call_type, slow_call, leaf_name, adr_type);
+  CallNode *call =  (CallNode*)new (C) CallLeafNode( slow_call_type, slow_call, leaf_name, adr_type);
 
   // Set fixed predefined input arguments
   call->init_req( TypeFunc::Control, ctrl() );
@@ -535,10 +534,10 @@ void IdealKit::make_leaf_call(const TypeFunc *slow_call_type,
 
   // Slow leaf call has no side-effects, sets few values
 
-  set_ctrl(transform( new (C, 1) ProjNode(call,TypeFunc::Control) ));
+  set_ctrl(transform( new (C) ProjNode(call,TypeFunc::Control) ));
 
   // Make memory for the call
-  Node* mem = _gvn.transform( new (C, 1) ProjNode(call, TypeFunc::Memory) );
+  Node* mem = _gvn.transform( new (C) ProjNode(call, TypeFunc::Memory) );
 
   // Set the RawPtr memory state only.
   set_memory(mem, adr_idx);
@@ -561,8 +560,7 @@ void IdealKit::make_leaf_call_no_fp(const TypeFunc *slow_call_type,
   uint adr_idx = C->get_alias_index(adr_type);
 
   // Slow-path leaf call
-  int size = slow_call_type->domain()->cnt();
-  CallNode *call =  (CallNode*)new (C, size) CallLeafNoFPNode( slow_call_type, slow_call, leaf_name, adr_type);
+  CallNode *call =  (CallNode*)new (C) CallLeafNoFPNode( slow_call_type, slow_call, leaf_name, adr_type);
 
   // Set fixed predefined input arguments
   call->init_req( TypeFunc::Control, ctrl() );
@@ -583,10 +581,10 @@ void IdealKit::make_leaf_call_no_fp(const TypeFunc *slow_call_type,
 
   // Slow leaf call has no side-effects, sets few values
 
-  set_ctrl(transform( new (C, 1) ProjNode(call,TypeFunc::Control) ));
+  set_ctrl(transform( new (C) ProjNode(call,TypeFunc::Control) ));
 
   // Make memory for the call
-  Node* mem = _gvn.transform( new (C, 1) ProjNode(call, TypeFunc::Memory) );
+  Node* mem = _gvn.transform( new (C) ProjNode(call, TypeFunc::Memory) );
 
   // Set the RawPtr memory state only.
   set_memory(mem, adr_idx);
diff --git a/src/share/vm/opto/idealKit.hpp b/src/share/vm/opto/idealKit.hpp
index 6787e533a..c6cc3c2af 100644
--- a/src/share/vm/opto/idealKit.hpp
+++ b/src/share/vm/opto/idealKit.hpp
@@ -175,39 +175,39 @@ class IdealKit: public StackObj {
   void declarations_done();
   void drain_delay_transform();
 
-  Node* IfTrue(IfNode* iff)  { return transform(new (C,1) IfTrueNode(iff)); }
-  Node* IfFalse(IfNode* iff) { return transform(new (C,1) IfFalseNode(iff)); }
+  Node* IfTrue(IfNode* iff)  { return transform(new (C) IfTrueNode(iff)); }
+  Node* IfFalse(IfNode* iff) { return transform(new (C) IfFalseNode(iff)); }
 
   // Data
   Node* ConI(jint k) { return (Node*)gvn().intcon(k); }
   Node* makecon(const Type *t)  const { return _gvn.makecon(t); }
 
-  Node* AddI(Node* l, Node* r) { return transform(new (C,3) AddINode(l, r)); }
-  Node* SubI(Node* l, Node* r) { return transform(new (C,3) SubINode(l, r)); }
-  Node* AndI(Node* l, Node* r) { return transform(new (C,3) AndINode(l, r)); }
-  Node* MaxI(Node* l, Node* r) { return transform(new (C,3) MaxINode(l, r)); }
-  Node* LShiftI(Node* l, Node* r) { return transform(new (C,3) LShiftINode(l, r)); }
-  Node* CmpI(Node* l, Node* r) { return transform(new (C,3) CmpINode(l, r)); }
-  Node* Bool(Node* cmp, BoolTest::mask relop) { return transform(new (C,2) BoolNode(cmp, relop)); }
+  Node* AddI(Node* l, Node* r) { return transform(new (C) AddINode(l, r)); }
+  Node* SubI(Node* l, Node* r) { return transform(new (C) SubINode(l, r)); }
+  Node* AndI(Node* l, Node* r) { return transform(new (C) AndINode(l, r)); }
+  Node* MaxI(Node* l, Node* r) { return transform(new (C) MaxINode(l, r)); }
+  Node* LShiftI(Node* l, Node* r) { return transform(new (C) LShiftINode(l, r)); }
+  Node* CmpI(Node* l, Node* r) { return transform(new (C) CmpINode(l, r)); }
+  Node* Bool(Node* cmp, BoolTest::mask relop) { return transform(new (C) BoolNode(cmp, relop)); }
   void  increment(IdealVariable& v, Node* j)  { set(v, AddI(value(v), j)); }
   void  decrement(IdealVariable& v, Node* j)  { set(v, SubI(value(v), j)); }
 
-  Node* CmpL(Node* l, Node* r) { return transform(new (C,3) CmpLNode(l, r)); }
+  Node* CmpL(Node* l, Node* r) { return transform(new (C) CmpLNode(l, r)); }
 
   // TLS
-  Node* thread()  {  return gvn().transform(new (C, 1) ThreadLocalNode()); }
+  Node* thread()  {  return gvn().transform(new (C) ThreadLocalNode()); }
 
   // Pointers
-  Node* AddP(Node *base, Node *ptr, Node *off) { return transform(new (C,4) AddPNode(base, ptr, off)); }
-  Node* CmpP(Node* l, Node* r) { return transform(new (C,3) CmpPNode(l, r)); }
+  Node* AddP(Node *base, Node *ptr, Node *off) { return transform(new (C) AddPNode(base, ptr, off)); }
+  Node* CmpP(Node* l, Node* r) { return transform(new (C) CmpPNode(l, r)); }
 #ifdef _LP64
-  Node* XorX(Node* l, Node* r) { return transform(new (C,3) XorLNode(l, r)); }
+  Node* XorX(Node* l, Node* r) { return transform(new (C) XorLNode(l, r)); }
 #else // _LP64
-  Node* XorX(Node* l, Node* r) { return transform(new (C,3) XorINode(l, r)); }
+  Node* XorX(Node* l, Node* r) { return transform(new (C) XorINode(l, r)); }
 #endif // _LP64
-  Node* URShiftX(Node* l, Node* r) { return transform(new (C,3) URShiftXNode(l, r)); }
+  Node* URShiftX(Node* l, Node* r) { return transform(new (C) URShiftXNode(l, r)); }
   Node* ConX(jint k) { return (Node*)gvn().MakeConX(k); }
-  Node* CastPX(Node* ctl, Node* p) { return transform(new (C,2) CastP2XNode(ctl, p)); }
+  Node* CastPX(Node* ctl, Node* p) { return transform(new (C) CastP2XNode(ctl, p)); }
   // Add a fixed offset to a pointer
   Node* basic_plus_adr(Node* base, Node* ptr, intptr_t offset);
 
diff --git a/src/share/vm/opto/ifnode.cpp b/src/share/vm/opto/ifnode.cpp
index c48e982ba..bc9106b53 100644
--- a/src/share/vm/opto/ifnode.cpp
+++ b/src/share/vm/opto/ifnode.cpp
@@ -238,10 +238,10 @@ static Node* split_if(IfNode *iff, PhaseIterGVN *igvn) {
   Node* predicate_x = NULL;
   bool counted_loop = r->is_CountedLoop();
 
-  Node *region_c = new (igvn->C, req_c + 1) RegionNode(req_c + 1);
+  Node *region_c = new (igvn->C) RegionNode(req_c + 1);
   Node *phi_c    = con1;
   uint  len      = r->req();
-  Node *region_x = new (igvn->C, len - req_c) RegionNode(len - req_c);
+  Node *region_x = new (igvn->C) RegionNode(len - req_c);
   Node *phi_x    = PhiNode::make_blank(region_x, phi);
   for (uint i = 1, i_c = 1, i_x = 1; i < len; i++) {
     if (phi->in(i) == con1) {
@@ -272,7 +272,7 @@ static Node* split_if(IfNode *iff, PhaseIterGVN *igvn) {
   // Prevent the untimely death of phi_x.  Currently he has no uses.  He is
   // about to get one.  If this only use goes away, then phi_x will look dead.
   // However, he will be picking up some more uses down below.
-  Node *hook = new (igvn->C, 4) Node(4);
+  Node *hook = new (igvn->C) Node(4);
   hook->init_req(0, phi_x);
   hook->init_req(1, phi_c);
   phi_x = phase->transform( phi_x );
@@ -284,30 +284,30 @@ static Node* split_if(IfNode *iff, PhaseIterGVN *igvn) {
   cmp_x->set_req(2,con2);
   cmp_x = phase->transform(cmp_x);
   // Make the bool
-  Node *b_c = phase->transform(new (igvn->C, 2) BoolNode(cmp_c,b->_test._test));
-  Node *b_x = phase->transform(new (igvn->C, 2) BoolNode(cmp_x,b->_test._test));
+  Node *b_c = phase->transform(new (igvn->C) BoolNode(cmp_c,b->_test._test));
+  Node *b_x = phase->transform(new (igvn->C) BoolNode(cmp_x,b->_test._test));
   // Make the IfNode
-  IfNode *iff_c = new (igvn->C, 2) IfNode(region_c,b_c,iff->_prob,iff->_fcnt);
+  IfNode *iff_c = new (igvn->C) IfNode(region_c,b_c,iff->_prob,iff->_fcnt);
   igvn->set_type_bottom(iff_c);
   igvn->_worklist.push(iff_c);
   hook->init_req(2, iff_c);
 
-  IfNode *iff_x = new (igvn->C, 2) IfNode(region_x,b_x,iff->_prob, iff->_fcnt);
+  IfNode *iff_x = new (igvn->C) IfNode(region_x,b_x,iff->_prob, iff->_fcnt);
   igvn->set_type_bottom(iff_x);
   igvn->_worklist.push(iff_x);
   hook->init_req(3, iff_x);
 
   // Make the true/false arms
-  Node *iff_c_t = phase->transform(new (igvn->C, 1) IfTrueNode (iff_c));
-  Node *iff_c_f = phase->transform(new (igvn->C, 1) IfFalseNode(iff_c));
+  Node *iff_c_t = phase->transform(new (igvn->C) IfTrueNode (iff_c));
+  Node *iff_c_f = phase->transform(new (igvn->C) IfFalseNode(iff_c));
   if (predicate_c != NULL) {
     assert(predicate_x == NULL, "only one predicate entry expected");
     // Clone loop predicates to each path
     iff_c_t = igvn->clone_loop_predicates(predicate_c, iff_c_t, !counted_loop);
     iff_c_f = igvn->clone_loop_predicates(predicate_c, iff_c_f, !counted_loop);
   }
-  Node *iff_x_t = phase->transform(new (igvn->C, 1) IfTrueNode (iff_x));
-  Node *iff_x_f = phase->transform(new (igvn->C, 1) IfFalseNode(iff_x));
+  Node *iff_x_t = phase->transform(new (igvn->C) IfTrueNode (iff_x));
+  Node *iff_x_f = phase->transform(new (igvn->C) IfFalseNode(iff_x));
   if (predicate_x != NULL) {
     assert(predicate_c == NULL, "only one predicate entry expected");
     // Clone loop predicates to each path
@@ -316,14 +316,14 @@ static Node* split_if(IfNode *iff, PhaseIterGVN *igvn) {
   }
 
   // Merge the TRUE paths
-  Node *region_s = new (igvn->C, 3) RegionNode(3);
+  Node *region_s = new (igvn->C) RegionNode(3);
   igvn->_worklist.push(region_s);
   region_s->init_req(1, iff_c_t);
   region_s->init_req(2, iff_x_t);
   igvn->register_new_node_with_optimizer( region_s );
 
   // Merge the FALSE paths
-  Node *region_f = new (igvn->C, 3) RegionNode(3);
+  Node *region_f = new (igvn->C) RegionNode(3);
   igvn->_worklist.push(region_f);
   region_f->init_req(1, iff_c_f);
   region_f->init_req(2, iff_x_f);
@@ -438,7 +438,7 @@ static Node* split_if(IfNode *iff, PhaseIterGVN *igvn) {
 
   // Must return either the original node (now dead) or a new node
   // (Do not return a top here, since that would break the uniqueness of top.)
-  return new (igvn->C, 1) ConINode(TypeInt::ZERO);
+  return new (igvn->C) ConINode(TypeInt::ZERO);
 }
 
 //------------------------------is_range_check---------------------------------
@@ -541,16 +541,16 @@ static void adjust_check(Node* proj, Node* range, Node* index,
   // Compute a new check
   Node *new_add = gvn->intcon(off_lo);
   if( index ) {
-    new_add = off_lo ? gvn->transform(new (gvn->C, 3) AddINode( index, new_add )) : index;
+    new_add = off_lo ? gvn->transform(new (gvn->C) AddINode( index, new_add )) : index;
   }
   Node *new_cmp = (flip == 1)
-    ? new (gvn->C, 3) CmpUNode( new_add, range )
-    : new (gvn->C, 3) CmpUNode( range, new_add );
+    ? new (gvn->C) CmpUNode( new_add, range )
+    : new (gvn->C) CmpUNode( range, new_add );
   new_cmp = gvn->transform(new_cmp);
   // See if no need to adjust the existing check
   if( new_cmp == cmp ) return;
   // Else, adjust existing check
-  Node *new_bol = gvn->transform( new (gvn->C, 2) BoolNode( new_cmp, bol->as_Bool()->_test._test ) );
+  Node *new_bol = gvn->transform( new (gvn->C) BoolNode( new_cmp, bol->as_Bool()->_test._test ) );
   igvn->rehash_node_delayed( iff );
   iff->set_req_X( 1, new_bol, igvn );
 }
@@ -727,9 +727,9 @@ Node* IfNode::fold_compares(PhaseGVN* phase) {
             if (failtype->_hi != max_jint && failtype->_lo != min_jint && bound > 1) {
               // Merge the two compares into a single unsigned compare by building  (CmpU (n - lo) hi)
               BoolTest::mask cond = fail->as_Proj()->_con ? BoolTest::lt : BoolTest::ge;
-              Node* adjusted = phase->transform(new (phase->C, 3) SubINode(n, phase->intcon(failtype->_lo)));
-              Node* newcmp = phase->transform(new (phase->C, 3) CmpUNode(adjusted, phase->intcon(bound)));
-              Node* newbool = phase->transform(new (phase->C, 2) BoolNode(newcmp, cond));
+              Node* adjusted = phase->transform(new (phase->C) SubINode(n, phase->intcon(failtype->_lo)));
+              Node* newcmp = phase->transform(new (phase->C) CmpUNode(adjusted, phase->intcon(bound)));
+              Node* newbool = phase->transform(new (phase->C) BoolNode(newcmp, cond));
               phase->is_IterGVN()->replace_input_of(dom_iff, 1, phase->intcon(ctrl->as_Proj()->_con));
               phase->hash_delete(this);
               set_req(1, newbool);
@@ -1002,7 +1002,7 @@ Node *IfNode::Ideal(PhaseGVN *phase, bool can_reshape) {
 
   // Must return either the original node (now dead) or a new node
   // (Do not return a top here, since that would break the uniqueness of top.)
-  return new (phase->C, 1) ConINode(TypeInt::ZERO);
+  return new (phase->C) ConINode(TypeInt::ZERO);
 }
 
 //------------------------------dominated_by-----------------------------------
@@ -1098,7 +1098,7 @@ static IfNode* idealize_test(PhaseGVN* phase, IfNode* iff) {
 
   // Flip test to be canonical.  Requires flipping the IfFalse/IfTrue and
   // cloning the IfNode.
-  Node* new_b = phase->transform( new (phase->C, 2) BoolNode(b->in(1), bt.negate()) );
+  Node* new_b = phase->transform( new (phase->C) BoolNode(b->in(1), bt.negate()) );
   if( !new_b->is_Bool() ) return NULL;
   b = new_b->as_Bool();
 
@@ -1106,7 +1106,7 @@ static IfNode* idealize_test(PhaseGVN* phase, IfNode* iff) {
   assert( igvn, "Test is not canonical in parser?" );
 
   // The IF node never really changes, but it needs to be cloned
-  iff = new (phase->C, 2) IfNode( iff->in(0), b, 1.0-iff->_prob, iff->_fcnt);
+  iff = new (phase->C) IfNode( iff->in(0), b, 1.0-iff->_prob, iff->_fcnt);
 
   Node *prior = igvn->hash_find_insert(iff);
   if( prior ) {
@@ -1119,8 +1119,8 @@ static IfNode* idealize_test(PhaseGVN* phase, IfNode* iff) {
   igvn->_worklist.push(iff);
 
   // Now handle projections.  Cloning not required.
-  Node* new_if_f = (Node*)(new (phase->C, 1) IfFalseNode( iff ));
-  Node* new_if_t = (Node*)(new (phase->C, 1) IfTrueNode ( iff ));
+  Node* new_if_f = (Node*)(new (phase->C) IfFalseNode( iff ));
+  Node* new_if_t = (Node*)(new (phase->C) IfTrueNode ( iff ));
 
   igvn->register_new_node_with_optimizer(new_if_f);
   igvn->register_new_node_with_optimizer(new_if_t);
diff --git a/src/share/vm/opto/lcm.cpp b/src/share/vm/opto/lcm.cpp
index 2f272eb55..4cca9c580 100644
--- a/src/share/vm/opto/lcm.cpp
+++ b/src/share/vm/opto/lcm.cpp
@@ -369,7 +369,7 @@ void Block::implicit_null_check(PhaseCFG *cfg, Node *proj, Node *val, int allowe
     Node *tmp2 = _nodes[end_idx()+2];
     _nodes.map(end_idx()+1, tmp2);
     _nodes.map(end_idx()+2, tmp1);
-    Node *tmp = new (C, 1) Node(C->top()); // Use not NULL input
+    Node *tmp = new (C) Node(C->top()); // Use not NULL input
     tmp1->replace_by(tmp);
     tmp2->replace_by(tmp1);
     tmp->replace_by(tmp2);
@@ -612,7 +612,7 @@ uint Block::sched_call( Matcher &matcher, Block_Array &bbs, uint node_cnt, Node_
   // Set all registers killed and not already defined by the call.
   uint r_cnt = mcall->tf()->range()->cnt();
   int op = mcall->ideal_Opcode();
-  MachProjNode *proj = new (matcher.C, 1) MachProjNode( mcall, r_cnt+1, RegMask::Empty, MachProjNode::fat_proj );
+  MachProjNode *proj = new (matcher.C) MachProjNode( mcall, r_cnt+1, RegMask::Empty, MachProjNode::fat_proj );
   bbs.map(proj->_idx,this);
   _nodes.insert(node_cnt++, proj);
 
@@ -839,7 +839,7 @@ bool Block::schedule_local(PhaseCFG *cfg, Matcher &matcher, GrowableArray<int> &
       regs.Insert(matcher.c_frame_pointer());
       regs.OR(n->out_RegMask());
 
-      MachProjNode *proj = new (matcher.C, 1) MachProjNode( n, 1, RegMask::Empty, MachProjNode::fat_proj );
+      MachProjNode *proj = new (matcher.C) MachProjNode( n, 1, RegMask::Empty, MachProjNode::fat_proj );
       cfg->_bbs.map(proj->_idx,this);
       _nodes.insert(phi_cnt++, proj);
 
diff --git a/src/share/vm/opto/library_call.cpp b/src/share/vm/opto/library_call.cpp
index 832799a58..386662e15 100644
--- a/src/share/vm/opto/library_call.cpp
+++ b/src/share/vm/opto/library_call.cpp
@@ -65,6 +65,8 @@ class LibraryCallKit : public GraphKit {
  private:
   LibraryIntrinsic* _intrinsic;   // the library intrinsic being called
 
+  const TypeOopPtr* sharpen_unsafe_type(Compile::AliasType* alias_type, const TypePtr *adr_type, bool is_native_ptr = false);
+
  public:
   LibraryCallKit(JVMState* caller, LibraryIntrinsic* intrinsic)
     : GraphKit(caller),
@@ -241,7 +243,8 @@ class LibraryCallKit : public GraphKit {
                                     Node* src,  Node* src_offset,
                                     Node* dest, Node* dest_offset,
                                     Node* copy_length, bool dest_uninitialized);
-  bool inline_unsafe_CAS(BasicType type);
+  typedef enum { LS_xadd, LS_xchg, LS_cmpxchg } LoadStoreKind;
+  bool inline_unsafe_load_store(BasicType type,  LoadStoreKind kind);
   bool inline_unsafe_ordered_store(BasicType type);
   bool inline_fp_conversions(vmIntrinsics::ID id);
   bool inline_numberOfLeadingZeros(vmIntrinsics::ID id);
@@ -290,6 +293,11 @@ CallGenerator* Compile::make_vm_intrinsic(ciMethod* m, bool is_virtual) {
     case vmIntrinsics::_compareTo:
     case vmIntrinsics::_equals:
     case vmIntrinsics::_equalsC:
+    case vmIntrinsics::_getAndAddInt:
+    case vmIntrinsics::_getAndAddLong:
+    case vmIntrinsics::_getAndSetInt:
+    case vmIntrinsics::_getAndSetLong:
+    case vmIntrinsics::_getAndSetObject:
       break;  // InlineNatives does not control String.compareTo
     case vmIntrinsics::_Reference_get:
       break;  // InlineNatives does not control Reference.get
@@ -369,6 +377,42 @@ CallGenerator* Compile::make_vm_intrinsic(ciMethod* m, bool is_virtual) {
     // across safepoint since GC can change it value.
     break;
 
+  case vmIntrinsics::_compareAndSwapObject:
+#ifdef _LP64
+    if (!UseCompressedOops && !Matcher::match_rule_supported(Op_CompareAndSwapP)) return NULL;
+#endif
+    break;
+
+  case vmIntrinsics::_compareAndSwapLong:
+    if (!Matcher::match_rule_supported(Op_CompareAndSwapL)) return NULL;
+    break;
+
+  case vmIntrinsics::_getAndAddInt:
+    if (!Matcher::match_rule_supported(Op_GetAndAddI)) return NULL;
+    break;
+
+  case vmIntrinsics::_getAndAddLong:
+    if (!Matcher::match_rule_supported(Op_GetAndAddL)) return NULL;
+    break;
+
+  case vmIntrinsics::_getAndSetInt:
+    if (!Matcher::match_rule_supported(Op_GetAndSetI)) return NULL;
+    break;
+
+  case vmIntrinsics::_getAndSetLong:
+    if (!Matcher::match_rule_supported(Op_GetAndSetL)) return NULL;
+    break;
+
+  case vmIntrinsics::_getAndSetObject:
+#ifdef _LP64
+    if (!UseCompressedOops && !Matcher::match_rule_supported(Op_GetAndSetP)) return NULL;
+    if (UseCompressedOops && !Matcher::match_rule_supported(Op_GetAndSetN)) return NULL;
+    break;
+#else
+    if (!Matcher::match_rule_supported(Op_GetAndSetP)) return NULL;
+    break;
+#endif
+
  default:
     assert(id <= vmIntrinsics::LAST_COMPILER_INLINE, "caller responsibility");
     assert(id != vmIntrinsics::_Object_init && id != vmIntrinsics::_invoke, "enum out of order?");
@@ -620,11 +664,11 @@ bool LibraryCallKit::try_to_inline() {
     return inline_unsafe_prefetch(!is_native_ptr, is_store, is_static);
 
   case vmIntrinsics::_compareAndSwapObject:
-    return inline_unsafe_CAS(T_OBJECT);
+    return inline_unsafe_load_store(T_OBJECT, LS_cmpxchg);
   case vmIntrinsics::_compareAndSwapInt:
-    return inline_unsafe_CAS(T_INT);
+    return inline_unsafe_load_store(T_INT, LS_cmpxchg);
   case vmIntrinsics::_compareAndSwapLong:
-    return inline_unsafe_CAS(T_LONG);
+    return inline_unsafe_load_store(T_LONG, LS_cmpxchg);
 
   case vmIntrinsics::_putOrderedObject:
     return inline_unsafe_ordered_store(T_OBJECT);
@@ -633,6 +677,17 @@ bool LibraryCallKit::try_to_inline() {
   case vmIntrinsics::_putOrderedLong:
     return inline_unsafe_ordered_store(T_LONG);
 
+  case vmIntrinsics::_getAndAddInt:
+    return inline_unsafe_load_store(T_INT, LS_xadd);
+  case vmIntrinsics::_getAndAddLong:
+    return inline_unsafe_load_store(T_LONG, LS_xadd);
+  case vmIntrinsics::_getAndSetInt:
+    return inline_unsafe_load_store(T_INT, LS_xchg);
+  case vmIntrinsics::_getAndSetLong:
+    return inline_unsafe_load_store(T_LONG, LS_xchg);
+  case vmIntrinsics::_getAndSetObject:
+    return inline_unsafe_load_store(T_OBJECT, LS_xchg);
+
   case vmIntrinsics::_currentThread:
     return inline_native_currentThread();
   case vmIntrinsics::_isInterrupted:
@@ -759,7 +814,7 @@ Node* LibraryCallKit::generate_guard(Node* test, RegionNode* region, float true_
 
   IfNode* iff = create_and_map_if(control(), test, true_prob, COUNT_UNKNOWN);
 
-  Node* if_slow = _gvn.transform( new (C, 1) IfTrueNode(iff) );
+  Node* if_slow = _gvn.transform( new (C) IfTrueNode(iff) );
   if (if_slow == top()) {
     // The slow branch is never taken.  No need to build this guard.
     return NULL;
@@ -768,7 +823,7 @@ Node* LibraryCallKit::generate_guard(Node* test, RegionNode* region, float true_
   if (region != NULL)
     region->add_req(if_slow);
 
-  Node* if_fast = _gvn.transform( new (C, 1) IfFalseNode(iff) );
+  Node* if_fast = _gvn.transform( new (C) IfFalseNode(iff) );
   set_control(if_fast);
 
   return if_slow;
@@ -787,12 +842,12 @@ inline Node* LibraryCallKit::generate_negative_guard(Node* index, RegionNode* re
     return NULL;                // already stopped
   if (_gvn.type(index)->higher_equal(TypeInt::POS)) // [0,maxint]
     return NULL;                // index is already adequately typed
-  Node* cmp_lt = _gvn.transform( new (C, 3) CmpINode(index, intcon(0)) );
-  Node* bol_lt = _gvn.transform( new (C, 2) BoolNode(cmp_lt, BoolTest::lt) );
+  Node* cmp_lt = _gvn.transform( new (C) CmpINode(index, intcon(0)) );
+  Node* bol_lt = _gvn.transform( new (C) BoolNode(cmp_lt, BoolTest::lt) );
   Node* is_neg = generate_guard(bol_lt, region, PROB_MIN);
   if (is_neg != NULL && pos_index != NULL) {
     // Emulate effect of Parse::adjust_map_after_if.
-    Node* ccast = new (C, 2) CastIINode(index, TypeInt::POS);
+    Node* ccast = new (C) CastIINode(index, TypeInt::POS);
     ccast->set_req(0, control());
     (*pos_index) = _gvn.transform(ccast);
   }
@@ -805,13 +860,13 @@ inline Node* LibraryCallKit::generate_nonpositive_guard(Node* index, bool never_
     return NULL;                // already stopped
   if (_gvn.type(index)->higher_equal(TypeInt::POS1)) // [1,maxint]
     return NULL;                // index is already adequately typed
-  Node* cmp_le = _gvn.transform( new (C, 3) CmpINode(index, intcon(0)) );
+  Node* cmp_le = _gvn.transform( new (C) CmpINode(index, intcon(0)) );
   BoolTest::mask le_or_eq = (never_negative ? BoolTest::eq : BoolTest::le);
-  Node* bol_le = _gvn.transform( new (C, 2) BoolNode(cmp_le, le_or_eq) );
+  Node* bol_le = _gvn.transform( new (C) BoolNode(cmp_le, le_or_eq) );
   Node* is_notp = generate_guard(bol_le, NULL, PROB_MIN);
   if (is_notp != NULL && pos_index != NULL) {
     // Emulate effect of Parse::adjust_map_after_if.
-    Node* ccast = new (C, 2) CastIINode(index, TypeInt::POS1);
+    Node* ccast = new (C) CastIINode(index, TypeInt::POS1);
     ccast->set_req(0, control());
     (*pos_index) = _gvn.transform(ccast);
   }
@@ -843,9 +898,9 @@ inline Node* LibraryCallKit::generate_limit_guard(Node* offset,
     return NULL;                // common case of whole-array copy
   Node* last = subseq_length;
   if (!zero_offset)             // last += offset
-    last = _gvn.transform( new (C, 3) AddINode(last, offset));
-  Node* cmp_lt = _gvn.transform( new (C, 3) CmpUNode(array_length, last) );
-  Node* bol_lt = _gvn.transform( new (C, 2) BoolNode(cmp_lt, BoolTest::lt) );
+    last = _gvn.transform( new (C) AddINode(last, offset));
+  Node* cmp_lt = _gvn.transform( new (C) CmpUNode(array_length, last) );
+  Node* bol_lt = _gvn.transform( new (C) BoolNode(cmp_lt, BoolTest::lt) );
   Node* is_over = generate_guard(bol_lt, region, PROB_MIN);
   return is_over;
 }
@@ -855,7 +910,7 @@ inline Node* LibraryCallKit::generate_limit_guard(Node* offset,
 Node* LibraryCallKit::generate_current_thread(Node* &tls_output) {
   ciKlass*    thread_klass = env()->Thread_klass();
   const Type* thread_type  = TypeOopPtr::make_from_klass(thread_klass)->cast_to_ptr_type(TypePtr::NotNull);
-  Node* thread = _gvn.transform(new (C, 1) ThreadLocalNode());
+  Node* thread = _gvn.transform(new (C) ThreadLocalNode());
   Node* p = basic_plus_adr(top()/*!oop*/, thread, in_bytes(JavaThread::threadObj_offset()));
   Node* threadObj = make_load(NULL, p, thread_type, T_OBJECT);
   tls_output = thread;
@@ -890,18 +945,18 @@ Node* LibraryCallKit::make_string_method_node(int opcode, Node* str1, Node* str2
     // Get length of string 2
     str2_len = load_String_length(no_ctrl, str2);
 
-    result = new (C, 6) StrIndexOfNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrIndexOfNode(control(), memory(TypeAryPtr::CHARS),
                                  str1_start, str1_len, str2_start, str2_len);
     break;
   case Op_StrComp:
     // Get length of string 2
     str2_len = load_String_length(no_ctrl, str2);
 
-    result = new (C, 6) StrCompNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrCompNode(control(), memory(TypeAryPtr::CHARS),
                                  str1_start, str1_len, str2_start, str2_len);
     break;
   case Op_StrEquals:
-    result = new (C, 5) StrEqualsNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrEqualsNode(control(), memory(TypeAryPtr::CHARS),
                                str1_start, str2_start, str1_len);
     break;
   default:
@@ -924,15 +979,15 @@ Node* LibraryCallKit::make_string_method_node(int opcode, Node* str1_start, Node
   Node* result = NULL;
   switch (opcode) {
   case Op_StrIndexOf:
-    result = new (C, 6) StrIndexOfNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrIndexOfNode(control(), memory(TypeAryPtr::CHARS),
                                  str1_start, cnt1, str2_start, cnt2);
     break;
   case Op_StrComp:
-    result = new (C, 6) StrCompNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrCompNode(control(), memory(TypeAryPtr::CHARS),
                                  str1_start, cnt1, str2_start, cnt2);
     break;
   case Op_StrEquals:
-    result = new (C, 5) StrEqualsNode(control(), memory(TypeAryPtr::CHARS),
+    result = new (C) StrEqualsNode(control(), memory(TypeAryPtr::CHARS),
                                  str1_start, str2_start, cnt1);
     break;
   default:
@@ -997,12 +1052,12 @@ bool LibraryCallKit::inline_string_equals() {
   }
 
   // paths (plus control) merge
-  RegionNode* region = new (C, 5) RegionNode(5);
-  Node* phi = new (C, 5) PhiNode(region, TypeInt::BOOL);
+  RegionNode* region = new (C) RegionNode(5);
+  Node* phi = new (C) PhiNode(region, TypeInt::BOOL);
 
   // does source == target string?
-  Node* cmp = _gvn.transform(new (C, 3) CmpPNode(receiver, argument));
-  Node* bol = _gvn.transform(new (C, 2) BoolNode(cmp, BoolTest::eq));
+  Node* cmp = _gvn.transform(new (C) CmpPNode(receiver, argument));
+  Node* bol = _gvn.transform(new (C) BoolNode(cmp, BoolTest::eq));
 
   Node* if_eq = generate_slow_guard(bol, NULL);
   if (if_eq != NULL) {
@@ -1018,8 +1073,8 @@ bool LibraryCallKit::inline_string_equals() {
     _sp += nargs;          // gen_instanceof might do an uncommon trap
     Node* inst = gen_instanceof(argument, makecon(TypeKlassPtr::make(klass)));
     _sp -= nargs;
-    Node* cmp  = _gvn.transform(new (C, 3) CmpINode(inst, intcon(1)));
-    Node* bol  = _gvn.transform(new (C, 2) BoolNode(cmp, BoolTest::ne));
+    Node* cmp  = _gvn.transform(new (C) CmpINode(inst, intcon(1)));
+    Node* bol  = _gvn.transform(new (C) BoolNode(cmp, BoolTest::ne));
 
     Node* inst_false = generate_guard(bol, NULL, PROB_MIN);
     //instanceOf == true, fallthrough
@@ -1034,7 +1089,7 @@ bool LibraryCallKit::inline_string_equals() {
     const TypeOopPtr* string_type = TypeOopPtr::make_from_klass(klass);
 
     // Properly cast the argument to String
-    argument = _gvn.transform(new (C, 2) CheckCastPPNode(control(), argument, string_type));
+    argument = _gvn.transform(new (C) CheckCastPPNode(control(), argument, string_type));
     // This path is taken only when argument's type is String:NotNull.
     argument = cast_not_null(argument, false);
 
@@ -1057,8 +1112,8 @@ bool LibraryCallKit::inline_string_equals() {
     Node* argument_cnt  = load_String_length(no_ctrl, argument);
 
     // Check for receiver count != argument count
-    Node* cmp = _gvn.transform( new(C, 3) CmpINode(receiver_cnt, argument_cnt) );
-    Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::ne) );
+    Node* cmp = _gvn.transform( new(C) CmpINode(receiver_cnt, argument_cnt) );
+    Node* bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::ne) );
     Node* if_ne = generate_slow_guard(bol, NULL);
     if (if_ne != NULL) {
       phi->init_req(4, intcon(0));
@@ -1093,7 +1148,7 @@ bool LibraryCallKit::inline_array_equals() {
   Node *argument1 = pop();
 
   Node* equals =
-    _gvn.transform(new (C, 4) AryEqNode(control(), memory(TypeAryPtr::CHARS),
+    _gvn.transform(new (C) AryEqNode(control(), memory(TypeAryPtr::CHARS),
                                         argument1, argument2) );
   push(equals);
   return true;
@@ -1268,8 +1323,8 @@ bool LibraryCallKit::inline_string_indexOf() {
     const TypeOopPtr* string_type = TypeOopPtr::make_from_klass(str_klass);
 
     // Make the merge point
-    RegionNode* result_rgn = new (C, 4) RegionNode(4);
-    Node*       result_phi = new (C, 4) PhiNode(result_rgn, TypeInt::INT);
+    RegionNode* result_rgn = new (C) RegionNode(4);
+    Node*       result_phi = new (C) PhiNode(result_rgn, TypeInt::INT);
     Node* no_ctrl  = NULL;
 
     // Get start addr of source string
@@ -1289,8 +1344,8 @@ bool LibraryCallKit::inline_string_indexOf() {
     Node* substr_cnt  = load_String_length(no_ctrl, argument);
 
     // Check for substr count > string count
-    Node* cmp = _gvn.transform( new(C, 3) CmpINode(substr_cnt, source_cnt) );
-    Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::gt) );
+    Node* cmp = _gvn.transform( new(C) CmpINode(substr_cnt, source_cnt) );
+    Node* bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::gt) );
     Node* if_gt = generate_slow_guard(bol, NULL);
     if (if_gt != NULL) {
       result_phi->init_req(2, intcon(-1));
@@ -1299,8 +1354,8 @@ bool LibraryCallKit::inline_string_indexOf() {
 
     if (!stopped()) {
       // Check for substr count == 0
-      cmp = _gvn.transform( new(C, 3) CmpINode(substr_cnt, intcon(0)) );
-      bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::eq) );
+      cmp = _gvn.transform( new(C) CmpINode(substr_cnt, intcon(0)) );
+      bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::eq) );
       Node* if_zero = generate_slow_guard(bol, NULL);
       if (if_zero != NULL) {
         result_phi->init_req(3, intcon(0));
@@ -1401,7 +1456,7 @@ bool LibraryCallKit::inline_string_indexOf() {
 Node * LibraryCallKit::pop_math_arg() {
   Node *arg = pop_pair();
   if( Matcher::strict_fp_requires_explicit_rounding && UseSSE<=1 )
-    arg = _gvn.transform( new (C, 2) RoundDoubleNode(0, arg) );
+    arg = _gvn.transform( new (C) RoundDoubleNode(0, arg) );
   return arg;
 }
 
@@ -1415,13 +1470,13 @@ bool LibraryCallKit::inline_trig(vmIntrinsics::ID id) {
 
   switch (id) {
   case vmIntrinsics::_dsin:
-    trig = _gvn.transform((Node*)new (C, 2) SinDNode(arg));
+    trig = _gvn.transform((Node*)new (C) SinDNode(arg));
     break;
   case vmIntrinsics::_dcos:
-    trig = _gvn.transform((Node*)new (C, 2) CosDNode(arg));
+    trig = _gvn.transform((Node*)new (C) CosDNode(arg));
     break;
   case vmIntrinsics::_dtan:
-    trig = _gvn.transform((Node*)new (C, 2) TanDNode(arg));
+    trig = _gvn.transform((Node*)new (C) TanDNode(arg));
     break;
   default:
     assert(false, "bad intrinsic was passed in");
@@ -1465,17 +1520,17 @@ bool LibraryCallKit::inline_trig(vmIntrinsics::ID id) {
     // probably do the math inside the SIN encoding.
 
     // Make the merge point
-    RegionNode *r = new (C, 3) RegionNode(3);
-    Node *phi = new (C, 3) PhiNode(r,Type::DOUBLE);
+    RegionNode *r = new (C) RegionNode(3);
+    Node *phi = new (C) PhiNode(r,Type::DOUBLE);
 
     // Flatten arg so we need only 1 test
-    Node *abs = _gvn.transform(new (C, 2) AbsDNode(arg));
+    Node *abs = _gvn.transform(new (C) AbsDNode(arg));
     // Node for PI/4 constant
     Node *pi4 = makecon(TypeD::make(pi_4));
     // Check PI/4 : abs(arg)
-    Node *cmp = _gvn.transform(new (C, 3) CmpDNode(pi4,abs));
+    Node *cmp = _gvn.transform(new (C) CmpDNode(pi4,abs));
     // Check: If PI/4 < abs(arg) then go slow
-    Node *bol = _gvn.transform( new (C, 2) BoolNode( cmp, BoolTest::lt ) );
+    Node *bol = _gvn.transform( new (C) BoolNode( cmp, BoolTest::lt ) );
     // Branch either way
     IfNode *iff = create_and_xform_if(control(),bol, PROB_STATIC_FREQUENT, COUNT_UNKNOWN);
     set_control(opt_iff(r,iff));
@@ -1503,7 +1558,7 @@ bool LibraryCallKit::inline_trig(vmIntrinsics::ID id) {
       break;
     }
     assert(control()->in(0) == call, "");
-    Node* slow_result = _gvn.transform(new (C, 1) ProjNode(call,TypeFunc::Parms));
+    Node* slow_result = _gvn.transform(new (C) ProjNode(call,TypeFunc::Parms));
     r->init_req(1,control());
     phi->init_req(1,slow_result);
 
@@ -1524,7 +1579,7 @@ bool LibraryCallKit::inline_trig(vmIntrinsics::ID id) {
 bool LibraryCallKit::inline_sqrt(vmIntrinsics::ID id) {
   assert(id == vmIntrinsics::_dsqrt, "Not square root");
   _sp += arg_size();        // restore stack pointer
-  push_pair(_gvn.transform(new (C, 2) SqrtDNode(0, pop_math_arg())));
+  push_pair(_gvn.transform(new (C) SqrtDNode(0, pop_math_arg())));
   return true;
 }
 
@@ -1533,7 +1588,7 @@ bool LibraryCallKit::inline_sqrt(vmIntrinsics::ID id) {
 bool LibraryCallKit::inline_abs(vmIntrinsics::ID id) {
   assert(id == vmIntrinsics::_dabs, "Not absolute value");
   _sp += arg_size();        // restore stack pointer
-  push_pair(_gvn.transform(new (C, 2) AbsDNode(pop_math_arg())));
+  push_pair(_gvn.transform(new (C) AbsDNode(pop_math_arg())));
   return true;
 }
 
@@ -1542,9 +1597,9 @@ void LibraryCallKit::finish_pow_exp(Node* result, Node* x, Node* y, const TypeFu
   //result=(result.isNaN())? funcAddr():result;
   // Check: If isNaN() by checking result!=result? then either trap
   // or go to runtime
-  Node* cmpisnan = _gvn.transform(new (C, 3) CmpDNode(result,result));
+  Node* cmpisnan = _gvn.transform(new (C) CmpDNode(result,result));
   // Build the boolean node
-  Node* bolisnum = _gvn.transform( new (C, 2) BoolNode(cmpisnan, BoolTest::eq) );
+  Node* bolisnum = _gvn.transform( new (C) BoolNode(cmpisnan, BoolTest::eq) );
 
   if (!too_many_traps(Deoptimization::Reason_intrinsic)) {
     {
@@ -1565,12 +1620,12 @@ void LibraryCallKit::finish_pow_exp(Node* result, Node* x, Node* y, const TypeFu
     // to the runtime to properly handle corner cases
 
     IfNode* iff = create_and_xform_if(control(), bolisnum, PROB_STATIC_FREQUENT, COUNT_UNKNOWN);
-    Node* if_slow = _gvn.transform( new (C, 1) IfFalseNode(iff) );
-    Node* if_fast = _gvn.transform( new (C, 1) IfTrueNode(iff) );
+    Node* if_slow = _gvn.transform( new (C) IfFalseNode(iff) );
+    Node* if_fast = _gvn.transform( new (C) IfTrueNode(iff) );
 
     if (!if_slow->is_top()) {
-      RegionNode* result_region = new(C, 3) RegionNode(3);
-      PhiNode*    result_val = new (C, 3) PhiNode(result_region, Type::DOUBLE);
+      RegionNode* result_region = new(C) RegionNode(3);
+      PhiNode*    result_val = new (C) PhiNode(result_region, Type::DOUBLE);
 
       result_region->init_req(1, if_fast);
       result_val->init_req(1, result);
@@ -1581,9 +1636,9 @@ void LibraryCallKit::finish_pow_exp(Node* result, Node* x, Node* y, const TypeFu
       Node* rt = make_runtime_call(RC_LEAF, call_type, funcAddr, funcName,
                                    no_memory_effects,
                                    x, top(), y, y ? top() : NULL);
-      Node* value = _gvn.transform(new (C, 1) ProjNode(rt, TypeFunc::Parms+0));
+      Node* value = _gvn.transform(new (C) ProjNode(rt, TypeFunc::Parms+0));
 #ifdef ASSERT
-      Node* value_top = _gvn.transform(new (C, 1) ProjNode(rt, TypeFunc::Parms+1));
+      Node* value_top = _gvn.transform(new (C) ProjNode(rt, TypeFunc::Parms+1));
       assert(value_top == top(), "second value must be top");
 #endif
 
@@ -1604,7 +1659,7 @@ bool LibraryCallKit::inline_exp(vmIntrinsics::ID id) {
 
   _sp += arg_size();        // restore stack pointer
   Node *x = pop_math_arg();
-  Node *result = _gvn.transform(new (C, 2) ExpDNode(0,x));
+  Node *result = _gvn.transform(new (C) ExpDNode(0,x));
 
   finish_pow_exp(result, x, NULL, OptoRuntime::Math_D_D_Type(), CAST_FROM_FN_PTR(address, SharedRuntime::dexp), "EXP");
 
@@ -1643,58 +1698,58 @@ bool LibraryCallKit::inline_pow(vmIntrinsics::ID id) {
 
   if (!too_many_traps(Deoptimization::Reason_intrinsic)) {
     // Short form: skip the fancy tests and just check for NaN result.
-    result = _gvn.transform( new (C, 3) PowDNode(0, x, y) );
+    result = _gvn.transform( new (C) PowDNode(0, x, y) );
   } else {
     // If this inlining ever returned NaN in the past, include all
     // checks + call to the runtime.
 
     // Set the merge point for If node with condition of (x <= 0.0)
     // There are four possible paths to region node and phi node
-    RegionNode *r = new (C, 4) RegionNode(4);
-    Node *phi = new (C, 4) PhiNode(r, Type::DOUBLE);
+    RegionNode *r = new (C) RegionNode(4);
+    Node *phi = new (C) PhiNode(r, Type::DOUBLE);
 
     // Build the first if node: if (x <= 0.0)
     // Node for 0 constant
     Node *zeronode = makecon(TypeD::ZERO);
     // Check x:0
-    Node *cmp = _gvn.transform(new (C, 3) CmpDNode(x, zeronode));
+    Node *cmp = _gvn.transform(new (C) CmpDNode(x, zeronode));
     // Check: If (x<=0) then go complex path
-    Node *bol1 = _gvn.transform( new (C, 2) BoolNode( cmp, BoolTest::le ) );
+    Node *bol1 = _gvn.transform( new (C) BoolNode( cmp, BoolTest::le ) );
     // Branch either way
     IfNode *if1 = create_and_xform_if(control(),bol1, PROB_STATIC_INFREQUENT, COUNT_UNKNOWN);
     // Fast path taken; set region slot 3
-    Node *fast_taken = _gvn.transform( new (C, 1) IfFalseNode(if1) );
+    Node *fast_taken = _gvn.transform( new (C) IfFalseNode(if1) );
     r->init_req(3,fast_taken); // Capture fast-control
 
     // Fast path not-taken, i.e. slow path
-    Node *complex_path = _gvn.transform( new (C, 1) IfTrueNode(if1) );
+    Node *complex_path = _gvn.transform( new (C) IfTrueNode(if1) );
 
     // Set fast path result
-    Node *fast_result = _gvn.transform( new (C, 3) PowDNode(0, x, y) );
+    Node *fast_result = _gvn.transform( new (C) PowDNode(0, x, y) );
     phi->init_req(3, fast_result);
 
     // Complex path
     // Build the second if node (if y is long)
     // Node for (long)y
-    Node *longy = _gvn.transform( new (C, 2) ConvD2LNode(y));
+    Node *longy = _gvn.transform( new (C) ConvD2LNode(y));
     // Node for (double)((long) y)
-    Node *doublelongy= _gvn.transform( new (C, 2) ConvL2DNode(longy));
+    Node *doublelongy= _gvn.transform( new (C) ConvL2DNode(longy));
     // Check (double)((long) y) : y
-    Node *cmplongy= _gvn.transform(new (C, 3) CmpDNode(doublelongy, y));
+    Node *cmplongy= _gvn.transform(new (C) CmpDNode(doublelongy, y));
     // Check if (y isn't long) then go to slow path
 
-    Node *bol2 = _gvn.transform( new (C, 2) BoolNode( cmplongy, BoolTest::ne ) );
+    Node *bol2 = _gvn.transform( new (C) BoolNode( cmplongy, BoolTest::ne ) );
     // Branch either way
     IfNode *if2 = create_and_xform_if(complex_path,bol2, PROB_STATIC_INFREQUENT, COUNT_UNKNOWN);
-    Node* ylong_path = _gvn.transform( new (C, 1) IfFalseNode(if2));
+    Node* ylong_path = _gvn.transform( new (C) IfFalseNode(if2));
 
-    Node *slow_path = _gvn.transform( new (C, 1) IfTrueNode(if2) );
+    Node *slow_path = _gvn.transform( new (C) IfTrueNode(if2) );
 
     // Calculate DPow(abs(x), y)*(1 & (long)y)
     // Node for constant 1
     Node *conone = longcon(1);
     // 1& (long)y
-    Node *signnode= _gvn.transform( new (C, 3) AndLNode(conone, longy) );
+    Node *signnode= _gvn.transform( new (C) AndLNode(conone, longy) );
 
     // A huge number is always even. Detect a huge number by checking
     // if y + 1 == y and set integer to be tested for parity to 0.
@@ -1702,18 +1757,18 @@ bool LibraryCallKit::inline_pow(vmIntrinsics::ID id) {
     // (long)9.223372036854776E18 = max_jlong
     // (double)(long)9.223372036854776E18 = 9.223372036854776E18
     // max_jlong is odd but 9.223372036854776E18 is even
-    Node* yplus1 = _gvn.transform( new (C, 3) AddDNode(y, makecon(TypeD::make(1))));
-    Node *cmpyplus1= _gvn.transform(new (C, 3) CmpDNode(yplus1, y));
-    Node *bolyplus1 = _gvn.transform( new (C, 2) BoolNode( cmpyplus1, BoolTest::eq ) );
+    Node* yplus1 = _gvn.transform( new (C) AddDNode(y, makecon(TypeD::make(1))));
+    Node *cmpyplus1= _gvn.transform(new (C) CmpDNode(yplus1, y));
+    Node *bolyplus1 = _gvn.transform( new (C) BoolNode( cmpyplus1, BoolTest::eq ) );
     Node* correctedsign = NULL;
     if (ConditionalMoveLimit != 0) {
       correctedsign = _gvn.transform( CMoveNode::make(C, NULL, bolyplus1, signnode, longcon(0), TypeLong::LONG));
     } else {
       IfNode *ifyplus1 = create_and_xform_if(ylong_path,bolyplus1, PROB_FAIR, COUNT_UNKNOWN);
-      RegionNode *r = new (C, 3) RegionNode(3);
-      Node *phi = new (C, 3) PhiNode(r, TypeLong::LONG);
-      r->init_req(1, _gvn.transform( new (C, 1) IfFalseNode(ifyplus1)));
-      r->init_req(2, _gvn.transform( new (C, 1) IfTrueNode(ifyplus1)));
+      RegionNode *r = new (C) RegionNode(3);
+      Node *phi = new (C) PhiNode(r, TypeLong::LONG);
+      r->init_req(1, _gvn.transform( new (C) IfFalseNode(ifyplus1)));
+      r->init_req(2, _gvn.transform( new (C) IfTrueNode(ifyplus1)));
       phi->init_req(1, signnode);
       phi->init_req(2, longcon(0));
       correctedsign = _gvn.transform(phi);
@@ -1724,25 +1779,25 @@ bool LibraryCallKit::inline_pow(vmIntrinsics::ID id) {
     // zero node
     Node *conzero = longcon(0);
     // Check (1&(long)y)==0?
-    Node *cmpeq1 = _gvn.transform(new (C, 3) CmpLNode(correctedsign, conzero));
+    Node *cmpeq1 = _gvn.transform(new (C) CmpLNode(correctedsign, conzero));
     // Check if (1&(long)y)!=0?, if so the result is negative
-    Node *bol3 = _gvn.transform( new (C, 2) BoolNode( cmpeq1, BoolTest::ne ) );
+    Node *bol3 = _gvn.transform( new (C) BoolNode( cmpeq1, BoolTest::ne ) );
     // abs(x)
-    Node *absx=_gvn.transform( new (C, 2) AbsDNode(x));
+    Node *absx=_gvn.transform( new (C) AbsDNode(x));
     // abs(x)^y
-    Node *absxpowy = _gvn.transform( new (C, 3) PowDNode(0, absx, y) );
+    Node *absxpowy = _gvn.transform( new (C) PowDNode(0, absx, y) );
     // -abs(x)^y
-    Node *negabsxpowy = _gvn.transform(new (C, 2) NegDNode (absxpowy));
+    Node *negabsxpowy = _gvn.transform(new (C) NegDNode (absxpowy));
     // (1&(long)y)==1?-DPow(abs(x), y):DPow(abs(x), y)
     Node *signresult = NULL;
     if (ConditionalMoveLimit != 0) {
       signresult = _gvn.transform( CMoveNode::make(C, NULL, bol3, absxpowy, negabsxpowy, Type::DOUBLE));
     } else {
       IfNode *ifyeven = create_and_xform_if(ylong_path,bol3, PROB_FAIR, COUNT_UNKNOWN);
-      RegionNode *r = new (C, 3) RegionNode(3);
-      Node *phi = new (C, 3) PhiNode(r, Type::DOUBLE);
-      r->init_req(1, _gvn.transform( new (C, 1) IfFalseNode(ifyeven)));
-      r->init_req(2, _gvn.transform( new (C, 1) IfTrueNode(ifyeven)));
+      RegionNode *r = new (C) RegionNode(3);
+      Node *phi = new (C) PhiNode(r, Type::DOUBLE);
+      r->init_req(1, _gvn.transform( new (C) IfFalseNode(ifyeven)));
+      r->init_req(2, _gvn.transform( new (C) IfTrueNode(ifyeven)));
       phi->init_req(1, absxpowy);
       phi->init_req(2, negabsxpowy);
       signresult = _gvn.transform(phi);
@@ -1781,10 +1836,10 @@ bool LibraryCallKit::inline_trans(vmIntrinsics::ID id) {
 
   switch (id) {
   case vmIntrinsics::_dlog:
-    trans = _gvn.transform((Node*)new (C, 2) LogDNode(arg));
+    trans = _gvn.transform((Node*)new (C) LogDNode(arg));
     break;
   case vmIntrinsics::_dlog10:
-    trans = _gvn.transform((Node*)new (C, 2) Log10DNode(arg));
+    trans = _gvn.transform((Node*)new (C) Log10DNode(arg));
     break;
   default:
     assert(false, "bad intrinsic was passed in");
@@ -1815,9 +1870,9 @@ bool LibraryCallKit::runtime_math(const TypeFunc* call_type, address funcAddr, c
   Node* trig = make_runtime_call(RC_LEAF, call_type, funcAddr, funcName,
                                  no_memory_effects,
                                  a, top(), b, b ? top() : NULL);
-  Node* value = _gvn.transform(new (C, 1) ProjNode(trig, TypeFunc::Parms+0));
+  Node* value = _gvn.transform(new (C) ProjNode(trig, TypeFunc::Parms+0));
 #ifdef ASSERT
-  Node* value_top = _gvn.transform(new (C, 1) ProjNode(trig, TypeFunc::Parms+1));
+  Node* value_top = _gvn.transform(new (C) ProjNode(trig, TypeFunc::Parms+1));
   assert(value_top == top(), "second value must be top");
 #endif
 
@@ -1908,7 +1963,7 @@ LibraryCallKit::generate_min_max(vmIntrinsics::ID id, Node* x0, Node* y0) {
   int   cmp_op = Op_CmpI;
   Node* xkey = xvalue;
   Node* ykey = yvalue;
-  Node* ideal_cmpxy = _gvn.transform( new(C, 3) CmpINode(xkey, ykey) );
+  Node* ideal_cmpxy = _gvn.transform( new(C) CmpINode(xkey, ykey) );
   if (ideal_cmpxy->is_Cmp()) {
     // E.g., if we have CmpI(length - offset, count),
     // it might idealize to CmpI(length, count + offset)
@@ -2001,7 +2056,7 @@ LibraryCallKit::generate_min_max(vmIntrinsics::ID id, Node* x0, Node* y0) {
   default:
     if (cmpxy == NULL)
       cmpxy = ideal_cmpxy;
-    best_bol = _gvn.transform( new(C, 2) BoolNode(cmpxy, BoolTest::lt) );
+    best_bol = _gvn.transform( new(C) BoolNode(cmpxy, BoolTest::lt) );
     // and fall through:
   case BoolTest::lt:          // x < y
   case BoolTest::le:          // x <= y
@@ -2061,7 +2116,7 @@ LibraryCallKit::classify_unsafe_addr(Node* &base, Node* &offset) {
     return Type::AnyPtr;
   } else if (base_type == TypePtr::NULL_PTR) {
     // Since this is a NULL+long form, we have to switch to a rawptr.
-    base   = _gvn.transform( new (C, 2) CastX2PNode(offset) );
+    base   = _gvn.transform( new (C) CastX2PNode(offset) );
     offset = MakeConX(0);
     return Type::RawPtr;
   } else if (base_type->base() == Type::RawPtr) {
@@ -2106,10 +2161,10 @@ bool LibraryCallKit::inline_numberOfLeadingZeros(vmIntrinsics::ID id) {
   _sp += arg_size();  // restore stack pointer
   switch (id) {
   case vmIntrinsics::_numberOfLeadingZeros_i:
-    push(_gvn.transform(new (C, 2) CountLeadingZerosINode(pop())));
+    push(_gvn.transform(new (C) CountLeadingZerosINode(pop())));
     break;
   case vmIntrinsics::_numberOfLeadingZeros_l:
-    push(_gvn.transform(new (C, 2) CountLeadingZerosLNode(pop_pair())));
+    push(_gvn.transform(new (C) CountLeadingZerosLNode(pop_pair())));
     break;
   default:
     ShouldNotReachHere();
@@ -2127,10 +2182,10 @@ bool LibraryCallKit::inline_numberOfTrailingZeros(vmIntrinsics::ID id) {
   _sp += arg_size();  // restore stack pointer
   switch (id) {
   case vmIntrinsics::_numberOfTrailingZeros_i:
-    push(_gvn.transform(new (C, 2) CountTrailingZerosINode(pop())));
+    push(_gvn.transform(new (C) CountTrailingZerosINode(pop())));
     break;
   case vmIntrinsics::_numberOfTrailingZeros_l:
-    push(_gvn.transform(new (C, 2) CountTrailingZerosLNode(pop_pair())));
+    push(_gvn.transform(new (C) CountTrailingZerosLNode(pop_pair())));
     break;
   default:
     ShouldNotReachHere();
@@ -2148,10 +2203,10 @@ bool LibraryCallKit::inline_bitCount(vmIntrinsics::ID id) {
   _sp += arg_size();  // restore stack pointer
   switch (id) {
   case vmIntrinsics::_bitCount_i:
-    push(_gvn.transform(new (C, 2) PopCountINode(pop())));
+    push(_gvn.transform(new (C) PopCountINode(pop())));
     break;
   case vmIntrinsics::_bitCount_l:
-    push(_gvn.transform(new (C, 2) PopCountLNode(pop_pair())));
+    push(_gvn.transform(new (C) PopCountLNode(pop_pair())));
     break;
   default:
     ShouldNotReachHere();
@@ -2175,16 +2230,16 @@ bool LibraryCallKit::inline_reverseBytes(vmIntrinsics::ID id) {
   _sp += arg_size();  // restore stack pointer
   switch (id) {
   case vmIntrinsics::_reverseBytes_i:
-    push(_gvn.transform(new (C, 2) ReverseBytesINode(0, pop())));
+    push(_gvn.transform(new (C) ReverseBytesINode(0, pop())));
     break;
   case vmIntrinsics::_reverseBytes_l:
-    push_pair(_gvn.transform(new (C, 2) ReverseBytesLNode(0, pop_pair())));
+    push_pair(_gvn.transform(new (C) ReverseBytesLNode(0, pop_pair())));
     break;
   case vmIntrinsics::_reverseBytes_c:
-    push(_gvn.transform(new (C, 2) ReverseBytesUSNode(0, pop())));
+    push(_gvn.transform(new (C) ReverseBytesUSNode(0, pop())));
     break;
   case vmIntrinsics::_reverseBytes_s:
-    push(_gvn.transform(new (C, 2) ReverseBytesSNode(0, pop())));
+    push(_gvn.transform(new (C) ReverseBytesSNode(0, pop())));
     break;
   default:
     ;
@@ -2301,6 +2356,43 @@ void LibraryCallKit::insert_pre_barrier(Node* base_oop, Node* offset,
 // Interpret Unsafe.fieldOffset cookies correctly:
 extern jlong Unsafe_field_offset_to_byte_offset(jlong field_offset);
 
+const TypeOopPtr* LibraryCallKit::sharpen_unsafe_type(Compile::AliasType* alias_type, const TypePtr *adr_type, bool is_native_ptr) {
+  // Attempt to infer a sharper value type from the offset and base type.
+  ciKlass* sharpened_klass = NULL;
+
+  // See if it is an instance field, with an object type.
+  if (alias_type->field() != NULL) {
+    assert(!is_native_ptr, "native pointer op cannot use a java address");
+    if (alias_type->field()->type()->is_klass()) {
+      sharpened_klass = alias_type->field()->type()->as_klass();
+    }
+  }
+
+  // See if it is a narrow oop array.
+  if (adr_type->isa_aryptr()) {
+    if (adr_type->offset() >= objArrayOopDesc::base_offset_in_bytes()) {
+      const TypeOopPtr *elem_type = adr_type->is_aryptr()->elem()->isa_oopptr();
+      if (elem_type != NULL) {
+        sharpened_klass = elem_type->klass();
+      }
+    }
+  }
+
+  if (sharpened_klass != NULL) {
+    const TypeOopPtr* tjp = TypeOopPtr::make_from_klass(sharpened_klass);
+
+#ifndef PRODUCT
+    if (PrintIntrinsics || PrintInlining || PrintOptoInlining) {
+      tty->print("  from base type:  ");   adr_type->dump();
+      tty->print("  sharpened value: ");   tjp->dump();
+    }
+#endif
+    // Sharpen the value type.
+    return tjp;
+  }
+  return NULL;
+}
+
 bool LibraryCallKit::inline_unsafe_access(bool is_native_ptr, bool is_store, BasicType type, bool is_volatile) {
   if (callee()->is_static())  return false;  // caller must have the capability!
 
@@ -2430,39 +2522,9 @@ bool LibraryCallKit::inline_unsafe_access(bool is_native_ptr, bool is_store, Bas
                            offset != top() && heap_base_oop != top();
 
   if (!is_store && type == T_OBJECT) {
-    // Attempt to infer a sharper value type from the offset and base type.
-    ciKlass* sharpened_klass = NULL;
-
-    // See if it is an instance field, with an object type.
-    if (alias_type->field() != NULL) {
-      assert(!is_native_ptr, "native pointer op cannot use a java address");
-      if (alias_type->field()->type()->is_klass()) {
-        sharpened_klass = alias_type->field()->type()->as_klass();
-      }
-    }
-
-    // See if it is a narrow oop array.
-    if (adr_type->isa_aryptr()) {
-      if (adr_type->offset() >= objArrayOopDesc::base_offset_in_bytes()) {
-        const TypeOopPtr *elem_type = adr_type->is_aryptr()->elem()->isa_oopptr();
-        if (elem_type != NULL) {
-          sharpened_klass = elem_type->klass();
-        }
-      }
-    }
-
-    if (sharpened_klass != NULL) {
-      const TypeOopPtr* tjp = TypeOopPtr::make_from_klass(sharpened_klass);
-
-      // Sharpen the value type.
+    const TypeOopPtr* tjp = sharpen_unsafe_type(alias_type, adr_type, is_native_ptr);
+    if (tjp != NULL) {
       value_type = tjp;
-
-#ifndef PRODUCT
-      if (PrintIntrinsics || PrintInlining || PrintOptoInlining) {
-        tty->print("  from base type:  ");   adr_type->dump();
-        tty->print("  sharpened value: "); value_type->dump();
-      }
-#endif
     }
   }
 
@@ -2522,7 +2584,7 @@ bool LibraryCallKit::inline_unsafe_access(bool is_native_ptr, bool is_store, Bas
       break;
     case T_ADDRESS:
       // Cast to an int type.
-      p = _gvn.transform( new (C, 2) CastP2XNode(NULL,p) );
+      p = _gvn.transform( new (C) CastP2XNode(NULL,p) );
       p = ConvX2L(p);
       push_pair(p);
       break;
@@ -2541,7 +2603,7 @@ bool LibraryCallKit::inline_unsafe_access(bool is_native_ptr, bool is_store, Bas
     case T_ADDRESS:
       // Repackage the long as a pointer.
       val = ConvL2X(val);
-      val = _gvn.transform( new (C, 2) CastX2PNode(val) );
+      val = _gvn.transform( new (C) CastX2PNode(val) );
       break;
     }
 
@@ -2663,9 +2725,9 @@ bool LibraryCallKit::inline_unsafe_prefetch(bool is_native_ptr, bool is_store, b
   // Generate the read or write prefetch
   Node *prefetch;
   if (is_store) {
-    prefetch = new (C, 3) PrefetchWriteNode(i_o(), adr);
+    prefetch = new (C) PrefetchWriteNode(i_o(), adr);
   } else {
-    prefetch = new (C, 3) PrefetchReadNode(i_o(), adr);
+    prefetch = new (C) PrefetchReadNode(i_o(), adr);
   }
   prefetch->init_req(0, control());
   set_i_o(_gvn.transform(prefetch));
@@ -2673,9 +2735,9 @@ bool LibraryCallKit::inline_unsafe_prefetch(bool is_native_ptr, bool is_store, b
   return true;
 }
 
-//----------------------------inline_unsafe_CAS----------------------------
+//----------------------------inline_unsafe_load_store----------------------------
 
-bool LibraryCallKit::inline_unsafe_CAS(BasicType type) {
+bool LibraryCallKit::inline_unsafe_load_store(BasicType type, LoadStoreKind kind) {
   // This basic scheme here is the same as inline_unsafe_access, but
   // differs in enough details that combining them would make the code
   // overly confusing.  (This is a true fact! I originally combined
@@ -2686,37 +2748,47 @@ bool LibraryCallKit::inline_unsafe_CAS(BasicType type) {
   if (callee()->is_static())  return false;  // caller must have the capability!
 
 #ifndef PRODUCT
+  BasicType rtype;
   {
     ResourceMark rm;
-    // Check the signatures.
     ciSignature* sig = signature();
+    rtype = sig->return_type()->basic_type();
+    if (kind == LS_xadd || kind == LS_xchg) {
+      // Check the signatures.
 #ifdef ASSERT
-    BasicType rtype = sig->return_type()->basic_type();
-    assert(rtype == T_BOOLEAN, "CAS must return boolean");
-    assert(sig->count() == 4, "CAS has 4 arguments");
-    assert(sig->type_at(0)->basic_type() == T_OBJECT, "CAS base is object");
-    assert(sig->type_at(1)->basic_type() == T_LONG, "CAS offset is long");
+      assert(rtype == type, "get and set must return the expected type");
+      assert(sig->count() == 3, "get and set has 3 arguments");
+      assert(sig->type_at(0)->basic_type() == T_OBJECT, "get and set base is object");
+      assert(sig->type_at(1)->basic_type() == T_LONG, "get and set offset is long");
+      assert(sig->type_at(2)->basic_type() == type, "get and set must take expected type as new value/delta");
 #endif // ASSERT
+    } else if (kind == LS_cmpxchg) {
+      // Check the signatures.
+#ifdef ASSERT
+      assert(rtype == T_BOOLEAN, "CAS must return boolean");
+      assert(sig->count() == 4, "CAS has 4 arguments");
+      assert(sig->type_at(0)->basic_type() == T_OBJECT, "CAS base is object");
+      assert(sig->type_at(1)->basic_type() == T_LONG, "CAS offset is long");
+#endif // ASSERT
+    } else {
+      ShouldNotReachHere();
+    }
   }
 #endif //PRODUCT
 
   // number of stack slots per value argument (1 or 2)
   int type_words = type2size[type];
 
-  // Cannot inline wide CAS on machines that don't support it natively
-  if (type2aelembytes(type) > BytesPerInt && !VM_Version::supports_cx8())
-    return false;
-
   C->set_has_unsafe_access(true);  // Mark eventual nmethod as "unsafe".
 
-  // Argument words:  "this" plus oop plus offset plus oldvalue plus newvalue;
-  int nargs = 1 + 1 + 2  + type_words + type_words;
+  // Argument words:  "this" plus oop plus offset (plus oldvalue) plus newvalue/delta;
+  int nargs = 1 + 1 + 2  + ((kind == LS_cmpxchg) ? type_words : 0) + type_words;
 
-  // pop arguments: newval, oldval, offset, base, and receiver
+  // pop arguments: newval, offset, base, and receiver
   debug_only(int saved_sp = _sp);
   _sp += nargs;
   Node* newval   = (type_words == 1) ? pop() : pop_pair();
-  Node* oldval   = (type_words == 1) ? pop() : pop_pair();
+  Node* oldval   = (kind == LS_cmpxchg) ? ((type_words == 1) ? pop() : pop_pair()) : NULL;
   Node *offset   = pop_pair();
   Node *base     = pop();
   Node *receiver = pop();
@@ -2740,16 +2812,24 @@ bool LibraryCallKit::inline_unsafe_CAS(BasicType type) {
   Node* adr = make_unsafe_address(base, offset);
   const TypePtr *adr_type = _gvn.type(adr)->isa_ptr();
 
-  // (Unlike inline_unsafe_access, there seems no point in trying
-  // to refine types. Just use the coarse types here.
+  // For CAS, unlike inline_unsafe_access, there seems no point in
+  // trying to refine types. Just use the coarse types here.
   const Type *value_type = Type::get_const_basic_type(type);
   Compile::AliasType* alias_type = C->alias_type(adr_type);
   assert(alias_type->index() != Compile::AliasIdxBot, "no bare pointers here");
+
+  if (kind == LS_xchg && type == T_OBJECT) {
+    const TypeOopPtr* tjp = sharpen_unsafe_type(alias_type, adr_type);
+    if (tjp != NULL) {
+      value_type = tjp;
+    }
+  }
+
   int alias_idx = C->get_alias_index(adr_type);
 
-  // Memory-model-wise, a CAS acts like a little synchronized block,
-  // so needs barriers on each side.  These don't translate into
-  // actual barriers on most machines, but we still need rest of
+  // Memory-model-wise, a LoadStore acts like a little synchronized
+  // block, so needs barriers on each side.  These don't translate
+  // into actual barriers on most machines, but we still need rest of
   // compiler to respect ordering.
 
   insert_mem_bar(Op_MemBarRelease);
@@ -2762,13 +2842,29 @@ bool LibraryCallKit::inline_unsafe_CAS(BasicType type) {
 
   // For now, we handle only those cases that actually exist: ints,
   // longs, and Object. Adding others should be straightforward.
-  Node* cas;
+  Node* load_store;
   switch(type) {
   case T_INT:
-    cas = _gvn.transform(new (C, 5) CompareAndSwapINode(control(), mem, adr, newval, oldval));
+    if (kind == LS_xadd) {
+      load_store = _gvn.transform(new (C) GetAndAddINode(control(), mem, adr, newval, adr_type));
+    } else if (kind == LS_xchg) {
+      load_store = _gvn.transform(new (C) GetAndSetINode(control(), mem, adr, newval, adr_type));
+    } else if (kind == LS_cmpxchg) {
+      load_store = _gvn.transform(new (C) CompareAndSwapINode(control(), mem, adr, newval, oldval));
+    } else {
+      ShouldNotReachHere();
+    }
     break;
   case T_LONG:
-    cas = _gvn.transform(new (C, 5) CompareAndSwapLNode(control(), mem, adr, newval, oldval));
+    if (kind == LS_xadd) {
+      load_store = _gvn.transform(new (C) GetAndAddLNode(control(), mem, adr, newval, adr_type));
+    } else if (kind == LS_xchg) {
+      load_store = _gvn.transform(new (C) GetAndSetLNode(control(), mem, adr, newval, adr_type));
+    } else if (kind == LS_cmpxchg) {
+      load_store = _gvn.transform(new (C) CompareAndSwapLNode(control(), mem, adr, newval, oldval));
+    } else {
+      ShouldNotReachHere();
+    }
     break;
   case T_OBJECT:
     // Transformation of a value which could be NULL pointer (CastPP #NULL)
@@ -2778,40 +2874,57 @@ bool LibraryCallKit::inline_unsafe_CAS(BasicType type) {
       newval = _gvn.makecon(TypePtr::NULL_PTR);
 
     // Reference stores need a store barrier.
-    // (They don't if CAS fails, but it isn't worth checking.)
     pre_barrier(true /* do_load*/,
                 control(), base, adr, alias_idx, newval, value_type->make_oopptr(),
                 NULL /* pre_val*/,
                 T_OBJECT);
 #ifdef _LP64
     if (adr->bottom_type()->is_ptr_to_narrowoop()) {
-      Node *newval_enc = _gvn.transform(new (C, 2) EncodePNode(newval, newval->bottom_type()->make_narrowoop()));
-      Node *oldval_enc = _gvn.transform(new (C, 2) EncodePNode(oldval, oldval->bottom_type()->make_narrowoop()));
-      cas = _gvn.transform(new (C, 5) CompareAndSwapNNode(control(), mem, adr,
-                                                          newval_enc, oldval_enc));
+      Node *newval_enc = _gvn.transform(new (C) EncodePNode(newval, newval->bottom_type()->make_narrowoop()));
+      if (kind == LS_xchg) {
+        load_store = _gvn.transform(new (C) GetAndSetNNode(control(), mem, adr,
+                                                              newval_enc, adr_type, value_type->make_narrowoop()));
+      } else {
+        assert(kind == LS_cmpxchg, "wrong LoadStore operation");
+        Node *oldval_enc = _gvn.transform(new (C) EncodePNode(oldval, oldval->bottom_type()->make_narrowoop()));
+        load_store = _gvn.transform(new (C) CompareAndSwapNNode(control(), mem, adr,
+                                                                   newval_enc, oldval_enc));
+      }
     } else
 #endif
     {
-      cas = _gvn.transform(new (C, 5) CompareAndSwapPNode(control(), mem, adr, newval, oldval));
+      if (kind == LS_xchg) {
+        load_store = _gvn.transform(new (C) GetAndSetPNode(control(), mem, adr, newval, adr_type, value_type->is_oopptr()));
+      } else {
+        assert(kind == LS_cmpxchg, "wrong LoadStore operation");
+        load_store = _gvn.transform(new (C) CompareAndSwapPNode(control(), mem, adr, newval, oldval));
+      }
     }
-    post_barrier(control(), cas, base, adr, alias_idx, newval, T_OBJECT, true);
+    post_barrier(control(), load_store, base, adr, alias_idx, newval, T_OBJECT, true);
     break;
   default:
     ShouldNotReachHere();
     break;
   }
 
-  // SCMemProjNodes represent the memory state of CAS. Their main
-  // role is to prevent CAS nodes from being optimized away when their
-  // results aren't used.
-  Node* proj = _gvn.transform( new (C, 1) SCMemProjNode(cas));
+  // SCMemProjNodes represent the memory state of a LoadStore. Their
+  // main role is to prevent LoadStore nodes from being optimized away
+  // when their results aren't used.
+  Node* proj = _gvn.transform( new (C) SCMemProjNode(load_store));
   set_memory(proj, alias_idx);
 
   // Add the trailing membar surrounding the access
   insert_mem_bar(Op_MemBarCPUOrder);
   insert_mem_bar(Op_MemBarAcquire);
 
-  push(cas);
+#ifdef _LP64
+  if (type == T_OBJECT && adr->bottom_type()->is_ptr_to_narrowoop() && kind == LS_xchg) {
+    load_store = _gvn.transform(new (C) DecodeNNode(load_store, load_store->bottom_type()->make_ptr()));
+  }
+#endif
+
+  assert(type2size[load_store->bottom_type()->basic_type()] == type2size[rtype], "result type should match");
+  push_node(load_store->bottom_type()->basic_type(), load_store);
   return true;
 }
 
@@ -2909,7 +3022,7 @@ bool LibraryCallKit::inline_unsafe_allocate() {
   // can generate code to load it as unsigned byte.
   Node* inst = make_load(NULL, insp, TypeInt::UBYTE, T_BOOLEAN);
   Node* bits = intcon(InstanceKlass::fully_initialized);
-  Node* test = _gvn.transform( new (C, 3) SubINode(inst, bits) );
+  Node* test = _gvn.transform( new (C) SubINode(inst, bits) );
   // The 'test' is non-zero if we need to take a slow path.
 
   Node* obj = new_instance(kls, test);
@@ -2938,9 +3051,9 @@ bool LibraryCallKit::inline_native_classID() {
   Node* insp = basic_plus_adr(kls, in_bytes(offset));
   Node* tvalue = make_load(NULL, insp, TypeLong::LONG, T_LONG);
   Node* bits = longcon(~0x03l); // ignore bit 0 & 1
-  Node* andl = _gvn.transform(new (C, 3) AndLNode(tvalue, bits));
+  Node* andl = _gvn.transform(new (C) AndLNode(tvalue, bits));
   Node* clsused = longcon(0x01l); // set the class bit
-  Node* orl = _gvn.transform(new (C, 3) OrLNode(tvalue, clsused));
+  Node* orl = _gvn.transform(new (C) OrLNode(tvalue, clsused));
 
   const TypePtr *adr_type = _gvn.type(insp)->isa_ptr();
   store_to_memory(control(), insp, orl, T_LONG, adr_type);
@@ -2977,9 +3090,9 @@ bool LibraryCallKit::inline_native_time_funcs(address funcAddr, const char* func
   const TypeFunc *tf = OptoRuntime::void_long_Type();
   const TypePtr* no_memory_effects = NULL;
   Node* time = make_runtime_call(RC_LEAF, tf, funcAddr, funcName, no_memory_effects);
-  Node* value = _gvn.transform(new (C, 1) ProjNode(time, TypeFunc::Parms+0));
+  Node* value = _gvn.transform(new (C) ProjNode(time, TypeFunc::Parms+0));
 #ifdef ASSERT
-  Node* value_top = _gvn.transform(new (C, 1) ProjNode(time, TypeFunc::Parms + 1));
+  Node* value_top = _gvn.transform(new (C) ProjNode(time, TypeFunc::Parms + 1));
   assert(value_top == top(), "second value must be top");
 #endif
   push_pair(value);
@@ -3008,10 +3121,10 @@ bool LibraryCallKit::inline_native_isInterrupted() {
 
   // We only go to the fast case code if we pass two guards.
   // Paths which do not pass are accumulated in the slow_region.
-  RegionNode* slow_region = new (C, 1) RegionNode(1);
+  RegionNode* slow_region = new (C) RegionNode(1);
   record_for_igvn(slow_region);
-  RegionNode* result_rgn = new (C, 4) RegionNode(1+3); // fast1, fast2, slow
-  PhiNode*    result_val = new (C, 4) PhiNode(result_rgn, TypeInt::BOOL);
+  RegionNode* result_rgn = new (C) RegionNode(1+3); // fast1, fast2, slow
+  PhiNode*    result_val = new (C) PhiNode(result_rgn, TypeInt::BOOL);
   enum { no_int_result_path   = 1,
          no_clear_result_path = 2,
          slow_result_path     = 3
@@ -3021,8 +3134,8 @@ bool LibraryCallKit::inline_native_isInterrupted() {
   Node* rec_thr = argument(0);
   Node* tls_ptr = NULL;
   Node* cur_thr = generate_current_thread(tls_ptr);
-  Node* cmp_thr = _gvn.transform( new (C, 3) CmpPNode(cur_thr, rec_thr) );
-  Node* bol_thr = _gvn.transform( new (C, 2) BoolNode(cmp_thr, BoolTest::ne) );
+  Node* cmp_thr = _gvn.transform( new (C) CmpPNode(cur_thr, rec_thr) );
+  Node* bol_thr = _gvn.transform( new (C) BoolNode(cmp_thr, BoolTest::ne) );
 
   bool known_current_thread = (_gvn.type(bol_thr) == TypeInt::ZERO);
   if (!known_current_thread)
@@ -3034,32 +3147,32 @@ bool LibraryCallKit::inline_native_isInterrupted() {
   p = basic_plus_adr(top()/*!oop*/, osthread, in_bytes(OSThread::interrupted_offset()));
   // Set the control input on the field _interrupted read to prevent it floating up.
   Node* int_bit = make_load(control(), p, TypeInt::BOOL, T_INT);
-  Node* cmp_bit = _gvn.transform( new (C, 3) CmpINode(int_bit, intcon(0)) );
-  Node* bol_bit = _gvn.transform( new (C, 2) BoolNode(cmp_bit, BoolTest::ne) );
+  Node* cmp_bit = _gvn.transform( new (C) CmpINode(int_bit, intcon(0)) );
+  Node* bol_bit = _gvn.transform( new (C) BoolNode(cmp_bit, BoolTest::ne) );
 
   IfNode* iff_bit = create_and_map_if(control(), bol_bit, PROB_UNLIKELY_MAG(3), COUNT_UNKNOWN);
 
   // First fast path:  if (!TLS._interrupted) return false;
-  Node* false_bit = _gvn.transform( new (C, 1) IfFalseNode(iff_bit) );
+  Node* false_bit = _gvn.transform( new (C) IfFalseNode(iff_bit) );
   result_rgn->init_req(no_int_result_path, false_bit);
   result_val->init_req(no_int_result_path, intcon(0));
 
   // drop through to next case
-  set_control( _gvn.transform(new (C, 1) IfTrueNode(iff_bit)) );
+  set_control( _gvn.transform(new (C) IfTrueNode(iff_bit)) );
 
   // (c) Or, if interrupt bit is set and clear_int is false, use 2nd fast path.
   Node* clr_arg = argument(1);
-  Node* cmp_arg = _gvn.transform( new (C, 3) CmpINode(clr_arg, intcon(0)) );
-  Node* bol_arg = _gvn.transform( new (C, 2) BoolNode(cmp_arg, BoolTest::ne) );
+  Node* cmp_arg = _gvn.transform( new (C) CmpINode(clr_arg, intcon(0)) );
+  Node* bol_arg = _gvn.transform( new (C) BoolNode(cmp_arg, BoolTest::ne) );
   IfNode* iff_arg = create_and_map_if(control(), bol_arg, PROB_FAIR, COUNT_UNKNOWN);
 
   // Second fast path:  ... else if (!clear_int) return true;
-  Node* false_arg = _gvn.transform( new (C, 1) IfFalseNode(iff_arg) );
+  Node* false_arg = _gvn.transform( new (C) IfFalseNode(iff_arg) );
   result_rgn->init_req(no_clear_result_path, false_arg);
   result_val->init_req(no_clear_result_path, intcon(1));
 
   // drop through to next case
-  set_control( _gvn.transform(new (C, 1) IfTrueNode(iff_arg)) );
+  set_control( _gvn.transform(new (C) IfTrueNode(iff_arg)) );
 
   // (d) Otherwise, go to the slow path.
   slow_region->add_req(control());
@@ -3147,9 +3260,9 @@ Node* LibraryCallKit::generate_access_flags_guard(Node* kls, int modifier_mask,
   Node* mods = make_load(NULL, modp, TypeInt::INT, T_INT);
   Node* mask = intcon(modifier_mask);
   Node* bits = intcon(modifier_bits);
-  Node* mbit = _gvn.transform( new (C, 3) AndINode(mods, mask) );
-  Node* cmp  = _gvn.transform( new (C, 3) CmpINode(mbit, bits) );
-  Node* bol  = _gvn.transform( new (C, 2) BoolNode(cmp, BoolTest::ne) );
+  Node* mbit = _gvn.transform( new (C) AndINode(mods, mask) );
+  Node* cmp  = _gvn.transform( new (C) CmpINode(mbit, bits) );
+  Node* bol  = _gvn.transform( new (C) BoolNode(cmp, BoolTest::ne) );
   return generate_fair_guard(bol, region);
 }
 Node* LibraryCallKit::generate_interface_guard(Node* kls, RegionNode* region) {
@@ -3222,9 +3335,9 @@ bool LibraryCallKit::inline_native_Class_query(vmIntrinsics::ID id) {
 #endif
 
   // Null-check the mirror, and the mirror's klass ptr (in case it is a primitive).
-  RegionNode* region = new (C, PATH_LIMIT) RegionNode(PATH_LIMIT);
+  RegionNode* region = new (C) RegionNode(PATH_LIMIT);
   record_for_igvn(region);
-  PhiNode* phi = new (C, PATH_LIMIT) PhiNode(region, return_type);
+  PhiNode* phi = new (C) PhiNode(region, return_type);
 
   // The mirror will never be null of Reflection.getClassAccessFlags, however
   // it may be null for Class.isInstance or Class.getModifiers. Throw a NPE
@@ -3372,8 +3485,8 @@ bool LibraryCallKit::inline_native_subtype_check() {
     PATH_LIMIT
   };
 
-  RegionNode* region = new (C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  Node*       phi    = new (C, PATH_LIMIT) PhiNode(region, TypeInt::BOOL);
+  RegionNode* region = new (C) RegionNode(PATH_LIMIT);
+  Node*       phi    = new (C) PhiNode(region, TypeInt::BOOL);
   record_for_igvn(region);
 
   const TypePtr* adr_type = TypeRawPtr::BOTTOM;   // memory type of loads
@@ -3424,8 +3537,8 @@ bool LibraryCallKit::inline_native_subtype_check() {
   set_control(region->in(_prim_0_path)); // go back to first null check
   if (!stopped()) {
     // Since superc is primitive, make a guard for the superc==subc case.
-    Node* cmp_eq = _gvn.transform( new (C, 3) CmpPNode(args[0], args[1]) );
-    Node* bol_eq = _gvn.transform( new (C, 2) BoolNode(cmp_eq, BoolTest::eq) );
+    Node* cmp_eq = _gvn.transform( new (C) CmpPNode(args[0], args[1]) );
+    Node* bol_eq = _gvn.transform( new (C) BoolNode(cmp_eq, BoolTest::eq) );
     generate_guard(bol_eq, region, PROB_FAIR);
     if (region->req() == PATH_LIMIT+1) {
       // A guard was added.  If the added guard is taken, superc==subc.
@@ -3491,11 +3604,11 @@ Node* LibraryCallKit::generate_array_guard_common(Node* kls, RegionNode* region,
                 ? ((jint)Klass::_lh_array_tag_type_value
                    <<    Klass::_lh_array_tag_shift)
                 : Klass::_lh_neutral_value);
-  Node* cmp = _gvn.transform( new(C, 3) CmpINode(layout_val, intcon(nval)) );
+  Node* cmp = _gvn.transform( new(C) CmpINode(layout_val, intcon(nval)) );
   BoolTest::mask btest = BoolTest::lt;  // correct for testing is_[obj]array
   // invert the test if we are looking for a non-array
   if (not_array)  btest = BoolTest(btest).negate();
-  Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, btest) );
+  Node* bol = _gvn.transform( new(C) BoolNode(cmp, btest) );
   return generate_fair_guard(bol, region);
 }
 
@@ -3513,12 +3626,12 @@ bool LibraryCallKit::inline_native_newArray() {
   if (stopped())  return true;
 
   enum { _normal_path = 1, _slow_path = 2, PATH_LIMIT };
-  RegionNode* result_reg = new(C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  PhiNode*    result_val = new(C, PATH_LIMIT) PhiNode(result_reg,
-                                                      TypeInstPtr::NOTNULL);
-  PhiNode*    result_io  = new(C, PATH_LIMIT) PhiNode(result_reg, Type::ABIO);
-  PhiNode*    result_mem = new(C, PATH_LIMIT) PhiNode(result_reg, Type::MEMORY,
-                                                      TypePtr::BOTTOM);
+  RegionNode* result_reg = new(C) RegionNode(PATH_LIMIT);
+  PhiNode*    result_val = new(C) PhiNode(result_reg,
+                                          TypeInstPtr::NOTNULL);
+  PhiNode*    result_io  = new(C) PhiNode(result_reg, Type::ABIO);
+  PhiNode*    result_mem = new(C) PhiNode(result_reg, Type::MEMORY,
+                                          TypePtr::BOTTOM);
 
   bool never_see_null = !too_many_traps(Deoptimization::Reason_null_check);
   Node* klass_node = load_array_klass_from_mirror(mirror, never_see_null,
@@ -3633,7 +3746,7 @@ bool LibraryCallKit::inline_array_copyOf(bool is_copyOfRange) {
                                               NULL, 0);
     klass_node = do_null_check(klass_node, T_OBJECT);
 
-    RegionNode* bailout = new (C, 1) RegionNode(1);
+    RegionNode* bailout = new (C) RegionNode(1);
     record_for_igvn(bailout);
 
     // Despite the generic type of Arrays.copyOf, the mirror might be int, int[], etc.
@@ -3643,7 +3756,7 @@ bool LibraryCallKit::inline_array_copyOf(bool is_copyOfRange) {
       // Improve the klass node's type from the new optimistic assumption:
       ciKlass* ak = ciArrayKlass::make(env()->Object_klass());
       const Type* akls = TypeKlassPtr::make(TypePtr::NotNull, ak, 0/*offset*/);
-      Node* cast = new (C, 2) CastPPNode(klass_node, akls);
+      Node* cast = new (C) CastPPNode(klass_node, akls);
       cast->init_req(0, control());
       klass_node = _gvn.transform(cast);
     }
@@ -3654,7 +3767,7 @@ bool LibraryCallKit::inline_array_copyOf(bool is_copyOfRange) {
 
     Node* length = end;
     if (_gvn.type(start) != TypeInt::ZERO) {
-      length = _gvn.transform( new (C, 3) SubINode(end, start) );
+      length = _gvn.transform( new (C) SubINode(end, start) );
     }
 
     // Bail out if length is negative.
@@ -3674,7 +3787,7 @@ bool LibraryCallKit::inline_array_copyOf(bool is_copyOfRange) {
 
       // How many elements will we copy from the original?
       // The answer is MinI(orig_length - start, length).
-      Node* orig_tail = _gvn.transform( new(C, 3) SubINode(orig_length, start) );
+      Node* orig_tail = _gvn.transform( new(C) SubINode(orig_length, start) );
       Node* moved = generate_min_max(vmIntrinsics::_min, orig_tail, length);
 
       newcopy = new_array(klass_node, length, 0);
@@ -3721,8 +3834,8 @@ Node* LibraryCallKit::generate_virtual_guard(Node* obj_klass,
   const TypePtr* native_call_addr = TypeMetadataPtr::make(method);
 
   Node* native_call = makecon(native_call_addr);
-  Node* chk_native  = _gvn.transform( new(C, 3) CmpPNode(target_call, native_call) );
-  Node* test_native = _gvn.transform( new(C, 2) BoolNode(chk_native, BoolTest::ne) );
+  Node* chk_native  = _gvn.transform( new(C) CmpPNode(target_call, native_call) );
+  Node* test_native = _gvn.transform( new(C) BoolNode(chk_native, BoolTest::ne) );
 
   return generate_slow_guard(test_native, slow_region);
 }
@@ -3744,13 +3857,12 @@ LibraryCallKit::generate_method_call(vmIntrinsics::ID method_id, bool is_virtual
   guarantee(method_id == method->intrinsic_id(), "must match");
 
   const TypeFunc* tf = TypeFunc::make(method);
-  int tfdc = tf->domain()->cnt();
   CallJavaNode* slow_call;
   if (is_static) {
     assert(!is_virtual, "");
-    slow_call = new(C, tfdc) CallStaticJavaNode(tf,
-                                SharedRuntime::get_resolve_static_call_stub(),
-                                method, bci());
+    slow_call = new(C) CallStaticJavaNode(tf,
+                           SharedRuntime::get_resolve_static_call_stub(),
+                           method, bci());
   } else if (is_virtual) {
     null_check_receiver(method);
     int vtable_index = Method::invalid_vtable_index;
@@ -3762,12 +3874,12 @@ LibraryCallKit::generate_method_call(vmIntrinsics::ID method_id, bool is_virtual
       // No need to use the linkResolver to get it.
        vtable_index = method->vtable_index();
     }
-    slow_call = new(C, tfdc) CallDynamicJavaNode(tf,
-                                SharedRuntime::get_resolve_virtual_call_stub(),
-                                method, vtable_index, bci());
+    slow_call = new(C) CallDynamicJavaNode(tf,
+                          SharedRuntime::get_resolve_virtual_call_stub(),
+                          method, vtable_index, bci());
   } else {  // neither virtual nor static:  opt_virtual
     null_check_receiver(method);
-    slow_call = new(C, tfdc) CallStaticJavaNode(tf,
+    slow_call = new(C) CallStaticJavaNode(tf,
                                 SharedRuntime::get_resolve_opt_virtual_call_stub(),
                                 method, bci());
     slow_call->set_optimized_virtual(true);
@@ -3786,12 +3898,12 @@ bool LibraryCallKit::inline_native_hashcode(bool is_virtual, bool is_static) {
 
   enum { _slow_path = 1, _fast_path, _null_path, PATH_LIMIT };
 
-  RegionNode* result_reg = new(C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  PhiNode*    result_val = new(C, PATH_LIMIT) PhiNode(result_reg,
-                                                      TypeInt::INT);
-  PhiNode*    result_io  = new(C, PATH_LIMIT) PhiNode(result_reg, Type::ABIO);
-  PhiNode*    result_mem = new(C, PATH_LIMIT) PhiNode(result_reg, Type::MEMORY,
-                                                      TypePtr::BOTTOM);
+  RegionNode* result_reg = new(C) RegionNode(PATH_LIMIT);
+  PhiNode*    result_val = new(C) PhiNode(result_reg,
+                                          TypeInt::INT);
+  PhiNode*    result_io  = new(C) PhiNode(result_reg, Type::ABIO);
+  PhiNode*    result_mem = new(C) PhiNode(result_reg, Type::MEMORY,
+                                          TypePtr::BOTTOM);
   Node* obj = NULL;
   if (!is_static) {
     // Check for hashing null object
@@ -3825,7 +3937,7 @@ bool LibraryCallKit::inline_native_hashcode(bool is_virtual, bool is_static) {
 
   // We only go to the fast case code if we pass a number of guards.  The
   // paths which do not pass are accumulated in the slow_region.
-  RegionNode* slow_region = new (C, 1) RegionNode(1);
+  RegionNode* slow_region = new (C) RegionNode(1);
   record_for_igvn(slow_region);
 
   // If this is a virtual call, we generate a funny guard.  We pull out
@@ -3844,10 +3956,10 @@ bool LibraryCallKit::inline_native_hashcode(bool is_virtual, bool is_static) {
 
   // Test the header to see if it is unlocked.
   Node *lock_mask      = _gvn.MakeConX(markOopDesc::biased_lock_mask_in_place);
-  Node *lmasked_header = _gvn.transform( new (C, 3) AndXNode(header, lock_mask) );
+  Node *lmasked_header = _gvn.transform( new (C) AndXNode(header, lock_mask) );
   Node *unlocked_val   = _gvn.MakeConX(markOopDesc::unlocked_value);
-  Node *chk_unlocked   = _gvn.transform( new (C, 3) CmpXNode( lmasked_header, unlocked_val));
-  Node *test_unlocked  = _gvn.transform( new (C, 2) BoolNode( chk_unlocked, BoolTest::ne) );
+  Node *chk_unlocked   = _gvn.transform( new (C) CmpXNode( lmasked_header, unlocked_val));
+  Node *test_unlocked  = _gvn.transform( new (C) BoolNode( chk_unlocked, BoolTest::ne) );
 
   generate_slow_guard(test_unlocked, slow_region);
 
@@ -3857,17 +3969,17 @@ bool LibraryCallKit::inline_native_hashcode(bool is_virtual, bool is_static) {
   // vm: see markOop.hpp.
   Node *hash_mask      = _gvn.intcon(markOopDesc::hash_mask);
   Node *hash_shift     = _gvn.intcon(markOopDesc::hash_shift);
-  Node *hshifted_header= _gvn.transform( new (C, 3) URShiftXNode(header, hash_shift) );
+  Node *hshifted_header= _gvn.transform( new (C) URShiftXNode(header, hash_shift) );
   // This hack lets the hash bits live anywhere in the mark object now, as long
   // as the shift drops the relevant bits into the low 32 bits.  Note that
   // Java spec says that HashCode is an int so there's no point in capturing
   // an 'X'-sized hashcode (32 in 32-bit build or 64 in 64-bit build).
   hshifted_header      = ConvX2I(hshifted_header);
-  Node *hash_val       = _gvn.transform( new (C, 3) AndINode(hshifted_header, hash_mask) );
+  Node *hash_val       = _gvn.transform( new (C) AndINode(hshifted_header, hash_mask) );
 
   Node *no_hash_val    = _gvn.intcon(markOopDesc::no_hash);
-  Node *chk_assigned   = _gvn.transform( new (C, 3) CmpINode( hash_val, no_hash_val));
-  Node *test_assigned  = _gvn.transform( new (C, 2) BoolNode( chk_assigned, BoolTest::eq) );
+  Node *chk_assigned   = _gvn.transform( new (C) CmpINode( hash_val, no_hash_val));
+  Node *test_assigned  = _gvn.transform( new (C) BoolNode( chk_assigned, BoolTest::eq) );
 
   generate_slow_guard(test_assigned, slow_region);
 
@@ -4071,31 +4183,31 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
 
   switch (id) {
   case vmIntrinsics::_floatToRawIntBits:
-    push(_gvn.transform( new (C, 2) MoveF2INode(pop())));
+    push(_gvn.transform( new (C) MoveF2INode(pop())));
     break;
 
   case vmIntrinsics::_intBitsToFloat:
-    push(_gvn.transform( new (C, 2) MoveI2FNode(pop())));
+    push(_gvn.transform( new (C) MoveI2FNode(pop())));
     break;
 
   case vmIntrinsics::_doubleToRawLongBits:
-    push_pair(_gvn.transform( new (C, 2) MoveD2LNode(pop_pair())));
+    push_pair(_gvn.transform( new (C) MoveD2LNode(pop_pair())));
     break;
 
   case vmIntrinsics::_longBitsToDouble:
-    push_pair(_gvn.transform( new (C, 2) MoveL2DNode(pop_pair())));
+    push_pair(_gvn.transform( new (C) MoveL2DNode(pop_pair())));
     break;
 
   case vmIntrinsics::_doubleToLongBits: {
     Node* value = pop_pair();
 
     // two paths (plus control) merge in a wood
-    RegionNode *r = new (C, 3) RegionNode(3);
-    Node *phi = new (C, 3) PhiNode(r, TypeLong::LONG);
+    RegionNode *r = new (C) RegionNode(3);
+    Node *phi = new (C) PhiNode(r, TypeLong::LONG);
 
-    Node *cmpisnan = _gvn.transform( new (C, 3) CmpDNode(value, value));
+    Node *cmpisnan = _gvn.transform( new (C) CmpDNode(value, value));
     // Build the boolean node
-    Node *bolisnan = _gvn.transform( new (C, 2) BoolNode( cmpisnan, BoolTest::ne ) );
+    Node *bolisnan = _gvn.transform( new (C) BoolNode( cmpisnan, BoolTest::ne ) );
 
     // Branch either way.
     // NaN case is less traveled, which makes all the difference.
@@ -4103,7 +4215,7 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
     Node *opt_isnan = _gvn.transform(ifisnan);
     assert( opt_isnan->is_If(), "Expect an IfNode");
     IfNode *opt_ifisnan = (IfNode*)opt_isnan;
-    Node *iftrue = _gvn.transform( new (C, 1) IfTrueNode(opt_ifisnan) );
+    Node *iftrue = _gvn.transform( new (C) IfTrueNode(opt_ifisnan) );
 
     set_control(iftrue);
 
@@ -4113,10 +4225,10 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
     r->init_req(1, iftrue);
 
     // Else fall through
-    Node *iffalse = _gvn.transform( new (C, 1) IfFalseNode(opt_ifisnan) );
+    Node *iffalse = _gvn.transform( new (C) IfFalseNode(opt_ifisnan) );
     set_control(iffalse);
 
-    phi->init_req(2, _gvn.transform( new (C, 2) MoveD2LNode(value)));
+    phi->init_req(2, _gvn.transform( new (C) MoveD2LNode(value)));
     r->init_req(2, iffalse);
 
     // Post merge
@@ -4136,12 +4248,12 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
     Node* value = pop();
 
     // two paths (plus control) merge in a wood
-    RegionNode *r = new (C, 3) RegionNode(3);
-    Node *phi = new (C, 3) PhiNode(r, TypeInt::INT);
+    RegionNode *r = new (C) RegionNode(3);
+    Node *phi = new (C) PhiNode(r, TypeInt::INT);
 
-    Node *cmpisnan = _gvn.transform( new (C, 3) CmpFNode(value, value));
+    Node *cmpisnan = _gvn.transform( new (C) CmpFNode(value, value));
     // Build the boolean node
-    Node *bolisnan = _gvn.transform( new (C, 2) BoolNode( cmpisnan, BoolTest::ne ) );
+    Node *bolisnan = _gvn.transform( new (C) BoolNode( cmpisnan, BoolTest::ne ) );
 
     // Branch either way.
     // NaN case is less traveled, which makes all the difference.
@@ -4149,7 +4261,7 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
     Node *opt_isnan = _gvn.transform(ifisnan);
     assert( opt_isnan->is_If(), "Expect an IfNode");
     IfNode *opt_ifisnan = (IfNode*)opt_isnan;
-    Node *iftrue = _gvn.transform( new (C, 1) IfTrueNode(opt_ifisnan) );
+    Node *iftrue = _gvn.transform( new (C) IfTrueNode(opt_ifisnan) );
 
     set_control(iftrue);
 
@@ -4159,10 +4271,10 @@ bool LibraryCallKit::inline_fp_conversions(vmIntrinsics::ID id) {
     r->init_req(1, iftrue);
 
     // Else fall through
-    Node *iffalse = _gvn.transform( new (C, 1) IfFalseNode(opt_ifisnan) );
+    Node *iffalse = _gvn.transform( new (C) IfFalseNode(opt_ifisnan) );
     set_control(iffalse);
 
-    phi->init_req(2, _gvn.transform( new (C, 2) MoveF2INode(value)));
+    phi->init_req(2, _gvn.transform( new (C) MoveF2INode(value)));
     r->init_req(2, iffalse);
 
     // Post merge
@@ -4284,8 +4396,8 @@ void LibraryCallKit::copy_to_clone(Node* obj, Node* alloc_obj, Node* obj_size, b
 
   // Compute the length also, if needed:
   Node* countx = size;
-  countx = _gvn.transform( new (C, 3) SubXNode(countx, MakeConX(base_off)) );
-  countx = _gvn.transform( new (C, 3) URShiftXNode(countx, intcon(LogBytesPerLong) ));
+  countx = _gvn.transform( new (C) SubXNode(countx, MakeConX(base_off)) );
+  countx = _gvn.transform( new (C) URShiftXNode(countx, intcon(LogBytesPerLong) ));
 
   const TypePtr* raw_adr_type = TypeRawPtr::BOTTOM;
   bool disjoint_bases = true;
@@ -4376,12 +4488,12 @@ bool LibraryCallKit::inline_native_clone(bool is_virtual) {
       _instance_path,     // plain instance allocation, plus arrayof_long_arraycopy
       PATH_LIMIT
     };
-    RegionNode* result_reg = new(C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-    result_val             = new(C, PATH_LIMIT) PhiNode(result_reg,
-                                                        TypeInstPtr::NOTNULL);
-    PhiNode*    result_i_o = new(C, PATH_LIMIT) PhiNode(result_reg, Type::ABIO);
-    PhiNode*    result_mem = new(C, PATH_LIMIT) PhiNode(result_reg, Type::MEMORY,
-                                                        TypePtr::BOTTOM);
+    RegionNode* result_reg = new(C) RegionNode(PATH_LIMIT);
+    result_val             = new(C) PhiNode(result_reg,
+                                            TypeInstPtr::NOTNULL);
+    PhiNode*    result_i_o = new(C) PhiNode(result_reg, Type::ABIO);
+    PhiNode*    result_mem = new(C) PhiNode(result_reg, Type::MEMORY,
+                                            TypePtr::BOTTOM);
     record_for_igvn(result_reg);
 
     const TypePtr* raw_adr_type = TypeRawPtr::BOTTOM;
@@ -4437,7 +4549,7 @@ bool LibraryCallKit::inline_native_clone(bool is_virtual) {
 
     // We only go to the instance fast case code if we pass a number of guards.
     // The paths which do not pass are accumulated in the slow_region.
-    RegionNode* slow_region = new (C, 1) RegionNode(1);
+    RegionNode* slow_region = new (C) RegionNode(1);
     record_for_igvn(slow_region);
     if (!stopped()) {
       // It's an instance (we did array above).  Make the slow-path tests.
@@ -4607,7 +4719,7 @@ bool LibraryCallKit::inline_arraycopy() {
   // (8) dest_offset + length must not exceed length of dest.
   // (9) each element of an oop array must be assignable
 
-  RegionNode* slow_region = new (C, 1) RegionNode(1);
+  RegionNode* slow_region = new (C) RegionNode(1);
   record_for_igvn(slow_region);
 
   // (3) operands must not be null
@@ -4697,7 +4809,7 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
                                    RegionNode* slow_region) {
 
   if (slow_region == NULL) {
-    slow_region = new(C,1) RegionNode(1);
+    slow_region = new(C) RegionNode(1);
     record_for_igvn(slow_region);
   }
 
@@ -4745,9 +4857,9 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
          bcopy_path       = 5,  // copy primitive array by 64-bit blocks
          PATH_LIMIT       = 6
   };
-  RegionNode* result_region = new(C, PATH_LIMIT) RegionNode(PATH_LIMIT);
-  PhiNode*    result_i_o    = new(C, PATH_LIMIT) PhiNode(result_region, Type::ABIO);
-  PhiNode*    result_memory = new(C, PATH_LIMIT) PhiNode(result_region, Type::MEMORY, adr_type);
+  RegionNode* result_region = new(C) RegionNode(PATH_LIMIT);
+  PhiNode*    result_i_o    = new(C) PhiNode(result_region, Type::ABIO);
+  PhiNode*    result_memory = new(C) PhiNode(result_region, Type::MEMORY, adr_type);
   record_for_igvn(result_region);
   _gvn.set_type_bottom(result_i_o);
   _gvn.set_type_bottom(result_memory);
@@ -4821,7 +4933,7 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
     // are dest_head = dest[0..off] and dest_tail = dest[off+len..dest.length].
     Node* dest_size   = alloc->in(AllocateNode::AllocSize);
     Node* dest_length = alloc->in(AllocateNode::ALength);
-    Node* dest_tail   = _gvn.transform( new(C,3) AddINode(dest_offset,
+    Node* dest_tail   = _gvn.transform( new(C) AddINode(dest_offset,
                                                           copy_length) );
 
     // If there is a head section that needs zeroing, do it now.
@@ -4838,8 +4950,8 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
     // the copy to a more hardware-friendly word size of 64 bits.
     Node* tail_ctl = NULL;
     if (!stopped() && !dest_tail->eqv_uncast(dest_length)) {
-      Node* cmp_lt   = _gvn.transform( new(C,3) CmpINode(dest_tail, dest_length) );
-      Node* bol_lt   = _gvn.transform( new(C,2) BoolNode(cmp_lt, BoolTest::lt) );
+      Node* cmp_lt   = _gvn.transform( new(C) CmpINode(dest_tail, dest_length) );
+      Node* bol_lt   = _gvn.transform( new(C) BoolNode(cmp_lt, BoolTest::lt) );
       tail_ctl = generate_slow_guard(bol_lt, NULL);
       assert(tail_ctl != NULL || !stopped(), "must be an outcome");
     }
@@ -4873,8 +4985,8 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
                              dest_size);
       } else {
         // Make a local merge.
-        Node* done_ctl = new(C,3) RegionNode(3);
-        Node* done_mem = new(C,3) PhiNode(done_ctl, Type::MEMORY, adr_type);
+        Node* done_ctl = new(C) RegionNode(3);
+        Node* done_mem = new(C) PhiNode(done_ctl, Type::MEMORY, adr_type);
         done_ctl->init_req(1, notail_ctl);
         done_mem->init_req(1, memory(adr_type));
         generate_clear_array(adr_type, dest, basic_elem_type,
@@ -4969,21 +5081,21 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
     // Clean up after the checked call.
     // The returned value is either 0 or -1^K,
     // where K = number of partially transferred array elements.
-    Node* cmp = _gvn.transform( new(C, 3) CmpINode(checked_value, intcon(0)) );
-    Node* bol = _gvn.transform( new(C, 2) BoolNode(cmp, BoolTest::eq) );
+    Node* cmp = _gvn.transform( new(C) CmpINode(checked_value, intcon(0)) );
+    Node* bol = _gvn.transform( new(C) BoolNode(cmp, BoolTest::eq) );
     IfNode* iff = create_and_map_if(control(), bol, PROB_MAX, COUNT_UNKNOWN);
 
     // If it is 0, we are done, so transfer to the end.
-    Node* checks_done = _gvn.transform( new(C, 1) IfTrueNode(iff) );
+    Node* checks_done = _gvn.transform( new(C) IfTrueNode(iff) );
     result_region->init_req(checked_path, checks_done);
     result_i_o   ->init_req(checked_path, checked_i_o);
     result_memory->init_req(checked_path, checked_mem);
 
     // If it is not zero, merge into the slow call.
-    set_control( _gvn.transform( new(C, 1) IfFalseNode(iff) ));
-    RegionNode* slow_reg2 = new(C, 3) RegionNode(3);
-    PhiNode*    slow_i_o2 = new(C, 3) PhiNode(slow_reg2, Type::ABIO);
-    PhiNode*    slow_mem2 = new(C, 3) PhiNode(slow_reg2, Type::MEMORY, adr_type);
+    set_control( _gvn.transform( new(C) IfFalseNode(iff) ));
+    RegionNode* slow_reg2 = new(C) RegionNode(3);
+    PhiNode*    slow_i_o2 = new(C) PhiNode(slow_reg2, Type::ABIO);
+    PhiNode*    slow_mem2 = new(C) PhiNode(slow_reg2, Type::MEMORY, adr_type);
     record_for_igvn(slow_reg2);
     slow_reg2  ->init_req(1, slow_control);
     slow_i_o2  ->init_req(1, slow_i_o);
@@ -5003,16 +5115,16 @@ LibraryCallKit::generate_arraycopy(const TypePtr* adr_type,
     } else {
       // We must continue the copy exactly where it failed, or else
       // another thread might see the wrong number of writes to dest.
-      Node* checked_offset = _gvn.transform( new(C, 3) XorINode(checked_value, intcon(-1)) );
-      Node* slow_offset    = new(C, 3) PhiNode(slow_reg2, TypeInt::INT);
+      Node* checked_offset = _gvn.transform( new(C) XorINode(checked_value, intcon(-1)) );
+      Node* slow_offset    = new(C) PhiNode(slow_reg2, TypeInt::INT);
       slow_offset->init_req(1, intcon(0));
       slow_offset->init_req(2, checked_offset);
       slow_offset  = _gvn.transform(slow_offset);
 
       // Adjust the arguments by the conditionally incoming offset.
-      Node* src_off_plus  = _gvn.transform( new(C, 3) AddINode(src_offset,  slow_offset) );
-      Node* dest_off_plus = _gvn.transform( new(C, 3) AddINode(dest_offset, slow_offset) );
-      Node* length_minus  = _gvn.transform( new(C, 3) SubINode(copy_length, slow_offset) );
+      Node* src_off_plus  = _gvn.transform( new(C) AddINode(src_offset,  slow_offset) );
+      Node* dest_off_plus = _gvn.transform( new(C) AddINode(dest_offset, slow_offset) );
+      Node* length_minus  = _gvn.transform( new(C) SubINode(copy_length, slow_offset) );
 
       // Tweak the node variables to adjust the code produced below:
       src_offset  = src_off_plus;
@@ -5233,10 +5345,10 @@ LibraryCallKit::generate_clear_array(const TypePtr* adr_type,
     int      end_round = (-1 << scale) & (BytesPerLong  - 1);
     Node*    end       = ConvI2X(slice_len);
     if (scale != 0)
-      end = _gvn.transform( new(C,3) LShiftXNode(end, intcon(scale) ));
+      end = _gvn.transform( new(C) LShiftXNode(end, intcon(scale) ));
     end_base += end_round;
-    end = _gvn.transform( new(C,3) AddXNode(end, MakeConX(end_base)) );
-    end = _gvn.transform( new(C,3) AndXNode(end, MakeConX(~end_round)) );
+    end = _gvn.transform( new(C) AddXNode(end, MakeConX(end_base)) );
+    end = _gvn.transform( new(C) AndXNode(end, MakeConX(~end_round)) );
     mem = ClearArrayNode::clear_memory(control(), mem, dest,
                                        start_con, end, &_gvn);
   } else if (start_con < 0 && dest_size != top()) {
@@ -5245,8 +5357,8 @@ LibraryCallKit::generate_clear_array(const TypePtr* adr_type,
     Node* start = slice_idx;
     start = ConvI2X(start);
     if (scale != 0)
-      start = _gvn.transform( new(C,3) LShiftXNode( start, intcon(scale) ));
-    start = _gvn.transform( new(C,3) AddXNode(start, MakeConX(abase)) );
+      start = _gvn.transform( new(C) LShiftXNode( start, intcon(scale) ));
+    start = _gvn.transform( new(C) AddXNode(start, MakeConX(abase)) );
     if ((bump_bit | clear_low) != 0) {
       int to_clear = (bump_bit | clear_low);
       // Align up mod 8, then store a jint zero unconditionally
@@ -5257,14 +5369,14 @@ LibraryCallKit::generate_clear_array(const TypePtr* adr_type,
         assert((abase & to_clear) == 0, "array base must be long-aligned");
       } else {
         // Bump 'start' up to (or past) the next jint boundary:
-        start = _gvn.transform( new(C,3) AddXNode(start, MakeConX(bump_bit)) );
+        start = _gvn.transform( new(C) AddXNode(start, MakeConX(bump_bit)) );
         assert((abase & clear_low) == 0, "array base must be int-aligned");
       }
       // Round bumped 'start' down to jlong boundary in body of array.
-      start = _gvn.transform( new(C,3) AndXNode(start, MakeConX(~to_clear)) );
+      start = _gvn.transform( new(C) AndXNode(start, MakeConX(~to_clear)) );
       if (bump_bit != 0) {
         // Store a zero to the immediately preceding jint:
-        Node* x1 = _gvn.transform( new(C,3) AddXNode(start, MakeConX(-bump_bit)) );
+        Node* x1 = _gvn.transform( new(C) AddXNode(start, MakeConX(-bump_bit)) );
         Node* p1 = basic_plus_adr(dest, x1);
         mem = StoreNode::make(_gvn, control(), mem, p1, adr_type, intcon(0), T_INT);
         mem = _gvn.transform(mem);
@@ -5331,8 +5443,8 @@ LibraryCallKit::generate_block_arraycopy(const TypePtr* adr_type,
   Node* sptr  = basic_plus_adr(src,  src_off);
   Node* dptr  = basic_plus_adr(dest, dest_off);
   Node* countx = dest_size;
-  countx = _gvn.transform( new (C, 3) SubXNode(countx, MakeConX(dest_off)) );
-  countx = _gvn.transform( new (C, 3) URShiftXNode(countx, intcon(LogBytesPerLong)) );
+  countx = _gvn.transform( new (C) SubXNode(countx, MakeConX(dest_off)) );
+  countx = _gvn.transform( new (C) URShiftXNode(countx, intcon(LogBytesPerLong)) );
 
   bool disjoint_bases = true;   // since alloc != NULL
   generate_unchecked_arraycopy(adr_type, T_LONG, disjoint_bases,
@@ -5382,7 +5494,7 @@ LibraryCallKit::generate_checkcast_arraycopy(const TypePtr* adr_type,
   // super_check_offset, for the desired klass.
   int sco_offset = in_bytes(Klass::super_check_offset_offset());
   Node* p3 = basic_plus_adr(dest_elem_klass, sco_offset);
-  Node* n3 = new(C, 3) LoadINode(NULL, memory(p3), p3, _gvn.type(p3)->is_ptr());
+  Node* n3 = new(C) LoadINode(NULL, memory(p3), p3, _gvn.type(p3)->is_ptr());
   Node* check_offset = ConvI2X(_gvn.transform(n3));
   Node* check_value  = dest_elem_klass;
 
@@ -5400,7 +5512,7 @@ LibraryCallKit::generate_checkcast_arraycopy(const TypePtr* adr_type,
                                  check_offset XTOP,
                                  check_value);
 
-  return _gvn.transform(new (C, 1) ProjNode(call, TypeFunc::Parms));
+  return _gvn.transform(new (C) ProjNode(call, TypeFunc::Parms));
 }
 
 
@@ -5422,7 +5534,7 @@ LibraryCallKit::generate_generic_arraycopy(const TypePtr* adr_type,
                     copyfunc_addr, "generic_arraycopy", adr_type,
                     src, src_offset, dest, dest_offset, copy_length);
 
-  return _gvn.transform(new (C, 1) ProjNode(call, TypeFunc::Parms));
+  return _gvn.transform(new (C) ProjNode(call, TypeFunc::Parms));
 }
 
 // Helper function; generates the fast out-of-line call to an arraycopy stub.
diff --git a/src/share/vm/opto/loopPredicate.cpp b/src/share/vm/opto/loopPredicate.cpp
index 048cba894..6dd45e00e 100644
--- a/src/share/vm/opto/loopPredicate.cpp
+++ b/src/share/vm/opto/loopPredicate.cpp
@@ -159,7 +159,7 @@ ProjNode* PhaseIdealLoop::create_new_if_for_predicate(ProjNode* cont_proj, Node*
     assert(rgn->is_Call(), "must be call uct");
     CallNode* call = rgn->as_Call();
     IdealLoopTree* loop = get_loop(call);
-    rgn = new (C, 1) RegionNode(1);
+    rgn = new (C) RegionNode(1);
     rgn->add_req(uncommon_proj);
     register_control(rgn, loop, uncommon_proj);
     _igvn.hash_delete(call);
@@ -185,8 +185,8 @@ ProjNode* PhaseIdealLoop::create_new_if_for_predicate(ProjNode* cont_proj, Node*
   IfNode *new_iff = iff->clone()->as_If();
   new_iff->set_req(0, entry);
   register_control(new_iff, lp, entry);
-  Node *if_cont = new (C, 1) IfTrueNode(new_iff);
-  Node *if_uct  = new (C, 1) IfFalseNode(new_iff);
+  Node *if_cont = new (C) IfTrueNode(new_iff);
+  Node *if_uct  = new (C) IfFalseNode(new_iff);
   if (cont_proj->is_IfFalse()) {
     // Swap
     Node* tmp = if_uct; if_uct = if_cont; if_cont = tmp;
@@ -246,7 +246,7 @@ ProjNode* PhaseIterGVN::create_new_if_for_predicate(ProjNode* cont_proj, Node* n
   if (!rgn->is_Region()) { // create a region to guard the call
     assert(rgn->is_Call(), "must be call uct");
     CallNode* call = rgn->as_Call();
-    rgn = new (C, 1) RegionNode(1);
+    rgn = new (C) RegionNode(1);
     register_new_node_with_optimizer(rgn);
     rgn->add_req(uncommon_proj);
     hash_delete(call);
@@ -263,8 +263,8 @@ ProjNode* PhaseIterGVN::create_new_if_for_predicate(ProjNode* cont_proj, Node* n
   new_iff->set_req(0, new_entry);
 
   register_new_node_with_optimizer(new_iff);
-  Node *if_cont = new (C, 1) IfTrueNode(new_iff);
-  Node *if_uct  = new (C, 1) IfFalseNode(new_iff);
+  Node *if_cont = new (C) IfTrueNode(new_iff);
+  Node *if_uct  = new (C) IfFalseNode(new_iff);
   if (cont_proj->is_IfFalse()) {
     // Swap
     Node* tmp = if_uct; if_uct = if_cont; if_cont = tmp;
@@ -309,10 +309,10 @@ ProjNode* PhaseIdealLoop::clone_predicate(ProjNode* predicate_proj, Node* new_en
 
   // Match original condition since predicate's projections could be swapped.
   assert(predicate_proj->in(0)->in(1)->in(1)->Opcode()==Op_Opaque1, "must be");
-  Node* opq = new (igvn->C, 2) Opaque1Node(igvn->C, predicate_proj->in(0)->in(1)->in(1)->in(1));
+  Node* opq = new (igvn->C) Opaque1Node(igvn->C, predicate_proj->in(0)->in(1)->in(1)->in(1));
   igvn->C->add_predicate_opaq(opq);
 
-  Node* bol = new (igvn->C, 2) Conv2BNode(opq);
+  Node* bol = new (igvn->C) Conv2BNode(opq);
   if (loop_phase != NULL) {
     loop_phase->register_new_node(opq, ctrl);
     loop_phase->register_new_node(bol, ctrl);
@@ -660,11 +660,11 @@ BoolNode* PhaseIdealLoop::rc_predicate(IdealLoopTree *loop, Node* ctrl,
       // Calculate exact limit here.
       // Note, counted loop's test is '<' or '>'.
       limit = exact_limit(loop);
-      max_idx_expr = new (C, 3) SubINode(limit, stride);
+      max_idx_expr = new (C) SubINode(limit, stride);
       register_new_node(max_idx_expr, ctrl);
       if (TraceLoopPredicate) predString->print("(limit - stride) ");
     } else {
-      max_idx_expr = new (C, 3) SubINode(limit, stride);
+      max_idx_expr = new (C) SubINode(limit, stride);
       register_new_node(max_idx_expr, ctrl);
       if (TraceLoopPredicate) predString->print("(limit - stride) ");
     }
@@ -674,22 +674,22 @@ BoolNode* PhaseIdealLoop::rc_predicate(IdealLoopTree *loop, Node* ctrl,
 
   if (scale != 1) {
     ConNode* con_scale = _igvn.intcon(scale);
-    max_idx_expr = new (C, 3) MulINode(max_idx_expr, con_scale);
+    max_idx_expr = new (C) MulINode(max_idx_expr, con_scale);
     register_new_node(max_idx_expr, ctrl);
     if (TraceLoopPredicate) predString->print("* %d ", scale);
   }
 
   if (offset && (!offset->is_Con() || offset->get_int() != 0)){
-    max_idx_expr = new (C, 3) AddINode(max_idx_expr, offset);
+    max_idx_expr = new (C) AddINode(max_idx_expr, offset);
     register_new_node(max_idx_expr, ctrl);
     if (TraceLoopPredicate)
       if (offset->is_Con()) predString->print("+ %d ", offset->get_int());
       else predString->print("+ offset ");
   }
 
-  CmpUNode* cmp = new (C, 3) CmpUNode(max_idx_expr, range);
+  CmpUNode* cmp = new (C) CmpUNode(max_idx_expr, range);
   register_new_node(cmp, ctrl);
-  BoolNode* bol = new (C, 2) BoolNode(cmp, BoolTest::lt);
+  BoolNode* bol = new (C) BoolNode(cmp, BoolTest::lt);
   register_new_node(bol, ctrl);
 
   if (TraceLoopPredicate) {
@@ -805,7 +805,7 @@ bool PhaseIdealLoop::loop_predication_impl(IdealLoopTree *loop) {
       // Negate test if necessary
       bool negated = false;
       if (proj->_con != predicate_proj->_con) {
-        new_predicate_bol = new (C, 2) BoolNode(new_predicate_bol->in(1), new_predicate_bol->_test.negate());
+        new_predicate_bol = new (C) BoolNode(new_predicate_bol->in(1), new_predicate_bol->_test.negate());
         register_new_node(new_predicate_bol, ctrl);
         negated = true;
       }
diff --git a/src/share/vm/opto/loopTransform.cpp b/src/share/vm/opto/loopTransform.cpp
index 8d88242ff..ba6a86569 100644
--- a/src/share/vm/opto/loopTransform.cpp
+++ b/src/share/vm/opto/loopTransform.cpp
@@ -224,24 +224,24 @@ Node* IdealLoopTree::reassociate_add_sub(Node* n1, PhaseIdealLoop *phase) {
   if (neg_inv1) {
     Node *zero = phase->_igvn.intcon(0);
     phase->set_ctrl(zero, phase->C->root());
-    n_inv1 = new (phase->C, 3) SubINode(zero, inv1);
+    n_inv1 = new (phase->C) SubINode(zero, inv1);
     phase->register_new_node(n_inv1, inv1_c);
   } else {
     n_inv1 = inv1;
   }
   Node* inv;
   if (neg_inv2) {
-    inv = new (phase->C, 3) SubINode(n_inv1, inv2);
+    inv = new (phase->C) SubINode(n_inv1, inv2);
   } else {
-    inv = new (phase->C, 3) AddINode(n_inv1, inv2);
+    inv = new (phase->C) AddINode(n_inv1, inv2);
   }
   phase->register_new_node(inv, phase->get_early_ctrl(inv));
 
   Node* addx;
   if (neg_x) {
-    addx = new (phase->C, 3) SubINode(inv, x);
+    addx = new (phase->C) SubINode(inv, x);
   } else {
-    addx = new (phase->C, 3) AddINode(x, inv);
+    addx = new (phase->C) AddINode(x, inv);
   }
   phase->register_new_node(addx, phase->get_ctrl(x));
   phase->_igvn.replace_node(n1, addx);
@@ -932,7 +932,7 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   post_end->_prob = PROB_FAIR;
 
   // Build the main-loop normal exit.
-  IfFalseNode *new_main_exit = new (C, 1) IfFalseNode(main_end);
+  IfFalseNode *new_main_exit = new (C) IfFalseNode(main_end);
   _igvn.register_new_node_with_optimizer( new_main_exit );
   set_idom(new_main_exit, main_end, dd_main_exit );
   set_loop(new_main_exit, loop->_parent);
@@ -942,15 +942,15 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   // (the main-loop trip-counter exit value) because we will be changing
   // the exit value (via unrolling) so we cannot constant-fold away the zero
   // trip guard until all unrolling is done.
-  Node *zer_opaq = new (C, 2) Opaque1Node(C, incr);
-  Node *zer_cmp  = new (C, 3) CmpINode( zer_opaq, limit );
-  Node *zer_bol  = new (C, 2) BoolNode( zer_cmp, b_test );
+  Node *zer_opaq = new (C) Opaque1Node(C, incr);
+  Node *zer_cmp  = new (C) CmpINode( zer_opaq, limit );
+  Node *zer_bol  = new (C) BoolNode( zer_cmp, b_test );
   register_new_node( zer_opaq, new_main_exit );
   register_new_node( zer_cmp , new_main_exit );
   register_new_node( zer_bol , new_main_exit );
 
   // Build the IfNode
-  IfNode *zer_iff = new (C, 2) IfNode( new_main_exit, zer_bol, PROB_FAIR, COUNT_UNKNOWN );
+  IfNode *zer_iff = new (C) IfNode( new_main_exit, zer_bol, PROB_FAIR, COUNT_UNKNOWN );
   _igvn.register_new_node_with_optimizer( zer_iff );
   set_idom(zer_iff, new_main_exit, dd_main_exit);
   set_loop(zer_iff, loop->_parent);
@@ -960,7 +960,7 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   set_idom(main_exit, zer_iff, dd_main_exit);
   set_idom(main_exit->unique_out(), zer_iff, dd_main_exit);
   // Make the true-path, must enter the post loop
-  Node *zer_taken = new (C, 1) IfTrueNode( zer_iff );
+  Node *zer_taken = new (C) IfTrueNode( zer_iff );
   _igvn.register_new_node_with_optimizer( zer_taken );
   set_idom(zer_taken, zer_iff, dd_main_exit);
   set_loop(zer_taken, loop->_parent);
@@ -1008,7 +1008,7 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   // Find the pre-loop normal exit.
   Node* pre_exit = pre_end->proj_out(false);
   assert( pre_exit->Opcode() == Op_IfFalse, "" );
-  IfFalseNode *new_pre_exit = new (C, 1) IfFalseNode(pre_end);
+  IfFalseNode *new_pre_exit = new (C) IfFalseNode(pre_end);
   _igvn.register_new_node_with_optimizer( new_pre_exit );
   set_idom(new_pre_exit, pre_end, dd_main_head);
   set_loop(new_pre_exit, loop->_parent);
@@ -1017,15 +1017,15 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   // pre-loop, the main-loop may not execute at all.  Later in life this
   // zero-trip guard will become the minimum-trip guard when we unroll
   // the main-loop.
-  Node *min_opaq = new (C, 2) Opaque1Node(C, limit);
-  Node *min_cmp  = new (C, 3) CmpINode( pre_incr, min_opaq );
-  Node *min_bol  = new (C, 2) BoolNode( min_cmp, b_test );
+  Node *min_opaq = new (C) Opaque1Node(C, limit);
+  Node *min_cmp  = new (C) CmpINode( pre_incr, min_opaq );
+  Node *min_bol  = new (C) BoolNode( min_cmp, b_test );
   register_new_node( min_opaq, new_pre_exit );
   register_new_node( min_cmp , new_pre_exit );
   register_new_node( min_bol , new_pre_exit );
 
   // Build the IfNode (assume the main-loop is executed always).
-  IfNode *min_iff = new (C, 2) IfNode( new_pre_exit, min_bol, PROB_ALWAYS, COUNT_UNKNOWN );
+  IfNode *min_iff = new (C) IfNode( new_pre_exit, min_bol, PROB_ALWAYS, COUNT_UNKNOWN );
   _igvn.register_new_node_with_optimizer( min_iff );
   set_idom(min_iff, new_pre_exit, dd_main_head);
   set_loop(min_iff, loop->_parent);
@@ -1036,7 +1036,7 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   set_idom(pre_exit, min_iff, dd_main_head);
   set_idom(pre_exit->unique_out(), min_iff, dd_main_head);
   // Make the true-path, must enter the main loop
-  Node *min_taken = new (C, 1) IfTrueNode( min_iff );
+  Node *min_taken = new (C) IfTrueNode( min_iff );
   _igvn.register_new_node_with_optimizer( min_taken );
   set_idom(min_taken, min_iff, dd_main_head);
   set_loop(min_taken, loop->_parent);
@@ -1066,11 +1066,11 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
   // RCE and alignment may change this later.
   Node *cmp_end = pre_end->cmp_node();
   assert( cmp_end->in(2) == limit, "" );
-  Node *pre_limit = new (C, 3) AddINode( init, stride );
+  Node *pre_limit = new (C) AddINode( init, stride );
 
   // Save the original loop limit in this Opaque1 node for
   // use by range check elimination.
-  Node *pre_opaq  = new (C, 3) Opaque1Node(C, pre_limit, limit);
+  Node *pre_opaq  = new (C) Opaque1Node(C, pre_limit, limit);
 
   register_new_node( pre_limit, pre_head->in(0) );
   register_new_node( pre_opaq , pre_head->in(0) );
@@ -1095,19 +1095,19 @@ void PhaseIdealLoop::insert_pre_post_loops( IdealLoopTree *loop, Node_List &old_
     BoolTest::mask new_test = (main_end->stride_con() > 0) ? BoolTest::lt : BoolTest::gt;
     // Modify pre loop end condition
     Node* pre_bol = pre_end->in(CountedLoopEndNode::TestValue)->as_Bool();
-    BoolNode* new_bol0 = new (C, 2) BoolNode(pre_bol->in(1), new_test);
+    BoolNode* new_bol0 = new (C) BoolNode(pre_bol->in(1), new_test);
     register_new_node( new_bol0, pre_head->in(0) );
     _igvn.hash_delete(pre_end);
     pre_end->set_req(CountedLoopEndNode::TestValue, new_bol0);
     // Modify main loop guard condition
     assert(min_iff->in(CountedLoopEndNode::TestValue) == min_bol, "guard okay");
-    BoolNode* new_bol1 = new (C, 2) BoolNode(min_bol->in(1), new_test);
+    BoolNode* new_bol1 = new (C) BoolNode(min_bol->in(1), new_test);
     register_new_node( new_bol1, new_pre_exit );
     _igvn.hash_delete(min_iff);
     min_iff->set_req(CountedLoopEndNode::TestValue, new_bol1);
     // Modify main loop end condition
     BoolNode* main_bol = main_end->in(CountedLoopEndNode::TestValue)->as_Bool();
-    BoolNode* new_bol2 = new (C, 2) BoolNode(main_bol->in(1), new_test);
+    BoolNode* new_bol2 = new (C) BoolNode(main_bol->in(1), new_test);
     register_new_node( new_bol2, main_end->in(CountedLoopEndNode::TestControl) );
     _igvn.hash_delete(main_end);
     main_end->set_req(CountedLoopEndNode::TestValue, new_bol2);
@@ -1257,13 +1257,13 @@ void PhaseIdealLoop::do_unroll( IdealLoopTree *loop, Node_List &old_new, bool ad
           // zero trip guard limit will be different from loop limit.
           assert(has_ctrl(opaq), "should have it");
           Node* opaq_ctrl = get_ctrl(opaq);
-          limit = new (C, 2) Opaque2Node( C, limit );
+          limit = new (C) Opaque2Node( C, limit );
           register_new_node( limit, opaq_ctrl );
         }
         if (stride_con > 0 && ((limit_type->_lo - stride_con) < limit_type->_lo) ||
                    stride_con < 0 && ((limit_type->_hi - stride_con) > limit_type->_hi)) {
           // No underflow.
-          new_limit = new (C, 3) SubINode(limit, stride);
+          new_limit = new (C) SubINode(limit, stride);
         } else {
           // (limit - stride) may underflow.
           // Clamp the adjustment value with MININT or MAXINT:
@@ -1293,18 +1293,18 @@ void PhaseIdealLoop::do_unroll( IdealLoopTree *loop, Node_List &old_new, bool ad
             old_limit = bol->in(1)->in(1);
             // Adjust previous adjusted limit.
             adj_limit = limit->in(CMoveNode::IfFalse);
-            adj_limit = new (C, 3) SubINode(adj_limit, stride);
+            adj_limit = new (C) SubINode(adj_limit, stride);
           } else {
             old_limit = limit;
-            adj_limit = new (C, 3) SubINode(limit, stride);
+            adj_limit = new (C) SubINode(limit, stride);
           }
           assert(old_limit != NULL && adj_limit != NULL, "");
           register_new_node( adj_limit, ctrl ); // adjust amount
-          Node* adj_cmp = new (C, 3) CmpINode(old_limit, adj_limit);
+          Node* adj_cmp = new (C) CmpINode(old_limit, adj_limit);
           register_new_node( adj_cmp, ctrl );
-          Node* adj_bool = new (C, 2) BoolNode(adj_cmp, bt);
+          Node* adj_bool = new (C) BoolNode(adj_cmp, bt);
           register_new_node( adj_bool, ctrl );
-          new_limit = new (C, 4) CMoveINode(adj_bool, adj_limit, adj_max, TypeInt::INT);
+          new_limit = new (C) CMoveINode(adj_bool, adj_limit, adj_max, TypeInt::INT);
         }
         register_new_node(new_limit, ctrl);
       }
@@ -1366,24 +1366,24 @@ void PhaseIdealLoop::do_unroll( IdealLoopTree *loop, Node_List &old_new, bool ad
     // CountedLoop this is exact (stride divides limit-init exactly).
     // We are going to double the loop body, so we want to knock off any
     // odd iteration: (trip_cnt & ~1).  Then back compute a new limit.
-    Node *span = new (C, 3) SubINode( limit, init );
+    Node *span = new (C) SubINode( limit, init );
     register_new_node( span, ctrl );
-    Node *trip = new (C, 3) DivINode( 0, span, stride );
+    Node *trip = new (C) DivINode( 0, span, stride );
     register_new_node( trip, ctrl );
     Node *mtwo = _igvn.intcon(-2);
     set_ctrl(mtwo, C->root());
-    Node *rond = new (C, 3) AndINode( trip, mtwo );
+    Node *rond = new (C) AndINode( trip, mtwo );
     register_new_node( rond, ctrl );
-    Node *spn2 = new (C, 3) MulINode( rond, stride );
+    Node *spn2 = new (C) MulINode( rond, stride );
     register_new_node( spn2, ctrl );
-    new_limit = new (C, 3) AddINode( spn2, init );
+    new_limit = new (C) AddINode( spn2, init );
     register_new_node( new_limit, ctrl );
 
     // Hammer in the new limit
     Node *ctrl2 = loop_end->in(0);
-    Node *cmp2 = new (C, 3) CmpINode( loop_head->incr(), new_limit );
+    Node *cmp2 = new (C) CmpINode( loop_head->incr(), new_limit );
     register_new_node( cmp2, ctrl2 );
-    Node *bol2 = new (C, 2) BoolNode( cmp2, loop_end->test_trip() );
+    Node *bol2 = new (C) BoolNode( cmp2, loop_end->test_trip() );
     register_new_node( bol2, ctrl2 );
     _igvn.hash_delete(loop_end);
     loop_end->set_req(CountedLoopEndNode::TestValue, bol2);
@@ -1489,15 +1489,15 @@ bool IdealLoopTree::dominates_backedge(Node* ctrl) {
 // Helper function for add_constraint().
 Node* PhaseIdealLoop::adjust_limit(int stride_con, Node * scale, Node *offset, Node *rc_limit, Node *loop_limit, Node *pre_ctrl) {
   // Compute "I :: (limit-offset)/scale"
-  Node *con = new (C, 3) SubINode(rc_limit, offset);
+  Node *con = new (C) SubINode(rc_limit, offset);
   register_new_node(con, pre_ctrl);
-  Node *X = new (C, 3) DivINode(0, con, scale);
+  Node *X = new (C) DivINode(0, con, scale);
   register_new_node(X, pre_ctrl);
 
   // Adjust loop limit
   loop_limit = (stride_con > 0)
-               ? (Node*)(new (C, 3) MinINode(loop_limit, X))
-               : (Node*)(new (C, 3) MaxINode(loop_limit, X));
+               ? (Node*)(new (C) MinINode(loop_limit, X))
+               : (Node*)(new (C) MaxINode(loop_limit, X));
   register_new_node(loop_limit, pre_ctrl);
   return loop_limit;
 }
@@ -1558,9 +1558,9 @@ void PhaseIdealLoop::add_constraint( int stride_con, int scale_con, Node *offset
       // to avoid problem with scale == -1 (min_int/(-1) == min_int).
       Node* shift = _igvn.intcon(31);
       set_ctrl(shift, C->root());
-      Node* sign = new (C, 3) RShiftINode(offset, shift);
+      Node* sign = new (C) RShiftINode(offset, shift);
       register_new_node(sign, pre_ctrl);
-      offset = new (C, 3) AndINode(offset, sign);
+      offset = new (C) AndINode(offset, sign);
       register_new_node(offset, pre_ctrl);
     } else {
       assert(low_limit->get_int() == 0, "wrong low limit for range check");
@@ -1593,7 +1593,7 @@ void PhaseIdealLoop::add_constraint( int stride_con, int scale_con, Node *offset
     Node *one  = _igvn.intcon(1);
     set_ctrl(one, C->root());
 
-    Node *plus_one = new (C, 3) AddINode(offset, one);
+    Node *plus_one = new (C) AddINode(offset, one);
     register_new_node( plus_one, pre_ctrl );
     // Pass (-stride) to indicate pre_loop_cond = NOT(main_loop_cond);
     *pre_limit = adjust_limit((-stride_con), scale, plus_one, upper_limit, *pre_limit, pre_ctrl);
@@ -1611,9 +1611,9 @@ void PhaseIdealLoop::add_constraint( int stride_con, int scale_con, Node *offset
       // to avoid problem with scale == -1 (min_int/(-1) == min_int).
       Node* shift = _igvn.intcon(31);
       set_ctrl(shift, C->root());
-      Node* sign = new (C, 3) RShiftINode(plus_one, shift);
+      Node* sign = new (C) RShiftINode(plus_one, shift);
       register_new_node(sign, pre_ctrl);
-      plus_one = new (C, 3) AndINode(plus_one, sign);
+      plus_one = new (C) AndINode(plus_one, sign);
       register_new_node(plus_one, pre_ctrl);
     } else {
       assert(low_limit->get_int() == 0, "wrong low limit for range check");
@@ -1696,7 +1696,7 @@ bool PhaseIdealLoop::is_scaled_iv_plus_offset(Node* exp, Node* iv, int* p_scale,
                                    p_offset != NULL ? &offset2 : NULL, depth+1)) {
         if (p_offset != NULL) {
           Node *ctrl_off2 = get_ctrl(offset2);
-          Node* offset = new (C, 3) AddINode(offset2, exp->in(2));
+          Node* offset = new (C) AddINode(offset2, exp->in(2));
           register_new_node(offset, ctrl_off2);
           *p_offset = offset;
         }
@@ -1709,7 +1709,7 @@ bool PhaseIdealLoop::is_scaled_iv_plus_offset(Node* exp, Node* iv, int* p_scale,
         Node *zero = _igvn.intcon(0);
         set_ctrl(zero, C->root());
         Node *ctrl_off = get_ctrl(exp->in(2));
-        Node* offset = new (C, 3) SubINode(zero, exp->in(2));
+        Node* offset = new (C) SubINode(zero, exp->in(2));
         register_new_node(offset, ctrl_off);
         *p_offset = offset;
       }
@@ -1912,15 +1912,15 @@ void PhaseIdealLoop::do_range_check( IdealLoopTree *loop, Node_List &old_new ) {
         case BoolTest::ge:
           // Convert (I*scale+offset) >= Limit to (I*(-scale)+(-offset)) <= -Limit
           scale_con = -scale_con;
-          offset = new (C, 3) SubINode( zero, offset );
+          offset = new (C) SubINode( zero, offset );
           register_new_node( offset, pre_ctrl );
-          limit  = new (C, 3) SubINode( zero, limit  );
+          limit  = new (C) SubINode( zero, limit  );
           register_new_node( limit, pre_ctrl );
           // Fall into LE case
         case BoolTest::le:
           if (b_test._test != BoolTest::gt) {
             // Convert X <= Y to X < Y+1
-            limit = new (C, 3) AddINode( limit, one );
+            limit = new (C) AddINode( limit, one );
             register_new_node( limit, pre_ctrl );
           }
           // Fall into LT case
@@ -1971,8 +1971,8 @@ void PhaseIdealLoop::do_range_check( IdealLoopTree *loop, Node_List &old_new ) {
 
   // Update loop limits
   if (conditional_rc) {
-    pre_limit = (stride_con > 0) ? (Node*)new (C,3) MinINode(pre_limit, orig_limit)
-                                 : (Node*)new (C,3) MaxINode(pre_limit, orig_limit);
+    pre_limit = (stride_con > 0) ? (Node*)new (C) MinINode(pre_limit, orig_limit)
+                                 : (Node*)new (C) MaxINode(pre_limit, orig_limit);
     register_new_node(pre_limit, pre_ctrl);
   }
   _igvn.hash_delete(pre_opaq);
@@ -1987,16 +1987,16 @@ void PhaseIdealLoop::do_range_check( IdealLoopTree *loop, Node_List &old_new ) {
     Node *ctrl = get_ctrl(main_limit);
     Node *stride = cl->stride();
     Node *init = cl->init_trip();
-    Node *span = new (C, 3) SubINode(main_limit,init);
+    Node *span = new (C) SubINode(main_limit,init);
     register_new_node(span,ctrl);
     Node *rndup = _igvn.intcon(stride_con + ((stride_con>0)?-1:1));
-    Node *add = new (C, 3) AddINode(span,rndup);
+    Node *add = new (C) AddINode(span,rndup);
     register_new_node(add,ctrl);
-    Node *div = new (C, 3) DivINode(0,add,stride);
+    Node *div = new (C) DivINode(0,add,stride);
     register_new_node(div,ctrl);
-    Node *mul = new (C, 3) MulINode(div,stride);
+    Node *mul = new (C) MulINode(div,stride);
     register_new_node(mul,ctrl);
-    Node *newlim = new (C, 3) AddINode(mul,init);
+    Node *newlim = new (C) AddINode(mul,init);
     register_new_node(newlim,ctrl);
     main_limit = newlim;
   }
@@ -2167,7 +2167,7 @@ bool IdealLoopTree::policy_do_remove_empty_loop( PhaseIdealLoop *phase ) {
   }
   // Note: the final value after increment should not overflow since
   // counted loop has limit check predicate.
-  Node *final = new (phase->C, 3) SubINode( exact_limit, cl->stride() );
+  Node *final = new (phase->C) SubINode( exact_limit, cl->stride() );
   phase->register_new_node(final,cl->in(LoopNode::EntryControl));
   phase->_igvn.replace_node(phi,final);
   phase->C->set_major_progress();
@@ -2651,20 +2651,20 @@ bool PhaseIdealLoop::intrinsify_fill(IdealLoopTree* lpt) {
   // Build an expression for the beginning of the copy region
   Node* index = head->init_trip();
 #ifdef _LP64
-  index = new (C, 2) ConvI2LNode(index);
+  index = new (C) ConvI2LNode(index);
   _igvn.register_new_node_with_optimizer(index);
 #endif
   if (shift != NULL) {
     // byte arrays don't require a shift but others do.
-    index = new (C, 3) LShiftXNode(index, shift->in(2));
+    index = new (C) LShiftXNode(index, shift->in(2));
     _igvn.register_new_node_with_optimizer(index);
   }
-  index = new (C, 4) AddPNode(base, base, index);
+  index = new (C) AddPNode(base, base, index);
   _igvn.register_new_node_with_optimizer(index);
-  Node* from = new (C, 4) AddPNode(base, index, offset);
+  Node* from = new (C) AddPNode(base, index, offset);
   _igvn.register_new_node_with_optimizer(from);
   // Compute the number of elements to copy
-  Node* len = new (C, 3) SubINode(head->limit(), head->init_trip());
+  Node* len = new (C) SubINode(head->limit(), head->init_trip());
   _igvn.register_new_node_with_optimizer(len);
 
   BasicType t = store->as_Mem()->memory_type();
@@ -2681,10 +2681,10 @@ bool PhaseIdealLoop::intrinsify_fill(IdealLoopTree* lpt) {
 
   // Convert float/double to int/long for fill routines
   if (t == T_FLOAT) {
-    store_value = new (C, 2) MoveF2INode(store_value);
+    store_value = new (C) MoveF2INode(store_value);
     _igvn.register_new_node_with_optimizer(store_value);
   } else if (t == T_DOUBLE) {
-    store_value = new (C, 2) MoveD2LNode(store_value);
+    store_value = new (C) MoveD2LNode(store_value);
     _igvn.register_new_node_with_optimizer(store_value);
   }
 
@@ -2692,13 +2692,12 @@ bool PhaseIdealLoop::intrinsify_fill(IdealLoopTree* lpt) {
   Node* result_ctrl;
   Node* result_mem;
   const TypeFunc* call_type = OptoRuntime::array_fill_Type();
-  int size = call_type->domain()->cnt();
-  CallLeafNode *call = new (C, size) CallLeafNoFPNode(call_type, fill,
-                                                      fill_name, TypeAryPtr::get_array_body_type(t));
+  CallLeafNode *call = new (C) CallLeafNoFPNode(call_type, fill,
+                                                fill_name, TypeAryPtr::get_array_body_type(t));
   call->init_req(TypeFunc::Parms+0, from);
   call->init_req(TypeFunc::Parms+1, store_value);
 #ifdef _LP64
-  len = new (C, 2) ConvI2LNode(len);
+  len = new (C) ConvI2LNode(len);
   _igvn.register_new_node_with_optimizer(len);
 #endif
   call->init_req(TypeFunc::Parms+2, len);
@@ -2711,9 +2710,9 @@ bool PhaseIdealLoop::intrinsify_fill(IdealLoopTree* lpt) {
   call->init_req( TypeFunc::ReturnAdr, C->start()->proj_out(TypeFunc::ReturnAdr) );
   call->init_req( TypeFunc::FramePtr, C->start()->proj_out(TypeFunc::FramePtr) );
   _igvn.register_new_node_with_optimizer(call);
-  result_ctrl = new (C, 1) ProjNode(call,TypeFunc::Control);
+  result_ctrl = new (C) ProjNode(call,TypeFunc::Control);
   _igvn.register_new_node_with_optimizer(result_ctrl);
-  result_mem = new (C, 1) ProjNode(call,TypeFunc::Memory);
+  result_mem = new (C) ProjNode(call,TypeFunc::Memory);
   _igvn.register_new_node_with_optimizer(result_mem);
 
   // If this fill is tightly coupled to an allocation and overwrites
diff --git a/src/share/vm/opto/loopUnswitch.cpp b/src/share/vm/opto/loopUnswitch.cpp
index 47e3990fd..1470598fc 100644
--- a/src/share/vm/opto/loopUnswitch.cpp
+++ b/src/share/vm/opto/loopUnswitch.cpp
@@ -223,15 +223,15 @@ ProjNode* PhaseIdealLoop::create_slow_version_of_loop(IdealLoopTree *loop,
 
   Node *cont      = _igvn.intcon(1);
   set_ctrl(cont, C->root());
-  Node* opq       = new (C, 2) Opaque1Node(C, cont);
+  Node* opq       = new (C) Opaque1Node(C, cont);
   register_node(opq, outer_loop, entry, dom_depth(entry));
-  Node *bol       = new (C, 2) Conv2BNode(opq);
+  Node *bol       = new (C) Conv2BNode(opq);
   register_node(bol, outer_loop, entry, dom_depth(entry));
-  IfNode* iff = new (C, 2) IfNode(entry, bol, PROB_MAX, COUNT_UNKNOWN);
+  IfNode* iff = new (C) IfNode(entry, bol, PROB_MAX, COUNT_UNKNOWN);
   register_node(iff, outer_loop, entry, dom_depth(entry));
-  ProjNode* iffast = new (C, 1) IfTrueNode(iff);
+  ProjNode* iffast = new (C) IfTrueNode(iff);
   register_node(iffast, outer_loop, iff, dom_depth(iff));
-  ProjNode* ifslow = new (C, 1) IfFalseNode(iff);
+  ProjNode* ifslow = new (C) IfFalseNode(iff);
   register_node(ifslow, outer_loop, iff, dom_depth(iff));
 
   // Clone the loop body.  The clone becomes the fast loop.  The
diff --git a/src/share/vm/opto/loopnode.cpp b/src/share/vm/opto/loopnode.cpp
index 511cca449..631e39a70 100644
--- a/src/share/vm/opto/loopnode.cpp
+++ b/src/share/vm/opto/loopnode.cpp
@@ -343,7 +343,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   assert(x->Opcode() == Op_Loop, "regular loops only");
   C->print_method("Before CountedLoop", 3);
 
-  Node *hook = new (C, 6) Node(6);
+  Node *hook = new (C) Node(6);
 
   if (LoopLimitCheck) {
 
@@ -406,11 +406,11 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
     Node* bol;
 
     if (stride_con > 0) {
-      cmp_limit = new (C, 3) CmpINode(limit, _igvn.intcon(max_jint - stride_m));
-      bol = new (C, 2) BoolNode(cmp_limit, BoolTest::le);
+      cmp_limit = new (C) CmpINode(limit, _igvn.intcon(max_jint - stride_m));
+      bol = new (C) BoolNode(cmp_limit, BoolTest::le);
     } else {
-      cmp_limit = new (C, 3) CmpINode(limit, _igvn.intcon(min_jint - stride_m));
-      bol = new (C, 2) BoolNode(cmp_limit, BoolTest::ge);
+      cmp_limit = new (C) CmpINode(limit, _igvn.intcon(min_jint - stride_m));
+      bol = new (C) BoolNode(cmp_limit, BoolTest::ge);
     }
     cmp_limit = _igvn.register_new_node_with_optimizer(cmp_limit);
     bol = _igvn.register_new_node_with_optimizer(bol);
@@ -447,7 +447,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
     // is converted to
     //   i = init; do {} while(++i < limit+1);
     //
-    limit = gvn->transform(new (C, 3) AddINode(limit, stride));
+    limit = gvn->transform(new (C) AddINode(limit, stride));
   }
 
   // Now we need to canonicalize loop condition.
@@ -466,7 +466,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
     // we can convert 'i <= limit' to 'i < limit+1' since stride != 0.
     //
     Node* one = (stride_con > 0) ? gvn->intcon( 1) : gvn->intcon(-1);
-    limit = gvn->transform(new (C, 3) AddINode(limit, one));
+    limit = gvn->transform(new (C) AddINode(limit, one));
     if (bt == BoolTest::le)
       bt = BoolTest::lt;
     else if (bt == BoolTest::ge)
@@ -482,7 +482,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   // can directly point to the phi; in this case adjust the compare so that
   // it points to the incr by adjusting the limit.
   if (cmp->in(1) == phi || cmp->in(2) == phi)
-    limit = gvn->transform(new (C, 3) AddINode(limit,stride));
+    limit = gvn->transform(new (C) AddINode(limit,stride));
 
   // trip-count for +-tive stride should be: (limit - init_trip + stride - 1)/stride.
   // Final value for iterator should be: trip_count * stride + init_trip.
@@ -495,16 +495,16 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
     ShouldNotReachHere();
   case BoolTest::ne:            // Ahh, the case we desire
     if (stride_con == 1)
-      trip_count = gvn->transform(new (C, 3) SubINode(limit,init_trip));
+      trip_count = gvn->transform(new (C) SubINode(limit,init_trip));
     else if (stride_con == -1)
-      trip_count = gvn->transform(new (C, 3) SubINode(init_trip,limit));
+      trip_count = gvn->transform(new (C) SubINode(init_trip,limit));
     else
       ShouldNotReachHere();
     set_subtree_ctrl(trip_count);
     //_loop.map(trip_count->_idx,loop(limit));
     break;
   case BoolTest::le:            // Maybe convert to '<' case
-    limit = gvn->transform(new (C, 3) AddINode(limit,one_p));
+    limit = gvn->transform(new (C) AddINode(limit,one_p));
     set_subtree_ctrl( limit );
     hook->init_req(4, limit);
 
@@ -515,26 +515,26 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   case BoolTest::lt: {          // Maybe convert to '!=' case
     if (stride_con < 0) // Count down loop rolls through MAXINT
       ShouldNotReachHere();
-    Node *range = gvn->transform(new (C, 3) SubINode(limit,init_trip));
+    Node *range = gvn->transform(new (C) SubINode(limit,init_trip));
     set_subtree_ctrl( range );
     hook->init_req(0, range);
 
-    Node *bias  = gvn->transform(new (C, 3) AddINode(range,stride));
+    Node *bias  = gvn->transform(new (C) AddINode(range,stride));
     set_subtree_ctrl( bias );
     hook->init_req(1, bias);
 
-    Node *bias1 = gvn->transform(new (C, 3) AddINode(bias,one_m));
+    Node *bias1 = gvn->transform(new (C) AddINode(bias,one_m));
     set_subtree_ctrl( bias1 );
     hook->init_req(2, bias1);
 
-    trip_count  = gvn->transform(new (C, 3) DivINode(0,bias1,stride));
+    trip_count  = gvn->transform(new (C) DivINode(0,bias1,stride));
     set_subtree_ctrl( trip_count );
     hook->init_req(3, trip_count);
     break;
   }
 
   case BoolTest::ge:            // Maybe convert to '>' case
-    limit = gvn->transform(new (C, 3) AddINode(limit,one_m));
+    limit = gvn->transform(new (C) AddINode(limit,one_m));
     set_subtree_ctrl( limit );
     hook->init_req(4 ,limit);
 
@@ -545,30 +545,30 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   case BoolTest::gt: {          // Maybe convert to '!=' case
     if (stride_con > 0) // count up loop rolls through MININT
       ShouldNotReachHere();
-    Node *range = gvn->transform(new (C, 3) SubINode(limit,init_trip));
+    Node *range = gvn->transform(new (C) SubINode(limit,init_trip));
     set_subtree_ctrl( range );
     hook->init_req(0, range);
 
-    Node *bias  = gvn->transform(new (C, 3) AddINode(range,stride));
+    Node *bias  = gvn->transform(new (C) AddINode(range,stride));
     set_subtree_ctrl( bias );
     hook->init_req(1, bias);
 
-    Node *bias1 = gvn->transform(new (C, 3) AddINode(bias,one_p));
+    Node *bias1 = gvn->transform(new (C) AddINode(bias,one_p));
     set_subtree_ctrl( bias1 );
     hook->init_req(2, bias1);
 
-    trip_count  = gvn->transform(new (C, 3) DivINode(0,bias1,stride));
+    trip_count  = gvn->transform(new (C) DivINode(0,bias1,stride));
     set_subtree_ctrl( trip_count );
     hook->init_req(3, trip_count);
     break;
   }
   } // switch( bt )
 
-  Node *span = gvn->transform(new (C, 3) MulINode(trip_count,stride));
+  Node *span = gvn->transform(new (C) MulINode(trip_count,stride));
   set_subtree_ctrl( span );
   hook->init_req(5, span);
 
-  limit = gvn->transform(new (C, 3) AddINode(span,init_trip));
+  limit = gvn->transform(new (C) AddINode(span,init_trip));
   set_subtree_ctrl( limit );
 
   } // LoopLimitCheck
@@ -617,7 +617,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   set_ctrl(test, iff->in(0));
 
   // Replace the old IfNode with a new LoopEndNode
-  Node *lex = _igvn.register_new_node_with_optimizer(new (C, 2) CountedLoopEndNode( iff->in(0), test, cl_prob, iff->as_If()->_fcnt ));
+  Node *lex = _igvn.register_new_node_with_optimizer(new (C) CountedLoopEndNode( iff->in(0), test, cl_prob, iff->as_If()->_fcnt ));
   IfNode *le = lex->as_If();
   uint dd = dom_depth(iff);
   set_idom(le, le->in(0), dd); // Update dominance for loop exit
@@ -628,8 +628,8 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
 
   // Need to swap loop-exit and loop-back control?
   if (iftrue_op == Op_IfFalse) {
-    Node *ift2=_igvn.register_new_node_with_optimizer(new (C, 1) IfTrueNode (le));
-    Node *iff2=_igvn.register_new_node_with_optimizer(new (C, 1) IfFalseNode(le));
+    Node *ift2=_igvn.register_new_node_with_optimizer(new (C) IfTrueNode (le));
+    Node *iff2=_igvn.register_new_node_with_optimizer(new (C) IfFalseNode(le));
 
     loop->_tail = back_control = ift2;
     set_loop(ift2, loop);
@@ -655,7 +655,7 @@ bool PhaseIdealLoop::is_counted_loop( Node *x, IdealLoopTree *loop ) {
   lazy_replace( iff, le ); // fix 'get_ctrl'
 
   // Now setup a new CountedLoopNode to replace the existing LoopNode
-  CountedLoopNode *l = new (C, 3) CountedLoopNode(init_control, back_control);
+  CountedLoopNode *l = new (C) CountedLoopNode(init_control, back_control);
   l->set_unswitch_count(x->as_Loop()->unswitch_count()); // Preserve
   // The following assert is approximately true, and defines the intention
   // of can_be_counted_loop.  It fails, however, because phase->type
@@ -729,7 +729,7 @@ Node* PhaseIdealLoop::exact_limit( IdealLoopTree *loop ) {
     limit = _igvn.intcon(final_int);
   } else {
     // Create new LoopLimit node to get exact limit (final iv value).
-    limit = new (C, 4) LoopLimitNode(C, cl->init_trip(), cl->limit(), cl->stride());
+    limit = new (C) LoopLimitNode(C, cl->init_trip(), cl->limit(), cl->stride());
     register_new_node(limit, cl->in(LoopNode::EntryControl));
   }
   assert(limit != NULL, "sanity");
@@ -846,11 +846,11 @@ Node *LoopLimitNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if (range <= max) {
     // Convert to integer expression if it is not overflow.
     Node* stride_m = phase->intcon(stride_con - (stride_con > 0 ? 1 : -1));
-    Node *range = phase->transform(new (phase->C, 3) SubINode(in(Limit), in(Init)));
-    Node *bias  = phase->transform(new (phase->C, 3) AddINode(range, stride_m));
-    Node *trip  = phase->transform(new (phase->C, 3) DivINode(0, bias, in(Stride)));
-    Node *span  = phase->transform(new (phase->C, 3) MulINode(trip, in(Stride)));
-    return new (phase->C, 3) AddINode(span, in(Init)); // exact limit
+    Node *range = phase->transform(new (phase->C) SubINode(in(Limit), in(Init)));
+    Node *bias  = phase->transform(new (phase->C) AddINode(range, stride_m));
+    Node *trip  = phase->transform(new (phase->C) DivINode(0, bias, in(Stride)));
+    Node *span  = phase->transform(new (phase->C) MulINode(trip, in(Stride)));
+    return new (phase->C) AddINode(span, in(Init)); // exact limit
   }
 
   if (is_power_of_2(stride_p) ||                // divisor is 2^n
@@ -858,13 +858,13 @@ Node *LoopLimitNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     // Convert to long expression to avoid integer overflow
     // and let igvn optimizer convert this division.
     //
-    Node*   init   = phase->transform( new (phase->C, 2) ConvI2LNode(in(Init)));
-    Node*  limit   = phase->transform( new (phase->C, 2) ConvI2LNode(in(Limit)));
+    Node*   init   = phase->transform( new (phase->C) ConvI2LNode(in(Init)));
+    Node*  limit   = phase->transform( new (phase->C) ConvI2LNode(in(Limit)));
     Node* stride   = phase->longcon(stride_con);
     Node* stride_m = phase->longcon(stride_con - (stride_con > 0 ? 1 : -1));
 
-    Node *range = phase->transform(new (phase->C, 3) SubLNode(limit, init));
-    Node *bias  = phase->transform(new (phase->C, 3) AddLNode(range, stride_m));
+    Node *range = phase->transform(new (phase->C) SubLNode(limit, init));
+    Node *bias  = phase->transform(new (phase->C) AddLNode(range, stride_m));
     Node *span;
     if (stride_con > 0 && is_power_of_2(stride_p)) {
       // bias >= 0 if stride >0, so if stride is 2^n we can use &(-stride)
@@ -875,14 +875,14 @@ Node *LoopLimitNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       // only RCE predicate where exact limit is used and the predicate
       // will simply fail forcing recompilation.
       Node* neg_stride   = phase->longcon(-stride_con);
-      span = phase->transform(new (phase->C, 3) AndLNode(bias, neg_stride));
+      span = phase->transform(new (phase->C) AndLNode(bias, neg_stride));
     } else {
-      Node *trip  = phase->transform(new (phase->C, 3) DivLNode(0, bias, stride));
-      span = phase->transform(new (phase->C, 3) MulLNode(trip, stride));
+      Node *trip  = phase->transform(new (phase->C) DivLNode(0, bias, stride));
+      span = phase->transform(new (phase->C) MulLNode(trip, stride));
     }
     // Convert back to int
-    Node *span_int = phase->transform(new (phase->C, 2) ConvL2INode(span));
-    return new (phase->C, 3) AddINode(span_int, in(Init)); // exact limit
+    Node *span_int = phase->transform(new (phase->C) ConvL2INode(span));
+    return new (phase->C) AddINode(span_int, in(Init)); // exact limit
   }
 
   return NULL;    // No progress
@@ -1088,7 +1088,7 @@ void IdealLoopTree::split_fall_in( PhaseIdealLoop *phase, int fall_in_cnt ) {
   uint i;
 
   // Make a new RegionNode to be the landing pad.
-  Node *landing_pad = new (phase->C, fall_in_cnt+1) RegionNode( fall_in_cnt+1 );
+  Node *landing_pad = new (phase->C) RegionNode( fall_in_cnt+1 );
   phase->set_loop(landing_pad,_parent);
   // Gather all the fall-in control paths into the landing pad
   uint icnt = fall_in_cnt;
@@ -1174,7 +1174,7 @@ void IdealLoopTree::split_outer_loop( PhaseIdealLoop *phase ) {
 
   // Make a LoopNode for the outermost loop.
   Node *ctl = _head->in(LoopNode::EntryControl);
-  Node *outer = new (phase->C, 3) LoopNode( ctl, _head->in(outer_idx) );
+  Node *outer = new (phase->C) LoopNode( ctl, _head->in(outer_idx) );
   outer = igvn.register_new_node_with_optimizer(outer, _head);
   phase->set_created_loop_node();
 
@@ -1288,7 +1288,7 @@ void IdealLoopTree::merge_many_backedges( PhaseIdealLoop *phase ) {
 
   Node *hot_tail = NULL;
   // Make a Region for the merge point
-  Node *r = new (phase->C, 1) RegionNode(1);
+  Node *r = new (phase->C) RegionNode(1);
   for( i = 2; i < _head->req(); i++ ) {
     if( i != hot_idx )
       r->add_req( _head->in(i) );
@@ -1307,7 +1307,7 @@ void IdealLoopTree::merge_many_backedges( PhaseIdealLoop *phase ) {
       PhiNode* n = out->as_Phi();
       igvn.hash_delete(n);      // Delete from hash before hacking edges
       Node *hot_phi = NULL;
-      Node *phi = new (phase->C, r->req()) PhiNode(r, n->type(), n->adr_type());
+      Node *phi = new (phase->C) PhiNode(r, n->type(), n->adr_type());
       // Check all inputs for the ones to peel out
       uint j = 1;
       for( uint i = 2; i < n->req(); i++ ) {
@@ -1429,7 +1429,7 @@ bool IdealLoopTree::beautify_loops( PhaseIdealLoop *phase ) {
 
   } else if( !_head->is_Loop() && !_irreducible ) {
     // Make a new LoopNode to replace the old loop head
-    Node *l = new (phase->C, 3) LoopNode( _head->in(1), _head->in(2) );
+    Node *l = new (phase->C) LoopNode( _head->in(1), _head->in(2) );
     l = igvn.register_new_node_with_optimizer(l, _head);
     phase->set_created_loop_node();
     // Go ahead and replace _head
@@ -1671,16 +1671,16 @@ void PhaseIdealLoop::replace_parallel_iv(IdealLoopTree *loop) {
       // It is scaled by the 'ratio_con'.
       Node* ratio = _igvn.intcon(ratio_con);
       set_ctrl(ratio, C->root());
-      Node* ratio_init = new (C, 3) MulINode(init, ratio);
+      Node* ratio_init = new (C) MulINode(init, ratio);
       _igvn.register_new_node_with_optimizer(ratio_init, init);
       set_early_ctrl(ratio_init);
-      Node* diff = new (C, 3) SubINode(init2, ratio_init);
+      Node* diff = new (C) SubINode(init2, ratio_init);
       _igvn.register_new_node_with_optimizer(diff, init2);
       set_early_ctrl(diff);
-      Node* ratio_idx = new (C, 3) MulINode(phi, ratio);
+      Node* ratio_idx = new (C) MulINode(phi, ratio);
       _igvn.register_new_node_with_optimizer(ratio_idx, phi);
       set_ctrl(ratio_idx, cl);
-      Node* add = new (C, 3) AddINode(ratio_idx, diff);
+      Node* add = new (C) AddINode(ratio_idx, diff);
       _igvn.register_new_node_with_optimizer(add);
       set_ctrl(add, cl);
       _igvn.replace_node( phi2, add );
@@ -2677,10 +2677,10 @@ int PhaseIdealLoop::build_loop_tree_impl( Node *n, int pre_order ) {
 
         if (!_verify_only) {
           // Insert the NeverBranch between 'm' and it's control user.
-          NeverBranchNode *iff = new (C, 1) NeverBranchNode( m );
+          NeverBranchNode *iff = new (C) NeverBranchNode( m );
           _igvn.register_new_node_with_optimizer(iff);
           set_loop(iff, l);
-          Node *if_t = new (C, 1) CProjNode( iff, 0 );
+          Node *if_t = new (C) CProjNode( iff, 0 );
           _igvn.register_new_node_with_optimizer(if_t);
           set_loop(if_t, l);
 
@@ -2696,16 +2696,16 @@ int PhaseIdealLoop::build_loop_tree_impl( Node *n, int pre_order ) {
           cfg->set_req( k, if_t ); // Now point to NeverBranch
 
           // Now create the never-taken loop exit
-          Node *if_f = new (C, 1) CProjNode( iff, 1 );
+          Node *if_f = new (C) CProjNode( iff, 1 );
           _igvn.register_new_node_with_optimizer(if_f);
           set_loop(if_f, l);
           // Find frame ptr for Halt.  Relies on the optimizer
           // V-N'ing.  Easier and quicker than searching through
           // the program structure.
-          Node *frame = new (C, 1) ParmNode( C->start(), TypeFunc::FramePtr );
+          Node *frame = new (C) ParmNode( C->start(), TypeFunc::FramePtr );
           _igvn.register_new_node_with_optimizer(frame);
           // Halt & Catch Fire
-          Node *halt = new (C, TypeFunc::Parms) HaltNode( if_f, frame );
+          Node *halt = new (C) HaltNode( if_f, frame );
           _igvn.register_new_node_with_optimizer(halt);
           set_loop(halt, l);
           C->root()->add_req(halt);
diff --git a/src/share/vm/opto/loopopts.cpp b/src/share/vm/opto/loopopts.cpp
index ee6c36d22..5c49a0444 100644
--- a/src/share/vm/opto/loopopts.cpp
+++ b/src/share/vm/opto/loopopts.cpp
@@ -53,7 +53,7 @@ Node *PhaseIdealLoop::split_thru_phi( Node *n, Node *region, int policy ) {
     int iid    = t_oop->instance_id();
     int index  = C->get_alias_index(t_oop);
     int offset = t_oop->offset();
-    phi = new (C,region->req()) PhiNode(region, type, NULL, iid, index, offset);
+    phi = new (C) PhiNode(region, type, NULL, iid, index, offset);
   } else {
     phi = PhiNode::make_blank(region, n);
   }
@@ -356,9 +356,9 @@ Node *PhaseIdealLoop::remix_address_expressions( Node *n ) {
         _igvn.type( add->in(1) ) != TypeInt::ZERO ) {
       Node *zero = _igvn.intcon(0);
       set_ctrl(zero, C->root());
-      Node *neg = new (C, 3) SubINode( _igvn.intcon(0), add->in(2) );
+      Node *neg = new (C) SubINode( _igvn.intcon(0), add->in(2) );
       register_new_node( neg, get_ctrl(add->in(2) ) );
-      add = new (C, 3) AddINode( add->in(1), neg );
+      add = new (C) AddINode( add->in(1), neg );
       register_new_node( add, add_ctrl );
     }
     if( add->Opcode() != Op_AddI ) return NULL;
@@ -384,14 +384,14 @@ Node *PhaseIdealLoop::remix_address_expressions( Node *n ) {
       return NULL;              // No invariant part of the add?
 
     // Yes!  Reshape address expression!
-    Node *inv_scale = new (C, 3) LShiftINode( add_invar, scale );
+    Node *inv_scale = new (C) LShiftINode( add_invar, scale );
     Node *inv_scale_ctrl =
       dom_depth(add_invar_ctrl) > dom_depth(scale_ctrl) ?
       add_invar_ctrl : scale_ctrl;
     register_new_node( inv_scale, inv_scale_ctrl );
-    Node *var_scale = new (C, 3) LShiftINode( add_var, scale );
+    Node *var_scale = new (C) LShiftINode( add_var, scale );
     register_new_node( var_scale, n_ctrl );
-    Node *var_add = new (C, 3) AddINode( var_scale, inv_scale );
+    Node *var_add = new (C) AddINode( var_scale, inv_scale );
     register_new_node( var_add, n_ctrl );
     _igvn.replace_node( n, var_add );
     return var_add;
@@ -423,10 +423,10 @@ Node *PhaseIdealLoop::remix_address_expressions( Node *n ) {
         IdealLoopTree *n23_loop = get_loop( n23_ctrl );
         if( n22loop != n_loop && n22loop->is_member(n_loop) &&
             n23_loop == n_loop ) {
-          Node *add1 = new (C, 4) AddPNode( n->in(1), n->in(2)->in(2), n->in(3) );
+          Node *add1 = new (C) AddPNode( n->in(1), n->in(2)->in(2), n->in(3) );
           // Stuff new AddP in the loop preheader
           register_new_node( add1, n_loop->_head->in(LoopNode::EntryControl) );
-          Node *add2 = new (C, 4) AddPNode( n->in(1), add1, n->in(2)->in(3) );
+          Node *add2 = new (C) AddPNode( n->in(1), add1, n->in(2)->in(3) );
           register_new_node( add2, n_ctrl );
           _igvn.replace_node( n, add2 );
           return add2;
@@ -444,10 +444,10 @@ Node *PhaseIdealLoop::remix_address_expressions( Node *n ) {
           Node *tmp = V; V = I; I = tmp;
         }
         if( !is_member(n_loop,get_ctrl(I)) ) {
-          Node *add1 = new (C, 4) AddPNode( n->in(1), n->in(2), I );
+          Node *add1 = new (C) AddPNode( n->in(1), n->in(2), I );
           // Stuff new AddP in the loop preheader
           register_new_node( add1, n_loop->_head->in(LoopNode::EntryControl) );
-          Node *add2 = new (C, 4) AddPNode( n->in(1), add1, V );
+          Node *add2 = new (C) AddPNode( n->in(1), add1, V );
           register_new_node( add2, n_ctrl );
           _igvn.replace_node( n, add2 );
           return add2;
@@ -1094,9 +1094,8 @@ BoolNode *PhaseIdealLoop::clone_iff( PhiNode *phi, IdealLoopTree *loop ) {
   Node *sample_cmp  = sample_bool->in(1);
 
   // Make Phis to merge the Cmp's inputs.
-  int size = phi->in(0)->req();
-  PhiNode *phi1 = new (C, size) PhiNode( phi->in(0), Type::TOP );
-  PhiNode *phi2 = new (C, size) PhiNode( phi->in(0), Type::TOP );
+  PhiNode *phi1 = new (C) PhiNode( phi->in(0), Type::TOP );
+  PhiNode *phi2 = new (C) PhiNode( phi->in(0), Type::TOP );
   for( i = 1; i < phi->req(); i++ ) {
     Node *n1 = phi->in(i)->in(1)->in(1);
     Node *n2 = phi->in(i)->in(1)->in(2);
@@ -1163,9 +1162,8 @@ CmpNode *PhaseIdealLoop::clone_bool( PhiNode *phi, IdealLoopTree *loop ) {
   Node *sample_cmp = phi->in(1);
 
   // Make Phis to merge the Cmp's inputs.
-  int size = phi->in(0)->req();
-  PhiNode *phi1 = new (C, size) PhiNode( phi->in(0), Type::TOP );
-  PhiNode *phi2 = new (C, size) PhiNode( phi->in(0), Type::TOP );
+  PhiNode *phi1 = new (C) PhiNode( phi->in(0), Type::TOP );
+  PhiNode *phi2 = new (C) PhiNode( phi->in(0), Type::TOP );
   for( uint j = 1; j < phi->req(); j++ ) {
     Node *cmp_top = phi->in(j); // Inputs are all Cmp or TOP
     Node *n1, *n2;
@@ -1329,7 +1327,7 @@ void PhaseIdealLoop::clone_loop( IdealLoopTree *loop, Node_List &old_new, int dd
 
         // We need a Region to merge the exit from the peeled body and the
         // exit from the old loop body.
-        RegionNode *r = new (C, 3) RegionNode(3);
+        RegionNode *r = new (C) RegionNode(3);
         // Map the old use to the new merge point
         old_new.map( use->_idx, r );
         uint dd_r = MIN2(dom_depth(newuse),dom_depth(use));
@@ -1675,13 +1673,13 @@ ProjNode* PhaseIdealLoop::insert_if_before_proj(Node* left, bool Signed, BoolTes
   ProjNode* proj2 = proj_clone(proj, iff);
   register_node(proj2, loop, iff, ddepth);
 
-  Node* cmp = Signed ? (Node*) new (C,3)CmpINode(left, right) : (Node*) new (C,3)CmpUNode(left, right);
+  Node* cmp = Signed ? (Node*) new (C)CmpINode(left, right) : (Node*) new (C)CmpUNode(left, right);
   register_node(cmp, loop, proj2, ddepth);
 
-  BoolNode* bol = new (C,2)BoolNode(cmp, relop);
+  BoolNode* bol = new (C)BoolNode(cmp, relop);
   register_node(bol, loop, proj2, ddepth);
 
-  IfNode* new_if = new (C,2)IfNode(proj2, bol, iff->_prob, iff->_fcnt);
+  IfNode* new_if = new (C)IfNode(proj2, bol, iff->_prob, iff->_fcnt);
   register_node(new_if, loop, proj2, ddepth);
 
   proj->set_req(0, new_if); // reattach
@@ -1732,11 +1730,11 @@ RegionNode* PhaseIdealLoop::insert_region_before_proj(ProjNode* proj) {
   ProjNode* proj2 = proj_clone(proj, iff);
   register_node(proj2, loop, iff, ddepth);
 
-  RegionNode* reg = new (C,2)RegionNode(2);
+  RegionNode* reg = new (C)RegionNode(2);
   reg->set_req(1, proj2);
   register_node(reg, loop, iff, ddepth);
 
-  IfNode* dum_if = new (C,2)IfNode(reg, short_circuit_if(NULL, proj), iff->_prob, iff->_fcnt);
+  IfNode* dum_if = new (C)IfNode(reg, short_circuit_if(NULL, proj), iff->_prob, iff->_fcnt);
   register_node(dum_if, loop, reg, ddepth);
 
   proj->set_req(0, dum_if); // reattach
@@ -2547,7 +2545,7 @@ bool PhaseIdealLoop::partial_peel( IdealLoopTree *loop, Node_List &old_new ) {
 
   // Create new loop head for new phis and to hang
   // the nodes being moved (sinked) from the peel region.
-  LoopNode* new_head = new (C, 3) LoopNode(last_peel, last_peel);
+  LoopNode* new_head = new (C) LoopNode(last_peel, last_peel);
   new_head->set_unswitch_count(head->unswitch_count()); // Preserve
   _igvn.register_new_node_with_optimizer(new_head);
   assert(first_not_peeled->in(0) == last_peel, "last_peel <- first_not_peeled");
@@ -2746,11 +2744,11 @@ void PhaseIdealLoop::reorg_offsets(IdealLoopTree *loop) {
       if (dom_lca(exit, u_ctrl) != exit) continue;
       // Hit!  Refactor use to use the post-incremented tripcounter.
       // Compute a post-increment tripcounter.
-      Node *opaq = new (C, 2) Opaque2Node( C, cle->incr() );
+      Node *opaq = new (C) Opaque2Node( C, cle->incr() );
       register_new_node( opaq, u_ctrl );
       Node *neg_stride = _igvn.intcon(-cle->stride_con());
       set_ctrl(neg_stride, C->root());
-      Node *post = new (C, 3) AddINode( opaq, neg_stride);
+      Node *post = new (C) AddINode( opaq, neg_stride);
       register_new_node( post, u_ctrl );
       _igvn.rehash_node_delayed(use);
       for (uint j = 1; j < use->req(); j++) {
diff --git a/src/share/vm/opto/macro.cpp b/src/share/vm/opto/macro.cpp
index a5250292b..90f1f0299 100644
--- a/src/share/vm/opto/macro.cpp
+++ b/src/share/vm/opto/macro.cpp
@@ -103,20 +103,20 @@ void PhaseMacroExpand::copy_call_debug_info(CallNode *oldcall, CallNode * newcal
 Node* PhaseMacroExpand::opt_bits_test(Node* ctrl, Node* region, int edge, Node* word, int mask, int bits, bool return_fast_path) {
   Node* cmp;
   if (mask != 0) {
-    Node* and_node = transform_later(new (C, 3) AndXNode(word, MakeConX(mask)));
-    cmp = transform_later(new (C, 3) CmpXNode(and_node, MakeConX(bits)));
+    Node* and_node = transform_later(new (C) AndXNode(word, MakeConX(mask)));
+    cmp = transform_later(new (C) CmpXNode(and_node, MakeConX(bits)));
   } else {
     cmp = word;
   }
-  Node* bol = transform_later(new (C, 2) BoolNode(cmp, BoolTest::ne));
-  IfNode* iff = new (C, 2) IfNode( ctrl, bol, PROB_MIN, COUNT_UNKNOWN );
+  Node* bol = transform_later(new (C) BoolNode(cmp, BoolTest::ne));
+  IfNode* iff = new (C) IfNode( ctrl, bol, PROB_MIN, COUNT_UNKNOWN );
   transform_later(iff);
 
   // Fast path taken.
-  Node *fast_taken = transform_later( new (C, 1) IfFalseNode(iff) );
+  Node *fast_taken = transform_later( new (C) IfFalseNode(iff) );
 
   // Fast path not-taken, i.e. slow path
-  Node *slow_taken = transform_later( new (C, 1) IfTrueNode(iff) );
+  Node *slow_taken = transform_later( new (C) IfTrueNode(iff) );
 
   if (return_fast_path) {
     region->init_req(edge, slow_taken); // Capture slow-control
@@ -141,10 +141,9 @@ void PhaseMacroExpand::copy_predefined_input_for_runtime_call(Node * ctrl, CallN
 CallNode* PhaseMacroExpand::make_slow_call(CallNode *oldcall, const TypeFunc* slow_call_type, address slow_call, const char* leaf_name, Node* slow_path, Node* parm0, Node* parm1) {
 
   // Slow-path call
-  int size = slow_call_type->domain()->cnt();
  CallNode *call = leaf_name
-   ? (CallNode*)new (C, size) CallLeafNode      ( slow_call_type, slow_call, leaf_name, TypeRawPtr::BOTTOM )
-   : (CallNode*)new (C, size) CallStaticJavaNode( slow_call_type, slow_call, OptoRuntime::stub_name(slow_call), oldcall->jvms()->bci(), TypeRawPtr::BOTTOM );
+   ? (CallNode*)new (C) CallLeafNode      ( slow_call_type, slow_call, leaf_name, TypeRawPtr::BOTTOM )
+   : (CallNode*)new (C) CallStaticJavaNode( slow_call_type, slow_call, OptoRuntime::stub_name(slow_call), oldcall->jvms()->bci(), TypeRawPtr::BOTTOM );
 
   // Slow path call has no side-effects, uses few values
   copy_predefined_input_for_runtime_call(slow_path, oldcall, call );
@@ -412,7 +411,7 @@ Node *PhaseMacroExpand::value_from_mem_phi(Node *mem, BasicType ft, const Type *
   GrowableArray <Node *> values(length, length, NULL, false);
 
   // create a new Phi for the value
-  PhiNode *phi = new (C, length) PhiNode(mem->in(0), phi_type, NULL, instance_id, alias_idx, offset);
+  PhiNode *phi = new (C) PhiNode(mem->in(0), phi_type, NULL, instance_id, alias_idx, offset);
   transform_later(phi);
   value_phis->push(phi, mem->_idx);
 
@@ -720,7 +719,7 @@ bool PhaseMacroExpand::scalar_replacement(AllocateNode *alloc, GrowableArray <Sa
     SafePointNode* sfpt = safepoints.pop();
     Node* mem = sfpt->memory();
     uint first_ind = sfpt->req();
-    SafePointScalarObjectNode* sobj = new (C, 1) SafePointScalarObjectNode(res_type,
+    SafePointScalarObjectNode* sobj = new (C) SafePointScalarObjectNode(res_type,
 #ifdef ASSERT
                                                  alloc,
 #endif
@@ -828,7 +827,7 @@ bool PhaseMacroExpand::scalar_replacement(AllocateNode *alloc, GrowableArray <Sa
         if (field_val->is_EncodeP()) {
           field_val = field_val->in(1);
         } else {
-          field_val = transform_later(new (C, 2) DecodeNNode(field_val, field_val->bottom_type()->make_ptr()));
+          field_val = transform_later(new (C) DecodeNNode(field_val, field_val->bottom_type()->make_ptr()));
         }
       }
       sfpt->add_req(field_val);
@@ -995,7 +994,7 @@ bool PhaseMacroExpand::eliminate_allocate_node(AllocateNode *alloc) {
 //---------------------------set_eden_pointers-------------------------
 void PhaseMacroExpand::set_eden_pointers(Node* &eden_top_adr, Node* &eden_end_adr) {
   if (UseTLAB) {                // Private allocation: load from TLS
-    Node* thread = transform_later(new (C, 1) ThreadLocalNode());
+    Node* thread = transform_later(new (C) ThreadLocalNode());
     int tlab_top_offset = in_bytes(JavaThread::tlab_top_offset());
     int tlab_end_offset = in_bytes(JavaThread::tlab_end_offset());
     eden_top_adr = basic_plus_adr(top()/*not oop*/, thread, tlab_top_offset);
@@ -1137,18 +1136,18 @@ void PhaseMacroExpand::expand_allocate_common(
   assert (initial_slow_test == NULL || !always_slow, "arguments must be consistent");
   // generate the initial test if necessary
   if (initial_slow_test != NULL ) {
-    slow_region = new (C, 3) RegionNode(3);
+    slow_region = new (C) RegionNode(3);
 
     // Now make the initial failure test.  Usually a too-big test but
     // might be a TRUE for finalizers or a fancy class check for
     // newInstance0.
-    IfNode *toobig_iff = new (C, 2) IfNode(ctrl, initial_slow_test, PROB_MIN, COUNT_UNKNOWN);
+    IfNode *toobig_iff = new (C) IfNode(ctrl, initial_slow_test, PROB_MIN, COUNT_UNKNOWN);
     transform_later(toobig_iff);
     // Plug the failing-too-big test into the slow-path region
-    Node *toobig_true = new (C, 1) IfTrueNode( toobig_iff );
+    Node *toobig_true = new (C) IfTrueNode( toobig_iff );
     transform_later(toobig_true);
     slow_region    ->init_req( too_big_or_final_path, toobig_true );
-    toobig_false = new (C, 1) IfFalseNode( toobig_iff );
+    toobig_false = new (C) IfFalseNode( toobig_iff );
     transform_later(toobig_false);
   } else {         // No initial test, just fall into next case
     toobig_false = ctrl;
@@ -1181,10 +1180,10 @@ void PhaseMacroExpand::expand_allocate_common(
     Node *eden_end = make_load(ctrl, mem, eden_end_adr, 0, TypeRawPtr::BOTTOM, T_ADDRESS);
 
     // allocate the Region and Phi nodes for the result
-    result_region = new (C, 3) RegionNode(3);
-    result_phi_rawmem = new (C, 3) PhiNode(result_region, Type::MEMORY, TypeRawPtr::BOTTOM);
-    result_phi_rawoop = new (C, 3) PhiNode(result_region, TypeRawPtr::BOTTOM);
-    result_phi_i_o    = new (C, 3) PhiNode(result_region, Type::ABIO); // I/O is used for Prefetch
+    result_region = new (C) RegionNode(3);
+    result_phi_rawmem = new (C) PhiNode(result_region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    result_phi_rawoop = new (C) PhiNode(result_region, TypeRawPtr::BOTTOM);
+    result_phi_i_o    = new (C) PhiNode(result_region, Type::ABIO); // I/O is used for Prefetch
 
     // We need a Region for the loop-back contended case.
     enum { fall_in_path = 1, contended_loopback_path = 2 };
@@ -1194,8 +1193,8 @@ void PhaseMacroExpand::expand_allocate_common(
       contended_region = toobig_false;
       contended_phi_rawmem = mem;
     } else {
-      contended_region = new (C, 3) RegionNode(3);
-      contended_phi_rawmem = new (C, 3) PhiNode(contended_region, Type::MEMORY, TypeRawPtr::BOTTOM);
+      contended_region = new (C) RegionNode(3);
+      contended_phi_rawmem = new (C) PhiNode(contended_region, Type::MEMORY, TypeRawPtr::BOTTOM);
       // Now handle the passing-too-big test.  We fall into the contended
       // loop-back merge point.
       contended_region    ->init_req(fall_in_path, toobig_false);
@@ -1207,23 +1206,23 @@ void PhaseMacroExpand::expand_allocate_common(
     // Load(-locked) the heap top.
     // See note above concerning the control input when using a TLAB
     Node *old_eden_top = UseTLAB
-      ? new (C, 3) LoadPNode      (ctrl, contended_phi_rawmem, eden_top_adr, TypeRawPtr::BOTTOM, TypeRawPtr::BOTTOM)
-      : new (C, 3) LoadPLockedNode(contended_region, contended_phi_rawmem, eden_top_adr);
+      ? new (C) LoadPNode      (ctrl, contended_phi_rawmem, eden_top_adr, TypeRawPtr::BOTTOM, TypeRawPtr::BOTTOM)
+      : new (C) LoadPLockedNode(contended_region, contended_phi_rawmem, eden_top_adr);
 
     transform_later(old_eden_top);
     // Add to heap top to get a new heap top
-    Node *new_eden_top = new (C, 4) AddPNode(top(), old_eden_top, size_in_bytes);
+    Node *new_eden_top = new (C) AddPNode(top(), old_eden_top, size_in_bytes);
     transform_later(new_eden_top);
     // Check for needing a GC; compare against heap end
-    Node *needgc_cmp = new (C, 3) CmpPNode(new_eden_top, eden_end);
+    Node *needgc_cmp = new (C) CmpPNode(new_eden_top, eden_end);
     transform_later(needgc_cmp);
-    Node *needgc_bol = new (C, 2) BoolNode(needgc_cmp, BoolTest::ge);
+    Node *needgc_bol = new (C) BoolNode(needgc_cmp, BoolTest::ge);
     transform_later(needgc_bol);
-    IfNode *needgc_iff = new (C, 2) IfNode(contended_region, needgc_bol, PROB_UNLIKELY_MAG(4), COUNT_UNKNOWN);
+    IfNode *needgc_iff = new (C) IfNode(contended_region, needgc_bol, PROB_UNLIKELY_MAG(4), COUNT_UNKNOWN);
     transform_later(needgc_iff);
 
     // Plug the failing-heap-space-need-gc test into the slow-path region
-    Node *needgc_true = new (C, 1) IfTrueNode(needgc_iff);
+    Node *needgc_true = new (C) IfTrueNode(needgc_iff);
     transform_later(needgc_true);
     if (initial_slow_test) {
       slow_region->init_req(need_gc_path, needgc_true);
@@ -1234,7 +1233,7 @@ void PhaseMacroExpand::expand_allocate_common(
       slow_region = needgc_true;
     }
     // No need for a GC.  Setup for the Store-Conditional
-    Node *needgc_false = new (C, 1) IfFalseNode(needgc_iff);
+    Node *needgc_false = new (C) IfFalseNode(needgc_iff);
     transform_later(needgc_false);
 
     // Grab regular I/O before optional prefetch may change it.
@@ -1254,37 +1253,37 @@ void PhaseMacroExpand::expand_allocate_common(
     // memory state.
     if (UseTLAB) {
       Node* store_eden_top =
-        new (C, 4) StorePNode(needgc_false, contended_phi_rawmem, eden_top_adr,
+        new (C) StorePNode(needgc_false, contended_phi_rawmem, eden_top_adr,
                               TypeRawPtr::BOTTOM, new_eden_top);
       transform_later(store_eden_top);
       fast_oop_ctrl = needgc_false; // No contention, so this is the fast path
       fast_oop_rawmem = store_eden_top;
     } else {
       Node* store_eden_top =
-        new (C, 5) StorePConditionalNode(needgc_false, contended_phi_rawmem, eden_top_adr,
+        new (C) StorePConditionalNode(needgc_false, contended_phi_rawmem, eden_top_adr,
                                          new_eden_top, fast_oop/*old_eden_top*/);
       transform_later(store_eden_top);
-      Node *contention_check = new (C, 2) BoolNode(store_eden_top, BoolTest::ne);
+      Node *contention_check = new (C) BoolNode(store_eden_top, BoolTest::ne);
       transform_later(contention_check);
-      store_eden_top = new (C, 1) SCMemProjNode(store_eden_top);
+      store_eden_top = new (C) SCMemProjNode(store_eden_top);
       transform_later(store_eden_top);
 
       // If not using TLABs, check to see if there was contention.
-      IfNode *contention_iff = new (C, 2) IfNode (needgc_false, contention_check, PROB_MIN, COUNT_UNKNOWN);
+      IfNode *contention_iff = new (C) IfNode (needgc_false, contention_check, PROB_MIN, COUNT_UNKNOWN);
       transform_later(contention_iff);
-      Node *contention_true = new (C, 1) IfTrueNode(contention_iff);
+      Node *contention_true = new (C) IfTrueNode(contention_iff);
       transform_later(contention_true);
       // If contention, loopback and try again.
       contended_region->init_req(contended_loopback_path, contention_true);
       contended_phi_rawmem->init_req(contended_loopback_path, store_eden_top);
 
       // Fast-path succeeded with no contention!
-      Node *contention_false = new (C, 1) IfFalseNode(contention_iff);
+      Node *contention_false = new (C) IfFalseNode(contention_iff);
       transform_later(contention_false);
       fast_oop_ctrl = contention_false;
 
       // Bump total allocated bytes for this thread
-      Node* thread = new (C, 1) ThreadLocalNode();
+      Node* thread = new (C) ThreadLocalNode();
       transform_later(thread);
       Node* alloc_bytes_adr = basic_plus_adr(top()/*not oop*/, thread,
                                              in_bytes(JavaThread::allocated_bytes_offset()));
@@ -1293,10 +1292,10 @@ void PhaseMacroExpand::expand_allocate_common(
 #ifdef _LP64
       Node* alloc_size = size_in_bytes;
 #else
-      Node* alloc_size = new (C, 2) ConvI2LNode(size_in_bytes);
+      Node* alloc_size = new (C) ConvI2LNode(size_in_bytes);
       transform_later(alloc_size);
 #endif
-      Node* new_alloc_bytes = new (C, 3) AddLNode(alloc_bytes, alloc_size);
+      Node* new_alloc_bytes = new (C) AddLNode(alloc_bytes, alloc_size);
       transform_later(new_alloc_bytes);
       fast_oop_rawmem = make_store(fast_oop_ctrl, store_eden_top, alloc_bytes_adr,
                                    0, new_alloc_bytes, T_LONG);
@@ -1323,9 +1322,9 @@ void PhaseMacroExpand::expand_allocate_common(
 
         mb->init_req(TypeFunc::Memory, fast_oop_rawmem);
         mb->init_req(TypeFunc::Control, fast_oop_ctrl);
-        fast_oop_ctrl = new (C, 1) ProjNode(mb,TypeFunc::Control);
+        fast_oop_ctrl = new (C) ProjNode(mb,TypeFunc::Control);
         transform_later(fast_oop_ctrl);
-        fast_oop_rawmem = new (C, 1) ProjNode(mb,TypeFunc::Memory);
+        fast_oop_rawmem = new (C) ProjNode(mb,TypeFunc::Memory);
         transform_later(fast_oop_rawmem);
       } else {
         // Add the MemBarStoreStore after the InitializeNode so that
@@ -1339,9 +1338,9 @@ void PhaseMacroExpand::expand_allocate_common(
         MemBarNode* mb = MemBarNode::make(C, Op_MemBarStoreStore, Compile::AliasIdxBot);
         transform_later(mb);
 
-        Node* ctrl = new (C, 1) ProjNode(init,TypeFunc::Control);
+        Node* ctrl = new (C) ProjNode(init,TypeFunc::Control);
         transform_later(ctrl);
-        Node* mem = new (C, 1) ProjNode(init,TypeFunc::Memory);
+        Node* mem = new (C) ProjNode(init,TypeFunc::Memory);
         transform_later(mem);
 
         // The MemBarStoreStore depends on control and memory coming
@@ -1349,9 +1348,9 @@ void PhaseMacroExpand::expand_allocate_common(
         mb->init_req(TypeFunc::Memory, mem);
         mb->init_req(TypeFunc::Control, ctrl);
 
-        ctrl = new (C, 1) ProjNode(mb,TypeFunc::Control);
+        ctrl = new (C) ProjNode(mb,TypeFunc::Control);
         transform_later(ctrl);
-        mem = new (C, 1) ProjNode(mb,TypeFunc::Memory);
+        mem = new (C) ProjNode(mb,TypeFunc::Memory);
         transform_later(mem);
 
         // All nodes that depended on the InitializeNode for control
@@ -1365,13 +1364,13 @@ void PhaseMacroExpand::expand_allocate_common(
     if (C->env()->dtrace_extended_probes()) {
       // Slow-path call
       int size = TypeFunc::Parms + 2;
-      CallLeafNode *call = new (C, size) CallLeafNode(OptoRuntime::dtrace_object_alloc_Type(),
-                                                      CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_object_alloc_base),
-                                                      "dtrace_object_alloc",
-                                                      TypeRawPtr::BOTTOM);
+      CallLeafNode *call = new (C) CallLeafNode(OptoRuntime::dtrace_object_alloc_Type(),
+                                                CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_object_alloc_base),
+                                                "dtrace_object_alloc",
+                                                TypeRawPtr::BOTTOM);
 
       // Get base of thread-local storage area
-      Node* thread = new (C, 1) ThreadLocalNode();
+      Node* thread = new (C) ThreadLocalNode();
       transform_later(thread);
 
       call->init_req(TypeFunc::Parms+0, thread);
@@ -1382,9 +1381,9 @@ void PhaseMacroExpand::expand_allocate_common(
       call->init_req(TypeFunc::ReturnAdr, alloc->in(TypeFunc::ReturnAdr));
       call->init_req(TypeFunc::FramePtr, alloc->in(TypeFunc::FramePtr));
       transform_later(call);
-      fast_oop_ctrl = new (C, 1) ProjNode(call,TypeFunc::Control);
+      fast_oop_ctrl = new (C) ProjNode(call,TypeFunc::Control);
       transform_later(fast_oop_ctrl);
-      fast_oop_rawmem = new (C, 1) ProjNode(call,TypeFunc::Memory);
+      fast_oop_rawmem = new (C) ProjNode(call,TypeFunc::Memory);
       transform_later(fast_oop_rawmem);
     }
 
@@ -1399,11 +1398,10 @@ void PhaseMacroExpand::expand_allocate_common(
   }
 
   // Generate slow-path call
-  CallNode *call = new (C, slow_call_type->domain()->cnt())
-    CallStaticJavaNode(slow_call_type, slow_call_address,
-                       OptoRuntime::stub_name(slow_call_address),
-                       alloc->jvms()->bci(),
-                       TypePtr::BOTTOM);
+  CallNode *call = new (C) CallStaticJavaNode(slow_call_type, slow_call_address,
+                               OptoRuntime::stub_name(slow_call_address),
+                               alloc->jvms()->bci(),
+                               TypePtr::BOTTOM);
   call->init_req( TypeFunc::Control, slow_region );
   call->init_req( TypeFunc::I_O    , top() )     ;   // does no i/o
   call->init_req( TypeFunc::Memory , slow_mem ); // may gc ptrs
@@ -1457,7 +1455,7 @@ void PhaseMacroExpand::expand_allocate_common(
   // _memproj_catchall so we end up with a call that has only 1 memory projection.
   if (_memproj_catchall != NULL ) {
     if (_memproj_fallthrough == NULL) {
-      _memproj_fallthrough = new (C, 1) ProjNode(call, TypeFunc::Memory);
+      _memproj_fallthrough = new (C) ProjNode(call, TypeFunc::Memory);
       transform_later(_memproj_fallthrough);
     }
     for (DUIterator_Fast imax, i = _memproj_catchall->fast_outs(imax); i < imax; i++) {
@@ -1489,7 +1487,7 @@ void PhaseMacroExpand::expand_allocate_common(
   // _ioproj_catchall so we end up with a call that has only 1 i_o projection.
   if (_ioproj_catchall != NULL ) {
     if (_ioproj_fallthrough == NULL) {
-      _ioproj_fallthrough = new (C, 1) ProjNode(call, TypeFunc::I_O);
+      _ioproj_fallthrough = new (C) ProjNode(call, TypeFunc::I_O);
       transform_later(_ioproj_fallthrough);
     }
     for (DUIterator_Fast imax, i = _ioproj_catchall->fast_outs(imax); i < imax; i++) {
@@ -1623,46 +1621,46 @@ Node* PhaseMacroExpand::prefetch_allocation(Node* i_o, Node*& needgc_false,
       // As an allocation hits the watermark, we will prefetch starting
       // at a "distance" away from watermark.
 
-      Node *pf_region = new (C, 3) RegionNode(3);
-      Node *pf_phi_rawmem = new (C, 3) PhiNode( pf_region, Type::MEMORY,
+      Node *pf_region = new (C) RegionNode(3);
+      Node *pf_phi_rawmem = new (C) PhiNode( pf_region, Type::MEMORY,
                                                 TypeRawPtr::BOTTOM );
       // I/O is used for Prefetch
-      Node *pf_phi_abio = new (C, 3) PhiNode( pf_region, Type::ABIO );
+      Node *pf_phi_abio = new (C) PhiNode( pf_region, Type::ABIO );
 
-      Node *thread = new (C, 1) ThreadLocalNode();
+      Node *thread = new (C) ThreadLocalNode();
       transform_later(thread);
 
-      Node *eden_pf_adr = new (C, 4) AddPNode( top()/*not oop*/, thread,
+      Node *eden_pf_adr = new (C) AddPNode( top()/*not oop*/, thread,
                    _igvn.MakeConX(in_bytes(JavaThread::tlab_pf_top_offset())) );
       transform_later(eden_pf_adr);
 
-      Node *old_pf_wm = new (C, 3) LoadPNode( needgc_false,
+      Node *old_pf_wm = new (C) LoadPNode( needgc_false,
                                    contended_phi_rawmem, eden_pf_adr,
                                    TypeRawPtr::BOTTOM, TypeRawPtr::BOTTOM );
       transform_later(old_pf_wm);
 
       // check against new_eden_top
-      Node *need_pf_cmp = new (C, 3) CmpPNode( new_eden_top, old_pf_wm );
+      Node *need_pf_cmp = new (C) CmpPNode( new_eden_top, old_pf_wm );
       transform_later(need_pf_cmp);
-      Node *need_pf_bol = new (C, 2) BoolNode( need_pf_cmp, BoolTest::ge );
+      Node *need_pf_bol = new (C) BoolNode( need_pf_cmp, BoolTest::ge );
       transform_later(need_pf_bol);
-      IfNode *need_pf_iff = new (C, 2) IfNode( needgc_false, need_pf_bol,
+      IfNode *need_pf_iff = new (C) IfNode( needgc_false, need_pf_bol,
                                        PROB_UNLIKELY_MAG(4), COUNT_UNKNOWN );
       transform_later(need_pf_iff);
 
       // true node, add prefetchdistance
-      Node *need_pf_true = new (C, 1) IfTrueNode( need_pf_iff );
+      Node *need_pf_true = new (C) IfTrueNode( need_pf_iff );
       transform_later(need_pf_true);
 
-      Node *need_pf_false = new (C, 1) IfFalseNode( need_pf_iff );
+      Node *need_pf_false = new (C) IfFalseNode( need_pf_iff );
       transform_later(need_pf_false);
 
-      Node *new_pf_wmt = new (C, 4) AddPNode( top(), old_pf_wm,
+      Node *new_pf_wmt = new (C) AddPNode( top(), old_pf_wm,
                                     _igvn.MakeConX(AllocatePrefetchDistance) );
       transform_later(new_pf_wmt );
       new_pf_wmt->set_req(0, need_pf_true);
 
-      Node *store_new_wmt = new (C, 4) StorePNode( need_pf_true,
+      Node *store_new_wmt = new (C) StorePNode( need_pf_true,
                                        contended_phi_rawmem, eden_pf_adr,
                                        TypeRawPtr::BOTTOM, new_pf_wmt );
       transform_later(store_new_wmt);
@@ -1677,10 +1675,10 @@ Node* PhaseMacroExpand::prefetch_allocation(Node* i_o, Node*& needgc_false,
       uint distance = 0;
 
       for ( uint i = 0; i < lines; i++ ) {
-        prefetch_adr = new (C, 4) AddPNode( old_pf_wm, new_pf_wmt,
+        prefetch_adr = new (C) AddPNode( old_pf_wm, new_pf_wmt,
                                             _igvn.MakeConX(distance) );
         transform_later(prefetch_adr);
-        prefetch = new (C, 3) PrefetchAllocationNode( i_o, prefetch_adr );
+        prefetch = new (C) PrefetchAllocationNode( i_o, prefetch_adr );
         transform_later(prefetch);
         distance += step_size;
         i_o = prefetch;
@@ -1703,9 +1701,9 @@ Node* PhaseMacroExpand::prefetch_allocation(Node* i_o, Node*& needgc_false,
    } else if( UseTLAB && AllocatePrefetchStyle == 3 ) {
       // Insert a prefetch for each allocation.
       // This code is used for Sparc with BIS.
-      Node *pf_region = new (C, 3) RegionNode(3);
-      Node *pf_phi_rawmem = new (C, 3) PhiNode( pf_region, Type::MEMORY,
-                                                TypeRawPtr::BOTTOM );
+      Node *pf_region = new (C) RegionNode(3);
+      Node *pf_phi_rawmem = new (C) PhiNode( pf_region, Type::MEMORY,
+                                             TypeRawPtr::BOTTOM );
 
       // Generate several prefetch instructions.
       uint lines = (length != NULL) ? AllocatePrefetchLines : AllocateInstancePrefetchLines;
@@ -1713,29 +1711,29 @@ Node* PhaseMacroExpand::prefetch_allocation(Node* i_o, Node*& needgc_false,
       uint distance = AllocatePrefetchDistance;
 
       // Next cache address.
-      Node *cache_adr = new (C, 4) AddPNode(old_eden_top, old_eden_top,
+      Node *cache_adr = new (C) AddPNode(old_eden_top, old_eden_top,
                                             _igvn.MakeConX(distance));
       transform_later(cache_adr);
-      cache_adr = new (C, 2) CastP2XNode(needgc_false, cache_adr);
+      cache_adr = new (C) CastP2XNode(needgc_false, cache_adr);
       transform_later(cache_adr);
       Node* mask = _igvn.MakeConX(~(intptr_t)(step_size-1));
-      cache_adr = new (C, 3) AndXNode(cache_adr, mask);
+      cache_adr = new (C) AndXNode(cache_adr, mask);
       transform_later(cache_adr);
-      cache_adr = new (C, 2) CastX2PNode(cache_adr);
+      cache_adr = new (C) CastX2PNode(cache_adr);
       transform_later(cache_adr);
 
       // Prefetch
-      Node *prefetch = new (C, 3) PrefetchAllocationNode( contended_phi_rawmem, cache_adr );
+      Node *prefetch = new (C) PrefetchAllocationNode( contended_phi_rawmem, cache_adr );
       prefetch->set_req(0, needgc_false);
       transform_later(prefetch);
       contended_phi_rawmem = prefetch;
       Node *prefetch_adr;
       distance = step_size;
       for ( uint i = 1; i < lines; i++ ) {
-        prefetch_adr = new (C, 4) AddPNode( cache_adr, cache_adr,
+        prefetch_adr = new (C) AddPNode( cache_adr, cache_adr,
                                             _igvn.MakeConX(distance) );
         transform_later(prefetch_adr);
-        prefetch = new (C, 3) PrefetchAllocationNode( contended_phi_rawmem, prefetch_adr );
+        prefetch = new (C) PrefetchAllocationNode( contended_phi_rawmem, prefetch_adr );
         transform_later(prefetch);
         distance += step_size;
         contended_phi_rawmem = prefetch;
@@ -1749,10 +1747,10 @@ Node* PhaseMacroExpand::prefetch_allocation(Node* i_o, Node*& needgc_false,
       uint step_size = AllocatePrefetchStepSize;
       uint distance = AllocatePrefetchDistance;
       for ( uint i = 0; i < lines; i++ ) {
-        prefetch_adr = new (C, 4) AddPNode( old_eden_top, new_eden_top,
+        prefetch_adr = new (C) AddPNode( old_eden_top, new_eden_top,
                                             _igvn.MakeConX(distance) );
         transform_later(prefetch_adr);
-        prefetch = new (C, 3) PrefetchAllocationNode( i_o, prefetch_adr );
+        prefetch = new (C) PrefetchAllocationNode( i_o, prefetch_adr );
         // Do not let it float too high, since if eden_top == eden_end,
         // both might be null.
         if( i == 0 ) { // Set control for first prefetch, next follows it
@@ -2101,12 +2099,12 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
      *  }
      */
 
-    region  = new (C, 5) RegionNode(5);
+    region  = new (C) RegionNode(5);
     // create a Phi for the memory state
-    mem_phi = new (C, 5) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    mem_phi = new (C) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
 
-    Node* fast_lock_region  = new (C, 3) RegionNode(3);
-    Node* fast_lock_mem_phi = new (C, 3) PhiNode( fast_lock_region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    Node* fast_lock_region  = new (C) RegionNode(3);
+    Node* fast_lock_mem_phi = new (C) PhiNode( fast_lock_region, Type::MEMORY, TypeRawPtr::BOTTOM);
 
     // First, check mark word for the biased lock pattern.
     Node* mark_node = make_load(ctrl, mem, obj, oopDesc::mark_offset_in_bytes(), TypeX_X, TypeX_X->basic_type());
@@ -2136,10 +2134,10 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
     }
     Node *proto_node = make_load(ctrl, mem, klass_node, in_bytes(Klass::prototype_header_offset()), TypeX_X, TypeX_X->basic_type());
 
-    Node* thread = transform_later(new (C, 1) ThreadLocalNode());
-    Node* cast_thread = transform_later(new (C, 2) CastP2XNode(ctrl, thread));
-    Node* o_node = transform_later(new (C, 3) OrXNode(cast_thread, proto_node));
-    Node* x_node = transform_later(new (C, 3) XorXNode(o_node, mark_node));
+    Node* thread = transform_later(new (C) ThreadLocalNode());
+    Node* cast_thread = transform_later(new (C) CastP2XNode(ctrl, thread));
+    Node* o_node = transform_later(new (C) OrXNode(cast_thread, proto_node));
+    Node* x_node = transform_later(new (C) XorXNode(o_node, mark_node));
 
     // Get slow path - mark word does NOT match the value.
     Node* not_biased_ctrl =  opt_bits_test(ctrl, region, 3, x_node,
@@ -2162,17 +2160,17 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
     // We are going to try to reset the mark of this object to the prototype
     // value and fall through to the CAS-based locking scheme.
     Node* adr = basic_plus_adr(obj, oopDesc::mark_offset_in_bytes());
-    Node* cas = new (C, 5) StoreXConditionalNode(not_biased_ctrl, mem, adr,
-                                                 proto_node, mark_node);
+    Node* cas = new (C) StoreXConditionalNode(not_biased_ctrl, mem, adr,
+                                              proto_node, mark_node);
     transform_later(cas);
-    Node* proj = transform_later( new (C, 1) SCMemProjNode(cas));
+    Node* proj = transform_later( new (C) SCMemProjNode(cas));
     fast_lock_mem_phi->init_req(2, proj);
 
 
     // Second, check epoch bits.
-    Node* rebiased_region  = new (C, 3) RegionNode(3);
-    Node* old_phi = new (C, 3) PhiNode( rebiased_region, TypeX_X);
-    Node* new_phi = new (C, 3) PhiNode( rebiased_region, TypeX_X);
+    Node* rebiased_region  = new (C) RegionNode(3);
+    Node* old_phi = new (C) PhiNode( rebiased_region, TypeX_X);
+    Node* new_phi = new (C) PhiNode( rebiased_region, TypeX_X);
 
     // Get slow path - mark word does NOT match epoch bits.
     Node* epoch_ctrl =  opt_bits_test(ctrl, rebiased_region, 1, x_node,
@@ -2189,9 +2187,9 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
     Node* cmask   = MakeConX(markOopDesc::biased_lock_mask_in_place |
                              markOopDesc::age_mask_in_place |
                              markOopDesc::epoch_mask_in_place);
-    Node* old = transform_later(new (C, 3) AndXNode(mark_node, cmask));
-    cast_thread = transform_later(new (C, 2) CastP2XNode(ctrl, thread));
-    Node* new_mark = transform_later(new (C, 3) OrXNode(cast_thread, old));
+    Node* old = transform_later(new (C) AndXNode(mark_node, cmask));
+    cast_thread = transform_later(new (C) CastP2XNode(ctrl, thread));
+    Node* new_mark = transform_later(new (C) OrXNode(cast_thread, old));
     old_phi->init_req(1, old);
     new_phi->init_req(1, new_mark);
 
@@ -2201,10 +2199,10 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
 
     // Try to acquire the bias of the object using an atomic operation.
     // If this fails we will go in to the runtime to revoke the object's bias.
-    cas = new (C, 5) StoreXConditionalNode(rebiased_region, mem, adr,
+    cas = new (C) StoreXConditionalNode(rebiased_region, mem, adr,
                                            new_phi, old_phi);
     transform_later(cas);
-    proj = transform_later( new (C, 1) SCMemProjNode(cas));
+    proj = transform_later( new (C) SCMemProjNode(cas));
 
     // Get slow path - Failed to CAS.
     not_biased_ctrl = opt_bits_test(rebiased_region, region, 4, cas, 0, 0);
@@ -2212,8 +2210,8 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
     // region->in(4) is set to fast path - the object is rebiased to the current thread.
 
     // Failed to CAS.
-    slow_path  = new (C, 3) RegionNode(3);
-    Node *slow_mem = new (C, 3) PhiNode( slow_path, Type::MEMORY, TypeRawPtr::BOTTOM);
+    slow_path  = new (C) RegionNode(3);
+    Node *slow_mem = new (C) PhiNode( slow_path, Type::MEMORY, TypeRawPtr::BOTTOM);
 
     slow_path->init_req(1, not_biased_ctrl); // Capture slow-control
     slow_mem->init_req(1, proj);
@@ -2237,9 +2235,9 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
     lock->set_req(TypeFunc::Memory, slow_mem);
 
   } else {
-    region  = new (C, 3) RegionNode(3);
+    region  = new (C) RegionNode(3);
     // create a Phi for the memory state
-    mem_phi = new (C, 3) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    mem_phi = new (C) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
 
     // Optimize test; set region slot 2
     slow_path = opt_bits_test(ctrl, region, 2, flock, 0, 0);
@@ -2270,7 +2268,7 @@ void PhaseMacroExpand::expand_lock_node(LockNode *lock) {
   transform_later(region);
   _igvn.replace_node(_fallthroughproj, region);
 
-  Node *memproj = transform_later( new(C, 1) ProjNode(call, TypeFunc::Memory) );
+  Node *memproj = transform_later( new(C) ProjNode(call, TypeFunc::Memory) );
   mem_phi->init_req(1, memproj );
   transform_later(mem_phi);
   _igvn.replace_node(_memproj_fallthrough, mem_phi);
@@ -2295,9 +2293,9 @@ void PhaseMacroExpand::expand_unlock_node(UnlockNode *unlock) {
   if (UseOptoBiasInlining) {
     // Check for biased locking unlock case, which is a no-op.
     // See the full description in MacroAssembler::biased_locking_exit().
-    region  = new (C, 4) RegionNode(4);
+    region  = new (C) RegionNode(4);
     // create a Phi for the memory state
-    mem_phi = new (C, 4) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    mem_phi = new (C) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
     mem_phi->init_req(3, mem);
 
     Node* mark_node = make_load(ctrl, mem, obj, oopDesc::mark_offset_in_bytes(), TypeX_X, TypeX_X->basic_type());
@@ -2305,12 +2303,12 @@ void PhaseMacroExpand::expand_unlock_node(UnlockNode *unlock) {
                          markOopDesc::biased_lock_mask_in_place,
                          markOopDesc::biased_lock_pattern);
   } else {
-    region  = new (C, 3) RegionNode(3);
+    region  = new (C) RegionNode(3);
     // create a Phi for the memory state
-    mem_phi = new (C, 3) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
+    mem_phi = new (C) PhiNode( region, Type::MEMORY, TypeRawPtr::BOTTOM);
   }
 
-  FastUnlockNode *funlock = new (C, 3) FastUnlockNode( ctrl, obj, box );
+  FastUnlockNode *funlock = new (C) FastUnlockNode( ctrl, obj, box );
   funlock = transform_later( funlock )->as_FastUnlock();
   // Optimize test; set region slot 2
   Node *slow_path = opt_bits_test(ctrl, region, 2, funlock, 0, 0);
@@ -2335,7 +2333,7 @@ void PhaseMacroExpand::expand_unlock_node(UnlockNode *unlock) {
   transform_later(region);
   _igvn.replace_node(_fallthroughproj, region);
 
-  Node *memproj = transform_later( new(C, 1) ProjNode(call, TypeFunc::Memory) );
+  Node *memproj = transform_later( new(C) ProjNode(call, TypeFunc::Memory) );
   mem_phi->init_req(1, memproj );
   mem_phi->init_req(2, mem);
   transform_later(mem_phi);
diff --git a/src/share/vm/opto/macro.hpp b/src/share/vm/opto/macro.hpp
index f521e3d35..b84d8d7d7 100644
--- a/src/share/vm/opto/macro.hpp
+++ b/src/share/vm/opto/macro.hpp
@@ -52,7 +52,7 @@ private:
     return basic_plus_adr(base, base, offset);
   }
   Node* basic_plus_adr(Node* base, Node* ptr, Node* offset) {
-    Node* adr = new (C, 4) AddPNode(base, ptr, offset);
+    Node* adr = new (C) AddPNode(base, ptr, offset);
     return transform_later(adr);
   }
   Node* transform_later(Node* n) {
diff --git a/src/share/vm/opto/matcher.cpp b/src/share/vm/opto/matcher.cpp
index bf4a345ce..8ee42d13d 100644
--- a/src/share/vm/opto/matcher.cpp
+++ b/src/share/vm/opto/matcher.cpp
@@ -727,7 +727,7 @@ void Matcher::Fixup_Save_On_Entry( ) {
         tail_call_rms[tail_call_edge_cnt].Insert(OptoReg::Name(i+1));
         tail_jump_rms[tail_jump_edge_cnt].Insert(OptoReg::Name(i+1));
         halt_rms     [     halt_edge_cnt].Insert(OptoReg::Name(i+1));
-        mproj = new (C, 1) MachProjNode( start, proj_cnt, ret_rms[ret_edge_cnt], Op_RegD );
+        mproj = new (C) MachProjNode( start, proj_cnt, ret_rms[ret_edge_cnt], Op_RegD );
         proj_cnt += 2;          // Skip 2 for doubles
       }
       else if( (i&1) == 1 &&    // Else check for high half of double
@@ -753,7 +753,7 @@ void Matcher::Fixup_Save_On_Entry( ) {
         tail_call_rms[tail_call_edge_cnt].Insert(OptoReg::Name(i+1));
         tail_jump_rms[tail_jump_edge_cnt].Insert(OptoReg::Name(i+1));
         halt_rms     [     halt_edge_cnt].Insert(OptoReg::Name(i+1));
-        mproj = new (C, 1) MachProjNode( start, proj_cnt, ret_rms[ret_edge_cnt], Op_RegL );
+        mproj = new (C) MachProjNode( start, proj_cnt, ret_rms[ret_edge_cnt], Op_RegL );
         proj_cnt += 2;          // Skip 2 for longs
       }
       else if( (i&1) == 1 &&    // Else check for high half of long
@@ -768,7 +768,7 @@ void Matcher::Fixup_Save_On_Entry( ) {
         mproj = C->top();
       } else {
         // Make a projection for it off the Start
-        mproj = new (C, 1) MachProjNode( start, proj_cnt++, ret_rms[ret_edge_cnt], _register_save_type[i] );
+        mproj = new (C) MachProjNode( start, proj_cnt++, ret_rms[ret_edge_cnt], _register_save_type[i] );
       }
 
       ret_edge_cnt ++;
@@ -821,13 +821,13 @@ void Matcher::init_spill_mask( Node *ret ) {
 
   // Compute generic short-offset Loads
 #ifdef _LP64
-  MachNode *spillCP = match_tree(new (C, 3) LoadNNode(NULL,mem,fp,atp,TypeInstPtr::BOTTOM));
+  MachNode *spillCP = match_tree(new (C) LoadNNode(NULL,mem,fp,atp,TypeInstPtr::BOTTOM));
 #endif
-  MachNode *spillI  = match_tree(new (C, 3) LoadINode(NULL,mem,fp,atp));
-  MachNode *spillL  = match_tree(new (C, 3) LoadLNode(NULL,mem,fp,atp));
-  MachNode *spillF  = match_tree(new (C, 3) LoadFNode(NULL,mem,fp,atp));
-  MachNode *spillD  = match_tree(new (C, 3) LoadDNode(NULL,mem,fp,atp));
-  MachNode *spillP  = match_tree(new (C, 3) LoadPNode(NULL,mem,fp,atp,TypeInstPtr::BOTTOM));
+  MachNode *spillI  = match_tree(new (C) LoadINode(NULL,mem,fp,atp));
+  MachNode *spillL  = match_tree(new (C) LoadLNode(NULL,mem,fp,atp));
+  MachNode *spillF  = match_tree(new (C) LoadFNode(NULL,mem,fp,atp));
+  MachNode *spillD  = match_tree(new (C) LoadDNode(NULL,mem,fp,atp));
+  MachNode *spillP  = match_tree(new (C) LoadPNode(NULL,mem,fp,atp,TypeInstPtr::BOTTOM));
   assert(spillI != NULL && spillL != NULL && spillF != NULL &&
          spillD != NULL && spillP != NULL, "");
 
@@ -844,19 +844,19 @@ void Matcher::init_spill_mask( Node *ret ) {
   // Vector regmasks.
   if (Matcher::vector_size_supported(T_BYTE,4)) {
     TypeVect::VECTS = TypeVect::make(T_BYTE, 4);
-    MachNode *spillVectS = match_tree(new (C, 3) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTS));
+    MachNode *spillVectS = match_tree(new (C) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTS));
     idealreg2regmask[Op_VecS] = &spillVectS->out_RegMask();
   }
   if (Matcher::vector_size_supported(T_FLOAT,2)) {
-    MachNode *spillVectD = match_tree(new (C, 3) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTD));
+    MachNode *spillVectD = match_tree(new (C) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTD));
     idealreg2regmask[Op_VecD] = &spillVectD->out_RegMask();
   }
   if (Matcher::vector_size_supported(T_FLOAT,4)) {
-    MachNode *spillVectX = match_tree(new (C, 3) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTX));
+    MachNode *spillVectX = match_tree(new (C) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTX));
     idealreg2regmask[Op_VecX] = &spillVectX->out_RegMask();
   }
   if (Matcher::vector_size_supported(T_FLOAT,8)) {
-    MachNode *spillVectY = match_tree(new (C, 3) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTY));
+    MachNode *spillVectY = match_tree(new (C) LoadVectorNode(NULL,mem,fp,atp,TypeVect::VECTY));
     idealreg2regmask[Op_VecY] = &spillVectY->out_RegMask();
   }
 }
@@ -1303,7 +1303,7 @@ MachNode *Matcher::match_sfpt( SafePointNode *sfpt ) {
     // is excluded on the max-per-method basis, debug info cannot land in
     // this killed area.
     uint r_cnt = mcall->tf()->range()->cnt();
-    MachProjNode *proj = new (C, 1) MachProjNode( mcall, r_cnt+10000, RegMask::Empty, MachProjNode::fat_proj );
+    MachProjNode *proj = new (C) MachProjNode( mcall, r_cnt+10000, RegMask::Empty, MachProjNode::fat_proj );
     if (!RegMask::can_represent_arg(OptoReg::Name(out_arg_limit_per_call-1))) {
       C->record_method_not_compilable_all_tiers("unsupported outgoing calling sequence");
     } else {
@@ -2134,10 +2134,10 @@ void Matcher::find_shared( Node *n ) {
       case Op_CompareAndSwapP:
       case Op_CompareAndSwapN: {   // Convert trinary to binary-tree
         Node *newval = n->in(MemNode::ValueIn );
-        Node *oldval  = n->in(LoadStoreNode::ExpectedIn);
-        Node *pair = new (C, 3) BinaryNode( oldval, newval );
+        Node *oldval  = n->in(LoadStoreConditionalNode::ExpectedIn);
+        Node *pair = new (C) BinaryNode( oldval, newval );
         n->set_req(MemNode::ValueIn,pair);
-        n->del_req(LoadStoreNode::ExpectedIn);
+        n->del_req(LoadStoreConditionalNode::ExpectedIn);
         break;
       }
       case Op_CMoveD:              // Convert trinary to binary-tree
@@ -2150,22 +2150,22 @@ void Matcher::find_shared( Node *n ) {
         // we could move this code up next to the graph reshaping for IfNodes
         // or vice-versa, but I do not want to debug this for Ladybird.
         // 10/2/2000 CNC.
-        Node *pair1 = new (C, 3) BinaryNode(n->in(1),n->in(1)->in(1));
+        Node *pair1 = new (C) BinaryNode(n->in(1),n->in(1)->in(1));
         n->set_req(1,pair1);
-        Node *pair2 = new (C, 3) BinaryNode(n->in(2),n->in(3));
+        Node *pair2 = new (C) BinaryNode(n->in(2),n->in(3));
         n->set_req(2,pair2);
         n->del_req(3);
         break;
       }
       case Op_LoopLimit: {
-        Node *pair1 = new (C, 3) BinaryNode(n->in(1),n->in(2));
+        Node *pair1 = new (C) BinaryNode(n->in(1),n->in(2));
         n->set_req(1,pair1);
         n->set_req(2,n->in(3));
         n->del_req(3);
         break;
       }
       case Op_StrEquals: {
-        Node *pair1 = new (C, 3) BinaryNode(n->in(2),n->in(3));
+        Node *pair1 = new (C) BinaryNode(n->in(2),n->in(3));
         n->set_req(2,pair1);
         n->set_req(3,n->in(4));
         n->del_req(4);
@@ -2173,9 +2173,9 @@ void Matcher::find_shared( Node *n ) {
       }
       case Op_StrComp:
       case Op_StrIndexOf: {
-        Node *pair1 = new (C, 3) BinaryNode(n->in(2),n->in(3));
+        Node *pair1 = new (C) BinaryNode(n->in(2),n->in(3));
         n->set_req(2,pair1);
-        Node *pair2 = new (C, 3) BinaryNode(n->in(4),n->in(5));
+        Node *pair2 = new (C) BinaryNode(n->in(4),n->in(5));
         n->set_req(3,pair2);
         n->del_req(5);
         n->del_req(4);
diff --git a/src/share/vm/opto/memnode.cpp b/src/share/vm/opto/memnode.cpp
index 1cbe31756..657d33105 100644
--- a/src/share/vm/opto/memnode.cpp
+++ b/src/share/vm/opto/memnode.cpp
@@ -883,25 +883,25 @@ Node *LoadNode::make( PhaseGVN& gvn, Node *ctl, Node *mem, Node *adr, const Type
           rt->isa_oopptr() || is_immutable_value(adr),
           "raw memory operations should have control edge");
   switch (bt) {
-  case T_BOOLEAN: return new (C, 3) LoadUBNode(ctl, mem, adr, adr_type, rt->is_int()    );
-  case T_BYTE:    return new (C, 3) LoadBNode (ctl, mem, adr, adr_type, rt->is_int()    );
-  case T_INT:     return new (C, 3) LoadINode (ctl, mem, adr, adr_type, rt->is_int()    );
-  case T_CHAR:    return new (C, 3) LoadUSNode(ctl, mem, adr, adr_type, rt->is_int()    );
-  case T_SHORT:   return new (C, 3) LoadSNode (ctl, mem, adr, adr_type, rt->is_int()    );
-  case T_LONG:    return new (C, 3) LoadLNode (ctl, mem, adr, adr_type, rt->is_long()   );
-  case T_FLOAT:   return new (C, 3) LoadFNode (ctl, mem, adr, adr_type, rt              );
-  case T_DOUBLE:  return new (C, 3) LoadDNode (ctl, mem, adr, adr_type, rt              );
-  case T_ADDRESS: return new (C, 3) LoadPNode (ctl, mem, adr, adr_type, rt->is_ptr()    );
+  case T_BOOLEAN: return new (C) LoadUBNode(ctl, mem, adr, adr_type, rt->is_int()    );
+  case T_BYTE:    return new (C) LoadBNode (ctl, mem, adr, adr_type, rt->is_int()    );
+  case T_INT:     return new (C) LoadINode (ctl, mem, adr, adr_type, rt->is_int()    );
+  case T_CHAR:    return new (C) LoadUSNode(ctl, mem, adr, adr_type, rt->is_int()    );
+  case T_SHORT:   return new (C) LoadSNode (ctl, mem, adr, adr_type, rt->is_int()    );
+  case T_LONG:    return new (C) LoadLNode (ctl, mem, adr, adr_type, rt->is_long()   );
+  case T_FLOAT:   return new (C) LoadFNode (ctl, mem, adr, adr_type, rt              );
+  case T_DOUBLE:  return new (C) LoadDNode (ctl, mem, adr, adr_type, rt              );
+  case T_ADDRESS: return new (C) LoadPNode (ctl, mem, adr, adr_type, rt->is_ptr()    );
   case T_OBJECT:
 #ifdef _LP64
     if (adr->bottom_type()->is_ptr_to_narrowoop()) {
-      Node* load  = gvn.transform(new (C, 3) LoadNNode(ctl, mem, adr, adr_type, rt->make_narrowoop()));
-      return new (C, 2) DecodeNNode(load, load->bottom_type()->make_ptr());
+      Node* load  = gvn.transform(new (C) LoadNNode(ctl, mem, adr, adr_type, rt->make_narrowoop()));
+      return new (C) DecodeNNode(load, load->bottom_type()->make_ptr());
     } else
 #endif
     {
       assert(!adr->bottom_type()->is_ptr_to_narrowoop(), "should have got back a narrow oop");
-      return new (C, 3) LoadPNode(ctl, mem, adr, adr_type, rt->is_oopptr());
+      return new (C) LoadPNode(ctl, mem, adr, adr_type, rt->is_oopptr());
     }
   }
   ShouldNotReachHere();
@@ -910,7 +910,7 @@ Node *LoadNode::make( PhaseGVN& gvn, Node *ctl, Node *mem, Node *adr, const Type
 
 LoadLNode* LoadLNode::make_atomic(Compile *C, Node* ctl, Node* mem, Node* adr, const TypePtr* adr_type, const Type* rt) {
   bool require_atomic = true;
-  return new (C, 3) LoadLNode(ctl, mem, adr, adr_type, rt->is_long(), require_atomic);
+  return new (C) LoadLNode(ctl, mem, adr, adr_type, rt->is_long(), require_atomic);
 }
 
 
@@ -1244,20 +1244,20 @@ Node* LoadNode::eliminate_autobox(PhaseGVN* phase) {
           // Add up all the offsets making of the address of the load
           Node* result = elements[0];
           for (int i = 1; i < count; i++) {
-            result = phase->transform(new (phase->C, 3) AddXNode(result, elements[i]));
+            result = phase->transform(new (phase->C) AddXNode(result, elements[i]));
           }
           // Remove the constant offset from the address and then
           // remove the scaling of the offset to recover the original index.
-          result = phase->transform(new (phase->C, 3) AddXNode(result, phase->MakeConX(-offset)));
+          result = phase->transform(new (phase->C) AddXNode(result, phase->MakeConX(-offset)));
           if (result->Opcode() == Op_LShiftX && result->in(2) == phase->intcon(shift)) {
             // Peel the shift off directly but wrap it in a dummy node
             // since Ideal can't return existing nodes
-            result = new (phase->C, 3) RShiftXNode(result->in(1), phase->intcon(0));
+            result = new (phase->C) RShiftXNode(result->in(1), phase->intcon(0));
           } else {
-            result = new (phase->C, 3) RShiftXNode(result, phase->intcon(shift));
+            result = new (phase->C) RShiftXNode(result, phase->intcon(shift));
           }
 #ifdef _LP64
-          result = new (phase->C, 2) ConvL2INode(phase->transform(result));
+          result = new (phase->C) ConvL2INode(phase->transform(result));
 #endif
           return result;
         }
@@ -1320,7 +1320,7 @@ Node *LoadNode::split_through_phi(PhaseGVN *phase) {
   int this_offset = addr_t->offset();
   int this_iid    = addr_t->is_oopptr()->instance_id();
   PhaseIterGVN *igvn = phase->is_IterGVN();
-  Node *phi = new (igvn->C, region->req()) PhiNode(region, this_type, NULL, this_iid, this_index, this_offset);
+  Node *phi = new (igvn->C) PhiNode(region, this_type, NULL, this_iid, this_index, this_offset);
   for (uint i = 1; i < region->req(); i++) {
     Node *x;
     Node* the_clone = NULL;
@@ -1771,8 +1771,8 @@ Node *LoadBNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node* mem = in(MemNode::Memory);
   Node* value = can_see_stored_value(mem,phase);
   if( value && !phase->type(value)->higher_equal( _type ) ) {
-    Node *result = phase->transform( new (phase->C, 3) LShiftINode(value, phase->intcon(24)) );
-    return new (phase->C, 3) RShiftINode(result, phase->intcon(24));
+    Node *result = phase->transform( new (phase->C) LShiftINode(value, phase->intcon(24)) );
+    return new (phase->C) RShiftINode(result, phase->intcon(24));
   }
   // Identity call will handle the case where truncation is not needed.
   return LoadNode::Ideal(phase, can_reshape);
@@ -1803,7 +1803,7 @@ Node* LoadUBNode::Ideal(PhaseGVN* phase, bool can_reshape) {
   Node* mem = in(MemNode::Memory);
   Node* value = can_see_stored_value(mem, phase);
   if (value && !phase->type(value)->higher_equal(_type))
-    return new (phase->C, 3) AndINode(value, phase->intcon(0xFF));
+    return new (phase->C) AndINode(value, phase->intcon(0xFF));
   // Identity call will handle the case where truncation is not needed.
   return LoadNode::Ideal(phase, can_reshape);
 }
@@ -1833,7 +1833,7 @@ Node *LoadUSNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node* mem = in(MemNode::Memory);
   Node* value = can_see_stored_value(mem,phase);
   if( value && !phase->type(value)->higher_equal( _type ) )
-    return new (phase->C, 3) AndINode(value,phase->intcon(0xFFFF));
+    return new (phase->C) AndINode(value,phase->intcon(0xFFFF));
   // Identity call will handle the case where truncation is not needed.
   return LoadNode::Ideal(phase, can_reshape);
 }
@@ -1863,8 +1863,8 @@ Node *LoadSNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node* mem = in(MemNode::Memory);
   Node* value = can_see_stored_value(mem,phase);
   if( value && !phase->type(value)->higher_equal( _type ) ) {
-    Node *result = phase->transform( new (phase->C, 3) LShiftINode(value, phase->intcon(16)) );
-    return new (phase->C, 3) RShiftINode(result, phase->intcon(16));
+    Node *result = phase->transform( new (phase->C) LShiftINode(value, phase->intcon(16)) );
+    return new (phase->C) RShiftINode(result, phase->intcon(16));
   }
   // Identity call will handle the case where truncation is not needed.
   return LoadNode::Ideal(phase, can_reshape);
@@ -1896,12 +1896,12 @@ Node *LoadKlassNode::make( PhaseGVN& gvn, Node *mem, Node *adr, const TypePtr* a
 #ifdef _LP64
   if (adr_type->is_ptr_to_narrowoop()) {
     assert(UseCompressedKlassPointers, "no compressed klasses");
-    Node* load_klass = gvn.transform(new (C, 3) LoadNKlassNode(ctl, mem, adr, at, tk->make_narrowoop()));
-    return new (C, 2) DecodeNNode(load_klass, load_klass->bottom_type()->make_ptr());
+    Node* load_klass = gvn.transform(new (C) LoadNKlassNode(ctl, mem, adr, at, tk->make_narrowoop()));
+    return new (C) DecodeNNode(load_klass, load_klass->bottom_type()->make_ptr());
   }
 #endif
   assert(!adr_type->is_ptr_to_narrowoop(), "should have got back a narrow oop");
-  return new (C, 3) LoadKlassNode(ctl, mem, adr, at, tk);
+  return new (C) LoadKlassNode(ctl, mem, adr, at, tk);
 }
 
 //------------------------------Value------------------------------------------
@@ -2123,7 +2123,7 @@ Node* LoadNKlassNode::Identity( PhaseTransform *phase ) {
   if( t == Type::TOP ) return x;
   if( t->isa_narrowoop()) return x;
 
-  return phase->transform(new (phase->C, 2) EncodePNode(x, t->make_narrowoop()));
+  return phase->transform(new (phase->C) EncodePNode(x, t->make_narrowoop()));
 }
 
 //------------------------------Value-----------------------------------------
@@ -2217,13 +2217,13 @@ StoreNode* StoreNode::make( PhaseGVN& gvn, Node* ctl, Node* mem, Node* adr, cons
 
   switch (bt) {
   case T_BOOLEAN:
-  case T_BYTE:    return new (C, 4) StoreBNode(ctl, mem, adr, adr_type, val);
-  case T_INT:     return new (C, 4) StoreINode(ctl, mem, adr, adr_type, val);
+  case T_BYTE:    return new (C) StoreBNode(ctl, mem, adr, adr_type, val);
+  case T_INT:     return new (C) StoreINode(ctl, mem, adr, adr_type, val);
   case T_CHAR:
-  case T_SHORT:   return new (C, 4) StoreCNode(ctl, mem, adr, adr_type, val);
-  case T_LONG:    return new (C, 4) StoreLNode(ctl, mem, adr, adr_type, val);
-  case T_FLOAT:   return new (C, 4) StoreFNode(ctl, mem, adr, adr_type, val);
-  case T_DOUBLE:  return new (C, 4) StoreDNode(ctl, mem, adr, adr_type, val);
+  case T_SHORT:   return new (C) StoreCNode(ctl, mem, adr, adr_type, val);
+  case T_LONG:    return new (C) StoreLNode(ctl, mem, adr, adr_type, val);
+  case T_FLOAT:   return new (C) StoreFNode(ctl, mem, adr, adr_type, val);
+  case T_DOUBLE:  return new (C) StoreDNode(ctl, mem, adr, adr_type, val);
   case T_METADATA:
   case T_ADDRESS:
   case T_OBJECT:
@@ -2231,12 +2231,12 @@ StoreNode* StoreNode::make( PhaseGVN& gvn, Node* ctl, Node* mem, Node* adr, cons
     if (adr->bottom_type()->is_ptr_to_narrowoop() ||
         (UseCompressedKlassPointers && val->bottom_type()->isa_klassptr() &&
          adr->bottom_type()->isa_rawptr())) {
-      val = gvn.transform(new (C, 2) EncodePNode(val, val->bottom_type()->make_narrowoop()));
-      return new (C, 4) StoreNNode(ctl, mem, adr, adr_type, val);
+      val = gvn.transform(new (C) EncodePNode(val, val->bottom_type()->make_narrowoop()));
+      return new (C) StoreNNode(ctl, mem, adr, adr_type, val);
     } else
 #endif
     {
-      return new (C, 4) StorePNode(ctl, mem, adr, adr_type, val);
+      return new (C) StorePNode(ctl, mem, adr, adr_type, val);
     }
   }
   ShouldNotReachHere();
@@ -2245,7 +2245,7 @@ StoreNode* StoreNode::make( PhaseGVN& gvn, Node* ctl, Node* mem, Node* adr, cons
 
 StoreLNode* StoreLNode::make_atomic(Compile *C, Node* ctl, Node* mem, Node* adr, const TypePtr* adr_type, Node* val) {
   bool require_atomic = true;
-  return new (C, 4) StoreLNode(ctl, mem, adr, adr_type, val, require_atomic);
+  return new (C) StoreLNode(ctl, mem, adr, adr_type, val, require_atomic);
 }
 
 
@@ -2552,14 +2552,38 @@ const Type * SCMemProjNode::Value( PhaseTransform *phase ) const
 }
 
 //=============================================================================
-LoadStoreNode::LoadStoreNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex ) : Node(5) {
+//----------------------------------LoadStoreNode------------------------------
+LoadStoreNode::LoadStoreNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at, const Type* rt, uint required )
+  : Node(required),
+    _type(rt),
+    _adr_type(at)
+{
   init_req(MemNode::Control, c  );
   init_req(MemNode::Memory , mem);
   init_req(MemNode::Address, adr);
   init_req(MemNode::ValueIn, val);
-  init_req(         ExpectedIn, ex );
   init_class_id(Class_LoadStore);
+}
+
+uint LoadStoreNode::ideal_reg() const {
+  return _type->ideal_reg();
+}
+
+bool LoadStoreNode::result_not_used() const {
+  for( DUIterator_Fast imax, i = fast_outs(imax); i < imax; i++ ) {
+    Node *x = fast_out(i);
+    if (x->Opcode() == Op_SCMemProj) continue;
+    return false;
+  }
+  return true;
+}
+
+uint LoadStoreNode::size_of() const { return sizeof(*this); }
 
+//=============================================================================
+//----------------------------------LoadStoreConditionalNode--------------------
+LoadStoreConditionalNode::LoadStoreConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex ) : LoadStoreNode(c, mem, adr, val, NULL, TypeInt::BOOL, 5) {
+  init_req(ExpectedIn, ex );
 }
 
 //=============================================================================
@@ -2614,12 +2638,12 @@ Node *ClearArrayNode::Ideal(PhaseGVN *phase, bool can_reshape){
 
   Node *zero = phase->makecon(TypeLong::ZERO);
   Node *off  = phase->MakeConX(BytesPerLong);
-  mem = new (phase->C, 4) StoreLNode(in(0),mem,adr,atp,zero);
+  mem = new (phase->C) StoreLNode(in(0),mem,adr,atp,zero);
   count--;
   while( count-- ) {
     mem = phase->transform(mem);
-    adr = phase->transform(new (phase->C, 4) AddPNode(base,adr,off));
-    mem = new (phase->C, 4) StoreLNode(in(0),mem,adr,atp,zero);
+    adr = phase->transform(new (phase->C) AddPNode(base,adr,off));
+    mem = new (phase->C) StoreLNode(in(0),mem,adr,atp,zero);
   }
   return mem;
 }
@@ -2660,7 +2684,7 @@ Node* ClearArrayNode::clear_memory(Node* ctl, Node* mem, Node* dest,
 
   int unit = BytesPerLong;
   if ((offset % unit) != 0) {
-    Node* adr = new (C, 4) AddPNode(dest, dest, phase->MakeConX(offset));
+    Node* adr = new (C) AddPNode(dest, dest, phase->MakeConX(offset));
     adr = phase->transform(adr);
     const TypePtr* atp = TypeRawPtr::BOTTOM;
     mem = StoreNode::make(*phase, ctl, mem, adr, atp, phase->zerocon(T_INT), T_INT);
@@ -2690,16 +2714,16 @@ Node* ClearArrayNode::clear_memory(Node* ctl, Node* mem, Node* dest,
   // Scale to the unit required by the CPU:
   if (!Matcher::init_array_count_is_in_bytes) {
     Node* shift = phase->intcon(exact_log2(unit));
-    zbase = phase->transform( new(C,3) URShiftXNode(zbase, shift) );
-    zend  = phase->transform( new(C,3) URShiftXNode(zend,  shift) );
+    zbase = phase->transform( new(C) URShiftXNode(zbase, shift) );
+    zend  = phase->transform( new(C) URShiftXNode(zend,  shift) );
   }
 
-  Node* zsize = phase->transform( new(C,3) SubXNode(zend, zbase) );
+  Node* zsize = phase->transform( new(C) SubXNode(zend, zbase) );
   Node* zinit = phase->zerocon((unit == BytesPerLong) ? T_LONG : T_INT);
 
   // Bulk clear double-words
-  Node* adr = phase->transform( new(C,4) AddPNode(dest, dest, start_offset) );
-  mem = new (C, 4) ClearArrayNode(ctl, mem, zsize, adr);
+  Node* adr = phase->transform( new(C) AddPNode(dest, dest, start_offset) );
+  mem = new (C) ClearArrayNode(ctl, mem, zsize, adr);
   return phase->transform(mem);
 }
 
@@ -2723,7 +2747,7 @@ Node* ClearArrayNode::clear_memory(Node* ctl, Node* mem, Node* dest,
                        start_offset, phase->MakeConX(done_offset), phase);
   }
   if (done_offset < end_offset) { // emit the final 32-bit store
-    Node* adr = new (C, 4) AddPNode(dest, dest, phase->MakeConX(done_offset));
+    Node* adr = new (C) AddPNode(dest, dest, phase->MakeConX(done_offset));
     adr = phase->transform(adr);
     const TypePtr* atp = TypeRawPtr::BOTTOM;
     mem = StoreNode::make(*phase, ctl, mem, adr, atp, phase->zerocon(T_INT), T_INT);
@@ -2789,16 +2813,15 @@ uint MemBarNode::cmp( const Node &n ) const {
 
 //------------------------------make-------------------------------------------
 MemBarNode* MemBarNode::make(Compile* C, int opcode, int atp, Node* pn) {
-  int len = Precedent + (pn == NULL? 0: 1);
   switch (opcode) {
-  case Op_MemBarAcquire:   return new(C, len) MemBarAcquireNode(C,  atp, pn);
-  case Op_MemBarRelease:   return new(C, len) MemBarReleaseNode(C,  atp, pn);
-  case Op_MemBarAcquireLock: return new(C, len) MemBarAcquireLockNode(C,  atp, pn);
-  case Op_MemBarReleaseLock: return new(C, len) MemBarReleaseLockNode(C,  atp, pn);
-  case Op_MemBarVolatile:  return new(C, len) MemBarVolatileNode(C, atp, pn);
-  case Op_MemBarCPUOrder:  return new(C, len) MemBarCPUOrderNode(C, atp, pn);
-  case Op_Initialize:      return new(C, len) InitializeNode(C,     atp, pn);
-  case Op_MemBarStoreStore: return new(C, len) MemBarStoreStoreNode(C,  atp, pn);
+  case Op_MemBarAcquire:   return new(C) MemBarAcquireNode(C,  atp, pn);
+  case Op_MemBarRelease:   return new(C) MemBarReleaseNode(C,  atp, pn);
+  case Op_MemBarAcquireLock: return new(C) MemBarAcquireLockNode(C,  atp, pn);
+  case Op_MemBarReleaseLock: return new(C) MemBarReleaseLockNode(C,  atp, pn);
+  case Op_MemBarVolatile:  return new(C) MemBarVolatileNode(C, atp, pn);
+  case Op_MemBarCPUOrder:  return new(C) MemBarCPUOrderNode(C, atp, pn);
+  case Op_Initialize:      return new(C) InitializeNode(C,     atp, pn);
+  case Op_MemBarStoreStore: return new(C) MemBarStoreStoreNode(C,  atp, pn);
   default:                 ShouldNotReachHere(); return NULL;
   }
 }
@@ -2828,7 +2851,7 @@ Node *MemBarNode::Ideal(PhaseGVN *phase, bool can_reshape) {
         igvn->replace_node(proj_out(TypeFunc::Control), in(TypeFunc::Control));
         // Must return either the original node (now dead) or a new node
         // (Do not return a top here, since that would break the uniqueness of top.)
-        return new (phase->C, 1) ConINode(TypeInt::ZERO);
+        return new (phase->C) ConINode(TypeInt::ZERO);
       }
     }
   }
@@ -2849,7 +2872,7 @@ Node *MemBarNode::match( const ProjNode *proj, const Matcher *m ) {
   switch (proj->_con) {
   case TypeFunc::Control:
   case TypeFunc::Memory:
-    return new (m->C, 1) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
+    return new (m->C) MachProjNode(this,proj->_con,RegMask::Empty,MachProjNode::unmatched_proj);
   }
   ShouldNotReachHere();
   return NULL;
@@ -3193,7 +3216,7 @@ Node* InitializeNode::make_raw_address(intptr_t offset,
   Node* addr = in(RawAddress);
   if (offset != 0) {
     Compile* C = phase->C;
-    addr = phase->transform( new (C, 4) AddPNode(C->top(), addr,
+    addr = phase->transform( new (C) AddPNode(C->top(), addr,
                                                  phase->MakeConX(offset)) );
   }
   return addr;
@@ -3882,7 +3905,7 @@ MergeMemNode::MergeMemNode(Node *new_base) : Node(1+Compile::AliasIdxRaw) {
 // Make a new, untransformed MergeMem with the same base as 'mem'.
 // If mem is itself a MergeMem, populate the result with the same edges.
 MergeMemNode* MergeMemNode::make(Compile* C, Node* mem) {
-  return new(C, 1+Compile::AliasIdxRaw) MergeMemNode(mem);
+  return new(C) MergeMemNode(mem);
 }
 
 //------------------------------cmp--------------------------------------------
diff --git a/src/share/vm/opto/memnode.hpp b/src/share/vm/opto/memnode.hpp
index f79f615d1..6f722f938 100644
--- a/src/share/vm/opto/memnode.hpp
+++ b/src/share/vm/opto/memnode.hpp
@@ -657,23 +657,36 @@ public:
 //------------------------------LoadStoreNode---------------------------
 // Note: is_Mem() method returns 'true' for this class.
 class LoadStoreNode : public Node {
+private:
+  const Type* const _type;      // What kind of value is loaded?
+  const TypePtr* _adr_type;     // What kind of memory is being addressed?
+  virtual uint size_of() const; // Size is bigger
+public:
+  LoadStoreNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at, const Type* rt, uint required );
+  virtual bool depends_only_on_test() const { return false; }
+  virtual uint match_edge(uint idx) const { return idx == MemNode::Address || idx == MemNode::ValueIn; }
+
+  virtual const Type *bottom_type() const { return _type; }
+  virtual uint ideal_reg() const;
+  virtual const class TypePtr *adr_type() const { return _adr_type; }  // returns bottom_type of address
+
+  bool result_not_used() const;
+};
+
+class LoadStoreConditionalNode : public LoadStoreNode {
 public:
   enum {
     ExpectedIn = MemNode::ValueIn+1 // One more input than MemNode
   };
-  LoadStoreNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex);
-  virtual bool depends_only_on_test() const { return false; }
-  virtual const Type *bottom_type() const { return TypeInt::BOOL; }
-  virtual uint ideal_reg() const { return Op_RegI; }
-  virtual uint match_edge(uint idx) const { return idx == MemNode::Address || idx == MemNode::ValueIn; }
+  LoadStoreConditionalNode(Node *c, Node *mem, Node *adr, Node *val, Node *ex);
 };
 
 //------------------------------StorePConditionalNode---------------------------
 // Conditionally store pointer to memory, if no change since prior
 // load-locked.  Sets flags for success or failure of the store.
-class StorePConditionalNode : public LoadStoreNode {
+class StorePConditionalNode : public LoadStoreConditionalNode {
 public:
-  StorePConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ll ) : LoadStoreNode(c, mem, adr, val, ll) { }
+  StorePConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ll ) : LoadStoreConditionalNode(c, mem, adr, val, ll) { }
   virtual int Opcode() const;
   // Produces flags
   virtual uint ideal_reg() const { return Op_RegFlags; }
@@ -682,9 +695,9 @@ public:
 //------------------------------StoreIConditionalNode---------------------------
 // Conditionally store int to memory, if no change since prior
 // load-locked.  Sets flags for success or failure of the store.
-class StoreIConditionalNode : public LoadStoreNode {
+class StoreIConditionalNode : public LoadStoreConditionalNode {
 public:
-  StoreIConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ii ) : LoadStoreNode(c, mem, adr, val, ii) { }
+  StoreIConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ii ) : LoadStoreConditionalNode(c, mem, adr, val, ii) { }
   virtual int Opcode() const;
   // Produces flags
   virtual uint ideal_reg() const { return Op_RegFlags; }
@@ -693,9 +706,9 @@ public:
 //------------------------------StoreLConditionalNode---------------------------
 // Conditionally store long to memory, if no change since prior
 // load-locked.  Sets flags for success or failure of the store.
-class StoreLConditionalNode : public LoadStoreNode {
+class StoreLConditionalNode : public LoadStoreConditionalNode {
 public:
-  StoreLConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ll ) : LoadStoreNode(c, mem, adr, val, ll) { }
+  StoreLConditionalNode( Node *c, Node *mem, Node *adr, Node *val, Node *ll ) : LoadStoreConditionalNode(c, mem, adr, val, ll) { }
   virtual int Opcode() const;
   // Produces flags
   virtual uint ideal_reg() const { return Op_RegFlags; }
@@ -703,32 +716,75 @@ public:
 
 
 //------------------------------CompareAndSwapLNode---------------------------
-class CompareAndSwapLNode : public LoadStoreNode {
+class CompareAndSwapLNode : public LoadStoreConditionalNode {
 public:
-  CompareAndSwapLNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreNode(c, mem, adr, val, ex) { }
+  CompareAndSwapLNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreConditionalNode(c, mem, adr, val, ex) { }
   virtual int Opcode() const;
 };
 
 
 //------------------------------CompareAndSwapINode---------------------------
-class CompareAndSwapINode : public LoadStoreNode {
+class CompareAndSwapINode : public LoadStoreConditionalNode {
 public:
-  CompareAndSwapINode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreNode(c, mem, adr, val, ex) { }
+  CompareAndSwapINode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreConditionalNode(c, mem, adr, val, ex) { }
   virtual int Opcode() const;
 };
 
 
 //------------------------------CompareAndSwapPNode---------------------------
-class CompareAndSwapPNode : public LoadStoreNode {
+class CompareAndSwapPNode : public LoadStoreConditionalNode {
 public:
-  CompareAndSwapPNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreNode(c, mem, adr, val, ex) { }
+  CompareAndSwapPNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreConditionalNode(c, mem, adr, val, ex) { }
   virtual int Opcode() const;
 };
 
 //------------------------------CompareAndSwapNNode---------------------------
-class CompareAndSwapNNode : public LoadStoreNode {
+class CompareAndSwapNNode : public LoadStoreConditionalNode {
+public:
+  CompareAndSwapNNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreConditionalNode(c, mem, adr, val, ex) { }
+  virtual int Opcode() const;
+};
+
+//------------------------------GetAndAddINode---------------------------
+class GetAndAddINode : public LoadStoreNode {
+public:
+  GetAndAddINode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at ) : LoadStoreNode(c, mem, adr, val, at, TypeInt::INT, 4) { }
+  virtual int Opcode() const;
+};
+
+//------------------------------GetAndAddLNode---------------------------
+class GetAndAddLNode : public LoadStoreNode {
+public:
+  GetAndAddLNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at ) : LoadStoreNode(c, mem, adr, val, at, TypeLong::LONG, 4) { }
+  virtual int Opcode() const;
+};
+
+
+//------------------------------GetAndSetINode---------------------------
+class GetAndSetINode : public LoadStoreNode {
+public:
+  GetAndSetINode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at ) : LoadStoreNode(c, mem, adr, val, at, TypeInt::INT, 4) { }
+  virtual int Opcode() const;
+};
+
+//------------------------------GetAndSetINode---------------------------
+class GetAndSetLNode : public LoadStoreNode {
+public:
+  GetAndSetLNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at ) : LoadStoreNode(c, mem, adr, val, at, TypeLong::LONG, 4) { }
+  virtual int Opcode() const;
+};
+
+//------------------------------GetAndSetPNode---------------------------
+class GetAndSetPNode : public LoadStoreNode {
+public:
+  GetAndSetPNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at, const Type* t ) : LoadStoreNode(c, mem, adr, val, at, t, 4) { }
+  virtual int Opcode() const;
+};
+
+//------------------------------GetAndSetNNode---------------------------
+class GetAndSetNNode : public LoadStoreNode {
 public:
-  CompareAndSwapNNode( Node *c, Node *mem, Node *adr, Node *val, Node *ex) : LoadStoreNode(c, mem, adr, val, ex) { }
+  GetAndSetNNode( Node *c, Node *mem, Node *adr, Node *val, const TypePtr* at, const Type* t ) : LoadStoreNode(c, mem, adr, val, at, t, 4) { }
   virtual int Opcode() const;
 };
 
diff --git a/src/share/vm/opto/mulnode.cpp b/src/share/vm/opto/mulnode.cpp
index f08631928..f53310e81 100644
--- a/src/share/vm/opto/mulnode.cpp
+++ b/src/share/vm/opto/mulnode.cpp
@@ -198,22 +198,22 @@ Node *MulINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node *res = NULL;
   jint bit1 = con & -con;       // Extract low bit
   if( bit1 == con ) {           // Found a power of 2?
-    res = new (phase->C, 3) LShiftINode( in(1), phase->intcon(log2_intptr(bit1)) );
+    res = new (phase->C) LShiftINode( in(1), phase->intcon(log2_intptr(bit1)) );
   } else {
 
     // Check for constant with 2 bits set
     jint bit2 = con-bit1;
     bit2 = bit2 & -bit2;          // Extract 2nd bit
     if( bit2 + bit1 == con ) {    // Found all bits in con?
-      Node *n1 = phase->transform( new (phase->C, 3) LShiftINode( in(1), phase->intcon(log2_intptr(bit1)) ) );
-      Node *n2 = phase->transform( new (phase->C, 3) LShiftINode( in(1), phase->intcon(log2_intptr(bit2)) ) );
-      res = new (phase->C, 3) AddINode( n2, n1 );
+      Node *n1 = phase->transform( new (phase->C) LShiftINode( in(1), phase->intcon(log2_intptr(bit1)) ) );
+      Node *n2 = phase->transform( new (phase->C) LShiftINode( in(1), phase->intcon(log2_intptr(bit2)) ) );
+      res = new (phase->C) AddINode( n2, n1 );
 
     } else if (is_power_of_2(con+1)) {
       // Sleezy: power-of-2 -1.  Next time be generic.
       jint temp = (jint) (con + 1);
-      Node *n1 = phase->transform( new (phase->C, 3) LShiftINode( in(1), phase->intcon(log2_intptr(temp)) ) );
-      res = new (phase->C, 3) SubINode( n1, in(1) );
+      Node *n1 = phase->transform( new (phase->C) LShiftINode( in(1), phase->intcon(log2_intptr(temp)) ) );
+      res = new (phase->C) SubINode( n1, in(1) );
     } else {
       return MulNode::Ideal(phase, can_reshape);
     }
@@ -221,7 +221,7 @@ Node *MulINode::Ideal(PhaseGVN *phase, bool can_reshape) {
 
   if( sign_flip ) {             // Need to negate result?
     res = phase->transform(res);// Transform, before making the zero con
-    res = new (phase->C, 3) SubINode(phase->intcon(0),res);
+    res = new (phase->C) SubINode(phase->intcon(0),res);
   }
 
   return res;                   // Return final result
@@ -294,22 +294,22 @@ Node *MulLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node *res = NULL;
   jlong bit1 = con & -con;      // Extract low bit
   if( bit1 == con ) {           // Found a power of 2?
-    res = new (phase->C, 3) LShiftLNode( in(1), phase->intcon(log2_long(bit1)) );
+    res = new (phase->C) LShiftLNode( in(1), phase->intcon(log2_long(bit1)) );
   } else {
 
     // Check for constant with 2 bits set
     jlong bit2 = con-bit1;
     bit2 = bit2 & -bit2;          // Extract 2nd bit
     if( bit2 + bit1 == con ) {    // Found all bits in con?
-      Node *n1 = phase->transform( new (phase->C, 3) LShiftLNode( in(1), phase->intcon(log2_long(bit1)) ) );
-      Node *n2 = phase->transform( new (phase->C, 3) LShiftLNode( in(1), phase->intcon(log2_long(bit2)) ) );
-      res = new (phase->C, 3) AddLNode( n2, n1 );
+      Node *n1 = phase->transform( new (phase->C) LShiftLNode( in(1), phase->intcon(log2_long(bit1)) ) );
+      Node *n2 = phase->transform( new (phase->C) LShiftLNode( in(1), phase->intcon(log2_long(bit2)) ) );
+      res = new (phase->C) AddLNode( n2, n1 );
 
     } else if (is_power_of_2_long(con+1)) {
       // Sleezy: power-of-2 -1.  Next time be generic.
       jlong temp = (jlong) (con + 1);
-      Node *n1 = phase->transform( new (phase->C, 3) LShiftLNode( in(1), phase->intcon(log2_long(temp)) ) );
-      res = new (phase->C, 3) SubLNode( n1, in(1) );
+      Node *n1 = phase->transform( new (phase->C) LShiftLNode( in(1), phase->intcon(log2_long(temp)) ) );
+      res = new (phase->C) SubLNode( n1, in(1) );
     } else {
       return MulNode::Ideal(phase, can_reshape);
     }
@@ -317,7 +317,7 @@ Node *MulLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
 
   if( sign_flip ) {             // Need to negate result?
     res = phase->transform(res);// Transform, before making the zero con
-    res = new (phase->C, 3) SubLNode(phase->longcon(0),res);
+    res = new (phase->C) SubLNode(phase->longcon(0),res);
   }
 
   return res;                   // Return final result
@@ -476,27 +476,27 @@ Node *AndINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // Masking bits off of a Character?  Hi bits are already zero.
   if( lop == Op_LoadUS &&
       (mask & 0xFFFF0000) )     // Can we make a smaller mask?
-    return new (phase->C, 3) AndINode(load,phase->intcon(mask&0xFFFF));
+    return new (phase->C) AndINode(load,phase->intcon(mask&0xFFFF));
 
   // Masking bits off of a Short?  Loading a Character does some masking
   if (lop == Op_LoadS && (mask & 0xFFFF0000) == 0 ) {
-    Node *ldus = new (phase->C, 3) LoadUSNode(load->in(MemNode::Control),
+    Node *ldus = new (phase->C) LoadUSNode(load->in(MemNode::Control),
                                               load->in(MemNode::Memory),
                                               load->in(MemNode::Address),
                                               load->adr_type());
     ldus = phase->transform(ldus);
-    return new (phase->C, 3) AndINode(ldus, phase->intcon(mask & 0xFFFF));
+    return new (phase->C) AndINode(ldus, phase->intcon(mask & 0xFFFF));
   }
 
   // Masking sign bits off of a Byte?  Do an unsigned byte load plus
   // an and.
   if (lop == Op_LoadB && (mask & 0xFFFFFF00) == 0) {
-    Node* ldub = new (phase->C, 3) LoadUBNode(load->in(MemNode::Control),
+    Node* ldub = new (phase->C) LoadUBNode(load->in(MemNode::Control),
                                               load->in(MemNode::Memory),
                                               load->in(MemNode::Address),
                                               load->adr_type());
     ldub = phase->transform(ldub);
-    return new (phase->C, 3) AndINode(ldub, phase->intcon(mask));
+    return new (phase->C) AndINode(ldub, phase->intcon(mask));
   }
 
   // Masking off sign bits?  Dont make them!
@@ -510,8 +510,8 @@ Node *AndINode::Ideal(PhaseGVN *phase, bool can_reshape) {
       // bits survive.  NO sign-extension bits survive the maskings.
       if( (sign_bits_mask & mask) == 0 ) {
         // Use zero-fill shift instead
-        Node *zshift = phase->transform(new (phase->C, 3) URShiftINode(load->in(1),load->in(2)));
-        return new (phase->C, 3) AndINode( zshift, in(2) );
+        Node *zshift = phase->transform(new (phase->C) URShiftINode(load->in(1),load->in(2)));
+        return new (phase->C) AndINode( zshift, in(2) );
       }
     }
   }
@@ -521,7 +521,7 @@ Node *AndINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // plus 1) and the mask is of the low order bit.  Skip the negate.
   if( lop == Op_SubI && mask == 1 && load->in(1) &&
       phase->type(load->in(1)) == TypeInt::ZERO )
-    return new (phase->C, 3) AndINode( load->in(2), in(2) );
+    return new (phase->C) AndINode( load->in(2), in(2) );
 
   return MulNode::Ideal(phase, can_reshape);
 }
@@ -607,7 +607,7 @@ Node *AndLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // which is wrong.
   if (op == Op_ConvI2L && in1->in(1)->Opcode() == Op_LoadI && mask == CONST64(0x00000000FFFFFFFF)) {
     Node* load = in1->in(1);
-    return new (phase->C, 3) LoadUI2LNode(load->in(MemNode::Control),
+    return new (phase->C) LoadUI2LNode(load->in(MemNode::Control),
                                           load->in(MemNode::Memory),
                                           load->in(MemNode::Address),
                                           load->adr_type());
@@ -619,9 +619,9 @@ Node *AndLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // value.  This check includes UI2L masks (0x00000000FFFFFFFF) which
   // would be optimized away later in Identity.
   if (op == Op_ConvI2L && (mask & CONST64(0xFFFFFFFF80000000)) == 0) {
-    Node* andi = new (phase->C, 3) AndINode(in1->in(1), phase->intcon(mask));
+    Node* andi = new (phase->C) AndINode(in1->in(1), phase->intcon(mask));
     andi = phase->transform(andi);
-    return new (phase->C, 2) ConvI2LNode(andi);
+    return new (phase->C) ConvI2LNode(andi);
   }
 
   // Masking off sign bits?  Dont make them!
@@ -635,8 +635,8 @@ Node *AndLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       // bits survive.  NO sign-extension bits survive the maskings.
       if( (sign_bits_mask & mask) == 0 ) {
         // Use zero-fill shift instead
-        Node *zshift = phase->transform(new (phase->C, 3) URShiftLNode(in1->in(1), in1->in(2)));
-        return new (phase->C, 3) AndLNode(zshift, in(2));
+        Node *zshift = phase->transform(new (phase->C) URShiftLNode(in1->in(1), in1->in(2)));
+        return new (phase->C) AndLNode(zshift, in(2));
       }
     }
   }
@@ -674,9 +674,9 @@ Node *LShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
       // and 'i2b' patterns which typically fold into 'StoreC/StoreB'.
       if( con < 16 ) {
         // Compute X << con0
-        Node *lsh = phase->transform( new (phase->C, 3) LShiftINode( add1->in(1), in(2) ) );
+        Node *lsh = phase->transform( new (phase->C) LShiftINode( add1->in(1), in(2) ) );
         // Compute X<<con0 + (con1<<con0)
-        return new (phase->C, 3) AddINode( lsh, phase->intcon(t12->get_con() << con));
+        return new (phase->C) AddINode( lsh, phase->intcon(t12->get_con() << con));
       }
     }
   }
@@ -685,7 +685,7 @@ Node *LShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if( (add1_op == Op_RShiftI || add1_op == Op_URShiftI ) &&
       add1->in(2) == in(2) )
     // Convert to "(x & -(1<<c0))"
-    return new (phase->C, 3) AndINode(add1->in(1),phase->intcon( -(1<<con)));
+    return new (phase->C) AndINode(add1->in(1),phase->intcon( -(1<<con)));
 
   // Check for "((x>>c0) & Y)<<c0" which just masks off more low bits
   if( add1_op == Op_AndI ) {
@@ -694,8 +694,8 @@ Node *LShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     if( (add2_op == Op_RShiftI || add2_op == Op_URShiftI ) &&
         add2->in(2) == in(2) ) {
       // Convert to "(x & (Y<<c0))"
-      Node *y_sh = phase->transform( new (phase->C, 3) LShiftINode( add1->in(2), in(2) ) );
-      return new (phase->C, 3) AndINode( add2->in(1), y_sh );
+      Node *y_sh = phase->transform( new (phase->C) LShiftINode( add1->in(2), in(2) ) );
+      return new (phase->C) AndINode( add2->in(1), y_sh );
     }
   }
 
@@ -704,7 +704,7 @@ Node *LShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   const jint bits_mask = right_n_bits(BitsPerJavaInteger-con);
   if( add1_op == Op_AndI &&
       phase->type(add1->in(2)) == TypeInt::make( bits_mask ) )
-    return new (phase->C, 3) LShiftINode( add1->in(1), in(2) );
+    return new (phase->C) LShiftINode( add1->in(1), in(2) );
 
   return NULL;
 }
@@ -784,9 +784,9 @@ Node *LShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     const TypeLong *t12 = phase->type(add1->in(2))->isa_long();
     if( t12 && t12->is_con() ){ // Left input is an add of a con?
       // Compute X << con0
-      Node *lsh = phase->transform( new (phase->C, 3) LShiftLNode( add1->in(1), in(2) ) );
+      Node *lsh = phase->transform( new (phase->C) LShiftLNode( add1->in(1), in(2) ) );
       // Compute X<<con0 + (con1<<con0)
-      return new (phase->C, 3) AddLNode( lsh, phase->longcon(t12->get_con() << con));
+      return new (phase->C) AddLNode( lsh, phase->longcon(t12->get_con() << con));
     }
   }
 
@@ -794,7 +794,7 @@ Node *LShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   if( (add1_op == Op_RShiftL || add1_op == Op_URShiftL ) &&
       add1->in(2) == in(2) )
     // Convert to "(x & -(1<<c0))"
-    return new (phase->C, 3) AndLNode(add1->in(1),phase->longcon( -(CONST64(1)<<con)));
+    return new (phase->C) AndLNode(add1->in(1),phase->longcon( -(CONST64(1)<<con)));
 
   // Check for "((x>>c0) & Y)<<c0" which just masks off more low bits
   if( add1_op == Op_AndL ) {
@@ -803,8 +803,8 @@ Node *LShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     if( (add2_op == Op_RShiftL || add2_op == Op_URShiftL ) &&
         add2->in(2) == in(2) ) {
       // Convert to "(x & (Y<<c0))"
-      Node *y_sh = phase->transform( new (phase->C, 3) LShiftLNode( add1->in(2), in(2) ) );
-      return new (phase->C, 3) AndLNode( add2->in(1), y_sh );
+      Node *y_sh = phase->transform( new (phase->C) LShiftLNode( add1->in(2), in(2) ) );
+      return new (phase->C) AndLNode( add2->in(1), y_sh );
     }
   }
 
@@ -813,7 +813,7 @@ Node *LShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   const jlong bits_mask = ((jlong)CONST64(1) << (jlong)(BitsPerJavaLong - con)) - CONST64(1);
   if( add1_op == Op_AndL &&
       phase->type(add1->in(2)) == TypeLong::make( bits_mask ) )
-    return new (phase->C, 3) LShiftLNode( add1->in(1), in(2) );
+    return new (phase->C) LShiftLNode( add1->in(1), in(2) );
 
   return NULL;
 }
@@ -915,8 +915,8 @@ Node *RShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *x = mask->in(1);
     jint maskbits = t3->get_con();
     // Convert to "(x >> shift) & (mask >> shift)"
-    Node *shr_nomask = phase->transform( new (phase->C, 3) RShiftINode(mask->in(1), in(2)) );
-    return new (phase->C, 3) AndINode(shr_nomask, phase->intcon( maskbits >> shift));
+    Node *shr_nomask = phase->transform( new (phase->C) RShiftINode(mask->in(1), in(2)) );
+    return new (phase->C) AndINode(shr_nomask, phase->intcon( maskbits >> shift));
   }
 
   // Check for "(short[i] <<16)>>16" which simply sign-extends
@@ -939,7 +939,7 @@ Node *RShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     }
     else if( ld->Opcode() == Op_LoadUS )
       // Replace zero-extension-load with sign-extension-load
-      return new (phase->C, 3) LoadSNode( ld->in(MemNode::Control),
+      return new (phase->C) LoadSNode( ld->in(MemNode::Control),
                                 ld->in(MemNode::Memory),
                                 ld->in(MemNode::Address),
                                 ld->adr_type());
@@ -1124,7 +1124,7 @@ Node *URShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
       const int con2 = t12->get_con() & 31; // Shift count is always masked
       const int con3 = con+con2;
       if( con3 < 32 )           // Only merge shifts if total is < 32
-        return new (phase->C, 3) URShiftINode( in(1)->in(1), phase->intcon(con3) );
+        return new (phase->C) URShiftINode( in(1)->in(1), phase->intcon(con3) );
     }
   }
 
@@ -1137,9 +1137,9 @@ Node *URShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *lshl = add->in(1);
     if( lshl->Opcode() == Op_LShiftI &&
         phase->type(lshl->in(2)) == t2 ) {
-      Node *y_z = phase->transform( new (phase->C, 3) URShiftINode(add->in(2),in(2)) );
-      Node *sum = phase->transform( new (phase->C, 3) AddINode( lshl->in(1), y_z ) );
-      return new (phase->C, 3) AndINode( sum, phase->intcon(mask) );
+      Node *y_z = phase->transform( new (phase->C) URShiftINode(add->in(2),in(2)) );
+      Node *sum = phase->transform( new (phase->C) AddINode( lshl->in(1), y_z ) );
+      return new (phase->C) AndINode( sum, phase->intcon(mask) );
     }
   }
 
@@ -1152,8 +1152,8 @@ Node *URShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
     if( t3 && t3->is_con() ) { // Right input is a constant
       jint mask2 = t3->get_con();
       mask2 >>= con;  // *signed* shift downward (high-order zeroes do not help)
-      Node *newshr = phase->transform( new (phase->C, 3) URShiftINode(andi->in(1), in(2)) );
-      return new (phase->C, 3) AndINode(newshr, phase->intcon(mask2));
+      Node *newshr = phase->transform( new (phase->C) URShiftINode(andi->in(1), in(2)) );
+      return new (phase->C) AndINode(newshr, phase->intcon(mask2));
       // The negative values are easier to materialize than positive ones.
       // A typical case from address arithmetic is ((x & ~15) >> 4).
       // It's better to change that to ((x >> 4) & ~0) versus
@@ -1165,7 +1165,7 @@ Node *URShiftINode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node *shl = in(1);
   if( in1_op == Op_LShiftI &&
       phase->type(shl->in(2)) == t2 )
-    return new (phase->C, 3) AndINode( shl->in(1), phase->intcon(mask) );
+    return new (phase->C) AndINode( shl->in(1), phase->intcon(mask) );
 
   return NULL;
 }
@@ -1270,9 +1270,9 @@ Node *URShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *lshl = add->in(1);
     if( lshl->Opcode() == Op_LShiftL &&
         phase->type(lshl->in(2)) == t2 ) {
-      Node *y_z = phase->transform( new (phase->C, 3) URShiftLNode(add->in(2),in(2)) );
-      Node *sum = phase->transform( new (phase->C, 3) AddLNode( lshl->in(1), y_z ) );
-      return new (phase->C, 3) AndLNode( sum, phase->longcon(mask) );
+      Node *y_z = phase->transform( new (phase->C) URShiftLNode(add->in(2),in(2)) );
+      Node *sum = phase->transform( new (phase->C) AddLNode( lshl->in(1), y_z ) );
+      return new (phase->C) AndLNode( sum, phase->longcon(mask) );
     }
   }
 
@@ -1285,8 +1285,8 @@ Node *URShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     if( t3 && t3->is_con() ) { // Right input is a constant
       jlong mask2 = t3->get_con();
       mask2 >>= con;  // *signed* shift downward (high-order zeroes do not help)
-      Node *newshr = phase->transform( new (phase->C, 3) URShiftLNode(andi->in(1), in(2)) );
-      return new (phase->C, 3) AndLNode(newshr, phase->longcon(mask2));
+      Node *newshr = phase->transform( new (phase->C) URShiftLNode(andi->in(1), in(2)) );
+      return new (phase->C) AndLNode(newshr, phase->longcon(mask2));
     }
   }
 
@@ -1294,7 +1294,7 @@ Node *URShiftLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   Node *shl = in(1);
   if( shl->Opcode() == Op_LShiftL &&
       phase->type(shl->in(2)) == t2 )
-    return new (phase->C, 3) AndLNode( shl->in(1), phase->longcon(mask) );
+    return new (phase->C) AndLNode( shl->in(1), phase->longcon(mask) );
 
   return NULL;
 }
diff --git a/src/share/vm/opto/node.cpp b/src/share/vm/opto/node.cpp
index 880576016..cc35098e8 100644
--- a/src/share/vm/opto/node.cpp
+++ b/src/share/vm/opto/node.cpp
@@ -296,6 +296,14 @@ inline int Node::Init(int req, Compile* C) {
   assert(Compile::current() == C, "must use operator new(Compile*)");
   int idx = C->next_unique();
 
+  // Allocate memory for the necessary number of edges.
+  if (req > 0) {
+    // Allocate space for _in array to have double alignment.
+    _in = (Node **) ((char *) (C->node_arena()->Amalloc_D(req * sizeof(void*))));
+#ifdef ASSERT
+    _in[req-1] = this; // magic cookie for assertion check
+#endif
+  }
   // If there are default notes floating around, capture them:
   Node_Notes* nn = C->default_node_notes();
   if (nn != NULL)  init_node_notes(C, idx, nn);
@@ -1004,15 +1012,15 @@ const Type *Node::Value( PhaseTransform * ) const {
 //    set_req(2, phase->intcon(7));
 //    return this;
 // Example: reshape "X*4" into "X<<2"
-//    return new (C,3) LShiftINode(in(1), phase->intcon(2));
+//    return new (C) LShiftINode(in(1), phase->intcon(2));
 //
 // You must call 'phase->transform(X)' on any new Nodes X you make, except
 // for the returned root node.  Example: reshape "X*31" with "(X<<5)-X".
-//    Node *shift=phase->transform(new(C,3)LShiftINode(in(1),phase->intcon(5)));
-//    return new (C,3) AddINode(shift, in(1));
+//    Node *shift=phase->transform(new(C)LShiftINode(in(1),phase->intcon(5)));
+//    return new (C) AddINode(shift, in(1));
 //
 // When making a Node for a constant use 'phase->makecon' or 'phase->intcon'.
-// These forms are faster than 'phase->transform(new (C,1) ConNode())' and Do
+// These forms are faster than 'phase->transform(new (C) ConNode())' and Do
 // The Right Thing with def-use info.
 //
 // You cannot bury the 'this' Node inside of a graph reshape.  If the reshaped
diff --git a/src/share/vm/opto/node.hpp b/src/share/vm/opto/node.hpp
index 759c0bcd6..a6033c68b 100644
--- a/src/share/vm/opto/node.hpp
+++ b/src/share/vm/opto/node.hpp
@@ -217,18 +217,6 @@ public:
     return (void*)n;
   }
 
-  // New Operator that takes a Compile pointer, this will eventually
-  // be the "new" New operator.
-  inline void* operator new( size_t x, Compile* C, int y) {
-    Node* n = (Node*)C->node_arena()->Amalloc_D(x + y*sizeof(void*));
-    n->_in = (Node**)(((char*)n) + x);
-#ifdef ASSERT
-    n->_in[y-1] = n; // magic cookie for assertion check
-#endif
-    n->_out = (Node**)C;
-    return (void*)n;
-  }
-
   // Delete is a NOP
   void operator delete( void *ptr ) {}
   // Fancy destructor; eagerly attempt to reclaim Node numberings and storage
diff --git a/src/share/vm/opto/output.cpp b/src/share/vm/opto/output.cpp
index 89b62127e..f16bbc6fb 100644
--- a/src/share/vm/opto/output.cpp
+++ b/src/share/vm/opto/output.cpp
@@ -265,9 +265,9 @@ void Compile::Insert_zap_nodes() {
 Node* Compile::call_zap_node(MachSafePointNode* node_to_check, int block_no) {
   const TypeFunc *tf = OptoRuntime::zap_dead_locals_Type();
   CallStaticJavaNode* ideal_node =
-    new (this, tf->domain()->cnt()) CallStaticJavaNode( tf,
+    new (this) CallStaticJavaNode( tf,
          OptoRuntime::zap_dead_locals_stub(_method->flags().is_native()),
-                            "call zap dead locals stub", 0, TypePtr::BOTTOM);
+                       "call zap dead locals stub", 0, TypePtr::BOTTOM);
   // We need to copy the OopMap from the site we're zapping at.
   // We have to make a copy, because the zap site might not be
   // a call site, and zap_dead is a call site.
@@ -1869,7 +1869,9 @@ void Compile::ScheduleAndBundle() {
   if (!do_scheduling())
     return;
 
-  assert(MaxVectorSize <= 8, "scheduling code works only with pairs");
+  // Scheduling code works only with pairs (8 bytes) maximum.
+  if (max_vector_size() > 8)
+    return;
 
   NOT_PRODUCT( TracePhase t2("isched", &_t_instrSched, TimeCompiler); )
 
@@ -2682,7 +2684,7 @@ void Scheduling::anti_do_def( Block *b, Node *def, OptoReg::Name def_reg, int is
     if ( _pinch_free_list.size() > 0) {
       pinch = _pinch_free_list.pop();
     } else {
-      pinch = new (_cfg->C, 1) Node(1); // Pinch point to-be
+      pinch = new (_cfg->C) Node(1); // Pinch point to-be
     }
     if (pinch->_idx >= _regalloc->node_regs_max_index()) {
       _cfg->C->record_method_not_compilable("too many D-U pinch points");
diff --git a/src/share/vm/opto/parse1.cpp b/src/share/vm/opto/parse1.cpp
index 8eb86d7bf..cdd1cff68 100644
--- a/src/share/vm/opto/parse1.cpp
+++ b/src/share/vm/opto/parse1.cpp
@@ -107,10 +107,10 @@ Node *Parse::fetch_interpreter_state(int index,
   // doubles on Sparc.  Intel can handle them just fine directly.
   Node *l;
   switch( bt ) {                // Signature is flattened
-  case T_INT:     l = new (C, 3) LoadINode( ctl, mem, adr, TypeRawPtr::BOTTOM ); break;
-  case T_FLOAT:   l = new (C, 3) LoadFNode( ctl, mem, adr, TypeRawPtr::BOTTOM ); break;
-  case T_ADDRESS: l = new (C, 3) LoadPNode( ctl, mem, adr, TypeRawPtr::BOTTOM, TypeRawPtr::BOTTOM  ); break;
-  case T_OBJECT:  l = new (C, 3) LoadPNode( ctl, mem, adr, TypeRawPtr::BOTTOM, TypeInstPtr::BOTTOM ); break;
+  case T_INT:     l = new (C) LoadINode( ctl, mem, adr, TypeRawPtr::BOTTOM ); break;
+  case T_FLOAT:   l = new (C) LoadFNode( ctl, mem, adr, TypeRawPtr::BOTTOM ); break;
+  case T_ADDRESS: l = new (C) LoadPNode( ctl, mem, adr, TypeRawPtr::BOTTOM, TypeRawPtr::BOTTOM  ); break;
+  case T_OBJECT:  l = new (C) LoadPNode( ctl, mem, adr, TypeRawPtr::BOTTOM, TypeInstPtr::BOTTOM ); break;
   case T_LONG:
   case T_DOUBLE: {
     // Since arguments are in reverse order, the argument address 'adr'
@@ -118,12 +118,12 @@ Node *Parse::fetch_interpreter_state(int index,
     adr = basic_plus_adr( local_addrs_base, local_addrs, -(index+1)*wordSize );
     if( Matcher::misaligned_doubles_ok ) {
       l = (bt == T_DOUBLE)
-        ? (Node*)new (C, 3) LoadDNode( ctl, mem, adr, TypeRawPtr::BOTTOM )
-        : (Node*)new (C, 3) LoadLNode( ctl, mem, adr, TypeRawPtr::BOTTOM );
+        ? (Node*)new (C) LoadDNode( ctl, mem, adr, TypeRawPtr::BOTTOM )
+        : (Node*)new (C) LoadLNode( ctl, mem, adr, TypeRawPtr::BOTTOM );
     } else {
       l = (bt == T_DOUBLE)
-        ? (Node*)new (C, 3) LoadD_unalignedNode( ctl, mem, adr, TypeRawPtr::BOTTOM )
-        : (Node*)new (C, 3) LoadL_unalignedNode( ctl, mem, adr, TypeRawPtr::BOTTOM );
+        ? (Node*)new (C) LoadD_unalignedNode( ctl, mem, adr, TypeRawPtr::BOTTOM )
+        : (Node*)new (C) LoadL_unalignedNode( ctl, mem, adr, TypeRawPtr::BOTTOM );
     }
     break;
   }
@@ -147,11 +147,11 @@ Node* Parse::check_interpreter_type(Node* l, const Type* type,
   if (type == TypePtr::NULL_PTR ||
       (tp != NULL && !tp->klass()->is_loaded())) {
     // Value must be null, not a real oop.
-    Node* chk = _gvn.transform( new (C, 3) CmpPNode(l, null()) );
-    Node* tst = _gvn.transform( new (C, 2) BoolNode(chk, BoolTest::eq) );
+    Node* chk = _gvn.transform( new (C) CmpPNode(l, null()) );
+    Node* tst = _gvn.transform( new (C) BoolNode(chk, BoolTest::eq) );
     IfNode* iff = create_and_map_if(control(), tst, PROB_MAX, COUNT_UNKNOWN);
-    set_control(_gvn.transform( new (C, 1) IfTrueNode(iff) ));
-    Node* bad_type = _gvn.transform( new (C, 1) IfFalseNode(iff) );
+    set_control(_gvn.transform( new (C) IfTrueNode(iff) ));
+    Node* bad_type = _gvn.transform( new (C) IfFalseNode(iff) );
     bad_type_exit->control()->add_req(bad_type);
     l = null();
   }
@@ -218,7 +218,7 @@ void Parse::load_interpreter_state(Node* osr_buf) {
   Node *monitors_addr = basic_plus_adr(osr_buf, osr_buf, (max_locals+mcnt*2-1)*wordSize);
   for (index = 0; index < mcnt; index++) {
     // Make a BoxLockNode for the monitor.
-    Node *box = _gvn.transform(new (C, 1) BoxLockNode(next_monitor()));
+    Node *box = _gvn.transform(new (C) BoxLockNode(next_monitor()));
 
 
     // Displaced headers and locked objects are interleaved in the
@@ -233,7 +233,7 @@ void Parse::load_interpreter_state(Node* osr_buf) {
 
     // Build a bogus FastLockNode (no code will be generated) and push the
     // monitor into our debug info.
-    const FastLockNode *flock = _gvn.transform(new (C, 3) FastLockNode( 0, lock_object, box ))->as_FastLock();
+    const FastLockNode *flock = _gvn.transform(new (C) FastLockNode( 0, lock_object, box ))->as_FastLock();
     map()->push_monitor(flock);
 
     // If the lock is our method synchronization lock, tuck it away in
@@ -323,7 +323,7 @@ void Parse::load_interpreter_state(Node* osr_buf) {
   // Now that the interpreter state is loaded, make sure it will match
   // at execution time what the compiler is expecting now:
   SafePointNode* bad_type_exit = clone_map();
-  bad_type_exit->set_control(new (C, 1) RegionNode(1));
+  bad_type_exit->set_control(new (C) RegionNode(1));
 
   assert(osr_block->flow()->jsrs()->size() == 0, "should be no jsrs live at osr point");
   for (index = 0; index < max_locals; index++) {
@@ -647,7 +647,7 @@ void Parse::do_all_blocks() {
           add_predicate();
           // Add new region for back branches.
           int edges = block->pred_count() - block->preds_parsed() + 1; // +1 for original region
-          RegionNode *r = new (C, edges+1) RegionNode(edges+1);
+          RegionNode *r = new (C) RegionNode(edges+1);
           _gvn.set_type(r, Type::CONTROL);
           record_for_igvn(r);
           r->init_req(edges, control());
@@ -714,14 +714,14 @@ void Parse::build_exits() {
   _exits.clean_stack(_exits.sp());
   _exits.sync_jvms();
 
-  RegionNode* region = new (C, 1) RegionNode(1);
+  RegionNode* region = new (C) RegionNode(1);
   record_for_igvn(region);
   gvn().set_type_bottom(region);
   _exits.set_control(region);
 
   // Note:  iophi and memphi are not transformed until do_exits.
-  Node* iophi  = new (C, region->req()) PhiNode(region, Type::ABIO);
-  Node* memphi = new (C, region->req()) PhiNode(region, Type::MEMORY, TypePtr::BOTTOM);
+  Node* iophi  = new (C) PhiNode(region, Type::ABIO);
+  Node* memphi = new (C) PhiNode(region, Type::MEMORY, TypePtr::BOTTOM);
   _exits.set_i_o(iophi);
   _exits.set_all_memory(memphi);
 
@@ -736,7 +736,7 @@ void Parse::build_exits() {
       ret_type = TypeOopPtr::BOTTOM;
     }
     int         ret_size = type2size[ret_type->basic_type()];
-    Node*       ret_phi  = new (C, region->req()) PhiNode(region, ret_type);
+    Node*       ret_phi  = new (C) PhiNode(region, ret_type);
     _exits.ensure_stack(ret_size);
     assert((int)(tf()->range()->cnt() - TypeFunc::Parms) == ret_size, "good tf range");
     assert(method()->return_type()->size() == ret_size, "tf agrees w/ method");
@@ -753,7 +753,7 @@ JVMState* Compile::build_start_state(StartNode* start, const TypeFunc* tf) {
   int        arg_size = tf->domain()->cnt();
   int        max_size = MAX2(arg_size, (int)tf->range()->cnt());
   JVMState*  jvms     = new (this) JVMState(max_size - TypeFunc::Parms);
-  SafePointNode* map  = new (this, max_size) SafePointNode(max_size, NULL);
+  SafePointNode* map  = new (this) SafePointNode(max_size, NULL);
   record_for_igvn(map);
   assert(arg_size == TypeFunc::Parms + (is_osr_compilation() ? 1 : method()->arg_size()), "correct arg_size");
   Node_Notes* old_nn = default_node_notes();
@@ -767,7 +767,7 @@ JVMState* Compile::build_start_state(StartNode* start, const TypeFunc* tf) {
   }
   uint i;
   for (i = 0; i < (uint)arg_size; i++) {
-    Node* parm = initial_gvn()->transform(new (this, 1) ParmNode(start, i));
+    Node* parm = initial_gvn()->transform(new (this) ParmNode(start, i));
     map->init_req(i, parm);
     // Record all these guys for later GVN.
     record_for_igvn(parm);
@@ -798,7 +798,7 @@ Node_Notes* Parse::make_node_notes(Node_Notes* caller_nn) {
 //--------------------------return_values--------------------------------------
 void Compile::return_values(JVMState* jvms) {
   GraphKit kit(jvms);
-  Node* ret = new (this, TypeFunc::Parms) ReturnNode(TypeFunc::Parms,
+  Node* ret = new (this) ReturnNode(TypeFunc::Parms,
                              kit.control(),
                              kit.i_o(),
                              kit.reset_memory(),
@@ -826,7 +826,7 @@ void Compile::rethrow_exceptions(JVMState* jvms) {
   // Load my combined exception state into the kit, with all phis transformed:
   SafePointNode* ex_map = kit.combine_and_pop_all_exception_states();
   Node* ex_oop = kit.use_exception_state(ex_map);
-  RethrowNode* exit = new (this, TypeFunc::Parms + 1) RethrowNode(kit.control(),
+  RethrowNode* exit = new (this) RethrowNode(kit.control(),
                                       kit.i_o(), kit.reset_memory(),
                                       kit.frameptr(), kit.returnadr(),
                                       // like a return but with exception input
@@ -1020,7 +1020,7 @@ SafePointNode* Parse::create_entry_map() {
 
   // Create an initial safepoint to hold JVM state during parsing
   JVMState* jvms = new (C) JVMState(method(), _caller->has_method() ? _caller : NULL);
-  set_map(new (C, len) SafePointNode(len, jvms));
+  set_map(new (C) SafePointNode(len, jvms));
   jvms->set_map(map());
   record_for_igvn(map());
   assert(jvms->endoff() == len, "correct jvms sizing");
@@ -1528,7 +1528,7 @@ void Parse::merge_common(Parse::Block* target, int pnum) {
       // later lazily.
       int edges = target->pred_count();
       if (edges < pnum)  edges = pnum;  // might be a new path!
-      RegionNode *r = new (C, edges+1) RegionNode(edges+1);
+      RegionNode *r = new (C) RegionNode(edges+1);
       gvn().set_type(r, Type::CONTROL);
       record_for_igvn(r);
       // zap all inputs to NULL for debugging (done in Node(uint) constructor)
@@ -1923,19 +1923,19 @@ void Parse::call_register_finalizer() {
   Node* access_flags_addr = basic_plus_adr(klass, klass, in_bytes(Klass::access_flags_offset()));
   Node* access_flags = make_load(NULL, access_flags_addr, TypeInt::INT, T_INT);
 
-  Node* mask  = _gvn.transform(new (C, 3) AndINode(access_flags, intcon(JVM_ACC_HAS_FINALIZER)));
-  Node* check = _gvn.transform(new (C, 3) CmpINode(mask, intcon(0)));
-  Node* test  = _gvn.transform(new (C, 2) BoolNode(check, BoolTest::ne));
+  Node* mask  = _gvn.transform(new (C) AndINode(access_flags, intcon(JVM_ACC_HAS_FINALIZER)));
+  Node* check = _gvn.transform(new (C) CmpINode(mask, intcon(0)));
+  Node* test  = _gvn.transform(new (C) BoolNode(check, BoolTest::ne));
 
   IfNode* iff = create_and_map_if(control(), test, PROB_MAX, COUNT_UNKNOWN);
 
-  RegionNode* result_rgn = new (C, 3) RegionNode(3);
+  RegionNode* result_rgn = new (C) RegionNode(3);
   record_for_igvn(result_rgn);
 
-  Node *skip_register = _gvn.transform(new (C, 1) IfFalseNode(iff));
+  Node *skip_register = _gvn.transform(new (C) IfFalseNode(iff));
   result_rgn->init_req(1, skip_register);
 
-  Node *needs_register = _gvn.transform(new (C, 1) IfTrueNode(iff));
+  Node *needs_register = _gvn.transform(new (C) IfTrueNode(iff));
   set_control(needs_register);
   if (stopped()) {
     // There is no slow path.
@@ -2013,7 +2013,7 @@ void Parse::return_current(Node* value) {
         // sharpen the type eagerly; this eases certain assert checking
         if (tp->higher_equal(TypeInstPtr::NOTNULL))
           tr = tr->join(TypeInstPtr::NOTNULL)->is_instptr();
-        value = _gvn.transform(new (C, 2) CheckCastPPNode(0,value,tr));
+        value = _gvn.transform(new (C) CheckCastPPNode(0,value,tr));
       }
     }
     phi->add_req(value);
@@ -2048,7 +2048,7 @@ void Parse::add_safepoint() {
   kill_dead_locals();
 
   // Clone the JVM State
-  SafePointNode *sfpnt = new (C, parms) SafePointNode(parms, NULL);
+  SafePointNode *sfpnt = new (C) SafePointNode(parms, NULL);
 
   // Capture memory state BEFORE a SafePoint.  Since we can block at a
   // SafePoint we need our GC state to be safe; i.e. we need all our current
diff --git a/src/share/vm/opto/parse2.cpp b/src/share/vm/opto/parse2.cpp
index af514850b..0373ebfc9 100644
--- a/src/share/vm/opto/parse2.cpp
+++ b/src/share/vm/opto/parse2.cpp
@@ -127,9 +127,9 @@ Node* Parse::array_addressing(BasicType type, int vals, const Type* *result2) {
       Node* len = load_array_length(ary);
 
       // Test length vs index (standard trick using unsigned compare)
-      Node* chk = _gvn.transform( new (C, 3) CmpUNode(idx, len) );
+      Node* chk = _gvn.transform( new (C) CmpUNode(idx, len) );
       BoolTest::mask btest = BoolTest::lt;
-      tst = _gvn.transform( new (C, 2) BoolNode(chk, btest) );
+      tst = _gvn.transform( new (C) BoolNode(chk, btest) );
     }
     // Branch to failure if out of bounds
     { BuildCutout unless(this, tst, PROB_MAX);
@@ -165,15 +165,15 @@ Node* Parse::array_addressing(BasicType type, int vals, const Type* *result2) {
 
 // returns IfNode
 IfNode* Parse::jump_if_fork_int(Node* a, Node* b, BoolTest::mask mask) {
-  Node   *cmp = _gvn.transform( new (C, 3) CmpINode( a, b)); // two cases: shiftcount > 32 and shiftcount <= 32
-  Node   *tst = _gvn.transform( new (C, 2) BoolNode( cmp, mask));
+  Node   *cmp = _gvn.transform( new (C) CmpINode( a, b)); // two cases: shiftcount > 32 and shiftcount <= 32
+  Node   *tst = _gvn.transform( new (C) BoolNode( cmp, mask));
   IfNode *iff = create_and_map_if( control(), tst, ((mask == BoolTest::eq) ? PROB_STATIC_INFREQUENT : PROB_FAIR), COUNT_UNKNOWN );
   return iff;
 }
 
 // return Region node
 Node* Parse::jump_if_join(Node* iffalse, Node* iftrue) {
-  Node *region  = new (C, 3) RegionNode(3); // 2 results
+  Node *region  = new (C) RegionNode(3); // 2 results
   record_for_igvn(region);
   region->init_req(1, iffalse);
   region->init_req(2, iftrue );
@@ -188,28 +188,28 @@ Node* Parse::jump_if_join(Node* iffalse, Node* iftrue) {
 void Parse::jump_if_true_fork(IfNode *iff, int dest_bci_if_true, int prof_table_index) {
   // True branch, use existing map info
   { PreserveJVMState pjvms(this);
-    Node *iftrue  = _gvn.transform( new (C, 1) IfTrueNode (iff) );
+    Node *iftrue  = _gvn.transform( new (C) IfTrueNode (iff) );
     set_control( iftrue );
     profile_switch_case(prof_table_index);
     merge_new_path(dest_bci_if_true);
   }
 
   // False branch
-  Node *iffalse = _gvn.transform( new (C, 1) IfFalseNode(iff) );
+  Node *iffalse = _gvn.transform( new (C) IfFalseNode(iff) );
   set_control( iffalse );
 }
 
 void Parse::jump_if_false_fork(IfNode *iff, int dest_bci_if_true, int prof_table_index) {
   // True branch, use existing map info
   { PreserveJVMState pjvms(this);
-    Node *iffalse  = _gvn.transform( new (C, 1) IfFalseNode (iff) );
+    Node *iffalse  = _gvn.transform( new (C) IfFalseNode (iff) );
     set_control( iffalse );
     profile_switch_case(prof_table_index);
     merge_new_path(dest_bci_if_true);
   }
 
   // False branch
-  Node *iftrue = _gvn.transform( new (C, 1) IfTrueNode(iff) );
+  Node *iftrue = _gvn.transform( new (C) IfTrueNode(iff) );
   set_control( iftrue );
 }
 
@@ -437,14 +437,14 @@ bool Parse::create_jump_tables(Node* key_val, SwitchRange* lo, SwitchRange* hi)
 
   // Normalize table lookups to zero
   int lowval = lo->lo();
-  key_val = _gvn.transform( new (C, 3) SubINode(key_val, _gvn.intcon(lowval)) );
+  key_val = _gvn.transform( new (C) SubINode(key_val, _gvn.intcon(lowval)) );
 
   // Generate a guard to protect against input keyvals that aren't
   // in the switch domain.
   if (needs_guard) {
     Node*   size = _gvn.intcon(num_cases);
-    Node*   cmp = _gvn.transform( new (C, 3) CmpUNode(key_val, size) );
-    Node*   tst = _gvn.transform( new (C, 2) BoolNode(cmp, BoolTest::ge) );
+    Node*   cmp = _gvn.transform( new (C) CmpUNode(key_val, size) );
+    Node*   tst = _gvn.transform( new (C) BoolNode(cmp, BoolTest::ge) );
     IfNode* iff = create_and_map_if( control(), tst, PROB_FAIR, COUNT_UNKNOWN);
     jump_if_true_fork(iff, default_dest, NullTableIndex);
   }
@@ -457,21 +457,21 @@ bool Parse::create_jump_tables(Node* key_val, SwitchRange* lo, SwitchRange* hi)
   // Clean the 32-bit int into a real 64-bit offset.
   // Otherwise, the jint value 0 might turn into an offset of 0x0800000000.
   const TypeLong* lkeytype = TypeLong::make(CONST64(0), num_cases-1, Type::WidenMin);
-  key_val       = _gvn.transform( new (C, 2) ConvI2LNode(key_val, lkeytype) );
+  key_val       = _gvn.transform( new (C) ConvI2LNode(key_val, lkeytype) );
 #endif
   // Shift the value by wordsize so we have an index into the table, rather
   // than a switch value
   Node *shiftWord = _gvn.MakeConX(wordSize);
-  key_val = _gvn.transform( new (C, 3) MulXNode( key_val, shiftWord));
+  key_val = _gvn.transform( new (C) MulXNode( key_val, shiftWord));
 
   // Create the JumpNode
-  Node* jtn = _gvn.transform( new (C, 2) JumpNode(control(), key_val, num_cases) );
+  Node* jtn = _gvn.transform( new (C) JumpNode(control(), key_val, num_cases) );
 
   // These are the switch destinations hanging off the jumpnode
   int i = 0;
   for (SwitchRange* r = lo; r <= hi; r++) {
     for (int j = r->lo(); j <= r->hi(); j++, i++) {
-      Node* input = _gvn.transform(new (C, 1) JumpProjNode(jtn, i, r->dest(), j - lowval));
+      Node* input = _gvn.transform(new (C) JumpProjNode(jtn, i, r->dest(), j - lowval));
       {
         PreserveJVMState pjvms(this);
         set_control(input);
@@ -572,8 +572,8 @@ void Parse::jump_switch_ranges(Node* key_val, SwitchRange *lo, SwitchRange *hi,
         // two comparisons of same values--should enable 1 test for 2 branches
         // Use BoolTest::le instead of BoolTest::gt
         IfNode *iff_le  = jump_if_fork_int(key_val, test_val, BoolTest::le);
-        Node   *iftrue  = _gvn.transform( new (C, 1) IfTrueNode(iff_le) );
-        Node   *iffalse = _gvn.transform( new (C, 1) IfFalseNode(iff_le) );
+        Node   *iftrue  = _gvn.transform( new (C) IfTrueNode(iff_le) );
+        Node   *iffalse = _gvn.transform( new (C) IfFalseNode(iff_le) );
         { PreserveJVMState pjvms(this);
           set_control(iffalse);
           jump_switch_ranges(key_val, mid+1, hi, switch_depth+1);
@@ -589,8 +589,8 @@ void Parse::jump_switch_ranges(Node* key_val, SwitchRange *lo, SwitchRange *hi,
       if (mid == hi) {
         jump_if_true_fork(iff_ge, mid->dest(), mid->table_index());
       } else {
-        Node *iftrue  = _gvn.transform( new (C, 1) IfTrueNode(iff_ge) );
-        Node *iffalse = _gvn.transform( new (C, 1) IfFalseNode(iff_ge) );
+        Node *iftrue  = _gvn.transform( new (C) IfTrueNode(iff_ge) );
+        Node *iffalse = _gvn.transform( new (C) IfFalseNode(iff_ge) );
         { PreserveJVMState pjvms(this);
           set_control(iftrue);
           jump_switch_ranges(key_val, mid, hi, switch_depth+1);
@@ -645,7 +645,7 @@ void Parse::modf() {
                               CAST_FROM_FN_PTR(address, SharedRuntime::frem),
                               "frem", NULL, //no memory effects
                               f1, f2);
-  Node* res = _gvn.transform(new (C, 1) ProjNode(c, TypeFunc::Parms + 0));
+  Node* res = _gvn.transform(new (C) ProjNode(c, TypeFunc::Parms + 0));
 
   push(res);
 }
@@ -657,10 +657,10 @@ void Parse::modd() {
                               CAST_FROM_FN_PTR(address, SharedRuntime::drem),
                               "drem", NULL, //no memory effects
                               d1, top(), d2, top());
-  Node* res_d   = _gvn.transform(new (C, 1) ProjNode(c, TypeFunc::Parms + 0));
+  Node* res_d   = _gvn.transform(new (C) ProjNode(c, TypeFunc::Parms + 0));
 
 #ifdef ASSERT
-  Node* res_top = _gvn.transform(new (C, 1) ProjNode(c, TypeFunc::Parms + 1));
+  Node* res_top = _gvn.transform(new (C) ProjNode(c, TypeFunc::Parms + 1));
   assert(res_top == top(), "second value must be top");
 #endif
 
@@ -674,7 +674,7 @@ void Parse::l2f() {
                               CAST_FROM_FN_PTR(address, SharedRuntime::l2f),
                               "l2f", NULL, //no memory effects
                               f1, f2);
-  Node* res = _gvn.transform(new (C, 1) ProjNode(c, TypeFunc::Parms + 0));
+  Node* res = _gvn.transform(new (C) ProjNode(c, TypeFunc::Parms + 0));
 
   push(res);
 }
@@ -701,17 +701,17 @@ void Parse::do_irem() {
         // Sigh, must handle negative dividends
         Node *zero = _gvn.intcon(0);
         IfNode *ifff = jump_if_fork_int(a, zero, BoolTest::lt);
-        Node *iff = _gvn.transform( new (C, 1) IfFalseNode(ifff) );
-        Node *ift = _gvn.transform( new (C, 1) IfTrueNode (ifff) );
+        Node *iff = _gvn.transform( new (C) IfFalseNode(ifff) );
+        Node *ift = _gvn.transform( new (C) IfTrueNode (ifff) );
         Node *reg = jump_if_join(ift, iff);
         Node *phi = PhiNode::make(reg, NULL, TypeInt::INT);
         // Negative path; negate/and/negate
-        Node *neg = _gvn.transform( new (C, 3) SubINode(zero, a) );
-        Node *andn= _gvn.transform( new (C, 3) AndINode(neg, mask) );
-        Node *negn= _gvn.transform( new (C, 3) SubINode(zero, andn) );
+        Node *neg = _gvn.transform( new (C) SubINode(zero, a) );
+        Node *andn= _gvn.transform( new (C) AndINode(neg, mask) );
+        Node *negn= _gvn.transform( new (C) SubINode(zero, andn) );
         phi->init_req(1, negn);
         // Fast positive case
-        Node *andx = _gvn.transform( new (C, 3) AndINode(a, mask) );
+        Node *andx = _gvn.transform( new (C) AndINode(a, mask) );
         phi->init_req(2, andx);
         // Push the merge
         push( _gvn.transform(phi) );
@@ -720,7 +720,7 @@ void Parse::do_irem() {
     }
   }
   // Default case
-  push( _gvn.transform( new (C, 3) ModINode(control(),a,b) ) );
+  push( _gvn.transform( new (C) ModINode(control(),a,b) ) );
 }
 
 // Handle jsr and jsr_w bytecode
@@ -997,7 +997,7 @@ void Parse::do_ifnull(BoolTest::mask btest, Node *c) {
   explicit_null_checks_inserted++;
 
   // Generate real control flow
-  Node   *tst = _gvn.transform( new (C, 2) BoolNode( c, btest ) );
+  Node   *tst = _gvn.transform( new (C) BoolNode( c, btest ) );
 
   // Sanity check the probability value
   assert(prob > 0.0f,"Bad probability in Parser");
@@ -1006,7 +1006,7 @@ void Parse::do_ifnull(BoolTest::mask btest, Node *c) {
   assert(iff->_prob > 0.0f,"Optimizer made bad probability in parser");
   // True branch
   { PreserveJVMState pjvms(this);
-    Node* iftrue  = _gvn.transform( new (C, 1) IfTrueNode (iff) );
+    Node* iftrue  = _gvn.transform( new (C) IfTrueNode (iff) );
     set_control(iftrue);
 
     if (stopped()) {            // Path is dead?
@@ -1026,7 +1026,7 @@ void Parse::do_ifnull(BoolTest::mask btest, Node *c) {
   }
 
   // False branch
-  Node* iffalse = _gvn.transform( new (C, 1) IfFalseNode(iff) );
+  Node* iffalse = _gvn.transform( new (C) IfFalseNode(iff) );
   set_control(iffalse);
 
   if (stopped()) {              // Path is dead?
@@ -1089,7 +1089,7 @@ void Parse::do_if(BoolTest::mask btest, Node* c) {
   }
   assert(btest != BoolTest::eq, "!= is the only canonical exact test");
 
-  Node* tst0 = new (C, 2) BoolNode(c, btest);
+  Node* tst0 = new (C) BoolNode(c, btest);
   Node* tst = _gvn.transform(tst0);
   BoolTest::mask taken_btest   = BoolTest::illegal;
   BoolTest::mask untaken_btest = BoolTest::illegal;
@@ -1120,8 +1120,8 @@ void Parse::do_if(BoolTest::mask btest, Node* c) {
   float true_prob = (taken_if_true ? prob : untaken_prob);
   IfNode* iff = create_and_map_if(control(), tst, true_prob, cnt);
   assert(iff->_prob > 0.0f,"Optimizer made bad probability in parser");
-  Node* taken_branch   = new (C, 1) IfTrueNode(iff);
-  Node* untaken_branch = new (C, 1) IfFalseNode(iff);
+  Node* taken_branch   = new (C) IfTrueNode(iff);
+  Node* untaken_branch = new (C) IfFalseNode(iff);
   if (!taken_if_true) {  // Finish conversion to canonical form
     Node* tmp      = taken_branch;
     taken_branch   = untaken_branch;
@@ -1285,7 +1285,7 @@ void Parse::sharpen_type_after_if(BoolTest::mask btest,
           JVMState* jvms = this->jvms();
           if (obj_in_map >= 0 &&
               (jvms->is_loc(obj_in_map) || jvms->is_stk(obj_in_map))) {
-            TypeNode* ccast = new (C, 2) CheckCastPPNode(control(), obj, tboth);
+            TypeNode* ccast = new (C) CheckCastPPNode(control(), obj, tboth);
             const Type* tcc = ccast->as_Type()->type();
             assert(tcc != obj_type && tcc->higher_equal(obj_type), "must improve");
             // Delay transform() call to allow recovery of pre-cast value
@@ -1320,10 +1320,10 @@ void Parse::sharpen_type_after_if(BoolTest::mask btest,
       const Type* tboth = tcon->join(tval);
       if (tboth == tval)  break;        // Nothing to gain.
       if (tcon->isa_int()) {
-        ccast = new (C, 2) CastIINode(val, tboth);
+        ccast = new (C) CastIINode(val, tboth);
       } else if (tcon == TypePtr::NULL_PTR) {
         // Cast to null, but keep the pointer identity temporarily live.
-        ccast = new (C, 2) CastPPNode(val, tboth);
+        ccast = new (C) CastPPNode(val, tboth);
       } else {
         const TypeF* tf = tcon->isa_float_constant();
         const TypeD* td = tcon->isa_double_constant();
@@ -1738,59 +1738,59 @@ void Parse::do_one_bytecode() {
     if (stopped())  return;
     b = pop();
     a = pop();
-    push( _gvn.transform( new (C, 3) DivINode(control(),a,b) ) );
+    push( _gvn.transform( new (C) DivINode(control(),a,b) ) );
     break;
   case Bytecodes::_imul:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) MulINode(a,b) ) );
+    push( _gvn.transform( new (C) MulINode(a,b) ) );
     break;
   case Bytecodes::_iadd:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) AddINode(a,b) ) );
+    push( _gvn.transform( new (C) AddINode(a,b) ) );
     break;
   case Bytecodes::_ineg:
     a = pop();
-    push( _gvn.transform( new (C, 3) SubINode(_gvn.intcon(0),a)) );
+    push( _gvn.transform( new (C) SubINode(_gvn.intcon(0),a)) );
     break;
   case Bytecodes::_isub:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) SubINode(a,b) ) );
+    push( _gvn.transform( new (C) SubINode(a,b) ) );
     break;
   case Bytecodes::_iand:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) AndINode(a,b) ) );
+    push( _gvn.transform( new (C) AndINode(a,b) ) );
     break;
   case Bytecodes::_ior:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) OrINode(a,b) ) );
+    push( _gvn.transform( new (C) OrINode(a,b) ) );
     break;
   case Bytecodes::_ixor:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) XorINode(a,b) ) );
+    push( _gvn.transform( new (C) XorINode(a,b) ) );
     break;
   case Bytecodes::_ishl:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) LShiftINode(a,b) ) );
+    push( _gvn.transform( new (C) LShiftINode(a,b) ) );
     break;
   case Bytecodes::_ishr:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) RShiftINode(a,b) ) );
+    push( _gvn.transform( new (C) RShiftINode(a,b) ) );
     break;
   case Bytecodes::_iushr:
     b = pop(); a = pop();
-    push( _gvn.transform( new (C, 3) URShiftINode(a,b) ) );
+    push( _gvn.transform( new (C) URShiftINode(a,b) ) );
     break;
 
   case Bytecodes::_fneg:
     a = pop();
-    b = _gvn.transform(new (C, 2) NegFNode (a));
+    b = _gvn.transform(new (C) NegFNode (a));
     push(b);
     break;
 
   case Bytecodes::_fsub:
     b = pop();
     a = pop();
-    c = _gvn.transform( new (C, 3) SubFNode(a,b) );
+    c = _gvn.transform( new (C) SubFNode(a,b) );
     d = precision_rounding(c);
     push( d );
     break;
@@ -1798,7 +1798,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_fadd:
     b = pop();
     a = pop();
-    c = _gvn.transform( new (C, 3) AddFNode(a,b) );
+    c = _gvn.transform( new (C) AddFNode(a,b) );
     d = precision_rounding(c);
     push( d );
     break;
@@ -1806,7 +1806,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_fmul:
     b = pop();
     a = pop();
-    c = _gvn.transform( new (C, 3) MulFNode(a,b) );
+    c = _gvn.transform( new (C) MulFNode(a,b) );
     d = precision_rounding(c);
     push( d );
     break;
@@ -1814,7 +1814,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_fdiv:
     b = pop();
     a = pop();
-    c = _gvn.transform( new (C, 3) DivFNode(0,a,b) );
+    c = _gvn.transform( new (C) DivFNode(0,a,b) );
     d = precision_rounding(c);
     push( d );
     break;
@@ -1824,7 +1824,7 @@ void Parse::do_one_bytecode() {
       // Generate a ModF node.
       b = pop();
       a = pop();
-      c = _gvn.transform( new (C, 3) ModFNode(0,a,b) );
+      c = _gvn.transform( new (C) ModFNode(0,a,b) );
       d = precision_rounding(c);
       push( d );
     }
@@ -1837,7 +1837,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_fcmpl:
     b = pop();
     a = pop();
-    c = _gvn.transform( new (C, 3) CmpF3Node( a, b));
+    c = _gvn.transform( new (C) CmpF3Node( a, b));
     push(c);
     break;
   case Bytecodes::_fcmpg:
@@ -1849,40 +1849,40 @@ void Parse::do_one_bytecode() {
     // as well by using CmpF3 which implements unordered-lesser instead of
     // unordered-greater semantics.  Finally, commute the result bits.  Result
     // is same as using a CmpF3Greater except we did it with CmpF3 alone.
-    c = _gvn.transform( new (C, 3) CmpF3Node( b, a));
-    c = _gvn.transform( new (C, 3) SubINode(_gvn.intcon(0),c) );
+    c = _gvn.transform( new (C) CmpF3Node( b, a));
+    c = _gvn.transform( new (C) SubINode(_gvn.intcon(0),c) );
     push(c);
     break;
 
   case Bytecodes::_f2i:
     a = pop();
-    push(_gvn.transform(new (C, 2) ConvF2INode(a)));
+    push(_gvn.transform(new (C) ConvF2INode(a)));
     break;
 
   case Bytecodes::_d2i:
     a = pop_pair();
-    b = _gvn.transform(new (C, 2) ConvD2INode(a));
+    b = _gvn.transform(new (C) ConvD2INode(a));
     push( b );
     break;
 
   case Bytecodes::_f2d:
     a = pop();
-    b = _gvn.transform( new (C, 2) ConvF2DNode(a));
+    b = _gvn.transform( new (C) ConvF2DNode(a));
     push_pair( b );
     break;
 
   case Bytecodes::_d2f:
     a = pop_pair();
-    b = _gvn.transform( new (C, 2) ConvD2FNode(a));
+    b = _gvn.transform( new (C) ConvD2FNode(a));
     // This breaks _227_mtrt (speed & correctness) and _222_mpegaudio (speed)
-    //b = _gvn.transform(new (C, 2) RoundFloatNode(0, b) );
+    //b = _gvn.transform(new (C) RoundFloatNode(0, b) );
     push( b );
     break;
 
   case Bytecodes::_l2f:
     if (Matcher::convL2FSupported()) {
       a = pop_pair();
-      b = _gvn.transform( new (C, 2) ConvL2FNode(a));
+      b = _gvn.transform( new (C) ConvL2FNode(a));
       // For i486.ad, FILD doesn't restrict precision to 24 or 53 bits.
       // Rather than storing the result into an FP register then pushing
       // out to memory to round, the machine instruction that implements
@@ -1897,7 +1897,7 @@ void Parse::do_one_bytecode() {
 
   case Bytecodes::_l2d:
     a = pop_pair();
-    b = _gvn.transform( new (C, 2) ConvL2DNode(a));
+    b = _gvn.transform( new (C) ConvL2DNode(a));
     // For i486.ad, rounding is always necessary (see _l2f above).
     // c = dprecision_rounding(b);
     c = _gvn.transform(b);
@@ -1906,20 +1906,20 @@ void Parse::do_one_bytecode() {
 
   case Bytecodes::_f2l:
     a = pop();
-    b = _gvn.transform( new (C, 2) ConvF2LNode(a));
+    b = _gvn.transform( new (C) ConvF2LNode(a));
     push_pair(b);
     break;
 
   case Bytecodes::_d2l:
     a = pop_pair();
-    b = _gvn.transform( new (C, 2) ConvD2LNode(a));
+    b = _gvn.transform( new (C) ConvD2LNode(a));
     push_pair(b);
     break;
 
   case Bytecodes::_dsub:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) SubDNode(a,b) );
+    c = _gvn.transform( new (C) SubDNode(a,b) );
     d = dprecision_rounding(c);
     push_pair( d );
     break;
@@ -1927,7 +1927,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_dadd:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) AddDNode(a,b) );
+    c = _gvn.transform( new (C) AddDNode(a,b) );
     d = dprecision_rounding(c);
     push_pair( d );
     break;
@@ -1935,7 +1935,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_dmul:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) MulDNode(a,b) );
+    c = _gvn.transform( new (C) MulDNode(a,b) );
     d = dprecision_rounding(c);
     push_pair( d );
     break;
@@ -1943,14 +1943,14 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_ddiv:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) DivDNode(0,a,b) );
+    c = _gvn.transform( new (C) DivDNode(0,a,b) );
     d = dprecision_rounding(c);
     push_pair( d );
     break;
 
   case Bytecodes::_dneg:
     a = pop_pair();
-    b = _gvn.transform(new (C, 2) NegDNode (a));
+    b = _gvn.transform(new (C) NegDNode (a));
     push_pair(b);
     break;
 
@@ -1961,7 +1961,7 @@ void Parse::do_one_bytecode() {
       a = pop_pair();
       // a % b
 
-      c = _gvn.transform( new (C, 3) ModDNode(0,a,b) );
+      c = _gvn.transform( new (C) ModDNode(0,a,b) );
       d = dprecision_rounding(c);
       push_pair( d );
     }
@@ -1974,7 +1974,7 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_dcmpl:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) CmpD3Node( a, b));
+    c = _gvn.transform( new (C) CmpD3Node( a, b));
     push(c);
     break;
 
@@ -1987,8 +1987,8 @@ void Parse::do_one_bytecode() {
     // unordered-lesser instead of unordered-greater semantics.
     // Finally, negate the result bits.  Result is same as using a
     // CmpD3Greater except we did it with CmpD3 alone.
-    c = _gvn.transform( new (C, 3) CmpD3Node( b, a));
-    c = _gvn.transform( new (C, 3) SubINode(_gvn.intcon(0),c) );
+    c = _gvn.transform( new (C) CmpD3Node( b, a));
+    c = _gvn.transform( new (C) SubINode(_gvn.intcon(0),c) );
     push(c);
     break;
 
@@ -1997,44 +1997,44 @@ void Parse::do_one_bytecode() {
   case Bytecodes::_land:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) AndLNode(a,b) );
+    c = _gvn.transform( new (C) AndLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lor:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) OrLNode(a,b) );
+    c = _gvn.transform( new (C) OrLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lxor:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) XorLNode(a,b) );
+    c = _gvn.transform( new (C) XorLNode(a,b) );
     push_pair(c);
     break;
 
   case Bytecodes::_lshl:
     b = pop();                  // the shift count
     a = pop_pair();             // value to be shifted
-    c = _gvn.transform( new (C, 3) LShiftLNode(a,b) );
+    c = _gvn.transform( new (C) LShiftLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lshr:
     b = pop();                  // the shift count
     a = pop_pair();             // value to be shifted
-    c = _gvn.transform( new (C, 3) RShiftLNode(a,b) );
+    c = _gvn.transform( new (C) RShiftLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lushr:
     b = pop();                  // the shift count
     a = pop_pair();             // value to be shifted
-    c = _gvn.transform( new (C, 3) URShiftLNode(a,b) );
+    c = _gvn.transform( new (C) URShiftLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lmul:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) MulLNode(a,b) );
+    c = _gvn.transform( new (C) MulLNode(a,b) );
     push_pair(c);
     break;
 
@@ -2046,7 +2046,7 @@ void Parse::do_one_bytecode() {
     if (stopped())  return;
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) ModLNode(control(),a,b) );
+    c = _gvn.transform( new (C) ModLNode(control(),a,b) );
     push_pair(c);
     break;
 
@@ -2058,20 +2058,20 @@ void Parse::do_one_bytecode() {
     if (stopped())  return;
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) DivLNode(control(),a,b) );
+    c = _gvn.transform( new (C) DivLNode(control(),a,b) );
     push_pair(c);
     break;
 
   case Bytecodes::_ladd:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) AddLNode(a,b) );
+    c = _gvn.transform( new (C) AddLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lsub:
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) SubLNode(a,b) );
+    c = _gvn.transform( new (C) SubLNode(a,b) );
     push_pair(c);
     break;
   case Bytecodes::_lcmp:
@@ -2102,58 +2102,58 @@ void Parse::do_one_bytecode() {
     }
     b = pop_pair();
     a = pop_pair();
-    c = _gvn.transform( new (C, 3) CmpL3Node( a, b ));
+    c = _gvn.transform( new (C) CmpL3Node( a, b ));
     push(c);
     break;
 
   case Bytecodes::_lneg:
     a = pop_pair();
-    b = _gvn.transform( new (C, 3) SubLNode(longcon(0),a));
+    b = _gvn.transform( new (C) SubLNode(longcon(0),a));
     push_pair(b);
     break;
   case Bytecodes::_l2i:
     a = pop_pair();
-    push( _gvn.transform( new (C, 2) ConvL2INode(a)));
+    push( _gvn.transform( new (C) ConvL2INode(a)));
     break;
   case Bytecodes::_i2l:
     a = pop();
-    b = _gvn.transform( new (C, 2) ConvI2LNode(a));
+    b = _gvn.transform( new (C) ConvI2LNode(a));
     push_pair(b);
     break;
   case Bytecodes::_i2b:
     // Sign extend
     a = pop();
-    a = _gvn.transform( new (C, 3) LShiftINode(a,_gvn.intcon(24)) );
-    a = _gvn.transform( new (C, 3) RShiftINode(a,_gvn.intcon(24)) );
+    a = _gvn.transform( new (C) LShiftINode(a,_gvn.intcon(24)) );
+    a = _gvn.transform( new (C) RShiftINode(a,_gvn.intcon(24)) );
     push( a );
     break;
   case Bytecodes::_i2s:
     a = pop();
-    a = _gvn.transform( new (C, 3) LShiftINode(a,_gvn.intcon(16)) );
-    a = _gvn.transform( new (C, 3) RShiftINode(a,_gvn.intcon(16)) );
+    a = _gvn.transform( new (C) LShiftINode(a,_gvn.intcon(16)) );
+    a = _gvn.transform( new (C) RShiftINode(a,_gvn.intcon(16)) );
     push( a );
     break;
   case Bytecodes::_i2c:
     a = pop();
-    push( _gvn.transform( new (C, 3) AndINode(a,_gvn.intcon(0xFFFF)) ) );
+    push( _gvn.transform( new (C) AndINode(a,_gvn.intcon(0xFFFF)) ) );
     break;
 
   case Bytecodes::_i2f:
     a = pop();
-    b = _gvn.transform( new (C, 2) ConvI2FNode(a) ) ;
+    b = _gvn.transform( new (C) ConvI2FNode(a) ) ;
     c = precision_rounding(b);
     push (b);
     break;
 
   case Bytecodes::_i2d:
     a = pop();
-    b = _gvn.transform( new (C, 2) ConvI2DNode(a));
+    b = _gvn.transform( new (C) ConvI2DNode(a));
     push_pair(b);
     break;
 
   case Bytecodes::_iinc:        // Increment local
     i = iter().get_index();     // Get local index
-    set_local( i, _gvn.transform( new (C, 3) AddINode( _gvn.intcon(iter().get_iinc_con()), local(i) ) ) );
+    set_local( i, _gvn.transform( new (C) AddINode( _gvn.intcon(iter().get_iinc_con()), local(i) ) ) );
     break;
 
   // Exit points of synchronized methods must have an unlock node
@@ -2225,7 +2225,7 @@ void Parse::do_one_bytecode() {
     maybe_add_safepoint(iter().get_dest());
     a = null();
     b = pop();
-    c = _gvn.transform( new (C, 3) CmpPNode(b, a) );
+    c = _gvn.transform( new (C) CmpPNode(b, a) );
     do_ifnull(btest, c);
     break;
 
@@ -2236,7 +2236,7 @@ void Parse::do_one_bytecode() {
     maybe_add_safepoint(iter().get_dest());
     a = pop();
     b = pop();
-    c = _gvn.transform( new (C, 3) CmpPNode(b, a) );
+    c = _gvn.transform( new (C) CmpPNode(b, a) );
     do_if(btest, c);
     break;
 
@@ -2251,7 +2251,7 @@ void Parse::do_one_bytecode() {
     maybe_add_safepoint(iter().get_dest());
     a = _gvn.intcon(0);
     b = pop();
-    c = _gvn.transform( new (C, 3) CmpINode(b, a) );
+    c = _gvn.transform( new (C) CmpINode(b, a) );
     do_if(btest, c);
     break;
 
@@ -2266,7 +2266,7 @@ void Parse::do_one_bytecode() {
     maybe_add_safepoint(iter().get_dest());
     a = pop();
     b = pop();
-    c = _gvn.transform( new (C, 3) CmpINode( b, a ) );
+    c = _gvn.transform( new (C) CmpINode( b, a ) );
     do_if(btest, c);
     break;
 
diff --git a/src/share/vm/opto/parse3.cpp b/src/share/vm/opto/parse3.cpp
index 6d6fd07c6..1a2f0520a 100644
--- a/src/share/vm/opto/parse3.cpp
+++ b/src/share/vm/opto/parse3.cpp
@@ -510,7 +510,7 @@ void Parse::do_multianewarray() {
                           dims);
   }
 
-  Node* res = _gvn.transform(new (C, 1) ProjNode(c, TypeFunc::Parms));
+  Node* res = _gvn.transform(new (C) ProjNode(c, TypeFunc::Parms));
 
   const Type* type = TypeOopPtr::make_from_klass_raw(array_klass);
 
@@ -524,7 +524,7 @@ void Parse::do_multianewarray() {
 
     // We cannot sharpen the nested sub-arrays, since the top level is mutable.
 
-  Node* cast = _gvn.transform( new (C, 2) CheckCastPPNode(control(), res, type) );
+  Node* cast = _gvn.transform( new (C) CheckCastPPNode(control(), res, type) );
   push(cast);
 
   // Possible improvements:
diff --git a/src/share/vm/opto/parseHelper.cpp b/src/share/vm/opto/parseHelper.cpp
index e562d2772..5cb6bd241 100644
--- a/src/share/vm/opto/parseHelper.cpp
+++ b/src/share/vm/opto/parseHelper.cpp
@@ -43,7 +43,7 @@ void GraphKit::make_dtrace_method_entry_exit(ciMethod* method, bool is_entry) {
   const char     *call_name    = is_entry ? "dtrace_method_entry" : "dtrace_method_exit";
 
   // Get base of thread-local storage area
-  Node* thread = _gvn.transform( new (C, 1) ThreadLocalNode() );
+  Node* thread = _gvn.transform( new (C) ThreadLocalNode() );
 
   // Get method
   const TypePtr* method_type = TypeMetadataPtr::make(method);
@@ -175,8 +175,8 @@ void Parse::array_store_check() {
     // Make a constant out of the inexact array klass
     const TypeKlassPtr *extak = tak->cast_to_exactness(true)->is_klassptr();
     Node* con = makecon(extak);
-    Node* cmp = _gvn.transform(new (C, 3) CmpPNode( array_klass, con ));
-    Node* bol = _gvn.transform(new (C, 2) BoolNode( cmp, BoolTest::eq ));
+    Node* cmp = _gvn.transform(new (C) CmpPNode( array_klass, con ));
+    Node* bol = _gvn.transform(new (C) BoolNode( cmp, BoolTest::eq ));
     Node* ctrl= control();
     { BuildCutout unless(this, bol, PROB_MAX);
       uncommon_trap(Deoptimization::Reason_array_check,
@@ -215,8 +215,8 @@ void Parse::emit_guard_for_new(ciInstanceKlass* klass) {
   //   if (klass->_init_thread != current_thread ||
   //       klass->_init_state != being_initialized)
   //      uncommon_trap
-  Node* cur_thread = _gvn.transform( new (C, 1) ThreadLocalNode() );
-  Node* merge = new (C, 3) RegionNode(3);
+  Node* cur_thread = _gvn.transform( new (C) ThreadLocalNode() );
+  Node* merge = new (C) RegionNode(3);
   _gvn.set_type(merge, Type::CONTROL);
   Node* kls = makecon(TypeKlassPtr::make(klass));
 
@@ -322,9 +322,9 @@ void Parse::test_counter_against_threshold(Node* cnt, int limit) {
 
   // Test invocation count vs threshold
   Node *threshold = makecon(TypeInt::make(limit));
-  Node *chk   = _gvn.transform( new (C, 3) CmpUNode( cnt, threshold) );
+  Node *chk   = _gvn.transform( new (C) CmpUNode( cnt, threshold) );
   BoolTest::mask btest = BoolTest::lt;
-  Node *tst   = _gvn.transform( new (C, 2) BoolNode( chk, btest) );
+  Node *tst   = _gvn.transform( new (C) BoolNode( chk, btest) );
   // Branch to failure if threshold exceeded
   { BuildCutout unless(this, tst, PROB_ALWAYS);
     uncommon_trap(Deoptimization::Reason_age,
@@ -348,7 +348,7 @@ void Parse::increment_and_test_invocation_counter(int limit) {
   test_counter_against_threshold(cnt, limit);
 
   // Add one to the counter and store
-  Node* incr = _gvn.transform(new (C, 3) AddINode(cnt, _gvn.intcon(1)));
+  Node* incr = _gvn.transform(new (C) AddINode(cnt, _gvn.intcon(1)));
   store_to_memory( NULL, adr_node, incr, T_INT, adr_type );
 }
 
@@ -369,8 +369,8 @@ Node* Parse::method_data_addressing(ciMethodData* md, ciProfileData* data, ByteS
 
   if (stride != 0) {
     Node* str = _gvn.MakeConX(stride);
-    Node* scale = _gvn.transform( new (C, 3) MulXNode( idx, str ) );
-    ptr   = _gvn.transform( new (C, 4) AddPNode( mdo, ptr, scale ) );
+    Node* scale = _gvn.transform( new (C) MulXNode( idx, str ) );
+    ptr   = _gvn.transform( new (C) AddPNode( mdo, ptr, scale ) );
   }
 
   return ptr;
@@ -382,7 +382,7 @@ void Parse::increment_md_counter_at(ciMethodData* md, ciProfileData* data, ByteS
 
   const TypePtr* adr_type = _gvn.type(adr_node)->is_ptr();
   Node* cnt  = make_load(NULL, adr_node, TypeInt::INT, T_INT, adr_type);
-  Node* incr = _gvn.transform(new (C, 3) AddINode(cnt, _gvn.intcon(DataLayout::counter_increment)));
+  Node* incr = _gvn.transform(new (C) AddINode(cnt, _gvn.intcon(DataLayout::counter_increment)));
   store_to_memory(NULL, adr_node, incr, T_INT, adr_type );
 }
 
@@ -402,7 +402,7 @@ void Parse::set_md_flag_at(ciMethodData* md, ciProfileData* data, int flag_const
 
   const TypePtr* adr_type = _gvn.type(adr_node)->is_ptr();
   Node* flags = make_load(NULL, adr_node, TypeInt::BYTE, T_BYTE, adr_type);
-  Node* incr = _gvn.transform(new (C, 3) OrINode(flags, _gvn.intcon(flag_constant)));
+  Node* incr = _gvn.transform(new (C) OrINode(flags, _gvn.intcon(flag_constant)));
   store_to_memory(NULL, adr_node, incr, T_BYTE, adr_type);
 }
 
diff --git a/src/share/vm/opto/phaseX.cpp b/src/share/vm/opto/phaseX.cpp
index 0fbcf4d09..037359667 100644
--- a/src/share/vm/opto/phaseX.cpp
+++ b/src/share/vm/opto/phaseX.cpp
@@ -48,7 +48,7 @@ NodeHash::NodeHash(uint est_max_size) :
   _total_insert_probes(0), _total_inserts(0),
   _insert_probes(0), _grows(0) {
   // _sentinel must be in the current node space
-  _sentinel = new (Compile::current(), 1) ProjNode(NULL, TypeFunc::Control);
+  _sentinel = new (Compile::current()) ProjNode(NULL, TypeFunc::Control);
   memset(_table,0,sizeof(Node*)*_max);
 }
 
@@ -63,7 +63,7 @@ NodeHash::NodeHash(Arena *arena, uint est_max_size) :
   _total_insert_probes(0), _total_inserts(0),
   _insert_probes(0), _grows(0) {
   // _sentinel must be in the current node space
-  _sentinel = new (Compile::current(), 1) ProjNode(NULL, TypeFunc::Control);
+  _sentinel = new (Compile::current()) ProjNode(NULL, TypeFunc::Control);
   memset(_table,0,sizeof(Node*)*_max);
 }
 
@@ -1246,7 +1246,7 @@ void PhaseIterGVN::subsume_node( Node *old, Node *nn ) {
   }
 
   // Smash all inputs to 'old', isolating him completely
-  Node *temp = new (C, 1) Node(1);
+  Node *temp = new (C) Node(1);
   temp->init_req(0,nn);     // Add a use to nn to prevent him from dying
   remove_dead_node( old );
   temp->del_req(0);         // Yank bogus edge
diff --git a/src/share/vm/opto/reg_split.cpp b/src/share/vm/opto/reg_split.cpp
index 5e0dad80d..fe720e893 100644
--- a/src/share/vm/opto/reg_split.cpp
+++ b/src/share/vm/opto/reg_split.cpp
@@ -646,7 +646,7 @@ uint PhaseChaitin::Split(uint maxlrg, ResourceArea* split_arena) {
           // create a new phi node and insert it into the block
           // type is taken from left over pointer to a predecessor
           assert(n3,"No non-NULL reaching DEF for a Phi");
-          phi = new (C, b->num_preds()) PhiNode(b->head(), n3->bottom_type());
+          phi = new (C) PhiNode(b->head(), n3->bottom_type());
           // initialize the Reaches entry for this LRG
           Reachblock[slidx] = phi;
 
diff --git a/src/share/vm/opto/split_if.cpp b/src/share/vm/opto/split_if.cpp
index bd5eec942..c2f89df36 100644
--- a/src/share/vm/opto/split_if.cpp
+++ b/src/share/vm/opto/split_if.cpp
@@ -35,7 +35,7 @@ Node *PhaseIdealLoop::split_thru_region( Node *n, Node *region ) {
   uint wins = 0;
   assert( n->is_CFG(), "" );
   assert( region->is_Region(), "" );
-  Node *r = new (C, region->req()) RegionNode( region->req() );
+  Node *r = new (C) RegionNode( region->req() );
   IdealLoopTree *loop = get_loop( n );
   for( uint i = 1; i < region->req(); i++ ) {
     Node *x = n->clone();
diff --git a/src/share/vm/opto/stringopts.cpp b/src/share/vm/opto/stringopts.cpp
index 803b13899..6be806acc 100644
--- a/src/share/vm/opto/stringopts.cpp
+++ b/src/share/vm/opto/stringopts.cpp
@@ -69,7 +69,7 @@ class StringConcat : public ResourceObj {
     _multiple(false),
     _string_alloc(NULL),
     _stringopts(stringopts) {
-    _arguments = new (_stringopts->C, 1) Node(1);
+    _arguments = new (_stringopts->C) Node(1);
     _arguments->del_req(0);
   }
 
@@ -220,11 +220,10 @@ class StringConcat : public ResourceObj {
       // Build a new call using the jvms state of the allocate
       address call_addr = SharedRuntime::uncommon_trap_blob()->entry_point();
       const TypeFunc* call_type = OptoRuntime::uncommon_trap_Type();
-      int size = call_type->domain()->cnt();
       const TypePtr* no_memory_effects = NULL;
       Compile* C = _stringopts->C;
-      CallStaticJavaNode* call = new (C, size) CallStaticJavaNode(call_type, call_addr, "uncommon_trap",
-                                                                  jvms->bci(), no_memory_effects);
+      CallStaticJavaNode* call = new (C) CallStaticJavaNode(call_type, call_addr, "uncommon_trap",
+                                                            jvms->bci(), no_memory_effects);
       for (int e = 0; e < TypeFunc::Parms; e++) {
         call->init_req(e, uct->in(e));
       }
@@ -969,9 +968,9 @@ Node* PhaseStringOpts::fetch_static_field(GraphKit& kit, ciField* field) {
 }
 
 Node* PhaseStringOpts::int_stringSize(GraphKit& kit, Node* arg) {
-  RegionNode *final_merge = new (C, 3) RegionNode(3);
+  RegionNode *final_merge = new (C) RegionNode(3);
   kit.gvn().set_type(final_merge, Type::CONTROL);
-  Node* final_size = new (C, 3) PhiNode(final_merge, TypeInt::INT);
+  Node* final_size = new (C) PhiNode(final_merge, TypeInt::INT);
   kit.gvn().set_type(final_size, TypeInt::INT);
 
   IfNode* iff = kit.create_and_map_if(kit.control(),
@@ -988,11 +987,11 @@ Node* PhaseStringOpts::int_stringSize(GraphKit& kit, Node* arg) {
   } else {
 
     // int size = (i < 0) ? stringSize(-i) + 1 : stringSize(i);
-    RegionNode *r = new (C, 3) RegionNode(3);
+    RegionNode *r = new (C) RegionNode(3);
     kit.gvn().set_type(r, Type::CONTROL);
-    Node *phi = new (C, 3) PhiNode(r, TypeInt::INT);
+    Node *phi = new (C) PhiNode(r, TypeInt::INT);
     kit.gvn().set_type(phi, TypeInt::INT);
-    Node *size = new (C, 3) PhiNode(r, TypeInt::INT);
+    Node *size = new (C) PhiNode(r, TypeInt::INT);
     kit.gvn().set_type(size, TypeInt::INT);
     Node* chk = __ CmpI(arg, __ intcon(0));
     Node* p = __ Bool(chk, BoolTest::lt);
@@ -1017,11 +1016,11 @@ Node* PhaseStringOpts::int_stringSize(GraphKit& kit, Node* arg) {
     // Add loop predicate first.
     kit.add_predicate();
 
-    RegionNode *loop = new (C, 3) RegionNode(3);
+    RegionNode *loop = new (C) RegionNode(3);
     loop->init_req(1, kit.control());
     kit.gvn().set_type(loop, Type::CONTROL);
 
-    Node *index = new (C, 3) PhiNode(loop, TypeInt::INT);
+    Node *index = new (C) PhiNode(loop, TypeInt::INT);
     index->init_req(1, __ intcon(0));
     kit.gvn().set_type(index, TypeInt::INT);
     kit.set_control(loop);
@@ -1054,7 +1053,7 @@ Node* PhaseStringOpts::int_stringSize(GraphKit& kit, Node* arg) {
 }
 
 void PhaseStringOpts::int_getChars(GraphKit& kit, Node* arg, Node* char_array, Node* start, Node* end) {
-  RegionNode *final_merge = new (C, 4) RegionNode(4);
+  RegionNode *final_merge = new (C) RegionNode(4);
   kit.gvn().set_type(final_merge, Type::CONTROL);
   Node *final_mem = PhiNode::make(final_merge, kit.memory(char_adr_idx), Type::MEMORY, TypeAryPtr::CHARS);
   kit.gvn().set_type(final_mem, Type::MEMORY);
@@ -1104,11 +1103,11 @@ void PhaseStringOpts::int_getChars(GraphKit& kit, Node* arg, Node* char_array, N
                                         __ Bool(__ CmpI(arg, __ intcon(0)), BoolTest::lt),
                                         PROB_FAIR, COUNT_UNKNOWN);
 
-    RegionNode *merge = new (C, 3) RegionNode(3);
+    RegionNode *merge = new (C) RegionNode(3);
     kit.gvn().set_type(merge, Type::CONTROL);
-    i = new (C, 3) PhiNode(merge, TypeInt::INT);
+    i = new (C) PhiNode(merge, TypeInt::INT);
     kit.gvn().set_type(i, TypeInt::INT);
-    sign = new (C, 3) PhiNode(merge, TypeInt::INT);
+    sign = new (C) PhiNode(merge, TypeInt::INT);
     kit.gvn().set_type(sign, TypeInt::INT);
 
     merge->init_req(1, __ IfTrue(iff));
@@ -1137,10 +1136,10 @@ void PhaseStringOpts::int_getChars(GraphKit& kit, Node* arg, Node* char_array, N
     // Add loop predicate first.
     kit.add_predicate();
 
-    RegionNode *head = new (C, 3) RegionNode(3);
+    RegionNode *head = new (C) RegionNode(3);
     head->init_req(1, kit.control());
     kit.gvn().set_type(head, Type::CONTROL);
-    Node *i_phi = new (C, 3) PhiNode(head, TypeInt::INT);
+    Node *i_phi = new (C) PhiNode(head, TypeInt::INT);
     i_phi->init_req(1, i);
     kit.gvn().set_type(i_phi, TypeInt::INT);
     charPos = PhiNode::make(head, charPos);
@@ -1261,7 +1260,7 @@ void PhaseStringOpts::replace_string_concat(StringConcat* sc) {
   // as a shim for the insertion of the new code.
   JVMState* jvms     = sc->begin()->jvms()->clone_shallow(C);
   uint size = sc->begin()->req();
-  SafePointNode* map = new (C, size) SafePointNode(size, jvms);
+  SafePointNode* map = new (C) SafePointNode(size, jvms);
 
   // copy the control and memory state from the final call into our
   // new starting state.  This allows any preceeding tests to feed
@@ -1306,12 +1305,12 @@ void PhaseStringOpts::replace_string_concat(StringConcat* sc) {
 
   // Create a region for the overflow checks to merge into.
   int args = MAX2(sc->num_arguments(), 1);
-  RegionNode* overflow = new (C, args) RegionNode(args);
+  RegionNode* overflow = new (C) RegionNode(args);
   kit.gvn().set_type(overflow, Type::CONTROL);
 
   // Create a hook node to hold onto the individual sizes since they
   // are need for the copying phase.
-  Node* string_sizes = new (C, args) Node(args);
+  Node* string_sizes = new (C) Node(args);
 
   Node* length = __ intcon(0);
   for (int argi = 0; argi < sc->num_arguments(); argi++) {
@@ -1355,9 +1354,9 @@ void PhaseStringOpts::replace_string_concat(StringConcat* sc) {
         } else if (!type->higher_equal(TypeInstPtr::NOTNULL)) {
           // s = s != null ? s : "null";
           // length = length + (s.count - s.offset);
-          RegionNode *r = new (C, 3) RegionNode(3);
+          RegionNode *r = new (C) RegionNode(3);
           kit.gvn().set_type(r, Type::CONTROL);
-          Node *phi = new (C, 3) PhiNode(r, type);
+          Node *phi = new (C) PhiNode(r, type);
           kit.gvn().set_type(phi, phi->bottom_type());
           Node* p = __ Bool(__ CmpP(arg, kit.null()), BoolTest::ne);
           IfNode* iff = kit.create_and_map_if(kit.control(), p, PROB_MIN, COUNT_UNKNOWN);
diff --git a/src/share/vm/opto/subnode.cpp b/src/share/vm/opto/subnode.cpp
index 9ecd2db7d..81d698526 100644
--- a/src/share/vm/opto/subnode.cpp
+++ b/src/share/vm/opto/subnode.cpp
@@ -149,7 +149,7 @@ Node *SubINode::Ideal(PhaseGVN *phase, bool can_reshape){
   if( t2->base() == Type::Int ){        // Might be bottom or top...
     const TypeInt *i = t2->is_int();
     if( i->is_con() )
-      return new (phase->C, 3) AddINode(in1, phase->intcon(-i->get_con()));
+      return new (phase->C) AddINode(in1, phase->intcon(-i->get_con()));
   }
 
   // Convert "(x+c0) - y" into (x-y) + c0"
@@ -158,8 +158,8 @@ Node *SubINode::Ideal(PhaseGVN *phase, bool can_reshape){
   if( op1 == Op_AddI && ok_to_convert(in1, in2) ) {
     const Type *tadd = phase->type( in1->in(2) );
     if( tadd->singleton() && tadd != Type::TOP ) {
-      Node *sub2 = phase->transform( new (phase->C, 3) SubINode( in1->in(1), in2 ));
-      return new (phase->C, 3) AddINode( sub2, in1->in(2) );
+      Node *sub2 = phase->transform( new (phase->C) SubINode( in1->in(1), in2 ));
+      return new (phase->C) AddINode( sub2, in1->in(2) );
     }
   }
 
@@ -171,9 +171,9 @@ Node *SubINode::Ideal(PhaseGVN *phase, bool can_reshape){
     Node* in22 = in2->in(2);
     const TypeInt* tcon = phase->type(in22)->isa_int();
     if (tcon != NULL && tcon->is_con()) {
-      Node* sub2 = phase->transform( new (phase->C, 3) SubINode(in1, in21) );
+      Node* sub2 = phase->transform( new (phase->C) SubINode(in1, in21) );
       Node* neg_c0 = phase->intcon(- tcon->get_con());
-      return new (phase->C, 3) AddINode(sub2, neg_c0);
+      return new (phase->C) AddINode(sub2, neg_c0);
     }
   }
 
@@ -191,47 +191,47 @@ Node *SubINode::Ideal(PhaseGVN *phase, bool can_reshape){
   // Convert "x - (x+y)" into "-y"
   if( op2 == Op_AddI &&
       phase->eqv( in1, in2->in(1) ) )
-    return new (phase->C, 3) SubINode( phase->intcon(0),in2->in(2));
+    return new (phase->C) SubINode( phase->intcon(0),in2->in(2));
   // Convert "(x-y) - x" into "-y"
   if( op1 == Op_SubI &&
       phase->eqv( in1->in(1), in2 ) )
-    return new (phase->C, 3) SubINode( phase->intcon(0),in1->in(2));
+    return new (phase->C) SubINode( phase->intcon(0),in1->in(2));
   // Convert "x - (y+x)" into "-y"
   if( op2 == Op_AddI &&
       phase->eqv( in1, in2->in(2) ) )
-    return new (phase->C, 3) SubINode( phase->intcon(0),in2->in(1));
+    return new (phase->C) SubINode( phase->intcon(0),in2->in(1));
 
   // Convert "0 - (x-y)" into "y-x"
   if( t1 == TypeInt::ZERO && op2 == Op_SubI )
-    return new (phase->C, 3) SubINode( in2->in(2), in2->in(1) );
+    return new (phase->C) SubINode( in2->in(2), in2->in(1) );
 
   // Convert "0 - (x+con)" into "-con-x"
   jint con;
   if( t1 == TypeInt::ZERO && op2 == Op_AddI &&
       (con = in2->in(2)->find_int_con(0)) != 0 )
-    return new (phase->C, 3) SubINode( phase->intcon(-con), in2->in(1) );
+    return new (phase->C) SubINode( phase->intcon(-con), in2->in(1) );
 
   // Convert "(X+A) - (X+B)" into "A - B"
   if( op1 == Op_AddI && op2 == Op_AddI && in1->in(1) == in2->in(1) )
-    return new (phase->C, 3) SubINode( in1->in(2), in2->in(2) );
+    return new (phase->C) SubINode( in1->in(2), in2->in(2) );
 
   // Convert "(A+X) - (B+X)" into "A - B"
   if( op1 == Op_AddI && op2 == Op_AddI && in1->in(2) == in2->in(2) )
-    return new (phase->C, 3) SubINode( in1->in(1), in2->in(1) );
+    return new (phase->C) SubINode( in1->in(1), in2->in(1) );
 
   // Convert "(A+X) - (X+B)" into "A - B"
   if( op1 == Op_AddI && op2 == Op_AddI && in1->in(2) == in2->in(1) )
-    return new (phase->C, 3) SubINode( in1->in(1), in2->in(2) );
+    return new (phase->C) SubINode( in1->in(1), in2->in(2) );
 
   // Convert "(X+A) - (B+X)" into "A - B"
   if( op1 == Op_AddI && op2 == Op_AddI && in1->in(1) == in2->in(2) )
-    return new (phase->C, 3) SubINode( in1->in(2), in2->in(1) );
+    return new (phase->C) SubINode( in1->in(2), in2->in(1) );
 
   // Convert "A-(B-C)" into (A+C)-B", since add is commutative and generally
   // nicer to optimize than subtract.
   if( op2 == Op_SubI && in2->outcnt() == 1) {
-    Node *add1 = phase->transform( new (phase->C, 3) AddINode( in1, in2->in(2) ) );
-    return new (phase->C, 3) SubINode( add1, in2->in(1) );
+    Node *add1 = phase->transform( new (phase->C) AddINode( in1, in2->in(2) ) );
+    return new (phase->C) SubINode( add1, in2->in(1) );
   }
 
   return NULL;
@@ -278,7 +278,7 @@ Node *SubLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // Convert "x-c0" into "x+ -c0".
   if( i &&                      // Might be bottom or top...
       i->is_con() )
-    return new (phase->C, 3) AddLNode(in1, phase->longcon(-i->get_con()));
+    return new (phase->C) AddLNode(in1, phase->longcon(-i->get_con()));
 
   // Convert "(x+c0) - y" into (x-y) + c0"
   // Do not collapse (x+c0)-y if "+" is a loop increment or
@@ -287,8 +287,8 @@ Node *SubLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node *in11 = in1->in(1);
     const Type *tadd = phase->type( in1->in(2) );
     if( tadd->singleton() && tadd != Type::TOP ) {
-      Node *sub2 = phase->transform( new (phase->C, 3) SubLNode( in11, in2 ));
-      return new (phase->C, 3) AddLNode( sub2, in1->in(2) );
+      Node *sub2 = phase->transform( new (phase->C) SubLNode( in11, in2 ));
+      return new (phase->C) AddLNode( sub2, in1->in(2) );
     }
   }
 
@@ -299,9 +299,9 @@ Node *SubLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     Node* in22 = in2->in(2);
     const TypeLong* tcon = phase->type(in22)->isa_long();
     if (tcon != NULL && tcon->is_con()) {
-      Node* sub2 = phase->transform( new (phase->C, 3) SubLNode(in1, in21) );
+      Node* sub2 = phase->transform( new (phase->C) SubLNode(in1, in21) );
       Node* neg_c0 = phase->longcon(- tcon->get_con());
-      return new (phase->C, 3) AddLNode(sub2, neg_c0);
+      return new (phase->C) AddLNode(sub2, neg_c0);
     }
   }
 
@@ -319,28 +319,28 @@ Node *SubLNode::Ideal(PhaseGVN *phase, bool can_reshape) {
   // Convert "x - (x+y)" into "-y"
   if( op2 == Op_AddL &&
       phase->eqv( in1, in2->in(1) ) )
-    return new (phase->C, 3) SubLNode( phase->makecon(TypeLong::ZERO), in2->in(2));
+    return new (phase->C) SubLNode( phase->makecon(TypeLong::ZERO), in2->in(2));
   // Convert "x - (y+x)" into "-y"
   if( op2 == Op_AddL &&
       phase->eqv( in1, in2->in(2) ) )
-    return new (phase->C, 3) SubLNode( phase->makecon(TypeLong::ZERO),in2->in(1));
+    return new (phase->C) SubLNode( phase->makecon(TypeLong::ZERO),in2->in(1));
 
   // Convert "0 - (x-y)" into "y-x"
   if( phase->type( in1 ) == TypeLong::ZERO && op2 == Op_SubL )
-    return new (phase->C, 3) SubLNode( in2->in(2), in2->in(1) );
+    return new (phase->C) SubLNode( in2->in(2), in2->in(1) );
 
   // Convert "(X+A) - (X+B)" into "A - B"
   if( op1 == Op_AddL && op2 == Op_AddL && in1->in(1) == in2->in(1) )
-    return new (phase->C, 3) SubLNode( in1->in(2), in2->in(2) );
+    return new (phase->C) SubLNode( in1->in(2), in2->in(2) );
 
   // Convert "(A+X) - (B+X)" into "A - B"
   if( op1 == Op_AddL && op2 == Op_AddL && in1->in(2) == in2->in(2) )
-    return new (phase->C, 3) SubLNode( in1->in(1), in2->in(1) );
+    return new (phase->C) SubLNode( in1->in(1), in2->in(1) );
 
   // Convert "A-(B-C)" into (A+C)-B"
   if( op2 == Op_SubL && in2->outcnt() == 1) {
-    Node *add1 = phase->transform( new (phase->C, 3) AddLNode( in1, in2->in(2) ) );
-    return new (phase->C, 3) SubLNode( add1, in2->in(1) );
+    Node *add1 = phase->transform( new (phase->C) AddLNode( in1, in2->in(2) ) );
+    return new (phase->C) SubLNode( add1, in2->in(1) );
   }
 
   return NULL;
@@ -407,7 +407,7 @@ Node *SubFNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     // Convert "x - (x+y)" into "-y"
     if( in(2)->is_Add() &&
         phase->eqv(in(1),in(2)->in(1) ) )
-      return new (phase->C, 3) SubFNode( phase->makecon(TypeF::ZERO),in(2)->in(2));
+      return new (phase->C) SubFNode( phase->makecon(TypeF::ZERO),in(2)->in(2));
   }
 
   // Cannot replace 0.0-X with -X because a 'fsub' bytecode computes
@@ -450,7 +450,7 @@ Node *SubDNode::Ideal(PhaseGVN *phase, bool can_reshape){
     // Convert "x - (x+y)" into "-y"
     if( in(2)->is_Add() &&
         phase->eqv(in(1),in(2)->in(1) ) )
-      return new (phase->C, 3) SubDNode( phase->makecon(TypeD::ZERO),in(2)->in(2));
+      return new (phase->C) SubDNode( phase->makecon(TypeD::ZERO),in(2)->in(2));
   }
 
   // Cannot replace 0.0-X with -X because a 'dsub' bytecode computes
@@ -581,11 +581,11 @@ Node *CmpINode::Ideal( PhaseGVN *phase, bool can_reshape ) {
   if (phase->type(in(2))->higher_equal(TypeInt::ZERO)) {
     switch (in(1)->Opcode()) {
     case Op_CmpL3:              // Collapse a CmpL3/CmpI into a CmpL
-      return new (phase->C, 3) CmpLNode(in(1)->in(1),in(1)->in(2));
+      return new (phase->C) CmpLNode(in(1)->in(1),in(1)->in(2));
     case Op_CmpF3:              // Collapse a CmpF3/CmpI into a CmpF
-      return new (phase->C, 3) CmpFNode(in(1)->in(1),in(1)->in(2));
+      return new (phase->C) CmpFNode(in(1)->in(1),in(1)->in(2));
     case Op_CmpD3:              // Collapse a CmpD3/CmpI into a CmpD
-      return new (phase->C, 3) CmpDNode(in(1)->in(1),in(1)->in(2));
+      return new (phase->C) CmpDNode(in(1)->in(1),in(1)->in(2));
     //case Op_SubI:
       // If (x - y) cannot overflow, then ((x - y) <?> 0)
       // can be turned into (x <?> y).
@@ -1023,8 +1023,8 @@ Node *CmpDNode::Ideal(PhaseGVN *phase, bool can_reshape){
         new_in2 = tmp;
       }
       CmpFNode *new_cmp = (Opcode() == Op_CmpD3)
-        ? new (phase->C, 3) CmpF3Node( new_in1, new_in2 )
-        : new (phase->C, 3) CmpFNode ( new_in1, new_in2 ) ;
+        ? new (phase->C) CmpF3Node( new_in1, new_in2 )
+        : new (phase->C) CmpFNode ( new_in1, new_in2 ) ;
       return new_cmp;           // Changed to CmpFNode
     }
     // Testing value required the precision of a double
@@ -1085,7 +1085,7 @@ static Node *clone_cmp( Node *cmp, Node *cmp1, Node *cmp2, PhaseGVN *gvn, BoolTe
   ncmp->set_req(1,cmp1);
   ncmp->set_req(2,cmp2);
   ncmp = gvn->transform( ncmp );
-  return new (gvn->C, 2) BoolNode( ncmp, test );
+  return new (gvn->C) BoolNode( ncmp, test );
 }
 
 //-------------------------------make_predicate--------------------------------
@@ -1106,9 +1106,9 @@ Node* BoolNode::make_predicate(Node* test_value, PhaseGVN* phase) {
     // Else fall through.  The CMove gets in the way of the test.
     // It should be the case that make_predicate(bol->as_int_value()) == bol.
   }
-  Node* cmp = new (C, 3) CmpINode(test_value, phase->intcon(0));
+  Node* cmp = new (C) CmpINode(test_value, phase->intcon(0));
   cmp = phase->transform(cmp);
-  Node* bol = new (C, 2) BoolNode(cmp, BoolTest::ne);
+  Node* bol = new (C) BoolNode(cmp, BoolTest::ne);
   return phase->transform(bol);
 }
 
@@ -1124,7 +1124,7 @@ Node* BoolNode::as_int_value(PhaseGVN* phase) {
 //----------------------------------negate-------------------------------------
 BoolNode* BoolNode::negate(PhaseGVN* phase) {
   Compile* C = phase->C;
-  return new (C, 2) BoolNode(in(1), _test.negate());
+  return new (C) BoolNode(in(1), _test.negate());
 }
 
 
@@ -1158,7 +1158,7 @@ Node *BoolNode::Ideal(PhaseGVN *phase, bool can_reshape) {
     // Swap inputs to the clone
     cmp->swap_edges(1, 2);
     cmp = phase->transform( cmp );
-    return new (phase->C, 2) BoolNode( cmp, _test.commute() );
+    return new (phase->C) BoolNode( cmp, _test.commute() );
   }
 
   // Change "bool eq/ne (cmp (xor X 1) 0)" into "bool ne/eq (cmp X 0)".
@@ -1175,8 +1175,8 @@ Node *BoolNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       phase->type( j_xor->in(2) ) == TypeInt::ONE &&
       (_test._test == BoolTest::eq ||
        _test._test == BoolTest::ne) ) {
-    Node *ncmp = phase->transform(new (phase->C, 3) CmpINode(j_xor->in(1),cmp2));
-    return new (phase->C, 2) BoolNode( ncmp, _test.negate() );
+    Node *ncmp = phase->transform(new (phase->C) CmpINode(j_xor->in(1),cmp2));
+    return new (phase->C) BoolNode( ncmp, _test.negate() );
   }
 
   // Change "bool eq/ne (cmp (Conv2B X) 0)" into "bool eq/ne (cmp X 0)".
@@ -1187,10 +1187,10 @@ Node *BoolNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       (_test._test == BoolTest::eq ||
        _test._test == BoolTest::ne) ) {
     Node *ncmp = phase->transform(phase->type(c2b->in(1))->isa_int()
-       ? (Node*)new (phase->C, 3) CmpINode(c2b->in(1),cmp2)
-       : (Node*)new (phase->C, 3) CmpPNode(c2b->in(1),phase->makecon(TypePtr::NULL_PTR))
+       ? (Node*)new (phase->C) CmpINode(c2b->in(1),cmp2)
+       : (Node*)new (phase->C) CmpPNode(c2b->in(1),phase->makecon(TypePtr::NULL_PTR))
     );
-    return new (phase->C, 2) BoolNode( ncmp, _test._test );
+    return new (phase->C) BoolNode( ncmp, _test._test );
   }
 
   // Comparing a SubI against a zero is equal to comparing the SubI
@@ -1200,8 +1200,8 @@ Node *BoolNode::Ideal(PhaseGVN *phase, bool can_reshape) {
         (cop == Op_CmpI) &&
         (cmp1->Opcode() == Op_SubI) &&
         ( cmp2_type == TypeInt::ZERO ) ) {
-    Node *ncmp = phase->transform( new (phase->C, 3) CmpINode(cmp1->in(1),cmp1->in(2)));
-    return new (phase->C, 2) BoolNode( ncmp, _test._test );
+    Node *ncmp = phase->transform( new (phase->C) CmpINode(cmp1->in(1),cmp1->in(2)));
+    return new (phase->C) BoolNode( ncmp, _test._test );
   }
 
   // Change (-A vs 0) into (A vs 0) by commuting the test.  Disallow in the
@@ -1212,8 +1212,8 @@ Node *BoolNode::Ideal(PhaseGVN *phase, bool can_reshape) {
       cmp2_type == TypeInt::ZERO &&
       phase->type( cmp1->in(1) ) == TypeInt::ZERO &&
       phase->type( cmp1->in(2) )->higher_equal(TypeInt::SYMINT) ) {
-    Node *ncmp = phase->transform( new (phase->C, 3) CmpINode(cmp1->in(2),cmp2));
-    return new (phase->C, 2) BoolNode( ncmp, _test.commute() );
+    Node *ncmp = phase->transform( new (phase->C) CmpINode(cmp1->in(2),cmp2));
+    return new (phase->C) BoolNode( ncmp, _test.commute() );
   }
 
   //  The transformation below is not valid for either signed or unsigned
diff --git a/src/share/vm/opto/superword.cpp b/src/share/vm/opto/superword.cpp
index 8d640c1e6..20e287dc0 100644
--- a/src/share/vm/opto/superword.cpp
+++ b/src/share/vm/opto/superword.cpp
@@ -179,6 +179,7 @@ void SuperWord::find_adjacent_refs() {
   for (int i = 0; i < _block.length(); i++) {
     Node* n = _block.at(i);
     if (n->is_Mem() && !n->is_LoadStore() && in_bb(n) &&
+        n->Opcode() != Op_LoadUI2L &&
         is_java_primitive(n->as_Mem()->memory_type())) {
       int align = memory_alignment(n->as_Mem(), 0);
       if (align != bottom_align) {
@@ -481,12 +482,19 @@ int SuperWord::get_iv_adjustment(MemNode* mem_ref) {
   int vw       = vector_width_in_bytes(mem_ref);
   assert(vw > 1, "sanity");
   int stride_sign   = (scale * iv_stride()) > 0 ? 1 : -1;
-  int iv_adjustment = (stride_sign * vw - (offset % vw)) % vw;
+  // At least one iteration is executed in pre-loop by default. As result
+  // several iterations are needed to align memory operations in main-loop even
+  // if offset is 0.
+  int iv_adjustment_in_bytes = (stride_sign * vw - (offset % vw));
+  int elt_size = align_to_ref_p.memory_size();
+  assert(((ABS(iv_adjustment_in_bytes) % elt_size) == 0),
+         err_msg_res("(%d) should be divisible by (%d)", iv_adjustment_in_bytes, elt_size));
+  int iv_adjustment = iv_adjustment_in_bytes/elt_size;
 
 #ifndef PRODUCT
   if (TraceSuperWord)
     tty->print_cr("\noffset = %d iv_adjust = %d elt_size = %d scale = %d iv_stride = %d vect_size %d",
-                  offset, iv_adjustment, align_to_ref_p.memory_size(), scale, iv_stride(), vw);
+                  offset, iv_adjustment, elt_size, scale, iv_stride(), vw);
 #endif
   return iv_adjustment;
 }
@@ -1080,23 +1088,22 @@ bool SuperWord::profitable(Node_List* p) {
   uint start, end;
   VectorNode::vector_operands(p0, &start, &end);
 
-  // Return false if some input is not vector and inside block
+  // Return false if some inputs are not vectors or vectors with different
+  // size or alignment.
+  // Also, for now, return false if not scalar promotion case when inputs are
+  // the same. Later, implement PackNode and allow differing, non-vector inputs
+  // (maybe just the ones from outside the block.)
   for (uint i = start; i < end; i++) {
-    if (!is_vector_use(p0, i)) {
-      // For now, return false if not scalar promotion case (inputs are the same.)
-      // Later, implement PackNode and allow differing, non-vector inputs
-      // (maybe just the ones from outside the block.)
-      if (!same_inputs(p, i)) {
-        return false;
-      }
-    }
+    if (!is_vector_use(p0, i))
+      return false;
   }
   if (VectorNode::is_shift(p0)) {
-    // For now, return false if shift count is vector because
-    // hw does not support it.
-    if (is_vector_use(p0, 2))
+    // For now, return false if shift count is vector or not scalar promotion
+    // case (different shift counts) because it is not supported yet.
+    Node* cnt = p0->in(2);
+    Node_List* cnt_pk = my_pack(cnt);
+    if (cnt_pk != NULL)
       return false;
-    // For the same reason return false if different shift counts.
     if (!same_inputs(p, 2))
       return false;
   }
@@ -1350,11 +1357,14 @@ void SuperWord::output() {
     insert_extracts(_packset.at(i));
   }
 
+  Compile* C = _phase->C;
+  uint max_vlen_in_bytes = 0;
   for (int i = 0; i < _block.length(); i++) {
     Node* n = _block.at(i);
     Node_List* p = my_pack(n);
     if (p && n == executed_last(p)) {
       uint vlen = p->size();
+      uint vlen_in_bytes = 0;
       Node* vn = NULL;
       Node* low_adr = p->at(0);
       Node* first   = executed_first(p);
@@ -1364,7 +1374,8 @@ void SuperWord::output() {
         Node* mem = first->in(MemNode::Memory);
         Node* adr = low_adr->in(MemNode::Address);
         const TypePtr* atyp = n->adr_type();
-        vn = LoadVectorNode::make(_phase->C, opc, ctl, mem, adr, atyp, vlen, velt_basic_type(n));
+        vn = LoadVectorNode::make(C, opc, ctl, mem, adr, atyp, vlen, velt_basic_type(n));
+        vlen_in_bytes = vn->as_LoadVector()->memory_size();
       } else if (n->is_Store()) {
         // Promote value to be stored to vector
         Node* val = vector_opd(p, MemNode::ValueIn);
@@ -1372,7 +1383,8 @@ void SuperWord::output() {
         Node* mem = first->in(MemNode::Memory);
         Node* adr = low_adr->in(MemNode::Address);
         const TypePtr* atyp = n->adr_type();
-        vn = StoreVectorNode::make(_phase->C, opc, ctl, mem, adr, atyp, val, vlen);
+        vn = StoreVectorNode::make(C, opc, ctl, mem, adr, atyp, val, vlen);
+        vlen_in_bytes = vn->as_StoreVector()->memory_size();
       } else if (n->req() == 3) {
         // Promote operands to vector
         Node* in1 = vector_opd(p, 1);
@@ -1383,18 +1395,23 @@ void SuperWord::output() {
           in1 = in2;
           in2 = tmp;
         }
-        vn = VectorNode::make(_phase->C, opc, in1, in2, vlen, velt_basic_type(n));
+        vn = VectorNode::make(C, opc, in1, in2, vlen, velt_basic_type(n));
+        vlen_in_bytes = vn->as_Vector()->length_in_bytes();
       } else {
         ShouldNotReachHere();
       }
       assert(vn != NULL, "sanity");
-      _phase->_igvn.register_new_node_with_optimizer(vn);
+      _igvn.register_new_node_with_optimizer(vn);
       _phase->set_ctrl(vn, _phase->get_ctrl(p->at(0)));
       for (uint j = 0; j < p->size(); j++) {
         Node* pm = p->at(j);
         _igvn.replace_node(pm, vn);
       }
       _igvn._worklist.push(vn);
+
+      if (vlen_in_bytes > max_vlen_in_bytes) {
+        max_vlen_in_bytes = vlen_in_bytes;
+      }
 #ifdef ASSERT
       if (TraceNewVectors) {
         tty->print("new Vector node: ");
@@ -1403,6 +1420,7 @@ void SuperWord::output() {
 #endif
     }
   }
+  C->set_max_vector_size(max_vlen_in_bytes);
 }
 
 //------------------------------vector_opd---------------------------
@@ -1432,17 +1450,17 @@ Node* SuperWord::vector_opd(Node_List* p, int opd_idx) {
       } else {
         if (t == NULL || t->_lo < 0 || t->_hi > (int)mask) {
           cnt = ConNode::make(C, TypeInt::make(mask));
-          _phase->_igvn.register_new_node_with_optimizer(cnt);
-          cnt = new (C, 3) AndINode(opd, cnt);
-          _phase->_igvn.register_new_node_with_optimizer(cnt);
+          _igvn.register_new_node_with_optimizer(cnt);
+          cnt = new (C) AndINode(opd, cnt);
+          _igvn.register_new_node_with_optimizer(cnt);
           _phase->set_ctrl(cnt, _phase->get_ctrl(opd));
         }
         assert(opd->bottom_type()->isa_int(), "int type only");
         // Move non constant shift count into XMM register.
-        cnt = new (_phase->C, 2) MoveI2FNode(cnt);
+        cnt = new (C) MoveI2FNode(cnt);
       }
       if (cnt != opd) {
-        _phase->_igvn.register_new_node_with_optimizer(cnt);
+        _igvn.register_new_node_with_optimizer(cnt);
         _phase->set_ctrl(cnt, _phase->get_ctrl(opd));
       }
       return cnt;
@@ -1454,7 +1472,7 @@ Node* SuperWord::vector_opd(Node_List* p, int opd_idx) {
     const Type* p0_t = velt_type(p0);
     VectorNode* vn = VectorNode::scalar2vector(_phase->C, opd, vlen, p0_t);
 
-    _phase->_igvn.register_new_node_with_optimizer(vn);
+    _igvn.register_new_node_with_optimizer(vn);
     _phase->set_ctrl(vn, _phase->get_ctrl(opd));
 #ifdef ASSERT
     if (TraceNewVectors) {
@@ -1477,13 +1495,13 @@ Node* SuperWord::vector_opd(Node_List* p, int opd_idx) {
     assert(opd_bt == in->bottom_type()->basic_type(), "all same type");
     pk->add_opd(in);
   }
-  _phase->_igvn.register_new_node_with_optimizer(pk);
+  _igvn.register_new_node_with_optimizer(pk);
   _phase->set_ctrl(pk, _phase->get_ctrl(opd));
 #ifdef ASSERT
-    if (TraceNewVectors) {
-      tty->print("new Vector node: ");
-      pk->dump();
-    }
+  if (TraceNewVectors) {
+    tty->print("new Vector node: ");
+    pk->dump();
+  }
 #endif
   return pk;
 }
@@ -1524,7 +1542,7 @@ void SuperWord::insert_extracts(Node_List* p) {
     int def_pos = alignment(def) / data_size(def);
 
     Node* ex = ExtractNode::make(_phase->C, def, def_pos, velt_basic_type(def));
-    _phase->_igvn.register_new_node_with_optimizer(ex);
+    _igvn.register_new_node_with_optimizer(ex);
     _phase->set_ctrl(ex, _phase->get_ctrl(def));
     _igvn.replace_input_of(use, idx, ex);
     _igvn._worklist.push(def);
@@ -1805,7 +1823,7 @@ void SuperWord::compute_vector_element_type() {
 
 //------------------------------memory_alignment---------------------------
 // Alignment within a vector memory reference
-int SuperWord::memory_alignment(MemNode* s, int iv_adjust_in_bytes) {
+int SuperWord::memory_alignment(MemNode* s, int iv_adjust) {
   SWPointer p(s, this);
   if (!p.valid()) {
     return bottom_align;
@@ -1815,7 +1833,7 @@ int SuperWord::memory_alignment(MemNode* s, int iv_adjust_in_bytes) {
     return bottom_align; // No vectors for this type
   }
   int offset  = p.offset_in_bytes();
-  offset     += iv_adjust_in_bytes;
+  offset     += iv_adjust*p.memory_size();
   int off_rem = offset % vw;
   int off_mod = off_rem >= 0 ? off_rem : off_rem + vw;
   return off_mod;
@@ -1838,7 +1856,7 @@ const Type* SuperWord::container_type(Node* n) {
 
 bool SuperWord::same_velt_type(Node* n1, Node* n2) {
   const Type* vt1 = velt_type(n1);
-  const Type* vt2 = velt_type(n1);
+  const Type* vt2 = velt_type(n2);
   if (vt1->basic_type() == T_INT && vt2->basic_type() == T_INT) {
     // Compare vectors element sizes for integer types.
     return data_size(n1) == data_size(n2);
@@ -2003,73 +2021,73 @@ void SuperWord::align_initial_loop_index(MemNode* align_to_ref) {
   if (align_to_ref_p.invar() != NULL) {
     // incorporate any extra invariant piece producing (offset +/- invar) >>> log2(elt)
     Node* log2_elt = _igvn.intcon(exact_log2(elt_size));
-    Node* aref     = new (_phase->C, 3) URShiftINode(align_to_ref_p.invar(), log2_elt);
-    _phase->_igvn.register_new_node_with_optimizer(aref);
+    Node* aref     = new (_phase->C) URShiftINode(align_to_ref_p.invar(), log2_elt);
+    _igvn.register_new_node_with_optimizer(aref);
     _phase->set_ctrl(aref, pre_ctrl);
     if (align_to_ref_p.negate_invar()) {
-      e = new (_phase->C, 3) SubINode(e, aref);
+      e = new (_phase->C) SubINode(e, aref);
     } else {
-      e = new (_phase->C, 3) AddINode(e, aref);
+      e = new (_phase->C) AddINode(e, aref);
     }
-    _phase->_igvn.register_new_node_with_optimizer(e);
+    _igvn.register_new_node_with_optimizer(e);
     _phase->set_ctrl(e, pre_ctrl);
   }
   if (vw > ObjectAlignmentInBytes) {
     // incorporate base e +/- base && Mask >>> log2(elt)
-    Node* xbase = new(_phase->C, 2) CastP2XNode(NULL, align_to_ref_p.base());
-    _phase->_igvn.register_new_node_with_optimizer(xbase);
+    Node* xbase = new(_phase->C) CastP2XNode(NULL, align_to_ref_p.base());
+    _igvn.register_new_node_with_optimizer(xbase);
 #ifdef _LP64
-    xbase  = new (_phase->C, 2) ConvL2INode(xbase);
-    _phase->_igvn.register_new_node_with_optimizer(xbase);
+    xbase  = new (_phase->C) ConvL2INode(xbase);
+    _igvn.register_new_node_with_optimizer(xbase);
 #endif
     Node* mask = _igvn.intcon(vw-1);
-    Node* masked_xbase  = new (_phase->C, 3) AndINode(xbase, mask);
-    _phase->_igvn.register_new_node_with_optimizer(masked_xbase);
+    Node* masked_xbase  = new (_phase->C) AndINode(xbase, mask);
+    _igvn.register_new_node_with_optimizer(masked_xbase);
     Node* log2_elt = _igvn.intcon(exact_log2(elt_size));
-    Node* bref     = new (_phase->C, 3) URShiftINode(masked_xbase, log2_elt);
-    _phase->_igvn.register_new_node_with_optimizer(bref);
+    Node* bref     = new (_phase->C) URShiftINode(masked_xbase, log2_elt);
+    _igvn.register_new_node_with_optimizer(bref);
     _phase->set_ctrl(bref, pre_ctrl);
-    e = new (_phase->C, 3) AddINode(e, bref);
-    _phase->_igvn.register_new_node_with_optimizer(e);
+    e = new (_phase->C) AddINode(e, bref);
+    _igvn.register_new_node_with_optimizer(e);
     _phase->set_ctrl(e, pre_ctrl);
   }
 
   // compute e +/- lim0
   if (scale < 0) {
-    e = new (_phase->C, 3) SubINode(e, lim0);
+    e = new (_phase->C) SubINode(e, lim0);
   } else {
-    e = new (_phase->C, 3) AddINode(e, lim0);
+    e = new (_phase->C) AddINode(e, lim0);
   }
-  _phase->_igvn.register_new_node_with_optimizer(e);
+  _igvn.register_new_node_with_optimizer(e);
   _phase->set_ctrl(e, pre_ctrl);
 
   if (stride * scale > 0) {
     // compute V - (e +/- lim0)
     Node* va  = _igvn.intcon(v_align);
-    e = new (_phase->C, 3) SubINode(va, e);
-    _phase->_igvn.register_new_node_with_optimizer(e);
+    e = new (_phase->C) SubINode(va, e);
+    _igvn.register_new_node_with_optimizer(e);
     _phase->set_ctrl(e, pre_ctrl);
   }
   // compute N = (exp) % V
   Node* va_msk = _igvn.intcon(v_align - 1);
-  Node* N = new (_phase->C, 3) AndINode(e, va_msk);
-  _phase->_igvn.register_new_node_with_optimizer(N);
+  Node* N = new (_phase->C) AndINode(e, va_msk);
+  _igvn.register_new_node_with_optimizer(N);
   _phase->set_ctrl(N, pre_ctrl);
 
   //   substitute back into (1), so that new limit
   //     lim = lim0 + N
   Node* lim;
   if (stride < 0) {
-    lim = new (_phase->C, 3) SubINode(lim0, N);
+    lim = new (_phase->C) SubINode(lim0, N);
   } else {
-    lim = new (_phase->C, 3) AddINode(lim0, N);
+    lim = new (_phase->C) AddINode(lim0, N);
   }
-  _phase->_igvn.register_new_node_with_optimizer(lim);
+  _igvn.register_new_node_with_optimizer(lim);
   _phase->set_ctrl(lim, pre_ctrl);
   Node* constrained =
-    (stride > 0) ? (Node*) new (_phase->C,3) MinINode(lim, orig_limit)
-                 : (Node*) new (_phase->C,3) MaxINode(lim, orig_limit);
-  _phase->_igvn.register_new_node_with_optimizer(constrained);
+    (stride > 0) ? (Node*) new (_phase->C) MinINode(lim, orig_limit)
+                 : (Node*) new (_phase->C) MaxINode(lim, orig_limit);
+  _igvn.register_new_node_with_optimizer(constrained);
   _phase->set_ctrl(constrained, pre_ctrl);
   _igvn.hash_delete(pre_opaq);
   pre_opaq->set_req(1, constrained);
diff --git a/src/share/vm/opto/superword.hpp b/src/share/vm/opto/superword.hpp
index 050ce9ea2..45c0f9a83 100644
--- a/src/share/vm/opto/superword.hpp
+++ b/src/share/vm/opto/superword.hpp
@@ -400,7 +400,7 @@ class SuperWord : public ResourceObj {
   // Return the node executed last in pack p.
   Node* executed_last(Node_List* p);
   // Alignment within a vector memory reference
-  int memory_alignment(MemNode* s, int iv_adjust_in_bytes);
+  int memory_alignment(MemNode* s, int iv_adjust);
   // (Start, end] half-open range defining which operands are vector
   void vector_opd_range(Node* n, uint* start, uint* end);
   // Smallest type containing range of values
diff --git a/src/share/vm/opto/vectornode.cpp b/src/share/vm/opto/vectornode.cpp
index 7c8e52199..180b72277 100644
--- a/src/share/vm/opto/vectornode.cpp
+++ b/src/share/vm/opto/vectornode.cpp
@@ -245,46 +245,46 @@ VectorNode* VectorNode::make(Compile* C, int opc, Node* n1, Node* n2, uint vlen,
   int vopc = VectorNode::opcode(opc, bt);
 
   switch (vopc) {
-  case Op_AddVB: return new (C, 3) AddVBNode(n1, n2, vt);
-  case Op_AddVS: return new (C, 3) AddVSNode(n1, n2, vt);
-  case Op_AddVI: return new (C, 3) AddVINode(n1, n2, vt);
-  case Op_AddVL: return new (C, 3) AddVLNode(n1, n2, vt);
-  case Op_AddVF: return new (C, 3) AddVFNode(n1, n2, vt);
-  case Op_AddVD: return new (C, 3) AddVDNode(n1, n2, vt);
+  case Op_AddVB: return new (C) AddVBNode(n1, n2, vt);
+  case Op_AddVS: return new (C) AddVSNode(n1, n2, vt);
+  case Op_AddVI: return new (C) AddVINode(n1, n2, vt);
+  case Op_AddVL: return new (C) AddVLNode(n1, n2, vt);
+  case Op_AddVF: return new (C) AddVFNode(n1, n2, vt);
+  case Op_AddVD: return new (C) AddVDNode(n1, n2, vt);
 
-  case Op_SubVB: return new (C, 3) SubVBNode(n1, n2, vt);
-  case Op_SubVS: return new (C, 3) SubVSNode(n1, n2, vt);
-  case Op_SubVI: return new (C, 3) SubVINode(n1, n2, vt);
-  case Op_SubVL: return new (C, 3) SubVLNode(n1, n2, vt);
-  case Op_SubVF: return new (C, 3) SubVFNode(n1, n2, vt);
-  case Op_SubVD: return new (C, 3) SubVDNode(n1, n2, vt);
+  case Op_SubVB: return new (C) SubVBNode(n1, n2, vt);
+  case Op_SubVS: return new (C) SubVSNode(n1, n2, vt);
+  case Op_SubVI: return new (C) SubVINode(n1, n2, vt);
+  case Op_SubVL: return new (C) SubVLNode(n1, n2, vt);
+  case Op_SubVF: return new (C) SubVFNode(n1, n2, vt);
+  case Op_SubVD: return new (C) SubVDNode(n1, n2, vt);
 
-  case Op_MulVS: return new (C, 3) MulVSNode(n1, n2, vt);
-  case Op_MulVI: return new (C, 3) MulVINode(n1, n2, vt);
-  case Op_MulVF: return new (C, 3) MulVFNode(n1, n2, vt);
-  case Op_MulVD: return new (C, 3) MulVDNode(n1, n2, vt);
+  case Op_MulVS: return new (C) MulVSNode(n1, n2, vt);
+  case Op_MulVI: return new (C) MulVINode(n1, n2, vt);
+  case Op_MulVF: return new (C) MulVFNode(n1, n2, vt);
+  case Op_MulVD: return new (C) MulVDNode(n1, n2, vt);
 
-  case Op_DivVF: return new (C, 3) DivVFNode(n1, n2, vt);
-  case Op_DivVD: return new (C, 3) DivVDNode(n1, n2, vt);
+  case Op_DivVF: return new (C) DivVFNode(n1, n2, vt);
+  case Op_DivVD: return new (C) DivVDNode(n1, n2, vt);
 
-  case Op_LShiftVB: return new (C, 3) LShiftVBNode(n1, n2, vt);
-  case Op_LShiftVS: return new (C, 3) LShiftVSNode(n1, n2, vt);
-  case Op_LShiftVI: return new (C, 3) LShiftVINode(n1, n2, vt);
-  case Op_LShiftVL: return new (C, 3) LShiftVLNode(n1, n2, vt);
+  case Op_LShiftVB: return new (C) LShiftVBNode(n1, n2, vt);
+  case Op_LShiftVS: return new (C) LShiftVSNode(n1, n2, vt);
+  case Op_LShiftVI: return new (C) LShiftVINode(n1, n2, vt);
+  case Op_LShiftVL: return new (C) LShiftVLNode(n1, n2, vt);
 
-  case Op_RShiftVB: return new (C, 3) RShiftVBNode(n1, n2, vt);
-  case Op_RShiftVS: return new (C, 3) RShiftVSNode(n1, n2, vt);
-  case Op_RShiftVI: return new (C, 3) RShiftVINode(n1, n2, vt);
-  case Op_RShiftVL: return new (C, 3) RShiftVLNode(n1, n2, vt);
+  case Op_RShiftVB: return new (C) RShiftVBNode(n1, n2, vt);
+  case Op_RShiftVS: return new (C) RShiftVSNode(n1, n2, vt);
+  case Op_RShiftVI: return new (C) RShiftVINode(n1, n2, vt);
+  case Op_RShiftVL: return new (C) RShiftVLNode(n1, n2, vt);
 
-  case Op_URShiftVB: return new (C, 3) URShiftVBNode(n1, n2, vt);
-  case Op_URShiftVS: return new (C, 3) URShiftVSNode(n1, n2, vt);
-  case Op_URShiftVI: return new (C, 3) URShiftVINode(n1, n2, vt);
-  case Op_URShiftVL: return new (C, 3) URShiftVLNode(n1, n2, vt);
+  case Op_URShiftVB: return new (C) URShiftVBNode(n1, n2, vt);
+  case Op_URShiftVS: return new (C) URShiftVSNode(n1, n2, vt);
+  case Op_URShiftVI: return new (C) URShiftVINode(n1, n2, vt);
+  case Op_URShiftVL: return new (C) URShiftVLNode(n1, n2, vt);
 
-  case Op_AndV: return new (C, 3) AndVNode(n1, n2, vt);
-  case Op_OrV:  return new (C, 3) OrVNode (n1, n2, vt);
-  case Op_XorV: return new (C, 3) XorVNode(n1, n2, vt);
+  case Op_AndV: return new (C) AndVNode(n1, n2, vt);
+  case Op_OrV:  return new (C) OrVNode (n1, n2, vt);
+  case Op_XorV: return new (C) XorVNode(n1, n2, vt);
   }
   ShouldNotReachHere();
   return NULL;
@@ -299,18 +299,18 @@ VectorNode* VectorNode::scalar2vector(Compile* C, Node* s, uint vlen, const Type
   switch (bt) {
   case T_BOOLEAN:
   case T_BYTE:
-    return new (C, 2) ReplicateBNode(s, vt);
+    return new (C) ReplicateBNode(s, vt);
   case T_CHAR:
   case T_SHORT:
-    return new (C, 2) ReplicateSNode(s, vt);
+    return new (C) ReplicateSNode(s, vt);
   case T_INT:
-    return new (C, 2) ReplicateINode(s, vt);
+    return new (C) ReplicateINode(s, vt);
   case T_LONG:
-    return new (C, 2) ReplicateLNode(s, vt);
+    return new (C) ReplicateLNode(s, vt);
   case T_FLOAT:
-    return new (C, 2) ReplicateFNode(s, vt);
+    return new (C) ReplicateFNode(s, vt);
   case T_DOUBLE:
-    return new (C, 2) ReplicateDNode(s, vt);
+    return new (C) ReplicateDNode(s, vt);
   }
   ShouldNotReachHere();
   return NULL;
@@ -322,18 +322,18 @@ PackNode* PackNode::make(Compile* C, Node* s, uint vlen, BasicType bt) {
   switch (bt) {
   case T_BOOLEAN:
   case T_BYTE:
-    return new (C, 2) PackBNode(s, vt);
+    return new (C) PackBNode(s, vt);
   case T_CHAR:
   case T_SHORT:
-    return new (C, 2) PackSNode(s, vt);
+    return new (C) PackSNode(s, vt);
   case T_INT:
-    return new (C, 2) PackINode(s, vt);
+    return new (C) PackINode(s, vt);
   case T_LONG:
-    return new (C, 2) PackLNode(s, vt);
+    return new (C) PackLNode(s, vt);
   case T_FLOAT:
-    return new (C, 2) PackFNode(s, vt);
+    return new (C) PackFNode(s, vt);
   case T_DOUBLE:
-    return new (C, 2) PackDNode(s, vt);
+    return new (C) PackDNode(s, vt);
   }
   ShouldNotReachHere();
   return NULL;
@@ -358,18 +358,18 @@ PackNode* PackNode::binary_tree_pack(Compile* C, int lo, int hi) {
     switch (bt) {
     case T_BOOLEAN:
     case T_BYTE:
-      return new (C, 3) PackSNode(n1, n2, TypeVect::make(T_SHORT, 2));
+      return new (C) PackSNode(n1, n2, TypeVect::make(T_SHORT, 2));
     case T_CHAR:
     case T_SHORT:
-      return new (C, 3) PackINode(n1, n2, TypeVect::make(T_INT, 2));
+      return new (C) PackINode(n1, n2, TypeVect::make(T_INT, 2));
     case T_INT:
-      return new (C, 3) PackLNode(n1, n2, TypeVect::make(T_LONG, 2));
+      return new (C) PackLNode(n1, n2, TypeVect::make(T_LONG, 2));
     case T_LONG:
-      return new (C, 3) Pack2LNode(n1, n2, TypeVect::make(T_LONG, 2));
+      return new (C) Pack2LNode(n1, n2, TypeVect::make(T_LONG, 2));
     case T_FLOAT:
-      return new (C, 3) PackDNode(n1, n2, TypeVect::make(T_DOUBLE, 2));
+      return new (C) PackDNode(n1, n2, TypeVect::make(T_DOUBLE, 2));
     case T_DOUBLE:
-      return new (C, 3) Pack2DNode(n1, n2, TypeVect::make(T_DOUBLE, 2));
+      return new (C) Pack2DNode(n1, n2, TypeVect::make(T_DOUBLE, 2));
     }
     ShouldNotReachHere();
   }
@@ -380,7 +380,7 @@ PackNode* PackNode::binary_tree_pack(Compile* C, int lo, int hi) {
 LoadVectorNode* LoadVectorNode::make(Compile* C, int opc, Node* ctl, Node* mem,
                                      Node* adr, const TypePtr* atyp, uint vlen, BasicType bt) {
   const TypeVect* vt = TypeVect::make(bt, vlen);
-  return new (C, 3) LoadVectorNode(ctl, mem, adr, atyp, vt);
+  return new (C) LoadVectorNode(ctl, mem, adr, atyp, vt);
   return NULL;
 }
 
@@ -388,7 +388,7 @@ LoadVectorNode* LoadVectorNode::make(Compile* C, int opc, Node* ctl, Node* mem,
 StoreVectorNode* StoreVectorNode::make(Compile* C, int opc, Node* ctl, Node* mem,
                                        Node* adr, const TypePtr* atyp, Node* val,
                                        uint vlen) {
-  return new (C, 4) StoreVectorNode(ctl, mem, adr, atyp, val);
+  return new (C) StoreVectorNode(ctl, mem, adr, atyp, val);
 }
 
 // Extract a scalar element of vector.
@@ -397,21 +397,21 @@ Node* ExtractNode::make(Compile* C, Node* v, uint position, BasicType bt) {
   ConINode* pos = ConINode::make(C, (int)position);
   switch (bt) {
   case T_BOOLEAN:
-    return new (C, 3) ExtractUBNode(v, pos);
+    return new (C) ExtractUBNode(v, pos);
   case T_BYTE:
-    return new (C, 3) ExtractBNode(v, pos);
+    return new (C) ExtractBNode(v, pos);
   case T_CHAR:
-    return new (C, 3) ExtractCNode(v, pos);
+    return new (C) ExtractCNode(v, pos);
   case T_SHORT:
-    return new (C, 3) ExtractSNode(v, pos);
+    return new (C) ExtractSNode(v, pos);
   case T_INT:
-    return new (C, 3) ExtractINode(v, pos);
+    return new (C) ExtractINode(v, pos);
   case T_LONG:
-    return new (C, 3) ExtractLNode(v, pos);
+    return new (C) ExtractLNode(v, pos);
   case T_FLOAT:
-    return new (C, 3) ExtractFNode(v, pos);
+    return new (C) ExtractFNode(v, pos);
   case T_DOUBLE:
-    return new (C, 3) ExtractDNode(v, pos);
+    return new (C) ExtractDNode(v, pos);
   }
   ShouldNotReachHere();
   return NULL;
diff --git a/src/share/vm/runtime/advancedThresholdPolicy.cpp b/src/share/vm/runtime/advancedThresholdPolicy.cpp
index 9f4135f92..e111ea7b7 100644
--- a/src/share/vm/runtime/advancedThresholdPolicy.cpp
+++ b/src/share/vm/runtime/advancedThresholdPolicy.cpp
@@ -30,12 +30,12 @@
 // Print an event.
 void AdvancedThresholdPolicy::print_specific(EventType type, methodHandle mh, methodHandle imh,
                                              int bci, CompLevel level) {
-  tty->print(" rate: ");
+  tty->print(" rate=");
   if (mh->prev_time() == 0) tty->print("n/a");
   else tty->print("%f", mh->rate());
 
-  tty->print(" k: %.2lf,%.2lf", threshold_scale(CompLevel_full_profile, Tier3LoadFeedback),
-                                threshold_scale(CompLevel_full_optimization, Tier4LoadFeedback));
+  tty->print(" k=%.2lf,%.2lf", threshold_scale(CompLevel_full_profile, Tier3LoadFeedback),
+                               threshold_scale(CompLevel_full_optimization, Tier4LoadFeedback));
 
 }
 
diff --git a/src/share/vm/runtime/compilationPolicy.cpp b/src/share/vm/runtime/compilationPolicy.cpp
index 8689c8c1b..940978cf7 100644
--- a/src/share/vm/runtime/compilationPolicy.cpp
+++ b/src/share/vm/runtime/compilationPolicy.cpp
@@ -394,28 +394,27 @@ void NonTieredCompPolicy::trace_osr_request(methodHandle method, nmethod* osr, i
 // SimpleCompPolicy - compile current method
 
 void SimpleCompPolicy::method_invocation_event(methodHandle m, JavaThread* thread) {
-  int hot_count = m->invocation_count();
+  const int comp_level = CompLevel_highest_tier;
+  const int hot_count = m->invocation_count();
   reset_counter_for_invocation_event(m);
   const char* comment = "count";
 
   if (is_compilation_enabled() && can_be_compiled(m)) {
     nmethod* nm = m->code();
     if (nm == NULL ) {
-      const char* comment = "count";
-      CompileBroker::compile_method(m, InvocationEntryBci, CompLevel_highest_tier,
-                                    m, hot_count, comment, thread);
+      CompileBroker::compile_method(m, InvocationEntryBci, comp_level, m, hot_count, comment, thread);
     }
   }
 }
 
 void SimpleCompPolicy::method_back_branch_event(methodHandle m, int bci, JavaThread* thread) {
-  int hot_count = m->backedge_count();
+  const int comp_level = CompLevel_highest_tier;
+  const int hot_count = m->backedge_count();
   const char* comment = "backedge_count";
 
-  if (is_compilation_enabled() && !m->is_not_osr_compilable() && can_be_compiled(m)) {
-    CompileBroker::compile_method(m, bci, CompLevel_highest_tier,
-                                  m, hot_count, comment, thread);
-    NOT_PRODUCT(trace_osr_completion(m->lookup_osr_nmethod_for(bci, CompLevel_highest_tier, true));)
+  if (is_compilation_enabled() && !m->is_not_osr_compilable(comp_level) && can_be_compiled(m)) {
+    CompileBroker::compile_method(m, bci, comp_level, m, hot_count, comment, thread);
+    NOT_PRODUCT(trace_osr_completion(m->lookup_osr_nmethod_for(bci, comp_level, true));)
   }
 }
 // StackWalkCompPolicy - walk up stack to find a suitable method to compile
@@ -426,7 +425,8 @@ const char* StackWalkCompPolicy::_msg = NULL;
 
 // Consider m for compilation
 void StackWalkCompPolicy::method_invocation_event(methodHandle m, JavaThread* thread) {
-  int hot_count = m->invocation_count();
+  const int comp_level = CompLevel_highest_tier;
+  const int hot_count = m->invocation_count();
   reset_counter_for_invocation_event(m);
   const char* comment = "count";
 
@@ -457,20 +457,20 @@ void StackWalkCompPolicy::method_invocation_event(methodHandle m, JavaThread* th
       if (TimeCompilationPolicy) accumulated_time()->stop();
       assert(top != NULL, "findTopInlinableFrame returned null");
       if (TraceCompilationPolicy) top->print();
-      CompileBroker::compile_method(top->top_method(), InvocationEntryBci, CompLevel_highest_tier,
+      CompileBroker::compile_method(top->top_method(), InvocationEntryBci, comp_level,
                                     m, hot_count, comment, thread);
     }
   }
 }
 
 void StackWalkCompPolicy::method_back_branch_event(methodHandle m, int bci, JavaThread* thread) {
-  int hot_count = m->backedge_count();
+  const int comp_level = CompLevel_highest_tier;
+  const int hot_count = m->backedge_count();
   const char* comment = "backedge_count";
 
-  if (is_compilation_enabled() && !m->is_not_osr_compilable() && can_be_compiled(m)) {
-    CompileBroker::compile_method(m, bci, CompLevel_highest_tier, m, hot_count, comment, thread);
-
-    NOT_PRODUCT(trace_osr_completion(m->lookup_osr_nmethod_for(bci, CompLevel_highest_tier, true));)
+  if (is_compilation_enabled() && !m->is_not_osr_compilable(comp_level) && can_be_compiled(m)) {
+    CompileBroker::compile_method(m, bci, comp_level, m, hot_count, comment, thread);
+    NOT_PRODUCT(trace_osr_completion(m->lookup_osr_nmethod_for(bci, comp_level, true));)
   }
 }
 
diff --git a/src/share/vm/runtime/sharedRuntime.cpp b/src/share/vm/runtime/sharedRuntime.cpp
index d730251b7..a0ef0abc2 100644
--- a/src/share/vm/runtime/sharedRuntime.cpp
+++ b/src/share/vm/runtime/sharedRuntime.cpp
@@ -88,6 +88,7 @@ RuntimeStub*        SharedRuntime::_resolve_virtual_call_blob;
 RuntimeStub*        SharedRuntime::_resolve_static_call_blob;
 
 DeoptimizationBlob* SharedRuntime::_deopt_blob;
+SafepointBlob*      SharedRuntime::_polling_page_vectors_safepoint_handler_blob;
 SafepointBlob*      SharedRuntime::_polling_page_safepoint_handler_blob;
 SafepointBlob*      SharedRuntime::_polling_page_return_handler_blob;
 
@@ -104,8 +105,14 @@ void SharedRuntime::generate_stubs() {
   _resolve_virtual_call_blob           = generate_resolve_blob(CAST_FROM_FN_PTR(address, SharedRuntime::resolve_virtual_call_C),      "resolve_virtual_call");
   _resolve_static_call_blob            = generate_resolve_blob(CAST_FROM_FN_PTR(address, SharedRuntime::resolve_static_call_C),       "resolve_static_call");
 
-  _polling_page_safepoint_handler_blob = generate_handler_blob(CAST_FROM_FN_PTR(address, SafepointSynchronize::handle_polling_page_exception), false);
-  _polling_page_return_handler_blob    = generate_handler_blob(CAST_FROM_FN_PTR(address, SafepointSynchronize::handle_polling_page_exception), true);
+#ifdef COMPILER2
+  // Vectors are generated only by C2.
+  if (is_wide_vector(MaxVectorSize)) {
+    _polling_page_vectors_safepoint_handler_blob = generate_handler_blob(CAST_FROM_FN_PTR(address, SafepointSynchronize::handle_polling_page_exception), POLL_AT_VECTOR_LOOP);
+  }
+#endif // COMPILER2
+  _polling_page_safepoint_handler_blob = generate_handler_blob(CAST_FROM_FN_PTR(address, SafepointSynchronize::handle_polling_page_exception), POLL_AT_LOOP);
+  _polling_page_return_handler_blob    = generate_handler_blob(CAST_FROM_FN_PTR(address, SafepointSynchronize::handle_polling_page_exception), POLL_AT_RETURN);
 
   generate_deopt_blob();
 
@@ -535,10 +542,15 @@ address SharedRuntime::get_poll_stub(address pc) {
     "Only polling locations are used for safepoint");
 
   bool at_poll_return = ((nmethod*)cb)->is_at_poll_return(pc);
+  bool has_wide_vectors = ((nmethod*)cb)->has_wide_vectors();
   if (at_poll_return) {
     assert(SharedRuntime::polling_page_return_handler_blob() != NULL,
            "polling page return stub not created yet");
     stub = SharedRuntime::polling_page_return_handler_blob()->entry_point();
+  } else if (has_wide_vectors) {
+    assert(SharedRuntime::polling_page_vectors_safepoint_handler_blob() != NULL,
+           "polling page vectors safepoint stub not created yet");
+    stub = SharedRuntime::polling_page_vectors_safepoint_handler_blob()->entry_point();
   } else {
     assert(SharedRuntime::polling_page_safepoint_handler_blob() != NULL,
            "polling page safepoint stub not created yet");
@@ -1618,6 +1630,31 @@ methodHandle SharedRuntime::reresolve_call_site(JavaThread *thread, TRAPS) {
   return callee_method;
 }
 
+#ifdef ASSERT
+void SharedRuntime::check_member_name_argument_is_last_argument(methodHandle method,
+                                                                const BasicType* sig_bt,
+                                                                const VMRegPair* regs) {
+  ResourceMark rm;
+  const int total_args_passed = method->size_of_parameters();
+  const VMRegPair*    regs_with_member_name = regs;
+        VMRegPair* regs_without_member_name = NEW_RESOURCE_ARRAY(VMRegPair, total_args_passed - 1);
+
+  const int member_arg_pos = total_args_passed - 1;
+  assert(member_arg_pos >= 0 && member_arg_pos < total_args_passed, "oob");
+  assert(sig_bt[member_arg_pos] == T_OBJECT, "dispatch argument must be an object");
+
+  const bool is_outgoing = method->is_method_handle_intrinsic();
+  int comp_args_on_stack = java_calling_convention(sig_bt, regs_without_member_name, total_args_passed - 1, is_outgoing);
+
+  for (int i = 0; i < member_arg_pos; i++) {
+    VMReg a =    regs_with_member_name[i].first();
+    VMReg b = regs_without_member_name[i].first();
+    assert(a->value() == b->value(), err_msg_res("register allocation mismatch: a=%d, b=%d", a->value(), b->value()));
+  }
+  assert(regs_with_member_name[member_arg_pos].first()->is_valid(), "bad member arg");
+}
+#endif
+
 // ---------------------------------------------------------------------------
 // We are calling the interpreter via a c2i. Normally this would mean that
 // we were called by a compiled method. However we could have lost a race
@@ -2423,6 +2460,7 @@ AdapterHandlerEntry* AdapterHandlerLibrary::get_adapter(methodHandle method) {
 #ifndef PRODUCT
     // debugging suppport
     if (PrintAdapterHandlers || PrintStubCode) {
+      ttyLocker ttyl;
       entry->print_adapter_on(tty);
       tty->print_cr("i2c argument handler #%d for: %s %s (%d bytes generated)",
                     _adapters->number_of_entries(), (method->is_static() ? "static" : "receiver"),
@@ -2430,8 +2468,10 @@ AdapterHandlerEntry* AdapterHandlerLibrary::get_adapter(methodHandle method) {
       tty->print_cr("c2i argument handler starts at %p",entry->get_c2i_entry());
       if (Verbose || PrintStubCode) {
         address first_pc = entry->base_address();
-        if (first_pc != NULL)
+        if (first_pc != NULL) {
           Disassembler::decode(first_pc, first_pc + insts_size);
+          tty->cr();
+        }
       }
     }
 #endif
@@ -2546,10 +2586,10 @@ nmethod *AdapterHandlerLibrary::create_native_wrapper(methodHandle method, int c
       MacroAssembler _masm(&buffer);
 
       // Fill in the signature array, for the calling-convention call.
-      int total_args_passed = method->size_of_parameters();
+      const int total_args_passed = method->size_of_parameters();
 
-      BasicType* sig_bt = NEW_RESOURCE_ARRAY(BasicType,total_args_passed);
-      VMRegPair*   regs = NEW_RESOURCE_ARRAY(VMRegPair,total_args_passed);
+      BasicType* sig_bt = NEW_RESOURCE_ARRAY(BasicType, total_args_passed);
+      VMRegPair*   regs = NEW_RESOURCE_ARRAY(VMRegPair, total_args_passed);
       int i=0;
       if( !method->is_static() )  // Pass in receiver first
         sig_bt[i++] = T_OBJECT;
@@ -2559,7 +2599,7 @@ nmethod *AdapterHandlerLibrary::create_native_wrapper(methodHandle method, int c
         if( ss.type() == T_LONG || ss.type() == T_DOUBLE )
           sig_bt[i++] = T_VOID;   // Longs & doubles take 2 Java slots
       }
-      assert( i==total_args_passed, "" );
+      assert(i == total_args_passed, "");
       BasicType ret_type = ss.type();
 
       // Now get the compiled-Java layout as input (or output) arguments.
@@ -2572,9 +2612,8 @@ nmethod *AdapterHandlerLibrary::create_native_wrapper(methodHandle method, int c
       nm = SharedRuntime::generate_native_wrapper(&_masm,
                                                   method,
                                                   compile_id,
-                                                  total_args_passed,
-                                                  comp_args_on_stack,
-                                                  sig_bt,regs,
+                                                  sig_bt,
+                                                  regs,
                                                   ret_type);
     }
   }
diff --git a/src/share/vm/runtime/sharedRuntime.hpp b/src/share/vm/runtime/sharedRuntime.hpp
index 014ec503c..54448890e 100644
--- a/src/share/vm/runtime/sharedRuntime.hpp
+++ b/src/share/vm/runtime/sharedRuntime.hpp
@@ -62,6 +62,7 @@ class SharedRuntime: AllStatic {
 
   static DeoptimizationBlob* _deopt_blob;
 
+  static SafepointBlob*      _polling_page_vectors_safepoint_handler_blob;
   static SafepointBlob*      _polling_page_safepoint_handler_blob;
   static SafepointBlob*      _polling_page_return_handler_blob;
 
@@ -75,7 +76,8 @@ class SharedRuntime: AllStatic {
 #endif // !PRODUCT
 
  private:
-  static SafepointBlob* generate_handler_blob(address call_ptr, bool cause_return);
+  enum { POLL_AT_RETURN,  POLL_AT_LOOP, POLL_AT_VECTOR_LOOP };
+  static SafepointBlob* generate_handler_blob(address call_ptr, int poll_type);
   static RuntimeStub*   generate_resolve_blob(address destination, const char* name);
 
  public:
@@ -223,6 +225,7 @@ class SharedRuntime: AllStatic {
 
   static SafepointBlob* polling_page_return_handler_blob()     { return _polling_page_return_handler_blob; }
   static SafepointBlob* polling_page_safepoint_handler_blob()  { return _polling_page_safepoint_handler_blob; }
+  static SafepointBlob* polling_page_vectors_safepoint_handler_blob()  { return _polling_page_vectors_safepoint_handler_blob; }
 
   // Counters
 #ifndef PRODUCT
@@ -345,7 +348,11 @@ class SharedRuntime: AllStatic {
   // the bottom of the frame the first 16 words will be skipped and SharedInfo::stack0
   // will be just above it. (
   // return value is the maximum number of VMReg stack slots the convention will use.
-  static int java_calling_convention(const BasicType *sig_bt, VMRegPair *regs, int total_args_passed, int is_outgoing);
+  static int java_calling_convention(const BasicType* sig_bt, VMRegPair* regs, int total_args_passed, int is_outgoing);
+
+  static void check_member_name_argument_is_last_argument(methodHandle method,
+                                                          const BasicType* sig_bt,
+                                                          const VMRegPair* regs) NOT_DEBUG_RETURN;
 
   // Ditto except for calling C
   static int c_calling_convention(const BasicType *sig_bt, VMRegPair *regs, int total_args_passed);
@@ -412,6 +419,10 @@ class SharedRuntime: AllStatic {
   // when an interrupt occurs.
   static uint out_preserve_stack_slots();
 
+  // Is vector's size (in bytes) bigger than a size saved by default?
+  // For example, on x86 16 bytes XMM registers are saved by default.
+  static bool is_wide_vector(int size);
+
   // Save and restore a native result
   static void    save_native_result(MacroAssembler *_masm, BasicType ret_type, int frame_slots );
   static void restore_native_result(MacroAssembler *_masm, BasicType ret_type, int frame_slots );
@@ -425,13 +436,11 @@ class SharedRuntime: AllStatic {
   // The wrapper may contain special-case code if the given method
   // is a JNI critical method, or a compiled method handle adapter,
   // such as _invokeBasic, _linkToVirtual, etc.
-  static nmethod *generate_native_wrapper(MacroAssembler* masm,
+  static nmethod* generate_native_wrapper(MacroAssembler* masm,
                                           methodHandle method,
                                           int compile_id,
-                                          int total_args_passed,
-                                          int max_arg,
-                                          BasicType *sig_bt,
-                                          VMRegPair *regs,
+                                          BasicType* sig_bt,
+                                          VMRegPair* regs,
                                           BasicType ret_type );
 
   // Block before entering a JNI critical method
diff --git a/src/share/vm/runtime/simpleThresholdPolicy.cpp b/src/share/vm/runtime/simpleThresholdPolicy.cpp
index ecf3d01df..fbb0f89c3 100644
--- a/src/share/vm/runtime/simpleThresholdPolicy.cpp
+++ b/src/share/vm/runtime/simpleThresholdPolicy.cpp
@@ -43,11 +43,11 @@ void SimpleThresholdPolicy::print_counters(const char* prefix, methodHandle mh)
     mdo_invocations_start = mdh->invocation_count_start();
     mdo_backedges_start = mdh->backedge_count_start();
   }
-  tty->print(" %stotal: %d,%d %smdo: %d(%d),%d(%d)", prefix,
+  tty->print(" %stotal=%d,%d %smdo=%d(%d),%d(%d)", prefix,
       invocation_count, backedge_count, prefix,
       mdo_invocations, mdo_invocations_start,
       mdo_backedges, mdo_backedges_start);
-  tty->print(" %smax levels: %d,%d", prefix,
+  tty->print(" %smax levels=%d,%d", prefix,
       mh->highest_comp_level(), mh->highest_osr_comp_level());
 }
 
@@ -85,7 +85,7 @@ void SimpleThresholdPolicy::print_event(EventType type, methodHandle mh, methodH
     tty->print("unknown");
   }
 
-  tty->print(" level: %d ", level);
+  tty->print(" level=%d ", level);
 
   ResourceMark rm;
   char *method_name = mh->name_and_sig_as_C_string();
@@ -95,8 +95,8 @@ void SimpleThresholdPolicy::print_event(EventType type, methodHandle mh, methodH
     tty->print(" [%s]] ", inlinee_name);
   }
   else tty->print("] ");
-  tty->print("@%d queues: %d,%d", bci, CompileBroker::queue_size(CompLevel_full_profile),
-                                       CompileBroker::queue_size(CompLevel_full_optimization));
+  tty->print("@%d queues=%d,%d", bci, CompileBroker::queue_size(CompLevel_full_profile),
+                                      CompileBroker::queue_size(CompLevel_full_optimization));
 
   print_specific(type, mh, imh, bci, level);
 
@@ -105,25 +105,30 @@ void SimpleThresholdPolicy::print_event(EventType type, methodHandle mh, methodH
     if (inlinee_event) {
       print_counters("inlinee ", imh);
     }
-    tty->print(" compilable: ");
+    tty->print(" compilable=");
     bool need_comma = false;
     if (!mh->is_not_compilable(CompLevel_full_profile)) {
       tty->print("c1");
       need_comma = true;
     }
+    if (!mh->is_not_osr_compilable(CompLevel_full_profile)) {
+      if (need_comma) tty->print(",");
+      tty->print("c1-osr");
+      need_comma = true;
+    }
     if (!mh->is_not_compilable(CompLevel_full_optimization)) {
-      if (need_comma) tty->print(", ");
+      if (need_comma) tty->print(",");
       tty->print("c2");
       need_comma = true;
     }
-    if (!mh->is_not_osr_compilable()) {
-      if (need_comma) tty->print(", ");
-      tty->print("osr");
+    if (!mh->is_not_osr_compilable(CompLevel_full_optimization)) {
+      if (need_comma) tty->print(",");
+      tty->print("c2-osr");
     }
-    tty->print(" status:");
+    tty->print(" status=");
     if (mh->queued_for_compilation()) {
-      tty->print(" in queue");
-    } else tty->print(" idle");
+      tty->print("in-queue");
+    } else tty->print("idle");
   }
   tty->print_cr("]");
 }
@@ -223,7 +228,7 @@ void SimpleThresholdPolicy::compile(methodHandle mh, int bci, CompLevel level, J
     }
     return;
   }
-  if (bci != InvocationEntryBci && mh->is_not_osr_compilable()) {
+  if (bci != InvocationEntryBci && mh->is_not_osr_compilable(level)) {
     return;
   }
   if (!CompileBroker::compilation_is_in_queue(mh, bci)) {
diff --git a/src/share/vm/runtime/vmStructs.cpp b/src/share/vm/runtime/vmStructs.cpp
index 2866e57b4..1f0f2becc 100644
--- a/src/share/vm/runtime/vmStructs.cpp
+++ b/src/share/vm/runtime/vmStructs.cpp
@@ -1905,6 +1905,8 @@ typedef TwoOopHashtable<Symbol*, mtClass>     SymbolTwoOopHashtable;
   declare_c2_type(SubVLNode, VectorNode)                                  \
   declare_c2_type(SubVFNode, VectorNode)                                  \
   declare_c2_type(SubVDNode, VectorNode)                                  \
+  declare_c2_type(MulVSNode, VectorNode)                                  \
+  declare_c2_type(MulVINode, VectorNode)                                  \
   declare_c2_type(MulVFNode, VectorNode)                                  \
   declare_c2_type(MulVDNode, VectorNode)                                  \
   declare_c2_type(DivVFNode, VectorNode)                                  \
@@ -1912,9 +1914,15 @@ typedef TwoOopHashtable<Symbol*, mtClass>     SymbolTwoOopHashtable;
   declare_c2_type(LShiftVBNode, VectorNode)                               \
   declare_c2_type(LShiftVSNode, VectorNode)                               \
   declare_c2_type(LShiftVINode, VectorNode)                               \
+  declare_c2_type(LShiftVLNode, VectorNode)                               \
   declare_c2_type(RShiftVBNode, VectorNode)                               \
   declare_c2_type(RShiftVSNode, VectorNode)                               \
   declare_c2_type(RShiftVINode, VectorNode)                               \
+  declare_c2_type(RShiftVLNode, VectorNode)                               \
+  declare_c2_type(URShiftVBNode, VectorNode)                              \
+  declare_c2_type(URShiftVSNode, VectorNode)                              \
+  declare_c2_type(URShiftVINode, VectorNode)                              \
+  declare_c2_type(URShiftVLNode, VectorNode)                              \
   declare_c2_type(AndVNode, VectorNode)                                   \
   declare_c2_type(OrVNode, VectorNode)                                    \
   declare_c2_type(XorVNode, VectorNode)                                   \
@@ -1937,6 +1945,8 @@ typedef TwoOopHashtable<Symbol*, mtClass>     SymbolTwoOopHashtable;
   declare_c2_type(Pack2DNode, PackNode)                                   \
   declare_c2_type(ExtractNode, Node)                                      \
   declare_c2_type(ExtractBNode, ExtractNode)                              \
+  declare_c2_type(ExtractUBNode, ExtractNode)                             \
+  declare_c2_type(ExtractCNode, ExtractNode)                              \
   declare_c2_type(ExtractSNode, ExtractNode)                              \
   declare_c2_type(ExtractINode, ExtractNode)                              \
   declare_c2_type(ExtractLNode, ExtractNode)                              \
@@ -2182,7 +2192,7 @@ typedef TwoOopHashtable<Symbol*, mtClass>     SymbolTwoOopHashtable;
   declare_constant(JVM_ACC_HAS_LOOPS)                                     \
   declare_constant(JVM_ACC_LOOPS_FLAG_INIT)                               \
   declare_constant(JVM_ACC_QUEUED)                                        \
-  declare_constant(JVM_ACC_NOT_OSR_COMPILABLE)                            \
+  declare_constant(JVM_ACC_NOT_C2_OSR_COMPILABLE)                         \
   declare_constant(JVM_ACC_HAS_LINE_NUMBER_TABLE)                         \
   declare_constant(JVM_ACC_HAS_CHECKED_EXCEPTIONS)                        \
   declare_constant(JVM_ACC_HAS_JSRS)                                      \
diff --git a/src/share/vm/runtime/vm_version.cpp b/src/share/vm/runtime/vm_version.cpp
index 2d51b6781..7bcd1d7d3 100644
--- a/src/share/vm/runtime/vm_version.cpp
+++ b/src/share/vm/runtime/vm_version.cpp
@@ -45,6 +45,10 @@
 const char* Abstract_VM_Version::_s_vm_release = Abstract_VM_Version::vm_release();
 const char* Abstract_VM_Version::_s_internal_vm_info_string = Abstract_VM_Version::internal_vm_info_string();
 bool Abstract_VM_Version::_supports_cx8 = false;
+bool Abstract_VM_Version::_supports_atomic_getset4 = false;
+bool Abstract_VM_Version::_supports_atomic_getset8 = false;
+bool Abstract_VM_Version::_supports_atomic_getadd4 = false;
+bool Abstract_VM_Version::_supports_atomic_getadd8 = false;
 unsigned int Abstract_VM_Version::_logical_processors_per_package = 1U;
 int Abstract_VM_Version::_reserve_for_allocation_prefetch = 0;
 
diff --git a/src/share/vm/runtime/vm_version.hpp b/src/share/vm/runtime/vm_version.hpp
index 0d35522b6..6506103b2 100644
--- a/src/share/vm/runtime/vm_version.hpp
+++ b/src/share/vm/runtime/vm_version.hpp
@@ -37,6 +37,10 @@ class Abstract_VM_Version: AllStatic {
   static const char*  _s_internal_vm_info_string;
   // These are set by machine-dependent initializations
   static bool         _supports_cx8;
+  static bool         _supports_atomic_getset4;
+  static bool         _supports_atomic_getset8;
+  static bool         _supports_atomic_getadd4;
+  static bool         _supports_atomic_getadd8;
   static unsigned int _logical_processors_per_package;
   static int          _vm_major_version;
   static int          _vm_minor_version;
@@ -75,6 +79,13 @@ class Abstract_VM_Version: AllStatic {
 
   // does HW support an 8-byte compare-exchange operation?
   static bool supports_cx8()  {return _supports_cx8;}
+  // does HW support atomic get-and-set or atomic get-and-add?  Used
+  // to guide intrinsification decisions for Unsafe atomic ops
+  static bool supports_atomic_getset4()  {return _supports_atomic_getset4;}
+  static bool supports_atomic_getset8()  {return _supports_atomic_getset8;}
+  static bool supports_atomic_getadd4()  {return _supports_atomic_getadd4;}
+  static bool supports_atomic_getadd8()  {return _supports_atomic_getadd8;}
+
   static unsigned int logical_processors_per_package() {
     return _logical_processors_per_package;
   }
diff --git a/src/share/vm/utilities/accessFlags.hpp b/src/share/vm/utilities/accessFlags.hpp
index 13f38c375..c2729eb1b 100644
--- a/src/share/vm/utilities/accessFlags.hpp
+++ b/src/share/vm/utilities/accessFlags.hpp
@@ -47,7 +47,7 @@ enum {
   JVM_ACC_QUEUED                  = 0x01000000,     // Queued for compilation
   JVM_ACC_NOT_C2_COMPILABLE       = 0x02000000,
   JVM_ACC_NOT_C1_COMPILABLE       = 0x04000000,
-  JVM_ACC_NOT_OSR_COMPILABLE      = 0x08000000,
+  JVM_ACC_NOT_C2_OSR_COMPILABLE   = 0x08000000,
   JVM_ACC_HAS_LINE_NUMBER_TABLE   = 0x00100000,
   JVM_ACC_HAS_CHECKED_EXCEPTIONS  = 0x00400000,
   JVM_ACC_HAS_JSRS                = 0x00800000,
@@ -121,9 +121,9 @@ class AccessFlags VALUE_OBJ_CLASS_SPEC {
   bool has_loops               () const { return (_flags & JVM_ACC_HAS_LOOPS              ) != 0; }
   bool loops_flag_init         () const { return (_flags & JVM_ACC_LOOPS_FLAG_INIT        ) != 0; }
   bool queued_for_compilation  () const { return (_flags & JVM_ACC_QUEUED                 ) != 0; }
-  bool is_not_c1_compilable () const    { return (_flags & JVM_ACC_NOT_C1_COMPILABLE      ) != 0; }
-  bool is_not_c2_compilable () const    { return (_flags & JVM_ACC_NOT_C2_COMPILABLE      ) != 0; }
-  bool is_not_osr_compilable   () const { return (_flags & JVM_ACC_NOT_OSR_COMPILABLE     ) != 0; }
+  bool is_not_c1_compilable    () const { return (_flags & JVM_ACC_NOT_C1_COMPILABLE      ) != 0; }
+  bool is_not_c2_compilable    () const { return (_flags & JVM_ACC_NOT_C2_COMPILABLE      ) != 0; }
+  bool is_not_c2_osr_compilable() const { return (_flags & JVM_ACC_NOT_C2_OSR_COMPILABLE  ) != 0; }
   bool has_linenumber_table    () const { return (_flags & JVM_ACC_HAS_LINE_NUMBER_TABLE  ) != 0; }
   bool has_checked_exceptions  () const { return (_flags & JVM_ACC_HAS_CHECKED_EXCEPTIONS ) != 0; }
   bool has_jsrs                () const { return (_flags & JVM_ACC_HAS_JSRS               ) != 0; }
@@ -186,7 +186,7 @@ class AccessFlags VALUE_OBJ_CLASS_SPEC {
   void set_loops_flag_init()           { atomic_set_bits(JVM_ACC_LOOPS_FLAG_INIT);         }
   void set_not_c1_compilable()         { atomic_set_bits(JVM_ACC_NOT_C1_COMPILABLE);       }
   void set_not_c2_compilable()         { atomic_set_bits(JVM_ACC_NOT_C2_COMPILABLE);       }
-  void set_not_osr_compilable()        { atomic_set_bits(JVM_ACC_NOT_OSR_COMPILABLE);      }
+  void set_not_c2_osr_compilable()     { atomic_set_bits(JVM_ACC_NOT_C2_OSR_COMPILABLE);   }
   void set_has_linenumber_table()      { atomic_set_bits(JVM_ACC_HAS_LINE_NUMBER_TABLE);   }
   void set_has_checked_exceptions()    { atomic_set_bits(JVM_ACC_HAS_CHECKED_EXCEPTIONS);  }
   void set_has_jsrs()                  { atomic_set_bits(JVM_ACC_HAS_JSRS);                }
diff --git a/test/compiler/7196199/Test7196199.java b/test/compiler/7196199/Test7196199.java
new file mode 100644
index 000000000..6aa35369a
--- /dev/null
+++ b/test/compiler/7196199/Test7196199.java
@@ -0,0 +1,189 @@
+/*
+ * Copyright (c) 2012, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+/**
+ * @test
+ * @bug 7196199
+ * @summary java/text/Bidi/Bug6665028.java failed: Bidi run count incorrect
+ *
+ * @run main/othervm/timeout=400 -Xmx32m -Xbatch -XX:+IgnoreUnrecognizedVMOptions -XX:-TieredCompilation -XX:CompileCommand=exclude,Test7196199.test -XX:+SafepointALot -XX:GuaranteedSafepointInterval=100 Test7196199
+ */
+
+
+public class Test7196199 {
+  private static final int ARRLEN = 97;
+  private static final int ITERS  = 5000;
+  private static final int INI_ITERS  = 1000;
+  private static final int SFP_ITERS  = 10000;
+  private static final float SFP_ITERS_F  = 10000.f;
+  private static final float VALUE = 15.f;
+  public static void main(String args[]) {
+    int errn = test();
+    if (errn > 0) {
+      System.err.println("FAILED: " + errn + " errors");
+      System.exit(97);
+    }
+    System.out.println("PASSED");
+  }
+
+  static int test() {
+    float[] a0 = new float[ARRLEN];
+    float[] a1 = new float[ARRLEN];
+    // Initialize
+    for (int i=0; i<ARRLEN; i++) {
+      a0[i] = 0.f;
+      a1[i] = (float)i;
+    }
+    System.out.println("Warmup");
+    for (int i=0; i<INI_ITERS; i++) {
+      test_incrc(a0);
+      test_incrv(a0, VALUE);
+      test_addc(a0, a1);
+      test_addv(a0, a1, VALUE);
+    }
+    // Test and verify results
+    System.out.println("Verification");
+    int errn = 0;
+    for (int i=0; i<ARRLEN; i++)
+      a0[i] = 0.f;
+
+    System.out.println("  test_incrc");
+    for (int j=0; j<ITERS; j++) {
+      test_incrc(a0);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_incrc: ", i, a0[i], VALUE*SFP_ITERS_F);
+        a0[i] = 0.f; // Reset
+      }
+    }
+
+    System.out.println("  test_incrv");
+    for (int j=0; j<ITERS; j++) {
+      test_incrv(a0, VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_incrv: ", i, a0[i], VALUE*SFP_ITERS_F);
+        a0[i] = 0.f; // Reset
+      }
+    }
+
+    System.out.println("  test_addc");
+    for (int j=0; j<ITERS; j++) {
+      test_addc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_addc: ", i, a0[i], ((float)i + VALUE)*SFP_ITERS_F);
+        a0[i] = 0.f; // Reset
+      }
+    }
+
+    System.out.println("  test_addv");
+    for (int j=0; j<ITERS; j++) {
+      test_addv(a0, a1, VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_addv: ", i, a0[i], ((float)i + VALUE)*SFP_ITERS_F);
+        a0[i] = 0.f; // Reset
+      }
+    }
+
+    if (errn > 0)
+      return errn;
+
+    System.out.println("Time");
+    long start, end;
+
+    start = System.currentTimeMillis();
+    for (int i=0; i<INI_ITERS; i++) {
+      test_incrc(a0);
+    }
+    end = System.currentTimeMillis();
+    System.out.println("test_incrc: " + (end - start));
+
+    start = System.currentTimeMillis();
+    for (int i=0; i<INI_ITERS; i++) {
+      test_incrv(a0, VALUE);
+    }
+    end = System.currentTimeMillis();
+    System.out.println("test_incrv: " + (end - start));
+
+    start = System.currentTimeMillis();
+    for (int i=0; i<INI_ITERS; i++) {
+      test_addc(a0, a1);
+    }
+    end = System.currentTimeMillis();
+    System.out.println("test_addc: " + (end - start));
+
+    start = System.currentTimeMillis();
+    for (int i=0; i<INI_ITERS; i++) {
+      test_addv(a0, a1, VALUE);
+    }
+    end = System.currentTimeMillis();
+    System.out.println("test_addv: " + (end - start));
+
+    return errn;
+  }
+
+  static void test_incrc(float[] a0) {
+    // Non-counted loop with safepoint.
+    for (long l = 0; l < SFP_ITERS; l++) {
+      // Counted and vectorized loop.
+      for (int i = 0; i < a0.length; i+=1) {
+        a0[i] += VALUE;
+      }
+    }
+  }
+  static void test_incrv(float[] a0, float b) {
+    // Non-counted loop with safepoint.
+    for (long l = 0; l < SFP_ITERS; l++) {
+      // Counted and vectorized loop.
+      for (int i = 0; i < a0.length; i+=1) {
+        a0[i] += b;
+      }
+    }
+  }
+  static void test_addc(float[] a0, float[] a1) {
+    // Non-counted loop with safepoint.
+    for (long l = 0; l < SFP_ITERS; l++) {
+      // Counted and vectorized loop.
+      for (int i = 0; i < a0.length; i+=1) {
+        a0[i] += a1[i]+VALUE;
+      }
+    }
+  }
+  static void test_addv(float[] a0, float[] a1, float b) {
+    // Non-counted loop with safepoint.
+    for (long l = 0; l < SFP_ITERS; l++) {
+      // Counted and vectorized loop.
+      for (int i = 0; i < a0.length; i+=1) {
+        a0[i] += a1[i]+b;
+      }
+    }
+  }
+
+  static int verify(String text, int i, float elem, float val) {
+    if (elem != val) {
+      System.err.println(text + "[" + i + "] = " + elem + " != " + val);
+      return 1;
+    }
+    return 0;
+  }
+}
+
diff --git a/test/compiler/7200264/Test7200264.sh b/test/compiler/7200264/Test7200264.sh
new file mode 100644
index 000000000..71b68d0f6
--- /dev/null
+++ b/test/compiler/7200264/Test7200264.sh
@@ -0,0 +1,150 @@
+#!/bin/sh
+# 
+# Copyright (c) 2012, Oracle and/or its affiliates. All rights reserved.
+# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+# 
+# This code is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License version 2 only, as
+# published by the Free Software Foundation.
+# 
+# This code is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+# FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+# version 2 for more details (a copy is included in the LICENSE file that
+# accompanied this code).
+# 
+# You should have received a copy of the GNU General Public License version
+# 2 along with this work; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+# 
+# Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+# or visit www.oracle.com if you need additional information or have any
+# questions.
+# 
+# 
+
+if [ "${TESTSRC}" = "" ]
+then
+  echo "TESTSRC not set.  Test cannot execute.  Failed."
+  exit 1
+fi
+echo "TESTSRC=${TESTSRC}"
+if [ "${TESTJAVA}" = "" ]
+then
+  echo "TESTJAVA not set.  Test cannot execute.  Failed."
+  exit 1
+fi
+echo "TESTJAVA=${TESTJAVA}"
+if [ "${TESTCLASSES}" = "" ]
+then
+  echo "TESTCLASSES not set.  Test cannot execute.  Failed."
+  exit 1
+fi
+echo "TESTCLASSES=${TESTCLASSES}"
+echo "CLASSPATH=${CLASSPATH}"
+
+# set platform-dependent variables
+OS=`uname -s`
+case "$OS" in
+  SunOS | Linux | Darwin )
+    NULL=/dev/null
+    PS=":"
+    FS="/"
+    ;;
+  Windows_* )
+    NULL=NUL
+    PS=";"
+    FS="\\"
+    ;;
+  CYGWIN_* )
+    NULL=/dev/null
+    PS=";"
+    FS="/"
+    ;;
+  * )
+    echo "Unrecognized system!"
+    exit 1;
+    ;;
+esac
+
+
+${TESTJAVA}${FS}bin${FS}java ${TESTVMOPTS} -Xinternalversion | sed 's/amd64/x86/' | grep "x86" | grep "Server VM" | grep "debug"
+
+# Only test fastdebug Server VM on x86
+if [ $? != 0 ]
+then
+    echo "Test Passed"
+    exit 0
+fi
+
+cp ${TESTSRC}${FS}TestIntVect.java .
+${TESTJAVA}${FS}bin${FS}javac -d . TestIntVect.java
+
+${TESTJAVA}${FS}bin${FS}java ${TESTVMOPTS} -Xbatch -XX:-TieredCompilation -XX:CICompilerCount=1 -XX:+PrintCompilation -XX:+TraceNewVectors TestIntVect > test.out 2>&1
+
+COUNT=`grep AddVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 4 ]
+then
+    echo "Test Failed: AddVI $COUNT < 4"
+    exit 1
+fi
+
+# AddVI is generated for test_subc
+COUNT=`grep SubVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 4 ]
+then
+    echo "Test Failed: SubVI $COUNT < 4"
+    exit 1
+fi
+
+# LShiftVI+SubVI is generated for test_mulc
+COUNT=`grep MulVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 2 ]
+then
+    echo "Test Failed: MulVI $COUNT < 2"
+    exit 1
+fi
+
+COUNT=`grep AndV test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 3 ]
+then
+    echo "Test Failed: AndV $COUNT < 3"
+    exit 1
+fi
+
+COUNT=`grep OrV test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 3 ]
+then
+    echo "Test Failed: OrV $COUNT < 3"
+    exit 1
+fi
+
+COUNT=`grep XorV test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 3 ]
+then
+    echo "Test Failed: XorV $COUNT < 3"
+    exit 1
+fi
+
+COUNT=`grep LShiftVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 5 ]
+then
+    echo "Test Failed: LShiftVI $COUNT < 5"
+    exit 1
+fi
+
+# RShiftVI + URShiftVI
+COUNT=`grep RShiftVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 6 ]
+then
+    echo "Test Failed: RShiftVI $COUNT < 6"
+    exit 1
+fi
+
+COUNT=`grep URShiftVI test.out | wc -l | awk '{print $1}'`
+if [ $COUNT -lt 3 ]
+then
+    echo "Test Failed: URShiftVI $COUNT < 3"
+    exit 1
+fi
+
diff --git a/test/compiler/7200264/TestIntVect.java b/test/compiler/7200264/TestIntVect.java
new file mode 100644
index 000000000..f85d4ff39
--- /dev/null
+++ b/test/compiler/7200264/TestIntVect.java
@@ -0,0 +1,650 @@
+/*
+ * Copyright (c) 2012, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+/**
+ * @test
+ * @bug 7200264
+ * @summary 7192963 changes disabled shift vectors
+ *
+ * @run shell Test7200264.sh
+ */
+
+/*
+ * Copy of test/compiler/6340864/TestIntVect.java without performance tests.
+ */
+public class TestIntVect {
+  private static final int ARRLEN = 997;
+  private static final int ITERS  = 11000;
+  private static final int ADD_INIT = Integer.MAX_VALUE-500;
+  private static final int BIT_MASK = 0xEC80F731;
+  private static final int VALUE = 15;
+  private static final int SHIFT = 32;
+
+  public static void main(String args[]) {
+    System.out.println("Testing Integer vectors");
+    int errn = test();
+    if (errn > 0) {
+      System.err.println("FAILED: " + errn + " errors");
+      System.exit(97);
+    }
+    System.out.println("PASSED");
+  }
+
+  static int test() {
+    int[] a0 = new int[ARRLEN];
+    int[] a1 = new int[ARRLEN];
+    int[] a2 = new int[ARRLEN];
+    int[] a3 = new int[ARRLEN];
+    int[] a4 = new int[ARRLEN];
+    long[] p2 = new long[ARRLEN/2];
+    // Initialize
+    int gold_sum = 0;
+    for (int i=0; i<ARRLEN; i++) {
+      int val = (int)(ADD_INIT+i);
+      gold_sum += val;
+      a1[i] = val;
+      a2[i] = (int)VALUE;
+      a3[i] = (int)-VALUE;
+      a4[i] = (int)BIT_MASK;
+    }
+    System.out.println("Warmup");
+    for (int i=0; i<ITERS; i++) {
+      test_sum(a1);
+      test_addc(a0, a1);
+      test_addv(a0, a1, (int)VALUE);
+      test_adda(a0, a1, a2);
+      test_subc(a0, a1);
+      test_subv(a0, a1, (int)VALUE);
+      test_suba(a0, a1, a2);
+      test_mulc(a0, a1);
+      test_mulv(a0, a1, (int)VALUE);
+      test_mula(a0, a1, a2);
+      test_divc(a0, a1);
+      test_divv(a0, a1, (int)VALUE);
+      test_diva(a0, a1, a2);
+      test_mulc_n(a0, a1);
+      test_mulv(a0, a1, (int)-VALUE);
+      test_mula(a0, a1, a3);
+      test_divc_n(a0, a1);
+      test_divv(a0, a1, (int)-VALUE);
+      test_diva(a0, a1, a3);
+      test_andc(a0, a1);
+      test_andv(a0, a1, (int)BIT_MASK);
+      test_anda(a0, a1, a4);
+      test_orc(a0, a1);
+      test_orv(a0, a1, (int)BIT_MASK);
+      test_ora(a0, a1, a4);
+      test_xorc(a0, a1);
+      test_xorv(a0, a1, (int)BIT_MASK);
+      test_xora(a0, a1, a4);
+      test_sllc(a0, a1);
+      test_sllv(a0, a1, VALUE);
+      test_srlc(a0, a1);
+      test_srlv(a0, a1, VALUE);
+      test_srac(a0, a1);
+      test_srav(a0, a1, VALUE);
+      test_sllc_n(a0, a1);
+      test_sllv(a0, a1, -VALUE);
+      test_srlc_n(a0, a1);
+      test_srlv(a0, a1, -VALUE);
+      test_srac_n(a0, a1);
+      test_srav(a0, a1, -VALUE);
+      test_sllc_o(a0, a1);
+      test_sllv(a0, a1, SHIFT);
+      test_srlc_o(a0, a1);
+      test_srlv(a0, a1, SHIFT);
+      test_srac_o(a0, a1);
+      test_srav(a0, a1, SHIFT);
+      test_sllc_on(a0, a1);
+      test_sllv(a0, a1, -SHIFT);
+      test_srlc_on(a0, a1);
+      test_srlv(a0, a1, -SHIFT);
+      test_srac_on(a0, a1);
+      test_srav(a0, a1, -SHIFT);
+      test_pack2(p2, a1);
+      test_unpack2(a0, p2);
+      test_pack2_swap(p2, a1);
+      test_unpack2_swap(a0, p2);
+    }
+    // Test and verify results
+    System.out.println("Verification");
+    int errn = 0;
+    {
+      int sum = test_sum(a1);
+      if (sum != gold_sum) {
+        System.err.println("test_sum:  " + sum + " != " + gold_sum);
+        errn++;
+      }
+
+      test_addc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_addc: ", i, a0[i], (int)((int)(ADD_INIT+i)+VALUE));
+      }
+      test_addv(a0, a1, (int)VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_addv: ", i, a0[i], (int)((int)(ADD_INIT+i)+VALUE));
+      }
+      test_adda(a0, a1, a2);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_adda: ", i, a0[i], (int)((int)(ADD_INIT+i)+VALUE));
+      }
+
+      test_subc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_subc: ", i, a0[i], (int)((int)(ADD_INIT+i)-VALUE));
+      }
+      test_subv(a0, a1, (int)VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_subv: ", i, a0[i], (int)((int)(ADD_INIT+i)-VALUE));
+      }
+      test_suba(a0, a1, a2);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_suba: ", i, a0[i], (int)((int)(ADD_INIT+i)-VALUE));
+      }
+
+      test_mulc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mulc: ", i, a0[i], (int)((int)(ADD_INIT+i)*VALUE));
+      }
+      test_mulv(a0, a1, (int)VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mulv: ", i, a0[i], (int)((int)(ADD_INIT+i)*VALUE));
+      }
+      test_mula(a0, a1, a2);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mula: ", i, a0[i], (int)((int)(ADD_INIT+i)*VALUE));
+      }
+
+      test_divc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_divc: ", i, a0[i], (int)((int)(ADD_INIT+i)/VALUE));
+      }
+      test_divv(a0, a1, (int)VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_divv: ", i, a0[i], (int)((int)(ADD_INIT+i)/VALUE));
+      }
+      test_diva(a0, a1, a2);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_diva: ", i, a0[i], (int)((int)(ADD_INIT+i)/VALUE));
+      }
+
+      test_mulc_n(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mulc_n: ", i, a0[i], (int)((int)(ADD_INIT+i)*(-VALUE)));
+      }
+      test_mulv(a0, a1, (int)-VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mulv_n: ", i, a0[i], (int)((int)(ADD_INIT+i)*(-VALUE)));
+      }
+      test_mula(a0, a1, a3);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_mula_n: ", i, a0[i], (int)((int)(ADD_INIT+i)*(-VALUE)));
+      }
+
+      test_divc_n(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_divc_n: ", i, a0[i], (int)((int)(ADD_INIT+i)/(-VALUE)));
+      }
+      test_divv(a0, a1, (int)-VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_divv_n: ", i, a0[i], (int)((int)(ADD_INIT+i)/(-VALUE)));
+      }
+      test_diva(a0, a1, a3);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_diva_n: ", i, a0[i], (int)((int)(ADD_INIT+i)/(-VALUE)));
+      }
+
+      test_andc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_andc: ", i, a0[i], (int)((int)(ADD_INIT+i)&BIT_MASK));
+      }
+      test_andv(a0, a1, (int)BIT_MASK);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_andv: ", i, a0[i], (int)((int)(ADD_INIT+i)&BIT_MASK));
+      }
+      test_anda(a0, a1, a4);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_anda: ", i, a0[i], (int)((int)(ADD_INIT+i)&BIT_MASK));
+      }
+
+      test_orc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_orc: ", i, a0[i], (int)((int)(ADD_INIT+i)|BIT_MASK));
+      }
+      test_orv(a0, a1, (int)BIT_MASK);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_orv: ", i, a0[i], (int)((int)(ADD_INIT+i)|BIT_MASK));
+      }
+      test_ora(a0, a1, a4);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_ora: ", i, a0[i], (int)((int)(ADD_INIT+i)|BIT_MASK));
+      }
+
+      test_xorc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_xorc: ", i, a0[i], (int)((int)(ADD_INIT+i)^BIT_MASK));
+      }
+      test_xorv(a0, a1, (int)BIT_MASK);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_xorv: ", i, a0[i], (int)((int)(ADD_INIT+i)^BIT_MASK));
+      }
+      test_xora(a0, a1, a4);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_xora: ", i, a0[i], (int)((int)(ADD_INIT+i)^BIT_MASK));
+      }
+
+      test_sllc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllc: ", i, a0[i], (int)((int)(ADD_INIT+i)<<VALUE));
+      }
+      test_sllv(a0, a1, VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllv: ", i, a0[i], (int)((int)(ADD_INIT+i)<<VALUE));
+      }
+
+      test_srlc(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlc: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>VALUE));
+      }
+      test_srlv(a0, a1, VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlv: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>VALUE));
+      }
+
+      test_srac(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srac: ", i, a0[i], (int)((int)(ADD_INIT+i)>>VALUE));
+      }
+      test_srav(a0, a1, VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srav: ", i, a0[i], (int)((int)(ADD_INIT+i)>>VALUE));
+      }
+
+      test_sllc_n(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllc_n: ", i, a0[i], (int)((int)(ADD_INIT+i)<<(-VALUE)));
+      }
+      test_sllv(a0, a1, -VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllv_n: ", i, a0[i], (int)((int)(ADD_INIT+i)<<(-VALUE)));
+      }
+
+      test_srlc_n(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlc_n: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>(-VALUE)));
+      }
+      test_srlv(a0, a1, -VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlv_n: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>(-VALUE)));
+      }
+
+      test_srac_n(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srac_n: ", i, a0[i], (int)((int)(ADD_INIT+i)>>(-VALUE)));
+      }
+      test_srav(a0, a1, -VALUE);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srav_n: ", i, a0[i], (int)((int)(ADD_INIT+i)>>(-VALUE)));
+      }
+
+      test_sllc_o(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllc_o: ", i, a0[i], (int)((int)(ADD_INIT+i)<<SHIFT));
+      }
+      test_sllv(a0, a1, SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllv_o: ", i, a0[i], (int)((int)(ADD_INIT+i)<<SHIFT));
+      }
+
+      test_srlc_o(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlc_o: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>SHIFT));
+      }
+      test_srlv(a0, a1, SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlv_o: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>SHIFT));
+      }
+
+      test_srac_o(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srac_o: ", i, a0[i], (int)((int)(ADD_INIT+i)>>SHIFT));
+      }
+      test_srav(a0, a1, SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srav_o: ", i, a0[i], (int)((int)(ADD_INIT+i)>>SHIFT));
+      }
+
+      test_sllc_on(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllc_on: ", i, a0[i], (int)((int)(ADD_INIT+i)<<(-SHIFT)));
+      }
+      test_sllv(a0, a1, -SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_sllv_on: ", i, a0[i], (int)((int)(ADD_INIT+i)<<(-SHIFT)));
+      }
+
+      test_srlc_on(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlc_on: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>(-SHIFT)));
+      }
+      test_srlv(a0, a1, -SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srlv_on: ", i, a0[i], (int)((int)(ADD_INIT+i)>>>(-SHIFT)));
+      }
+
+      test_srac_on(a0, a1);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srac_on: ", i, a0[i], (int)((int)(ADD_INIT+i)>>(-SHIFT)));
+      }
+      test_srav(a0, a1, -SHIFT);
+      for (int i=0; i<ARRLEN; i++) {
+        errn += verify("test_srav_on: ", i, a0[i], (int)((int)(ADD_INIT+i)>>(-SHIFT)));
+      }
+
+      test_pack2(p2, a1);
+      for (int i=0; i<ARRLEN/2; i++) {
+        errn += verify("test_pack2: ", i, p2[i], ((long)(ADD_INIT+2*i) & 0xFFFFFFFFl) | ((long)(ADD_INIT+2*i+1) << 32));
+      }
+      for (int i=0; i<ARRLEN; i++) {
+        a0[i] = -1;
+      }
+      test_unpack2(a0, p2);
+      for (int i=0; i<(ARRLEN&(-2)); i++) {
+        errn += verify("test_unpack2: ", i, a0[i], (ADD_INIT+i));
+      }
+
+      test_pack2_swap(p2, a1);
+      for (int i=0; i<ARRLEN/2; i++) {
+        errn += verify("test_pack2_swap: ", i, p2[i], ((long)(ADD_INIT+2*i+1) & 0xFFFFFFFFl) | ((long)(ADD_INIT+2*i) << 32));
+      }
+      for (int i=0; i<ARRLEN; i++) {
+        a0[i] = -1;
+      }
+      test_unpack2_swap(a0, p2);
+      for (int i=0; i<(ARRLEN&(-2)); i++) {
+        errn += verify("test_unpack2_swap: ", i, a0[i], (ADD_INIT+i));
+      }
+
+    }
+
+    return errn;
+  }
+
+  static int test_sum(int[] a1) {
+    int sum = 0;
+    for (int i = 0; i < a1.length; i+=1) {
+      sum += a1[i];
+    }
+    return sum;
+  }
+
+  static void test_addc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]+VALUE);
+    }
+  }
+  static void test_addv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]+b);
+    }
+  }
+  static void test_adda(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]+a2[i]);
+    }
+  }
+
+  static void test_subc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]-VALUE);
+    }
+  }
+  static void test_subv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]-b);
+    }
+  }
+  static void test_suba(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]-a2[i]);
+    }
+  }
+
+  static void test_mulc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]*VALUE);
+    }
+  }
+  static void test_mulc_n(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]*(-VALUE));
+    }
+  }
+  static void test_mulv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]*b);
+    }
+  }
+  static void test_mula(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]*a2[i]);
+    }
+  }
+
+  static void test_divc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]/VALUE);
+    }
+  }
+  static void test_divc_n(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]/(-VALUE));
+    }
+  }
+  static void test_divv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]/b);
+    }
+  }
+  static void test_diva(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]/a2[i]);
+    }
+  }
+
+  static void test_andc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]&BIT_MASK);
+    }
+  }
+  static void test_andv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]&b);
+    }
+  }
+  static void test_anda(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]&a2[i]);
+    }
+  }
+
+  static void test_orc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]|BIT_MASK);
+    }
+  }
+  static void test_orv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]|b);
+    }
+  }
+  static void test_ora(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]|a2[i]);
+    }
+  }
+
+  static void test_xorc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]^BIT_MASK);
+    }
+  }
+  static void test_xorv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]^b);
+    }
+  }
+  static void test_xora(int[] a0, int[] a1, int[] a2) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]^a2[i]);
+    }
+  }
+
+  static void test_sllc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]<<VALUE);
+    }
+  }
+  static void test_sllc_n(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]<<(-VALUE));
+    }
+  }
+  static void test_sllc_o(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]<<SHIFT);
+    }
+  }
+  static void test_sllc_on(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]<<(-SHIFT));
+    }
+  }
+  static void test_sllv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]<<b);
+    }
+  }
+
+  static void test_srlc(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>>VALUE);
+    }
+  }
+  static void test_srlc_n(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>>(-VALUE));
+    }
+  }
+  static void test_srlc_o(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>>SHIFT);
+    }
+  }
+  static void test_srlc_on(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>>(-SHIFT));
+    }
+  }
+  static void test_srlv(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>>b);
+    }
+  }
+
+  static void test_srac(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>VALUE);
+    }
+  }
+  static void test_srac_n(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>(-VALUE));
+    }
+  }
+  static void test_srac_o(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>SHIFT);
+    }
+  }
+  static void test_srac_on(int[] a0, int[] a1) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>(-SHIFT));
+    }
+  }
+  static void test_srav(int[] a0, int[] a1, int b) {
+    for (int i = 0; i < a0.length; i+=1) {
+      a0[i] = (int)(a1[i]>>b);
+    }
+  }
+
+  static void test_pack2(long[] p2, int[] a1) {
+    if (p2.length*2 > a1.length) return;
+    for (int i = 0; i < p2.length; i+=1) {
+      long l0 = (long)a1[i*2+0];
+      long l1 = (long)a1[i*2+1];
+      p2[i] = (l1 << 32) | (l0 & 0xFFFFFFFFl);
+    }
+  }
+  static void test_unpack2(int[] a0, long[] p2) {
+    if (p2.length*2 > a0.length) return;
+    for (int i = 0; i < p2.length; i+=1) {
+      long l = p2[i];
+      a0[i*2+0] = (int)(l & 0xFFFFFFFFl);
+      a0[i*2+1] = (int)(l >> 32);
+    }
+  }
+  static void test_pack2_swap(long[] p2, int[] a1) {
+    if (p2.length*2 > a1.length) return;
+    for (int i = 0; i < p2.length; i+=1) {
+      long l0 = (long)a1[i*2+0];
+      long l1 = (long)a1[i*2+1];
+      p2[i] = (l0 << 32) | (l1 & 0xFFFFFFFFl);
+    }
+  }
+  static void test_unpack2_swap(int[] a0, long[] p2) {
+    if (p2.length*2 > a0.length) return;
+    for (int i = 0; i < p2.length; i+=1) {
+      long l = p2[i];
+      a0[i*2+0] = (int)(l >> 32);
+      a0[i*2+1] = (int)(l & 0xFFFFFFFFl);
+    }
+  }
+
+  static int verify(String text, int i, int elem, int val) {
+    if (elem != val) {
+      System.err.println(text + "[" + i + "] = " + elem + " != " + val);
+      return 1;
+    }
+    return 0;
+  }
+
+  static int verify(String text, int i, long elem, long val) {
+    if (elem != val) {
+      System.err.println(text + "[" + i + "] = " + Long.toHexString(elem) + " != " + Long.toHexString(val));
+      return 1;
+    }
+    return 0;
+  }
+}
author	kvn <none@none>	2012-09-28 10:16:29 -0700
committer	kvn <none@none>	2012-09-28 10:16:29 -0700
commit	7ce94fd32972533edc4ecf0d2a270005abab7c10 (patch)
tree	add83032c0821b8b7210ba3d7e7b58b27a16fe96
parent	1bac024e917c46f2b5884889b8faffb60546d7c1 (diff)
parent	00e5c6d91a3db80e48359621f9b85332477b6e1b (diff)