# HG changeset patch
# User asaha
# Date 1452011281 28800
# Node ID c3091ebd28118acf1e5fdba27aa4006ec1dcb6e5
# Parent  f796867c1bcb37c7ec8f8095c7d1cc38511e075c# Parent  80959a760b85781672c1ceb76bfa860eae46ba34
Merge

diff -r f796867c1bcb -r c3091ebd2811 .hgtags
--- a/.hgtags	Tue Dec 15 22:59:18 2015 -0800
+++ b/.hgtags	Tue Jan 05 08:28:01 2016 -0800
@@ -792,3 +792,5 @@
 b8e7dd0e21173ad829b40361763d27cb6ac532e9 jdk8u72-b12
 a8e4754b89aecc388623394a20f6d43d4c58f083 jdk8u72-b13
 d7b01fb81aa8a5437cb03bc36afe15cf0e55fb89 jdk8u76-b00
+c1679cc87ba045219169cabb6b9b378c2b5cc578 jdk8u76-b01
+218483967e52b419d885d34af4488a81c5133804 jdk8u76-b02
diff -r f796867c1bcb -r c3091ebd2811 agent/src/os/linux/LinuxDebuggerLocal.c
--- a/agent/src/os/linux/LinuxDebuggerLocal.c	Tue Dec 15 22:59:18 2015 -0800
+++ b/agent/src/os/linux/LinuxDebuggerLocal.c	Tue Jan 05 08:28:01 2016 -0800
@@ -209,9 +209,12 @@
   verifyBitness(env, (char *) &buf);
   CHECK_EXCEPTION;
 
+  char err_buf[200];
   struct ps_prochandle* ph;
-  if ( (ph = Pgrab(jpid)) == NULL) {
-    THROW_NEW_DEBUGGER_EXCEPTION("Can't attach to the process");
+  if ( (ph = Pgrab(jpid, err_buf, sizeof(err_buf))) == NULL) {
+    char msg[230];
+    snprintf(msg, sizeof(msg), "Can't attach to the process: %s", err_buf);
+    THROW_NEW_DEBUGGER_EXCEPTION(msg);
   }
   (*env)->SetLongField(env, this_obj, p_ps_prochandle_ID, (jlong)(intptr_t)ph);
   fillThreadsAndLoadObjects(env, this_obj, ph);
diff -r f796867c1bcb -r c3091ebd2811 agent/src/os/linux/libproc.h
--- a/agent/src/os/linux/libproc.h	Tue Dec 15 22:59:18 2015 -0800
+++ b/agent/src/os/linux/libproc.h	Tue Jan 05 08:28:01 2016 -0800
@@ -69,6 +69,7 @@
 
 
 #if defined(sparc) || defined(sparcv9) || defined(ppc64)
+#include <asm/ptrace.h>
 #define user_regs_struct  pt_regs
 #endif
 
@@ -82,7 +83,7 @@
 struct ps_prochandle;
 
 // attach to a process
-struct ps_prochandle* Pgrab(pid_t pid);
+struct ps_prochandle* Pgrab(pid_t pid, char* err_buf, size_t err_buf_len);
 
 // attach to a core dump
 struct ps_prochandle* Pgrab_core(const char* execfile, const char* corefile);
diff -r f796867c1bcb -r c3091ebd2811 agent/src/os/linux/ps_proc.c
--- a/agent/src/os/linux/ps_proc.c	Tue Dec 15 22:59:18 2015 -0800
+++ b/agent/src/os/linux/ps_proc.c	Tue Jan 05 08:28:01 2016 -0800
@@ -215,9 +215,12 @@
 }
 
 // attach to a process/thread specified by "pid"
-static bool ptrace_attach(pid_t pid) {
+static bool ptrace_attach(pid_t pid, char* err_buf, size_t err_buf_len) {
   if (ptrace(PTRACE_ATTACH, pid, NULL, NULL) < 0) {
-    print_debug("ptrace(PTRACE_ATTACH, ..) failed for %d\n", pid);
+    char buf[200];
+    char* msg = strerror_r(errno, buf, sizeof(buf));
+    snprintf(err_buf, err_buf_len, "ptrace(PTRACE_ATTACH, ..) failed for %d: %s", pid, msg);
+    print_debug("%s\n", err_buf);
     return false;
   } else {
     return ptrace_waitpid(pid);
@@ -339,16 +342,17 @@
 };
 
 // attach to the process. One and only one exposed stuff
-struct ps_prochandle* Pgrab(pid_t pid) {
+struct ps_prochandle* Pgrab(pid_t pid, char* err_buf, size_t err_buf_len) {
   struct ps_prochandle* ph = NULL;
   thread_info* thr = NULL;
 
   if ( (ph = (struct ps_prochandle*) calloc(1, sizeof(struct ps_prochandle))) == NULL) {
-     print_debug("can't allocate memory for ps_prochandle\n");
+     snprintf(err_buf, err_buf_len, "can't allocate memory for ps_prochandle");
+     print_debug("%s\n", err_buf);
      return NULL;
   }
 
-  if (ptrace_attach(pid) != true) {
+  if (ptrace_attach(pid, err_buf, err_buf_len) != true) {
      free(ph);
      return NULL;
   }
@@ -371,7 +375,7 @@
   thr = ph->threads;
   while (thr) {
      // don't attach to the main thread again
-     if (ph->pid != thr->lwp_id && ptrace_attach(thr->lwp_id) != true) {
+    if (ph->pid != thr->lwp_id && ptrace_attach(thr->lwp_id, err_buf, err_buf_len) != true) {
         // even if one attach fails, we get return NULL
         Prelease(ph);
         return NULL;
diff -r f796867c1bcb -r c3091ebd2811 make/linux/Makefile
--- a/make/linux/Makefile	Tue Dec 15 22:59:18 2015 -0800
+++ b/make/linux/Makefile	Tue Jan 05 08:28:01 2016 -0800
@@ -67,8 +67,12 @@
   endif
 endif
 # C1 is not ported on ppc64, so we cannot build a tiered VM:
-ifeq ($(ARCH),ppc64)
-  FORCE_TIERED=0
+# Notice: after 8046471 ARCH will be 'ppc' for top-level ppc64 builds but
+# 'ppc64' for HotSpot-only ppc64 builds. Need to detect both variants here!
+ifneq (,$(findstring $(ARCH), ppc ppc64))
+  ifeq ($(ARCH_DATA_MODEL), 64)
+    FORCE_TIERED=0
+  endif
 endif
 
 ifdef LP64
diff -r f796867c1bcb -r c3091ebd2811 make/linux/makefiles/defs.make
--- a/make/linux/makefiles/defs.make	Tue Dec 15 22:59:18 2015 -0800
+++ b/make/linux/makefiles/defs.make	Tue Jan 05 08:28:01 2016 -0800
@@ -69,7 +69,7 @@
 endif
 
 # sparc
-ifeq ($(ARCH), sparc64)
+ifneq (,$(findstring $(ARCH), sparc))
   ifeq ($(ARCH_DATA_MODEL), 64)
     ARCH_DATA_MODEL  = 64
     MAKE_ARGS        += LP64=1
@@ -83,39 +83,35 @@
   HS_ARCH            = sparc
 endif
 
-# amd64/x86_64
-ifneq (,$(findstring $(ARCH), amd64 x86_64))
+# i686/i586 and amd64/x86_64
+ifneq (,$(findstring $(ARCH), amd64 x86_64 i686 i586))
   ifeq ($(ARCH_DATA_MODEL), 64)
     ARCH_DATA_MODEL = 64
     MAKE_ARGS       += LP64=1
     PLATFORM        = linux-amd64
     VM_PLATFORM     = linux_amd64
-    HS_ARCH         = x86
   else
     ARCH_DATA_MODEL = 32
     PLATFORM        = linux-i586
     VM_PLATFORM     = linux_i486
-    HS_ARCH         = x86
-    # We have to reset ARCH to i686 since SRCARCH relies on it
-    ARCH            = i686
   endif
+  HS_ARCH           = x86
 endif
 
-# i686/i586 ie 32-bit x86
-ifneq (,$(findstring $(ARCH), i686 i586))
-  ARCH_DATA_MODEL  = 32
-  PLATFORM         = linux-i586
-  VM_PLATFORM      = linux_i486
-  HS_ARCH          = x86
-endif
-
-# PPC64
-ifeq ($(ARCH), ppc64)
-  ARCH_DATA_MODEL  = 64
-  MAKE_ARGS        += LP64=1
-  PLATFORM         = linux-ppc64
-  VM_PLATFORM      = linux_ppc64
-  HS_ARCH          = ppc
+# PPC
+# Notice: after 8046471 ARCH will be 'ppc' for top-level ppc64 builds but
+# 'ppc64' for HotSpot-only ppc64 builds. Need to detect both variants here!
+ifneq (,$(findstring $(ARCH), ppc ppc64))
+  ifeq ($(ARCH_DATA_MODEL), 64)
+    MAKE_ARGS        += LP64=1
+    PLATFORM         = linux-ppc64
+    VM_PLATFORM      = linux_ppc64
+  else
+    ARCH_DATA_MODEL  = 32
+    PLATFORM         = linux-ppc
+    VM_PLATFORM      = linux_ppc
+  endif
+  HS_ARCH = ppc
 endif
 
 # On 32 bit linux we build server and client, on 64 bit just server.
diff -r f796867c1bcb -r c3091ebd2811 src/cpu/sparc/vm/frame_sparc.cpp
--- a/src/cpu/sparc/vm/frame_sparc.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/cpu/sparc/vm/frame_sparc.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2015, 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
@@ -447,32 +447,6 @@
 }
 #endif // CC_INTERP
 
-
-#ifdef ASSERT
-// Debugging aid
-static frame nth_sender(int n) {
-  frame f = JavaThread::current()->last_frame();
-
-  for(int i = 0; i < n; ++i)
-    f = f.sender((RegisterMap*)NULL);
-
-  printf("first frame %d\n",          f.is_first_frame()       ? 1 : 0);
-  printf("interpreted frame %d\n",    f.is_interpreted_frame() ? 1 : 0);
-  printf("java frame %d\n",           f.is_java_frame()        ? 1 : 0);
-  printf("entry frame %d\n",          f.is_entry_frame()       ? 1 : 0);
-  printf("native frame %d\n",         f.is_native_frame()      ? 1 : 0);
-  if (f.is_compiled_frame()) {
-    if (f.is_deoptimized_frame())
-      printf("deoptimized frame 1\n");
-    else
-      printf("compiled frame 1\n");
-  }
-
-  return f;
-}
-#endif
-
-
 frame frame::sender_for_entry_frame(RegisterMap *map) const {
   assert(map != NULL, "map must be set");
   // Java frame called from C; skip all C frames and return top C
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/c1/c1_ValueType.cpp
--- a/src/share/vm/c1/c1_ValueType.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/c1/c1_ValueType.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -153,7 +153,19 @@
     case T_FLOAT  : return new FloatConstant (value.as_float ());
     case T_DOUBLE : return new DoubleConstant(value.as_double());
     case T_ARRAY  : // fall through (ciConstant doesn't have an array accessor)
-    case T_OBJECT : return new ObjectConstant(value.as_object());
+    case T_OBJECT : {
+      // TODO: Common the code with GraphBuilder::load_constant?
+      ciObject* obj = value.as_object();
+      if (obj->is_null_object())
+        return objectNull;
+      if (obj->is_loaded()) {
+        if (obj->is_array())
+          return new ArrayConstant(obj->as_array());
+        else if (obj->is_instance())
+          return new InstanceConstant(obj->as_instance());
+      }
+      return new ObjectConstant(obj);
+    }
   }
   ShouldNotReachHere();
   return illegalType;
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp
--- a/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -2331,6 +2331,7 @@
     case GCCause::_java_lang_system_gc:     return ExplicitGCInvokesConcurrent;
     case GCCause::_g1_humongous_allocation: return true;
     case GCCause::_update_allocation_context_stats_inc: return true;
+    case GCCause::_wb_conc_mark:            return true;
     default:                                return false;
   }
 }
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/gc_implementation/g1/vm_operations_g1.cpp
--- a/src/share/vm/gc_implementation/g1/vm_operations_g1.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/gc_implementation/g1/vm_operations_g1.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -90,12 +90,8 @@
 
 void VM_G1IncCollectionPause::doit() {
   G1CollectedHeap* g1h = G1CollectedHeap::heap();
-  assert(!_should_initiate_conc_mark ||
-  ((_gc_cause == GCCause::_gc_locker && GCLockerInvokesConcurrent) ||
-   (_gc_cause == GCCause::_java_lang_system_gc && ExplicitGCInvokesConcurrent) ||
-    _gc_cause == GCCause::_g1_humongous_allocation ||
-    _gc_cause == GCCause::_update_allocation_context_stats_inc),
-      "only a GC locker, a System.gc(), stats update or a hum allocation induced GC should start a cycle");
+  assert(!_should_initiate_conc_mark || g1h->should_do_concurrent_full_gc(_gc_cause),
+      "only a GC locker, a System.gc(), stats update, whitebox, or a hum allocation induced GC should start a cycle");
 
   if (_word_size > 0) {
     // An allocation has been requested. So, try to do that first.
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/gc_interface/gcCause.cpp
--- a/src/share/vm/gc_interface/gcCause.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/gc_interface/gcCause.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -54,6 +54,9 @@
     case _wb_young_gc:
       return "WhiteBox Initiated Young GC";
 
+    case _wb_conc_mark:
+      return "WhiteBox Initiated Concurrent Mark";
+
     case _update_allocation_context_stats_inc:
     case _update_allocation_context_stats_full:
       return "Update Allocation Context Stats";
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/gc_interface/gcCause.hpp
--- a/src/share/vm/gc_interface/gcCause.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/gc_interface/gcCause.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -47,6 +47,7 @@
     _heap_inspection,
     _heap_dump,
     _wb_young_gc,
+    _wb_conc_mark,
     _update_allocation_context_stats_inc,
     _update_allocation_context_stats_full,
 
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/c2_globals.hpp
--- a/src/share/vm/opto/c2_globals.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/c2_globals.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -205,6 +205,9 @@
   notproduct(bool, TraceProfileTripCount, false,                            \
           "Trace profile loop trip count information")                      \
                                                                             \
+  product(bool, UseCountedLoopSafepoints, false,                            \
+          "Force counted loops to keep a safepoint")                        \
+                                                                            \
   product(bool, UseLoopPredicate, true,                                     \
           "Generate a predicate to select fast/slow loop versions")         \
                                                                             \
@@ -669,6 +672,9 @@
   product_pd(bool, TrapBasedRangeChecks,                                    \
           "Generate code for range checks that uses a cmp and trap "        \
           "instruction raising SIGTRAP. Used on PPC64.")                    \
+                                                                            \
+  develop(bool, RenumberLiveNodes, true,                                    \
+          "Renumber live nodes")                                            \
 
 C2_FLAGS(DECLARE_DEVELOPER_FLAG, DECLARE_PD_DEVELOPER_FLAG, DECLARE_PRODUCT_FLAG, DECLARE_PD_PRODUCT_FLAG, DECLARE_DIAGNOSTIC_FLAG, DECLARE_EXPERIMENTAL_FLAG, DECLARE_NOTPRODUCT_FLAG)
 
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/compile.cpp
--- a/src/share/vm/opto/compile.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/compile.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -2093,6 +2093,20 @@
   // so keep only the actual candidates for optimizations.
   cleanup_expensive_nodes(igvn);
 
+  if (!failing() && RenumberLiveNodes && live_nodes() + NodeLimitFudgeFactor < unique()) {
+    NOT_PRODUCT(Compile::TracePhase t2("", &_t_renumberLive, TimeCompiler);)
+    initial_gvn()->replace_with(&igvn);
+    for_igvn()->clear();
+    Unique_Node_List new_worklist(C->comp_arena());
+    {
+      ResourceMark rm;
+      PhaseRenumberLive prl = PhaseRenumberLive(initial_gvn(), for_igvn(), &new_worklist);
+    }
+    set_for_igvn(&new_worklist);
+    igvn = PhaseIterGVN(initial_gvn());
+    igvn.optimize();
+  }
+
   // Perform escape analysis
   if (_do_escape_analysis && ConnectionGraph::has_candidates(this)) {
     if (has_loops()) {
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/loopnode.cpp
--- a/src/share/vm/opto/loopnode.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/loopnode.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -685,14 +685,16 @@
 
   } // LoopLimitCheck
 
-  // Check for SafePoint on backedge and remove
-  Node *sfpt = x->in(LoopNode::LoopBackControl);
-  if (sfpt->Opcode() == Op_SafePoint && is_deleteable_safept(sfpt)) {
-    lazy_replace( sfpt, iftrue );
-    if (loop->_safepts != NULL) {
-      loop->_safepts->yank(sfpt);
+  if (!UseCountedLoopSafepoints) {
+    // Check for SafePoint on backedge and remove
+    Node *sfpt = x->in(LoopNode::LoopBackControl);
+    if (sfpt->Opcode() == Op_SafePoint && is_deleteable_safept(sfpt)) {
+      lazy_replace( sfpt, iftrue );
+      if (loop->_safepts != NULL) {
+        loop->_safepts->yank(sfpt);
+      }
+      loop->_tail = iftrue;
     }
-    loop->_tail = iftrue;
   }
 
   // Build a canonical trip test.
@@ -781,12 +783,14 @@
   lazy_replace( x, l );
   set_idom(l, init_control, dom_depth(x));
 
-  // Check for immediately preceding SafePoint and remove
-  Node *sfpt2 = le->in(0);
-  if (sfpt2->Opcode() == Op_SafePoint && is_deleteable_safept(sfpt2)) {
-    lazy_replace( sfpt2, sfpt2->in(TypeFunc::Control));
-    if (loop->_safepts != NULL) {
-      loop->_safepts->yank(sfpt2);
+  if (!UseCountedLoopSafepoints) {
+    // Check for immediately preceding SafePoint and remove
+    Node *sfpt2 = le->in(0);
+    if (sfpt2->Opcode() == Op_SafePoint && is_deleteable_safept(sfpt2)) {
+      lazy_replace( sfpt2, sfpt2->in(TypeFunc::Control));
+      if (loop->_safepts != NULL) {
+        loop->_safepts->yank(sfpt2);
+      }
     }
   }
 
@@ -1806,6 +1810,37 @@
   }
 }
 
+void IdealLoopTree::remove_safepoints(PhaseIdealLoop* phase, bool keep_one) {
+  Node* keep = NULL;
+  if (keep_one) {
+    // Look for a safepoint on the idom-path.
+    for (Node* i = tail(); i != _head; i = phase->idom(i)) {
+      if (i->Opcode() == Op_SafePoint && phase->get_loop(i) == this) {
+        keep = i;
+        break; // Found one
+      }
+    }
+  }
+
+  // Don't remove any safepoints if it is requested to keep a single safepoint and
+  // no safepoint was found on idom-path. It is not safe to remove any safepoint
+  // in this case since there's no safepoint dominating all paths in the loop body.
+  bool prune = !keep_one || keep != NULL;
+
+  // Delete other safepoints in this loop.
+  Node_List* sfpts = _safepts;
+  if (prune && sfpts != NULL) {
+    assert(keep == NULL || keep->Opcode() == Op_SafePoint, "not safepoint");
+    for (uint i = 0; i < sfpts->size(); i++) {
+      Node* n = sfpts->at(i);
+      assert(phase->get_loop(n) == this, "");
+      if (n != keep && phase->is_deleteable_safept(n)) {
+        phase->lazy_replace(n, n->in(TypeFunc::Control));
+      }
+    }
+  }
+}
+
 //------------------------------counted_loop-----------------------------------
 // Convert to counted loops where possible
 void IdealLoopTree::counted_loop( PhaseIdealLoop *phase ) {
@@ -1817,42 +1852,23 @@
 
   if (_head->is_CountedLoop() ||
       phase->is_counted_loop(_head, this)) {
-    _has_sfpt = 1;              // Indicate we do not need a safepoint here
-
-    // Look for safepoints to remove.
-    Node_List* sfpts = _safepts;
-    if (sfpts != NULL) {
-      for (uint i = 0; i < sfpts->size(); i++) {
-        Node* n = sfpts->at(i);
-        assert(phase->get_loop(n) == this, "");
-        if (phase->is_deleteable_safept(n)) {
-          phase->lazy_replace(n, n->in(TypeFunc::Control));
-        }
-      }
+
+    if (!UseCountedLoopSafepoints) {
+      // Indicate we do not need a safepoint here
+      _has_sfpt = 1;
     }
 
+    // Remove safepoints
+    bool keep_one_sfpt = !(_has_call || _has_sfpt);
+    remove_safepoints(phase, keep_one_sfpt);
+
     // Look for induction variables
     phase->replace_parallel_iv(this);
 
   } else if (_parent != NULL && !_irreducible) {
-    // Not a counted loop.
-    // Look for a safepoint on the idom-path.
-    Node* sfpt = tail();
-    for (; sfpt != _head; sfpt = phase->idom(sfpt)) {
-      if (sfpt->Opcode() == Op_SafePoint && phase->get_loop(sfpt) == this)
-        break; // Found one
-    }
-    // Delete other safepoints in this loop.
-    Node_List* sfpts = _safepts;
-    if (sfpts != NULL && sfpt != _head && sfpt->Opcode() == Op_SafePoint) {
-      for (uint i = 0; i < sfpts->size(); i++) {
-        Node* n = sfpts->at(i);
-        assert(phase->get_loop(n) == this, "");
-        if (n != sfpt && phase->is_deleteable_safept(n)) {
-          phase->lazy_replace(n, n->in(TypeFunc::Control));
-        }
-      }
-    }
+    // Not a counted loop. Keep one safepoint.
+    bool keep_one_sfpt = true;
+    remove_safepoints(phase, keep_one_sfpt);
   }
 
   // Recursively
@@ -1906,6 +1922,15 @@
     if (cl->is_main_loop()) tty->print(" main");
     if (cl->is_post_loop()) tty->print(" post");
   }
+  if (_has_call) tty->print(" has_call");
+  if (_has_sfpt) tty->print(" has_sfpt");
+  if (_rce_candidate) tty->print(" rce");
+  if (_safepts != NULL && _safepts->size() > 0) {
+    tty->print(" sfpts={"); _safepts->dump_simple(); tty->print(" }");
+  }
+  if (_required_safept != NULL && _required_safept->size() > 0) {
+    tty->print(" req={"); _required_safept->dump_simple(); tty->print(" }");
+  }
   tty->cr();
 }
 
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/loopnode.hpp
--- a/src/share/vm/opto/loopnode.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/loopnode.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1998, 2015, 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
@@ -403,6 +403,9 @@
   // encountered.
   void allpaths_check_safepts(VectorSet &visited, Node_List &stack);
 
+  // Remove safepoints from loop. Optionally keeping one.
+  void remove_safepoints(PhaseIdealLoop* phase, bool keep_one);
+
   // Convert to counted loops where possible
   void counted_loop( PhaseIdealLoop *phase );
 
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/node.cpp
--- a/src/share/vm/opto/node.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/node.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -325,6 +325,9 @@
 // Create a Node, with a given number of required edges.
 Node::Node(uint req)
   : _idx(IDX_INIT(req))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   assert( req < Compile::current()->max_node_limit() - NodeLimitFudgeFactor, "Input limit exceeded" );
   debug_only( verify_construction() );
@@ -344,6 +347,9 @@
 //------------------------------Node-------------------------------------------
 Node::Node(Node *n0)
   : _idx(IDX_INIT(1))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -356,6 +362,9 @@
 //------------------------------Node-------------------------------------------
 Node::Node(Node *n0, Node *n1)
   : _idx(IDX_INIT(2))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -370,6 +379,9 @@
 //------------------------------Node-------------------------------------------
 Node::Node(Node *n0, Node *n1, Node *n2)
   : _idx(IDX_INIT(3))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -386,6 +398,9 @@
 //------------------------------Node-------------------------------------------
 Node::Node(Node *n0, Node *n1, Node *n2, Node *n3)
   : _idx(IDX_INIT(4))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -404,6 +419,9 @@
 //------------------------------Node-------------------------------------------
 Node::Node(Node *n0, Node *n1, Node *n2, Node *n3, Node *n4)
   : _idx(IDX_INIT(5))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -425,6 +443,9 @@
 Node::Node(Node *n0, Node *n1, Node *n2, Node *n3,
                      Node *n4, Node *n5)
   : _idx(IDX_INIT(6))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -448,6 +469,9 @@
 Node::Node(Node *n0, Node *n1, Node *n2, Node *n3,
                      Node *n4, Node *n5, Node *n6)
   : _idx(IDX_INIT(7))
+#ifdef ASSERT
+  , _parse_idx(_idx)
+#endif
 {
   debug_only( verify_construction() );
   NOT_PRODUCT(nodes_created++);
@@ -2083,6 +2107,17 @@
 #endif
 }
 
+void Node_List::dump_simple() const {
+#ifndef PRODUCT
+  for( uint i = 0; i < _cnt; i++ )
+    if( _nodes[i] ) {
+      tty->print(" %d", _nodes[i]->_idx);
+    } else {
+      tty->print(" NULL");
+    }
+#endif
+}
+
 //=============================================================================
 //------------------------------remove-----------------------------------------
 void Unique_Node_List::remove( Node *n ) {
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/node.hpp
--- a/src/share/vm/opto/node.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/node.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -294,10 +294,16 @@
 
  public:
   // Each Node is assigned a unique small/dense number.  This number is used
-  // to index into auxiliary arrays of data and bitvectors.
-  // It is declared const to defend against inadvertant assignment,
-  // since it is used by clients as a naked field.
+  // to index into auxiliary arrays of data and bit vectors.
+  // The field _idx is declared constant to defend against inadvertent assignments,
+  // since it is used by clients as a naked field. However, the field's value can be
+  // changed using the set_idx() method.
+  //
+  // The PhaseRenumberLive phase renumbers nodes based on liveness information.
+  // Therefore, it updates the value of the _idx field. The parse-time _idx is
+  // preserved in _parse_idx.
   const node_idx_t _idx;
+  DEBUG_ONLY(const node_idx_t _parse_idx;)
 
   // Get the (read-only) number of input edges
   uint req() const { return _cnt; }
@@ -1368,6 +1374,7 @@
   void clear() { _cnt = 0; Node_Array::clear(); } // retain storage
   uint size() const { return _cnt; }
   void dump() const;
+  void dump_simple() const;
 };
 
 //------------------------------Unique_Node_List-------------------------------
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/phase.cpp
--- a/src/share/vm/opto/phase.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/phase.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -67,6 +67,8 @@
 elapsedTimer   Phase::_t_iterGVN;
 elapsedTimer   Phase::_t_iterGVN2;
 elapsedTimer   Phase::_t_incrInline;
+elapsedTimer   Phase::_t_renumberLive;
+
 
 // Subtimers for _t_registerAllocation
 elapsedTimer   Phase::_t_ctorChaitin;
@@ -115,13 +117,14 @@
     }
     tty->print_cr ("      iterGVN        : %3.3f sec", Phase::_t_iterGVN.seconds());
     tty->print_cr ("      incrInline     : %3.3f sec", Phase::_t_incrInline.seconds());
+    tty->print_cr ("      renumberLive   : %3.3f sec", Phase::_t_renumberLive.seconds());
     tty->print_cr ("      idealLoop      : %3.3f sec", Phase::_t_idealLoop.seconds());
     tty->print_cr ("      idealLoopVerify: %3.3f sec", Phase::_t_idealLoopVerify.seconds());
     tty->print_cr ("      ccp            : %3.3f sec", Phase::_t_ccp.seconds());
     tty->print_cr ("      iterGVN2       : %3.3f sec", Phase::_t_iterGVN2.seconds());
     tty->print_cr ("      macroExpand    : %3.3f sec", Phase::_t_macroExpand.seconds());
     tty->print_cr ("      graphReshape   : %3.3f sec", Phase::_t_graphReshaping.seconds());
-    double optimizer_subtotal = Phase::_t_iterGVN.seconds() + Phase::_t_iterGVN2.seconds() +
+    double optimizer_subtotal = Phase::_t_iterGVN.seconds() + Phase::_t_iterGVN2.seconds() + Phase::_t_renumberLive.seconds() +
       Phase::_t_escapeAnalysis.seconds() + Phase::_t_macroEliminate.seconds() +
       Phase::_t_idealLoop.seconds() + Phase::_t_ccp.seconds() +
       Phase::_t_macroExpand.seconds() + Phase::_t_graphReshaping.seconds();
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/phase.hpp
--- a/src/share/vm/opto/phase.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/phase.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -40,22 +40,23 @@
 class Phase : public StackObj {
 public:
   enum PhaseNumber {
-    Compiler,                   // Top-level compiler phase
-    Parser,                     // Parse bytecodes
-    Remove_Useless,             // Remove useless nodes
-    Optimistic,                 // Optimistic analysis phase
-    GVN,                        // Pessimistic global value numbering phase
-    Ins_Select,                 // Instruction selection phase
-    CFG,                        // Build a CFG
-    BlockLayout,                // Linear ordering of blocks
-    Register_Allocation,        // Register allocation, duh
-    LIVE,                       // Dragon-book LIVE range problem
-    StringOpts,                 // StringBuilder related optimizations
-    Interference_Graph,         // Building the IFG
-    Coalesce,                   // Coalescing copies
-    Ideal_Loop,                 // Find idealized trip-counted loops
-    Macro_Expand,               // Expand macro nodes
-    Peephole,                   // Apply peephole optimizations
+    Compiler,                         // Top-level compiler phase
+    Parser,                           // Parse bytecodes
+    Remove_Useless,                   // Remove useless nodes
+    Remove_Useless_And_Renumber_Live, // First, remove useless nodes from the graph. Then, renumber live nodes.
+    Optimistic,                       // Optimistic analysis phase
+    GVN,                              // Pessimistic global value numbering phase
+    Ins_Select,                       // Instruction selection phase
+    CFG,                              // Build a CFG
+    BlockLayout,                      // Linear ordering of blocks
+    Register_Allocation,              // Register allocation, duh
+    LIVE,                             // Dragon-book LIVE range problem
+    StringOpts,                       // StringBuilder related optimizations
+    Interference_Graph,               // Building the IFG
+    Coalesce,                         // Coalescing copies
+    Ideal_Loop,                       // Find idealized trip-counted loops
+    Macro_Expand,                     // Expand macro nodes
+    Peephole,                         // Apply peephole optimizations
     last_phase
   };
 protected:
@@ -102,6 +103,7 @@
   static elapsedTimer   _t_iterGVN;
   static elapsedTimer   _t_iterGVN2;
   static elapsedTimer   _t_incrInline;
+  static elapsedTimer   _t_renumberLive;
 
 // Subtimers for _t_registerAllocation
   static elapsedTimer   _t_ctorChaitin;
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/phaseX.cpp
--- a/src/share/vm/opto/phaseX.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/phaseX.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -398,7 +398,7 @@
 //=============================================================================
 //------------------------------PhaseRemoveUseless-----------------------------
 // 1) Use a breadthfirst walk to collect useful nodes reachable from root.
-PhaseRemoveUseless::PhaseRemoveUseless( PhaseGVN *gvn, Unique_Node_List *worklist ) : Phase(Remove_Useless),
+PhaseRemoveUseless::PhaseRemoveUseless(PhaseGVN *gvn, Unique_Node_List *worklist, PhaseNumber phase_num) : Phase(phase_num),
   _useful(Thread::current()->resource_area()) {
 
   // Implementation requires 'UseLoopSafepoints == true' and an edge from root
@@ -435,6 +435,82 @@
   }
 }
 
+//=============================================================================
+//------------------------------PhaseRenumberLive------------------------------
+// First, remove useless nodes (equivalent to identifying live nodes).
+// Then, renumber live nodes.
+//
+// The set of live nodes is returned by PhaseRemoveUseless in the _useful structure.
+// If the number of live nodes is 'x' (where 'x' == _useful.size()), then the
+// PhaseRenumberLive updates the node ID of each node (the _idx field) with a unique
+// value in the range [0, x).
+//
+// At the end of the PhaseRenumberLive phase, the compiler's count of unique nodes is
+// updated to 'x' and the list of dead nodes is reset (as there are no dead nodes).
+//
+// The PhaseRenumberLive phase updates two data structures with the new node IDs.
+// (1) The worklist is used by the PhaseIterGVN phase to identify nodes that must be
+// processed. A new worklist (with the updated node IDs) is returned in 'new_worklist'.
+// (2) Type information (the field PhaseGVN::_types) maps type information to each
+// node ID. The mapping is updated to use the new node IDs as well. Updated type
+// information is returned in PhaseGVN::_types.
+//
+// The PhaseRenumberLive phase does not preserve the order of elements in the worklist.
+//
+// Other data structures used by the compiler are not updated. The hash table for value
+// numbering (the field PhaseGVN::_table) is not updated because computing the hash
+// values is not based on node IDs. The field PhaseGVN::_nodes is not updated either
+// because it is empty wherever PhaseRenumberLive is used.
+PhaseRenumberLive::PhaseRenumberLive(PhaseGVN* gvn,
+                                     Unique_Node_List* worklist, Unique_Node_List* new_worklist,
+                                     PhaseNumber phase_num) :
+  PhaseRemoveUseless(gvn, worklist, Remove_Useless_And_Renumber_Live) {
+
+  assert(RenumberLiveNodes, "RenumberLiveNodes must be set to true for node renumbering to take place");
+  assert(C->live_nodes() == _useful.size(), "the number of live nodes must match the number of useful nodes");
+  assert(gvn->nodes_size() == 0, "GVN must not contain any nodes at this point");
+
+  uint old_unique_count = C->unique();
+  uint live_node_count = C->live_nodes();
+  uint worklist_size = worklist->size();
+
+  // Storage for the updated type information.
+  Type_Array new_type_array(C->comp_arena());
+
+  // Iterate over the set of live nodes.
+  uint current_idx = 0; // The current new node ID. Incremented after every assignment.
+  for (uint i = 0; i < _useful.size(); i++) {
+    Node* n = _useful.at(i);
+    const Type* type = gvn->type_or_null(n);
+    new_type_array.map(current_idx, type);
+
+    bool in_worklist = false;
+    if (worklist->member(n)) {
+      in_worklist = true;
+    }
+
+    n->set_idx(current_idx); // Update node ID.
+
+    if (in_worklist) {
+      new_worklist->push(n);
+    }
+
+    current_idx++;
+  }
+
+  assert(worklist_size == new_worklist->size(), "the new worklist must have the same size as the original worklist");
+  assert(live_node_count == current_idx, "all live nodes must be processed");
+
+  // Replace the compiler's type information with the updated type information.
+  gvn->replace_types(new_type_array);
+
+  // Update the unique node count of the compilation to the number of currently live nodes.
+  C->set_unique(live_node_count);
+
+  // Set the dead node count to 0 and reset dead node list.
+  C->reset_dead_node_list();
+}
+
 
 //=============================================================================
 //------------------------------PhaseTransform---------------------------------
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/opto/phaseX.hpp
--- a/src/share/vm/opto/phaseX.hpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/opto/phaseX.hpp	Tue Jan 05 08:28:01 2016 -0800
@@ -148,11 +148,21 @@
   Unique_Node_List _useful;   // Nodes reachable from root
                               // list is allocated from current resource area
 public:
-  PhaseRemoveUseless( PhaseGVN *gvn, Unique_Node_List *worklist );
+  PhaseRemoveUseless(PhaseGVN *gvn, Unique_Node_List *worklist, PhaseNumber phase_num = Remove_Useless);
 
   Unique_Node_List *get_useful() { return &_useful; }
 };
 
+//------------------------------PhaseRenumber----------------------------------
+// Phase that first performs a PhaseRemoveUseless, then it renumbers compiler
+// structures accordingly.
+class PhaseRenumberLive : public PhaseRemoveUseless {
+public:
+  PhaseRenumberLive(PhaseGVN* gvn,
+                    Unique_Node_List* worklist, Unique_Node_List* new_worklist,
+                    PhaseNumber phase_num = Remove_Useless_And_Renumber_Live);
+};
+
 
 //------------------------------PhaseTransform---------------------------------
 // Phases that analyze, then transform.  Constructing the Phase object does any
@@ -162,7 +172,7 @@
 class PhaseTransform : public Phase {
 protected:
   Arena*     _arena;
-  Node_Array _nodes;           // Map old node indices to new nodes.
+  Node_List  _nodes;           // Map old node indices to new nodes.
   Type_Array _types;           // Map old node indices to Types.
 
   // ConNode caches:
@@ -187,7 +197,13 @@
 
   Arena*      arena()   { return _arena; }
   Type_Array& types()   { return _types; }
+  void replace_types(Type_Array new_types) {
+    _types = new_types;
+  }
   // _nodes is used in varying ways by subclasses, which define local accessors
+  uint nodes_size() {
+    return _nodes.size();
+  }
 
 public:
   // Get a previously recorded type for the node n.
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/prims/whitebox.cpp
--- a/src/share/vm/prims/whitebox.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/prims/whitebox.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -45,6 +45,7 @@
 #if INCLUDE_ALL_GCS
 #include "gc_implementation/parallelScavenge/parallelScavengeHeap.inline.hpp"
 #include "gc_implementation/g1/concurrentMark.hpp"
+#include "gc_implementation/g1/concurrentMarkThread.hpp"
 #include "gc_implementation/g1/g1CollectedHeap.inline.hpp"
 #include "gc_implementation/g1/heapRegionRemSet.hpp"
 #endif // INCLUDE_ALL_GCS
@@ -323,8 +324,16 @@
 
 WB_ENTRY(jboolean, WB_G1InConcurrentMark(JNIEnv* env, jobject o))
   G1CollectedHeap* g1 = G1CollectedHeap::heap();
-  ConcurrentMark* cm = g1->concurrent_mark();
-  return cm->concurrent_marking_in_progress();
+  return g1->concurrent_mark()->cmThread()->during_cycle();
+WB_END
+
+WB_ENTRY(jboolean, WB_G1StartMarkCycle(JNIEnv* env, jobject o))
+  G1CollectedHeap* g1h = G1CollectedHeap::heap();
+  if (!g1h->concurrent_mark()->cmThread()->during_cycle()) {
+    g1h->collect(GCCause::_wb_conc_mark);
+    return true;
+  }
+  return false;
 WB_END
 
 WB_ENTRY(jint, WB_G1RegionSize(JNIEnv* env, jobject o))
@@ -1031,6 +1040,7 @@
   {CC"g1NumMaxRegions",    CC"()J",                   (void*)&WB_G1NumMaxRegions  },
   {CC"g1NumFreeRegions",   CC"()J",                   (void*)&WB_G1NumFreeRegions  },
   {CC"g1RegionSize",       CC"()I",                   (void*)&WB_G1RegionSize      },
+  {CC"g1StartConcMarkCycle",       CC"()Z",           (void*)&WB_G1StartMarkCycle  },
   {CC"g1AuxiliaryMemoryUsage", CC"()Ljava/lang/management/MemoryUsage;",
                                                       (void*)&WB_G1AuxiliaryMemoryUsage  },
 #endif // INCLUDE_ALL_GCS
diff -r f796867c1bcb -r c3091ebd2811 src/share/vm/runtime/safepoint.cpp
--- a/src/share/vm/runtime/safepoint.cpp	Tue Dec 15 22:59:18 2015 -0800
+++ b/src/share/vm/runtime/safepoint.cpp	Tue Jan 05 08:28:01 2016 -0800
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2015, 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
@@ -739,80 +739,12 @@
 // ------------------------------------------------------------------------------------------------------
 // Exception handlers
 
-#ifndef PRODUCT
-
-#ifdef SPARC
-
-#ifdef _LP64
-#define PTR_PAD ""
-#else
-#define PTR_PAD "        "
-#endif
-
-static void print_ptrs(intptr_t oldptr, intptr_t newptr, bool wasoop) {
-  bool is_oop = newptr ? (cast_to_oop(newptr))->is_oop() : false;
-  tty->print_cr(PTR_FORMAT PTR_PAD " %s %c " PTR_FORMAT PTR_PAD " %s %s",
-                oldptr, wasoop?"oop":"   ", oldptr == newptr ? ' ' : '!',
-                newptr, is_oop?"oop":"   ", (wasoop && !is_oop) ? "STALE" : ((wasoop==false&&is_oop==false&&oldptr !=newptr)?"STOMP":"     "));
-}
-
-static void print_longs(jlong oldptr, jlong newptr, bool wasoop) {
-  bool is_oop = newptr ? (cast_to_oop(newptr))->is_oop() : false;
-  tty->print_cr(PTR64_FORMAT " %s %c " PTR64_FORMAT " %s %s",
-                oldptr, wasoop?"oop":"   ", oldptr == newptr ? ' ' : '!',
-                newptr, is_oop?"oop":"   ", (wasoop && !is_oop) ? "STALE" : ((wasoop==false&&is_oop==false&&oldptr !=newptr)?"STOMP":"     "));
-}
-
-static void print_me(intptr_t *new_sp, intptr_t *old_sp, bool *was_oops) {
-#ifdef _LP64
-  tty->print_cr("--------+------address-----+------before-----------+-------after----------+");
-  const int incr = 1;           // Increment to skip a long, in units of intptr_t
-#else
-  tty->print_cr("--------+--address-+------before-----------+-------after----------+");
-  const int incr = 2;           // Increment to skip a long, in units of intptr_t
-#endif
-  tty->print_cr("---SP---|");
-  for( int i=0; i<16; i++ ) {
-    tty->print("blob %c%d |"PTR_FORMAT" ","LO"[i>>3],i&7,new_sp); print_ptrs(*old_sp++,*new_sp++,*was_oops++); }
-  tty->print_cr("--------|");
-  for( int i1=0; i1<frame::memory_parameter_word_sp_offset-16; i1++ ) {
-    tty->print("argv pad|"PTR_FORMAT" ",new_sp); print_ptrs(*old_sp++,*new_sp++,*was_oops++); }
-  tty->print("     pad|"PTR_FORMAT" ",new_sp); print_ptrs(*old_sp++,*new_sp++,*was_oops++);
-  tty->print_cr("--------|");
-  tty->print(" G1     |"PTR_FORMAT" ",new_sp); print_longs(*(jlong*)old_sp,*(jlong*)new_sp,was_oops[incr-1]); old_sp += incr; new_sp += incr; was_oops += incr;
-  tty->print(" G3     |"PTR_FORMAT" ",new_sp); print_longs(*(jlong*)old_sp,*(jlong*)new_sp,was_oops[incr-1]); old_sp += incr; new_sp += incr; was_oops += incr;
-  tty->print(" G4     |"PTR_FORMAT" ",new_sp); print_longs(*(jlong*)old_sp,*(jlong*)new_sp,was_oops[incr-1]); old_sp += incr; new_sp += incr; was_oops += incr;
-  tty->print(" G5     |"PTR_FORMAT" ",new_sp); print_longs(*(jlong*)old_sp,*(jlong*)new_sp,was_oops[incr-1]); old_sp += incr; new_sp += incr; was_oops += incr;
-  tty->print_cr(" FSR    |"PTR_FORMAT" "PTR64_FORMAT"       "PTR64_FORMAT,new_sp,*(jlong*)old_sp,*(jlong*)new_sp);
-  old_sp += incr; new_sp += incr; was_oops += incr;
-  // Skip the floats
-  tty->print_cr("--Float-|"PTR_FORMAT,new_sp);
-  tty->print_cr("---FP---|");
-  old_sp += incr*32;  new_sp += incr*32;  was_oops += incr*32;
-  for( int i2=0; i2<16; i2++ ) {
-    tty->print("call %c%d |"PTR_FORMAT" ","LI"[i2>>3],i2&7,new_sp); print_ptrs(*old_sp++,*new_sp++,*was_oops++); }
-  tty->cr();
-}
-#endif  // SPARC
-#endif  // PRODUCT
-
 
 void SafepointSynchronize::handle_polling_page_exception(JavaThread *thread) {
   assert(thread->is_Java_thread(), "polling reference encountered by VM thread");
   assert(thread->thread_state() == _thread_in_Java, "should come from Java code");
   assert(SafepointSynchronize::is_synchronizing(), "polling encountered outside safepoint synchronization");
 
-  // Uncomment this to get some serious before/after printing of the
-  // Sparc safepoint-blob frame structure.
-  /*
-  intptr_t* sp = thread->last_Java_sp();
-  intptr_t stack_copy[150];
-  for( int i=0; i<150; i++ ) stack_copy[i] = sp[i];
-  bool was_oops[150];
-  for( int i=0; i<150; i++ )
-    was_oops[i] = stack_copy[i] ? ((oop)stack_copy[i])->is_oop() : false;
-  */
-
   if (ShowSafepointMsgs) {
     tty->print("handle_polling_page_exception: ");
   }
@@ -824,7 +756,6 @@
   ThreadSafepointState* state = thread->safepoint_state();
 
   state->handle_polling_page_exception();
-  // print_me(sp,stack_copy,was_oops);
 }
 
 
diff -r f796867c1bcb -r c3091ebd2811 test/compiler/loopopts/UseCountedLoopSafepoints.java
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compiler/loopopts/UseCountedLoopSafepoints.java	Tue Jan 05 08:28:01 2016 -0800
@@ -0,0 +1,67 @@
+/*
+ * Copyright (c) 2015, 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 6869327
+ * @summary Test that C2 flag UseCountedLoopSafepoints ensures a safepoint is kept in a CountedLoop
+ * @library /testlibrary
+ * @run main UseCountedLoopSafepoints
+ */
+
+import java.util.concurrent.atomic.AtomicLong;
+import com.oracle.java.testlibrary.ProcessTools;
+import com.oracle.java.testlibrary.OutputAnalyzer;
+
+public class UseCountedLoopSafepoints {
+    private static final AtomicLong _num = new AtomicLong(0);
+
+    // Uses the fact that an EnableBiasedLocking vmop will be started
+    // after 500ms, while we are still in the loop. If there is a
+    // safepoint in the counted loop, then we will reach safepoint
+    // very quickly. Otherwise SafepointTimeout will be hit.
+    public static void main (String args[]) throws Exception {
+        if (args.length == 1) {
+            final int loops = Integer.parseInt(args[0]);
+            for (int i = 0; i < loops; i++) {
+                _num.addAndGet(1);
+            }
+        } else {
+            ProcessBuilder pb = ProcessTools.createJavaProcessBuilder(
+                    "-XX:+IgnoreUnrecognizedVMOptions",
+                    "-XX:-TieredCompilation",
+                    "-XX:+UseBiasedLocking",
+                    "-XX:BiasedLockingStartupDelay=500",
+                    "-XX:+SafepointTimeout",
+                    "-XX:SafepointTimeoutDelay=2000",
+                    "-XX:+UseCountedLoopSafepoints",
+                    "UseCountedLoopSafepoints",
+                    "2000000000"
+                    );
+            OutputAnalyzer output = new OutputAnalyzer(pb.start());
+            output.shouldNotContain("Timeout detected");
+            output.shouldHaveExitValue(0);
+        }
+    }
+}
diff -r f796867c1bcb -r c3091ebd2811 test/gc/whitebox/TestConcMarkCycleWB.java
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/gc/whitebox/TestConcMarkCycleWB.java	Tue Jan 05 08:28:01 2016 -0800
@@ -0,0 +1,57 @@
+/*
+ * Copyright (c) 2014, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+/*
+ * @test TestConMarkCycleWB
+ * @bug 8065579
+ * @requires vm.gc=="null" | vm.gc=="G1"
+ * @library /testlibrary /testlibrary/whitebox
+ * @build ClassFileInstaller com.oracle.java.testlibrary.* sun.hotspot.WhiteBox TestConcMarkCycleWB
+ * @run main ClassFileInstaller sun.hotspot.WhiteBox
+ *                              sun.hotspot.WhiteBox$WhiteBoxPermission
+ * @run main/othervm -Xbootclasspath/a:. -XX:+UnlockDiagnosticVMOptions -XX:+WhiteBoxAPI -XX:+UseG1GC TestConcMarkCycleWB
+ * @summary Verifies that ConcurrentMarking-related WB works properly
+ */
+import static com.oracle.java.testlibrary.Asserts.assertFalse;
+import static com.oracle.java.testlibrary.Asserts.assertTrue;
+import sun.hotspot.WhiteBox;
+
+public class TestConcMarkCycleWB {
+
+    public static void main(String[] args) throws Exception {
+        WhiteBox wb = WhiteBox.getWhiteBox();
+
+        wb.youngGC();
+        assertTrue(wb.g1StartConcMarkCycle());
+        while (wb.g1InConcurrentMark()) {
+            Thread.sleep(5);
+        }
+
+        wb.fullGC();
+        assertTrue(wb.g1StartConcMarkCycle());
+        while (wb.g1InConcurrentMark()) {
+            Thread.sleep(5);
+        }
+        assertTrue(wb.g1StartConcMarkCycle());
+    }
+}
diff -r f796867c1bcb -r c3091ebd2811 test/testlibrary/whitebox/sun/hotspot/WhiteBox.java
--- a/test/testlibrary/whitebox/sun/hotspot/WhiteBox.java	Tue Dec 15 22:59:18 2015 -0800
+++ b/test/testlibrary/whitebox/sun/hotspot/WhiteBox.java	Tue Jan 05 08:28:01 2016 -0800
@@ -174,12 +174,16 @@
   public native long incMetaspaceCapacityUntilGC(long increment);
   public native long metaspaceCapacityUntilGC();
 
-  // force Young GC
+  // Force Young GC
   public native void youngGC();
 
-  // force Full GC
+  // Force Full GC
   public native void fullGC();
 
+  // Method tries to start concurrent mark cycle.
+  // It returns false if CM Thread is always in concurrent cycle.
+  public native boolean g1StartConcMarkCycle();
+
   // Tests on ReservedSpace/VirtualSpace classes
   public native int stressVirtualSpaceResize(long reservedSpaceSize, long magnitude, long iterations);
   public native void runMemoryUnitTests();