--- a/make/hotspot_version	Thu Oct 04 14:34:51 2012 -0700
+++ b/make/hotspot_version	Fri Oct 05 13:28:16 2012 -0700
@@ -35,7 +35,7 @@
 
 HS_MAJOR_VER=25
 HS_MINOR_VER=0
-HS_BUILD_NUMBER=03
+HS_BUILD_NUMBER=04
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=8

--- a/src/cpu/sparc/vm/sparc.ad	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/cpu/sparc/vm/sparc.ad	Fri Oct 05 13:28:16 2012 -0700
@@ -1870,6 +1870,11 @@
   return Op_RegD;
 }
 
+const int Matcher::vector_shift_count_ideal_reg(int size) {
+  fatal("vector shift is not supported");
+  return Node::NotAMachineReg;
+}
+
 // Limits on vector size (number of elements) loaded into vector.
 const int Matcher::max_vector_size(const BasicType bt) {
   assert(is_java_primitive(bt), "only primitive type vectors");

--- a/src/cpu/sparc/vm/templateTable_sparc.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/cpu/sparc/vm/templateTable_sparc.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -2936,6 +2936,7 @@
     // Push the appendix as a trailing parameter.
     // This must be done before we get the receiver,
     // since the parameter_size includes it.
+    assert(ConstantPoolCacheEntry::_indy_resolved_references_appendix_offset == 0, "appendix expected at index+0");
     __ load_resolved_reference_at_index(temp, index);
     __ verify_oop(temp);
     __ push_ptr(temp);  // push appendix (MethodType, CallSite, etc.)
@@ -3235,15 +3236,15 @@
   }
 
   const Register Rret       = Lscratch;
-  const Register G4_mtype   = G4_scratch;  // f1
+  const Register G4_mtype   = G4_scratch;
   const Register O0_recv    = O0;
   const Register Rscratch   = G3_scratch;
 
   prepare_invoke(byte_no, G5_method, Rret, G4_mtype, O0_recv);
   __ null_check(O0_recv);
 
-  // G4: MethodType object (from cpool->resolved_references[])
-  // G5: MH.linkToCallSite method (from f2)
+  // G4: MethodType object (from cpool->resolved_references[f1], if necessary)
+  // G5: MH.invokeExact_MT method (from f2)
 
   // Note:  G4_mtype is already pushed (if necessary) by prepare_invoke
 
@@ -3275,8 +3276,8 @@
 
   prepare_invoke(byte_no, G5_method, Rret, G4_callsite);
 
-  // G4: CallSite object (from cpool->resolved_references[])
-  // G5: MH.linkToCallSite method (from f1)
+  // G4: CallSite object (from cpool->resolved_references[f1])
+  // G5: MH.linkToCallSite method (from f2)
 
   // Note:  G4_callsite is already pushed by prepare_invoke
 

--- a/src/cpu/x86/vm/templateTable_x86_32.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/cpu/x86/vm/templateTable_x86_32.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -2139,7 +2139,7 @@
   const int index_offset = in_bytes(ConstantPoolCache::base_offset() +
                                     ConstantPoolCacheEntry::f2_offset());
 
-    size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
+  size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
   resolve_cache_and_index(byte_no, cache, index, index_size);
     __ movptr(method, Address(cache, index, Address::times_ptr, method_offset));
 
@@ -2876,6 +2876,7 @@
     // since the parameter_size includes it.
     __ push(rbx);
     __ mov(rbx, index);
+    assert(ConstantPoolCacheEntry::_indy_resolved_references_appendix_offset == 0, "appendix expected at index+0");
     __ load_resolved_reference_at_index(index, rbx);
     __ pop(rbx);
     __ push(index);  // push appendix (MethodType, CallSite, etc.)
@@ -3093,8 +3094,8 @@
 void TemplateTable::invokehandle(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  const Register rbx_method = rbx;  // (from f2)
-  const Register rax_mtype  = rax;  // (from f1)
+  const Register rbx_method = rbx;
+  const Register rax_mtype  = rax;
   const Register rcx_recv   = rcx;
   const Register rdx_flags  = rdx;
 
@@ -3104,13 +3105,14 @@
     return;
   }
 
-  prepare_invoke(byte_no,
-                 rbx_method, rax_mtype,  // get f2 Method*, f1 MethodType
-                 rcx_recv);
+  prepare_invoke(byte_no, rbx_method, rax_mtype, rcx_recv);
   __ verify_method_ptr(rbx_method);
   __ verify_oop(rcx_recv);
   __ null_check(rcx_recv);
 
+  // rax: MethodType object (from cpool->resolved_references[f1], if necessary)
+  // rbx: MH.invokeExact_MT method (from f2)
+
   // Note:  rax_mtype is already pushed (if necessary) by prepare_invoke
 
   // FIXME: profile the LambdaForm also
@@ -3140,7 +3142,7 @@
 
   prepare_invoke(byte_no, rbx_method, rax_callsite);
 
-  // rax: CallSite object (from cpool->resolved_references[])
+  // rax: CallSite object (from cpool->resolved_references[f1])
   // rbx: MH.linkToCallSite method (from f2)
 
   // Note:  rax_callsite is already pushed by prepare_invoke

--- a/src/cpu/x86/vm/templateTable_x86_64.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/cpu/x86/vm/templateTable_x86_64.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -2184,7 +2184,7 @@
   const int index_offset = in_bytes(ConstantPoolCache::base_offset() +
                                     ConstantPoolCacheEntry::f2_offset());
 
-    size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
+  size_t index_size = (is_invokedynamic ? sizeof(u4) : sizeof(u2));
   resolve_cache_and_index(byte_no, cache, index, index_size);
     __ movptr(method, Address(cache, index, Address::times_ptr, method_offset));
 
@@ -2926,6 +2926,7 @@
     // since the parameter_size includes it.
     __ push(rbx);
     __ mov(rbx, index);
+    assert(ConstantPoolCacheEntry::_indy_resolved_references_appendix_offset == 0, "appendix expected at index+0");
     __ load_resolved_reference_at_index(index, rbx);
     __ pop(rbx);
     __ push(index);  // push appendix (MethodType, CallSite, etc.)
@@ -3144,8 +3145,8 @@
 void TemplateTable::invokehandle(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
-  const Register rbx_method = rbx;  // f2
-  const Register rax_mtype  = rax;  // f1
+  const Register rbx_method = rbx;
+  const Register rax_mtype  = rax;
   const Register rcx_recv   = rcx;
   const Register rdx_flags  = rdx;
 
@@ -3155,13 +3156,14 @@
     return;
   }
 
-  prepare_invoke(byte_no,
-                 rbx_method, rax_mtype,  // get f2 Method*, f1 MethodType
-                 rcx_recv);
+  prepare_invoke(byte_no, rbx_method, rax_mtype, rcx_recv);
   __ verify_method_ptr(rbx_method);
   __ verify_oop(rcx_recv);
   __ null_check(rcx_recv);
 
+  // rax: MethodType object (from cpool->resolved_references[f1], if necessary)
+  // rbx: MH.invokeExact_MT method (from f2)
+
   // Note:  rax_mtype is already pushed (if necessary) by prepare_invoke
 
   // FIXME: profile the LambdaForm also
@@ -3191,7 +3193,7 @@
 
   prepare_invoke(byte_no, rbx_method, rax_callsite);
 
-  // rax: CallSite object (from cpool->resolved_references[])
+  // rax: CallSite object (from cpool->resolved_references[f1])
   // rbx: MH.linkToCallSite method (from f2)
 
   // Note:  rax_callsite is already pushed by prepare_invoke

--- a/src/cpu/x86/vm/x86.ad	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/cpu/x86/vm/x86.ad	Fri Oct 05 13:28:16 2012 -0700
@@ -571,6 +571,11 @@
   return 0;
 }
 
+// Only lowest bits of xmm reg are used for vector shift count.
+const int Matcher::vector_shift_count_ideal_reg(int size) {
+  return Op_VecS;
+}
+
 // x86 supports misaligned vectors store/load.
 const bool Matcher::misaligned_vectors_ok() {
   return !AlignVector; // can be changed by flag
@@ -3758,10 +3763,24 @@
   ins_pipe( pipe_slow );
 %}
 
+// ------------------------------ Shift ---------------------------------------
+
+// Left and right shift count vectors are the same on x86
+// (only lowest bits of xmm reg are used for count).
+instruct vshiftcnt(vecS dst, rRegI cnt) %{
+  match(Set dst (LShiftCntV cnt));
+  match(Set dst (RShiftCntV cnt));
+  format %{ "movd    $dst,$cnt\t! load shift count" %}
+  ins_encode %{
+    __ movdl($dst$$XMMRegister, $cnt$$Register);
+  %}
+  ins_pipe( pipe_slow );
+%}
+
 // ------------------------------ LeftShift -----------------------------------
 
 // Shorts/Chars vector left shift
-instruct vsll2S(vecS dst, regF shift) %{
+instruct vsll2S(vecS dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (LShiftVS dst shift));
   format %{ "psllw   $dst,$shift\t! left shift packed2S" %}
@@ -3781,7 +3800,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll2S_reg(vecS dst, vecS src, regF shift) %{
+instruct vsll2S_reg(vecS dst, vecS src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (LShiftVS src shift));
   format %{ "vpsllw  $dst,$src,$shift\t! left shift packed2S" %}
@@ -3803,7 +3822,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll4S(vecD dst, regF shift) %{
+instruct vsll4S(vecD dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 4);
   match(Set dst (LShiftVS dst shift));
   format %{ "psllw   $dst,$shift\t! left shift packed4S" %}
@@ -3823,7 +3842,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll4S_reg(vecD dst, vecD src, regF shift) %{
+instruct vsll4S_reg(vecD dst, vecD src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 4);
   match(Set dst (LShiftVS src shift));
   format %{ "vpsllw  $dst,$src,$shift\t! left shift packed4S" %}
@@ -3845,7 +3864,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll8S(vecX dst, regF shift) %{
+instruct vsll8S(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 8);
   match(Set dst (LShiftVS dst shift));
   format %{ "psllw   $dst,$shift\t! left shift packed8S" %}
@@ -3865,7 +3884,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll8S_reg(vecX dst, vecX src, regF shift) %{
+instruct vsll8S_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 8);
   match(Set dst (LShiftVS src shift));
   format %{ "vpsllw  $dst,$src,$shift\t! left shift packed8S" %}
@@ -3887,7 +3906,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll16S_reg(vecY dst, vecY src, regF shift) %{
+instruct vsll16S_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 16);
   match(Set dst (LShiftVS src shift));
   format %{ "vpsllw  $dst,$src,$shift\t! left shift packed16S" %}
@@ -3910,7 +3929,7 @@
 %}
 
 // Integers vector left shift
-instruct vsll2I(vecD dst, regF shift) %{
+instruct vsll2I(vecD dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (LShiftVI dst shift));
   format %{ "pslld   $dst,$shift\t! left shift packed2I" %}
@@ -3930,7 +3949,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll2I_reg(vecD dst, vecD src, regF shift) %{
+instruct vsll2I_reg(vecD dst, vecD src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (LShiftVI src shift));
   format %{ "vpslld  $dst,$src,$shift\t! left shift packed2I" %}
@@ -3952,7 +3971,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll4I(vecX dst, regF shift) %{
+instruct vsll4I(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 4);
   match(Set dst (LShiftVI dst shift));
   format %{ "pslld   $dst,$shift\t! left shift packed4I" %}
@@ -3972,7 +3991,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll4I_reg(vecX dst, vecX src, regF shift) %{
+instruct vsll4I_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 4);
   match(Set dst (LShiftVI src shift));
   format %{ "vpslld  $dst,$src,$shift\t! left shift packed4I" %}
@@ -3994,7 +4013,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll8I_reg(vecY dst, vecY src, regF shift) %{
+instruct vsll8I_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 8);
   match(Set dst (LShiftVI src shift));
   format %{ "vpslld  $dst,$src,$shift\t! left shift packed8I" %}
@@ -4017,7 +4036,7 @@
 %}
 
 // Longs vector left shift
-instruct vsll2L(vecX dst, regF shift) %{
+instruct vsll2L(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (LShiftVL dst shift));
   format %{ "psllq   $dst,$shift\t! left shift packed2L" %}
@@ -4037,7 +4056,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll2L_reg(vecX dst, vecX src, regF shift) %{
+instruct vsll2L_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (LShiftVL src shift));
   format %{ "vpsllq  $dst,$src,$shift\t! left shift packed2L" %}
@@ -4059,7 +4078,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsll4L_reg(vecY dst, vecY src, regF shift) %{
+instruct vsll4L_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 4);
   match(Set dst (LShiftVL src shift));
   format %{ "vpsllq  $dst,$src,$shift\t! left shift packed4L" %}
@@ -4088,7 +4107,7 @@
 // sign extension before a shift.
 
 // Integers vector logical right shift
-instruct vsrl2I(vecD dst, regF shift) %{
+instruct vsrl2I(vecD dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (URShiftVI dst shift));
   format %{ "psrld   $dst,$shift\t! logical right shift packed2I" %}
@@ -4108,7 +4127,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl2I_reg(vecD dst, vecD src, regF shift) %{
+instruct vsrl2I_reg(vecD dst, vecD src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (URShiftVI src shift));
   format %{ "vpsrld  $dst,$src,$shift\t! logical right shift packed2I" %}
@@ -4130,7 +4149,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl4I(vecX dst, regF shift) %{
+instruct vsrl4I(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 4);
   match(Set dst (URShiftVI dst shift));
   format %{ "psrld   $dst,$shift\t! logical right shift packed4I" %}
@@ -4150,7 +4169,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl4I_reg(vecX dst, vecX src, regF shift) %{
+instruct vsrl4I_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 4);
   match(Set dst (URShiftVI src shift));
   format %{ "vpsrld  $dst,$src,$shift\t! logical right shift packed4I" %}
@@ -4172,7 +4191,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl8I_reg(vecY dst, vecY src, regF shift) %{
+instruct vsrl8I_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 8);
   match(Set dst (URShiftVI src shift));
   format %{ "vpsrld  $dst,$src,$shift\t! logical right shift packed8I" %}
@@ -4195,7 +4214,7 @@
 %}
 
 // Longs vector logical right shift
-instruct vsrl2L(vecX dst, regF shift) %{
+instruct vsrl2L(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (URShiftVL dst shift));
   format %{ "psrlq   $dst,$shift\t! logical right shift packed2L" %}
@@ -4215,7 +4234,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl2L_reg(vecX dst, vecX src, regF shift) %{
+instruct vsrl2L_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (URShiftVL src shift));
   format %{ "vpsrlq  $dst,$src,$shift\t! logical right shift packed2L" %}
@@ -4237,7 +4256,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsrl4L_reg(vecY dst, vecY src, regF shift) %{
+instruct vsrl4L_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 4);
   match(Set dst (URShiftVL src shift));
   format %{ "vpsrlq  $dst,$src,$shift\t! logical right shift packed4L" %}
@@ -4262,7 +4281,7 @@
 // ------------------- ArithmeticRightShift -----------------------------------
 
 // Shorts/Chars vector arithmetic right shift
-instruct vsra2S(vecS dst, regF shift) %{
+instruct vsra2S(vecS dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (RShiftVS dst shift));
   format %{ "psraw   $dst,$shift\t! arithmetic right shift packed2S" %}
@@ -4282,7 +4301,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra2S_reg(vecS dst, vecS src, regF shift) %{
+instruct vsra2S_reg(vecS dst, vecS src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (RShiftVS src shift));
   format %{ "vpsraw  $dst,$src,$shift\t! arithmetic right shift packed2S" %}
@@ -4304,7 +4323,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra4S(vecD dst, regF shift) %{
+instruct vsra4S(vecD dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 4);
   match(Set dst (RShiftVS dst shift));
   format %{ "psraw   $dst,$shift\t! arithmetic right shift packed4S" %}
@@ -4324,7 +4343,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra4S_reg(vecD dst, vecD src, regF shift) %{
+instruct vsra4S_reg(vecD dst, vecD src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 4);
   match(Set dst (RShiftVS src shift));
   format %{ "vpsraw  $dst,$src,$shift\t! arithmetic right shift packed4S" %}
@@ -4346,7 +4365,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra8S(vecX dst, regF shift) %{
+instruct vsra8S(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 8);
   match(Set dst (RShiftVS dst shift));
   format %{ "psraw   $dst,$shift\t! arithmetic right shift packed8S" %}
@@ -4366,7 +4385,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra8S_reg(vecX dst, vecX src, regF shift) %{
+instruct vsra8S_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 8);
   match(Set dst (RShiftVS src shift));
   format %{ "vpsraw  $dst,$src,$shift\t! arithmetic right shift packed8S" %}
@@ -4388,7 +4407,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra16S_reg(vecY dst, vecY src, regF shift) %{
+instruct vsra16S_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 16);
   match(Set dst (RShiftVS src shift));
   format %{ "vpsraw  $dst,$src,$shift\t! arithmetic right shift packed16S" %}
@@ -4411,7 +4430,7 @@
 %}
 
 // Integers vector arithmetic right shift
-instruct vsra2I(vecD dst, regF shift) %{
+instruct vsra2I(vecD dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 2);
   match(Set dst (RShiftVI dst shift));
   format %{ "psrad   $dst,$shift\t! arithmetic right shift packed2I" %}
@@ -4431,7 +4450,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra2I_reg(vecD dst, vecD src, regF shift) %{
+instruct vsra2I_reg(vecD dst, vecD src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 2);
   match(Set dst (RShiftVI src shift));
   format %{ "vpsrad  $dst,$src,$shift\t! arithmetic right shift packed2I" %}
@@ -4453,7 +4472,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra4I(vecX dst, regF shift) %{
+instruct vsra4I(vecX dst, vecS shift) %{
   predicate(n->as_Vector()->length() == 4);
   match(Set dst (RShiftVI dst shift));
   format %{ "psrad   $dst,$shift\t! arithmetic right shift packed4I" %}
@@ -4473,7 +4492,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra4I_reg(vecX dst, vecX src, regF shift) %{
+instruct vsra4I_reg(vecX dst, vecX src, vecS shift) %{
   predicate(UseAVX > 0 && n->as_Vector()->length() == 4);
   match(Set dst (RShiftVI src shift));
   format %{ "vpsrad  $dst,$src,$shift\t! arithmetic right shift packed4I" %}
@@ -4495,7 +4514,7 @@
   ins_pipe( pipe_slow );
 %}
 
-instruct vsra8I_reg(vecY dst, vecY src, regF shift) %{
+instruct vsra8I_reg(vecY dst, vecY src, vecS shift) %{
   predicate(UseAVX > 1 && n->as_Vector()->length() == 8);
   match(Set dst (RShiftVI src shift));
   format %{ "vpsrad  $dst,$src,$shift\t! arithmetic right shift packed8I" %}

--- a/src/os/bsd/vm/os_bsd.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os/bsd/vm/os_bsd.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -82,12 +82,6 @@
 # include "assembler_ppc.inline.hpp"
 # include "nativeInst_ppc.hpp"
 #endif
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <sys/types.h>

--- a/src/os/linux/vm/os_linux.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os/linux/vm/os_linux.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -82,12 +82,6 @@
 # include "assembler_ppc.inline.hpp"
 # include "nativeInst_ppc.hpp"
 #endif
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <sys/types.h>

--- a/src/os/solaris/vm/os_solaris.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os/solaris/vm/os_solaris.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -70,12 +70,6 @@
 # include "assembler_sparc.inline.hpp"
 # include "nativeInst_sparc.hpp"
 #endif
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <dlfcn.h>

--- a/src/os/windows/vm/os_windows.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os/windows/vm/os_windows.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -69,12 +69,6 @@
 # include "assembler_x86.inline.hpp"
 # include "nativeInst_x86.hpp"
 #endif
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 #ifdef _DEBUG
 #include <crtdbg.h>

--- a/src/os_cpu/bsd_x86/vm/os_bsd_x86.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/bsd_x86/vm/os_bsd_x86.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,12 +52,6 @@
 #include "thread_bsd.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <sys/types.h>

--- a/src/os_cpu/bsd_zero/vm/os_bsd_zero.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/bsd_zero/vm/os_bsd_zero.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -58,12 +58,6 @@
 #include "thread_bsd.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 address os::current_stack_pointer() {
   address dummy = (address) &dummy;

--- a/src/os_cpu/linux_sparc/vm/os_linux_sparc.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/linux_sparc/vm/os_linux_sparc.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,13 +52,6 @@
 #include "thread_linux.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
-
 
 // Linux/Sparc has rather obscure naming of registers in sigcontext
 // different between 32 and 64 bits

--- a/src/os_cpu/linux_x86/vm/os_linux_x86.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/linux_x86/vm/os_linux_x86.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,12 +52,6 @@
 #include "thread_linux.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <sys/types.h>

--- a/src/os_cpu/linux_zero/vm/os_linux_zero.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/linux_zero/vm/os_linux_zero.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -53,12 +53,6 @@
 #include "thread_linux.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 address os::current_stack_pointer() {
   address dummy = (address) &dummy;

--- a/src/os_cpu/solaris_sparc/vm/os_solaris_sparc.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/solaris_sparc/vm/os_solaris_sparc.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,13 +52,6 @@
 #include "thread_solaris.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
-
 
 # include <signal.h>        // needed first to avoid name collision for "std" with SC 5.0
 

--- a/src/os_cpu/solaris_x86/vm/os_solaris_x86.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/solaris_x86/vm/os_solaris_x86.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,12 +52,6 @@
 #include "thread_solaris.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 // put OS-includes here
 # include <sys/types.h>

--- a/src/os_cpu/windows_x86/vm/os_windows_x86.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/os_cpu/windows_x86/vm/os_windows_x86.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -52,12 +52,6 @@
 #include "thread_windows.inline.hpp"
 #include "utilities/events.hpp"
 #include "utilities/vmError.hpp"
-#ifdef COMPILER1
-#include "c1/c1_Runtime1.hpp"
-#endif
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 
 # include "unwind_windows_x86.hpp"
 #undef REG_SP

--- a/src/share/vm/adlc/formssel.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/adlc/formssel.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -4049,6 +4049,7 @@
     "MulVS","MulVI","MulVF","MulVD",
     "DivVF","DivVD",
     "AndV" ,"XorV" ,"OrV",
+    "LShiftCntV","RShiftCntV",
     "LShiftVB","LShiftVS","LShiftVI","LShiftVL",
     "RShiftVB","RShiftVS","RShiftVI","RShiftVL",
     "URShiftVB","URShiftVS","URShiftVI","URShiftVL",

--- a/src/share/vm/ci/ciClassList.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciClassList.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -49,6 +49,7 @@
 class     ciCallSite;
 class     ciMemberName;
 class     ciMethodHandle;
+class     ciMethodType;
 class     ciArray;
 class       ciObjArray;
 class       ciTypeArray;
@@ -99,6 +100,7 @@
 friend class ciMethod;                 \
 friend class ciMethodData;             \
 friend class ciMethodHandle;           \
+friend class ciMethodType;             \
 friend class ciReceiverTypeData;       \
 friend class ciSymbol;                 \
 friend class ciArray;                  \

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/src/share/vm/ci/ciMethodType.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -0,0 +1,76 @@
+/*
+ * 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.
+ *
+ */
+
+#ifndef SHARE_VM_CI_CIMETHODTYPE_HPP
+#define SHARE_VM_CI_CIMETHODTYPE_HPP
+
+#include "ci/ciInstance.hpp"
+#include "ci/ciUtilities.hpp"
+#include "classfile/javaClasses.hpp"
+
+// ciMethodType
+//
+// The class represents a java.lang.invoke.MethodType object.
+class ciMethodType : public ciInstance {
+private:
+  ciType* class_to_citype(oop klass_oop) const {
+    if (java_lang_Class::is_primitive(klass_oop)) {
+      BasicType bt = java_lang_Class::primitive_type(klass_oop);
+      return ciType::make(bt);
+    } else {
+      Klass* k = java_lang_Class::as_Klass(klass_oop);
+      return CURRENT_ENV->get_klass(k);
+    }
+  }
+
+public:
+  ciMethodType(instanceHandle h_i) : ciInstance(h_i) {}
+
+  // What kind of ciObject is this?
+  bool is_method_type() const { return true; }
+
+  ciType* rtype() const {
+    GUARDED_VM_ENTRY(
+      oop rtype = java_lang_invoke_MethodType::rtype(get_oop());
+      return class_to_citype(rtype);
+    )
+  }
+
+  int ptype_count() const {
+    GUARDED_VM_ENTRY(return java_lang_invoke_MethodType::ptype_count(get_oop());)
+  }
+
+  int ptype_slot_count() const {
+    GUARDED_VM_ENTRY(return java_lang_invoke_MethodType::ptype_slot_count(get_oop());)
+  }
+
+  ciType* ptype_at(int index) const {
+    GUARDED_VM_ENTRY(
+      oop ptype = java_lang_invoke_MethodType::ptype(get_oop(), index);
+      return class_to_citype(ptype);
+    )
+  }
+};
+
+#endif // SHARE_VM_CI_CIMETHODTYPE_HPP

--- a/src/share/vm/ci/ciObject.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciObject.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -123,6 +123,7 @@
   virtual bool is_instance()                { return false; }
   virtual bool is_member_name()       const { return false; }
   virtual bool is_method_handle()     const { return false; }
+  virtual bool is_method_type()       const { return false; }
   virtual bool is_array()                   { return false; }
   virtual bool is_obj_array()               { return false; }
   virtual bool is_type_array()              { return false; }
@@ -142,35 +143,39 @@
   }
 
   // Subclass casting with assertions.
-  ciNullObject*            as_null_object() {
+  ciNullObject* as_null_object() {
     assert(is_null_object(), "bad cast");
     return (ciNullObject*)this;
   }
-  ciCallSite*              as_call_site() {
+  ciCallSite* as_call_site() {
     assert(is_call_site(), "bad cast");
-    return (ciCallSite*) this;
+    return (ciCallSite*)this;
   }
-  ciInstance*              as_instance() {
+  ciInstance* as_instance() {
     assert(is_instance(), "bad cast");
     return (ciInstance*)this;
   }
-  ciMemberName*            as_member_name() {
+  ciMemberName* as_member_name() {
     assert(is_member_name(), "bad cast");
     return (ciMemberName*)this;
   }
-  ciMethodHandle*          as_method_handle() {
+  ciMethodHandle* as_method_handle() {
     assert(is_method_handle(), "bad cast");
-    return (ciMethodHandle*) this;
+    return (ciMethodHandle*)this;
   }
-  ciArray*                 as_array() {
+  ciMethodType* as_method_type() {
+    assert(is_method_type(), "bad cast");
+    return (ciMethodType*)this;
+  }
+  ciArray* as_array() {
     assert(is_array(), "bad cast");
     return (ciArray*)this;
   }
-  ciObjArray*              as_obj_array() {
+  ciObjArray* as_obj_array() {
     assert(is_obj_array(), "bad cast");
     return (ciObjArray*)this;
   }
-  ciTypeArray*             as_type_array() {
+  ciTypeArray* as_type_array() {
     assert(is_type_array(), "bad cast");
     return (ciTypeArray*)this;
   }

--- a/src/share/vm/ci/ciObjectFactory.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciObjectFactory.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -30,6 +30,7 @@
 #include "ci/ciMethod.hpp"
 #include "ci/ciMethodData.hpp"
 #include "ci/ciMethodHandle.hpp"
+#include "ci/ciMethodType.hpp"
 #include "ci/ciNullObject.hpp"
 #include "ci/ciObjArray.hpp"
 #include "ci/ciObjArrayKlass.hpp"
@@ -237,23 +238,23 @@
 
   assert(key == NULL || Universe::heap()->is_in_reserved(key), "must be");
 
-    NonPermObject* &bucket = find_non_perm(key);
-    if (bucket != NULL) {
-      return bucket->object();
-    }
+  NonPermObject* &bucket = find_non_perm(key);
+  if (bucket != NULL) {
+    return bucket->object();
+  }
 
-    // The ciObject does not yet exist.  Create it and insert it
-    // into the cache.
-    Handle keyHandle(key);
-    ciObject* new_object = create_new_object(keyHandle());
-    assert(keyHandle() == new_object->get_oop(), "must be properly recorded");
-    init_ident_of(new_object);
+  // The ciObject does not yet exist.  Create it and insert it
+  // into the cache.
+  Handle keyHandle(key);
+  ciObject* new_object = create_new_object(keyHandle());
+  assert(keyHandle() == new_object->get_oop(), "must be properly recorded");
+  init_ident_of(new_object);
   assert(Universe::heap()->is_in_reserved(new_object->get_oop()), "must be");
 
-      // Not a perm-space object.
-      insert_non_perm(bucket, keyHandle(), new_object);
-      return new_object;
-    }
+  // Not a perm-space object.
+  insert_non_perm(bucket, keyHandle(), new_object);
+  return new_object;
+}
 
 // ------------------------------------------------------------------
 // ciObjectFactory::get
@@ -324,6 +325,8 @@
       return new (arena()) ciMemberName(h_i);
     else if (java_lang_invoke_MethodHandle::is_instance(o))
       return new (arena()) ciMethodHandle(h_i);
+    else if (java_lang_invoke_MethodType::is_instance(o))
+      return new (arena()) ciMethodType(h_i);
     else
       return new (arena()) ciInstance(h_i);
   } else if (o->is_objArray()) {

--- a/src/share/vm/ci/ciSignature.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciSignature.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -23,6 +23,7 @@
  */
 
 #include "precompiled.hpp"
+#include "ci/ciMethodType.hpp"
 #include "ci/ciSignature.hpp"
 #include "ci/ciUtilities.hpp"
 #include "memory/allocation.inline.hpp"
@@ -80,6 +81,24 @@
 }
 
 // ------------------------------------------------------------------
+// ciSignature::ciSignature
+ciSignature::ciSignature(ciKlass* accessing_klass, ciSymbol* symbol, ciMethodType* method_type) :
+  _symbol(symbol),
+  _accessing_klass(accessing_klass),
+  _size( method_type->ptype_slot_count()),
+  _count(method_type->ptype_count())
+{
+  ASSERT_IN_VM;
+  EXCEPTION_CONTEXT;
+  Arena* arena = CURRENT_ENV->arena();
+  _types = new (arena) GrowableArray<ciType*>(arena, _count + 1, 0, NULL);
+  for (int i = 0; i < _count; i++) {
+    _types->append(method_type->ptype_at(i));
+  }
+  _types->append(method_type->rtype());
+}
+
+// ------------------------------------------------------------------
 // ciSignature::return_type
 //
 // What is the return type of this signature?

--- a/src/share/vm/ci/ciSignature.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciSignature.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -47,6 +47,7 @@
   friend class ciObjectFactory;
 
   ciSignature(ciKlass* accessing_klass, constantPoolHandle cpool, ciSymbol* signature);
+  ciSignature(ciKlass* accessing_klass,                           ciSymbol* signature, ciMethodType* method_type);
 
   void get_all_klasses();
 

--- a/src/share/vm/ci/ciStreams.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciStreams.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -364,12 +364,15 @@
   constantPoolHandle cpool(_method->get_Method()->constants());
   ciMethod* m = env->get_method_by_index(cpool, get_method_index(), cur_bc(), _holder);
   will_link = m->is_loaded();
-  // Get declared method signature and return it.
-  if (has_optional_appendix()) {
-    const int sig_index = get_method_signature_index();
-    Symbol* sig_sym = cpool->symbol_at(sig_index);
-    ciKlass* pool_holder = env->get_klass(cpool->pool_holder());
-    (*declared_signature_result) = new (env->arena()) ciSignature(pool_holder, cpool, env->get_symbol(sig_sym));
+
+  // Use the MethodType stored in the CP cache to create a signature
+  // with correct types (in respect to class loaders).
+  if (has_method_type()) {
+    ciSymbol*     sig_sym     = env->get_symbol(cpool->symbol_at(get_method_signature_index()));
+    ciKlass*      pool_holder = env->get_klass(cpool->pool_holder());
+    ciMethodType* method_type = get_method_type();
+    ciSignature* declared_signature = new (env->arena()) ciSignature(pool_holder, sig_sym, method_type);
+    (*declared_signature_result) = declared_signature;
   } else {
     (*declared_signature_result) = m->signature();
   }
@@ -400,6 +403,31 @@
 }
 
 // ------------------------------------------------------------------
+// ciBytecodeStream::has_method_type
+//
+// Returns true if there is a MethodType argument stored in the
+// constant pool cache at the current bci.
+bool ciBytecodeStream::has_method_type() {
+  GUARDED_VM_ENTRY(
+    constantPoolHandle cpool(_method->get_Method()->constants());
+    return ConstantPool::has_method_type_at_if_loaded(cpool, get_method_index());
+  )
+}
+
+// ------------------------------------------------------------------
+// ciBytecodeStream::get_method_type
+//
+// Return the MethodType stored in the constant pool cache at
+// the current bci.
+ciMethodType* ciBytecodeStream::get_method_type() {
+  GUARDED_VM_ENTRY(
+    constantPoolHandle cpool(_method->get_Method()->constants());
+    oop method_type_oop = ConstantPool::method_type_at_if_loaded(cpool, get_method_index());
+    return CURRENT_ENV->get_object(method_type_oop)->as_method_type();
+  )
+}
+
+// ------------------------------------------------------------------
 // ciBytecodeStream::get_declared_method_holder
 //
 // Get the declared holder of the currently referenced method.

--- a/src/share/vm/ci/ciStreams.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciStreams.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -257,12 +257,14 @@
   int      get_field_holder_index();
   int      get_field_signature_index();
 
-  ciMethod* get_method(bool& will_link, ciSignature* *declared_signature_result);
-  bool      has_appendix();
-  ciObject* get_appendix();
-  ciKlass*  get_declared_method_holder();
-  int       get_method_holder_index();
-  int       get_method_signature_index();
+  ciMethod*     get_method(bool& will_link, ciSignature* *declared_signature_result);
+  bool          has_appendix();
+  ciObject*     get_appendix();
+  bool          has_method_type();
+  ciMethodType* get_method_type();
+  ciKlass*      get_declared_method_holder();
+  int           get_method_holder_index();
+  int           get_method_signature_index();
 
   // Get the resolved references arrays from the constant pool
   ciObjArray* get_resolved_references();

--- a/src/share/vm/ci/ciTypeFlow.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/ci/ciTypeFlow.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -2194,6 +2194,10 @@
     if (head->backedge_copy_count() != 0)
       continue;
 
+    // Don't clone head of OSR loop to get correct types in start block.
+    if (is_osr_flow() && head->start() == start_bci())
+      continue;
+
     // check _no_ shared head below us
     Loop* ch;
     for (ch = lp->child(); ch != NULL && ch->head() != head; ch = ch->sibling());

--- a/src/share/vm/classfile/systemDictionary.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/classfile/systemDictionary.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -2429,7 +2429,8 @@
 methodHandle SystemDictionary::find_method_handle_invoker(Symbol* name,
                                                           Symbol* signature,
                                                           KlassHandle accessing_klass,
-                                                          Handle* appendix_result,
+                                                          Handle *appendix_result,
+                                                          Handle *method_type_result,
                                                           TRAPS) {
   methodHandle empty;
   assert(EnableInvokeDynamic, "");
@@ -2461,6 +2462,7 @@
                          vmSymbols::linkMethod_signature(),
                          &args, CHECK_(empty));
   Handle mname(THREAD, (oop) result.get_jobject());
+  (*method_type_result) = method_type;
   return unpack_method_and_appendix(mname, appendix_box, appendix_result, THREAD);
 }
 
@@ -2607,7 +2609,8 @@
                                                               Handle bootstrap_specifier,
                                                               Symbol* name,
                                                               Symbol* type,
-                                                              Handle* appendix_result,
+                                                              Handle *appendix_result,
+                                                              Handle *method_type_result,
                                                               TRAPS) {
   methodHandle empty;
   Handle bsm, info;
@@ -2650,6 +2653,7 @@
                          vmSymbols::linkCallSite_signature(),
                          &args, CHECK_(empty));
   Handle mname(THREAD, (oop) result.get_jobject());
+  (*method_type_result) = method_type;
   return unpack_method_and_appendix(mname, appendix_box, appendix_result, THREAD);
 }
 

--- a/src/share/vm/classfile/systemDictionary.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/classfile/systemDictionary.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -497,6 +497,7 @@
                                                  Symbol* signature,
                                                  KlassHandle accessing_klass,
                                                  Handle *appendix_result,
+                                                 Handle *method_type_result,
                                                  TRAPS);
   // for a given signature, find the internal MethodHandle method (linkTo* or invokeBasic)
   // (does not ask Java, since this is a low-level intrinsic defined by the JVM)
@@ -523,6 +524,7 @@
                                                      Symbol* name,
                                                      Symbol* type,
                                                      Handle *appendix_result,
+                                                     Handle *method_type_result,
                                                      TRAPS);
 
   // Utility for printing loader "name" as part of tracing constraints

--- a/src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -1195,9 +1195,9 @@
   set_promo_size(desired_promo_size);
 }
 
-int CMSAdaptiveSizePolicy::compute_survivor_space_size_and_threshold(
+uint CMSAdaptiveSizePolicy::compute_survivor_space_size_and_threshold(
                                              bool is_survivor_overflow,
-                                             int tenuring_threshold,
+                                             uint tenuring_threshold,
                                              size_t survivor_limit) {
   assert(survivor_limit >= generation_alignment(),
          "survivor_limit too small");
@@ -1315,7 +1315,7 @@
 
     gclog_or_tty->print( "  avg_promoted_padded_avg: %f"
                 "  avg_pretenured_padded_avg: %f"
-                "  tenuring_thresh: %d"
+                "  tenuring_thresh: %u"
                 "  target_size: " SIZE_FORMAT
                 "  survivor_limit: " SIZE_FORMAT,
                 gch->gc_stats(1)->avg_promoted()->padded_average(),

--- a/src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/concurrentMarkSweep/cmsAdaptiveSizePolicy.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -440,9 +440,9 @@
                                                    size_t max_eden_size);
   // Calculates new survivor space size;  returns a new tenuring threshold
   // value. Stores new survivor size in _survivor_size.
-  virtual int compute_survivor_space_size_and_threshold(
+  virtual uint compute_survivor_space_size_and_threshold(
                                                 bool   is_survivor_overflow,
-                                                int    tenuring_threshold,
+                                                uint   tenuring_threshold,
                                                 size_t survivor_limit);
 
   virtual void compute_tenured_generation_free_space(size_t cur_tenured_free,

--- a/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -1188,29 +1188,14 @@
 // liveness counting data.
 class CMCountDataClosureBase: public HeapRegionClosure {
 protected:
+  G1CollectedHeap* _g1h;
   ConcurrentMark* _cm;
+  CardTableModRefBS* _ct_bs;
+
   BitMap* _region_bm;
   BitMap* _card_bm;
 
-  void set_card_bitmap_range(BitMap::idx_t start_idx, BitMap::idx_t last_idx) {
-    assert(start_idx <= last_idx, "sanity");
-
-    // Set the inclusive bit range [start_idx, last_idx].
-    // For small ranges (up to 8 cards) use a simple loop; otherwise
-    // use par_at_put_range.
-    if ((last_idx - start_idx) < 8) {
-      for (BitMap::idx_t i = start_idx; i <= last_idx; i += 1) {
-        _card_bm->par_set_bit(i);
-      }
-    } else {
-      assert(last_idx < _card_bm->size(), "sanity");
-      // Note BitMap::par_at_put_range() is exclusive.
-      BitMap::idx_t max_idx = MAX2(last_idx+1, _card_bm->size());
-      _card_bm->par_at_put_range(start_idx, max_idx, true);
-    }
-  }
-
-  // It takes a region that's not empty (i.e., it has at least one
+  // Takes a region that's not empty (i.e., it has at least one
   // live object in it and sets its corresponding bit on the region
   // bitmap to 1. If the region is "starts humongous" it will also set
   // to 1 the bits on the region bitmap that correspond to its
@@ -1231,9 +1216,11 @@
   }
 
 public:
-  CMCountDataClosureBase(ConcurrentMark *cm,
+  CMCountDataClosureBase(G1CollectedHeap* g1h,
                          BitMap* region_bm, BitMap* card_bm):
-    _cm(cm), _region_bm(region_bm), _card_bm(card_bm) { }
+    _g1h(g1h), _cm(g1h->concurrent_mark()),
+    _ct_bs((CardTableModRefBS*) (g1h->barrier_set())),
+    _region_bm(region_bm), _card_bm(card_bm) { }
 };
 
 // Closure that calculates the # live objects per region. Used
@@ -1243,9 +1230,9 @@
   size_t _region_marked_bytes;
 
 public:
-  CalcLiveObjectsClosure(CMBitMapRO *bm, ConcurrentMark *cm,
+  CalcLiveObjectsClosure(CMBitMapRO *bm, G1CollectedHeap* g1h,
                          BitMap* region_bm, BitMap* card_bm) :
-    CMCountDataClosureBase(cm, region_bm, card_bm),
+    CMCountDataClosureBase(g1h, region_bm, card_bm),
     _bm(bm), _region_marked_bytes(0) { }
 
   bool doHeapRegion(HeapRegion* hr) {
@@ -1261,44 +1248,63 @@
       return false;
     }
 
-    HeapWord* nextTop = hr->next_top_at_mark_start();
-    HeapWord* start   = hr->bottom();
-
-    assert(start <= hr->end() && start <= nextTop && nextTop <= hr->end(),
+    HeapWord* ntams = hr->next_top_at_mark_start();
+    HeapWord* start = hr->bottom();
+
+    assert(start <= hr->end() && start <= ntams && ntams <= hr->end(),
            err_msg("Preconditions not met - "
-                   "start: "PTR_FORMAT", nextTop: "PTR_FORMAT", end: "PTR_FORMAT,
-                   start, nextTop, hr->end()));
+                   "start: "PTR_FORMAT", ntams: "PTR_FORMAT", end: "PTR_FORMAT,
+                   start, ntams, hr->end()));
 
     // Find the first marked object at or after "start".
-    start = _bm->getNextMarkedWordAddress(start, nextTop);
+    start = _bm->getNextMarkedWordAddress(start, ntams);
 
     size_t marked_bytes = 0;
 
-    while (start < nextTop) {
+    while (start < ntams) {
       oop obj = oop(start);
       int obj_sz = obj->size();
-      HeapWord* obj_last = start + obj_sz - 1;
+      HeapWord* obj_end = start + obj_sz;
 
       BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
-      BitMap::idx_t last_idx = _cm->card_bitmap_index_for(obj_last);
-
-      // Set the bits in the card BM for this object (inclusive).
-      set_card_bitmap_range(start_idx, last_idx);
+      BitMap::idx_t end_idx = _cm->card_bitmap_index_for(obj_end);
+
+      // Note: if we're looking at the last region in heap - obj_end
+      // could be actually just beyond the end of the heap; end_idx
+      // will then correspond to a (non-existent) card that is also
+      // just beyond the heap.
+      if (_g1h->is_in_g1_reserved(obj_end) && !_ct_bs->is_card_aligned(obj_end)) {
+        // end of object is not card aligned - increment to cover
+        // all the cards spanned by the object
+        end_idx += 1;
+      }
+
+      // Set the bits in the card BM for the cards spanned by this object.
+      _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
 
       // Add the size of this object to the number of marked bytes.
       marked_bytes += (size_t)obj_sz * HeapWordSize;
 
       // Find the next marked object after this one.
-      start = _bm->getNextMarkedWordAddress(obj_last + 1, nextTop);
+      start = _bm->getNextMarkedWordAddress(obj_end, ntams);
     }
 
     // Mark the allocated-since-marking portion...
     HeapWord* top = hr->top();
-    if (nextTop < top) {
-      BitMap::idx_t start_idx = _cm->card_bitmap_index_for(nextTop);
-      BitMap::idx_t last_idx = _cm->card_bitmap_index_for(top - 1);
-
-      set_card_bitmap_range(start_idx, last_idx);
+    if (ntams < top) {
+      BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
+      BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
+
+      // Note: if we're looking at the last region in heap - top
+      // could be actually just beyond the end of the heap; end_idx
+      // will then correspond to a (non-existent) card that is also
+      // just beyond the heap.
+      if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
+        // end of object is not card aligned - increment to cover
+        // all the cards spanned by the object
+        end_idx += 1;
+      }
+      _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
 
       // This definitely means the region has live objects.
       set_bit_for_region(hr);
@@ -1325,6 +1331,7 @@
 // regions during the STW cleanup pause.
 
 class VerifyLiveObjectDataHRClosure: public HeapRegionClosure {
+  G1CollectedHeap* _g1h;
   ConcurrentMark* _cm;
   CalcLiveObjectsClosure _calc_cl;
   BitMap* _region_bm;   // Region BM to be verified
@@ -1337,14 +1344,14 @@
   int _failures;
 
 public:
-  VerifyLiveObjectDataHRClosure(ConcurrentMark* cm,
+  VerifyLiveObjectDataHRClosure(G1CollectedHeap* g1h,
                                 BitMap* region_bm,
                                 BitMap* card_bm,
                                 BitMap* exp_region_bm,
                                 BitMap* exp_card_bm,
                                 bool verbose) :
-    _cm(cm),
-    _calc_cl(_cm->nextMarkBitMap(), _cm, exp_region_bm, exp_card_bm),
+    _g1h(g1h), _cm(g1h->concurrent_mark()),
+    _calc_cl(_cm->nextMarkBitMap(), g1h, exp_region_bm, exp_card_bm),
     _region_bm(region_bm), _card_bm(card_bm), _verbose(verbose),
     _exp_region_bm(exp_region_bm), _exp_card_bm(exp_card_bm),
     _failures(0) { }
@@ -1491,7 +1498,7 @@
   void work(uint worker_id) {
     assert(worker_id < _n_workers, "invariant");
 
-    VerifyLiveObjectDataHRClosure verify_cl(_cm,
+    VerifyLiveObjectDataHRClosure verify_cl(_g1h,
                                             _actual_region_bm, _actual_card_bm,
                                             _expected_region_bm,
                                             _expected_card_bm,
@@ -1521,10 +1528,10 @@
 
 class FinalCountDataUpdateClosure: public CMCountDataClosureBase {
  public:
-  FinalCountDataUpdateClosure(ConcurrentMark* cm,
+  FinalCountDataUpdateClosure(G1CollectedHeap* g1h,
                               BitMap* region_bm,
                               BitMap* card_bm) :
-    CMCountDataClosureBase(cm, region_bm, card_bm) { }
+    CMCountDataClosureBase(g1h, region_bm, card_bm) { }
 
   bool doHeapRegion(HeapRegion* hr) {
 
@@ -1548,24 +1555,29 @@
     if (ntams < top) {
       // This definitely means the region has live objects.
       set_bit_for_region(hr);
-    }
-
-    // Now set the bits for [ntams, top]
-    BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
-    // set_card_bitmap_range() expects the last_idx to be with
-    // the range of the bit map (see assertion in set_card_bitmap_range()),
-    // so limit it to that range with this application of MIN2.
-    BitMap::idx_t last_idx = MIN2(_cm->card_bitmap_index_for(top),
-                                  _card_bm->size()-1);
-    if (start_idx < _card_bm->size()) {
-    set_card_bitmap_range(start_idx, last_idx);
-    } else {
-      // To reach here start_idx must be beyond the end of
-      // the bit map and last_idx must have been limited by
-      // the MIN2().
-      assert(start_idx == last_idx + 1,
-        err_msg("Not beyond end start_idx " SIZE_FORMAT " last_idx "
-                SIZE_FORMAT, start_idx, last_idx));
+
+      // Now set the bits in the card bitmap for [ntams, top)
+      BitMap::idx_t start_idx = _cm->card_bitmap_index_for(ntams);
+      BitMap::idx_t end_idx = _cm->card_bitmap_index_for(top);
+
+      // Note: if we're looking at the last region in heap - top
+      // could be actually just beyond the end of the heap; end_idx
+      // will then correspond to a (non-existent) card that is also
+      // just beyond the heap.
+      if (_g1h->is_in_g1_reserved(top) && !_ct_bs->is_card_aligned(top)) {
+        // end of object is not card aligned - increment to cover
+        // all the cards spanned by the object
+        end_idx += 1;
+      }
+
+      assert(end_idx <= _card_bm->size(),
+             err_msg("oob: end_idx=  "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
+                     end_idx, _card_bm->size()));
+      assert(start_idx < _card_bm->size(),
+             err_msg("oob: start_idx=  "SIZE_FORMAT", bitmap size= "SIZE_FORMAT,
+                     start_idx, _card_bm->size()));
+
+      _cm->set_card_bitmap_range(_card_bm, start_idx, end_idx, true /* is_par */);
     }
 
     // Set the bit for the region if it contains live data
@@ -1606,7 +1618,7 @@
   void work(uint worker_id) {
     assert(worker_id < _n_workers, "invariant");
 
-    FinalCountDataUpdateClosure final_update_cl(_cm,
+    FinalCountDataUpdateClosure final_update_cl(_g1h,
                                                 _actual_region_bm,
                                                 _actual_card_bm);
 
@@ -2846,20 +2858,19 @@
 // Aggregate the counting data that was constructed concurrently
 // with marking.
 class AggregateCountDataHRClosure: public HeapRegionClosure {
+  G1CollectedHeap* _g1h;
   ConcurrentMark* _cm;
+  CardTableModRefBS* _ct_bs;
   BitMap* _cm_card_bm;
   size_t _max_task_num;
 
  public:
-  AggregateCountDataHRClosure(ConcurrentMark *cm,
+  AggregateCountDataHRClosure(G1CollectedHeap* g1h,
                               BitMap* cm_card_bm,
                               size_t max_task_num) :
-    _cm(cm), _cm_card_bm(cm_card_bm),
-    _max_task_num(max_task_num) { }
-
-  bool is_card_aligned(HeapWord* p) {
-    return ((uintptr_t(p) & (CardTableModRefBS::card_size - 1)) == 0);
-  }
+    _g1h(g1h), _cm(g1h->concurrent_mark()),
+    _ct_bs((CardTableModRefBS*) (g1h->barrier_set())),
+    _cm_card_bm(cm_card_bm), _max_task_num(max_task_num) { }
 
   bool doHeapRegion(HeapRegion* hr) {
     if (hr->continuesHumongous()) {
@@ -2890,16 +2901,23 @@
       return false;
     }
 
-    assert(is_card_aligned(start), "sanity");
-    assert(is_card_aligned(end), "sanity");
+    // 'start' should be in the heap.
+    assert(_g1h->is_in_g1_reserved(start) && _ct_bs->is_card_aligned(start), "sanity");
+    // 'end' *may* be just beyone the end of the heap (if hr is the last region)
+    assert(!_g1h->is_in_g1_reserved(end) || _ct_bs->is_card_aligned(end), "sanity");
 
     BitMap::idx_t start_idx = _cm->card_bitmap_index_for(start);
     BitMap::idx_t limit_idx = _cm->card_bitmap_index_for(limit);
     BitMap::idx_t end_idx = _cm->card_bitmap_index_for(end);
 
-    // If ntams is not card aligned then we bump the index for
-    // limit so that we get the card spanning ntams.
-    if (!is_card_aligned(limit)) {
+    // If ntams is not card aligned then we bump card bitmap index
+    // for limit so that we get the all the cards spanned by
+    // the object ending at ntams.
+    // Note: if this is the last region in the heap then ntams
+    // could be actually just beyond the end of the the heap;
+    // limit_idx will then  correspond to a (non-existent) card
+    // that is also outside the heap.
+    if (_g1h->is_in_g1_reserved(limit) && !_ct_bs->is_card_aligned(limit)) {
       limit_idx += 1;
     }
 
@@ -2928,7 +2946,7 @@
 
         // BitMap::get_next_one_offset() can handle the case when
         // its left_offset parameter is greater than its right_offset
-        // parameter. If does, however, have an early exit if
+        // parameter. It does, however, have an early exit if
         // left_offset == right_offset. So let's limit the value
         // passed in for left offset here.
         BitMap::idx_t next_idx = MIN2(scan_idx + 1, limit_idx);
@@ -2964,7 +2982,7 @@
     _active_workers(n_workers) { }
 
   void work(uint worker_id) {
-    AggregateCountDataHRClosure cl(_cm, _cm_card_bm, _max_task_num);
+    AggregateCountDataHRClosure cl(_g1h, _cm_card_bm, _max_task_num);
 
     if (G1CollectedHeap::use_parallel_gc_threads()) {
       _g1h->heap_region_par_iterate_chunked(&cl, worker_id,

--- a/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -806,7 +806,14 @@
     return _MARKING_VERBOSE_ && _verbose_level >= high_verbose;
   }
 
-  // Counting data structure accessors
+  // Liveness counting
+
+  // Utility routine to set an exclusive range of cards on the given
+  // card liveness bitmap
+  inline void set_card_bitmap_range(BitMap* card_bm,
+                                    BitMap::idx_t start_idx,
+                                    BitMap::idx_t end_idx,
+                                    bool is_par);
 
   // Returns the card number of the bottom of the G1 heap.
   // Used in biasing indices into accounting card bitmaps.

--- a/src/share/vm/gc_implementation/g1/concurrentMark.inline.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.inline.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -28,6 +28,42 @@
 #include "gc_implementation/g1/concurrentMark.hpp"
 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp"
 
+// Utility routine to set an exclusive range of cards on the given
+// card liveness bitmap
+inline void ConcurrentMark::set_card_bitmap_range(BitMap* card_bm,
+                                                  BitMap::idx_t start_idx,
+                                                  BitMap::idx_t end_idx,
+                                                  bool is_par) {
+
+  // Set the exclusive bit range [start_idx, end_idx).
+  assert((end_idx - start_idx) > 0, "at least one card");
+  assert(end_idx <= card_bm->size(), "sanity");
+
+  // Silently clip the end index
+  end_idx = MIN2(end_idx, card_bm->size());
+
+  // For small ranges use a simple loop; otherwise use set_range or
+  // use par_at_put_range (if parallel). The range is made up of the
+  // cards that are spanned by an object/mem region so 8 cards will
+  // allow up to object sizes up to 4K to be handled using the loop.
+  if ((end_idx - start_idx) <= 8) {
+    for (BitMap::idx_t i = start_idx; i < end_idx; i += 1) {
+      if (is_par) {
+        card_bm->par_set_bit(i);
+      } else {
+        card_bm->set_bit(i);
+      }
+    }
+  } else {
+    // Note BitMap::par_at_put_range() and BitMap::set_range() are exclusive.
+    if (is_par) {
+      card_bm->par_at_put_range(start_idx, end_idx, true);
+    } else {
+      card_bm->set_range(start_idx, end_idx);
+    }
+  }
+}
+
 // Returns the index in the liveness accounting card bitmap
 // for the given address
 inline BitMap::idx_t ConcurrentMark::card_bitmap_index_for(HeapWord* addr) {
@@ -35,7 +71,6 @@
   // by the card shift -- address 0 corresponds to card number 0.  One
   // must subtract the card num of the bottom of the heap to obtain a
   // card table index.
-
   intptr_t card_num = intptr_t(uintptr_t(addr) >> CardTableModRefBS::card_shift);
   return card_num - heap_bottom_card_num();
 }
@@ -46,8 +81,10 @@
                                          size_t* marked_bytes_array,
                                          BitMap* task_card_bm) {
   G1CollectedHeap* g1h = _g1h;
+  CardTableModRefBS* ct_bs = (CardTableModRefBS*) (g1h->barrier_set());
+
   HeapWord* start = mr.start();
-  HeapWord* last = mr.last();
+  HeapWord* end = mr.end();
   size_t region_size_bytes = mr.byte_size();
   uint index = hr->hrs_index();
 
@@ -61,24 +98,21 @@
   marked_bytes_array[index] += region_size_bytes;
 
   BitMap::idx_t start_idx = card_bitmap_index_for(start);
-  BitMap::idx_t last_idx = card_bitmap_index_for(last);
+  BitMap::idx_t end_idx = card_bitmap_index_for(end);
 
-  // The card bitmap is task/worker specific => no need to use 'par' routines.
-  // Set bits in the inclusive bit range [start_idx, last_idx].
-  //
-  // For small ranges use a simple loop; otherwise use set_range
-  // The range are the cards that are spanned by the object/region
-  // so 8 cards will allow objects/regions up to 4K to be handled
-  // using the loop.
-  if ((last_idx - start_idx) <= 8) {
-    for (BitMap::idx_t i = start_idx; i <= last_idx; i += 1) {
-     task_card_bm->set_bit(i);
-    }
-  } else {
-    assert(last_idx < task_card_bm->size(), "sanity");
-    // Note: BitMap::set_range() is exclusive.
-    task_card_bm->set_range(start_idx, last_idx+1);
+  // Note: if we're looking at the last region in heap - end
+  // could be actually just beyond the end of the heap; end_idx
+  // will then correspond to a (non-existent) card that is also
+  // just beyond the heap.
+  if (g1h->is_in_g1_reserved(end) && !ct_bs->is_card_aligned(end)) {
+    // end of region is not card aligned - incremement to cover
+    // all the cards spanned by the region.
+    end_idx += 1;
   }
+  // The card bitmap is task/worker specific => no need to use
+  // the 'par' BitMap routines.
+  // Set bits in the exclusive bit range [start_idx, end_idx).
+  set_card_bitmap_range(task_card_bm, start_idx, end_idx, false /* is_par */);
 }
 
 // Counts the given memory region in the task/worker counting

--- a/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -4151,7 +4151,7 @@
   }
 }
 
-void G1CollectedHeap::release_gc_alloc_regions() {
+void G1CollectedHeap::release_gc_alloc_regions(uint no_of_gc_workers) {
   _survivor_gc_alloc_region.release();
   // If we have an old GC alloc region to release, we'll save it in
   // _retained_old_gc_alloc_region. If we don't
@@ -4161,8 +4161,8 @@
   _retained_old_gc_alloc_region = _old_gc_alloc_region.release();
 
   if (ResizePLAB) {
-    _survivor_plab_stats.adjust_desired_plab_sz();
-    _old_plab_stats.adjust_desired_plab_sz();
+    _survivor_plab_stats.adjust_desired_plab_sz(no_of_gc_workers);
+    _old_plab_stats.adjust_desired_plab_sz(no_of_gc_workers);
   }
 }
 
@@ -5427,7 +5427,7 @@
 };
 
 // Weak Reference processing during an evacuation pause (part 1).
-void G1CollectedHeap::process_discovered_references() {
+void G1CollectedHeap::process_discovered_references(uint no_of_gc_workers) {
   double ref_proc_start = os::elapsedTime();
 
   ReferenceProcessor* rp = _ref_processor_stw;
@@ -5454,15 +5454,14 @@
   // referents points to another object which is also referenced by an
   // object discovered by the STW ref processor.
 
-  uint active_workers = (G1CollectedHeap::use_parallel_gc_threads() ?
-                        workers()->active_workers() : 1);
-
   assert(!G1CollectedHeap::use_parallel_gc_threads() ||
-           active_workers == workers()->active_workers(),
-           "Need to reset active_workers");
-
-  set_par_threads(active_workers);
-  G1ParPreserveCMReferentsTask keep_cm_referents(this, active_workers, _task_queues);
+           no_of_gc_workers == workers()->active_workers(),
+           "Need to reset active GC workers");
+
+  set_par_threads(no_of_gc_workers);
+  G1ParPreserveCMReferentsTask keep_cm_referents(this,
+                                                 no_of_gc_workers,
+                                                 _task_queues);
 
   if (G1CollectedHeap::use_parallel_gc_threads()) {
     workers()->run_task(&keep_cm_referents);
@@ -5528,10 +5527,10 @@
                                       NULL);
   } else {
     // Parallel reference processing
-    assert(rp->num_q() == active_workers, "sanity");
-    assert(active_workers <= rp->max_num_q(), "sanity");
-
-    G1STWRefProcTaskExecutor par_task_executor(this, workers(), _task_queues, active_workers);
+    assert(rp->num_q() == no_of_gc_workers, "sanity");
+    assert(no_of_gc_workers <= rp->max_num_q(), "sanity");
+
+    G1STWRefProcTaskExecutor par_task_executor(this, workers(), _task_queues, no_of_gc_workers);
     rp->process_discovered_references(&is_alive, &keep_alive, &drain_queue, &par_task_executor);
   }
 
@@ -5546,7 +5545,7 @@
 }
 
 // Weak Reference processing during an evacuation pause (part 2).
-void G1CollectedHeap::enqueue_discovered_references() {
+void G1CollectedHeap::enqueue_discovered_references(uint no_of_gc_workers) {
   double ref_enq_start = os::elapsedTime();
 
   ReferenceProcessor* rp = _ref_processor_stw;
@@ -5560,13 +5559,12 @@
   } else {
     // Parallel reference enqueuing
 
-    uint active_workers = (ParallelGCThreads > 0 ? workers()->active_workers() : 1);
-    assert(active_workers == workers()->active_workers(),
-           "Need to reset active_workers");
-    assert(rp->num_q() == active_workers, "sanity");
-    assert(active_workers <= rp->max_num_q(), "sanity");
-
-    G1STWRefProcTaskExecutor par_task_executor(this, workers(), _task_queues, active_workers);
+    assert(no_of_gc_workers == workers()->active_workers(),
+           "Need to reset active workers");
+    assert(rp->num_q() == no_of_gc_workers, "sanity");
+    assert(no_of_gc_workers <= rp->max_num_q(), "sanity");
+
+    G1STWRefProcTaskExecutor par_task_executor(this, workers(), _task_queues, no_of_gc_workers);
     rp->enqueue_discovered_references(&par_task_executor);
   }
 
@@ -5658,7 +5656,7 @@
   // as we may have to copy some 'reachable' referent
   // objects (and their reachable sub-graphs) that were
   // not copied during the pause.
-  process_discovered_references();
+  process_discovered_references(n_workers);
 
   // Weak root processing.
   // Note: when JSR 292 is enabled and code blobs can contain
@@ -5670,7 +5668,7 @@
     JNIHandles::weak_oops_do(&is_alive, &keep_alive);
   }
 
-  release_gc_alloc_regions();
+  release_gc_alloc_regions(n_workers);
   g1_rem_set()->cleanup_after_oops_into_collection_set_do();
 
   concurrent_g1_refine()->clear_hot_cache();
@@ -5694,7 +5692,7 @@
   // will log these updates (and dirty their associated
   // cards). We need these updates logged to update any
   // RSets.
-  enqueue_discovered_references();
+  enqueue_discovered_references(n_workers);
 
   if (G1DeferredRSUpdate) {
     RedirtyLoggedCardTableEntryFastClosure redirty;

--- a/src/share/vm/gc_implementation/g1/g1CollectedHeap.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/g1CollectedHeap.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -326,7 +326,7 @@
   void init_gc_alloc_regions();
 
   // It releases the GC alloc regions at the end of a GC.
-  void release_gc_alloc_regions();
+  void release_gc_alloc_regions(uint no_of_gc_workers);
 
   // It does any cleanup that needs to be done on the GC alloc regions
   // before a Full GC.
@@ -652,11 +652,11 @@
 
   // Process any reference objects discovered during
   // an incremental evacuation pause.
-  void process_discovered_references();
+  void process_discovered_references(uint no_of_gc_workers);
 
   // Enqueue any remaining discovered references
   // after processing.
-  void enqueue_discovered_references();
+  void enqueue_discovered_references(uint no_of_gc_workers);
 
 public:
 

--- a/src/share/vm/gc_implementation/g1/g1CollectorPolicy.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/g1/g1CollectorPolicy.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -840,8 +840,8 @@
   //
 
   // Current tenuring threshold, set to 0 if the collector reaches the
-  // maximum amount of suvivors regions.
-  int _tenuring_threshold;
+  // maximum amount of survivors regions.
+  uint _tenuring_threshold;
 
   // The limit on the number of regions allocated for survivors.
   uint _max_survivor_regions;
@@ -851,7 +851,7 @@
   size_t _survivor_bytes_before_gc;
   size_t _capacity_before_gc;
 
-  // The amount of survor regions after a collection.
+  // The amount of survivor regions after a collection.
   uint _recorded_survivor_regions;
   // List of survivor regions.
   HeapRegion* _recorded_survivor_head;
@@ -862,7 +862,7 @@
 public:
 
   inline GCAllocPurpose
-    evacuation_destination(HeapRegion* src_region, int age, size_t word_sz) {
+    evacuation_destination(HeapRegion* src_region, uint age, size_t word_sz) {
       if (age < _tenuring_threshold && src_region->is_young()) {
         return GCAllocForSurvived;
       } else {

--- a/src/share/vm/gc_implementation/parNew/parNewGeneration.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parNew/parNewGeneration.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -1037,7 +1037,7 @@
 
   adjust_desired_tenuring_threshold();
   if (ResizePLAB) {
-    plab_stats()->adjust_desired_plab_sz();
+    plab_stats()->adjust_desired_plab_sz(n_workers);
   }
 
   if (PrintGC && !PrintGCDetails) {

--- a/src/share/vm/gc_implementation/parallelScavenge/cardTableExtension.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/cardTableExtension.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -121,103 +121,12 @@
 
 // We get passed the space_top value to prevent us from traversing into
 // the old_gen promotion labs, which cannot be safely parsed.
-void CardTableExtension::scavenge_contents(ObjectStartArray* start_array,
-                                           MutableSpace* sp,
-                                           HeapWord* space_top,
-                                           PSPromotionManager* pm)
-{
-  assert(start_array != NULL && sp != NULL && pm != NULL, "Sanity");
-  assert(start_array->covered_region().contains(sp->used_region()),
-         "ObjectStartArray does not cover space");
 
-  if (sp->not_empty()) {
-    oop* sp_top = (oop*)space_top;
-    oop* prev_top = NULL;
-    jbyte* current_card = byte_for(sp->bottom());
-    jbyte* end_card     = byte_for(sp_top - 1);    // sp_top is exclusive
-    // scan card marking array
-    while (current_card <= end_card) {
-      jbyte value = *current_card;
-      // skip clean cards
-      if (card_is_clean(value)) {
-        current_card++;
-      } else {
-        // we found a non-clean card
-        jbyte* first_nonclean_card = current_card++;
-        oop* bottom = (oop*)addr_for(first_nonclean_card);
-        // find object starting on card
-        oop* bottom_obj = (oop*)start_array->object_start((HeapWord*)bottom);
-        // bottom_obj = (oop*)start_array->object_start((HeapWord*)bottom);
-        assert(bottom_obj <= bottom, "just checking");
-        // make sure we don't scan oops we already looked at
-        if (bottom < prev_top) bottom = prev_top;
-        // figure out when to stop scanning
-        jbyte* first_clean_card;
-        oop* top;
-        bool restart_scanning;
-        do {
-          restart_scanning = false;
-          // find a clean card
-          while (current_card <= end_card) {
-            value = *current_card;
-            if (card_is_clean(value)) break;
-            current_card++;
-          }
-          // check if we reached the end, if so we are done
-          if (current_card >= end_card) {
-            first_clean_card = end_card + 1;
-            current_card++;
-            top = sp_top;
-          } else {
-            // we have a clean card, find object starting on that card
-            first_clean_card = current_card++;
-            top = (oop*)addr_for(first_clean_card);
-            oop* top_obj = (oop*)start_array->object_start((HeapWord*)top);
-            // top_obj = (oop*)start_array->object_start((HeapWord*)top);
-            assert(top_obj <= top, "just checking");
-            if (oop(top_obj)->is_objArray() || oop(top_obj)->is_typeArray()) {
-              // an arrayOop is starting on the clean card - since we do exact store
-              // checks for objArrays we are done
-            } else {
-              // otherwise, it is possible that the object starting on the clean card
-              // spans the entire card, and that the store happened on a later card.
-              // figure out where the object ends
-              top = top_obj + oop(top_obj)->size();
-              jbyte* top_card = CardTableModRefBS::byte_for(top - 1);   // top is exclusive
-              if (top_card > first_clean_card) {
-                // object ends a different card
-                current_card = top_card + 1;
-                if (card_is_clean(*top_card)) {
-                  // the ending card is clean, we are done
-                  first_clean_card = top_card;
-                } else {
-                  // the ending card is not clean, continue scanning at start of do-while
-                  restart_scanning = true;
-                }
-              } else {
-                // object ends on the clean card, we are done.
-                assert(first_clean_card == top_card, "just checking");
-              }
-            }
-          }
-        } while (restart_scanning);
-        // we know which cards to scan, now clear them
-        while (first_nonclean_card < first_clean_card) {
-          *first_nonclean_card++ = clean_card;
-        }
-        // scan oops in objects
-        do {
-          oop(bottom_obj)->push_contents(pm);
-          bottom_obj += oop(bottom_obj)->size();
-          assert(bottom_obj <= sp_top, "just checking");
-        } while (bottom_obj < top);
-        pm->drain_stacks_cond_depth();
-        // remember top oop* scanned
-        prev_top = top;
-      }
-    }
-  }
-}
+// Do not call this method if the space is empty.
+// It is a waste to start tasks and get here only to
+// do no work.  If this method needs to be called
+// when the space is empty, fix the calculation of
+// end_card to allow sp_top == sp->bottom().
 
 void CardTableExtension::scavenge_contents_parallel(ObjectStartArray* start_array,
                                                     MutableSpace* sp,
@@ -228,10 +137,11 @@
   int ssize = 128; // Naked constant!  Work unit = 64k.
   int dirty_card_count = 0;
 
+  // It is a waste to get here if empty.
+  assert(sp->bottom() < sp->top(), "Should not be called if empty");
   oop* sp_top = (oop*)space_top;
-  oop* sp_last = sp->bottom() == space_top ? sp_top : sp_top - 1;
   jbyte* start_card = byte_for(sp->bottom());
-  jbyte* end_card   = byte_for(sp_last) + 1;
+  jbyte* end_card   = byte_for(sp_top - 1) + 1;
   oop* last_scanned = NULL; // Prevent scanning objects more than once
   // The width of the stripe ssize*stripe_total must be
   // consistent with the number of stripes so that the complete slice
@@ -255,6 +165,16 @@
     HeapWord* slice_start = addr_for(worker_start_card);
     HeapWord* slice_end = MIN2((HeapWord*) sp_top, addr_for(worker_end_card));
 
+#ifdef ASSERT
+    if (GCWorkerDelayMillis > 0) {
+      // Delay 1 worker so that it proceeds after all the work
+      // has been completed.
+      if (stripe_number < 2) {
+        os::sleep(Thread::current(), GCWorkerDelayMillis, false);
+      }
+    }
+#endif
+
     // If there are not objects starting within the chunk, skip it.
     if (!start_array->object_starts_in_range(slice_start, slice_end)) {
       continue;

--- a/src/share/vm/gc_implementation/parallelScavenge/cardTableExtension.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/cardTableExtension.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -60,11 +60,6 @@
   // BarrierSet::Name kind() { return BarrierSet::CardTableExtension; }
 
   // Scavenge support
-  void scavenge_contents(ObjectStartArray* start_array,
-                         MutableSpace* sp,
-                         HeapWord* space_top,
-                         PSPromotionManager* pm);
-
   void scavenge_contents_parallel(ObjectStartArray* start_array,
                                   MutableSpace* sp,
                                   HeapWord* space_top,

--- a/src/share/vm/gc_implementation/parallelScavenge/psAdaptiveSizePolicy.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psAdaptiveSizePolicy.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -941,9 +941,9 @@
   return promo_heap_delta;
 }
 
-int PSAdaptiveSizePolicy::compute_survivor_space_size_and_threshold(
+uint PSAdaptiveSizePolicy::compute_survivor_space_size_and_threshold(
                                              bool is_survivor_overflow,
-                                             int tenuring_threshold,
+                                             uint tenuring_threshold,
                                              size_t survivor_limit) {
   assert(survivor_limit >= _intra_generation_alignment,
          "survivor_limit too small");

--- a/src/share/vm/gc_implementation/parallelScavenge/psAdaptiveSizePolicy.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psAdaptiveSizePolicy.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -353,9 +353,9 @@
 
   // Calculates new survivor space size;  returns a new tenuring threshold
   // value. Stores new survivor size in _survivor_size.
-  int compute_survivor_space_size_and_threshold(bool   is_survivor_overflow,
-                                                int    tenuring_threshold,
-                                                size_t survivor_limit);
+  uint compute_survivor_space_size_and_threshold(bool   is_survivor_overflow,
+                                                 uint    tenuring_threshold,
+                                                 size_t survivor_limit);
 
   // Return the maximum size of a survivor space if the young generation were of
   // size gen_size.

--- a/src/share/vm/gc_implementation/parallelScavenge/psPromotionManager.inline.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psPromotionManager.inline.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -85,7 +85,7 @@
 
     if (!promote_immediately) {
       // Find the objects age, MT safe.
-      int age = (test_mark->has_displaced_mark_helper() /* o->has_displaced_mark() */) ?
+      uint age = (test_mark->has_displaced_mark_helper() /* o->has_displaced_mark() */) ?
         test_mark->displaced_mark_helper()->age() : test_mark->age();
 
       // Try allocating obj in to-space (unless too old)
@@ -136,6 +136,13 @@
 
             HeapWord* lab_base = old_gen()->cas_allocate(OldPLABSize);
             if(lab_base != NULL) {
+#ifdef ASSERT
+              // Delay the initialization of the promotion lab (plab).
+              // This exposes uninitialized plabs to card table processing.
+              if (GCWorkerDelayMillis > 0) {
+                os::sleep(Thread::current(), GCWorkerDelayMillis, false);
+              }
+#endif
               _old_lab.initialize(MemRegion(lab_base, OldPLABSize));
               // Try the old lab allocation again.
               new_obj = (oop) _old_lab.allocate(new_obj_size);

--- a/src/share/vm/gc_implementation/parallelScavenge/psScavenge.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psScavenge.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -59,7 +59,7 @@
 ReferenceProcessor*        PSScavenge::_ref_processor = NULL;
 CardTableExtension*        PSScavenge::_card_table = NULL;
 bool                       PSScavenge::_survivor_overflow = false;
-int                        PSScavenge::_tenuring_threshold = 0;
+uint                       PSScavenge::_tenuring_threshold = 0;
 HeapWord*                  PSScavenge::_young_generation_boundary = NULL;
 elapsedTimer               PSScavenge::_accumulated_time;
 Stack<markOop, mtGC>       PSScavenge::_preserved_mark_stack;
@@ -395,9 +395,13 @@
 
       GCTaskQueue* q = GCTaskQueue::create();
 
-      uint stripe_total = active_workers;
-      for(uint i=0; i < stripe_total; i++) {
-        q->enqueue(new OldToYoungRootsTask(old_gen, old_top, i, stripe_total));
+      if (!old_gen->object_space()->is_empty()) {
+        // There are only old-to-young pointers if there are objects
+        // in the old gen.
+        uint stripe_total = active_workers;
+        for(uint i=0; i < stripe_total; i++) {
+          q->enqueue(new OldToYoungRootsTask(old_gen, old_top, i, stripe_total));
+        }
       }
 
       q->enqueue(new ScavengeRootsTask(ScavengeRootsTask::universe));
@@ -525,7 +529,7 @@
 
        if (PrintTenuringDistribution) {
          gclog_or_tty->cr();
-         gclog_or_tty->print_cr("Desired survivor size %ld bytes, new threshold %d (max %d)",
+         gclog_or_tty->print_cr("Desired survivor size %ld bytes, new threshold %u (max %u)",
                                 size_policy->calculated_survivor_size_in_bytes(),
                                 _tenuring_threshold, MaxTenuringThreshold);
        }

--- a/src/share/vm/gc_implementation/parallelScavenge/psScavenge.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psScavenge.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -66,14 +66,14 @@
   static PSIsAliveClosure    _is_alive_closure;     // Closure used for reference processing
   static CardTableExtension* _card_table;           // We cache the card table for fast access.
   static bool                _survivor_overflow;    // Overflow this collection
-  static int                 _tenuring_threshold;   // tenuring threshold for next scavenge
+  static uint                _tenuring_threshold;   // tenuring threshold for next scavenge
   static elapsedTimer        _accumulated_time;     // total time spent on scavenge
   static HeapWord*           _young_generation_boundary; // The lowest address possible for the young_gen.
                                                          // This is used to decide if an oop should be scavenged,
                                                          // cards should be marked, etc.
   static Stack<markOop, mtGC> _preserved_mark_stack; // List of marks to be restored after failed promotion
   static Stack<oop, mtGC>     _preserved_oop_stack;  // List of oops that need their mark restored.
-  static CollectorCounters*      _counters;         // collector performance counters
+  static CollectorCounters*      _counters;          // collector performance counters
   static bool                    _promotion_failed;
 
   static void clean_up_failed_promotion();
@@ -88,7 +88,7 @@
 
  public:
   // Accessors
-  static int              tenuring_threshold()  { return _tenuring_threshold; }
+  static uint             tenuring_threshold()  { return _tenuring_threshold; }
   static elapsedTimer*    accumulated_time()    { return &_accumulated_time; }
   static bool             promotion_failed()    { return _promotion_failed; }
   static int              consecutive_skipped_scavenges()

--- a/src/share/vm/gc_implementation/parallelScavenge/psTasks.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psTasks.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -165,35 +165,13 @@
 }
 
 //
-// SerialOldToYoungRootsTask
-//
-
-void SerialOldToYoungRootsTask::do_it(GCTaskManager* manager, uint which) {
-  assert(_gen != NULL, "Sanity");
-  assert(_gen->object_space()->contains(_gen_top) || _gen_top == _gen->object_space()->top(), "Sanity");
-
-  {
-    PSPromotionManager* pm = PSPromotionManager::gc_thread_promotion_manager(which);
-
-    assert(Universe::heap()->kind() == CollectedHeap::ParallelScavengeHeap, "Sanity");
-    CardTableExtension* card_table = (CardTableExtension *)Universe::heap()->barrier_set();
-    // FIX ME! Assert that card_table is the type we believe it to be.
-
-    card_table->scavenge_contents(_gen->start_array(),
-                                  _gen->object_space(),
-                                  _gen_top,
-                                  pm);
-
-    // Do the real work
-    pm->drain_stacks(false);
-  }
-}
-
-//
 // OldToYoungRootsTask
 //
 
 void OldToYoungRootsTask::do_it(GCTaskManager* manager, uint which) {
+  // There are not old-to-young pointers if the old gen is empty.
+  assert(!_gen->object_space()->is_empty(),
+    "Should not be called is there is no work");
   assert(_gen != NULL, "Sanity");
   assert(_gen->object_space()->contains(_gen_top) || _gen_top == _gen->object_space()->top(), "Sanity");
   assert(_stripe_number < ParallelGCThreads, "Sanity");

--- a/src/share/vm/gc_implementation/parallelScavenge/psTasks.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/parallelScavenge/psTasks.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -113,25 +113,6 @@
 };
 
 //
-// SerialOldToYoungRootsTask
-//
-// This task is used to scan for roots in the perm gen
-
-class SerialOldToYoungRootsTask : public GCTask {
- private:
-  PSOldGen* _gen;
-  HeapWord* _gen_top;
-
- public:
-  SerialOldToYoungRootsTask(PSOldGen *gen, HeapWord* gen_top) :
-    _gen(gen), _gen_top(gen_top) { }
-
-  char* name() { return (char *)"serial-old-to-young-roots-task"; }
-
-  virtual void do_it(GCTaskManager* manager, uint which);
-};
-
-//
 // OldToYoungRootsTask
 //
 // This task is used to scan old to young roots in parallel

--- a/src/share/vm/gc_implementation/shared/adaptiveSizePolicy.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/adaptiveSizePolicy.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -642,7 +642,7 @@
 
 bool AdaptiveSizePolicy::print_adaptive_size_policy_on(
                                             outputStream* st,
-                                            int tenuring_threshold_arg) const {
+                                            uint tenuring_threshold_arg) const {
   if (!AdaptiveSizePolicy::print_adaptive_size_policy_on(st)) {
     return false;
   }
@@ -663,7 +663,7 @@
     assert(!tenuring_threshold_change(), "(no change was attempted)");
   }
   if (tenuring_threshold_changed) {
-    st->print_cr("%d", tenuring_threshold_arg);
+    st->print_cr("%u", tenuring_threshold_arg);
   }
   return true;
 }

--- a/src/share/vm/gc_implementation/shared/adaptiveSizePolicy.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/adaptiveSizePolicy.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -489,8 +489,8 @@
 
   // Printing support
   virtual bool print_adaptive_size_policy_on(outputStream* st) const;
-  bool print_adaptive_size_policy_on(outputStream* st, int
-                                  tenuring_threshold) const;
+  bool print_adaptive_size_policy_on(outputStream* st,
+                                     uint tenuring_threshold) const;
 };
 
 // Class that can be used to print information about the

--- a/src/share/vm/gc_implementation/shared/ageTable.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/ageTable.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -78,10 +78,10 @@
   }
 }
 
-int ageTable::compute_tenuring_threshold(size_t survivor_capacity) {
+uint ageTable::compute_tenuring_threshold(size_t survivor_capacity) {
   size_t desired_survivor_size = (size_t)((((double) survivor_capacity)*TargetSurvivorRatio)/100);
   size_t total = 0;
-  int age = 1;
+  uint age = 1;
   assert(sizes[0] == 0, "no objects with age zero should be recorded");
   while (age < table_size) {
     total += sizes[age];
@@ -90,13 +90,13 @@
     if (total > desired_survivor_size) break;
     age++;
   }
-  int result = age < MaxTenuringThreshold ? age : MaxTenuringThreshold;
+  uint result = age < MaxTenuringThreshold ? age : MaxTenuringThreshold;
 
   if (PrintTenuringDistribution || UsePerfData) {
 
     if (PrintTenuringDistribution) {
       gclog_or_tty->cr();
-      gclog_or_tty->print_cr("Desired survivor size %ld bytes, new threshold %d (max %d)",
+      gclog_or_tty->print_cr("Desired survivor size %ld bytes, new threshold %u (max %u)",
         desired_survivor_size*oopSize, result, MaxTenuringThreshold);
     }
 
@@ -106,7 +106,7 @@
       total += sizes[age];
       if (sizes[age] > 0) {
         if (PrintTenuringDistribution) {
-          gclog_or_tty->print_cr("- age %3d: %10ld bytes, %10ld total",
+          gclog_or_tty->print_cr("- age %3u: %10ld bytes, %10ld total",
             age, sizes[age]*oopSize, total*oopSize);
         }
       }

--- a/src/share/vm/gc_implementation/shared/ageTable.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/ageTable.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -55,7 +55,7 @@
 
   // add entry
   void add(oop p, size_t oop_size) {
-    int age = p->age();
+    uint age = p->age();
     assert(age > 0 && age < table_size, "invalid age of object");
     sizes[age] += oop_size;
   }
@@ -66,7 +66,7 @@
   void merge_par(ageTable* subTable);
 
   // calculate new tenuring threshold based on age information
-  int compute_tenuring_threshold(size_t survivor_capacity);
+  uint compute_tenuring_threshold(size_t survivor_capacity);
 
  private:
   PerfVariable* _perf_sizes[table_size];

--- a/src/share/vm/gc_implementation/shared/gcAdaptivePolicyCounters.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/gcAdaptivePolicyCounters.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -188,7 +188,7 @@
   inline void update_survivor_overflowed(bool survivor_overflowed) {
     _survivor_overflowed_counter->set_value(survivor_overflowed);
   }
-  inline void update_tenuring_threshold(int threshold) {
+  inline void update_tenuring_threshold(uint threshold) {
     tenuring_threshold()->set_value(threshold);
   }
   inline void update_increment_tenuring_threshold_for_gc_cost() {

--- a/src/share/vm/gc_implementation/shared/parGCAllocBuffer.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/parGCAllocBuffer.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -87,7 +87,7 @@
 // Compute desired plab size and latch result for later
 // use. This should be called once at the end of parallel
 // scavenge; it clears the sensor accumulators.
-void PLABStats::adjust_desired_plab_sz() {
+void PLABStats::adjust_desired_plab_sz(uint no_of_gc_workers) {
   assert(ResizePLAB, "Not set");
   if (_allocated == 0) {
     assert(_unused == 0,
@@ -107,7 +107,7 @@
     target_refills = 1;
   }
   _used = _allocated - _wasted - _unused;
-  size_t plab_sz = _used/(target_refills*ParallelGCThreads);
+  size_t plab_sz = _used/(target_refills*no_of_gc_workers);
   if (PrintPLAB) gclog_or_tty->print(" (plab_sz = %d ", plab_sz);
   // Take historical weighted average
   _filter.sample(plab_sz);

--- a/src/share/vm/gc_implementation/shared/parGCAllocBuffer.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/parGCAllocBuffer.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -204,7 +204,8 @@
     return _desired_plab_sz;
   }
 
-  void adjust_desired_plab_sz(); // filter computation, latches output to
+  void adjust_desired_plab_sz(uint no_of_gc_workers);
+                                 // filter computation, latches output to
                                  // _desired_plab_sz, clears sensor accumulators
 
   void add_allocated(size_t v) {

--- a/src/share/vm/gc_implementation/shared/vmGCOperations.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/gc_implementation/shared/vmGCOperations.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -198,8 +198,6 @@
   CollectedHeap* heap = Universe::heap();
   GCCauseSetter gccs(heap, _gc_cause);
 
-  bool do_cms_concurrent = false;
-
   // Check again if the space is available.  Another thread
   // may have similarly failed a metadata allocation and induced
   // a GC that freed space for the allocation.
@@ -208,23 +206,25 @@
     }
 
   if (_result == NULL) {
-    if (!UseConcMarkSweepGC) {
-      // Don't clear the soft refs the first time.
+    if (UseConcMarkSweepGC) {
+      if (CMSClassUnloadingEnabled) {
+        MetaspaceGC::set_should_concurrent_collect(true);
+      }
+      // For CMS expand since the collection is going to be concurrent.
+      _result =
+        _loader_data->metaspace_non_null()->expand_and_allocate(_size, _mdtype);
+    }
+    if (_result == NULL) {
+      // Don't clear the soft refs.  This GC is for reclaiming metadata
+      // and is unrelated to the fullness of the Java heap which should
+      // be the criteria for clearing SoftReferences.
+      if (Verbose && PrintGCDetails && UseConcMarkSweepGC) {
+        gclog_or_tty->print_cr("\nCMS full GC for Metaspace");
+      }
       heap->collect_as_vm_thread(GCCause::_metadata_GC_threshold);
       _result = _loader_data->metaspace_non_null()->allocate(_size, _mdtype);
-      // Don't do this for now
-      // This seems too costly to do a second full GC
-      // Let the metaspace grow instead
-      // if (_result == NULL) {
-      //  // If allocation fails again, clear soft refs
-      //  heap->collect_as_vm_thread(GCCause::_last_ditch_collection);
-      //  _result = _loader_data->metaspace_non_null()->allocate(_size, _mdtype);
-      // }
-    } else {
-      MetaspaceGC::set_should_concurrent_collect(true);
-      do_cms_concurrent = true;
     }
-    if (_result == NULL) {
+    if (_result == NULL && !UseConcMarkSweepGC /* CMS already tried */) {
       // If still failing, allow the Metaspace to expand.
       // See delta_capacity_until_GC() for explanation of the
       // amount of the expansion.
@@ -233,18 +233,10 @@
       _result =
         _loader_data->metaspace_non_null()->expand_and_allocate(_size, _mdtype);
 
-      if (do_cms_concurrent && _result == NULL) {
-        // Rather than fail with a metaspace out-of-memory, do a full
-        // GC for CMS.
-        heap->collect_as_vm_thread(GCCause::_metadata_GC_threshold);
-        _result = _loader_data->metaspace_non_null()->allocate(_size, _mdtype);
-      }
-      if (_result == NULL) {
-        if (PrintGCDetails) {
-          gclog_or_tty->print_cr("\nAfter Metaspace GC failed to allocate size "
-                                 SIZE_FORMAT, _size);
-        }
-      }
+    }
+    if (Verbose && PrintGCDetails && _result == NULL) {
+      gclog_or_tty->print_cr("\nAfter Metaspace GC failed to allocate size "
+                             SIZE_FORMAT, _size);
     }
   }
 

--- a/src/share/vm/interpreter/interpreterRuntime.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/interpreter/interpreterRuntime.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -737,6 +737,7 @@
       pool,
       info.resolved_method(),
       info.resolved_appendix(),
+      info.resolved_method_type(),
       pool->resolved_references());
 }
 IRT_END
@@ -765,6 +766,7 @@
       pool,
       info.resolved_method(),
       info.resolved_appendix(),
+      info.resolved_method_type(),
       pool->resolved_references());
 }
 IRT_END

--- a/src/share/vm/interpreter/linkResolver.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/interpreter/linkResolver.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -99,7 +99,7 @@
   assert(!resolved_method->is_compiled_lambda_form(), "these must be handled via an invokehandle call");
 }
 
-void CallInfo::set_handle(methodHandle resolved_method, Handle resolved_appendix, TRAPS) {
+void CallInfo::set_handle(methodHandle resolved_method, Handle resolved_appendix, Handle resolved_method_type, TRAPS) {
   if (resolved_method.is_null()) {
     THROW_MSG(vmSymbols::java_lang_InternalError(), "resolved method is null");
   }
@@ -110,7 +110,8 @@
   int vtable_index = Method::nonvirtual_vtable_index;
   assert(resolved_method->vtable_index() == vtable_index, "");
   set_common(resolved_klass, resolved_klass, resolved_method, resolved_method, vtable_index, CHECK);
-  _resolved_appendix = resolved_appendix;
+  _resolved_appendix    = resolved_appendix;
+  _resolved_method_type = resolved_method_type;
 }
 
 void CallInfo::set_common(KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method, int vtable_index, TRAPS) {
@@ -221,7 +222,8 @@
 void LinkResolver::lookup_polymorphic_method(methodHandle& result,
                                              KlassHandle klass, Symbol* name, Symbol* full_signature,
                                              KlassHandle current_klass,
-                                             Handle* appendix_result_or_null,
+                                             Handle *appendix_result_or_null,
+                                             Handle *method_type_result,
                                              TRAPS) {
   vmIntrinsics::ID iid = MethodHandles::signature_polymorphic_name_id(name);
   if (TraceMethodHandles) {
@@ -275,10 +277,12 @@
       }
 
       Handle appendix;
+      Handle method_type;
       result = SystemDictionary::find_method_handle_invoker(name,
                                                             full_signature,
                                                             current_klass,
                                                             &appendix,
+                                                            &method_type,
                                                             CHECK);
       if (TraceMethodHandles) {
         tty->print("lookup_polymorphic_method => (via Java) ");
@@ -307,6 +311,7 @@
 
         assert(appendix_result_or_null != NULL, "");
         (*appendix_result_or_null) = appendix;
+        (*method_type_result)      = method_type;
         return;
       }
     }
@@ -419,7 +424,7 @@
     if (resolved_method.is_null()) {
       // JSR 292:  see if this is an implicitly generated method MethodHandle.linkToVirtual(*...), etc
       lookup_polymorphic_method(resolved_method, resolved_klass, method_name, method_signature,
-                                current_klass, (Handle*)NULL, THREAD);
+                                current_klass, (Handle*)NULL, (Handle*)NULL, THREAD);
       if (HAS_PENDING_EXCEPTION) {
         nested_exception = Handle(THREAD, PENDING_EXCEPTION);
         CLEAR_PENDING_EXCEPTION;
@@ -1207,11 +1212,12 @@
   assert(resolved_klass() == SystemDictionary::MethodHandle_klass(), "");
   assert(MethodHandles::is_signature_polymorphic_name(method_name), "");
   methodHandle resolved_method;
-  Handle resolved_appendix;
+  Handle       resolved_appendix;
+  Handle       resolved_method_type;
   lookup_polymorphic_method(resolved_method, resolved_klass,
                             method_name, method_signature,
-                            current_klass, &resolved_appendix, CHECK);
-  result.set_handle(resolved_method, resolved_appendix, CHECK);
+                            current_klass, &resolved_appendix, &resolved_method_type, CHECK);
+  result.set_handle(resolved_method, resolved_appendix, resolved_method_type, CHECK);
 }
 
 
@@ -1219,7 +1225,7 @@
   assert(EnableInvokeDynamic, "");
   pool->set_invokedynamic();    // mark header to flag active call sites
 
-  //resolve_pool(<resolved_klass>, method_name,  method_signature, current_klass, pool, index, CHECK);
+  //resolve_pool(<resolved_klass>, method_name, method_signature, current_klass, pool, index, CHECK);
   Symbol* method_name       = pool->name_ref_at(index);
   Symbol* method_signature  = pool->signature_ref_at(index);
   KlassHandle current_klass = KlassHandle(THREAD, pool->pool_holder());
@@ -1236,9 +1242,10 @@
     bootstrap_specifier = Handle(THREAD, bsm_info);
   }
   if (!cpce->is_f1_null()) {
-    methodHandle method(THREAD, cpce->f1_as_method());
-    Handle appendix(THREAD, cpce->appendix_if_resolved(pool));
-    result.set_handle(method, appendix, CHECK);
+    methodHandle method(     THREAD, cpce->f1_as_method());
+    Handle       appendix(   THREAD, cpce->appendix_if_resolved(pool));
+    Handle       method_type(THREAD, cpce->method_type_if_resolved(pool));
+    result.set_handle(method, appendix, method_type, CHECK);
     return;
   }
 
@@ -1260,11 +1267,13 @@
   // JSR 292:  this must resolve to an implicitly generated method MH.linkToCallSite(*...)
   // The appendix argument is likely to be a freshly-created CallSite.
   Handle       resolved_appendix;
+  Handle       resolved_method_type;
   methodHandle resolved_method =
     SystemDictionary::find_dynamic_call_site_invoker(current_klass,
                                                      bootstrap_specifier,
                                                      method_name, method_signature,
                                                      &resolved_appendix,
+                                                     &resolved_method_type,
                                                      THREAD);
   if (HAS_PENDING_EXCEPTION) {
     if (TraceMethodHandles) {
@@ -1284,7 +1293,7 @@
     CLEAR_PENDING_EXCEPTION;
     THROW_CAUSE(vmSymbols::java_lang_BootstrapMethodError(), nested_exception)
   }
-  result.set_handle(resolved_method, resolved_appendix, CHECK);
+  result.set_handle(resolved_method, resolved_appendix, resolved_method_type, CHECK);
 }
 
 //------------------------------------------------------------------------------------------------------------------------

--- a/src/share/vm/interpreter/linkResolver.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/interpreter/linkResolver.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -76,12 +76,13 @@
   methodHandle _selected_method;        // dynamic (actual) target method
   int          _vtable_index;           // vtable index of selected method
   Handle       _resolved_appendix;      // extra argument in constant pool (if CPCE::has_appendix)
+  Handle       _resolved_method_type;   // MethodType (for invokedynamic and invokehandle call sites)
 
-  void         set_static(   KlassHandle resolved_klass,                             methodHandle resolved_method                                                , TRAPS);
-  void         set_interface(KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method                  , TRAPS);
-  void         set_virtual(  KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method, int vtable_index, TRAPS);
-  void         set_handle(                                                           methodHandle resolved_method,   Handle resolved_appendix,                     TRAPS);
-  void         set_common(   KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method, int vtable_index, TRAPS);
+  void         set_static(   KlassHandle resolved_klass,                             methodHandle resolved_method                                                       , TRAPS);
+  void         set_interface(KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method                         , TRAPS);
+  void         set_virtual(  KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method, int vtable_index       , TRAPS);
+  void         set_handle(                                                           methodHandle resolved_method, Handle resolved_appendix, Handle resolved_method_type, TRAPS);
+  void         set_common(   KlassHandle resolved_klass, KlassHandle selected_klass, methodHandle resolved_method, methodHandle selected_method, int vtable_index       , TRAPS);
 
   friend class LinkResolver;
 
@@ -91,6 +92,7 @@
   methodHandle resolved_method() const           { return _resolved_method; }
   methodHandle selected_method() const           { return _selected_method; }
   Handle       resolved_appendix() const         { return _resolved_appendix; }
+  Handle       resolved_method_type() const      { return _resolved_method_type; }
 
   BasicType    result_type() const               { return selected_method()->result_type(); }
   bool         has_vtable_index() const          { return _vtable_index >= 0; }
@@ -113,7 +115,7 @@
   static void lookup_instance_method_in_klasses (methodHandle& result, KlassHandle klass, Symbol* name, Symbol* signature, TRAPS);
   static void lookup_method_in_interfaces       (methodHandle& result, KlassHandle klass, Symbol* name, Symbol* signature, TRAPS);
   static void lookup_polymorphic_method         (methodHandle& result, KlassHandle klass, Symbol* name, Symbol* signature,
-                                                 KlassHandle current_klass, Handle* appendix_result_or_null, TRAPS);
+                                                 KlassHandle current_klass, Handle *appendix_result_or_null, Handle *method_type_result, TRAPS);
 
   static int vtable_index_of_miranda_method(KlassHandle klass, Symbol* name, Symbol* signature, TRAPS);
 

--- a/src/share/vm/interpreter/rewriter.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/interpreter/rewriter.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -179,7 +179,7 @@
             MethodHandles::is_signature_polymorphic_name(SystemDictionary::MethodHandle_klass(),
                                                          _pool->name_ref_at(cp_index))) {
           // we may need a resolved_refs entry for the appendix
-          add_invokedynamic_resolved_references_entry(cp_index, cache_index);
+          add_invokedynamic_resolved_references_entries(cp_index, cache_index);
           status = +1;
         } else {
           status = -1;
@@ -211,7 +211,7 @@
   if (!reverse) {
     int cp_index = Bytes::get_Java_u2(p);
     int cache_index = add_invokedynamic_cp_cache_entry(cp_index);
-    add_invokedynamic_resolved_references_entry(cp_index, cache_index);
+    add_invokedynamic_resolved_references_entries(cp_index, cache_index);
     // Replace the trailing four bytes with a CPC index for the dynamic
     // call site.  Unlike other CPC entries, there is one per bytecode,
     // not just one per distinct CP entry.  In other words, the

--- a/src/share/vm/interpreter/rewriter.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/interpreter/rewriter.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -113,12 +113,19 @@
     return ref_index;
   }
 
-  // add a new entry to the resolved_references map (for invokedynamic only)
-  int add_invokedynamic_resolved_references_entry(int cp_index, int cache_index) {
+  // add a new entries to the resolved_references map (for invokedynamic and invokehandle only)
+  int add_invokedynamic_resolved_references_entries(int cp_index, int cache_index) {
     assert(_resolved_reference_limit >= 0, "must add indy refs after first iteration");
-    int ref_index = _resolved_references_map.append(cp_index);  // many-to-one
-    assert(ref_index >= _resolved_reference_limit, "");
-    _invokedynamic_references_map.at_put_grow(ref_index, cache_index, -1);
+    int ref_index = -1;
+    for (int entry = 0; entry < ConstantPoolCacheEntry::_indy_resolved_references_entries; entry++) {
+      const int index = _resolved_references_map.append(cp_index);  // many-to-one
+      assert(index >= _resolved_reference_limit, "");
+      if (entry == 0) {
+        ref_index = index;
+      }
+      assert((index - entry) == ref_index, "entries must be consecutive");
+      _invokedynamic_references_map.at_put_grow(index, cache_index, -1);
+    }
     return ref_index;
   }
 
@@ -127,15 +134,6 @@
     return cp_index;
   }
 
-  // invokedynamic support - append the cpCache entry (encoded) in object map.
-  // The resolved_references_map should still be in ascending order
-  // The resolved_references has the invokedynamic call site objects appended after
-  // the objects that are resolved in the constant pool.
-  int add_callsite_entry(int main_cpc_entry) {
-    int ref_index = _resolved_references_map.append(main_cpc_entry);
-    return ref_index;
-  }
-
   // Access the contents of _cp_cache_map to determine CP cache layout.
   int cp_cache_entry_pool_index(int cache_index) {
     int cp_index = _cp_cache_map[cache_index];

--- a/src/share/vm/memory/defNewGeneration.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/memory/defNewGeneration.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -43,7 +43,7 @@
 
 protected:
   Generation* _next_gen;
-  int         _tenuring_threshold;   // Tenuring threshold for next collection.
+  uint        _tenuring_threshold;   // Tenuring threshold for next collection.
   ageTable    _age_table;
   // Size of object to pretenure in words; command line provides bytes
   size_t        _pretenure_size_threshold_words;
@@ -325,7 +325,7 @@
                                 bool parallel = false);
 
   oop copy_to_survivor_space(oop old);
-  int tenuring_threshold() { return _tenuring_threshold; }
+  uint tenuring_threshold() { return _tenuring_threshold; }
 
   // Performance Counter support
   void update_counters();

--- a/src/share/vm/memory/threadLocalAllocBuffer.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/memory/threadLocalAllocBuffer.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -35,7 +35,7 @@
 // the threads for allocation.
 //            It is thread-private at any time, but maybe multiplexed over
 //            time across multiple threads. The park()/unpark() pair is
-//            used to make it avaiable for such multiplexing.
+//            used to make it available for such multiplexing.
 class ThreadLocalAllocBuffer: public CHeapObj<mtThread> {
   friend class VMStructs;
 private:

--- a/src/share/vm/oops/constantPool.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/constantPool.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -445,7 +445,6 @@
   return e->has_appendix();
 }
 
-
 oop ConstantPool::appendix_at_if_loaded(constantPoolHandle cpool, int which) {
   if (cpool->cache() == NULL)  return NULL;  // nothing to load yet
   int cache_index = decode_cpcache_index(which, true);
@@ -454,6 +453,21 @@
 }
 
 
+bool ConstantPool::has_method_type_at_if_loaded(constantPoolHandle cpool, int which) {
+  if (cpool->cache() == NULL)  return false;  // nothing to load yet
+  int cache_index = decode_cpcache_index(which, true);
+  ConstantPoolCacheEntry* e = cpool->cache()->entry_at(cache_index);
+  return e->has_method_type();
+}
+
+oop ConstantPool::method_type_at_if_loaded(constantPoolHandle cpool, int which) {
+  if (cpool->cache() == NULL)  return NULL;  // nothing to load yet
+  int cache_index = decode_cpcache_index(which, true);
+  ConstantPoolCacheEntry* e = cpool->cache()->entry_at(cache_index);
+  return e->method_type_if_resolved(cpool);
+}
+
+
 Symbol* ConstantPool::impl_name_ref_at(int which, bool uncached) {
   int name_index = name_ref_index_at(impl_name_and_type_ref_index_at(which, uncached));
   return symbol_at(name_index);

--- a/src/share/vm/oops/constantPool.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/constantPool.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -694,13 +694,15 @@
   friend class SystemDictionary;
 
   // Used by compiler to prevent classloading.
-  static Method*         method_at_if_loaded      (constantPoolHandle this_oop, int which);
-  static bool      has_appendix_at_if_loaded      (constantPoolHandle this_oop, int which);
-  static oop           appendix_at_if_loaded      (constantPoolHandle this_oop, int which);
-  static Klass*           klass_at_if_loaded      (constantPoolHandle this_oop, int which);
-  static Klass*       klass_ref_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static Method*          method_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static bool       has_appendix_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static oop            appendix_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static bool    has_method_type_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static oop         method_type_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static Klass*            klass_at_if_loaded      (constantPoolHandle this_oop, int which);
+  static Klass*        klass_ref_at_if_loaded      (constantPoolHandle this_oop, int which);
   // Same as above - but does LinkResolving.
-  static Klass*       klass_ref_at_if_loaded_check(constantPoolHandle this_oop, int which, TRAPS);
+  static Klass*        klass_ref_at_if_loaded_check(constantPoolHandle this_oop, int which, TRAPS);
 
   // Routines currently used for annotations (only called by jvm.cpp) but which might be used in the
   // future by other Java code. These take constant pool indices rather than

--- a/src/share/vm/oops/cpCache.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/cpCache.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -244,21 +244,23 @@
 
 
 void ConstantPoolCacheEntry::set_method_handle(constantPoolHandle cpool,
-                                               methodHandle adapter, Handle appendix,
+                                               methodHandle adapter,
+                                               Handle appendix, Handle method_type,
                                                objArrayHandle resolved_references) {
-  set_method_handle_common(cpool, Bytecodes::_invokehandle, adapter, appendix, resolved_references);
+  set_method_handle_common(cpool, Bytecodes::_invokehandle, adapter, appendix, method_type, resolved_references);
 }
 
 void ConstantPoolCacheEntry::set_dynamic_call(constantPoolHandle cpool,
-                                              methodHandle adapter, Handle appendix,
+                                              methodHandle adapter,
+                                              Handle appendix, Handle method_type,
                                               objArrayHandle resolved_references) {
-  set_method_handle_common(cpool, Bytecodes::_invokedynamic, adapter, appendix, resolved_references);
+  set_method_handle_common(cpool, Bytecodes::_invokedynamic, adapter, appendix, method_type, resolved_references);
 }
 
 void ConstantPoolCacheEntry::set_method_handle_common(constantPoolHandle cpool,
                                                       Bytecodes::Code invoke_code,
                                                       methodHandle adapter,
-                                                      Handle appendix,
+                                                      Handle appendix, Handle method_type,
                                                       objArrayHandle resolved_references) {
   // NOTE: This CPCE can be the subject of data races.
   // There are three words to update: flags, refs[f2], f1 (in that order).
@@ -274,18 +276,21 @@
     return;
   }
 
-  bool has_appendix = appendix.not_null();
+  const bool has_appendix    = appendix.not_null();
+  const bool has_method_type = method_type.not_null();
 
   // Write the flags.
   set_method_flags(as_TosState(adapter->result_type()),
-                   ((has_appendix ?  1 : 0) << has_appendix_shift) |
-                   (                 1      << is_final_shift),
+                   ((has_appendix    ? 1 : 0) << has_appendix_shift   ) |
+                   ((has_method_type ? 1 : 0) << has_method_type_shift) |
+                   (                   1      << is_final_shift       ),
                    adapter->size_of_parameters());
 
   if (TraceInvokeDynamic) {
-    tty->print_cr("set_method_handle bc=%d appendix="PTR_FORMAT"%s method="PTR_FORMAT" ",
+    tty->print_cr("set_method_handle bc=%d appendix="PTR_FORMAT"%s method_type="PTR_FORMAT"%s method="PTR_FORMAT" ",
                   invoke_code,
-                  (intptr_t)appendix(), (has_appendix ? "" : " (unused)"),
+                  (intptr_t)appendix(),    (has_appendix    ? "" : " (unused)"),
+                  (intptr_t)method_type(), (has_method_type ? "" : " (unused)"),
                   (intptr_t)adapter());
     adapter->print();
     if (has_appendix)  appendix()->print();
@@ -310,17 +315,26 @@
   // This allows us to create fewer method oops, while keeping type safety.
   //
 
+  // Store appendix, if any.
   if (has_appendix) {
-    int ref_index = f2_as_index();
-    assert(ref_index >= 0 && ref_index < resolved_references->length(), "oob");
-    assert(resolved_references->obj_at(ref_index) == NULL, "init just once");
-    resolved_references->obj_at_put(ref_index, appendix());
+    const int appendix_index = f2_as_index() + _indy_resolved_references_appendix_offset;
+    assert(appendix_index >= 0 && appendix_index < resolved_references->length(), "oob");
+    assert(resolved_references->obj_at(appendix_index) == NULL, "init just once");
+    resolved_references->obj_at_put(appendix_index, appendix());
+  }
+
+  // Store MethodType, if any.
+  if (has_method_type) {
+    const int method_type_index = f2_as_index() + _indy_resolved_references_method_type_offset;
+    assert(method_type_index >= 0 && method_type_index < resolved_references->length(), "oob");
+    assert(resolved_references->obj_at(method_type_index) == NULL, "init just once");
+    resolved_references->obj_at_put(method_type_index, method_type());
   }
 
   release_set_f1(adapter());  // This must be the last one to set (see NOTE above)!
 
-    // The interpreter assembly code does not check byte_2,
-    // but it is used by is_resolved, method_if_resolved, etc.
+  // The interpreter assembly code does not check byte_2,
+  // but it is used by is_resolved, method_if_resolved, etc.
   set_bytecode_1(invoke_code);
   NOT_PRODUCT(verify(tty));
   if (TraceInvokeDynamic) {
@@ -376,7 +390,16 @@
 oop ConstantPoolCacheEntry::appendix_if_resolved(constantPoolHandle cpool) {
   if (is_f1_null() || !has_appendix())
     return NULL;
-  int ref_index = f2_as_index();
+  const int ref_index = f2_as_index() + _indy_resolved_references_appendix_offset;
+  objArrayOop resolved_references = cpool->resolved_references();
+  return resolved_references->obj_at(ref_index);
+}
+
+
+oop ConstantPoolCacheEntry::method_type_if_resolved(constantPoolHandle cpool) {
+  if (is_f1_null() || !has_method_type())
+    return NULL;
+  const int ref_index = f2_as_index() + _indy_resolved_references_method_type_offset;
   objArrayOop resolved_references = cpool->resolved_references();
   return resolved_references->obj_at(ref_index);
 }
@@ -513,13 +536,23 @@
   for (int i = 0; i < length(); i++) {
     ConstantPoolCacheEntry* e = entry_at(i);
     int original_index = inverse_index_map[i];
-      e->initialize_entry(original_index);
+    e->initialize_entry(original_index);
     assert(entry_at(i) == e, "sanity");
-    }
+  }
   for (int ref = 0; ref < invokedynamic_references_map.length(); ref++) {
-    int cpci = invokedynamic_references_map[ref];
-    if (cpci >= 0)
+    const int cpci = invokedynamic_references_map[ref];
+    if (cpci >= 0) {
+#ifdef ASSERT
+      // invokedynamic and invokehandle have more entries; check if they
+      // all point to the same constant pool cache entry.
+      for (int entry = 1; entry < ConstantPoolCacheEntry::_indy_resolved_references_entries; entry++) {
+        const int cpci_next = invokedynamic_references_map[ref + entry];
+        assert(cpci == cpci_next, err_msg_res("%d == %d", cpci, cpci_next));
+      }
+#endif
       entry_at(cpci)->initialize_resolved_reference_index(ref);
+      ref += ConstantPoolCacheEntry::_indy_resolved_references_entries - 1;  // skip extra entries
+    }
   }
 }
 

--- a/src/share/vm/oops/cpCache.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/cpCache.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -42,10 +42,10 @@
 // _indices   [ b2 | b1 |  index  ]  index = constant_pool_index
 // _f1        [  entry specific   ]  metadata ptr (method or klass)
 // _f2        [  entry specific   ]  vtable or res_ref index, or vfinal method ptr
-// _flags     [tos|0|F=1|0|0|f|v|0 |00000|field_index] (for field entries)
-// bit length [ 4 |1| 1 |1|1|1|1|1 |--5--|----16-----]
-// _flags     [tos|0|F=0|A|I|f|0|vf|00000|00000|psize] (for method entries)
-// bit length [ 4 |1| 1 |1|1|1|1|1 |--5--|--8--|--8--]
+// _flags     [tos|0|F=1|0|0|0|f|v|0 |0000|field_index] (for field entries)
+// bit length [ 4 |1| 1 |1|1|1|1|1|1 |-4--|----16-----]
+// _flags     [tos|0|F=0|M|A|I|f|0|vf|0000|00000|psize] (for method entries)
+// bit length [ 4 |1| 1 |1|1|1|1|1|1 |-4--|--8--|--8--]
 
 // --------------------------------
 //
@@ -166,11 +166,12 @@
     tos_state_shift            = BitsPerInt - tos_state_bits,  // see verify_tos_state_shift below
     // misc. option bits; can be any bit position in [16..27]
     is_field_entry_shift       = 26,  // (F) is it a field or a method?
-    has_appendix_shift         = 25,  // (A) does the call site have an appendix argument?
-    is_forced_virtual_shift    = 24,  // (I) is the interface reference forced to virtual mode?
-    is_final_shift             = 23,  // (f) is the field or method final?
-    is_volatile_shift          = 22,  // (v) is the field volatile?
-    is_vfinal_shift            = 21,  // (vf) did the call resolve to a final method?
+    has_method_type_shift      = 25,  // (M) does the call site have a MethodType?
+    has_appendix_shift         = 24,  // (A) does the call site have an appendix argument?
+    is_forced_virtual_shift    = 23,  // (I) is the interface reference forced to virtual mode?
+    is_final_shift             = 22,  // (f) is the field or method final?
+    is_volatile_shift          = 21,  // (v) is the field volatile?
+    is_vfinal_shift            = 20,  // (vf) did the call resolve to a final method?
     // low order bits give field index (for FieldInfo) or method parameter size:
     field_index_bits           = 16,
     field_index_mask           = right_n_bits(field_index_bits),
@@ -223,14 +224,16 @@
   void set_method_handle(
     constantPoolHandle cpool,                    // holding constant pool (required for locking)
     methodHandle method,                         // adapter for invokeExact, etc.
-    Handle appendix,                             // stored in refs[f2]; could be a java.lang.invoke.MethodType
+    Handle appendix,                             // stored in refs[f2+0]; could be a java.lang.invoke.MethodType
+    Handle method_type,                          // stored in refs[f2+1]; is a java.lang.invoke.MethodType
     objArrayHandle resolved_references
   );
 
   void set_dynamic_call(
     constantPoolHandle cpool,                    // holding constant pool (required for locking)
     methodHandle method,                         // adapter for this call site
-    Handle appendix,                             // stored in refs[f2]; could be a java.lang.invoke.CallSite
+    Handle appendix,                             // stored in refs[f2+0]; could be a java.lang.invoke.CallSite
+    Handle method_type,                          // stored in refs[f2+1]; is a java.lang.invoke.MethodType
     objArrayHandle resolved_references
   );
 
@@ -253,12 +256,24 @@
     constantPoolHandle cpool,                    // holding constant pool (required for locking)
     Bytecodes::Code invoke_code,                 // _invokehandle or _invokedynamic
     methodHandle adapter,                        // invoker method (f1)
-    Handle appendix,                             // appendix such as CallSite, MethodType, etc. (refs[f2])
+    Handle appendix,                             // appendix such as CallSite, MethodType, etc. (refs[f2+0])
+    Handle method_type,                          // MethodType (refs[f2+1])
     objArrayHandle resolved_references
   );
 
-  Method* method_if_resolved(constantPoolHandle cpool);
-  oop appendix_if_resolved(constantPoolHandle cpool);
+  // invokedynamic and invokehandle call sites have two entries in the
+  // resolved references array:
+  //   appendix   (at index+0)
+  //   MethodType (at index+1)
+  enum {
+    _indy_resolved_references_appendix_offset    = 0,
+    _indy_resolved_references_method_type_offset = 1,
+    _indy_resolved_references_entries
+  };
+
+  Method*      method_if_resolved(constantPoolHandle cpool);
+  oop        appendix_if_resolved(constantPoolHandle cpool);
+  oop     method_type_if_resolved(constantPoolHandle cpool);
 
   void set_parameter_size(int value);
 
@@ -270,11 +285,11 @@
       case Bytecodes::_getfield        :    // fall through
       case Bytecodes::_invokespecial   :    // fall through
       case Bytecodes::_invokestatic    :    // fall through
+      case Bytecodes::_invokehandle    :    // fall through
+      case Bytecodes::_invokedynamic   :    // fall through
       case Bytecodes::_invokeinterface : return 1;
       case Bytecodes::_putstatic       :    // fall through
       case Bytecodes::_putfield        :    // fall through
-      case Bytecodes::_invokehandle    :    // fall through
-      case Bytecodes::_invokedynamic   :    // fall through
       case Bytecodes::_invokevirtual   : return 2;
       default                          : break;
     }
@@ -307,6 +322,7 @@
   bool is_forced_virtual() const                 { return (_flags & (1 << is_forced_virtual_shift)) != 0; }
   bool is_vfinal() const                         { return (_flags & (1 << is_vfinal_shift))         != 0; }
   bool has_appendix() const                      { return (_flags & (1 << has_appendix_shift))      != 0; }
+  bool has_method_type() const                   { return (_flags & (1 << has_method_type_shift))   != 0; }
   bool is_method_entry() const                   { return (_flags & (1 << is_field_entry_shift))    == 0; }
   bool is_field_entry() const                    { return (_flags & (1 << is_field_entry_shift))    != 0; }
   bool is_byte() const                           { return flag_state() == btos; }

--- a/src/share/vm/oops/markOop.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/markOop.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -318,7 +318,7 @@
     intptr_t tmp = (intptr_t) monitor;
     return (markOop) (tmp | monitor_value);
   }
-  static markOop encode(JavaThread* thread, int age, int bias_epoch) {
+  static markOop encode(JavaThread* thread, uint age, int bias_epoch) {
     intptr_t tmp = (intptr_t) thread;
     assert(UseBiasedLocking && ((tmp & (epoch_mask_in_place | age_mask_in_place | biased_lock_mask_in_place)) == 0), "misaligned JavaThread pointer");
     assert(age <= max_age, "age too large");
@@ -333,10 +333,10 @@
   markOop set_marked()   { return markOop((value() & ~lock_mask_in_place) | marked_value); }
   markOop set_unmarked() { return markOop((value() & ~lock_mask_in_place) | unlocked_value); }
 
-  int     age()               const { return mask_bits(value() >> age_shift, age_mask); }
-  markOop set_age(int v) const {
+  uint    age()               const { return mask_bits(value() >> age_shift, age_mask); }
+  markOop set_age(uint v) const {
     assert((v & ~age_mask) == 0, "shouldn't overflow age field");
-    return markOop((value() & ~age_mask_in_place) | (((intptr_t)v & age_mask) << age_shift));
+    return markOop((value() & ~age_mask_in_place) | (((uintptr_t)v & age_mask) << age_shift));
   }
   markOop incr_age()          const { return age() == max_age ? markOop(this) : set_age(age() + 1); }
 

--- a/src/share/vm/oops/objArrayKlass.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/objArrayKlass.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -646,7 +646,7 @@
   }
   int remaining = oa->length() - print_len;
   if (remaining > 0) {
-    tty->print_cr(" - <%d more elements, increase MaxElementPrintSize to print>", remaining);
+    st->print_cr(" - <%d more elements, increase MaxElementPrintSize to print>", remaining);
   }
 }
 

--- a/src/share/vm/oops/oop.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/oop.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -327,7 +327,7 @@
   oop forwardee() const;
 
   // Age of object during scavenge
-  int age() const;
+  uint age() const;
   void incr_age();
 
   // Adjust all pointers in this object to point at it's forwarded location and

--- a/src/share/vm/oops/oop.inline.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/oop.inline.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -693,7 +693,7 @@
 }
 
 // The following method needs to be MT safe.
-inline int oopDesc::age() const {
+inline uint oopDesc::age() const {
   assert(!is_forwarded(), "Attempt to read age from forwarded mark");
   if (has_displaced_mark()) {
     return displaced_mark()->age();

--- a/src/share/vm/oops/typeArrayKlass.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/oops/typeArrayKlass.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -381,7 +381,7 @@
   }
   int remaining = ta->length() - print_len;
   if (remaining > 0) {
-    tty->print_cr(" - <%d more elements, increase MaxElementPrintSize to print>", remaining);
+    st->print_cr(" - <%d more elements, increase MaxElementPrintSize to print>", remaining);
   }
 }
 

--- a/src/share/vm/opto/classes.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/classes.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -268,6 +268,8 @@
 macro(MulVD)
 macro(DivVF)
 macro(DivVD)
+macro(LShiftCntV)
+macro(RShiftCntV)
 macro(LShiftVB)
 macro(LShiftVS)
 macro(LShiftVI)

--- a/src/share/vm/opto/matcher.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/matcher.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -261,6 +261,7 @@
 
   // Vector ideal reg
   static const int vector_ideal_reg(int len);
+  static const int vector_shift_count_ideal_reg(int len);
 
   // CPU supports misaligned vectors store/load.
   static const bool misaligned_vectors_ok();

--- a/src/share/vm/opto/parse1.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/parse1.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -372,7 +372,8 @@
     // in the CFG, which typeflow had previously ignored.
     // E.g., Object x = coldAtFirst() && notReached()? "str": new Integer(123).
     // This x will be typed as Integer if notReached is not yet linked.
-    uncommon_trap(Deoptimization::Reason_unreached,
+    // It could also happen due to a problem in ciTypeFlow analysis.
+    uncommon_trap(Deoptimization::Reason_constraint,
                   Deoptimization::Action_reinterpret);
     set_map(types_are_good);
   }

--- a/src/share/vm/opto/superword.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/superword.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -1436,10 +1436,9 @@
       return opd; // input is matching vector
     }
     if ((opd_idx == 2) && VectorNode::is_shift(p0)) {
-      // No vector is needed for shift count.
-      // Vector instructions do not mask shift count, do it here.
       Compile* C = _phase->C;
       Node* cnt = opd;
+      // Vector instructions do not mask shift count, do it here.
       juint mask = (p0->bottom_type() == TypeInt::INT) ? (BitsPerInt - 1) : (BitsPerLong - 1);
       const TypeInt* t = opd->find_int_type();
       if (t != NULL && t->is_con()) {
@@ -1456,8 +1455,8 @@
           _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 (C) MoveI2FNode(cnt);
+        // Move non constant shift count into vector register.
+        cnt = VectorNode::shift_count(C, p0, cnt, vlen, velt_basic_type(p0));
       }
       if (cnt != opd) {
         _igvn.register_new_node_with_optimizer(cnt);

--- a/src/share/vm/opto/vectornode.cpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/vectornode.cpp	Fri Oct 05 13:28:16 2012 -0700
@@ -243,6 +243,8 @@
 VectorNode* VectorNode::make(Compile* C, int opc, Node* n1, Node* n2, uint vlen, BasicType bt) {
   const TypeVect* vt = TypeVect::make(bt, vlen);
   int vopc = VectorNode::opcode(opc, bt);
+  // This method should not be called for unimplemented vectors.
+  guarantee(vopc > 0, err_msg_res("Vector for '%s' is not implemented", NodeClassNames[opc]));
 
   switch (vopc) {
   case Op_AddVB: return new (C) AddVBNode(n1, n2, vt);
@@ -286,7 +288,7 @@
   case Op_OrV:  return new (C) OrVNode (n1, n2, vt);
   case Op_XorV: return new (C) XorVNode(n1, n2, vt);
   }
-  ShouldNotReachHere();
+  fatal(err_msg_res("Missed vector creation for '%s'", NodeClassNames[vopc]));
   return NULL;
 
 }
@@ -312,7 +314,25 @@
   case T_DOUBLE:
     return new (C) ReplicateDNode(s, vt);
   }
-  ShouldNotReachHere();
+  fatal(err_msg_res("Type '%s' is not supported for vectors", type2name(bt)));
+  return NULL;
+}
+
+VectorNode* VectorNode::shift_count(Compile* C, Node* shift, Node* cnt, uint vlen, BasicType bt) {
+  assert(VectorNode::is_shift(shift) && !cnt->is_Con(), "only variable shift count");
+  // Match shift count type with shift vector type.
+  const TypeVect* vt = TypeVect::make(bt, vlen);
+  switch (shift->Opcode()) {
+  case Op_LShiftI:
+  case Op_LShiftL:
+    return new (C) LShiftCntVNode(cnt, vt);
+  case Op_RShiftI:
+  case Op_RShiftL:
+  case Op_URShiftI:
+  case Op_URShiftL:
+    return new (C) RShiftCntVNode(cnt, vt);
+  }
+  fatal(err_msg_res("Missed vector creation for '%s'", NodeClassNames[shift->Opcode()]));
   return NULL;
 }
 
@@ -335,7 +355,7 @@
   case T_DOUBLE:
     return new (C) PackDNode(s, vt);
   }
-  ShouldNotReachHere();
+  fatal(err_msg_res("Type '%s' is not supported for vectors", type2name(bt)));
   return NULL;
 }
 
@@ -371,7 +391,7 @@
     case T_DOUBLE:
       return new (C) Pack2DNode(n1, n2, TypeVect::make(T_DOUBLE, 2));
     }
-    ShouldNotReachHere();
+    fatal(err_msg_res("Type '%s' is not supported for vectors", type2name(bt)));
   }
   return NULL;
 }
@@ -381,7 +401,6 @@
                                      Node* adr, const TypePtr* atyp, uint vlen, BasicType bt) {
   const TypeVect* vt = TypeVect::make(bt, vlen);
   return new (C) LoadVectorNode(ctl, mem, adr, atyp, vt);
-  return NULL;
 }
 
 // Return the vector version of a scalar store node.
@@ -413,7 +432,7 @@
   case T_DOUBLE:
     return new (C) ExtractDNode(v, pos);
   }
-  ShouldNotReachHere();
+  fatal(err_msg_res("Type '%s' is not supported for vectors", type2name(bt)));
   return NULL;
 }
 

--- a/src/share/vm/opto/vectornode.hpp	Thu Oct 04 14:34:51 2012 -0700
+++ b/src/share/vm/opto/vectornode.hpp	Fri Oct 05 13:28:16 2012 -0700
@@ -29,7 +29,7 @@
 #include "opto/node.hpp"
 #include "opto/opcodes.hpp"
 
-//------------------------------VectorNode--------------------------------------
+//------------------------------VectorNode-------------------------------------
 // Vector Operation
 class VectorNode : public TypeNode {
  public:
@@ -53,7 +53,7 @@
   virtual uint ideal_reg() const { return Matcher::vector_ideal_reg(vect_type()->length_in_bytes()); }
 
   static VectorNode* scalar2vector(Compile* C, Node* s, uint vlen, const Type* opd_t);
-
+  static VectorNode* shift_count(Compile* C, Node* shift, Node* cnt, uint vlen, BasicType bt);
   static VectorNode* make(Compile* C, int opc, Node* n1, Node* n2, uint vlen, BasicType bt);
 
   static int  opcode(int opc, BasicType bt);
@@ -64,9 +64,9 @@
   static void vector_operands(Node* n, uint* start, uint* end);
 };
 
-//===========================Vector=ALU=Operations====================================
+//===========================Vector=ALU=Operations=============================
 
-//------------------------------AddVBNode---------------------------------------
+//------------------------------AddVBNode--------------------------------------
 // Vector add byte
 class AddVBNode : public VectorNode {
  public:
@@ -74,7 +74,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------AddVSNode---------------------------------------
+//------------------------------AddVSNode--------------------------------------
 // Vector add char/short
 class AddVSNode : public VectorNode {
  public:
@@ -82,7 +82,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------AddVINode---------------------------------------
+//------------------------------AddVINode--------------------------------------
 // Vector add int
 class AddVINode : public VectorNode {
  public:
@@ -90,7 +90,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------AddVLNode---------------------------------------
+//------------------------------AddVLNode--------------------------------------
 // Vector add long
 class AddVLNode : public VectorNode {
  public:
@@ -98,7 +98,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------AddVFNode---------------------------------------
+//------------------------------AddVFNode--------------------------------------
 // Vector add float
 class AddVFNode : public VectorNode {
  public:
@@ -106,7 +106,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------AddVDNode---------------------------------------
+//------------------------------AddVDNode--------------------------------------
 // Vector add double
 class AddVDNode : public VectorNode {
  public:
@@ -114,7 +114,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVBNode---------------------------------------
+//------------------------------SubVBNode--------------------------------------
 // Vector subtract byte
 class SubVBNode : public VectorNode {
  public:
@@ -122,7 +122,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVSNode---------------------------------------
+//------------------------------SubVSNode--------------------------------------
 // Vector subtract short
 class SubVSNode : public VectorNode {
  public:
@@ -130,7 +130,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVINode---------------------------------------
+//------------------------------SubVINode--------------------------------------
 // Vector subtract int
 class SubVINode : public VectorNode {
  public:
@@ -138,7 +138,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVLNode---------------------------------------
+//------------------------------SubVLNode--------------------------------------
 // Vector subtract long
 class SubVLNode : public VectorNode {
  public:
@@ -146,7 +146,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVFNode---------------------------------------
+//------------------------------SubVFNode--------------------------------------
 // Vector subtract float
 class SubVFNode : public VectorNode {
  public:
@@ -154,7 +154,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------SubVDNode---------------------------------------
+//------------------------------SubVDNode--------------------------------------
 // Vector subtract double
 class SubVDNode : public VectorNode {
  public:
@@ -162,7 +162,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------MulVSNode---------------------------------------
+//------------------------------MulVSNode--------------------------------------
 // Vector multiply short
 class MulVSNode : public VectorNode {
  public:
@@ -170,7 +170,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------MulVINode---------------------------------------
+//------------------------------MulVINode--------------------------------------
 // Vector multiply int
 class MulVINode : public VectorNode {
  public:
@@ -178,7 +178,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------MulVFNode---------------------------------------
+//------------------------------MulVFNode--------------------------------------
 // Vector multiply float
 class MulVFNode : public VectorNode {
  public:
@@ -186,7 +186,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------MulVDNode---------------------------------------
+//------------------------------MulVDNode--------------------------------------
 // Vector multiply double
 class MulVDNode : public VectorNode {
  public:
@@ -194,7 +194,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------DivVFNode---------------------------------------
+//------------------------------DivVFNode--------------------------------------
 // Vector divide float
 class DivVFNode : public VectorNode {
  public:
@@ -202,7 +202,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------DivVDNode---------------------------------------
+//------------------------------DivVDNode--------------------------------------
 // Vector Divide double
 class DivVDNode : public VectorNode {
  public:
@@ -210,7 +210,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------LShiftVBNode---------------------------------------
+//------------------------------LShiftVBNode-----------------------------------
 // Vector left shift bytes
 class LShiftVBNode : public VectorNode {
  public:
@@ -218,7 +218,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------LShiftVSNode---------------------------------------
+//------------------------------LShiftVSNode-----------------------------------
 // Vector left shift shorts
 class LShiftVSNode : public VectorNode {
  public:
@@ -226,7 +226,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------LShiftVINode---------------------------------------
+//------------------------------LShiftVINode-----------------------------------
 // Vector left shift ints
 class LShiftVINode : public VectorNode {
  public:
@@ -234,7 +234,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------LShiftVLNode---------------------------------------
+//------------------------------LShiftVLNode-----------------------------------
 // Vector left shift longs
 class LShiftVLNode : public VectorNode {
  public:
@@ -242,7 +242,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------RShiftVBNode---------------------------------------
+//------------------------------RShiftVBNode-----------------------------------
 // Vector right arithmetic (signed) shift bytes
 class RShiftVBNode : public VectorNode {
  public:
@@ -250,7 +250,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------RShiftVSNode---------------------------------------
+//------------------------------RShiftVSNode-----------------------------------
 // Vector right arithmetic (signed) shift shorts
 class RShiftVSNode : public VectorNode {
  public:
@@ -258,7 +258,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------RShiftVINode---------------------------------------
+//------------------------------RShiftVINode-----------------------------------
 // Vector right arithmetic (signed) shift ints
 class RShiftVINode : public VectorNode {
  public:
@@ -266,7 +266,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------RShiftVLNode---------------------------------------
+//------------------------------RShiftVLNode-----------------------------------
 // Vector right arithmetic (signed) shift longs
 class RShiftVLNode : public VectorNode {
  public:
@@ -274,7 +274,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------URShiftVBNode---------------------------------------
+//------------------------------URShiftVBNode----------------------------------
 // Vector right logical (unsigned) shift bytes
 class URShiftVBNode : public VectorNode {
  public:
@@ -282,7 +282,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------URShiftVSNode---------------------------------------
+//------------------------------URShiftVSNode----------------------------------
 // Vector right logical (unsigned) shift shorts
 class URShiftVSNode : public VectorNode {
  public:
@@ -290,7 +290,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------URShiftVINode---------------------------------------
+//------------------------------URShiftVINode----------------------------------
 // Vector right logical (unsigned) shift ints
 class URShiftVINode : public VectorNode {
  public:
@@ -298,7 +298,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------URShiftVLNode---------------------------------------
+//------------------------------URShiftVLNode----------------------------------
 // Vector right logical (unsigned) shift longs
 class URShiftVLNode : public VectorNode {
  public:
@@ -306,6 +306,24 @@
   virtual int Opcode() const;
 };
 
+//------------------------------LShiftCntVNode---------------------------------
+// Vector left shift count
+class LShiftCntVNode : public VectorNode {
+ public:
+  LShiftCntVNode(Node* cnt, const TypeVect* vt) : VectorNode(cnt,vt) {}
+  virtual int Opcode() const;
+  virtual uint ideal_reg() const { return Matcher::vector_shift_count_ideal_reg(vect_type()->length_in_bytes()); }
+};
+
+//------------------------------RShiftCntVNode---------------------------------
+// Vector right shift count
+class RShiftCntVNode : public VectorNode {
+ public:
+  RShiftCntVNode(Node* cnt, const TypeVect* vt) : VectorNode(cnt,vt) {}
+  virtual int Opcode() const;
+  virtual uint ideal_reg() const { return Matcher::vector_shift_count_ideal_reg(vect_type()->length_in_bytes()); }
+};
+
 
 //------------------------------AndVNode---------------------------------------
 // Vector and integer
@@ -452,7 +470,7 @@
   static PackNode* make(Compile* C, Node* s, uint vlen, BasicType bt);
 };
 
-//------------------------------PackBNode---------------------------------------
+//------------------------------PackBNode--------------------------------------
 // Pack byte scalars into vector
 class PackBNode : public PackNode {
  public:
@@ -460,7 +478,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------PackSNode---------------------------------------
+//------------------------------PackSNode--------------------------------------
 // Pack short scalars into a vector
 class PackSNode : public PackNode {
  public:
@@ -469,7 +487,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------PackINode---------------------------------------
+//------------------------------PackINode--------------------------------------
 // Pack integer scalars into a vector
 class PackINode : public PackNode {
  public:
@@ -478,7 +496,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------PackLNode---------------------------------------
+//------------------------------PackLNode--------------------------------------
 // Pack long scalars into a vector
 class PackLNode : public PackNode {
  public:
@@ -487,7 +505,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------Pack2LNode--------------------------------------
+//------------------------------Pack2LNode-------------------------------------
 // Pack 2 long scalars into a vector
 class Pack2LNode : public PackNode {
  public:
@@ -495,7 +513,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------PackFNode---------------------------------------
+//------------------------------PackFNode--------------------------------------
 // Pack float scalars into vector
 class PackFNode : public PackNode {
  public:
@@ -504,7 +522,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------PackDNode---------------------------------------
+//------------------------------PackDNode--------------------------------------
 // Pack double scalars into a vector
 class PackDNode : public PackNode {
  public:
@@ -513,7 +531,7 @@
   virtual int Opcode() const;
 };
 
-//------------------------------Pack2DNode--------------------------------------
+//------------------------------Pack2DNode-------------------------------------
 // Pack 2 double scalars into a vector
 class Pack2DNode : public PackNode {
  public:
@@ -522,9 +540,9 @@
 };
 
 
-//========================Extract_Scalar_from_Vector===============================
+//========================Extract_Scalar_from_Vector===========================
 
-//------------------------------ExtractNode---------------------------------------
+//------------------------------ExtractNode------------------------------------
 // Extract a scalar from a vector at position "pos"
 class ExtractNode : public Node {
  public:
@@ -537,7 +555,7 @@
   static Node* make(Compile* C, Node* v, uint position, BasicType bt);
 };
 
-//------------------------------ExtractBNode---------------------------------------
+//------------------------------ExtractBNode-----------------------------------
 // Extract a byte from a vector at position "pos"
 class ExtractBNode : public ExtractNode {
  public:
@@ -547,7 +565,7 @@
   virtual uint ideal_reg() const { return Op_RegI; }
 };
 
-//------------------------------ExtractUBNode--------------------------------------
+//------------------------------ExtractUBNode----------------------------------
 // Extract a boolean from a vector at position "pos"
 class ExtractUBNode : public ExtractNode {
  public:
@@ -557,7 +575,7 @@
   virtual uint ideal_reg() const { return Op_RegI; }
 };
 
-//------------------------------ExtractCNode---------------------------------------
+//------------------------------ExtractCNode-----------------------------------
 // Extract a char from a vector at position "pos"
 class ExtractCNode : public ExtractNode {
  public:
@@ -567,7 +585,7 @@
   virtual uint ideal_reg() const { return Op_RegI; }
 };
 
-//------------------------------ExtractSNode---------------------------------------
+//------------------------------ExtractSNode-----------------------------------
 // Extract a short from a vector at position "pos"
 class ExtractSNode : public ExtractNode {
  public:
@@ -577,7 +595,7 @@
   virtual uint ideal_reg() const { return Op_RegI; }
 };
 
-//------------------------------ExtractINode---------------------------------------
+//------------------------------ExtractINode-----------------------------------
 // Extract an int from a vector at position "pos"
 class ExtractINode : public ExtractNode {
  public:
@@ -587,7 +605,7 @@
   virtual uint ideal_reg() const { return Op_RegI; }
 };
 
-//------------------------------ExtractLNode---------------------------------------
+//------------------------------ExtractLNode-----------------------------------
 // Extract a long from a vector at position "pos"
 class ExtractLNode : public ExtractNode {
  public:
@@ -597,7 +615,7 @@
   virtual uint ideal_reg() const { return Op_RegL; }
 };
 
-//------------------------------ExtractFNode---------------------------------------
+//------------------------------ExtractFNode-----------------------------------
 // Extract a float from a vector at position "pos"
 class ExtractFNode : public ExtractNode {
  public:
@@ -607,7 +625,7 @@
   virtual uint ideal_reg() const { return Op_RegF; }
 };
 
-//------------------------------ExtractDNode---------------------------------------
+//------------------------------ExtractDNode-----------------------------------
 // Extract a double from a vector at position "pos"
 class ExtractDNode : public ExtractNode {
  public: