Merge

.hgtags

@@ -522,6 +522,8 @@ c9635cad4a5d794a96b4a26d3e7ad1d783133add hs25.40-b09
 b63d0e8bfc0738bba21ae67779780f59118a95f7 jdk8u40-b07
 5c1b5be2c69bcae610a790e9438da446c61d3361 hs25.40-b12
 905a16825d2931345a7d6dba9e427f98eb51761a jdk8u40-b08
+d96716f6cbba9f000dfb1da39d2b81264f4cdea7 hs25.40-b13
+7ff8d51e0d8fc71f3ad31fd15817083341416ca8 jdk8u40-b09
 a4d44dfb7d30eea54bc172e4429a655454ae0bbf jdk8u25-b00
 9a2152fbd929b0d8b2f5c326a5526214ae71731a jdk8u25-b01
 d3d5604ea0dea3812e87ba76ac199d0a8be6f49f jdk8u25-b02

make/aix/makefiles/fastdebug.make

@@ -67,7 +67,6 @@ MAPFILE = $(GAMMADIR)/make/aix/makefiles/mapfile-vers-debug
 #    not justified.
 LFLAGS_QIPA=
 
-G_SUFFIX = _g
 VERSION = optimized
 SYSDEFS += -DASSERT -DFASTDEBUG
 PICFLAGS = DEFAULT

make/hotspot_version

@@ -35,7 +35,7 @@ HOTSPOT_VM_COPYRIGHT=Copyright 2014
 
 HS_MAJOR_VER=25
 HS_MINOR_VER=40
-HS_BUILD_NUMBER=12
+HS_BUILD_NUMBER=13
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=8

src/cpu/ppc/vm/assembler_ppc.hpp

@@ -268,8 +268,35 @@ class Assembler : public AbstractAssembler {
 
     ISEL_OPCODE   = (31u << OPCODE_SHIFT |  15u << 1),
 
-    MTLR_OPCODE   = (31u << OPCODE_SHIFT | 467u << 1 | 8 << SPR_0_4_SHIFT),
-    MFLR_OPCODE   = (31u << OPCODE_SHIFT | 339u << 1 | 8 << SPR_0_4_SHIFT),
+    // Special purpose registers
+    MTSPR_OPCODE  = (31u << OPCODE_SHIFT | 467u << 1),
+    MFSPR_OPCODE  = (31u << OPCODE_SHIFT | 339u << 1),
+
+    MTXER_OPCODE  = (MTSPR_OPCODE | 1 << SPR_0_4_SHIFT),
+    MFXER_OPCODE  = (MFSPR_OPCODE | 1 << SPR_0_4_SHIFT),
+
+    MTDSCR_OPCODE = (MTSPR_OPCODE | 3 << SPR_0_4_SHIFT),
+    MFDSCR_OPCODE = (MFSPR_OPCODE | 3 << SPR_0_4_SHIFT),
+
+    MTLR_OPCODE   = (MTSPR_OPCODE | 8 << SPR_0_4_SHIFT),
+    MFLR_OPCODE   = (MFSPR_OPCODE | 8 << SPR_0_4_SHIFT),
+
+    MTCTR_OPCODE  = (MTSPR_OPCODE | 9 << SPR_0_4_SHIFT),
+    MFCTR_OPCODE  = (MFSPR_OPCODE | 9 << SPR_0_4_SHIFT),
+
+    MTTFHAR_OPCODE   = (MTSPR_OPCODE | 128 << SPR_0_4_SHIFT),
+    MFTFHAR_OPCODE   = (MFSPR_OPCODE | 128 << SPR_0_4_SHIFT),
+    MTTFIAR_OPCODE   = (MTSPR_OPCODE | 129 << SPR_0_4_SHIFT),
+    MFTFIAR_OPCODE   = (MFSPR_OPCODE | 129 << SPR_0_4_SHIFT),
+    MTTEXASR_OPCODE  = (MTSPR_OPCODE | 130 << SPR_0_4_SHIFT),
+    MFTEXASR_OPCODE  = (MFSPR_OPCODE | 130 << SPR_0_4_SHIFT),
+    MTTEXASRU_OPCODE = (MTSPR_OPCODE | 131 << SPR_0_4_SHIFT),
+    MFTEXASRU_OPCODE = (MFSPR_OPCODE | 131 << SPR_0_4_SHIFT),
+
+    MTVRSAVE_OPCODE  = (MTSPR_OPCODE | 256 << SPR_0_4_SHIFT),
+    MFVRSAVE_OPCODE  = (MFSPR_OPCODE | 256 << SPR_0_4_SHIFT),
+
+    MFTB_OPCODE   = (MFSPR_OPCODE | 268 << SPR_0_4_SHIFT),
 
     MTCRF_OPCODE  = (31u << OPCODE_SHIFT | 144u << 1),
     MFCR_OPCODE   = (31u << OPCODE_SHIFT | 19u << 1),
@@ -291,9 +318,6 @@ class Assembler : public AbstractAssembler {
 
     // CTR-related opcodes
     BCCTR_OPCODE  = (19u << OPCODE_SHIFT | 528u << 1),
-    MTCTR_OPCODE  = (31u << OPCODE_SHIFT | 467u << 1 | 9 << SPR_0_4_SHIFT),
-    MFCTR_OPCODE  = (31u << OPCODE_SHIFT | 339u << 1 | 9 << SPR_0_4_SHIFT),
-
 
     LWZ_OPCODE   = (32u << OPCODE_SHIFT),
     LWZX_OPCODE  = (31u << OPCODE_SHIFT |  23u << 1),
@@ -585,6 +609,37 @@ class Assembler : public AbstractAssembler {
     MTVSCR_OPCODE  = (4u  << OPCODE_SHIFT | 1604u     ),
     MFVSCR_OPCODE  = (4u  << OPCODE_SHIFT | 1540u     ),
 
+    // AES (introduced with Power 8)
+    VCIPHER_OPCODE      = (4u  << OPCODE_SHIFT | 1288u),
+    VCIPHERLAST_OPCODE  = (4u  << OPCODE_SHIFT | 1289u),
+    VNCIPHER_OPCODE     = (4u  << OPCODE_SHIFT | 1352u),
+    VNCIPHERLAST_OPCODE = (4u  << OPCODE_SHIFT | 1353u),
+    VSBOX_OPCODE        = (4u  << OPCODE_SHIFT | 1480u),
+
+    // SHA (introduced with Power 8)
+    VSHASIGMAD_OPCODE   = (4u  << OPCODE_SHIFT | 1730u),
+    VSHASIGMAW_OPCODE   = (4u  << OPCODE_SHIFT | 1666u),
+
+    // Vector Binary Polynomial Multiplication (introduced with Power 8)
+    VPMSUMB_OPCODE      = (4u  << OPCODE_SHIFT | 1032u),
+    VPMSUMD_OPCODE      = (4u  << OPCODE_SHIFT | 1224u),
+    VPMSUMH_OPCODE      = (4u  << OPCODE_SHIFT | 1096u),
+    VPMSUMW_OPCODE      = (4u  << OPCODE_SHIFT | 1160u),
+
+    // Vector Permute and Xor (introduced with Power 8)
+    VPERMXOR_OPCODE     = (4u  << OPCODE_SHIFT |   45u),
+
+    // Transactional Memory instructions (introduced with Power 8)
+    TBEGIN_OPCODE    = (31u << OPCODE_SHIFT |  654u << 1),
+    TEND_OPCODE      = (31u << OPCODE_SHIFT |  686u << 1),
+    TABORT_OPCODE    = (31u << OPCODE_SHIFT |  910u << 1),
+    TABORTWC_OPCODE  = (31u << OPCODE_SHIFT |  782u << 1),
+    TABORTWCI_OPCODE = (31u << OPCODE_SHIFT |  846u << 1),
+    TABORTDC_OPCODE  = (31u << OPCODE_SHIFT |  814u << 1),
+    TABORTDCI_OPCODE = (31u << OPCODE_SHIFT |  878u << 1),
+    TSR_OPCODE       = (31u << OPCODE_SHIFT |  750u << 1),
+    TCHECK_OPCODE    = (31u << OPCODE_SHIFT |  718u << 1),
+
     // Icache and dcache related instructions
     DCBA_OPCODE    = (31u << OPCODE_SHIFT |  758u << 1),
     DCBZ_OPCODE    = (31u << OPCODE_SHIFT | 1014u << 1),
@@ -1420,6 +1475,25 @@ class Assembler : public AbstractAssembler {
   inline void mcrf( ConditionRegister crd, ConditionRegister cra);
   inline void mtcr( Register s);
 
+  // Special purpose registers
+  // Exception Register
+  inline void mtxer(Register s1);
+  inline void mfxer(Register d);
+  // Vector Register Save Register
+  inline void mtvrsave(Register s1);
+  inline void mfvrsave(Register d);
+  // Timebase
+  inline void mftb(Register d);
+  // Introduced with Power 8:
+  // Data Stream Control Register
+  inline void mtdscr(Register s1);
+  inline void mfdscr(Register d );
+  // Transactional Memory Registers
+  inline void mftfhar(Register d);
+  inline void mftfiar(Register d);
+  inline void mftexasr(Register d);
+  inline void mftexasru(Register d);
+
   // PPC 1, section 2.4.1 Branch Instructions
   inline void b(  address a, relocInfo::relocType rt = relocInfo::none);
   inline void b(  Label& L);
@@ -1860,6 +1934,39 @@ class Assembler : public AbstractAssembler {
   inline void mtvscr(   VectorRegister b);
   inline void mfvscr(   VectorRegister d);
 
+  // AES (introduced with Power 8)
+  inline void vcipher(     VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vcipherlast( VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vncipher(    VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vncipherlast(VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vsbox(       VectorRegister d, VectorRegister a);
+
+  // SHA (introduced with Power 8)
+  // Not yet implemented.
+
+  // Vector Binary Polynomial Multiplication (introduced with Power 8)
+  inline void vpmsumb(  VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vpmsumd(  VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vpmsumh(  VectorRegister d, VectorRegister a, VectorRegister b);
+  inline void vpmsumw(  VectorRegister d, VectorRegister a, VectorRegister b);
+
+  // Vector Permute and Xor (introduced with Power 8)
+  inline void vpermxor( VectorRegister d, VectorRegister a, VectorRegister b, VectorRegister c);
+
+  // Transactional Memory instructions (introduced with Power 8)
+  inline void tbegin_();    // R=0
+  inline void tbeginrot_(); // R=1 Rollback-Only Transaction
+  inline void tend_();    // A=0
+  inline void tendall_(); // A=1
+  inline void tabort_(Register a);
+  inline void tabortwc_(int t, Register a, Register b);
+  inline void tabortwci_(int t, Register a, int si);
+  inline void tabortdc_(int t, Register a, Register b);
+  inline void tabortdci_(int t, Register a, int si);
+  inline void tsuspend_(); // tsr with L=0
+  inline void tresume_();  // tsr with L=1
+  inline void tcheck(int f);
+
   // The following encoders use r0 as second operand. These instructions
   // read r0 as '0'.
   inline void lwzx( Register d, Register s2);

src/cpu/ppc/vm/assembler_ppc.inline.hpp

@@ -312,6 +312,25 @@ inline void Assembler::mcrf( ConditionRegister crd, ConditionRegister cra)
                                                       { emit_int32(MCRF_OPCODE | bf(crd) | bfa(cra)); }
 inline void Assembler::mtcr( Register s)          { Assembler::mtcrf(0xff, s); }
 
+// Special purpose registers
+// Exception Register
+inline void Assembler::mtxer(Register s1)         { emit_int32(MTXER_OPCODE | rs(s1)); }
+inline void Assembler::mfxer(Register d )         { emit_int32(MFXER_OPCODE | rt(d)); }
+// Vector Register Save Register
+inline void Assembler::mtvrsave(Register s1)      { emit_int32(MTVRSAVE_OPCODE | rs(s1)); }
+inline void Assembler::mfvrsave(Register d )      { emit_int32(MFVRSAVE_OPCODE | rt(d)); }
+// Timebase
+inline void Assembler::mftb(Register d )          { emit_int32(MFTB_OPCODE  | rt(d)); }
+// Introduced with Power 8:
+// Data Stream Control Register
+inline void Assembler::mtdscr(Register s1)        { emit_int32(MTDSCR_OPCODE | rs(s1)); }
+inline void Assembler::mfdscr(Register d )        { emit_int32(MFDSCR_OPCODE | rt(d)); }
+// Transactional Memory Registers
+inline void Assembler::mftfhar(Register d )       { emit_int32(MFTFHAR_OPCODE   | rt(d)); }
+inline void Assembler::mftfiar(Register d )       { emit_int32(MFTFIAR_OPCODE   | rt(d)); }
+inline void Assembler::mftexasr(Register d )      { emit_int32(MFTEXASR_OPCODE  | rt(d)); }
+inline void Assembler::mftexasru(Register d )     { emit_int32(MFTEXASRU_OPCODE | rt(d)); }
+
 // SAP JVM 2006-02-13 PPC branch instruction.
 // PPC 1, section 2.4.1 Branch Instructions
 inline void Assembler::b( address a, relocInfo::relocType rt) { emit_data(BXX_OPCODE| li(disp( intptr_t(a), intptr_t(pc()))) |aa(0)|lk(0), rt); }
@@ -735,6 +754,39 @@ inline void Assembler::vsrah(   VectorRegister d, VectorRegister a, VectorRegist
 inline void Assembler::mtvscr(  VectorRegister b)                                     { emit_int32( MTVSCR_OPCODE   | vrb(b)); }
 inline void Assembler::mfvscr(  VectorRegister d)                                     { emit_int32( MFVSCR_OPCODE   | vrt(d)); }
 
+// AES (introduced with Power 8)
+inline void Assembler::vcipher(     VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VCIPHER_OPCODE      | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vcipherlast( VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VCIPHERLAST_OPCODE  | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vncipher(    VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VNCIPHER_OPCODE     | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vncipherlast(VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VNCIPHERLAST_OPCODE | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vsbox(       VectorRegister d, VectorRegister a)                   { emit_int32( VSBOX_OPCODE        | vrt(d) | vra(a)         ); }
+
+// SHA (introduced with Power 8)
+// Not yet implemented.
+
+// Vector Binary Polynomial Multiplication (introduced with Power 8)
+inline void Assembler::vpmsumb(  VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPMSUMB_OPCODE | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vpmsumd(  VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPMSUMD_OPCODE | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vpmsumh(  VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPMSUMH_OPCODE | vrt(d) | vra(a) | vrb(b)); }
+inline void Assembler::vpmsumw(  VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPMSUMW_OPCODE | vrt(d) | vra(a) | vrb(b)); }
+
+// Vector Permute and Xor (introduced with Power 8)
+inline void Assembler::vpermxor( VectorRegister d, VectorRegister a, VectorRegister b, VectorRegister c) { emit_int32( VPMSUMW_OPCODE | vrt(d) | vra(a) | vrb(b) | vrc(c)); }
+
+// Transactional Memory instructions (introduced with Power 8)
+inline void Assembler::tbegin_()                                { emit_int32( TBEGIN_OPCODE | rc(1)); }
+inline void Assembler::tbeginrot_()                             { emit_int32( TBEGIN_OPCODE | /*R=1*/ 1u << (31-10) | rc(1)); }
+inline void Assembler::tend_()                                  { emit_int32( TEND_OPCODE | rc(1)); }
+inline void Assembler::tendall_()                               { emit_int32( TEND_OPCODE | /*A=1*/ 1u << (31-6) | rc(1)); }
+inline void Assembler::tabort_(Register a)                      { emit_int32( TABORT_OPCODE | ra(a) | rc(1)); }
+inline void Assembler::tabortwc_(int t, Register a, Register b) { emit_int32( TABORTWC_OPCODE | to(t) | ra(a) | rb(b) | rc(1)); }
+inline void Assembler::tabortwci_(int t, Register a, int si)    { emit_int32( TABORTWCI_OPCODE | to(t) | ra(a) | sh1620(si) | rc(1)); }
+inline void Assembler::tabortdc_(int t, Register a, Register b) { emit_int32( TABORTDC_OPCODE | to(t) | ra(a) | rb(b) | rc(1)); }
+inline void Assembler::tabortdci_(int t, Register a, int si)    { emit_int32( TABORTDCI_OPCODE | to(t) | ra(a) | sh1620(si) | rc(1)); }
+inline void Assembler::tsuspend_()                              { emit_int32( TSR_OPCODE | rc(1)); }
+inline void Assembler::tresume_()                               { emit_int32( TSR_OPCODE | /*L=1*/ 1u << (31-10) | rc(1)); }
+inline void Assembler::tcheck(int f)                            { emit_int32( TCHECK_OPCODE | bf(f)); }
+
 // ra0 version
 inline void Assembler::lwzx( Register d, Register s2) { emit_int32( LWZX_OPCODE | rt(d) | rb(s2));}
 inline void Assembler::lwz(  Register d, int si16   ) { emit_int32( LWZ_OPCODE  | rt(d) | d1(si16));}

src/cpu/ppc/vm/globalDefinitions_ppc.hpp

@@ -37,6 +37,8 @@ const int StackAlignmentInBytes = 16;
 // signatures accordingly.
 const bool CCallingConventionRequiresIntsAsLongs = true;
 
+#define SUPPORTS_NATIVE_CX8
+
 // The PPC CPUs are NOT multiple-copy-atomic.
 #define CPU_NOT_MULTIPLE_COPY_ATOMIC
 

src/cpu/ppc/vm/interp_masm_ppc_64.cpp

@@ -25,7 +25,6 @@
 
 
 #include "precompiled.hpp"
-#include "asm/assembler.hpp"
 #include "asm/macroAssembler.inline.hpp"
 #include "interp_masm_ppc_64.hpp"
 #include "interpreter/interpreterRuntime.hpp"

src/cpu/ppc/vm/interpreter_ppc.cpp

@@ -24,7 +24,6 @@
  */
 
 #include "precompiled.hpp"
-#include "asm/assembler.hpp"
 #include "asm/macroAssembler.inline.hpp"
 #include "interpreter/bytecodeHistogram.hpp"
 #include "interpreter/interpreter.hpp"

src/cpu/ppc/vm/macroAssembler_ppc.cpp

@@ -2365,7 +2365,7 @@ void MacroAssembler::g1_write_barrier_post(Register Rstore_addr, Register Rnew_v
 #endif // INCLUDE_ALL_GCS
 
 // Values for last_Java_pc, and last_Java_sp must comply to the rules
-// in frame_ppc64.hpp.
+// in frame_ppc.hpp.
 void MacroAssembler::set_last_Java_frame(Register last_Java_sp, Register last_Java_pc) {
   // Always set last_Java_pc and flags first because once last_Java_sp
   // is visible has_last_Java_frame is true and users will look at the
@@ -2492,6 +2492,7 @@ int MacroAssembler::instr_size_for_decode_klass_not_null() {
 }
 
 void MacroAssembler::decode_klass_not_null(Register dst, Register src) {
+  assert(dst != R0, "Dst reg may not be R0, as R0 is used here.");
   if (src == noreg) src = dst;
   Register shifted_src = src;
   if (Universe::narrow_klass_shift() != 0 ||
@@ -2526,14 +2527,11 @@ void MacroAssembler::load_klass_with_trap_null_check(Register dst, Register src)
 
 void MacroAssembler::reinit_heapbase(Register d, Register tmp) {
   if (Universe::heap() != NULL) {
-    if (Universe::narrow_oop_base() == NULL) {
-      Assembler::xorr(R30, R30, R30);
-    } else {
-      load_const(R30, Universe::narrow_ptrs_base(), tmp);
-    }
+    load_const_optimized(R30, Universe::narrow_ptrs_base(), tmp);
   } else {
-    load_const(R30, Universe::narrow_ptrs_base_addr(), tmp);
-    ld(R30, 0, R30);
+    // Heap not yet allocated. Load indirectly.
+    int simm16_offset = load_const_optimized(R30, Universe::narrow_ptrs_base_addr(), tmp, true);
+    ld(R30, simm16_offset, R30);
   }
 }
 

src/cpu/ppc/vm/ppc.ad

@@ -1249,6 +1249,7 @@ EmitCallOffsets emit_call_with_trampoline_stub(MacroAssembler &_masm, address en
 
     // Emit the trampoline stub which will be related to the branch-and-link below.
     CallStubImpl::emit_trampoline_stub(_masm, entry_point_toc_offset, offsets.insts_call_instruction_offset);
+    if (Compile::current()->env()->failing()) { return offsets; } // Code cache may be full.
     __ relocate(rtype);
   }
 
@@ -1412,7 +1413,7 @@ void MachPrologNode::emit(CodeBuffer &cbuf, PhaseRegAlloc *ra_) const {
     while (bang_offset <= bang_end) {
       // Need at least one stack bang at end of shadow zone.
 
-      // Again I had to copy code, this time from assembler_ppc64.cpp,
+      // Again I had to copy code, this time from assembler_ppc.cpp,
       // bang_stack_with_offset - see there for comments.
 
       // Stack grows down, caller passes positive offset.
@@ -2002,7 +2003,7 @@ void MachUEPNode::emit(CodeBuffer &cbuf, PhaseRegAlloc *ra_) const {
 
   // Inline_cache contains a klass.
   Register ic_klass       = as_Register(Matcher::inline_cache_reg_encode());
-  Register receiver_klass = R0;  // tmp
+  Register receiver_klass = R12_scratch2;  // tmp
 
   assert_different_registers(ic_klass, receiver_klass, R11_scratch1, R3_ARG1);
   assert(R11_scratch1 == R11, "need prologue scratch register");
@@ -3486,6 +3487,7 @@ encode %{
 
         // Emit the trampoline stub which will be related to the branch-and-link below.
         CallStubImpl::emit_trampoline_stub(_masm, entry_point_toc_offset, start_offset);
+        if (Compile::current()->env()->failing()) { return; } // Code cache may be full.
         __ relocate(_optimized_virtual ?
                     relocInfo::opt_virtual_call_type : relocInfo::static_call_type);
       }
@@ -3529,6 +3531,7 @@ encode %{
 
       // Emit the trampoline stub which will be related to the branch-and-link below.
       CallStubImpl::emit_trampoline_stub(_masm, entry_point_toc_offset, start_offset);
+      if (ra_->C->env()->failing()) { return; } // Code cache may be full.
       assert(_optimized_virtual, "methodHandle call should be a virtual call");
       __ relocate(relocInfo::opt_virtual_call_type);
     }
@@ -3579,9 +3582,7 @@ encode %{
       const address entry_point_const = __ address_constant(entry_point, RelocationHolder::none);
       const int entry_point_const_toc_offset = __ offset_to_method_toc(entry_point_const);
       CallStubImpl::emit_trampoline_stub(_masm, entry_point_const_toc_offset, __ offset());
-
-      if (ra_->C->env()->failing())
-        return;
+      if (ra_->C->env()->failing()) { return; } // Code cache may be full.
 
       // Build relocation at call site with ic position as data.
       assert((_load_ic_hi_node != NULL && _load_ic_node == NULL) ||
@@ -5640,19 +5641,6 @@ instruct loadNKlass(iRegNdst dst, memory mem) %{
   ins_pipe(pipe_class_memory);
 %}
 
-//// Load compressed klass and decode it if narrow_klass_shift == 0.
-//// TODO: will narrow_klass_shift ever be 0?
-//instruct decodeNKlass2Klass(iRegPdst dst, memory mem) %{
-//  match(Set dst (DecodeNKlass (LoadNKlass mem)));
-//  predicate(false /* TODO: PPC port Universe::narrow_klass_shift() == 0*);
-//  ins_cost(MEMORY_REF_COST);
-//
-//  format %{ "LWZ     $dst, $mem \t// DecodeNKlass (unscaled)" %}
-//  size(4);
-//  ins_encode( enc_lwz(dst, mem) );
-//  ins_pipe(pipe_class_memory);
-//%}
-
 // Load Klass Pointer
 instruct loadKlass(iRegPdst dst, memoryAlg4 mem) %{
   match(Set dst (LoadKlass mem));
@@ -6072,11 +6060,15 @@ instruct loadConN_Ex(iRegNdst dst, immN src) %{
   %}
 %}
 
-instruct loadConNKlass_hi(iRegNdst dst, immNKlass src) %{
+// We have seen a safepoint between the hi and lo parts, and this node was handled
+// as an oop. Therefore this needs a match rule so that build_oop_map knows this is
+// not a narrow oop.
+instruct loadConNKlass_hi(iRegNdst dst, immNKlass_NM src) %{
+  match(Set dst src);
   effect(DEF dst, USE src);
   ins_cost(DEFAULT_COST);
 
-  format %{ "LIS     $dst, $src \t// narrow oop hi" %}
+  format %{ "LIS     $dst, $src \t// narrow klass hi" %}
   size(4);
   ins_encode %{
     // TODO: PPC port $archOpcode(ppc64Opcode_addis);
@@ -6086,6 +6078,21 @@ instruct loadConNKlass_hi(iRegNdst dst, immNKlass src) %{
   ins_pipe(pipe_class_default);
 %}
 
+// As loadConNKlass_hi this must be recognized as narrow klass, not oop!
+instruct loadConNKlass_mask(iRegNdst dst, immNKlass_NM src1, iRegNsrc src2) %{
+  match(Set dst src1);
+  effect(TEMP src2);
+  ins_cost(DEFAULT_COST);
+
+  format %{ "MASK    $dst, $src2, 0xFFFFFFFF" %} // mask
+  size(4);
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_rldicl);
+    __ clrldi($dst$$Register, $src2$$Register, 0x20);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
 // This needs a match rule so that build_oop_map knows this is
 // not a narrow oop.
 instruct loadConNKlass_lo(iRegNdst dst, immNKlass_NM src1, iRegNsrc src2) %{
@@ -6093,10 +6100,10 @@ instruct loadConNKlass_lo(iRegNdst dst, immNKlass_NM src1, iRegNsrc src2) %{
   effect(TEMP src2);
   ins_cost(DEFAULT_COST);
 
-  format %{ "ADDI    $dst, $src1, $src2 \t// narrow oop lo" %}
+  format %{ "ORI    $dst, $src1, $src2 \t// narrow klass lo" %}
   size(4);
   ins_encode %{
-    // TODO: PPC port $archOpcode(ppc64Opcode_addi);
+    // TODO: PPC port $archOpcode(ppc64Opcode_ori);
     intptr_t Csrc = Klass::encode_klass((Klass *)$src1$$constant);
     assert(__ oop_recorder() != NULL, "this assembler needs an OopRecorder");
     int klass_index = __ oop_recorder()->find_index((Klass *)$src1$$constant);
@@ -6127,10 +6134,11 @@ instruct loadConNKlass_Ex(iRegNdst dst, immNKlass src) %{
     MachNode *m2 = m1;
     if (!Assembler::is_uimm((jlong)Klass::encode_klass((Klass *)op_src->constant()), 31)) {
       // Value might be 1-extended. Mask out these bits.
-      m2 = new (C) clearMs32bNode();
+      m2 = new (C) loadConNKlass_maskNode();
       m2->add_req(NULL, m1);
       m2->_opnds[0] = op_dst;
-      m2->_opnds[1] = op_dst;
+      m2->_opnds[1] = op_src;
+      m2->_opnds[2] = op_dst;
       ra_->set_pair(m2->_idx, ra_->get_reg_second(this), ra_->get_reg_first(this));
       nodes->push(m2);
     }
@@ -6975,7 +6983,7 @@ instruct encodePKlass_32GAligned(iRegNdst dst, iRegPsrc src) %{
   size(4);
   ins_encode %{
     // TODO: PPC port $archOpcode(ppc64Opcode_rldicl);
-    __ rldicl($dst$$Register, $src$$Register, 64-Universe::narrow_oop_shift(), 32);
+    __ rldicl($dst$$Register, $src$$Register, 64-Universe::narrow_klass_shift(), 32);
   %}
   ins_pipe(pipe_class_default);
 %}

src/cpu/ppc/vm/stubGenerator_ppc.cpp

@@ -24,7 +24,6 @@
  */
 
 #include "precompiled.hpp"
-#include "asm/assembler.hpp"
 #include "asm/macroAssembler.inline.hpp"
 #include "interpreter/interpreter.hpp"
 #include "nativeInst_ppc.hpp"
@@ -39,9 +38,6 @@
 #include "runtime/stubCodeGenerator.hpp"
 #include "runtime/stubRoutines.hpp"
 #include "utilities/top.hpp"
-#ifdef COMPILER2
-#include "opto/runtime.hpp"
-#endif
 #include "runtime/thread.inline.hpp"
 
 #define __ _masm->
@@ -216,7 +212,7 @@ class StubGenerator: public StubCodeGenerator {
     {
       BLOCK_COMMENT("Call frame manager or native entry.");
       // Call frame manager or native entry.
-      Register r_new_arg_entry = R14; // PPC_state;
+      Register r_new_arg_entry = R14;
       assert_different_registers(r_new_arg_entry, r_top_of_arguments_addr,
                                  r_arg_method, r_arg_thread);
 

src/cpu/ppc/vm/templateTable_ppc_64.cpp

@@ -352,7 +352,6 @@ void TemplateTable::ldc(bool wide) {
   __ sldi(Rscratch1, Rscratch1, LogBytesPerWord);
   __ cmpdi(CCR0, Rscratch2, JVM_CONSTANT_Integer);
   __ bne(CCR0, notInt);
-  __ isync(); // Order load of constant wrt. tags.
   __ lwax(R17_tos, Rcpool, Rscratch1);
   __ push(itos);
   __ b(exit);
@@ -364,7 +363,6 @@ void TemplateTable::ldc(bool wide) {
   __ cmpdi(CCR0, Rscratch2, JVM_CONSTANT_Float);
   __ asm_assert_eq("unexpected type", 0x8765);
 #endif
-  __ isync(); // Order load of constant wrt. tags.
   __ lfsx(F15_ftos, Rcpool, Rscratch1);
   __ push(ftos);
 
@@ -423,13 +421,11 @@ void TemplateTable::ldc2_w() {
   // Check out Conversions.java for an example.
   // Also ConstantPool::header_size() is 20, which makes it very difficult
   // to double-align double on the constant pool. SG, 11/7/97
-  __ isync(); // Order load of constant wrt. tags.
   __ lfdx(F15_ftos, Rcpool, Rindex);
   __ push(dtos);
   __ b(Lexit);
 
   __ bind(Llong);
-  __ isync(); // Order load of constant wrt. tags.
   __ ldx(R17_tos, Rcpool, Rindex);
   __ push(ltos);
 

src/os_cpu/linux_ppc/vm/prefetch_linux_ppc.inline.hpp

@@ -47,4 +47,4 @@ inline void Prefetch::write(void *loc, intx interval) {
     );
 }
 
-#endif // OS_CPU_LINUX_PPC_VM_PREFETCH_LINUX_OJDKPPC_HPP
+#endif // OS_CPU_LINUX_PPC_VM_PREFETCH_LINUX_PPC_INLINE_HPP

src/share/vm/classfile/javaClasses.cpp

@@ -625,6 +625,7 @@ void java_lang_Class::create_mirror(KlassHandle k, Handle class_loader,
     }
 
     // set the classLoader field in the java_lang_Class instance
+    assert(class_loader() == k->class_loader(), "should be same");
     set_class_loader(mirror(), class_loader());
 
     // Setup indirection from klass->mirror last

src/share/vm/classfile/systemDictionary.cpp

@@ -2276,12 +2276,14 @@ methodHandle SystemDictionary::find_method_handle_intrinsic(vmIntrinsics::ID iid
     spe = NULL;
     // Must create lots of stuff here, but outside of the SystemDictionary lock.
     m = Method::make_method_handle_intrinsic(iid, signature, CHECK_(empty));
-    CompileBroker::compile_method(m, InvocationEntryBci, CompLevel_highest_tier,
-                                  methodHandle(), CompileThreshold, "MH", CHECK_(empty));
-    // Check if we need to have compiled code but we don't.
-    if (!Arguments::is_interpreter_only() && !m->has_compiled_code()) {
-      THROW_MSG_(vmSymbols::java_lang_VirtualMachineError(),
-                 "out of space in CodeCache for method handle intrinsic", empty);
+    if (!Arguments::is_interpreter_only()) {
+      // Generate a compiled form of the MH intrinsic.
+      AdapterHandlerLibrary::create_native_wrapper(m);
+      // Check if have the compiled code.
+      if (!m->has_compiled_code()) {
+        THROW_MSG_(vmSymbols::java_lang_VirtualMachineError(),
+                   "out of space in CodeCache for method handle intrinsic", empty);
+      }
     }
     // Now grab the lock.  We might have to throw away the new method,
     // if a racing thread has managed to install one at the same time.

src/share/vm/gc_implementation/g1/concurrentG1Refine.cpp

@@ -128,9 +128,7 @@ void ConcurrentG1Refine::worker_threads_do(ThreadClosure * tc) {
 }
 
 uint ConcurrentG1Refine::thread_num() {
-  uint n_threads = (G1ConcRefinementThreads > 0) ? G1ConcRefinementThreads
-                                                : ParallelGCThreads;
-  return MAX2<uint>(n_threads, 1);
+  return G1ConcRefinementThreads;
 }
 
 void ConcurrentG1Refine::print_worker_threads_on(outputStream* st) const {

src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp

@@ -1483,9 +1483,7 @@ bool G1CollectedHeap::do_collection(bool explicit_gc,
 
       // Discard all rset updates
       JavaThread::dirty_card_queue_set().abandon_logs();
-      assert(!G1DeferredRSUpdate
-             || (G1DeferredRSUpdate &&
-                (dirty_card_queue_set().completed_buffers_num() == 0)), "Should not be any");
+      assert(dirty_card_queue_set().completed_buffers_num() == 0, "DCQS should be empty");
 
       _young_list->reset_sampled_info();
       // At this point there should be no regions in the
@@ -2100,15 +2098,13 @@ jint G1CollectedHeap::initialize() {
                                                 concurrent_g1_refine()->red_zone(),
                                                 Shared_DirtyCardQ_lock);
 
-  if (G1DeferredRSUpdate) {
-    dirty_card_queue_set().initialize(NULL, // Should never be called by the Java code
-                                      DirtyCardQ_CBL_mon,
-                                      DirtyCardQ_FL_lock,
-                                      -1, // never trigger processing
-                                      -1, // no limit on length
-                                      Shared_DirtyCardQ_lock,
-                                      &JavaThread::dirty_card_queue_set());
-  }
+  dirty_card_queue_set().initialize(NULL, // Should never be called by the Java code
+                                    DirtyCardQ_CBL_mon,
+                                    DirtyCardQ_FL_lock,
+                                    -1, // never trigger processing
+                                    -1, // no limit on length
+                                    Shared_DirtyCardQ_lock,
+                                    &JavaThread::dirty_card_queue_set());
 
   // Initialize the card queue set used to hold cards containing
   // references into the collection set.
@@ -5303,7 +5299,6 @@ class G1RedirtyLoggedCardsTask : public AbstractGangTask {
 };
 
 void G1CollectedHeap::redirty_logged_cards() {
-  guarantee(G1DeferredRSUpdate, "Must only be called when using deferred RS updates.");
   double redirty_logged_cards_start = os::elapsedTime();
 
   uint n_workers = (G1CollectedHeap::use_parallel_gc_threads() ?
@@ -5358,9 +5353,10 @@ public:
   void do_oop(narrowOop* p) { guarantee(false, "Not needed"); }
   void do_oop(oop* p) {
     oop obj = *p;
+    assert(obj != NULL, "the caller should have filtered out NULL values");
 
     G1CollectedHeap::in_cset_state_t cset_state = _g1->in_cset_state(obj);
-    if (obj == NULL || cset_state == G1CollectedHeap::InNeither) {
+    if (cset_state == G1CollectedHeap::InNeither) {
       return;
     }
     if (cset_state == G1CollectedHeap::InCSet) {
@@ -5962,9 +5958,7 @@ void G1CollectedHeap::evacuate_collection_set(EvacuationInfo& evacuation_info) {
   // RSets.
   enqueue_discovered_references(n_workers);
 
-  if (G1DeferredRSUpdate) {
-    redirty_logged_cards();
-  }
+  redirty_logged_cards();
   COMPILER2_PRESENT(DerivedPointerTable::update_pointers());
 }
 

src/share/vm/gc_implementation/g1/g1EvacFailure.hpp

@@ -176,15 +176,17 @@ public:
 class RemoveSelfForwardPtrHRClosure: public HeapRegionClosure {
   G1CollectedHeap* _g1h;
   ConcurrentMark* _cm;
-  OopsInHeapRegionClosure *_update_rset_cl;
   uint _worker_id;
 
+  DirtyCardQueue _dcq;
+  UpdateRSetDeferred _update_rset_cl;
+
 public:
   RemoveSelfForwardPtrHRClosure(G1CollectedHeap* g1h,
-                                OopsInHeapRegionClosure* update_rset_cl,
                                 uint worker_id) :
-    _g1h(g1h), _update_rset_cl(update_rset_cl),
-    _worker_id(worker_id), _cm(_g1h->concurrent_mark()) { }
+    _g1h(g1h), _dcq(&g1h->dirty_card_queue_set()), _update_rset_cl(g1h, &_dcq),
+    _worker_id(worker_id), _cm(_g1h->concurrent_mark()) {
+    }
 
   bool doHeapRegion(HeapRegion *hr) {
     bool during_initial_mark = _g1h->g1_policy()->during_initial_mark_pause();
@@ -195,7 +197,7 @@ public:
 
     if (hr->claimHeapRegion(HeapRegion::ParEvacFailureClaimValue)) {
       if (hr->evacuation_failed()) {
-        RemoveSelfForwardPtrObjClosure rspc(_g1h, _cm, hr, _update_rset_cl,
+        RemoveSelfForwardPtrObjClosure rspc(_g1h, _cm, hr, &_update_rset_cl,
                                             during_initial_mark,
                                             during_conc_mark,
                                             _worker_id);
@@ -214,7 +216,7 @@ public:
         // whenever this might be required in the future.
         hr->rem_set()->reset_for_par_iteration();
         hr->reset_bot();
-        _update_rset_cl->set_region(hr);
+        _update_rset_cl.set_region(hr);
         hr->object_iterate(&rspc);
 
         hr->rem_set()->clean_strong_code_roots(hr);
@@ -238,16 +240,7 @@ public:
     _g1h(g1h) { }
 
   void work(uint worker_id) {
-    UpdateRSetImmediate immediate_update(_g1h->g1_rem_set());
-    DirtyCardQueue dcq(&_g1h->dirty_card_queue_set());
-    UpdateRSetDeferred deferred_update(_g1h, &dcq);
-
-    OopsInHeapRegionClosure *update_rset_cl = &deferred_update;
-    if (!G1DeferredRSUpdate) {
-      update_rset_cl = &immediate_update;
-    }
-
-    RemoveSelfForwardPtrHRClosure rsfp_cl(_g1h, update_rset_cl, worker_id);
+    RemoveSelfForwardPtrHRClosure rsfp_cl(_g1h, worker_id);
 
     HeapRegion* hr = _g1h->start_cset_region_for_worker(worker_id);
     _g1h->collection_set_iterate_from(hr, &rsfp_cl);

src/share/vm/gc_implementation/g1/g1GCPhaseTimes.cpp

@@ -236,10 +236,8 @@ void G1GCPhaseTimes::note_gc_end() {
   _last_gc_worker_times_ms.verify();
   _last_gc_worker_other_times_ms.verify();
 
-  if (G1DeferredRSUpdate) {
-    _last_redirty_logged_cards_time_ms.verify();
-    _last_redirty_logged_cards_processed_cards.verify();
-  }
+  _last_redirty_logged_cards_time_ms.verify();
+  _last_redirty_logged_cards_processed_cards.verify();
 }
 
 void G1GCPhaseTimes::note_string_dedup_fixup_start() {
@@ -351,12 +349,10 @@ void G1GCPhaseTimes::print(double pause_time_sec) {
     _recorded_non_young_cset_choice_time_ms));
   print_stats(2, "Ref Proc", _cur_ref_proc_time_ms);
   print_stats(2, "Ref Enq", _cur_ref_enq_time_ms);
-  if (G1DeferredRSUpdate) {
-    print_stats(2, "Redirty Cards", _recorded_redirty_logged_cards_time_ms);
-    if (G1Log::finest()) {
-      _last_redirty_logged_cards_time_ms.print(3, "Parallel Redirty");
-      _last_redirty_logged_cards_processed_cards.print(3, "Redirtied Cards");
-    }
+  print_stats(2, "Redirty Cards", _recorded_redirty_logged_cards_time_ms);
+  if (G1Log::finest()) {
+    _last_redirty_logged_cards_time_ms.print(3, "Parallel Redirty");
+    _last_redirty_logged_cards_processed_cards.print(3, "Redirtied Cards");
   }
   if (G1ReclaimDeadHumongousObjectsAtYoungGC) {
     print_stats(2, "Humongous Reclaim", _cur_fast_reclaim_humongous_time_ms);

src/share/vm/gc_implementation/g1/g1ParScanThreadState.hpp

@@ -82,20 +82,6 @@ class G1ParScanThreadState : public StackObj {
   DirtyCardQueue& dirty_card_queue()             { return _dcq;  }
   G1SATBCardTableModRefBS* ctbs()                { return _ct_bs; }
 
-  template <class T> inline void immediate_rs_update(HeapRegion* from, T* p, int tid);
-
-  template <class T> void deferred_rs_update(HeapRegion* from, T* p, int tid) {
-    // If the new value of the field points to the same region or
-    // is the to-space, we don't need to include it in the Rset updates.
-    if (!from->is_in_reserved(oopDesc::load_decode_heap_oop(p)) && !from->is_survivor()) {
-      size_t card_index = ctbs()->index_for(p);
-      // If the card hasn't been added to the buffer, do it.
-      if (ctbs()->mark_card_deferred(card_index)) {
-        dirty_card_queue().enqueue((jbyte*)ctbs()->byte_for_index(card_index));
-      }
-    }
-  }
-
  public:
   G1ParScanThreadState(G1CollectedHeap* g1h, uint queue_num, ReferenceProcessor* rp);
   ~G1ParScanThreadState();
@@ -115,8 +101,17 @@ class G1ParScanThreadState : public StackObj {
     _refs->push(ref);
   }
 
-  template <class T> inline void update_rs(HeapRegion* from, T* p, int tid);
-
+  template <class T> void update_rs(HeapRegion* from, T* p, int tid) {
+    // If the new value of the field points to the same region or
+    // is the to-space, we don't need to include it in the Rset updates.
+    if (!from->is_in_reserved(oopDesc::load_decode_heap_oop(p)) && !from->is_survivor()) {
+      size_t card_index = ctbs()->index_for(p);
+      // If the card hasn't been added to the buffer, do it.
+      if (ctbs()->mark_card_deferred(card_index)) {
+        dirty_card_queue().enqueue((jbyte*)ctbs()->byte_for_index(card_index));
+      }
+   }
+  }
  public:
 
   void set_evac_failure_closure(OopsInHeapRegionClosure* evac_failure_cl) {

src/share/vm/gc_implementation/g1/g1ParScanThreadState.inline.hpp

@@ -29,20 +29,6 @@
 #include "gc_implementation/g1/g1RemSet.inline.hpp"
 #include "oops/oop.inline.hpp"
 
-template <class T> inline void G1ParScanThreadState::immediate_rs_update(HeapRegion* from, T* p, int tid) {
-  if (!from->is_survivor()) {
-    _g1_rem->par_write_ref(from, p, tid);
-  }
-}
-
-template <class T> void G1ParScanThreadState::update_rs(HeapRegion* from, T* p, int tid) {
-  if (G1DeferredRSUpdate) {
-    deferred_rs_update(from, p, tid);
-  } else {
-    immediate_rs_update(from, p, tid);
-  }
-}
-
 template <class T> void G1ParScanThreadState::do_oop_evac(T* p, HeapRegion* from) {
   assert(!oopDesc::is_null(oopDesc::load_decode_heap_oop(p)),
          "Reference should not be NULL here as such are never pushed to the task queue.");

src/share/vm/gc_implementation/g1/g1RemSet.cpp

@@ -338,12 +338,8 @@ void G1RemSet::oops_into_collection_set_do(OopsInHeapRegionClosure* oc,
   // are just discarded (there's no need to update the RSets of regions
   // that were in the collection set - after the pause these regions
   // are wholly 'free' of live objects. In the event of an evacuation
-  // failure the cards/buffers in this queue set are:
-  // * passed to the DirtyCardQueueSet that is used to manage deferred
-  //   RSet updates, or
-  // * scanned for references that point into the collection set
-  //   and the RSet of the corresponding region in the collection set
-  //   is updated immediately.
+  // failure the cards/buffers in this queue set are passed to the
+  // DirtyCardQueueSet that is used to manage RSet updates
   DirtyCardQueue into_cset_dcq(&_g1->into_cset_dirty_card_queue_set());
 
   assert((ParallelGCThreads > 0) || worker_i == 0, "invariant");
@@ -372,7 +368,6 @@ void G1RemSet::oops_into_collection_set_do(OopsInHeapRegionClosure* oc,
 
 void G1RemSet::prepare_for_oops_into_collection_set_do() {
   cleanupHRRS();
-  ConcurrentG1Refine* cg1r = _g1->concurrent_g1_refine();
   _g1->set_refine_cte_cl_concurrency(false);
   DirtyCardQueueSet& dcqs = JavaThread::dirty_card_queue_set();
   dcqs.concatenate_logs();
@@ -385,66 +380,6 @@ void G1RemSet::prepare_for_oops_into_collection_set_do() {
   _total_cards_scanned = 0;
 }
 
-
-// This closure, applied to a DirtyCardQueueSet, is used to immediately
-// update the RSets for the regions in the CSet. For each card it iterates
-// through the oops which coincide with that card. It scans the reference
-// fields in each oop; when it finds an oop that points into the collection
-// set, the RSet for the region containing the referenced object is updated.
-class UpdateRSetCardTableEntryIntoCSetClosure: public CardTableEntryClosure {
-  G1CollectedHeap* _g1;
-  CardTableModRefBS* _ct_bs;
-public:
-  UpdateRSetCardTableEntryIntoCSetClosure(G1CollectedHeap* g1,
-                                          CardTableModRefBS* bs):
-    _g1(g1), _ct_bs(bs)
-  { }
-
-  bool do_card_ptr(jbyte* card_ptr, uint worker_i) {
-    // Construct the region representing the card.
-    HeapWord* start = _ct_bs->addr_for(card_ptr);
-    // And find the region containing it.
-    HeapRegion* r = _g1->heap_region_containing(start);
-
-    // Scan oops in the card looking for references into the collection set
-    // Don't use addr_for(card_ptr + 1) which can ask for
-    // a card beyond the heap.  This is not safe without a perm
-    // gen.
-    HeapWord* end   = start + CardTableModRefBS::card_size_in_words;
-    MemRegion scanRegion(start, end);
-
-    UpdateRSetImmediate update_rs_cl(_g1->g1_rem_set());
-    FilterIntoCSClosure update_rs_cset_oop_cl(NULL, _g1, &update_rs_cl);
-    FilterOutOfRegionClosure filter_then_update_rs_cset_oop_cl(r, &update_rs_cset_oop_cl);
-
-    // We can pass false as the "filter_young" parameter here as:
-    // * we should be in a STW pause,
-    // * the DCQS to which this closure is applied is used to hold
-    //   references that point into the collection set from the prior
-    //   RSet updating,
-    // * the post-write barrier shouldn't be logging updates to young
-    //   regions (but there is a situation where this can happen - see
-    //   the comment in G1RemSet::refine_card() below -
-    //   that should not be applicable here), and
-    // * during actual RSet updating, the filtering of cards in young
-    //   regions in HeapRegion::oops_on_card_seq_iterate_careful is
-    //   employed.
-    // As a result, when this closure is applied to "refs into cset"
-    // DCQS, we shouldn't see any cards in young regions.
-    update_rs_cl.set_region(r);
-    HeapWord* stop_point =
-      r->oops_on_card_seq_iterate_careful(scanRegion,
-                                          &filter_then_update_rs_cset_oop_cl,
-                                          false /* filter_young */,
-                                          NULL  /* card_ptr */);
-
-    // Since this is performed in the event of an evacuation failure, we
-    // we shouldn't see a non-null stop point
-    assert(stop_point == NULL, "saw an unallocated region");
-    return true;
-  }
-};
-
 void G1RemSet::cleanup_after_oops_into_collection_set_do() {
   guarantee( _cards_scanned != NULL, "invariant" );
   _total_cards_scanned = 0;
@@ -465,25 +400,10 @@ void G1RemSet::cleanup_after_oops_into_collection_set_do() {
     double restore_remembered_set_start = os::elapsedTime();
 
     // Restore remembered sets for the regions pointing into the collection set.
-    if (G1DeferredRSUpdate) {
-      // If deferred RS updates are enabled then we just need to transfer
-      // the completed buffers from (a) the DirtyCardQueueSet used to hold
-      // cards that contain references that point into the collection set
-      // to (b) the DCQS used to hold the deferred RS updates
-      _g1->dirty_card_queue_set().merge_bufferlists(&into_cset_dcqs);
-    } else {
-
-      CardTableModRefBS* bs = (CardTableModRefBS*)_g1->barrier_set();
-      UpdateRSetCardTableEntryIntoCSetClosure update_rs_cset_immediate(_g1, bs);
-
-      int n_completed_buffers = 0;
-      while (into_cset_dcqs.apply_closure_to_completed_buffer(&update_rs_cset_immediate,
-                                                    0, 0, true)) {
-        n_completed_buffers++;
-      }
-      assert(n_completed_buffers == into_cset_n_buffers, "missed some buffers");
-    }
-
+    // We just need to transfer the completed buffers from the DirtyCardQueueSet
+    // used to hold cards that contain references that point into the collection set
+    // to the DCQS used to hold the deferred RS updates.
+    _g1->dirty_card_queue_set().merge_bufferlists(&into_cset_dcqs);
     _g1->g1_policy()->phase_times()->record_evac_fail_restore_remsets((os::elapsedTime() - restore_remembered_set_start) * 1000.0);
   }
 

src/share/vm/gc_implementation/g1/g1RemSet.hpp

@@ -193,18 +193,4 @@ public:
   bool apply_to_weak_ref_discovered_field() { return true; }
 };
 
-class UpdateRSetImmediate: public OopsInHeapRegionClosure {
-private:
-  G1RemSet* _g1_rem_set;
-
-  template <class T> void do_oop_work(T* p);
-public:
-  UpdateRSetImmediate(G1RemSet* rs) :
-    _g1_rem_set(rs) {}
-
-  virtual void do_oop(narrowOop* p) { do_oop_work(p); }
-  virtual void do_oop(      oop* p) { do_oop_work(p); }
-};
-
-
 #endif // SHARE_VM_GC_IMPLEMENTATION_G1_G1REMSET_HPP

src/share/vm/gc_implementation/g1/g1RemSet.inline.hpp

@@ -79,13 +79,4 @@ inline void UpdateRSOopClosure::do_oop_work(T* p) {
   _rs->par_write_ref(_from, p, _worker_i);
 }
 
-template <class T>
-inline void UpdateRSetImmediate::do_oop_work(T* p) {
-  assert(_from->is_in_reserved(p), "paranoia");
-  T heap_oop = oopDesc::load_heap_oop(p);
-  if (!oopDesc::is_null(heap_oop) && !_from->is_survivor()) {
-    _g1_rem_set->par_write_ref(_from, p, 0);
-  }
-}
-
 #endif // SHARE_VM_GC_IMPLEMENTATION_G1_G1REMSET_INLINE_HPP

src/share/vm/gc_implementation/g1/g1_globals.hpp

@@ -108,9 +108,6 @@
   develop(bool, G1RSBarrierRegionFilter, true,                              \
           "If true, generate region filtering code in RS barrier")          \
                                                                             \
-  develop(bool, G1DeferredRSUpdate, true,                                   \
-          "If true, use deferred RS updates")                               \
-                                                                            \
   develop(bool, G1RSLogCheckCardTable, false,                               \
           "If true, verify that no dirty cards remain after RS log "        \
           "processing.")                                                    \

src/share/vm/interpreter/oopMapCache.cpp

@@ -180,7 +180,7 @@ InterpreterOopMap::~InterpreterOopMap() {
   }
 }
 
-bool InterpreterOopMap::is_empty() {
+bool InterpreterOopMap::is_empty() const {
   bool result = _method == NULL;
   assert(_method != NULL || (_bci == 0 &&
     (_mask_size == 0 || _mask_size == USHRT_MAX) &&
@@ -196,7 +196,7 @@ void InterpreterOopMap::initialize() {
   for (int i = 0; i < N; i++) _bit_mask[i] = 0;
 }
 
-void InterpreterOopMap::iterate_oop(OffsetClosure* oop_closure) {
+void InterpreterOopMap::iterate_oop(OffsetClosure* oop_closure) const {
   int n = number_of_entries();
   int word_index = 0;
   uintptr_t value = 0;
@@ -238,7 +238,7 @@ void InterpreterOopMap::iterate_all(OffsetClosure* oop_closure, OffsetClosure* v
 #endif
 
 
-void InterpreterOopMap::print() {
+void InterpreterOopMap::print() const {
   int n = number_of_entries();
   tty->print("oop map for ");
   method()->print_value();
@@ -469,7 +469,7 @@ void InterpreterOopMap::resource_copy(OopMapCacheEntry* from) {
   }
 }
 
-inline unsigned int OopMapCache::hash_value_for(methodHandle method, int bci) {
+inline unsigned int OopMapCache::hash_value_for(methodHandle method, int bci) const {
   // We use method->code_size() rather than method->identity_hash() below since
   // the mark may not be present if a pointer to the method is already reversed.
   return   ((unsigned int) bci)
@@ -522,7 +522,7 @@ void OopMapCache::flush_obsolete_entries() {
 
 void OopMapCache::lookup(methodHandle method,
                          int bci,
-                         InterpreterOopMap* entry_for) {
+                         InterpreterOopMap* entry_for) const {
   MutexLocker x(&_mut);
 
   OopMapCacheEntry* entry = NULL;

src/share/vm/interpreter/oopMapCache.hpp

@@ -101,32 +101,31 @@ class InterpreterOopMap: ResourceObj {
 
   // access methods
   Method*        method() const                  { return _method; }
-  void           set_method(Method* v)         { _method = v; }
+  void           set_method(Method* v)           { _method = v; }
   int            bci() const                     { return _bci; }
   void           set_bci(int v)                  { _bci = v; }
   int            mask_size() const               { return _mask_size; }
   void           set_mask_size(int v)            { _mask_size = v; }
-  int            number_of_entries() const       { return mask_size() / bits_per_entry; }
   // Test bit mask size and return either the in-line bit mask or allocated
   // bit mask.
-  uintptr_t*  bit_mask()                         { return (uintptr_t*)(mask_size() <= small_mask_limit ? (intptr_t)_bit_mask : _bit_mask[0]); }
+  uintptr_t*  bit_mask() const                   { return (uintptr_t*)(mask_size() <= small_mask_limit ? (intptr_t)_bit_mask : _bit_mask[0]); }
 
   // return the word size of_bit_mask.  mask_size() <= 4 * MAX_USHORT
-  size_t mask_word_size() {
+  size_t mask_word_size() const {
     return (mask_size() + BitsPerWord - 1) / BitsPerWord;
   }
 
-  uintptr_t entry_at(int offset)            { int i = offset * bits_per_entry; return bit_mask()[i / BitsPerWord] >> (i % BitsPerWord); }
+  uintptr_t entry_at(int offset) const           { int i = offset * bits_per_entry; return bit_mask()[i / BitsPerWord] >> (i % BitsPerWord); }
 
-  void set_expression_stack_size(int sz)    { _expression_stack_size = sz; }
+  void set_expression_stack_size(int sz)         { _expression_stack_size = sz; }
 
 #ifdef ENABLE_ZAP_DEAD_LOCALS
-  bool is_dead(int offset)                       { return (entry_at(offset) & (1 << dead_bit_number)) != 0; }
+  bool is_dead(int offset) const                 { return (entry_at(offset) & (1 << dead_bit_number)) != 0; }
 #endif
 
   // Lookup
-  bool match(methodHandle method, int bci)       { return _method == method() && _bci == bci; }
-  bool is_empty();
+  bool match(methodHandle method, int bci) const { return _method == method() && _bci == bci; }
+  bool is_empty() const;
 
   // Initialization
   void initialize();
@@ -141,12 +140,13 @@ class InterpreterOopMap: ResourceObj {
   // in-line), allocate the space from a Resource area.
   void resource_copy(OopMapCacheEntry* from);
 
-  void iterate_oop(OffsetClosure* oop_closure);
-  void print();
+  void iterate_oop(OffsetClosure* oop_closure) const;
+  void print() const;
 
-  bool is_oop  (int offset)                      { return (entry_at(offset) & (1 << oop_bit_number )) != 0; }
+  int number_of_entries() const                  { return mask_size() / bits_per_entry; }
+  bool is_oop (int offset) const                 { return (entry_at(offset) & (1 << oop_bit_number )) != 0; }
 
-  int expression_stack_size()                    { return _expression_stack_size; }
+  int expression_stack_size() const              { return _expression_stack_size; }
 
 #ifdef ENABLE_ZAP_DEAD_LOCALS
   void iterate_all(OffsetClosure* oop_closure, OffsetClosure* value_closure, OffsetClosure* dead_closure);
@@ -161,10 +161,10 @@ class OopMapCache : public CHeapObj<mtClass> {
 
   OopMapCacheEntry* _array;
 
-  unsigned int hash_value_for(methodHandle method, int bci);
+  unsigned int hash_value_for(methodHandle method, int bci) const;
   OopMapCacheEntry* entry_at(int i) const;
 
-  Mutex _mut;
+  mutable Mutex _mut;
 
   void flush();
 
@@ -177,7 +177,7 @@ class OopMapCache : public CHeapObj<mtClass> {
 
   // Returns the oopMap for (method, bci) in parameter "entry".
   // Returns false if an oop map was not found.
-  void lookup(methodHandle method, int bci, InterpreterOopMap* entry);
+  void lookup(methodHandle method, int bci, InterpreterOopMap* entry) const;
 
   // Compute an oop map without updating the cache or grabbing any locks (for debugging)
   static void compute_one_oop_map(methodHandle method, int bci, InterpreterOopMap* entry);

src/share/vm/oops/arrayKlass.cpp

@@ -100,7 +100,7 @@ void ArrayKlass::complete_create_array_klass(ArrayKlass* k, KlassHandle super_kl
   ResourceMark rm(THREAD);
   k->initialize_supers(super_klass(), CHECK);
   k->vtable()->initialize_vtable(false, CHECK);
-  java_lang_Class::create_mirror(k, Handle(NULL), Handle(NULL), CHECK);
+  java_lang_Class::create_mirror(k, Handle(THREAD, k->class_loader()), Handle(NULL), CHECK);
 }
 
 GrowableArray<Klass*>* ArrayKlass::compute_secondary_supers(int num_extra_slots) {

src/share/vm/runtime/arguments.cpp

@@ -1646,6 +1646,12 @@ void Arguments::set_g1_gc_flags() {
                      Abstract_VM_Version::parallel_worker_threads());
   }
 
+#if INCLUDE_ALL_GCS
+  if (G1ConcRefinementThreads == 0) {
+    FLAG_SET_DEFAULT(G1ConcRefinementThreads, ParallelGCThreads);
+  }
+#endif
+
   // MarkStackSize will be set (if it hasn't been set by the user)
   // when concurrent marking is initialized.
   // Its value will be based upon the number of parallel marking threads.

src/share/vm/runtime/deoptimization.hpp

@@ -60,8 +60,8 @@ class Deoptimization : AllStatic {
     Reason_predicate,             // compiler generated predicate failed
     Reason_loop_limit_check,      // compiler generated loop limits check failed
     Reason_speculate_class_check, // saw unexpected object class from type speculation
-    Reason_unstable_if,           // a branch predicted always false was taken
     Reason_rtm_state_change,      // rtm state change detected
+    Reason_unstable_if,           // a branch predicted always false was taken
     Reason_LIMIT,
     // Note:  Keep this enum in sync. with _trap_reason_name.
     Reason_RECORDED_LIMIT = Reason_bimorphic  // some are not recorded per bc

src/share/vm/runtime/vframe.cpp

@@ -260,65 +260,156 @@ Method* interpretedVFrame::method() const {
   return fr().interpreter_frame_method();
 }
 
-StackValueCollection* interpretedVFrame::locals() const {
-  int length = method()->max_locals();
+static StackValue* create_stack_value_from_oop_map(const InterpreterOopMap& oop_mask,
+                                                   int index,
+                                                   const intptr_t* const addr) {
+
+  assert(index >= 0 &&
+         index < oop_mask.number_of_entries(), "invariant");
+
+  // categorize using oop_mask
+  if (oop_mask.is_oop(index)) {
+    // reference (oop) "r"
+    Handle h(addr != NULL ? (*(oop*)addr) : (oop)NULL);
+    return new StackValue(h);
+  }
+  // value (integer) "v"
+  return new StackValue(addr != NULL ? *addr : 0);
+}
+
+static bool is_in_expression_stack(const frame& fr, const intptr_t* const addr) {
+  assert(addr != NULL, "invariant");
+
+  // Ensure to be 'inside' the expresion stack (i.e., addr >= sp for Intel).
+  // In case of exceptions, the expression stack is invalid and the sp
+  // will be reset to express this condition.
+  if (frame::interpreter_frame_expression_stack_direction() > 0) {
+    return addr <= fr.interpreter_frame_tos_address();
+  }
+
+  return addr >= fr.interpreter_frame_tos_address();
+}
+
+static void stack_locals(StackValueCollection* result,
+                         int length,
+                         const InterpreterOopMap& oop_mask,
+                         const frame& fr) {
+
+  assert(result != NULL, "invariant");
+
+  for (int i = 0; i < length; ++i) {
+    const intptr_t* const addr = fr.interpreter_frame_local_at(i);
+    assert(addr != NULL, "invariant");
+    assert(addr >= fr.sp(), "must be inside the frame");
+
+    StackValue* const sv = create_stack_value_from_oop_map(oop_mask, i, addr);
+    assert(sv != NULL, "sanity check");
+
+    result->add(sv);
+  }
+}
+
+static void stack_expressions(StackValueCollection* result,
+                              int length,
+                              int max_locals,
+                              const InterpreterOopMap& oop_mask,
+                              const frame& fr) {
+
+  assert(result != NULL, "invariant");
 
-  if (method()->is_native()) {
-    // If the method is native, max_locals is not telling the truth.
-    // maxlocals then equals the size of parameters
-    length = method()->size_of_parameters();
+  for (int i = 0; i < length; ++i) {
+    const intptr_t* addr = fr.interpreter_frame_expression_stack_at(i);
+    assert(addr != NULL, "invariant");
+    if (!is_in_expression_stack(fr, addr)) {
+      // Need to ensure no bogus escapes.
+      addr = NULL;
+    }
+
+    StackValue* const sv = create_stack_value_from_oop_map(oop_mask,
+                                                           i + max_locals,
+                                                           addr);
+    assert(sv != NULL, "sanity check");
+
+    result->add(sv);
   }
+}
 
-  StackValueCollection* result = new StackValueCollection(length);
+StackValueCollection* interpretedVFrame::locals() const {
+  return stack_data(false);
+}
+
+StackValueCollection* interpretedVFrame::expressions() const {
+  return stack_data(true);
+}
+
+/*
+ * Worker routine for fetching references and/or values
+ * for a particular bci in the interpretedVFrame.
+ *
+ * Returns data for either "locals" or "expressions",
+ * using bci relative oop_map (oop_mask) information.
+ *
+ * @param expressions  bool switch controlling what data to return
+                       (false == locals / true == expressions)
+ *
+ */
+StackValueCollection* interpretedVFrame::stack_data(bool expressions) const {
 
-  // Get oopmap describing oops and int for current bci
   InterpreterOopMap oop_mask;
+  // oopmap for current bci
   if (TraceDeoptimization && Verbose) {
-    methodHandle m_h(thread(), method());
+    methodHandle m_h(Thread::current(), method());
     OopMapCache::compute_one_oop_map(m_h, bci(), &oop_mask);
   } else {
     method()->mask_for(bci(), &oop_mask);
   }
-  // handle locals
-  for(int i=0; i < length; i++) {
-    // Find stack location
-    intptr_t *addr = locals_addr_at(i);
 
-    // Depending on oop/int put it in the right package
-    StackValue *sv;
-    if (oop_mask.is_oop(i)) {
-      // oop value
-      Handle h(*(oop *)addr);
-      sv = new StackValue(h);
-    } else {
-      // integer
-      sv = new StackValue(*addr);
-    }
-    assert(sv != NULL, "sanity check");
-    result->add(sv);
+  const int mask_len = oop_mask.number_of_entries();
+
+  // If the method is native, method()->max_locals() is not telling the truth.
+  // For our purposes, max locals instead equals the size of parameters.
+  const int max_locals = method()->is_native() ?
+    method()->size_of_parameters() : method()->max_locals();
+
+  assert(mask_len >= max_locals, "invariant");
+
+  const int length = expressions ? mask_len - max_locals : max_locals;
+  assert(length >= 0, "invariant");
+
+  StackValueCollection* const result = new StackValueCollection(length);
+
+  if (0 == length) {
+    return result;
   }
+
+  if (expressions) {
+    stack_expressions(result, length, max_locals, oop_mask, fr());
+  } else {
+    stack_locals(result, length, oop_mask, fr());
+  }
+
+  assert(length == result->size(), "invariant");
+
   return result;
 }
 
 void interpretedVFrame::set_locals(StackValueCollection* values) const {
   if (values == NULL || values->size() == 0) return;
 
-  int length = method()->max_locals();
-  if (method()->is_native()) {
-    // If the method is native, max_locals is not telling the truth.
-    // maxlocals then equals the size of parameters
-    length = method()->size_of_parameters();
-  }
+  // If the method is native, max_locals is not telling the truth.
+  // maxlocals then equals the size of parameters
+  const int max_locals = method()->is_native() ?
+    method()->size_of_parameters() : method()->max_locals();
 
-  assert(length == values->size(), "Mismatch between actual stack format and supplied data");
+  assert(max_locals == values->size(), "Mismatch between actual stack format and supplied data");
 
   // handle locals
-  for (int i = 0; i < length; i++) {
+  for (int i = 0; i < max_locals; i++) {
     // Find stack location
     intptr_t *addr = locals_addr_at(i);
 
     // Depending on oop/int put it in the right package
-    StackValue *sv = values->at(i);
+    const StackValue* const sv = values->at(i);
     assert(sv != NULL, "sanity check");
     if (sv->type() == T_OBJECT) {
       *(oop *) addr = (sv->get_obj())();
@@ -328,46 +419,6 @@ void interpretedVFrame::set_locals(StackValueCollection* values) const {
   }
 }
 
-StackValueCollection*  interpretedVFrame::expressions() const {
-  int length = fr().interpreter_frame_expression_stack_size();
-  if (method()->is_native()) {
-    // If the method is native, there is no expression stack
-    length = 0;
-  }
-
-  int nof_locals = method()->max_locals();
-  StackValueCollection* result = new StackValueCollection(length);
-
-  InterpreterOopMap oop_mask;
-  // Get oopmap describing oops and int for current bci
-  if (TraceDeoptimization && Verbose) {
-    methodHandle m_h(method());
-    OopMapCache::compute_one_oop_map(m_h, bci(), &oop_mask);
-  } else {
-    method()->mask_for(bci(), &oop_mask);
-  }
-  // handle expressions
-  for(int i=0; i < length; i++) {
-    // Find stack location
-    intptr_t *addr = fr().interpreter_frame_expression_stack_at(i);
-
-    // Depending on oop/int put it in the right package
-    StackValue *sv;
-    if (oop_mask.is_oop(i + nof_locals)) {
-      // oop value
-      Handle h(*(oop *)addr);
-      sv = new StackValue(h);
-    } else {
-      // integer
-      sv = new StackValue(*addr);
-    }
-    assert(sv != NULL, "sanity check");
-    result->add(sv);
-  }
-  return result;
-}
-
-
 // ------------- cChunk --------------
 
 entryVFrame::entryVFrame(const frame* fr, const RegisterMap* reg_map, JavaThread* thread)

src/share/vm/runtime/vframe.hpp

@@ -186,7 +186,7 @@ class interpretedVFrame: public javaVFrame {
  private:
   static const int bcp_offset;
   intptr_t* locals_addr_at(int offset) const;
-
+  StackValueCollection* stack_data(bool expressions) const;
   // returns where the parameters starts relative to the frame pointer
   int start_of_parameters() const;
 

src/share/vm/services/mallocTracker.cpp

@@ -140,11 +140,6 @@ void* MallocTracker::record_malloc(void* malloc_base, size_t size, MEMFLAGS flag
     return NULL;
   }
 
-  // Check malloc size, size has to <= MAX_MALLOC_SIZE. This is only possible on 32-bit
-  // systems, when malloc size >= 1GB, but is is safe to assume it won't happen.
-  if (size > MAX_MALLOC_SIZE) {
-    fatal("Should not use malloc for big memory block, use virtual memory instead");
-  }
   // Uses placement global new operator to initialize malloc header
   switch(level) {
     case NMT_off:
@@ -154,10 +149,12 @@ void* MallocTracker::record_malloc(void* malloc_base, size_t size, MEMFLAGS flag
       break;
     }
     case NMT_summary: {
+      assert(size <= MAX_MALLOC_SIZE, "malloc size overrun for NMT");
       header = ::new (malloc_base) MallocHeader(size, flags);
       break;
     }
     case NMT_detail: {
+      assert(size <= MAX_MALLOC_SIZE, "malloc size overrun for NMT");
       header = ::new (malloc_base) MallocHeader(size, flags, stack);
       break;
     }

test/gc/arguments/TestG1ConcRefinementThreads.java

+/*
+* 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 TestG1ConcRefinementThreads
+ * @key gc
+ * @bug 8047976
+ * @summary Tests argument processing for G1ConcRefinementThreads
+ * @library /testlibrary
+ */
+
+import com.oracle.java.testlibrary.*;
+import java.util.*;
+import java.util.regex.*;
+
+public class TestG1ConcRefinementThreads {
+
+  static final int AUTO_SELECT_THREADS_COUNT = 0;
+  static final int PASSED_THREADS_COUNT = 11;
+
+  public static void main(String args[]) throws Exception {
+    // default case
+    runG1ConcRefinementThreadsTest(
+        new String[]{}, // automatically selected
+        AUTO_SELECT_THREADS_COUNT /* use default setting */);
+
+    // zero setting case
+    runG1ConcRefinementThreadsTest(
+        new String[]{"-XX:G1ConcRefinementThreads=0"}, // automatically selected
+        AUTO_SELECT_THREADS_COUNT /* set to zero */);
+
+    // non-zero sestting case
+    runG1ConcRefinementThreadsTest(
+        new String[]{"-XX:G1ConcRefinementThreads="+Integer.toString(PASSED_THREADS_COUNT)},
+        PASSED_THREADS_COUNT);
+  }
+
+  private static void runG1ConcRefinementThreadsTest(String[] passedOpts,
+          int expectedValue) throws Exception {
+    List<String> vmOpts = new ArrayList<>();
+    if (passedOpts.length > 0) {
+      Collections.addAll(vmOpts, passedOpts);
+    }
+    Collections.addAll(vmOpts, "-XX:+UseG1GC", "-XX:+PrintFlagsFinal", "-version");
+
+    ProcessBuilder pb = ProcessTools.createJavaProcessBuilder(vmOpts.toArray(new String[vmOpts.size()]));
+    OutputAnalyzer output = new OutputAnalyzer(pb.start());
+
+    output.shouldHaveExitValue(0);
+    String stdout = output.getStdout();
+    checkG1ConcRefinementThreadsConsistency(stdout, expectedValue);
+  }
+
+  private static void checkG1ConcRefinementThreadsConsistency(String output, int expectedValue) {
+    int actualValue = getIntValue("G1ConcRefinementThreads", output);
+
+    if (expectedValue == 0) {
+      // If expectedValue is automatically selected, set it same as ParallelGCThreads.
+      expectedValue = getIntValue("ParallelGCThreads", output);
+    }
+
+    if (expectedValue != actualValue) {
+      throw new RuntimeException(
+            "Actual G1ConcRefinementThreads(" + Integer.toString(actualValue)
+            + ") is not equal to expected value(" + Integer.toString(expectedValue) + ")");
+    }
+  }
+
+  public static int getIntValue(String flag, String where) {
+    Matcher m = Pattern.compile(flag + "\\s+:?=\\s+\\d+").matcher(where);
+    if (!m.find()) {
+      throw new RuntimeException("Could not find value for flag " + flag + " in output string");
+    }
+    String match = m.group();
+    return Integer.parseInt(match.substring(match.lastIndexOf(" ") + 1, match.length()));
+  }
+}

test/runtime/LoadClass/ShowClassLoader.java

+/*
+ * 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
+ * @key regression
+ * @bug 8058927
+ * @summary Make sure array class has the right class loader
+ * @run main ShowClassLoader
+ */
+
+public class ShowClassLoader {
+
+    public static void main(String[] args) {
+        Object[] oa = new Object[0];
+        ShowClassLoader[] sa = new ShowClassLoader[0];
+
+        System.out.println("Classloader for Object[] is " + oa.getClass().getClassLoader());
+        System.out.println("Classloader for SCL[] is " + sa.getClass().getClassLoader() );
+
+        if (sa.getClass().getClassLoader() == null) {
+            throw new RuntimeException("Wrong class loader");
+        }
+    }
+}

test/gc/g1/TestDeferredRSUpdate.java → test/runtime/NMT/UnsafeMallocLimit2.java

@@ -22,58 +22,33 @@
  */
 
 /*
- * @test TestDeferredRSUpdate
- * @bug 8040977 8052170
- * @summary Ensure that running with -XX:-G1DeferredRSUpdate does not crash the VM
- * @key gc
+ * @test
+ * @bug 8058818
  * @library /testlibrary
+ * @build UnsafeMallocLimit2
+ * @run main/othervm -Xmx32m -XX:NativeMemoryTracking=off UnsafeMallocLimit2
  */
 
-import com.oracle.java.testlibrary.ProcessTools;
-import com.oracle.java.testlibrary.OutputAnalyzer;
-
-public class TestDeferredRSUpdate {
-  public static void main(String[] args) throws Exception {
-    GCTest.main(args);
-
-    ProcessBuilder pb = ProcessTools.createJavaProcessBuilder("-XX:+UseG1GC",
-                                                              "-Xmx10M",
-                                                              "-XX:+PrintGCDetails",
-                                                              // G1DeferredRSUpdate is a develop option, but we cannot limit execution of this test to only debug VMs.
-                                                              "-XX:+IgnoreUnrecognizedVMOptions",
-                                                              "-XX:-G1DeferredRSUpdate",
-                                                              GCTest.class.getName());
-
-    OutputAnalyzer output = new OutputAnalyzer(pb.start());
-    output.shouldHaveExitValue(0);
-  }
-
-  static class GCTest {
-    private static Object[] garbage = new Object[32];
-
-    public static void main(String [] args) {
-      System.out.println("Creating garbage");
-      // Create 128MB of garbage. This should result in at least one minor GC, with
-      // some objects copied to old gen. As references from old to young are installed,
-      // the crash due to the use before initialize occurs.
-      Object prev = null;
-      Object prevPrev = null;
-      for (int i = 0; i < 1024; i++) {
-        Object[] next = new Object[32 * 1024];
-        next[0] = prev;
-        next[1] = prevPrev;
-
-        Object[] cur = (Object[]) garbage[i % garbage.length];
-        if (cur != null) {
-          cur[0] = null;
-          cur[1] = null;
+import com.oracle.java.testlibrary.*;
+import sun.misc.Unsafe;
+
+public class UnsafeMallocLimit2 {
+
+    public static void main(String args[]) throws Exception {
+        if (Platform.is32bit()) {
+            Unsafe unsafe = Utils.getUnsafe();
+            try {
+                // Allocate greater than MALLOC_MAX and likely won't fail to allocate,
+                // so it hits the NMT code that asserted.
+                // Test that this doesn't cause an assertion with NMT off.
+                // The option above overrides if all the tests are run with NMT on.
+                unsafe.allocateMemory(0x40000000);
+                System.out.println("Allocation succeeded");
+            } catch (OutOfMemoryError e) {
+                System.out.println("Allocation failed");
+            }
+        } else {
+            System.out.println("Test only valid on 32-bit platforms");
         }
-        garbage[i % garbage.length] = next;
-
-        prevPrev = prev;
-        prev = next;
-      }
-      System.out.println("Done");
     }
-  }
 }