changeset 4647:482ff4e18ca3 hs24-b42

Merge
author amurillo
date Thu, 25 Apr 2013 08:49:05 -0700
parents e0f1aa072185 ab40cf42161d
children 34a398422e59
files make/hotspot_version test/runtime/NMT/AllocTestType.java
diffstat 54 files changed, 1641 insertions(+), 707 deletions(-) [+]
line wrap: on
line diff
--- a/make/hotspot_version	Wed Apr 24 12:20:45 2013 -0700
+++ b/make/hotspot_version	Thu Apr 25 08:49:05 2013 -0700
@@ -35,7 +35,7 @@
 
 HS_MAJOR_VER=24
 HS_MINOR_VER=0
-HS_BUILD_NUMBER=41
+HS_BUILD_NUMBER=42
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=7
--- a/src/cpu/sparc/vm/sparc.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/sparc/vm/sparc.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -1,5 +1,5 @@
 //
-// Copyright (c) 1998, 2012, Oracle and/or its affiliates. All rights reserved.
+// Copyright (c) 1998, 2013, Oracle and/or its affiliates. All rights reserved.
 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 //
 // This code is free software; you can redistribute it and/or modify it
@@ -8176,10 +8176,25 @@
   format %{ "SUBcc  $p,$q,$p\t! p' = p-q\n\t"
             "ADD    $p,$y,$tmp\t! g3=p-q+y\n\t"
             "MOVlt  $tmp,$p\t! p' < 0 ? p'+y : p'" %}
-  ins_encode( enc_cadd_cmpLTMask(p, q, y, tmp) );
-  ins_pipe( cadd_cmpltmask );
-%}
-
+  ins_encode(enc_cadd_cmpLTMask(p, q, y, tmp));
+  ins_pipe(cadd_cmpltmask);
+%}
+
+instruct and_cmpLTMask(iRegI p, iRegI q, iRegI y, flagsReg ccr) %{
+  match(Set p (AndI (CmpLTMask p q) y));
+  effect(KILL ccr);
+  ins_cost(DEFAULT_COST*3);
+
+  format %{ "CMP  $p,$q\n\t"
+            "MOV  $y,$p\n\t"
+            "MOVge G0,$p" %}
+  ins_encode %{
+    __ cmp($p$$Register, $q$$Register);
+    __ mov($y$$Register, $p$$Register);
+    __ movcc(Assembler::greaterEqual, false, Assembler::icc, G0, $p$$Register);
+  %}
+  ins_pipe(ialu_reg_reg_ialu);
+%}
 
 //-----------------------------------------------------------------
 // Direct raw moves between float and general registers using VIS3.
--- a/src/cpu/x86/vm/assembler_x86.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/assembler_x86.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -10154,6 +10154,31 @@
   pop_CPU_state();
 }
 
+void MacroAssembler::restore_cpu_control_state_after_jni() {
+  // Either restore the MXCSR register after returning from the JNI Call
+  // or verify that it wasn't changed (with -Xcheck:jni flag).
+  if (VM_Version::supports_sse()) {
+    if (RestoreMXCSROnJNICalls) {
+      ldmxcsr(ExternalAddress(StubRoutines::addr_mxcsr_std()));
+    } else if (CheckJNICalls) {
+      call(RuntimeAddress(StubRoutines::x86::verify_mxcsr_entry()));
+    }
+  }
+  if (VM_Version::supports_avx()) {
+    // Clear upper bits of YMM registers to avoid SSE <-> AVX transition penalty.
+    vzeroupper();
+  }
+
+#ifndef _LP64
+  // Either restore the x87 floating pointer control word after returning
+  // from the JNI call or verify that it wasn't changed.
+  if (CheckJNICalls) {
+    call(RuntimeAddress(StubRoutines::x86::verify_fpu_cntrl_wrd_entry()));
+  }
+#endif // _LP64
+}
+
+
 void MacroAssembler::load_klass(Register dst, Register src) {
 #ifdef _LP64
   if (UseCompressedOops) {
@@ -11052,6 +11077,8 @@
     addptr(result, stride2);
     subl(cnt2, stride2);
     jccb(Assembler::notZero, COMPARE_WIDE_VECTORS_LOOP);
+    // clean upper bits of YMM registers
+    vzeroupper();
 
     // compare wide vectors tail
     bind(COMPARE_WIDE_TAIL);
@@ -11065,6 +11092,8 @@
 
     // Identifies the mismatching (higher or lower)16-bytes in the 32-byte vectors.
     bind(VECTOR_NOT_EQUAL);
+    // clean upper bits of YMM registers
+    vzeroupper();
     lea(str1, Address(str1, result, scale));
     lea(str2, Address(str2, result, scale));
     jmp(COMPARE_16_CHARS);
@@ -11321,6 +11350,10 @@
 
   // That's it
   bind(DONE);
+  if (UseAVX >= 2) {
+    // clean upper bits of YMM registers
+    vzeroupper();
+  }
 }
 
 void MacroAssembler::generate_fill(BasicType t, bool aligned,
@@ -11450,6 +11483,10 @@
         vmovdqu(Address(to, 0), xtmp);
         addptr(to, 32);
         subl(count, 8 << shift);
+
+        BIND(L_check_fill_8_bytes);
+        // clean upper bits of YMM registers
+        vzeroupper();
       } else {
         // Fill 32-byte chunks
         pshufd(xtmp, xtmp, 0);
@@ -11473,8 +11510,9 @@
         addptr(to, 32);
         subl(count, 8 << shift);
         jcc(Assembler::greaterEqual, L_fill_32_bytes_loop);
+
+        BIND(L_check_fill_8_bytes);
       }
-      BIND(L_check_fill_8_bytes);
       addl(count, 8 << shift);
       jccb(Assembler::zero, L_exit);
       jmpb(L_fill_8_bytes);
--- a/src/cpu/x86/vm/assembler_x86.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/assembler_x86.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2315,6 +2315,9 @@
   // only if +VerifyFPU
   void verify_FPU(int stack_depth, const char* s = "illegal FPU state");
 
+  // Verify or restore cpu control state after JNI call
+  void restore_cpu_control_state_after_jni();
+
   // prints msg, dumps registers and stops execution
   void stop(const char* msg);
 
--- a/src/cpu/x86/vm/cppInterpreter_x86.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/cppInterpreter_x86.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1295,25 +1295,8 @@
   __ push(rdx);
 #endif // _LP64
 
-  // Either restore the MXCSR register after returning from the JNI Call
-  // or verify that it wasn't changed.
-  if (VM_Version::supports_sse()) {
-    if (RestoreMXCSROnJNICalls) {
-      __ ldmxcsr(ExternalAddress(StubRoutines::addr_mxcsr_std()));
-    }
-    else if (CheckJNICalls ) {
-      __ call(RuntimeAddress(StubRoutines::x86::verify_mxcsr_entry()));
-    }
-  }
-
-#ifndef _LP64
-  // Either restore the x87 floating pointer control word after returning
-  // from the JNI call or verify that it wasn't changed.
-  if (CheckJNICalls) {
-    __ call(RuntimeAddress(StubRoutines::x86::verify_fpu_cntrl_wrd_entry()));
-  }
-#endif // _LP64
-
+  // Verify or restore cpu control state after JNI call
+  __ restore_cpu_control_state_after_jni();
 
   // change thread state
   __ movl(Address(thread, JavaThread::thread_state_offset()), _thread_in_native_trans);
--- a/src/cpu/x86/vm/sharedRuntime_x86_32.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/sharedRuntime_x86_32.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2070,6 +2070,9 @@
 
   __ call(RuntimeAddress(native_func));
 
+  // Verify or restore cpu control state after JNI call
+  __ restore_cpu_control_state_after_jni();
+
   // WARNING - on Windows Java Natives use pascal calling convention and pop the
   // arguments off of the stack. We could just re-adjust the stack pointer here
   // and continue to do SP relative addressing but we instead switch to FP
--- a/src/cpu/x86/vm/sharedRuntime_x86_64.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/sharedRuntime_x86_64.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2320,16 +2320,8 @@
 
   __ call(RuntimeAddress(native_func));
 
-    // Either restore the MXCSR register after returning from the JNI Call
-    // or verify that it wasn't changed.
-    if (RestoreMXCSROnJNICalls) {
-      __ ldmxcsr(ExternalAddress(StubRoutines::x86::mxcsr_std()));
-
-    }
-    else if (CheckJNICalls ) {
-      __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, StubRoutines::x86::verify_mxcsr_entry())));
-    }
-
+  // Verify or restore cpu control state after JNI call
+  __ restore_cpu_control_state_after_jni();
 
   // Unpack native results.
   switch (ret_type) {
--- a/src/cpu/x86/vm/stubGenerator_x86_32.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/stubGenerator_x86_32.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -864,6 +864,11 @@
   __ BIND(L_copy_64_bytes);
     __ subl(qword_count, 8);
     __ jcc(Assembler::greaterEqual, L_copy_64_bytes_loop);
+
+    if (UseUnalignedLoadStores && (UseAVX >= 2)) {
+      // clean upper bits of YMM registers
+      __ vzeroupper();
+    }
     __ addl(qword_count, 8);
     __ jccb(Assembler::zero, L_exit);
     //
--- a/src/cpu/x86/vm/stubGenerator_x86_64.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/stubGenerator_x86_64.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1359,6 +1359,10 @@
       }
       __ addptr(qword_count, 4);
       __ BIND(L_end);
+      if (UseAVX >= 2) {
+        // clean upper bits of YMM registers
+        __ vzeroupper();
+      }
     } else {
       // Copy 32-bytes per iteration
       __ BIND(L_loop);
@@ -1432,6 +1436,10 @@
       }
       __ subptr(qword_count, 4);
       __ BIND(L_end);
+      if (UseAVX >= 2) {
+        // clean upper bits of YMM registers
+        __ vzeroupper();
+      }
     } else {
       // Copy 32-bytes per iteration
       __ BIND(L_loop);
--- a/src/cpu/x86/vm/templateInterpreter_x86_32.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/templateInterpreter_x86_32.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1080,22 +1080,8 @@
 
   // result potentially in rdx:rax or ST0
 
-  // Either restore the MXCSR register after returning from the JNI Call
-  // or verify that it wasn't changed.
-  if (VM_Version::supports_sse()) {
-    if (RestoreMXCSROnJNICalls) {
-      __ ldmxcsr(ExternalAddress(StubRoutines::addr_mxcsr_std()));
-    }
-    else if (CheckJNICalls ) {
-      __ call(RuntimeAddress(StubRoutines::x86::verify_mxcsr_entry()));
-    }
-  }
-
-  // Either restore the x87 floating pointer control word after returning
-  // from the JNI call or verify that it wasn't changed.
-  if (CheckJNICalls) {
-    __ call(RuntimeAddress(StubRoutines::x86::verify_fpu_cntrl_wrd_entry()));
-  }
+  // Verify or restore cpu control state after JNI call
+  __ restore_cpu_control_state_after_jni();
 
   // save potential result in ST(0) & rdx:rax
   // (if result handler is the T_FLOAT or T_DOUBLE handler, result must be in ST0 -
--- a/src/cpu/x86/vm/templateInterpreter_x86_64.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/templateInterpreter_x86_64.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1081,15 +1081,8 @@
   __ call(rax);
   // result potentially in rax or xmm0
 
-  // Depending on runtime options, either restore the MXCSR
-  // register after returning from the JNI Call or verify that
-  // it wasn't changed during -Xcheck:jni.
-  if (RestoreMXCSROnJNICalls) {
-    __ ldmxcsr(ExternalAddress(StubRoutines::x86::mxcsr_std()));
-  }
-  else if (CheckJNICalls) {
-    __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, StubRoutines::x86::verify_mxcsr_entry())));
-  }
+  // Verify or restore cpu control state after JNI call
+  __ restore_cpu_control_state_after_jni();
 
   // NOTE: The order of these pushes is known to frame::interpreter_frame_result
   // in order to extract the result of a method call. If the order of these
--- a/src/cpu/x86/vm/x86_32.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/x86_32.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -1,5 +1,5 @@
 //
-// Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
+// Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 //
 // This code is free software; you can redistribute it and/or modify it
@@ -228,10 +228,16 @@
 static jlong *double_signflip_pool = double_quadword(&fp_signmask_pool[4*2], CONST64(0x8000000000000000), CONST64(0x8000000000000000));
 
 // Offset hacking within calls.
-static int pre_call_FPU_size() {
-  if (Compile::current()->in_24_bit_fp_mode())
-    return 6; // fldcw
-  return 0;
+static int pre_call_resets_size() {
+  int size = 0;
+  Compile* C = Compile::current();
+  if (C->in_24_bit_fp_mode()) {
+    size += 6; // fldcw
+  }
+  if (C->max_vector_size() > 16) {
+    size += 3; // vzeroupper
+  }
+  return size;
 }
 
 static int preserve_SP_size() {
@@ -242,21 +248,21 @@
 //       from the start of the call to the point where the return address
 //       will point.
 int MachCallStaticJavaNode::ret_addr_offset() {
-  int offset = 5 + pre_call_FPU_size();  // 5 bytes from start of call to where return address points
+  int offset = 5 + pre_call_resets_size();  // 5 bytes from start of call to where return address points
   if (_method_handle_invoke)
     offset += preserve_SP_size();
   return offset;
 }
 
 int MachCallDynamicJavaNode::ret_addr_offset() {
-  return 10 + pre_call_FPU_size();  // 10 bytes from start of call to where return address points
+  return 10 + pre_call_resets_size();  // 10 bytes from start of call to where return address points
 }
 
 static int sizeof_FFree_Float_Stack_All = -1;
 
 int MachCallRuntimeNode::ret_addr_offset() {
   assert(sizeof_FFree_Float_Stack_All != -1, "must have been emitted already");
-  return sizeof_FFree_Float_Stack_All + 5 + pre_call_FPU_size();
+  return sizeof_FFree_Float_Stack_All + 5 + pre_call_resets_size();
 }
 
 // Indicate if the safepoint node needs the polling page as an input.
@@ -272,7 +278,7 @@
 // The address of the call instruction needs to be 4-byte aligned to
 // ensure that it does not span a cache line so that it can be patched.
 int CallStaticJavaDirectNode::compute_padding(int current_offset) const {
-  current_offset += pre_call_FPU_size();  // skip fldcw, if any
+  current_offset += pre_call_resets_size();  // skip fldcw, if any
   current_offset += 1;      // skip call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
 }
@@ -280,7 +286,7 @@
 // The address of the call instruction needs to be 4-byte aligned to
 // ensure that it does not span a cache line so that it can be patched.
 int CallStaticJavaHandleNode::compute_padding(int current_offset) const {
-  current_offset += pre_call_FPU_size();  // skip fldcw, if any
+  current_offset += pre_call_resets_size();  // skip fldcw, if any
   current_offset += preserve_SP_size();   // skip mov rbp, rsp
   current_offset += 1;      // skip call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
@@ -289,7 +295,7 @@
 // The address of the call instruction needs to be 4-byte aligned to
 // ensure that it does not span a cache line so that it can be patched.
 int CallDynamicJavaDirectNode::compute_padding(int current_offset) const {
-  current_offset += pre_call_FPU_size();  // skip fldcw, if any
+  current_offset += pre_call_resets_size();  // skip fldcw, if any
   current_offset += 5;      // skip MOV instruction
   current_offset += 1;      // skip call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
@@ -583,16 +589,20 @@
   // Remove two words for return addr and rbp,
   framesize -= 2*wordSize;
 
-  if( C->in_24_bit_fp_mode() ) {
+  if (C->max_vector_size() > 16) {
+    st->print("VZEROUPPER");
+    st->cr(); st->print("\t");
+  }
+  if (C->in_24_bit_fp_mode()) {
     st->print("FLDCW  standard control word");
     st->cr(); st->print("\t");
   }
-  if( framesize ) {
+  if (framesize) {
     st->print("ADD    ESP,%d\t# Destroy frame",framesize);
     st->cr(); st->print("\t");
   }
   st->print_cr("POPL   EBP"); st->print("\t");
-  if( do_polling() && C->is_method_compilation() ) {
+  if (do_polling() && C->is_method_compilation()) {
     st->print("TEST   PollPage,EAX\t! Poll Safepoint");
     st->cr(); st->print("\t");
   }
@@ -602,8 +612,14 @@
 void MachEpilogNode::emit(CodeBuffer &cbuf, PhaseRegAlloc *ra_) const {
   Compile *C = ra_->C;
 
+  if (C->max_vector_size() > 16) {
+    // Clear upper bits of YMM registers when current compiled code uses
+    // wide vectors to avoid AVX <-> SSE transition penalty during call.
+    MacroAssembler masm(&cbuf);
+    masm.vzeroupper();
+  }
   // If method set FPU control word, restore to standard control word
-  if( C->in_24_bit_fp_mode() ) {
+  if (C->in_24_bit_fp_mode()) {
     MacroAssembler masm(&cbuf);
     masm.fldcw(ExternalAddress(StubRoutines::addr_fpu_cntrl_wrd_std()));
   }
@@ -615,12 +631,11 @@
 
   // Note that VerifyStackAtCalls' Majik cookie does not change the frame size popped here
 
-  if( framesize >= 128 ) {
+  if (framesize >= 128) {
     emit_opcode(cbuf, 0x81); // add  SP, #framesize
     emit_rm(cbuf, 0x3, 0x00, ESP_enc);
     emit_d32(cbuf, framesize);
-  }
-  else if( framesize ) {
+  } else if (framesize) {
     emit_opcode(cbuf, 0x83); // add  SP, #framesize
     emit_rm(cbuf, 0x3, 0x00, ESP_enc);
     emit_d8(cbuf, framesize);
@@ -628,7 +643,7 @@
 
   emit_opcode(cbuf, 0x58 | EBP_enc);
 
-  if( do_polling() && C->is_method_compilation() ) {
+  if (do_polling() && C->is_method_compilation()) {
     cbuf.relocate(cbuf.insts_end(), relocInfo::poll_return_type, 0);
     emit_opcode(cbuf,0x85);
     emit_rm(cbuf, 0x0, EAX_enc, 0x5); // EAX
@@ -640,7 +655,8 @@
   Compile *C = ra_->C;
   // If method set FPU control word, restore to standard control word
   int size = C->in_24_bit_fp_mode() ? 6 : 0;
-  if( do_polling() && C->is_method_compilation() ) size += 6;
+  if (C->max_vector_size() > 16) size += 3; // vzeroupper
+  if (do_polling() && C->is_method_compilation()) size += 6;
 
   int framesize = C->frame_slots() << LogBytesPerInt;
   assert((framesize & (StackAlignmentInBytes-1)) == 0, "frame size not aligned");
@@ -649,7 +665,7 @@
 
   size++; // popl rbp,
 
-  if( framesize >= 128 ) {
+  if (framesize >= 128) {
     size += 6;
   } else {
     size += framesize ? 3 : 0;
@@ -1848,20 +1864,26 @@
   %}
 
 
-  enc_class pre_call_FPU %{
+  enc_class pre_call_resets %{
     // If method sets FPU control word restore it here
     debug_only(int off0 = cbuf.insts_size());
-    if( Compile::current()->in_24_bit_fp_mode() ) {
-      MacroAssembler masm(&cbuf);
-      masm.fldcw(ExternalAddress(StubRoutines::addr_fpu_cntrl_wrd_std()));
+    if (ra_->C->in_24_bit_fp_mode()) {
+      MacroAssembler _masm(&cbuf);
+      __ fldcw(ExternalAddress(StubRoutines::addr_fpu_cntrl_wrd_std()));
+    }
+    if (ra_->C->max_vector_size() > 16) {
+      // Clear upper bits of YMM registers when current compiled code uses
+      // wide vectors to avoid AVX <-> SSE transition penalty during call.
+      MacroAssembler _masm(&cbuf);
+      __ vzeroupper();
     }
     debug_only(int off1 = cbuf.insts_size());
-    assert(off1 - off0 == pre_call_FPU_size(), "correct size prediction");
+    assert(off1 - off0 == pre_call_resets_size(), "correct size prediction");
   %}
 
   enc_class post_call_FPU %{
     // If method sets FPU control word do it here also
-    if( Compile::current()->in_24_bit_fp_mode() ) {
+    if (Compile::current()->in_24_bit_fp_mode()) {
       MacroAssembler masm(&cbuf);
       masm.fldcw(ExternalAddress(StubRoutines::addr_fpu_cntrl_wrd_24()));
     }
@@ -1872,17 +1894,17 @@
     // who we intended to call.
     cbuf.set_insts_mark();
     $$$emit8$primary;
-    if ( !_method ) {
+    if (!_method) {
       emit_d32_reloc(cbuf, ($meth$$method - (int)(cbuf.insts_end()) - 4),
                      runtime_call_Relocation::spec(), RELOC_IMM32 );
-    } else if(_optimized_virtual) {
+    } else if (_optimized_virtual) {
       emit_d32_reloc(cbuf, ($meth$$method - (int)(cbuf.insts_end()) - 4),
                      opt_virtual_call_Relocation::spec(), RELOC_IMM32 );
     } else {
       emit_d32_reloc(cbuf, ($meth$$method - (int)(cbuf.insts_end()) - 4),
                      static_call_Relocation::spec(), RELOC_IMM32 );
     }
-    if( _method ) {  // Emit stub for static call
+    if (_method) {  // Emit stub for static call
       emit_java_to_interp(cbuf);
     }
   %}
@@ -2301,30 +2323,6 @@
     emit_rm(cbuf, 0x3, $p$$reg, tmpReg);
   %}
 
-  enc_class enc_cmpLTP_mem(rRegI p, rRegI q, memory mem, eCXRegI tmp) %{    // cadd_cmpLT
-    int tmpReg = $tmp$$reg;
-
-    // SUB $p,$q
-    emit_opcode(cbuf,0x2B);
-    emit_rm(cbuf, 0x3, $p$$reg, $q$$reg);
-    // SBB $tmp,$tmp
-    emit_opcode(cbuf,0x1B);
-    emit_rm(cbuf, 0x3, tmpReg, tmpReg);
-    // AND $tmp,$y
-    cbuf.set_insts_mark();       // Mark start of opcode for reloc info in mem operand
-    emit_opcode(cbuf,0x23);
-    int reg_encoding = tmpReg;
-    int base  = $mem$$base;
-    int index = $mem$$index;
-    int scale = $mem$$scale;
-    int displace = $mem$$disp;
-    bool disp_is_oop = $mem->disp_is_oop();
-    encode_RegMem(cbuf, reg_encoding, base, index, scale, displace, disp_is_oop);
-    // ADD $p,$tmp
-    emit_opcode(cbuf,0x03);
-    emit_rm(cbuf, 0x3, $p$$reg, tmpReg);
-  %}
-
   enc_class shift_left_long( eRegL dst, eCXRegI shift ) %{
     // TEST shift,32
     emit_opcode(cbuf,0xF7);
@@ -8906,9 +8904,9 @@
   %}
 %}
 
-instruct cmpLTMask( eCXRegI dst, ncxRegI p, ncxRegI q, eFlagsReg cr ) %{
+instruct cmpLTMask(eCXRegI dst, ncxRegI p, ncxRegI q, eFlagsReg cr) %{
   match(Set dst (CmpLTMask p q));
-  effect( KILL cr );
+  effect(KILL cr);
   ins_cost(400);
 
   // SETlt can only use low byte of EAX,EBX, ECX, or EDX as destination
@@ -8916,50 +8914,83 @@
             "CMP    $p,$q\n\t"
             "SETlt  $dst\n\t"
             "NEG    $dst" %}
-  ins_encode( OpcRegReg(0x33,dst,dst),
-              OpcRegReg(0x3B,p,q),
-              setLT_reg(dst), neg_reg(dst) );
-  ins_pipe( pipe_slow );
-%}
-
-instruct cmpLTMask0( rRegI dst, immI0 zero, eFlagsReg cr ) %{
+  ins_encode %{
+    Register Rp = $p$$Register;
+    Register Rq = $q$$Register;
+    Register Rd = $dst$$Register;
+    Label done;
+    __ xorl(Rd, Rd);
+    __ cmpl(Rp, Rq);
+    __ setb(Assembler::less, Rd);
+    __ negl(Rd);
+  %}
+
+  ins_pipe(pipe_slow);
+%}
+
+instruct cmpLTMask0(rRegI dst, immI0 zero, eFlagsReg cr) %{
   match(Set dst (CmpLTMask dst zero));
-  effect( DEF dst, KILL cr );
+  effect(DEF dst, KILL cr);
   ins_cost(100);
 
-  format %{ "SAR    $dst,31" %}
-  opcode(0xC1, 0x7);  /* C1 /7 ib */
-  ins_encode( RegOpcImm( dst, 0x1F ) );
-  ins_pipe( ialu_reg );
-%}
-
-
-instruct cadd_cmpLTMask( ncxRegI p, ncxRegI q, ncxRegI y, eCXRegI tmp, eFlagsReg cr ) %{
+  format %{ "SAR    $dst,31\t# cmpLTMask0" %}
+  ins_encode %{
+  __ sarl($dst$$Register, 31);
+  %}
+  ins_pipe(ialu_reg);
+%}
+
+/* better to save a register than avoid a branch */
+instruct cadd_cmpLTMask(rRegI p, rRegI q, rRegI y, eFlagsReg cr) %{
   match(Set p (AddI (AndI (CmpLTMask p q) y) (SubI p q)));
-  effect( KILL tmp, KILL cr );
+  effect(KILL cr);
   ins_cost(400);
-  // annoyingly, $tmp has no edges so you cant ask for it in
-  // any format or encoding
-  format %{ "SUB    $p,$q\n\t"
-            "SBB    ECX,ECX\n\t"
-            "AND    ECX,$y\n\t"
-            "ADD    $p,ECX" %}
-  ins_encode( enc_cmpLTP(p,q,y,tmp) );
-  ins_pipe( pipe_cmplt );
+  format %{ "SUB    $p,$q\t# cadd_cmpLTMask\n\t"
+            "JGE    done\n\t"
+            "ADD    $p,$y\n"
+            "done:  " %}
+  ins_encode %{
+    Register Rp = $p$$Register;
+    Register Rq = $q$$Register;
+    Register Ry = $y$$Register;
+    Label done;
+    __ subl(Rp, Rq);
+    __ jccb(Assembler::greaterEqual, done);
+    __ addl(Rp, Ry);
+    __ bind(done);
+  %}
+
+  ins_pipe(pipe_cmplt);
+%}
+
+/* better to save a register than avoid a branch */
+instruct and_cmpLTMask(rRegI p, rRegI q, rRegI y, eFlagsReg cr) %{
+  match(Set y (AndI (CmpLTMask p q) y));
+  effect(KILL cr);
+
+  ins_cost(300);
+
+  format %{ "CMPL     $p, $q\t# and_cmpLTMask\n\t"
+            "JLT      done\n\t"
+            "XORL     $y, $y\n"
+            "done:  " %}
+  ins_encode %{
+    Register Rp = $p$$Register;
+    Register Rq = $q$$Register;
+    Register Ry = $y$$Register;
+    Label done;
+    __ cmpl(Rp, Rq);
+    __ jccb(Assembler::less, done);
+    __ xorl(Ry, Ry);
+    __ bind(done);
+  %}
+
+  ins_pipe(pipe_cmplt);
 %}
 
 /* If I enable this, I encourage spilling in the inner loop of compress.
-instruct cadd_cmpLTMask_mem( ncxRegI p, ncxRegI q, memory y, eCXRegI tmp, eFlagsReg cr ) %{
+instruct cadd_cmpLTMask_mem(ncxRegI p, ncxRegI q, memory y, eCXRegI tmp, eFlagsReg cr) %{
   match(Set p (AddI (AndI (CmpLTMask p q) (LoadI y)) (SubI p q)));
-  effect( USE_KILL tmp, KILL cr );
-  ins_cost(400);
-
-  format %{ "SUB    $p,$q\n\t"
-            "SBB    ECX,ECX\n\t"
-            "AND    ECX,$y\n\t"
-            "ADD    $p,ECX" %}
-  ins_encode( enc_cmpLTP_mem(p,q,y,tmp) );
-%}
 */
 
 //----------Long Instructions------------------------------------------------
@@ -12817,7 +12848,7 @@
   ins_cost(300);
   format %{ "CALL,static " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode( pre_call_FPU,
+  ins_encode( pre_call_resets,
               Java_Static_Call( meth ),
               call_epilog,
               post_call_FPU );
@@ -12838,7 +12869,7 @@
   ins_cost(300);
   format %{ "CALL,static/MethodHandle " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode( pre_call_FPU,
+  ins_encode( pre_call_resets,
               preserve_SP,
               Java_Static_Call( meth ),
               restore_SP,
@@ -12859,7 +12890,7 @@
   format %{ "MOV    EAX,(oop)-1\n\t"
             "CALL,dynamic" %}
   opcode(0xE8); /* E8 cd */
-  ins_encode( pre_call_FPU,
+  ins_encode( pre_call_resets,
               Java_Dynamic_Call( meth ),
               call_epilog,
               post_call_FPU );
@@ -12876,7 +12907,7 @@
   format %{ "CALL,runtime " %}
   opcode(0xE8); /* E8 cd */
   // Use FFREEs to clear entries in float stack
-  ins_encode( pre_call_FPU,
+  ins_encode( pre_call_resets,
               FFree_Float_Stack_All,
               Java_To_Runtime( meth ),
               post_call_FPU );
@@ -12891,7 +12922,7 @@
   ins_cost(300);
   format %{ "CALL_LEAF,runtime " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode( pre_call_FPU,
+  ins_encode( pre_call_resets,
               FFree_Float_Stack_All,
               Java_To_Runtime( meth ),
               Verify_FPU_For_Leaf, post_call_FPU );
--- a/src/cpu/x86/vm/x86_64.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/cpu/x86/vm/x86_64.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -1,5 +1,5 @@
 //
-// Copyright (c) 2003, 2012, Oracle and/or its affiliates. All rights reserved.
+// Copyright (c) 2003, 2013, Oracle and/or its affiliates. All rights reserved.
 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 //
 // This code is free software; you can redistribute it and/or modify it
@@ -399,6 +399,9 @@
 static int preserve_SP_size() {
   return 3;  // rex.w, op, rm(reg/reg)
 }
+static int clear_avx_size() {
+  return (Compile::current()->max_vector_size() > 16) ? 3 : 0;  // vzeroupper
+}
 
 // !!!!! Special hack to get all types of calls to specify the byte offset
 //       from the start of the call to the point where the return address
@@ -406,6 +409,7 @@
 int MachCallStaticJavaNode::ret_addr_offset()
 {
   int offset = 5; // 5 bytes from start of call to where return address points
+  offset += clear_avx_size();
   if (_method_handle_invoke)
     offset += preserve_SP_size();
   return offset;
@@ -413,11 +417,16 @@
 
 int MachCallDynamicJavaNode::ret_addr_offset()
 {
-  return 15; // 15 bytes from start of call to where return address points
+  int offset = 15; // 15 bytes from start of call to where return address points
+  offset += clear_avx_size();
+  return offset;
 }
 
-// In os_cpu .ad file
-// int MachCallRuntimeNode::ret_addr_offset()
+int MachCallRuntimeNode::ret_addr_offset() {
+  int offset = 13; // movq r10,#addr; callq (r10)
+  offset += clear_avx_size();
+  return offset;
+}
 
 // Indicate if the safepoint node needs the polling page as an input,
 // it does if the polling page is more than disp32 away.
@@ -434,6 +443,7 @@
 // ensure that it does not span a cache line so that it can be patched.
 int CallStaticJavaDirectNode::compute_padding(int current_offset) const
 {
+  current_offset += clear_avx_size(); // skip vzeroupper
   current_offset += 1; // skip call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
 }
@@ -443,6 +453,7 @@
 int CallStaticJavaHandleNode::compute_padding(int current_offset) const
 {
   current_offset += preserve_SP_size();   // skip mov rbp, rsp
+  current_offset += clear_avx_size(); // skip vzeroupper
   current_offset += 1; // skip call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
 }
@@ -451,6 +462,7 @@
 // ensure that it does not span a cache line so that it can be patched.
 int CallDynamicJavaDirectNode::compute_padding(int current_offset) const
 {
+  current_offset += clear_avx_size(); // skip vzeroupper
   current_offset += 11; // skip movq instruction + call opcode byte
   return round_to(current_offset, alignment_required()) - current_offset;
 }
@@ -762,6 +774,11 @@
 void MachEpilogNode::format(PhaseRegAlloc* ra_, outputStream* st) const
 {
   Compile* C = ra_->C;
+  if (C->max_vector_size() > 16) {
+    st->print("vzeroupper");
+    st->cr(); st->print("\t");
+  }
+
   int framesize = C->frame_slots() << LogBytesPerInt;
   assert((framesize & (StackAlignmentInBytes-1)) == 0, "frame size not aligned");
   // Remove word for return adr already pushed
@@ -791,6 +808,13 @@
 void MachEpilogNode::emit(CodeBuffer& cbuf, PhaseRegAlloc* ra_) const
 {
   Compile* C = ra_->C;
+  if (C->max_vector_size() > 16) {
+    // Clear upper bits of YMM registers when current compiled code uses
+    // wide vectors to avoid AVX <-> SSE transition penalty during call.
+    MacroAssembler _masm(&cbuf);
+    __ vzeroupper();
+  }
+
   int framesize = C->frame_slots() << LogBytesPerInt;
   assert((framesize & (StackAlignmentInBytes-1)) == 0, "frame size not aligned");
   // Remove word for return adr already pushed
@@ -2001,6 +2025,25 @@
     __ bind(miss);
   %}
 
+  enc_class clear_avx %{
+    debug_only(int off0 = cbuf.insts_size());
+    if (ra_->C->max_vector_size() > 16) {
+      // Clear upper bits of YMM registers when current compiled code uses
+      // wide vectors to avoid AVX <-> SSE transition penalty during call.
+      MacroAssembler _masm(&cbuf);
+      __ vzeroupper();
+    }
+    debug_only(int off1 = cbuf.insts_size());
+    assert(off1 - off0 == clear_avx_size(), "correct size prediction");
+  %}
+
+  enc_class Java_To_Runtime(method meth) %{
+    // No relocation needed
+    MacroAssembler _masm(&cbuf);
+    __ mov64(r10, (int64_t) $meth$$method);
+    __ call(r10);
+  %}
+
   enc_class Java_To_Interpreter(method meth)
   %{
     // CALL Java_To_Interpreter
@@ -2203,12 +2246,6 @@
     $$$emit32$src$$constant;
   %}
 
-  enc_class Con64(immL src)
-  %{
-    // Output immediate
-    emit_d64($src$$constant);
-  %}
-
   enc_class Con32F_as_bits(immF src)
   %{
     // Output Float immediate bits
@@ -7349,7 +7386,7 @@
   ins_pipe( pipe_cmpxchg );
 %}
 
-instruct xaddL_no_res( memory mem, Universe dummy, immL add, rFlagsReg cr) %{
+instruct xaddL_no_res( memory mem, Universe dummy, immL32 add, rFlagsReg cr) %{
   predicate(n->as_LoadStore()->result_not_used());
   match(Set dummy (GetAndAddL mem add));
   effect(KILL cr);
@@ -9198,7 +9235,7 @@
   match(Set dst (CmpLTMask p q));
   effect(KILL cr);
 
-  ins_cost(400); // XXX
+  ins_cost(400);
   format %{ "cmpl    $p, $q\t# cmpLTMask\n\t"
             "setlt   $dst\n\t"
             "movzbl  $dst, $dst\n\t"
@@ -9216,37 +9253,63 @@
   match(Set dst (CmpLTMask dst zero));
   effect(KILL cr);
 
-  ins_cost(100); // XXX
+  ins_cost(100);
   format %{ "sarl    $dst, #31\t# cmpLTMask0" %}
-  opcode(0xC1, 0x7);  /* C1 /7 ib */
-  ins_encode(reg_opc_imm(dst, 0x1F));
+  ins_encode %{
+  __ sarl($dst$$Register, 31);
+  %}
   ins_pipe(ialu_reg);
 %}
 
-
-instruct cadd_cmpLTMask(rRegI p, rRegI q, rRegI y, rRegI tmp, rFlagsReg cr)
+/* Better to save a register than avoid a branch */
+instruct cadd_cmpLTMask(rRegI p, rRegI q, rRegI y, rFlagsReg cr)
 %{
   match(Set p (AddI (AndI (CmpLTMask p q) y) (SubI p q)));
-  effect(TEMP tmp, KILL cr);
-
-  ins_cost(400); // XXX
-  format %{ "subl    $p, $q\t# cadd_cmpLTMask1\n\t"
-            "sbbl    $tmp, $tmp\n\t"
-            "andl    $tmp, $y\n\t"
-            "addl    $p, $tmp" %}
+  effect(KILL cr);
+  ins_cost(300);
+  format %{ "subl   $p,$q\t# cadd_cmpLTMask\n\t"
+            "jge    done\n\t"
+            "addl   $p,$y\n"
+            "done:  " %}
   ins_encode %{
     Register Rp = $p$$Register;
     Register Rq = $q$$Register;
     Register Ry = $y$$Register;
-    Register Rt = $tmp$$Register;
+    Label done;
     __ subl(Rp, Rq);
-    __ sbbl(Rt, Rt);
-    __ andl(Rt, Ry);
-    __ addl(Rp, Rt);
+    __ jccb(Assembler::greaterEqual, done);
+    __ addl(Rp, Ry);
+    __ bind(done);
   %}
   ins_pipe(pipe_cmplt);
 %}
 
+/* Better to save a register than avoid a branch */
+instruct and_cmpLTMask(rRegI p, rRegI q, rRegI y, rFlagsReg cr)
+%{
+  match(Set y (AndI (CmpLTMask p q) y));
+  effect(KILL cr);
+
+  ins_cost(300);
+
+  format %{ "cmpl     $p, $q\t# and_cmpLTMask\n\t"
+            "jlt      done\n\t"
+            "xorl     $y, $y\n"
+            "done:  " %}
+  ins_encode %{
+    Register Rp = $p$$Register;
+    Register Rq = $q$$Register;
+    Register Ry = $y$$Register;
+    Label done;
+    __ cmpl(Rp, Rq);
+    __ jccb(Assembler::less, done);
+    __ xorl(Ry, Ry);
+    __ bind(done);
+  %}
+  ins_pipe(pipe_cmplt);
+%}
+
+
 //---------- FP Instructions------------------------------------------------
 
 instruct cmpF_cc_reg(rFlagsRegU cr, regF src1, regF src2)
@@ -11124,7 +11187,7 @@
   ins_cost(300);
   format %{ "call,static " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode(Java_Static_Call(meth), call_epilog);
+  ins_encode(clear_avx, Java_Static_Call(meth), call_epilog);
   ins_pipe(pipe_slow);
   ins_alignment(4);
 %}
@@ -11142,7 +11205,7 @@
   ins_cost(300);
   format %{ "call,static/MethodHandle " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode(preserve_SP,
+  ins_encode(clear_avx, preserve_SP,
              Java_Static_Call(meth),
              restore_SP,
              call_epilog);
@@ -11162,7 +11225,7 @@
   format %{ "movq    rax, #Universe::non_oop_word()\n\t"
             "call,dynamic " %}
   opcode(0xE8); /* E8 cd */
-  ins_encode(Java_Dynamic_Call(meth), call_epilog);
+  ins_encode(clear_avx, Java_Dynamic_Call(meth), call_epilog);
   ins_pipe(pipe_slow);
   ins_alignment(4);
 %}
@@ -11175,8 +11238,7 @@
 
   ins_cost(300);
   format %{ "call,runtime " %}
-  opcode(0xE8); /* E8 cd */
-  ins_encode(Java_To_Runtime(meth));
+  ins_encode(clear_avx, Java_To_Runtime(meth));
   ins_pipe(pipe_slow);
 %}
 
@@ -11188,8 +11250,7 @@
 
   ins_cost(300);
   format %{ "call_leaf,runtime " %}
-  opcode(0xE8); /* E8 cd */
-  ins_encode(Java_To_Runtime(meth));
+  ins_encode(clear_avx, Java_To_Runtime(meth));
   ins_pipe(pipe_slow);
 %}
 
@@ -11201,7 +11262,6 @@
 
   ins_cost(300);
   format %{ "call_leaf_nofp,runtime " %}
-  opcode(0xE8); /* E8 cd */
   ins_encode(Java_To_Runtime(meth));
   ins_pipe(pipe_slow);
 %}
--- a/src/os/bsd/vm/os_bsd.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/bsd/vm/os_bsd.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2667,17 +2667,118 @@
 
 // Bsd(POSIX) specific hand shaking semaphore.
 #ifdef __APPLE__
-static semaphore_t sig_sem;
+typedef semaphore_t os_semaphore_t;
 #define SEM_INIT(sem, value)    semaphore_create(mach_task_self(), &sem, SYNC_POLICY_FIFO, value)
-#define SEM_WAIT(sem)           semaphore_wait(sem);
-#define SEM_POST(sem)           semaphore_signal(sem);
+#define SEM_WAIT(sem)           semaphore_wait(sem)
+#define SEM_POST(sem)           semaphore_signal(sem)
+#define SEM_DESTROY(sem)        semaphore_destroy(mach_task_self(), sem)
 #else
-static sem_t sig_sem;
+typedef sem_t os_semaphore_t;
 #define SEM_INIT(sem, value)    sem_init(&sem, 0, value)
-#define SEM_WAIT(sem)           sem_wait(&sem);
-#define SEM_POST(sem)           sem_post(&sem);
+#define SEM_WAIT(sem)           sem_wait(&sem)
+#define SEM_POST(sem)           sem_post(&sem)
+#define SEM_DESTROY(sem)        sem_destroy(&sem)
 #endif
 
+class Semaphore : public StackObj {
+  public:
+    Semaphore();
+    ~Semaphore();
+    void signal();
+    void wait();
+    bool trywait();
+    bool timedwait(unsigned int sec, int nsec);
+  private:
+    jlong currenttime() const;
+    semaphore_t _semaphore;
+};
+
+Semaphore::Semaphore() : _semaphore(0) {
+  SEM_INIT(_semaphore, 0);
+}
+
+Semaphore::~Semaphore() {
+  SEM_DESTROY(_semaphore);
+}
+
+void Semaphore::signal() {
+  SEM_POST(_semaphore);
+}
+
+void Semaphore::wait() {
+  SEM_WAIT(_semaphore);
+}
+
+jlong Semaphore::currenttime() const {
+    struct timeval tv;
+    gettimeofday(&tv, NULL);
+    return (tv.tv_sec * NANOSECS_PER_SEC) + (tv.tv_usec * 1000);
+}
+
+#ifdef __APPLE__
+bool Semaphore::trywait() {
+  return timedwait(0, 0);
+}
+
+bool Semaphore::timedwait(unsigned int sec, int nsec) {
+  kern_return_t kr = KERN_ABORTED;
+  mach_timespec_t waitspec;
+  waitspec.tv_sec = sec;
+  waitspec.tv_nsec = nsec;
+
+  jlong starttime = currenttime();
+
+  kr = semaphore_timedwait(_semaphore, waitspec);
+  while (kr == KERN_ABORTED) {
+    jlong totalwait = (sec * NANOSECS_PER_SEC) + nsec;
+
+    jlong current = currenttime();
+    jlong passedtime = current - starttime;
+
+    if (passedtime >= totalwait) {
+      waitspec.tv_sec = 0;
+      waitspec.tv_nsec = 0;
+    } else {
+      jlong waittime = totalwait - (current - starttime);
+      waitspec.tv_sec = waittime / NANOSECS_PER_SEC;
+      waitspec.tv_nsec = waittime % NANOSECS_PER_SEC;
+    }
+
+    kr = semaphore_timedwait(_semaphore, waitspec);
+  }
+
+  return kr == KERN_SUCCESS;
+}
+
+#else
+
+bool Semaphore::trywait() {
+  return sem_trywait(&_semaphore) == 0;
+}
+
+bool Semaphore::timedwait(unsigned int sec, int nsec) {
+  struct timespec ts;
+  jlong endtime = unpackTime(&ts, false, (sec * NANOSECS_PER_SEC) + nsec);
+
+  while (1) {
+    int result = sem_timedwait(&_semaphore, &ts);
+    if (result == 0) {
+      return true;
+    } else if (errno == EINTR) {
+      continue;
+    } else if (errno == ETIMEDOUT) {
+      return false;
+    } else {
+      return false;
+    }
+  }
+}
+
+#endif // __APPLE__
+
+static os_semaphore_t sig_sem;
+static Semaphore sr_semaphore;
+
 void os::signal_init_pd() {
   // Initialize signal structures
   ::memset((void*)pending_signals, 0, sizeof(pending_signals));
@@ -3791,12 +3892,14 @@
       pthread_sigmask(SIG_BLOCK, NULL, &suspend_set);
       sigdelset(&suspend_set, SR_signum);
 
+      sr_semaphore.signal();
       // wait here until we are resumed
       while (1) {
         sigsuspend(&suspend_set);
 
         os::SuspendResume::State result = osthread->sr.running();
         if (result == os::SuspendResume::SR_RUNNING) {
+          sr_semaphore.signal();
           break;
         } else if (result != os::SuspendResume::SR_SUSPENDED) {
           ShouldNotReachHere();
@@ -3815,7 +3918,7 @@
   } else if (current == os::SuspendResume::SR_WAKEUP_REQUEST) {
     // ignore
   } else {
-    ShouldNotReachHere();
+    // ignore
   }
 
   errno = old_errno;
@@ -3879,8 +3982,9 @@
 // but this seems the normal response to library errors
 static bool do_suspend(OSThread* osthread) {
   assert(osthread->sr.is_running(), "thread should be running");
+  assert(!sr_semaphore.trywait(), "semaphore has invalid state");
+
   // mark as suspended and send signal
-
   if (osthread->sr.request_suspend() != os::SuspendResume::SR_SUSPEND_REQUEST) {
     // failed to switch, state wasn't running?
     ShouldNotReachHere();
@@ -3888,35 +3992,22 @@
   }
 
   if (sr_notify(osthread) != 0) {
-    // try to cancel, switch to running
-
-    os::SuspendResume::State result = osthread->sr.cancel_suspend();
-    if (result == os::SuspendResume::SR_RUNNING) {
-      // cancelled
-      return false;
-    } else if (result == os::SuspendResume::SR_SUSPENDED) {
-      // somehow managed to suspend
-      return true;
+    ShouldNotReachHere();
+  }
+
+  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
+  while (true) {
+    if (sr_semaphore.timedwait(0, 2 * NANOSECS_PER_MILLISEC)) {
+      break;
     } else {
-      ShouldNotReachHere();
-      return false;
-    }
-  }
-
-  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
-
-  for (int n = 0; !osthread->sr.is_suspended(); n++) {
-    for (int i = 0; i < RANDOMLY_LARGE_INTEGER2 && !osthread->sr.is_suspended(); i++) {
-      os::yield_all(i);
-    }
-
-    // timeout, try to cancel the request
-    if (n >= RANDOMLY_LARGE_INTEGER) {
+      // timeout
       os::SuspendResume::State cancelled = osthread->sr.cancel_suspend();
       if (cancelled == os::SuspendResume::SR_RUNNING) {
         return false;
       } else if (cancelled == os::SuspendResume::SR_SUSPENDED) {
-        return true;
+        // make sure that we consume the signal on the semaphore as well
+        sr_semaphore.wait();
+        break;
       } else {
         ShouldNotReachHere();
         return false;
@@ -3930,6 +4021,7 @@
 
 static void do_resume(OSThread* osthread) {
   assert(osthread->sr.is_suspended(), "thread should be suspended");
+  assert(!sr_semaphore.trywait(), "invalid semaphore state");
 
   if (osthread->sr.request_wakeup() != os::SuspendResume::SR_WAKEUP_REQUEST) {
     // failed to switch to WAKEUP_REQUEST
@@ -3937,11 +4029,11 @@
     return;
   }
 
-  while (!osthread->sr.is_running()) {
+  while (true) {
     if (sr_notify(osthread) == 0) {
-      for (int n = 0; n < RANDOMLY_LARGE_INTEGER && !osthread->sr.is_running(); n++) {
-        for (int i = 0; i < 100 && !osthread->sr.is_running(); i++) {
-          os::yield_all(i);
+      if (sr_semaphore.timedwait(0, 2 * NANOSECS_PER_MILLISEC)) {
+        if (osthread->sr.is_running()) {
+          return;
         }
       }
     } else {
@@ -5936,3 +6028,4 @@
 
   return n;
 }
+
--- a/src/os/bsd/vm/os_bsd.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/bsd/vm/os_bsd.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -304,7 +304,7 @@
     int  TryPark () ;
     int  park (jlong millis) ;
     void SetAssociation (Thread * a) { _Assoc = a ; }
-} ;
+};
 
 class PlatformParker : public CHeapObj<mtInternal> {
   protected:
@@ -322,6 +322,6 @@
       status = pthread_mutex_init (_mutex, NULL);
       assert_status(status == 0, status, "mutex_init");
     }
-} ;
+};
 
 #endif // OS_BSD_VM_OS_BSD_HPP
--- a/src/os/linux/vm/os_linux.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/linux/vm/os_linux.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -163,6 +163,9 @@
 /* Used to protect dlsym() calls */
 static pthread_mutex_t dl_mutex;
 
+// Declarations
+static void unpackTime(timespec* absTime, bool isAbsolute, jlong time);
+
 #ifdef JAVASE_EMBEDDED
 class MemNotifyThread: public Thread {
   friend class VMStructs;
@@ -2434,6 +2437,57 @@
   return CAST_FROM_FN_PTR(void*, UserHandler);
 }
 
+class Semaphore : public StackObj {
+  public:
+    Semaphore();
+    ~Semaphore();
+    void signal();
+    void wait();
+    bool trywait();
+    bool timedwait(unsigned int sec, int nsec);
+  private:
+    sem_t _semaphore;
+};
+
+
+Semaphore::Semaphore() {
+  sem_init(&_semaphore, 0, 0);
+}
+
+Semaphore::~Semaphore() {
+  sem_destroy(&_semaphore);
+}
+
+void Semaphore::signal() {
+  sem_post(&_semaphore);
+}
+
+void Semaphore::wait() {
+  sem_wait(&_semaphore);
+}
+
+bool Semaphore::trywait() {
+  return sem_trywait(&_semaphore) == 0;
+}
+
+bool Semaphore::timedwait(unsigned int sec, int nsec) {
+  struct timespec ts;
+  unpackTime(&ts, false, (sec * NANOSECS_PER_SEC) + nsec);
+
+  while (1) {
+    int result = sem_timedwait(&_semaphore, &ts);
+    if (result == 0) {
+      return true;
+    } else if (errno == EINTR) {
+      continue;
+    } else if (errno == ETIMEDOUT) {
+      return false;
+    } else {
+      return false;
+    }
+  }
+}
+
 extern "C" {
   typedef void (*sa_handler_t)(int);
   typedef void (*sa_sigaction_t)(int, siginfo_t *, void *);
@@ -2473,6 +2527,7 @@
 
 // Linux(POSIX) specific hand shaking semaphore.
 static sem_t sig_sem;
+static Semaphore sr_semaphore;
 
 void os::signal_init_pd() {
   // Initialize signal structures
@@ -3628,12 +3683,14 @@
       pthread_sigmask(SIG_BLOCK, NULL, &suspend_set);
       sigdelset(&suspend_set, SR_signum);
 
+      sr_semaphore.signal();
       // wait here until we are resumed
       while (1) {
         sigsuspend(&suspend_set);
 
         os::SuspendResume::State result = osthread->sr.running();
         if (result == os::SuspendResume::SR_RUNNING) {
+          sr_semaphore.signal();
           break;
         }
       }
@@ -3650,7 +3707,7 @@
   } else if (current == os::SuspendResume::SR_WAKEUP_REQUEST) {
     // ignore
   } else {
-    ShouldNotReachHere();
+    // ignore
   }
 
   errno = old_errno;
@@ -3714,8 +3771,9 @@
 // but this seems the normal response to library errors
 static bool do_suspend(OSThread* osthread) {
   assert(osthread->sr.is_running(), "thread should be running");
+  assert(!sr_semaphore.trywait(), "semaphore has invalid state");
+
   // mark as suspended and send signal
-
   if (osthread->sr.request_suspend() != os::SuspendResume::SR_SUSPEND_REQUEST) {
     // failed to switch, state wasn't running?
     ShouldNotReachHere();
@@ -3723,35 +3781,22 @@
   }
 
   if (sr_notify(osthread) != 0) {
-    // try to cancel, switch to running
-
-    os::SuspendResume::State result = osthread->sr.cancel_suspend();
-    if (result == os::SuspendResume::SR_RUNNING) {
-      // cancelled
-      return false;
-    } else if (result == os::SuspendResume::SR_SUSPENDED) {
-      // somehow managed to suspend
-      return true;
+    ShouldNotReachHere();
+  }
+
+  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
+  while (true) {
+    if (sr_semaphore.timedwait(0, 2 * NANOSECS_PER_MILLISEC)) {
+      break;
     } else {
-      ShouldNotReachHere();
-      return false;
-    }
-  }
-
-  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
-
-  for (int n = 0; !osthread->sr.is_suspended(); n++) {
-    for (int i = 0; i < RANDOMLY_LARGE_INTEGER2 && !osthread->sr.is_suspended(); i++) {
-      os::yield_all(i);
-    }
-
-    // timeout, try to cancel the request
-    if (n >= RANDOMLY_LARGE_INTEGER) {
+      // timeout
       os::SuspendResume::State cancelled = osthread->sr.cancel_suspend();
       if (cancelled == os::SuspendResume::SR_RUNNING) {
         return false;
       } else if (cancelled == os::SuspendResume::SR_SUSPENDED) {
-        return true;
+        // make sure that we consume the signal on the semaphore as well
+        sr_semaphore.wait();
+        break;
       } else {
         ShouldNotReachHere();
         return false;
@@ -3765,6 +3810,7 @@
 
 static void do_resume(OSThread* osthread) {
   assert(osthread->sr.is_suspended(), "thread should be suspended");
+  assert(!sr_semaphore.trywait(), "invalid semaphore state");
 
   if (osthread->sr.request_wakeup() != os::SuspendResume::SR_WAKEUP_REQUEST) {
     // failed to switch to WAKEUP_REQUEST
@@ -3772,11 +3818,11 @@
     return;
   }
 
-  while (!osthread->sr.is_running()) {
+  while (true) {
     if (sr_notify(osthread) == 0) {
-      for (int n = 0; n < RANDOMLY_LARGE_INTEGER && !osthread->sr.is_running(); n++) {
-        for (int i = 0; i < 100 && !osthread->sr.is_running(); i++) {
-          os::yield_all(i);
+      if (sr_semaphore.timedwait(0, 2 * NANOSECS_PER_MILLISEC)) {
+        if (osthread->sr.is_running()) {
+          return;
         }
       }
     } else {
@@ -5748,4 +5794,5 @@
     new MemNotifyThread(fd);
   }
 }
+
 #endif // JAVASE_EMBEDDED
--- a/src/os/linux/vm/os_linux.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/linux/vm/os_linux.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -301,6 +301,6 @@
       status = pthread_mutex_init (_mutex, NULL);
       assert_status(status == 0, status, "mutex_init");
     }
-} ;
+};
 
 #endif // OS_LINUX_VM_OS_LINUX_HPP
--- a/src/os/solaris/vm/osThread_solaris.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/solaris/vm/osThread_solaris.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -66,59 +66,6 @@
   }
 }
 
-static void resume_clear_context(OSThread *osthread) {
-  osthread->set_ucontext(NULL);
+void OSThread::SR_handler(Thread* thread, ucontext_t* uc) {
+  os::Solaris::SR_handler(thread, uc);
 }
-
-static void suspend_save_context(OSThread *osthread, ucontext_t* context) {
-  osthread->set_ucontext(context);
-}
-
-void OSThread::SR_handler(Thread* thread, ucontext_t* uc) {
-  // Save and restore errno to avoid confusing native code with EINTR
-  // after sigsuspend.
-  int old_errno = errno;
-
-  OSThread* osthread = thread->osthread();
-  assert(thread->is_VM_thread() || thread->is_Java_thread(), "Must be VMThread or JavaThread");
-
-  os::SuspendResume::State current = osthread->sr.state();
-  if (current == os::SuspendResume::SR_SUSPEND_REQUEST) {
-    suspend_save_context(osthread, uc);
-
-    // attempt to switch the state, we assume we had a SUSPEND_REQUEST
-    os::SuspendResume::State state = osthread->sr.suspended();
-    if (state == os::SuspendResume::SR_SUSPENDED) {
-      sigset_t suspend_set;  // signals for sigsuspend()
-
-      // get current set of blocked signals and unblock resume signal
-      thr_sigsetmask(SIG_BLOCK, NULL, &suspend_set);
-      sigdelset(&suspend_set, os::Solaris::SIGasync());
-
-      // wait here until we are resumed
-      while (1) {
-        sigsuspend(&suspend_set);
-
-        os::SuspendResume::State result = osthread->sr.running();
-        if (result == os::SuspendResume::SR_RUNNING) {
-          break;
-        }
-      }
-
-    } else if (state == os::SuspendResume::SR_RUNNING) {
-      // request was cancelled, continue
-    } else {
-      ShouldNotReachHere();
-    }
-
-    resume_clear_context(osthread);
-  } else if (current == os::SuspendResume::SR_RUNNING) {
-    // request was cancelled, continue
-  } else if (current == os::SuspendResume::SR_WAKEUP_REQUEST) {
-    // ignore
-  } else {
-    ShouldNotReachHere();
-  }
-
-  errno = old_errno;
-}
--- a/src/os/solaris/vm/os_solaris.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/solaris/vm/os_solaris.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -247,6 +247,8 @@
   static int pthread_cond_default_init(cond_t *cv, int scope, void *arg){ memset(cv, 0, sizeof(cond_t)); return 0; }
 }
 
+static void unpackTime(timespec* absTime, bool isAbsolute, jlong time);
+
 // Thread Local Storage
 // This is common to all Solaris platforms so it is defined here,
 // in this common file.
@@ -2604,6 +2606,57 @@
   return CAST_FROM_FN_PTR(void*, UserHandler);
 }
 
+class Semaphore : public StackObj {
+  public:
+    Semaphore();
+    ~Semaphore();
+    void signal();
+    void wait();
+    bool trywait();
+    bool timedwait(unsigned int sec, int nsec);
+  private:
+    sema_t _semaphore;
+};
+
+
+Semaphore::Semaphore() {
+  sema_init(&_semaphore, 0, NULL, NULL);
+}
+
+Semaphore::~Semaphore() {
+  sema_destroy(&_semaphore);
+}
+
+void Semaphore::signal() {
+  sema_post(&_semaphore);
+}
+
+void Semaphore::wait() {
+  sema_wait(&_semaphore);
+}
+
+bool Semaphore::trywait() {
+  return sema_trywait(&_semaphore) == 0;
+}
+
+bool Semaphore::timedwait(unsigned int sec, int nsec) {
+  struct timespec ts;
+  unpackTime(&ts, false, (sec * NANOSECS_PER_SEC) + nsec);
+
+  while (1) {
+    int result = sema_timedwait(&_semaphore, &ts);
+    if (result == 0) {
+      return true;
+    } else if (errno == EINTR) {
+      continue;
+    } else if (errno == ETIME) {
+      return false;
+    } else {
+      return false;
+    }
+  }
+}
+
 extern "C" {
   typedef void (*sa_handler_t)(int);
   typedef void (*sa_sigaction_t)(int, siginfo_t *, void *);
@@ -4188,6 +4241,68 @@
   schedctl_start(schedctl_init());
 }
 
+static void resume_clear_context(OSThread *osthread) {
+  osthread->set_ucontext(NULL);
+}
+
+static void suspend_save_context(OSThread *osthread, ucontext_t* context) {
+  osthread->set_ucontext(context);
+}
+
+static Semaphore sr_semaphore;
+
+void os::Solaris::SR_handler(Thread* thread, ucontext_t* uc) {
+  // Save and restore errno to avoid confusing native code with EINTR
+  // after sigsuspend.
+  int old_errno = errno;
+
+  OSThread* osthread = thread->osthread();
+  assert(thread->is_VM_thread() || thread->is_Java_thread(), "Must be VMThread or JavaThread");
+
+  os::SuspendResume::State current = osthread->sr.state();
+  if (current == os::SuspendResume::SR_SUSPEND_REQUEST) {
+    suspend_save_context(osthread, uc);
+
+    // attempt to switch the state, we assume we had a SUSPEND_REQUEST
+    os::SuspendResume::State state = osthread->sr.suspended();
+    if (state == os::SuspendResume::SR_SUSPENDED) {
+      sigset_t suspend_set;  // signals for sigsuspend()
+
+      // get current set of blocked signals and unblock resume signal
+      thr_sigsetmask(SIG_BLOCK, NULL, &suspend_set);
+      sigdelset(&suspend_set, os::Solaris::SIGasync());
+
+      sr_semaphore.signal();
+      // wait here until we are resumed
+      while (1) {
+        sigsuspend(&suspend_set);
+
+        os::SuspendResume::State result = osthread->sr.running();
+        if (result == os::SuspendResume::SR_RUNNING) {
+          sr_semaphore.signal();
+          break;
+        }
+      }
+
+    } else if (state == os::SuspendResume::SR_RUNNING) {
+      // request was cancelled, continue
+    } else {
+      ShouldNotReachHere();
+    }
+
+    resume_clear_context(osthread);
+  } else if (current == os::SuspendResume::SR_RUNNING) {
+    // request was cancelled, continue
+  } else if (current == os::SuspendResume::SR_WAKEUP_REQUEST) {
+    // ignore
+  } else {
+    // ignore
+  }
+
+  errno = old_errno;
+}
+
+
 void os::interrupt(Thread* thread) {
   assert(Thread::current() == thread || Threads_lock->owned_by_self(), "possibility of dangling Thread pointer");
 
@@ -4285,8 +4400,9 @@
 
 static bool do_suspend(OSThread* osthread) {
   assert(osthread->sr.is_running(), "thread should be running");
+  assert(!sr_semaphore.trywait(), "semaphore has invalid state");
+
   // mark as suspended and send signal
-
   if (osthread->sr.request_suspend() != os::SuspendResume::SR_SUSPEND_REQUEST) {
     // failed to switch, state wasn't running?
     ShouldNotReachHere();
@@ -4294,35 +4410,22 @@
   }
 
   if (sr_notify(osthread) != 0) {
-    // try to cancel, switch to running
-
-    os::SuspendResume::State result = osthread->sr.cancel_suspend();
-    if (result == os::SuspendResume::SR_RUNNING) {
-      // cancelled
-      return false;
-    } else if (result == os::SuspendResume::SR_SUSPENDED) {
-      // somehow managed to suspend
-      return true;
+    ShouldNotReachHere();
+  }
+
+  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
+  while (true) {
+    if (sr_semaphore.timedwait(0, 2000 * NANOSECS_PER_MILLISEC)) {
+      break;
     } else {
-      ShouldNotReachHere();
-      return false;
-    }
-  }
-
-  // managed to send the signal and switch to SUSPEND_REQUEST, now wait for SUSPENDED
-
-  for (int n = 0; !osthread->sr.is_suspended(); n++) {
-    for (int i = 0; i < RANDOMLY_LARGE_INTEGER2 && !osthread->sr.is_suspended(); i++) {
-      os::yield_all(i);
-    }
-
-    // timeout, try to cancel the request
-    if (n >= RANDOMLY_LARGE_INTEGER) {
+      // timeout
       os::SuspendResume::State cancelled = osthread->sr.cancel_suspend();
       if (cancelled == os::SuspendResume::SR_RUNNING) {
         return false;
       } else if (cancelled == os::SuspendResume::SR_SUSPENDED) {
-        return true;
+        // make sure that we consume the signal on the semaphore as well
+        sr_semaphore.wait();
+        break;
       } else {
         ShouldNotReachHere();
         return false;
@@ -4336,6 +4439,7 @@
 
 static void do_resume(OSThread* osthread) {
   assert(osthread->sr.is_suspended(), "thread should be suspended");
+  assert(!sr_semaphore.trywait(), "invalid semaphore state");
 
   if (osthread->sr.request_wakeup() != os::SuspendResume::SR_WAKEUP_REQUEST) {
     // failed to switch to WAKEUP_REQUEST
@@ -4343,11 +4447,11 @@
     return;
   }
 
-  while (!osthread->sr.is_running()) {
+  while (true) {
     if (sr_notify(osthread) == 0) {
-      for (int n = 0; n < RANDOMLY_LARGE_INTEGER && !osthread->sr.is_running(); n++) {
-        for (int i = 0; i < 100 && !osthread->sr.is_running(); i++) {
-          os::yield_all(i);
+      if (sr_semaphore.timedwait(0, 2 * NANOSECS_PER_MILLISEC)) {
+        if (osthread->sr.is_running()) {
+          return;
         }
       }
     } else {
--- a/src/os/solaris/vm/os_solaris.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os/solaris/vm/os_solaris.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -127,7 +127,6 @@
   static void set_SIGinterrupt(int newsig) { _SIGinterrupt = newsig; }
   static void set_SIGasync(int newsig) { _SIGasync = newsig; }
 
-
  public:
   // Large Page Support--ISM.
   static bool largepage_range(char* addr, size_t size);
@@ -158,6 +157,8 @@
 
   static void hotspot_sigmask(Thread* thread);
 
+  // SR_handler
+  static void SR_handler(Thread* thread, ucontext_t* uc);
  protected:
   // Solaris-specific interface goes here
   static julong available_memory();
--- a/src/os_cpu/bsd_x86/vm/bsd_x86_64.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os_cpu/bsd_x86/vm/bsd_x86_64.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -55,20 +55,6 @@
   // adding a syntax that specifies the sizes of fields in an order,
   // so that the adlc can build the emit functions automagically
 
-  enc_class Java_To_Runtime(method meth) %{
-    // No relocation needed
-
-    // movq r10, <meth>
-    emit_opcode(cbuf, Assembler::REX_WB);
-    emit_opcode(cbuf, 0xB8 | (R10_enc - 8));
-    emit_d64(cbuf, (int64_t) $meth$$method);
-
-    // call (r10)
-    emit_opcode(cbuf, Assembler::REX_B);
-    emit_opcode(cbuf, 0xFF);
-    emit_opcode(cbuf, 0xD0 | (R10_enc - 8));
-  %}
-
 %}
 
 
@@ -76,8 +62,4 @@
 
 source %{
 
-int MachCallRuntimeNode::ret_addr_offset() {
-  return 13; // movq r10,#addr; callq (r10)
-}
-
 %}
--- a/src/os_cpu/linux_x86/vm/linux_x86_64.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os_cpu/linux_x86/vm/linux_x86_64.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -55,20 +55,6 @@
   // adding a syntax that specifies the sizes of fields in an order,
   // so that the adlc can build the emit functions automagically
 
-  enc_class Java_To_Runtime(method meth) %{
-    // No relocation needed
-
-    // movq r10, <meth>
-    emit_opcode(cbuf, Assembler::REX_WB);
-    emit_opcode(cbuf, 0xB8 | (R10_enc - 8));
-    emit_d64(cbuf, (int64_t) $meth$$method);
-
-    // call (r10)
-    emit_opcode(cbuf, Assembler::REX_B);
-    emit_opcode(cbuf, 0xFF);
-    emit_opcode(cbuf, 0xD0 | (R10_enc - 8));
-  %}
-
 %}
 
 
@@ -76,8 +62,4 @@
 
 source %{
 
-int MachCallRuntimeNode::ret_addr_offset() {
-  return 13; // movq r10,#addr; callq (r10)
-}
-
 %}
--- a/src/os_cpu/solaris_x86/vm/solaris_x86_64.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os_cpu/solaris_x86/vm/solaris_x86_64.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -54,39 +54,10 @@
   // main source block for now.  In future, we can generalize this by
   // adding a syntax that specifies the sizes of fields in an order,
   // so that the adlc can build the emit functions automagically
-
-  enc_class Java_To_Runtime(method meth) %{
-    // No relocation needed
-
-    // movq r10, <meth>
-    emit_opcode(cbuf, Assembler::REX_WB);
-    emit_opcode(cbuf, 0xB8 | (R10_enc - 8));
-    emit_d64(cbuf, (int64_t) $meth$$method);
-
-    // call (r10)
-    emit_opcode(cbuf, Assembler::REX_B);
-    emit_opcode(cbuf, 0xFF);
-    emit_opcode(cbuf, 0xD0 | (R10_enc - 8));
-  %}
-
-  enc_class post_call_verify_mxcsr %{
-    MacroAssembler _masm(&cbuf);
-    if (RestoreMXCSROnJNICalls) {
-      __ ldmxcsr(ExternalAddress(StubRoutines::amd64::mxcsr_std()));
-    }
-    else if (CheckJNICalls) {
-      __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, StubRoutines::amd64::verify_mxcsr_entry())));
-    }
-  %}
 %}
 
 
 // Platform dependent source
 
 source %{
-
-int MachCallRuntimeNode::ret_addr_offset() {
-  return 13; // movq r10,#addr; callq (r10)
-}
-
 %}
--- a/src/os_cpu/windows_x86/vm/windows_x86_64.ad	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/os_cpu/windows_x86/vm/windows_x86_64.ad	Thu Apr 25 08:49:05 2013 -0700
@@ -53,30 +53,11 @@
   // adding a syntax that specifies the sizes of fields in an order,
   // so that the adlc can build the emit functions automagically
 
-  enc_class Java_To_Runtime (method meth) %{    // CALL Java_To_Runtime
-    // No relocation needed
+%}
 
-    // movq r10, <meth>
-    emit_opcode(cbuf, Assembler::REX_WB);
-    emit_opcode(cbuf, 0xB8 | (R10_enc - 8));
-    emit_d64(cbuf, (int64_t) $meth$$method);
 
-    // call (r10)
-    emit_opcode(cbuf, Assembler::REX_B);
-    emit_opcode(cbuf, 0xFF);
-    emit_opcode(cbuf, 0xD0 | (R10_enc - 8));
-  %}
+// Platform dependent source
+
+source %{
 
 %}
-
-//
-// Platform dependent source
-//
-source %{
-
-int MachCallRuntimeNode::ret_addr_offset()
-{
-  return 13; // movq r10,#addr; callq (r10)
-}
-
-%}
--- a/src/share/vm/adlc/archDesc.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/adlc/archDesc.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -832,6 +832,7 @@
     int         length  = (int)strlen(rc_name) + (int)strlen(mask) + 5;
     char       *regMask = new char[length];
     sprintf(regMask,"%s%s()", rc_name, mask);
+    delete[] rc_name;
     return regMask;
   }
 }
--- a/src/share/vm/adlc/dfa.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/adlc/dfa.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -191,12 +191,19 @@
 // Macro equivalent to: _kids[0]->valid(FOO) && _kids[1]->valid(BAR)
 //
 static void child_test(FILE *fp, MatchList &mList) {
-  if( mList._lchild )           // If left child, check it
-    fprintf(fp, "STATE__VALID_CHILD(_kids[0], %s)", ArchDesc::getMachOperEnum(mList._lchild));
-  if( mList._lchild && mList._rchild )      // If both, add the "&&"
-    fprintf(fp, " && " );
-  if( mList._rchild )           // If right child, check it
-    fprintf(fp, "STATE__VALID_CHILD(_kids[1], %s)", ArchDesc::getMachOperEnum(mList._rchild));
+  if (mList._lchild) { // If left child, check it
+    const char* lchild_to_upper = ArchDesc::getMachOperEnum(mList._lchild);
+    fprintf(fp, "STATE__VALID_CHILD(_kids[0], %s)", lchild_to_upper);
+    delete[] lchild_to_upper;
+  }
+  if (mList._lchild && mList._rchild) { // If both, add the "&&"
+    fprintf(fp, " && ");
+  }
+  if (mList._rchild) { // If right child, check it
+    const char* rchild_to_upper = ArchDesc::getMachOperEnum(mList._rchild);
+    fprintf(fp, "STATE__VALID_CHILD(_kids[1], %s)", rchild_to_upper);
+    delete[] rchild_to_upper;
+  }
 }
 
 //---------------------------calc_cost-----------------------------------------
@@ -206,13 +213,17 @@
 Expr *ArchDesc::calc_cost(FILE *fp, const char *spaces, MatchList &mList, ProductionState &status) {
   fprintf(fp, "%sunsigned int c = ", spaces);
   Expr *c = new Expr("0");
-  if (mList._lchild ) {                   // If left child, add it in
-    sprintf(Expr::buffer(), "_kids[0]->_cost[%s]", ArchDesc::getMachOperEnum(mList._lchild));
+  if (mList._lchild) { // If left child, add it in
+    const char* lchild_to_upper = ArchDesc::getMachOperEnum(mList._lchild);
+    sprintf(Expr::buffer(), "_kids[0]->_cost[%s]", lchild_to_upper);
     c->add(Expr::buffer());
+    delete[] lchild_to_upper;
 }
-  if (mList._rchild) {                    // If right child, add it in
-    sprintf(Expr::buffer(), "_kids[1]->_cost[%s]", ArchDesc::getMachOperEnum(mList._rchild));
+  if (mList._rchild) { // If right child, add it in
+    const char* rchild_to_upper = ArchDesc::getMachOperEnum(mList._rchild);
+    sprintf(Expr::buffer(), "_kids[1]->_cost[%s]", rchild_to_upper);
     c->add(Expr::buffer());
+    delete[] rchild_to_upper;
   }
   // Add in cost of this rule
   const char *mList_cost = mList.get_cost();
@@ -232,15 +243,17 @@
   fprintf(fp, "%s", spaces4);
   // Only generate child tests if this is not a leaf node
   bool has_child_constraints = mList._lchild || mList._rchild;
-  const char *predicate_test        = mList.get_pred();
-  if( has_child_constraints || predicate_test ) {
+  const char *predicate_test = mList.get_pred();
+  if (has_child_constraints || predicate_test) {
     // Open the child-and-predicate-test braces
     fprintf(fp, "if( ");
     status.set_constraint(hasConstraint);
     child_test(fp, mList);
     // Only generate predicate test if one exists for this match
-    if( predicate_test ) {
-      if( has_child_constraints ) { fprintf(fp," &&\n"); }
+    if (predicate_test) {
+      if (has_child_constraints) {
+        fprintf(fp," &&\n");
+      }
       fprintf(fp, "%s  %s", spaces6, predicate_test);
     }
     // End of outer tests
--- a/src/share/vm/adlc/output_c.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/adlc/output_c.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -63,11 +63,10 @@
     RegDef *reg_def = NULL;
     RegDef *next = NULL;
     registers->reset_RegDefs();
-    for( reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next ) {
+    for (reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next) {
       next = registers->iter_RegDefs();
       const char *comma = (next != NULL) ? "," : " // no trailing comma";
-      fprintf(fp,"  \"%s\"%s\n",
-                 reg_def->_regname, comma );
+      fprintf(fp,"  \"%s\"%s\n", reg_def->_regname, comma);
     }
 
     // Finish defining enumeration
@@ -79,10 +78,10 @@
     reg_def = NULL;
     next = NULL;
     registers->reset_RegDefs();
-    for( reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next ) {
+    for (reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next) {
       next = registers->iter_RegDefs();
       const char *comma = (next != NULL) ? "," : " // no trailing comma";
-      fprintf(fp,"\t%s%s\n", reg_def->_concrete, comma );
+      fprintf(fp,"\t%s%s\n", reg_def->_concrete, comma);
     }
     // Finish defining array
     fprintf(fp,"\t};\n");
@@ -104,19 +103,17 @@
     RegDef *reg_def = NULL;
     RegDef *next    = NULL;
     registers->reset_RegDefs();
-    for( reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next ) {
+    for (reg_def = registers->iter_RegDefs(); reg_def != NULL; reg_def = next) {
       next = registers->iter_RegDefs();
       const char* register_encode = reg_def->register_encode();
       const char *comma = (next != NULL) ? "," : " // no trailing comma";
       int encval;
       if (!ADLParser::is_int_token(register_encode, encval)) {
-        fprintf(fp,"  %s%s  // %s\n",
-                register_encode, comma, reg_def->_regname );
+        fprintf(fp,"  %s%s  // %s\n", register_encode, comma, reg_def->_regname);
       } else {
         // Output known constants in hex char format (backward compatibility).
         assert(encval < 256, "Exceeded supported width for register encoding");
-        fprintf(fp,"  (unsigned char)'\\x%X'%s  // %s\n",
-                encval,          comma, reg_def->_regname );
+        fprintf(fp,"  (unsigned char)'\\x%X'%s  // %s\n", encval, comma, reg_def->_regname);
       }
     }
     // Finish defining enumeration
@@ -133,9 +130,10 @@
     fprintf(fp,"// Enumeration of register class names\n");
     fprintf(fp, "enum machRegisterClass {\n");
     registers->_rclasses.reset();
-    for( const char *class_name = NULL;
-         (class_name = registers->_rclasses.iter()) != NULL; ) {
-      fprintf(fp,"  %s,\n", toUpper( class_name ));
+    for (const char *class_name = NULL; (class_name = registers->_rclasses.iter()) != NULL;) {
+      const char * class_name_to_upper = toUpper(class_name);
+      fprintf(fp,"  %s,\n", class_name_to_upper);
+      delete[] class_name_to_upper;
     }
     // Finish defining enumeration
     fprintf(fp, "  _last_Mach_Reg_Class\n");
@@ -148,7 +146,7 @@
 void ArchDesc::declare_register_masks(FILE *fp_hpp) {
   const char  *rc_name;
 
-  if( _register ) {
+  if (_register) {
     // Build enumeration of user-defined register classes.
     defineRegClassEnum(fp_hpp, _register);
 
@@ -156,24 +154,27 @@
     fprintf(fp_hpp,"\n");
     fprintf(fp_hpp,"// Register masks, one for each register class.\n");
     _register->_rclasses.reset();
-    for( rc_name = NULL;
-         (rc_name = _register->_rclasses.iter()) != NULL; ) {
-      const char *prefix    = "";
-      RegClass   *reg_class = _register->getRegClass(rc_name);
-      assert( reg_class, "Using an undefined register class");
+    for (rc_name = NULL; (rc_name = _register->_rclasses.iter()) != NULL;) {
+      const char *prefix = "";
+      RegClass *reg_class = _register->getRegClass(rc_name);
+      assert(reg_class, "Using an undefined register class");
+
+      const char* rc_name_to_upper = toUpper(rc_name);
 
       if (reg_class->_user_defined == NULL) {
-        fprintf(fp_hpp, "extern const RegMask _%s%s_mask;\n", prefix, toUpper( rc_name ) );
-        fprintf(fp_hpp, "inline const RegMask &%s%s_mask() { return _%s%s_mask; }\n", prefix, toUpper( rc_name ), prefix, toUpper( rc_name ));
+        fprintf(fp_hpp, "extern const RegMask _%s%s_mask;\n", prefix,  rc_name_to_upper);
+        fprintf(fp_hpp, "inline const RegMask &%s%s_mask() { return _%s%s_mask; }\n", prefix, rc_name_to_upper, prefix, rc_name_to_upper);
       } else {
-        fprintf(fp_hpp, "inline const RegMask &%s%s_mask() { %s }\n", prefix, toUpper( rc_name ), reg_class->_user_defined);
+        fprintf(fp_hpp, "inline const RegMask &%s%s_mask() { %s }\n", prefix, rc_name_to_upper, reg_class->_user_defined);
       }
 
-      if( reg_class->_stack_or_reg ) {
+      if (reg_class->_stack_or_reg) {
         assert(reg_class->_user_defined == NULL, "no user defined reg class here");
-        fprintf(fp_hpp, "extern const RegMask _%sSTACK_OR_%s_mask;\n", prefix, toUpper( rc_name ) );
-        fprintf(fp_hpp, "inline const RegMask &%sSTACK_OR_%s_mask() { return _%sSTACK_OR_%s_mask; }\n", prefix, toUpper( rc_name ), prefix, toUpper( rc_name ) );
+        fprintf(fp_hpp, "extern const RegMask _%sSTACK_OR_%s_mask;\n", prefix, rc_name_to_upper);
+        fprintf(fp_hpp, "inline const RegMask &%sSTACK_OR_%s_mask() { return _%sSTACK_OR_%s_mask; }\n", prefix, rc_name_to_upper, prefix, rc_name_to_upper);
       }
+      delete[] rc_name_to_upper;
+
     }
   }
 }
@@ -183,34 +184,41 @@
 void ArchDesc::build_register_masks(FILE *fp_cpp) {
   const char  *rc_name;
 
-  if( _register ) {
+  if (_register) {
     // Generate a list of register masks, one for each class.
     fprintf(fp_cpp,"\n");
     fprintf(fp_cpp,"// Register masks, one for each register class.\n");
     _register->_rclasses.reset();
-    for( rc_name = NULL;
-         (rc_name = _register->_rclasses.iter()) != NULL; ) {
-      const char *prefix    = "";
-      RegClass   *reg_class = _register->getRegClass(rc_name);
-      assert( reg_class, "Using an undefined register class");
-
-      if (reg_class->_user_defined != NULL) continue;
+    for (rc_name = NULL; (rc_name = _register->_rclasses.iter()) != NULL;) {
+      const char *prefix = "";
+      RegClass *reg_class = _register->getRegClass(rc_name);
+      assert(reg_class, "Using an undefined register class");
+
+      if (reg_class->_user_defined != NULL) {
+        continue;
+      }
 
       int len = RegisterForm::RegMask_Size();
-      fprintf(fp_cpp, "const RegMask _%s%s_mask(", prefix, toUpper( rc_name ) );
-      { int i;
-        for( i = 0; i < len-1; i++ )
-          fprintf(fp_cpp," 0x%x,",reg_class->regs_in_word(i,false));
-        fprintf(fp_cpp," 0x%x );\n",reg_class->regs_in_word(i,false));
+      const char* rc_name_to_upper = toUpper(rc_name);
+      fprintf(fp_cpp, "const RegMask _%s%s_mask(", prefix, rc_name_to_upper);
+
+      {
+        int i;
+        for(i = 0; i < len - 1; i++) {
+          fprintf(fp_cpp," 0x%x,", reg_class->regs_in_word(i, false));
+        }
+        fprintf(fp_cpp," 0x%x );\n", reg_class->regs_in_word(i, false));
       }
 
-      if( reg_class->_stack_or_reg ) {
+      if (reg_class->_stack_or_reg) {
         int i;
-        fprintf(fp_cpp, "const RegMask _%sSTACK_OR_%s_mask(", prefix, toUpper( rc_name ) );
-        for( i = 0; i < len-1; i++ )
-          fprintf(fp_cpp," 0x%x,",reg_class->regs_in_word(i,true));
-        fprintf(fp_cpp," 0x%x );\n",reg_class->regs_in_word(i,true));
+        fprintf(fp_cpp, "const RegMask _%sSTACK_OR_%s_mask(", prefix, rc_name_to_upper);
+        for(i = 0; i < len - 1; i++) {
+          fprintf(fp_cpp," 0x%x,",reg_class->regs_in_word(i, true));
+        }
+        fprintf(fp_cpp," 0x%x );\n",reg_class->regs_in_word(i, true));
       }
+      delete[] rc_name_to_upper;
     }
   }
 }
@@ -2685,7 +2693,9 @@
       if (strcmp(first_reg_class, "stack_slots") == 0) {
         fprintf(fp,"  return &(Compile::current()->FIRST_STACK_mask());\n");
       } else {
-        fprintf(fp,"  return &%s_mask();\n", toUpper(first_reg_class));
+        const char* first_reg_class_to_upper = toUpper(first_reg_class);
+        fprintf(fp,"  return &%s_mask();\n", first_reg_class_to_upper);
+        delete[] first_reg_class_to_upper;
       }
     } else {
       // Build a switch statement to return the desired mask.
@@ -2697,7 +2707,9 @@
         if( !strcmp(reg_class, "stack_slots") ) {
           fprintf(fp, "  case %d: return &(Compile::current()->FIRST_STACK_mask());\n", index);
         } else {
-          fprintf(fp, "  case %d: return &%s_mask();\n", index, toUpper(reg_class));
+          const char* reg_class_to_upper = toUpper(reg_class);
+          fprintf(fp, "  case %d: return &%s_mask();\n", index, reg_class_to_upper);
+          delete[] reg_class_to_upper;
         }
       }
       fprintf(fp,"  }\n");
--- a/src/share/vm/adlc/output_h.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/adlc/output_h.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2040,9 +2040,21 @@
   void closing()     { fprintf(_cpp, "  _LAST_MACH_OPER\n");
                        OutputMap::closing();
   }
-  void map(OpClassForm &opc)  { fprintf(_cpp, "  %s", _AD.machOperEnum(opc._ident) ); }
-  void map(OperandForm &oper) { fprintf(_cpp, "  %s", _AD.machOperEnum(oper._ident) ); }
-  void map(char        *name) { fprintf(_cpp, "  %s", _AD.machOperEnum(name)); }
+  void map(OpClassForm &opc)  {
+    const char* opc_ident_to_upper = _AD.machOperEnum(opc._ident);
+    fprintf(_cpp, "  %s", opc_ident_to_upper);
+    delete[] opc_ident_to_upper;
+  }
+  void map(OperandForm &oper) {
+    const char* oper_ident_to_upper = _AD.machOperEnum(oper._ident);
+    fprintf(_cpp, "  %s", oper_ident_to_upper);
+    delete[] oper_ident_to_upper;
+  }
+  void map(char *name) {
+    const char* name_to_upper = _AD.machOperEnum(name);
+    fprintf(_cpp, "  %s", name_to_upper);
+    delete[] name_to_upper;
+  }
 
   bool do_instructions()      { return false; }
   void map(InstructForm &inst){ assert( false, "ShouldNotCallThis()"); }
--- a/src/share/vm/classfile/systemDictionary.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/classfile/systemDictionary.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -2026,12 +2026,6 @@
     }
   }
 
-  // Assign a classid if one has not already been assigned.  The
-  // counter does not need to be atomically incremented since this
-  // is only done while holding the SystemDictionary_lock.
-  // All loaded classes get a unique ID.
-  TRACE_INIT_ID(k);
-
   // Check for a placeholder. If there, remove it and make a
   // new system dictionary entry.
   placeholders()->find_and_remove(p_index, p_hash, name, class_loader, THREAD);
--- a/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -533,19 +533,14 @@
     _sleep_factor             =     0.0;
     _marking_task_overhead    =     1.0;
   } else {
-    if (ConcGCThreads > 0) {
-      // notice that ConcGCThreads overwrites G1MarkingOverheadPercent
+    if (!FLAG_IS_DEFAULT(ConcGCThreads) && ConcGCThreads > 0) {
+      // Note: ConcGCThreads has precedence over G1MarkingOverheadPercent
       // if both are set
-
-      _parallel_marking_threads = (uint) ConcGCThreads;
-      _max_parallel_marking_threads = _parallel_marking_threads;
       _sleep_factor             = 0.0;
       _marking_task_overhead    = 1.0;
     } else if (G1MarkingOverheadPercent > 0) {
-      // we will calculate the number of parallel marking threads
-      // based on a target overhead with respect to the soft real-time
-      // goal
-
+      // We will calculate the number of parallel marking threads based
+      // on a target overhead with respect to the soft real-time goal
       double marking_overhead = (double) G1MarkingOverheadPercent / 100.0;
       double overall_cm_overhead =
         (double) MaxGCPauseMillis * marking_overhead /
@@ -558,17 +553,22 @@
       double sleep_factor =
                          (1.0 - marking_task_overhead) / marking_task_overhead;
 
-      _parallel_marking_threads = (uint) marking_thread_num;
-      _max_parallel_marking_threads = _parallel_marking_threads;
+      FLAG_SET_ERGO(uintx, ConcGCThreads, (uint) marking_thread_num);
       _sleep_factor             = sleep_factor;
       _marking_task_overhead    = marking_task_overhead;
     } else {
-      _parallel_marking_threads = scale_parallel_threads((uint)ParallelGCThreads);
-      _max_parallel_marking_threads = _parallel_marking_threads;
+      // Calculate the number of parallel marking threads by scaling
+      // the number of parallel GC threads.
+      uint marking_thread_num = scale_parallel_threads((uint) ParallelGCThreads);
+      FLAG_SET_ERGO(uintx, ConcGCThreads, marking_thread_num);
       _sleep_factor             = 0.0;
       _marking_task_overhead    = 1.0;
     }
 
+    assert(ConcGCThreads > 0, "Should have been set");
+    _parallel_marking_threads = (uint) ConcGCThreads;
+    _max_parallel_marking_threads = _parallel_marking_threads;
+
     if (parallel_marking_threads() > 1) {
       _cleanup_task_overhead = 1.0;
     } else {
--- a/src/share/vm/gc_implementation/g1/concurrentMarkThread.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/gc_implementation/g1/concurrentMarkThread.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -116,7 +116,7 @@
         if (G1Log::fine()) {
           gclog_or_tty->date_stamp(PrintGCDateStamps);
           gclog_or_tty->stamp(PrintGCTimeStamps);
-          gclog_or_tty->print_cr("[GC concurrent-root-region-scan-end, %1.7lf]",
+          gclog_or_tty->print_cr("[GC concurrent-root-region-scan-end, %1.7lf secs]",
                                  scan_end - scan_start);
         }
       }
@@ -149,7 +149,7 @@
           if (G1Log::fine()) {
             gclog_or_tty->date_stamp(PrintGCDateStamps);
             gclog_or_tty->stamp(PrintGCTimeStamps);
-            gclog_or_tty->print_cr("[GC concurrent-mark-end, %1.7lf sec]",
+            gclog_or_tty->print_cr("[GC concurrent-mark-end, %1.7lf secs]",
                                       mark_end_sec - mark_start_sec);
           }
 
@@ -231,7 +231,7 @@
         if (G1Log::fine()) {
           gclog_or_tty->date_stamp(PrintGCDateStamps);
           gclog_or_tty->stamp(PrintGCTimeStamps);
-          gclog_or_tty->print_cr("[GC concurrent-cleanup-end, %1.7lf]",
+          gclog_or_tty->print_cr("[GC concurrent-cleanup-end, %1.7lf secs]",
                                  cleanup_end_sec - cleanup_start_sec);
         }
       }
--- a/src/share/vm/gc_implementation/g1/g1GCPhaseTimes.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/gc_implementation/g1/g1GCPhaseTimes.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -131,17 +131,23 @@
 
 #ifdef ASSERT
 
+template <> const int WorkerDataArray<int>::_uninitialized = -1;
+template <> const double WorkerDataArray<double>::_uninitialized = -1.0;
+template <> const size_t WorkerDataArray<size_t>::_uninitialized = (size_t)-1;
+
 template <class T>
 void WorkerDataArray<T>::reset() {
   for (uint i = 0; i < _length; i++) {
-    _data[i] = (T)-1;
+    _data[i] = (T)_uninitialized;
   }
 }
 
 template <class T>
 void WorkerDataArray<T>::verify() {
   for (uint i = 0; i < _length; i++) {
-    assert(_data[i] >= (T)0, err_msg("Invalid data for worker %d", i));
+    assert(_data[i] != _uninitialized,
+        err_msg("Invalid data for worker " UINT32_FORMAT ", data: %lf, uninitialized: %lf",
+            i, (double)_data[i], (double)_uninitialized));
   }
 }
 
@@ -201,20 +207,20 @@
   _last_termination_attempts.verify();
   _last_gc_worker_end_times_ms.verify();
 
-    for (uint i = 0; i < _active_gc_threads; i++) {
-      double worker_time = _last_gc_worker_end_times_ms.get(i) - _last_gc_worker_start_times_ms.get(i);
-      _last_gc_worker_times_ms.set(i, worker_time);
+  for (uint i = 0; i < _active_gc_threads; i++) {
+    double worker_time = _last_gc_worker_end_times_ms.get(i) - _last_gc_worker_start_times_ms.get(i);
+    _last_gc_worker_times_ms.set(i, worker_time);
 
-      double worker_known_time = _last_ext_root_scan_times_ms.get(i) +
-        _last_satb_filtering_times_ms.get(i) +
-        _last_update_rs_times_ms.get(i) +
-        _last_scan_rs_times_ms.get(i) +
-        _last_obj_copy_times_ms.get(i) +
-        _last_termination_times_ms.get(i);
+    double worker_known_time = _last_ext_root_scan_times_ms.get(i) +
+      _last_satb_filtering_times_ms.get(i) +
+      _last_update_rs_times_ms.get(i) +
+      _last_scan_rs_times_ms.get(i) +
+      _last_obj_copy_times_ms.get(i) +
+      _last_termination_times_ms.get(i);
 
-      double worker_other_time = worker_time - worker_known_time;
-      _last_gc_worker_other_times_ms.set(i, worker_other_time);
-    }
+    double worker_other_time = worker_time - worker_known_time;
+    _last_gc_worker_other_times_ms.set(i, worker_other_time);
+  }
 
   _last_gc_worker_times_ms.verify();
   _last_gc_worker_other_times_ms.verify();
--- a/src/share/vm/gc_implementation/g1/g1GCPhaseTimes.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/gc_implementation/g1/g1GCPhaseTimes.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -35,6 +35,8 @@
   const char* _print_format;
   bool        _print_sum;
 
+  NOT_PRODUCT(static const T _uninitialized;)
+
   // We are caching the sum and average to only have to calculate them once.
   // This is not done in an MT-safe way. It is intended to allow single
   // threaded code to call sum() and average() multiple times in any order
--- a/src/share/vm/interpreter/linkResolver.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/interpreter/linkResolver.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -227,6 +227,7 @@
                                              TRAPS) {
   vmIntrinsics::ID iid = MethodHandles::signature_polymorphic_name_id(name);
   if (TraceMethodHandles) {
+    ResourceMark rm(THREAD);
     tty->print_cr("lookup_polymorphic_method iid=%s %s.%s%s",
                   vmIntrinsics::name_at(iid), klass->external_name(),
                   name->as_C_string(), full_signature->as_C_string());
@@ -241,6 +242,7 @@
       TempNewSymbol basic_signature =
         MethodHandles::lookup_basic_type_signature(full_signature, keep_last_arg, CHECK);
       if (TraceMethodHandles) {
+        ResourceMark rm(THREAD);
         tty->print_cr("lookup_polymorphic_method %s %s => basic %s",
                       name->as_C_string(),
                       full_signature->as_C_string(),
@@ -293,6 +295,8 @@
       }
       if (result.not_null()) {
 #ifdef ASSERT
+        ResourceMark rm(THREAD);
+
         TempNewSymbol basic_signature =
           MethodHandles::lookup_basic_type_signature(full_signature, CHECK);
         int actual_size_of_params = result->size_of_parameters();
@@ -1199,8 +1203,10 @@
   Symbol* method_signature = NULL;
   KlassHandle  current_klass;
   resolve_pool(resolved_klass, method_name,  method_signature, current_klass, pool, index, CHECK);
-  if (TraceMethodHandles)
+  if (TraceMethodHandles) {
+    ResourceMark rm(THREAD);
     tty->print_cr("resolve_invokehandle %s %s", method_name->as_C_string(), method_signature->as_C_string());
+  }
   resolve_handle_call(result, resolved_klass, method_name, method_signature, current_klass, CHECK);
 }
 
--- a/src/share/vm/oops/klass.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/oops/klass.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -159,7 +159,7 @@
   kl->set_next_sibling(NULL);
   kl->set_alloc_count(0);
   kl->set_alloc_size(0);
-  TRACE_SET_KLASS_TRACE_ID(kl, 0);
+  TRACE_INIT_ID(kl);
 
   kl->set_prototype_header(markOopDesc::prototype());
   kl->set_biased_lock_revocation_count(0);
--- a/src/share/vm/opto/cfgnode.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/opto/cfgnode.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -1306,10 +1306,11 @@
     return NULL;
 
   Node *x = n2;
-  Node *y = n1->in(1);
-  if( n2 == n1->in(1) ) {
+  Node *y = NULL;
+  if( x == n1->in(1) ) {
     y = n1->in(2);
-  } else if( n2 == n1->in(1) ) {
+  } else if( x == n1->in(2) ) {
+    y = n1->in(1);
   } else return NULL;
 
   // Not so profitable if compare and add are constants
--- a/src/share/vm/opto/output.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/opto/output.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -449,6 +449,17 @@
       int max_loop_pad = nb->code_alignment()-relocInfo::addr_unit();
       if (max_loop_pad > 0) {
         assert(is_power_of_2(max_loop_pad+relocInfo::addr_unit()), "");
+        // Adjust last_call_adr and/or last_avoid_back_to_back_adr.
+        // If either is the last instruction in this block, bump by
+        // max_loop_pad in lock-step with blk_size, so sizing
+        // calculations in subsequent blocks still can conservatively
+        // detect that it may the last instruction in this block.
+        if (last_call_adr == blk_starts[i]+blk_size) {
+          last_call_adr += max_loop_pad;
+        }
+        if (last_avoid_back_to_back_adr == blk_starts[i]+blk_size) {
+          last_avoid_back_to_back_adr += max_loop_pad;
+        }
         blk_size += max_loop_pad;
       }
     }
@@ -1194,8 +1205,6 @@
   int last_call_offset = -1;
   int last_avoid_back_to_back_offset = -1;
 #ifdef ASSERT
-  int block_alignment_padding = 0;
-
   uint* jmp_target = NEW_RESOURCE_ARRAY(uint,nblocks);
   uint* jmp_offset = NEW_RESOURCE_ARRAY(uint,nblocks);
   uint* jmp_size   = NEW_RESOURCE_ARRAY(uint,nblocks);
@@ -1229,8 +1238,6 @@
   Node *delay_slot = NULL;
 
   for (uint i=0; i < nblocks; i++) {
-    guarantee(blk_starts[i] >= (uint)cb->insts_size(),"should not increase size");
-
     Block *b = _cfg->_blocks[i];
 
     Node *head = b->head();
@@ -1251,14 +1258,6 @@
     jmp_offset[i] = 0;
     jmp_size[i]   = 0;
     jmp_rule[i]   = 0;
-
-    // Maximum alignment padding for loop block was used
-    // during first round of branches shortening, as result
-    // padding for nodes (sfpt after call) was not added.
-    // Take this into account for block's size change check
-    // and allow increase block's size by the difference
-    // of maximum and actual alignment paddings.
-    int orig_blk_size = blk_starts[i+1] - blk_starts[i] + block_alignment_padding;
 #endif
     int blk_offset = current_offset;
 
@@ -1558,8 +1557,6 @@
       }
 
     } // End for all instructions in block
-    assert((uint)blk_offset <= blk_starts[i], "shouldn't increase distance");
-    blk_starts[i] = blk_offset;
 
     // If the next block is the top of a loop, pad this block out to align
     // the loop top a little. Helps prevent pipe stalls at loop back branches.
@@ -1573,16 +1570,13 @@
         nop->emit(*cb, _regalloc);
         current_offset = cb->insts_size();
       }
-#ifdef ASSERT
-      int max_loop_pad = nb->code_alignment()-relocInfo::addr_unit();
-      block_alignment_padding = (max_loop_pad - padding);
-      assert(block_alignment_padding >= 0, "sanity");
-#endif
     }
     // Verify that the distance for generated before forward
     // short branches is still valid.
-    assert(orig_blk_size >= (current_offset - blk_offset), "shouldn't increase block size");
-
+    guarantee((int)(blk_starts[i+1] - blk_starts[i]) >= (current_offset - blk_offset), "shouldn't increase block size");
+
+    // Save new block start offset
+    blk_starts[i] = blk_offset;
   } // End of for all blocks
   blk_starts[nblocks] = current_offset;
 
--- a/src/share/vm/prims/whitebox.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/prims/whitebox.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -79,45 +79,46 @@
 WB_END
 #endif // !SERIALGC
 
-// Keep track of the 3 allocations in NMTAllocTest so we can free them later
-// on and verify that they're not visible anymore
-static void* nmtMtTest1 = NULL, *nmtMtTest2 = NULL, *nmtMtTest3 = NULL;
-
 // Alloc memory using the test memory type so that we can use that to see if
 // NMT picks it up correctly
-WB_ENTRY(jboolean, WB_NMTAllocTest(JNIEnv* env))
-  void *mem;
+WB_ENTRY(jlong, WB_NMTMalloc(JNIEnv* env, jobject o, jlong size))
+  jlong addr = 0;
 
-  if (!MemTracker::is_on() || MemTracker::shutdown_in_progress()) {
-    return false;
+  if (MemTracker::is_on() && !MemTracker::shutdown_in_progress()) {
+    addr = (jlong)(uintptr_t)os::malloc(size, mtTest);
   }
 
-  // Allocate 2 * 128k + 256k + 1024k and free the 1024k one to make sure we track
-  // everything correctly. Total should be 512k held alive.
-  nmtMtTest1 = os::malloc(128 * 1024, mtTest);
-  mem = os::malloc(1024 * 1024, mtTest);
-  nmtMtTest2 = os::malloc(256 * 1024, mtTest);
-  os::free(mem, mtTest);
-  nmtMtTest3 = os::malloc(128 * 1024, mtTest);
-
-  return true;
+  return addr;
 WB_END
 
 // Free the memory allocated by NMTAllocTest
-WB_ENTRY(jboolean, WB_NMTFreeTestMemory(JNIEnv* env))
+WB_ENTRY(void, WB_NMTFree(JNIEnv* env, jobject o, jlong mem))
+  os::free((void*)(uintptr_t)mem, mtTest);
+WB_END
 
-  if (nmtMtTest1 == NULL || nmtMtTest2 == NULL || nmtMtTest3 == NULL) {
-    return false;
+WB_ENTRY(jlong, WB_NMTReserveMemory(JNIEnv* env, jobject o, jlong size))
+  jlong addr = 0;
+
+  if (MemTracker::is_on() && !MemTracker::shutdown_in_progress()) {
+    addr = (jlong)(uintptr_t)os::reserve_memory(size);
+    MemTracker::record_virtual_memory_type((address)addr, mtTest);
   }
 
-  os::free(nmtMtTest1, mtTest);
-  nmtMtTest1 = NULL;
-  os::free(nmtMtTest2, mtTest);
-  nmtMtTest2 = NULL;
-  os::free(nmtMtTest3, mtTest);
-  nmtMtTest3 = NULL;
+  return addr;
+WB_END
 
-  return true;
+
+WB_ENTRY(void, WB_NMTCommitMemory(JNIEnv* env, jobject o, jlong addr, jlong size))
+  os::commit_memory((char *)(uintptr_t)addr, size);
+  MemTracker::record_virtual_memory_type((address)(uintptr_t)addr, mtTest);
+WB_END
+
+WB_ENTRY(void, WB_NMTUncommitMemory(JNIEnv* env, jobject o, jlong addr, jlong size))
+  os::uncommit_memory((char *)(uintptr_t)addr, size);
+WB_END
+
+WB_ENTRY(void, WB_NMTReleaseMemory(JNIEnv* env, jobject o, jlong addr, jlong size))
+  os::release_memory((char *)(uintptr_t)addr, size);
 WB_END
 
 // Block until the current generation of NMT data to be merged, used to reliably test the NMT feature
@@ -196,9 +197,13 @@
   {CC"g1NumFreeRegions",   CC"()J",                   (void*)&WB_G1NumFreeRegions  },
   {CC"g1RegionSize",       CC"()I",                   (void*)&WB_G1RegionSize      },
 #endif // !SERIALGC
-  {CC"NMTAllocTest",       CC"()Z",                   (void*)&WB_NMTAllocTest      },
-  {CC"NMTFreeTestMemory",  CC"()Z",                   (void*)&WB_NMTFreeTestMemory },
-  {CC"NMTWaitForDataMerge",CC"()Z",                   (void*)&WB_NMTWaitForDataMerge},
+  {CC"NMTMalloc",           CC"(J)J",                 (void*)&WB_NMTMalloc          },
+  {CC"NMTFree",             CC"(J)V",                 (void*)&WB_NMTFree            },
+  {CC"NMTReserveMemory",    CC"(J)J",                 (void*)&WB_NMTReserveMemory   },
+  {CC"NMTCommitMemory",     CC"(JJ)V",                (void*)&WB_NMTCommitMemory    },
+  {CC"NMTUncommitMemory",   CC"(JJ)V",                (void*)&WB_NMTUncommitMemory  },
+  {CC"NMTReleaseMemory",    CC"(JJ)V",                (void*)&WB_NMTReleaseMemory   },
+  {CC"NMTWaitForDataMerge", CC"()Z",                  (void*)&WB_NMTWaitForDataMerge},
 };
 
 #undef CC
--- a/src/share/vm/runtime/globals.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/runtime/globals.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -3616,7 +3616,7 @@
   diagnostic(bool, PrintDTraceDOF, false,                                   \
              "Print the DTrace DOF passed to the system for JSDT probes")   \
                                                                             \
-  product(uintx, StringTableSize, 1009,                                     \
+  product(uintx, StringTableSize, NOT_LP64(1009) LP64_ONLY(60013),          \
           "Number of buckets in the interned String table")                 \
                                                                             \
   product(bool, UseVMInterruptibleIO, false,                                \
--- a/src/share/vm/services/runtimeService.cpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/services/runtimeService.cpp	Thu Apr 25 08:49:05 2013 -0700
@@ -115,6 +115,8 @@
 
   // Print the time interval in which the app was executing
   if (PrintGCApplicationConcurrentTime) {
+    gclog_or_tty->date_stamp(PrintGCDateStamps);
+    gclog_or_tty->stamp(PrintGCTimeStamps);
     gclog_or_tty->print_cr("Application time: %3.7f seconds",
                                 last_application_time_sec());
   }
@@ -145,6 +147,8 @@
   // Print the time interval for which the app was stopped
   // during the current safepoint operation.
   if (PrintGCApplicationStoppedTime) {
+    gclog_or_tty->date_stamp(PrintGCDateStamps);
+    gclog_or_tty->stamp(PrintGCTimeStamps);
     gclog_or_tty->print_cr("Total time for which application threads "
                            "were stopped: %3.7f seconds",
                            last_safepoint_time_sec());
--- a/src/share/vm/trace/trace.xml	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/trace/trace.xml	Thu Apr 25 08:49:05 2013 -0700
@@ -138,21 +138,21 @@
     </struct>
 
     <event id="GCHeapSummary" path="vm/gc/heap/summary" label="Heap Summary" is_instant="true">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="GCWHEN" field="when" label="When" />
       <structvalue type="VirtualSpace" field="heapSpace" label="Heap Space"/>
       <value type="BYTES64" field="heapUsed" label="Heap Used" description="Bytes allocated by objects in the heap"/>
     </event>
 
     <event id="PermGenSummary" path="vm/gc/heap/perm_gen_summary" label="Perm Gen Summary" is_instant="true">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="GCWHEN" field="when" label="When" />
       <structvalue type="VirtualSpace" field="permSpace" label="PermGen Space"/>
       <structvalue type="ObjectSpace" field="objectSpace" label="Object Space"/>
     </event>
 
     <event id="PSHeapSummary" path="vm/gc/heap/ps_summary" label="ParallelScavengeHeap Summary" is_instant="true">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="GCWHEN" field="when" label="When" />
 
       <structvalue type="VirtualSpace" field="oldSpace" label="Old Space"/>
@@ -166,7 +166,7 @@
 
     <event id="GCGarbageCollection" path="vm/gc/collector/garbage_collection" label="Garbage Collection"
            description="Garbage collection performed by the JVM">
-      <value type="ULONG" field="gcId"  label="GC ID" relation="GC_ID" />
+      <value type="UINT" field="gcId"  label="GC ID" relation="GC_ID" />
       <value type="GCNAME" field="name" label="Name" description="The name of the Garbage Collector" />
       <value type="GCCAUSE" field="cause" label="Cause" description="The reason for triggering this Garbage Collection" />
       <value type="RELATIVE_TICKS" field="sumOfPauses" label="Sum of Pauses" description="Sum of all the times in which Java execution was paused during the garbage collection" />
@@ -175,29 +175,29 @@
 
     <event id="GCParallelOld" path="vm/gc/collector/parold_garbage_collection" label="Parallel Old Garbage Collection"
            description="Extra information specific to Parallel Old Garbage Collections">
-      <value type="ULONG" field="gcId"  label="GC ID" relation="GC_ID" />
+      <value type="UINT" field="gcId"  label="GC ID" relation="GC_ID" />
       <value type="ADDRESS" field="densePrefix" label="Dense Prefix" description="The address of the dense prefix, used when compacting" />
     </event>
 
     <event id="GCYoungGarbageCollection" path="vm/gc/collector/young_garbage_collection" label="Young Garbage Collection"
            description="Extra information specific to Young Garbage Collections">
-      <value type="ULONG" field="gcId"  label="GC ID" relation="GC_ID" />
+      <value type="UINT" field="gcId"  label="GC ID" relation="GC_ID" />
       <value type="UINT" field="tenuringThreshold" label="Tenuring Threshold" />
     </event>
 
     <event id="GCOldGarbageCollection" path="vm/gc/collector/old_garbage_collection" label="Old Garbage Collection"
            description="Extra information specific to Old Garbage Collections">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
     </event>
 
     <event id="GCG1GarbageCollection" path="vm/gc/collector/g1_garbage_collection" label="G1 Garbage Collection"
            description="Extra information specific to G1 Garbage Collections">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="G1YCTYPE" field="type" label="Type" />
     </event>
 
     <event id="EvacuationInfo" path="vm/gc/detailed/evacuation_info" label="Evacuation Info" is_instant="true">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="UINT" field="cSetRegions" label="Collection Set Regions"/>
       <value type="BYTES64" field="cSetUsedBefore" label="Collection Set Before" description="Memory usage before GC in the collection set regions"/>
       <value type="BYTES64" field="cSetUsedAfter" label="Collection Set After" description="Memory usage after GC in the collection set regions"/>
@@ -211,11 +211,18 @@
     <event id="GCReferenceStatistics" path="vm/gc/reference/statistics"
            label="GC Reference Processing" is_instant="true"
            description="Total count of processed references during GC">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="REFERENCETYPE" field="type" label="Type" />
       <value type="ULONG" field="count" label="Total Count" />
     </event>
 
+    <event id="ObjectCountAfterGC" path="vm/gc/detailed/object_count_after_gc" is_instant="true" label="Object Count After GC">
+      <value type="UINT" field="gcId"  label="GC ID" relation="GC_ID" />
+      <value type="CLASS" field="class" label="Class" />
+      <value type="LONG" field="count" label="Count" />
+      <value type="BYTES64" field="totalSize" label="Total Size" />
+    </event>
+
     <struct id="CopyFailed">
       <value type="ULONG" field="objectCount" label="Object Count"/>
       <value type="BYTES64" field="firstSize" label="First Failed Object Size"/>
@@ -223,48 +230,41 @@
       <value type="BYTES64" field="totalSize" label="Total Object Size"/>
     </struct>
 
-    <event id="ObjectCountAfterGC" path="vm/gc/detailed/object_count_after_gc" is_instant="true" label="Object Count After GC">
-      <value type="ULONG" field="gcId"  label="GC ID" relation="GC_ID" />
-      <value type="CLASS" field="class" label="Class" />
-      <value type="LONG" field="count" label="Count" />
-      <value type="BYTES64" field="totalSize" label="Total Size" />
-    </event>
-
     <event id="PromotionFailed" path="vm/gc/detailed/promotion_failed" label="Promotion Failed" is_instant="true"
            description="Promotion of an object failed">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <structvalue type="CopyFailed" field="data" label="data"/>
       <value type="OSTHREAD" field="thread" label="Running thread"/>
     </event>
 
     <event id="EvacuationFailed" path="vm/gc/detailed/evacuation_failed" label="Evacuation Failed" is_instant="true"
            description="Evacuation of an object failed">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <structvalue type="CopyFailed" field="data" label="data"/>
     </event>
 
     <event id="ConcurrentModeFailure" path="vm/gc/detailed/concurrent_mode_failure" label="Concurrent Mode Failure"
            is_instant="true" description="Concurrent Mode failed">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
     </event>
 
     <event id="GCPhasePause" path="vm/gc/phases/pause" label="GC Phase Pause">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="UTF8" field="name" label="Name" />
     </event>
     
     <event id="GCPhasePauseLevel1" path="vm/gc/phases/pause_level_1" label="GC Phase Pause Level 1">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="UTF8" field="name" label="Name" />
     </event>
     
     <event id="GCPhasePauseLevel2" path="vm/gc/phases/pause_level_2" label="GC Phase Pause Level 2">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="UTF8" field="name" label="Name" />
     </event>
     
     <event id="GCPhasePauseLevel3" path="vm/gc/phases/pause_level_3" label="GC Phase Pause Level 3">
-      <value type="ULONG" field="gcId" label="GC ID" relation="GC_ID"/>
+      <value type="UINT" field="gcId" label="GC ID" relation="GC_ID"/>
       <value type="UTF8" field="name" label="Name" />
     </event>
     
@@ -273,7 +273,7 @@
     <event id="Compilation" path="vm/compiler/compilation" label="Compilation"
          has_thread="true" is_requestable="false" is_constant="false">
       <value type="METHOD" field="method" label="Java Method"/>
-      <value type="INTEGER" field="compileID" label="Compilation ID" relation="COMP_ID"/>
+      <value type="UINT" field="compileID" label="Compilation ID" relation="COMP_ID"/>
       <value type="USHORT" field="compileLevel" label="Compilation Level"/>
       <value type="BOOLEAN" field="succeded" label="Succeeded"/>
       <value type="BOOLEAN" field="isOsr" label="On Stack Replacement"/>
@@ -284,14 +284,14 @@
     <event id="CompilerPhase" path="vm/compiler/phase" label="Compiler Phase"
             has_thread="true" is_requestable="false" is_constant="false">
       <value type="COMPILERPHASETYPE" field="phase" label="Compile Phase"/>
-      <value type="INTEGER" field="compileID" label="Compilation ID" relation="COMP_ID"/>
+      <value type="UINT" field="compileID" label="Compilation ID" relation="COMP_ID"/>
       <value type="USHORT" field="phaseLevel" label="Phase Level"/>
     </event>
     
     <event id="CompilerFailure" path="vm/compiler/failure" label="Compilation Failure"
             has_thread="true" is_requestable="false" is_constant="false" is_instant="true">
       <value type="UTF8" field="failure" label="Message"/>
-      <value type="INTEGER" field="compileID" label="Compilation ID" relation="COMP_ID"/>
+      <value type="UINT" field="compileID" label="Compilation ID" relation="COMP_ID"/>
     </event>
 
     <!-- Code sweeper events -->
--- a/src/share/vm/trace/traceMacros.hpp	Wed Apr 24 12:20:45 2013 -0700
+++ b/src/share/vm/trace/traceMacros.hpp	Thu Apr 25 08:49:05 2013 -0700
@@ -33,7 +33,6 @@
 #define TRACE_START() JNI_OK
 #define TRACE_INITIALIZE() JNI_OK
 
-#define TRACE_SET_KLASS_TRACE_ID(x1, x2) do { } while (0)
 #define TRACE_DEFINE_KLASS_METHODS typedef int ___IGNORED_hs_trace_type1
 #define TRACE_DEFINE_KLASS_TRACE_ID typedef int ___IGNORED_hs_trace_type2
 #define TRACE_DEFINE_OFFSET typedef int ___IGNORED_hs_trace_type3
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compiler/6443505/Test6443505.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,107 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+/**
+ * @test
+ * @bug 6443505
+ * @summary Some cases for CmpLTMask missed; also wrong code.
+ *
+ * @run main/othervm -Xcomp -XX:CompileOnly="Test6443505.compiled" Test6443505
+ */
+
+public class Test6443505 {
+
+    public static void main(String[] args) throws InterruptedException {
+        test(Integer.MIN_VALUE, 0);
+        test(0, Integer.MIN_VALUE);
+        test(Integer.MIN_VALUE, -1);
+        test(-1, Integer.MIN_VALUE);
+        test(Integer.MIN_VALUE, 1);
+        test(1, Integer.MIN_VALUE);
+
+        test(Integer.MAX_VALUE, 0);
+        test(0, Integer.MAX_VALUE);
+        test(Integer.MAX_VALUE, -1);
+        test(-1, Integer.MAX_VALUE);
+        test(Integer.MAX_VALUE, 1);
+        test(1, Integer.MAX_VALUE);
+
+        test(Integer.MIN_VALUE, Integer.MAX_VALUE);
+        test(Integer.MAX_VALUE, Integer.MIN_VALUE);
+
+        test(1, -1);
+        test(1, 0);
+        test(1, 1);
+        test(-1, -1);
+        test(-1, 0);
+        test(-1, 1);
+        test(0, -1);
+        test(0, 0);
+        test(0, 1);
+    }
+
+    public static void test(int a, int b) throws InterruptedException {
+        int C = compiled(4, a, b);
+        int I = interpreted(4, a, b);
+        if (C != I) {
+            System.err.println("#1 C = " + C + ", I = " + I);
+            System.err.println("#1 C != I, FAIL");
+            System.exit(97);
+        }
+
+        C = compiled(a, b, q, 4);
+        I = interpreted(a, b, q, 4);
+        if (C != I) {
+            System.err.println("#2 C = " + C + ", I = " + I);
+            System.err.println("#2 C != I, FAIL");
+            System.exit(97);
+        }
+
+    }
+
+    static int q = 4;
+
+    // If improperly compiled, uses carry/borrow bit, which is wrong.
+    // with -XX:+PrintOptoAssembly, look for cadd_cmpLTMask
+    static int compiled(int p, int x, int y) {
+        return (x < y) ? q + (x - y) : (x - y);
+    }
+
+    // interpreted reference
+    static int interpreted(int p, int x, int y) {
+        return (x < y) ? q + (x - y) : (x - y);
+    }
+
+    // Test new code with a range of cases
+    // with -XX:+PrintOptoAssembly, look for and_cmpLTMask
+    static int compiled(int x, int y, int q, int p) {
+        return (x < y) ? p + q : q;
+    }
+
+    // interpreted reference
+    static int interpreted(int x, int y, int q, int p) {
+        return (x < y) ? p + q : q;
+    }
+
+}
--- a/test/compiler/6863420/Test.java	Wed Apr 24 12:20:45 2013 -0700
+++ b/test/compiler/6863420/Test.java	Thu Apr 25 08:49:05 2013 -0700
@@ -27,17 +27,35 @@
  * @bug 6863420
  * @summary os::javaTimeNanos() go backward on Solaris x86
  *
- * @run main/othervm Test
+ * Notice the internal timeout in timeout thread Test.TOT.
+ * @run main/othervm/timeout=300 Test
  */
 
 public class Test {
+
+    static final int INTERNAL_TIMEOUT=240;
+    static class TOT extends Thread {
+       public void run() {
+           try {
+               Thread.sleep(INTERNAL_TIMEOUT*1000);
+           } catch (InterruptedException ex) {
+           }
+           done = true;
+       }
+    }
+
     static long value = 0;
     static boolean got_backward_time = false;
+    static volatile boolean done = false;
 
     public static void main(String args[]) {
         final int count = 100000;
 
-        for (int numThreads = 1; numThreads <= 32; numThreads++) {
+        TOT tot = new TOT();
+        tot.setDaemon(true);
+        tot.start();
+
+        for (int numThreads = 1; !done && numThreads <= 32; numThreads++) {
             final int numRuns = 1;
             for (int t=1; t <= numRuns; t++) {
                 final int curRun = t;
@@ -48,7 +66,7 @@
                     Runnable thread =
                         new Runnable() {
                             public void run() {
-                                for (long l = 0; l < 100000; l++) {
+                                for (long l = 0; !done && l < 100000; l++) {
                                     final long start = System.nanoTime();
                                     if (value == 12345678) {
                                         System.out.println("Wow!");
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compiler/8011901/Test8011901.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test
+ * @bug 8011901
+ * @summary instruct xaddL_no_res shouldn't allow 64 bit constants.
+ * @run main/othervm -XX:-BackgroundCompilation Test8011901
+ *
+ */
+
+import java.lang.reflect.*;
+import sun.misc.*;
+
+public class Test8011901 {
+
+    private long ctl;
+
+    private static final sun.misc.Unsafe U;
+    private static final long CTL;
+
+    static {
+        try {
+            Field unsafe = sun.misc.Unsafe.class.getDeclaredField("theUnsafe");
+            unsafe.setAccessible(true);
+            U = (sun.misc.Unsafe) unsafe.get(null);
+            CTL = U.objectFieldOffset(Test8011901.class.getDeclaredField("ctl"));
+        } catch (Exception e) {
+            throw new Error(e);
+        }
+    }
+
+    public static void main(String[] args) {
+        for(int c = 0; c < 20000; c++) {
+            new Test8011901().makeTest();
+        }
+        System.out.println("Test Passed");
+    }
+
+    public static final long EXPECTED = 1L << 42;
+
+    public void makeTest() {
+        U.getAndAddLong(this, CTL, EXPECTED);
+        if (ctl != EXPECTED) {
+            throw new RuntimeException("Test failed. Expected: " + EXPECTED + ", but got = " + ctl);
+        }
+    }
+}
--- a/test/runtime/NMT/AllocTestType.java	Wed Apr 24 12:20:45 2013 -0700
+++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
@@ -1,73 +0,0 @@
-/*
- * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This code is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- */
-
-/*
- * @test
- * @summary Test consistency of NMT by leaking a few select allocations of the Test type and then verify visibility with jcmd
- * @key nmt jcmd
- * @library /testlibrary /testlibrary/whitebox
- * @build AllocTestType
- * @run main ClassFileInstaller sun.hotspot.WhiteBox
- * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:NativeMemoryTracking=detail AllocTestType
- */
-
-import com.oracle.java.testlibrary.*;
-import sun.hotspot.WhiteBox;
-
-public class AllocTestType {
-
-  public static void main(String args[]) throws Exception {
-    OutputAnalyzer output;
-
-    // Grab my own PID
-    String pid = Integer.toString(ProcessTools.getProcessId());
-    ProcessBuilder pb = new ProcessBuilder();
-
-    // Use WB API to alloc with the mtTest type
-    if (!WhiteBox.getWhiteBox().NMTAllocTest()) {
-      throw new Exception("Call to WB API NMTAllocTest() failed");
-    }
-
-    // Use WB API to ensure that all data has been merged before we continue
-    if (!WhiteBox.getWhiteBox().NMTWaitForDataMerge()) {
-      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
-    }
-
-    // Run 'jcmd <pid> VM.native_memory summary'
-    pb.command(new String[] { JDKToolFinder.getJDKTool("jcmd"), pid, "VM.native_memory", "summary"});
-    output = new OutputAnalyzer(pb.start());
-    output.shouldContain("Test (reserved=512KB, committed=512KB)");
-
-    // Free the memory allocated by NMTAllocTest
-    if (!WhiteBox.getWhiteBox().NMTFreeTestMemory()) {
-      throw new Exception("Call to WB API NMTFreeTestMemory() failed");
-    }
-
-    // Use WB API to ensure that all data has been merged before we continue
-    if (!WhiteBox.getWhiteBox().NMTWaitForDataMerge()) {
-      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
-    }
-    output = new OutputAnalyzer(pb.start());
-    output.shouldNotContain("Test (reserved=");
-  }
-}
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/runtime/NMT/MallocTestType.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test
+ * @summary Test consistency of NMT by leaking a few select allocations of the Test type and then verify visibility with jcmd
+ * @key nmt jcmd
+ * @library /testlibrary /testlibrary/whitebox
+ * @build MallocTestType
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:NativeMemoryTracking=detail MallocTestType
+ */
+
+import com.oracle.java.testlibrary.*;
+import sun.hotspot.WhiteBox;
+
+public class MallocTestType {
+
+  public static void main(String args[]) throws Exception {
+    OutputAnalyzer output;
+    WhiteBox wb = WhiteBox.getWhiteBox();
+
+    // Grab my own PID
+    String pid = Integer.toString(ProcessTools.getProcessId());
+    ProcessBuilder pb = new ProcessBuilder();
+
+    // Use WB API to alloc and free with the mtTest type
+    long memAlloc3 = wb.NMTMalloc(128 * 1024);
+    long memAlloc2 = wb.NMTMalloc(256 * 1024);
+    wb.NMTFree(memAlloc3);
+    long memAlloc1 = wb.NMTMalloc(512 * 1024);
+    wb.NMTFree(memAlloc2);
+
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+
+    // Run 'jcmd <pid> VM.native_memory summary'
+    pb.command(new String[] { JDKToolFinder.getJDKTool("jcmd"), pid, "VM.native_memory", "summary"});
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=512KB, committed=512KB)");
+
+    // Free the memory allocated by NMTAllocTest
+    wb.NMTFree(memAlloc1);
+
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+    output = new OutputAnalyzer(pb.start());
+    output.shouldNotContain("Test (reserved=");
+  }
+}
--- a/test/runtime/NMT/PrintNMTStatistics.java	Wed Apr 24 12:20:45 2013 -0700
+++ b/test/runtime/NMT/PrintNMTStatistics.java	Thu Apr 25 08:49:05 2013 -0700
@@ -27,7 +27,9 @@
  * @bug 8005936
  * @summary Make sure PrintNMTStatistics works on normal JVM exit
  * @library /testlibrary /testlibrary/whitebox
- * @run compile PrintNMTStatistics.java
+ * @build PrintNMTStatistics
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ * @run main PrintNMTStatistics
  */
 
 import com.oracle.java.testlibrary.*;
@@ -52,13 +54,15 @@
 
     ProcessBuilder pb = ProcessTools.createJavaProcessBuilder(
         "-XX:+UnlockDiagnosticVMOptions",
+        "-Xbootclasspath/a:.",
+        "-XX:+WhiteBoxAPI",
         "-XX:NativeMemoryTracking=summary",
-        "+XX:+PrintNMTStatistics",
+        "-XX:+PrintNMTStatistics",
         "PrintNMTStatistics",
         "test");
 
     OutputAnalyzer output = new OutputAnalyzer(pb.start());
-    output.shouldContain("Java Heap  (reserved=");
+    output.shouldContain("Java Heap (reserved=");
     output.shouldNotContain("error");
     output.shouldNotContain("warning");
     output.shouldHaveExitValue(0);
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/runtime/NMT/ThreadedMallocTestType.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,91 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test
+ * @key nmt jcmd
+ * @library /testlibrary /testlibrary/whitebox
+ * @build ThreadedMallocTestType
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:NativeMemoryTracking=detail ThreadedMallocTestType
+ */
+
+import com.oracle.java.testlibrary.*;
+import sun.hotspot.WhiteBox;
+
+public class ThreadedMallocTestType {
+  public static long memAlloc1;
+  public static long memAlloc2;
+  public static long memAlloc3;
+
+  public static void main(String args[]) throws Exception {
+    OutputAnalyzer output;
+    final WhiteBox wb = WhiteBox.getWhiteBox();
+
+    // Grab my own PID
+    String pid = Integer.toString(ProcessTools.getProcessId());
+    ProcessBuilder pb = new ProcessBuilder();
+
+    Thread allocThread = new Thread() {
+      public void run() {
+        // Alloc memory using the WB api
+        memAlloc1 = wb.NMTMalloc(128 * 1024);
+        memAlloc2 = wb.NMTMalloc(256 * 1024);
+        memAlloc3 = wb.NMTMalloc(512 * 1024);
+      }
+    };
+
+    allocThread.start();
+    allocThread.join();
+
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+
+    // Run 'jcmd <pid> VM.native_memory summary'
+    pb.command(new String[] { JDKToolFinder.getJDKTool("jcmd"), pid, "VM.native_memory", "summary"});
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=896KB, committed=896KB)");
+
+    Thread freeThread = new Thread() {
+      public void run() {
+        // Free the memory allocated by NMTMalloc
+        wb.NMTFree(memAlloc1);
+        wb.NMTFree(memAlloc2);
+        wb.NMTFree(memAlloc3);
+      }
+    };
+
+    freeThread.start();
+    freeThread.join();
+
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldNotContain("Test (reserved=");
+  }
+}
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/runtime/NMT/ThreadedVirtualAllocTestType.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,112 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test
+ * @key nmt jcmd
+ * @library /testlibrary /testlibrary/whitebox
+ * @build ThreadedVirtualAllocTestType
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:NativeMemoryTracking=detail ThreadedVirtualAllocTestType
+ */
+
+import com.oracle.java.testlibrary.*;
+import sun.hotspot.WhiteBox;
+
+public class ThreadedVirtualAllocTestType {
+  public static long addr;
+  public static final WhiteBox wb = WhiteBox.getWhiteBox();
+  public static final long commitSize = 128 * 1024;
+  public static final long reserveSize = 512 * 1024;
+
+  public static void main(String args[]) throws Exception {
+    OutputAnalyzer output;
+
+    String pid = Integer.toString(ProcessTools.getProcessId());
+    ProcessBuilder pb = new ProcessBuilder();
+
+    Thread reserveThread = new Thread() {
+      public void run() {
+        addr = wb.NMTReserveMemory(reserveSize);
+      }
+    };
+    reserveThread.start();
+    reserveThread.join();
+
+    mergeData();
+
+    pb.command(new String[] { JDKToolFinder.getJDKTool("jcmd"), pid, "VM.native_memory", "detail"});
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=512KB, committed=0KB)");
+    output.shouldMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + reserveSize) + "\\] reserved 512KB for Test");
+
+    Thread commitThread = new Thread() {
+      public void run() {
+        wb.NMTCommitMemory(addr, commitSize);
+      }
+    };
+    commitThread.start();
+    commitThread.join();
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=512KB, committed=128KB)");
+    output.shouldMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + commitSize) + "\\] committed 128KB");
+
+    Thread uncommitThread = new Thread() {
+      public void run() {
+        wb.NMTUncommitMemory(addr, commitSize);
+      }
+    };
+    uncommitThread.start();
+    uncommitThread.join();
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=512KB, committed=0KB)");
+    output.shouldNotMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + commitSize) + "\\] committed");
+
+    Thread releaseThread = new Thread() {
+      public void run() {
+        wb.NMTReleaseMemory(addr, reserveSize);
+      }
+    };
+    releaseThread.start();
+    releaseThread.join();
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldNotContain("Test (reserved=");
+    output.shouldNotContain("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + reserveSize) + "\\] reserved");
+  }
+
+  public static void mergeData() throws Exception {
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+  }
+}
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/runtime/NMT/VirtualAllocTestType.java	Thu Apr 25 08:49:05 2013 -0700
@@ -0,0 +1,88 @@
+/*
+ * Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test
+ * @summary Test Reserve/Commit/Uncommit/Release of virtual memory and that we track it correctly
+ * @key nmt jcmd
+ * @library /testlibrary /testlibrary/whitebox
+ * @build VirtualAllocTestType
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:NativeMemoryTracking=detail VirtualAllocTestType
+ */
+
+import com.oracle.java.testlibrary.*;
+import sun.hotspot.WhiteBox;
+
+public class VirtualAllocTestType {
+
+  public static WhiteBox wb = WhiteBox.getWhiteBox();
+  public static void main(String args[]) throws Exception {
+    OutputAnalyzer output;
+    long commitSize = 128 * 1024;
+    long reserveSize = 256 * 1024;
+    long addr;
+
+    String pid = Integer.toString(ProcessTools.getProcessId());
+    ProcessBuilder pb = new ProcessBuilder();
+
+    addr = wb.NMTReserveMemory(reserveSize);
+    mergeData();
+
+    pb.command(new String[] { JDKToolFinder.getJDKTool("jcmd"), pid, "VM.native_memory", "detail"});
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=256KB, committed=0KB)");
+    output.shouldMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + reserveSize) + "\\] reserved 256KB for Test");
+
+    wb.NMTCommitMemory(addr, commitSize);
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=256KB, committed=128KB)");
+    output.shouldMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + commitSize) + "\\] committed 128KB");
+
+    wb.NMTUncommitMemory(addr, commitSize);
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldContain("Test (reserved=256KB, committed=0KB)");
+    output.shouldNotMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + commitSize) + "\\] committed");
+
+    wb.NMTReleaseMemory(addr, reserveSize);
+
+    mergeData();
+
+    output = new OutputAnalyzer(pb.start());
+    output.shouldNotContain("Test (reserved=");
+    output.shouldNotMatch("\\[0x[0]*" + Long.toHexString(addr) + " - 0x[0]*" + Long.toHexString(addr + reserveSize) + "\\] reserved");
+  }
+
+  public static void mergeData() throws Exception {
+    // Use WB API to ensure that all data has been merged before we continue
+    if (!wb.NMTWaitForDataMerge()) {
+      throw new Exception("Call to WB API NMTWaitForDataMerge() failed");
+    }
+  }
+}
--- a/test/testlibrary/OutputAnalyzerTest.java	Wed Apr 24 12:20:45 2013 -0700
+++ b/test/testlibrary/OutputAnalyzerTest.java	Thu Apr 25 08:49:05 2013 -0700
@@ -36,6 +36,11 @@
     String stdout = "aaaaaa";
     String stderr = "bbbbbb";
 
+    // Regexps used for testing pattern matching of the test input
+    String stdoutPattern = "[a]";
+    String stderrPattern = "[b]";
+    String nonExistingPattern = "[c]";
+
     OutputAnalyzer output = new OutputAnalyzer(stdout, stderr);
 
     if (!stdout.equals(output.getStdout())) {
@@ -99,10 +104,73 @@
     }
 
     try {
-      output.stderrShouldNotContain(stderr);
-      throw new Exception("shouldContain() failed to throw exception");
+        output.stderrShouldNotContain(stderr);
+        throw new Exception("shouldContain() failed to throw exception");
     } catch (RuntimeException e) {
-      // expected
+        // expected
+    }
+
+    // Should match
+    try {
+        output.shouldMatch(stdoutPattern);
+        output.stdoutShouldMatch(stdoutPattern);
+        output.shouldMatch(stderrPattern);
+        output.stderrShouldMatch(stderrPattern);
+    } catch (RuntimeException e) {
+        throw new Exception("shouldMatch() failed", e);
+    }
+
+    try {
+        output.shouldMatch(nonExistingPattern);
+        throw new Exception("shouldMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
+    }
+
+    try {
+        output.stdoutShouldMatch(stderrPattern);
+        throw new Exception(
+                "stdoutShouldMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
+    }
+
+    try {
+        output.stderrShouldMatch(stdoutPattern);
+        throw new Exception(
+                "stderrShouldMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
+    }
+
+    // Should not match
+    try {
+        output.shouldNotMatch(nonExistingPattern);
+        output.stdoutShouldNotMatch(nonExistingPattern);
+        output.stderrShouldNotMatch(nonExistingPattern);
+    } catch (RuntimeException e) {
+        throw new Exception("shouldNotMatch() failed", e);
+    }
+
+    try {
+        output.shouldNotMatch(stdoutPattern);
+        throw new Exception("shouldNotMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
+    }
+
+    try {
+        output.stdoutShouldNotMatch(stdoutPattern);
+        throw new Exception("shouldNotMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
+    }
+
+    try {
+        output.stderrShouldNotMatch(stderrPattern);
+        throw new Exception("shouldNotMatch() failed to throw exception");
+    } catch (RuntimeException e) {
+        // expected
     }
   }
 }
--- a/test/testlibrary/com/oracle/java/testlibrary/OutputAnalyzer.java	Wed Apr 24 12:20:45 2013 -0700
+++ b/test/testlibrary/com/oracle/java/testlibrary/OutputAnalyzer.java	Thu Apr 25 08:49:05 2013 -0700
@@ -24,6 +24,8 @@
 package com.oracle.java.testlibrary;
 
 import java.io.IOException;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 public final class OutputAnalyzer {
 
@@ -142,15 +144,112 @@
   }
 
   /**
+   * Verify that the stdout and stderr contents of output buffer matches
+   * the pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was not found
+   */
+  public void shouldMatch(String pattern) {
+      Matcher stdoutMatcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stdout);
+      Matcher stderrMatcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stderr);
+      if (!stdoutMatcher.find() && !stderrMatcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' missing from stdout/stderr: [" + stdout + stderr
+                  + "]\n");
+      }
+  }
+
+  /**
+   * Verify that the stdout contents of output buffer matches the
+   * pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was not found
+   */
+  public void stdoutShouldMatch(String pattern) {
+      Matcher matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stdout);
+      if (!matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' missing from stdout: [" + stdout + "]\n");
+      }
+  }
+
+  /**
+   * Verify that the stderr contents of output buffer matches the
+   * pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was not found
+   */
+  public void stderrShouldMatch(String pattern) {
+      Matcher matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stderr);
+      if (!matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' missing from stderr: [" + stderr + "]\n");
+      }
+  }
+
+  /**
+   * Verify that the stdout and stderr contents of output buffer does not
+   * match the pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was found
+   */
+  public void shouldNotMatch(String pattern) {
+      Matcher matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stdout);
+      if (matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' found in stdout: [" + stdout + "]\n");
+      }
+      matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stderr);
+      if (matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' found in stderr: [" + stderr + "]\n");
+      }
+  }
+
+  /**
+   * Verify that the stdout contents of output buffer does not match the
+   * pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was found
+   */
+  public void stdoutShouldNotMatch(String pattern) {
+      Matcher matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stdout);
+      if (matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' found in stdout: [" + stdout + "]\n");
+      }
+  }
+
+  /**
+   * Verify that the stderr contents of output buffer does not match the
+   * pattern
+   *
+   * @param pattern
+   * @throws RuntimeException If the pattern was found
+   */
+  public void stderrShouldNotMatch(String pattern) {
+      Matcher matcher = Pattern.compile(pattern, Pattern.MULTILINE).matcher(stderr);
+      if (matcher.find()) {
+          throw new RuntimeException("'" + pattern
+                  + "' found in stderr: [" + stderr + "]\n");
+      }
+  }
+
+  /**
    * Verifiy the exit value of the process
    *
    * @param expectedExitValue Expected exit value from process
    * @throws RuntimeException If the exit value from the process did not match the expected value
    */
   public void shouldHaveExitValue(int expectedExitValue) {
-    if (getExitValue() != expectedExitValue) {
-      throw new RuntimeException("Exit value " + getExitValue() + " , expected to get " + expectedExitValue);
-    }
+      if (getExitValue() != expectedExitValue) {
+          throw new RuntimeException("Exit value " + getExitValue() + " , expected to get " + expectedExitValue);
+      }
   }
 
   /**
--- a/test/testlibrary/whitebox/sun/hotspot/WhiteBox.java	Wed Apr 24 12:20:45 2013 -0700
+++ b/test/testlibrary/whitebox/sun/hotspot/WhiteBox.java	Thu Apr 25 08:49:05 2013 -0700
@@ -71,7 +71,11 @@
   public native Object[]    parseCommandLine(String commandline, DiagnosticCommand[] args);
 
   // NMT
-  public native boolean NMTAllocTest();
-  public native boolean NMTFreeTestMemory();
+  public native long NMTMalloc(long size);
+  public native void NMTFree(long mem);
+  public native long NMTReserveMemory(long size);
+  public native void NMTCommitMemory(long addr, long size);
+  public native void NMTUncommitMemory(long addr, long size);
+  public native void NMTReleaseMemory(long addr, long size);
   public native boolean NMTWaitForDataMerge();
 }