Merge

.hgtags

@@ -299,3 +299,7 @@ cfc5309f03b7bd6c1567618b63cf1fc74c0f2a8f hs25-b10
 b61d9c88b759d1594b8af1655598e8fa00393672 hs25-b11
 25bdce771bb3a7ae9825261a284d292cda700122 jdk8-b67
 a35a72dd2e1255239d31f796f9f693e49b36bc9f hs25-b12
+121aa71316af6cd877bf455e775fa3fdbcdd4b65 jdk8-b68
+b6c9c0109a608eedbb6b868d260952990e3c91fe hs25-b13
+cb8a4e04bc8c104de8a2f67463c7e31232bf8d68 jdk8-b69
+990bbd393c239d95310ccc38094e57923bbf1d4a hs25-b14

agent/src/share/classes/sun/jvm/hotspot/oops/ConstMethod.java

@@ -69,6 +69,8 @@ public class ConstMethod extends VMObject {
     signatureIndex             = new CIntField(type.getCIntegerField("_signature_index"), 0);
     idnum                      = new CIntField(type.getCIntegerField("_method_idnum"), 0);
     maxStack                   = new CIntField(type.getCIntegerField("_max_stack"), 0);
+    maxLocals                  = new CIntField(type.getCIntegerField("_max_locals"), 0);
+    sizeOfParameters           = new CIntField(type.getCIntegerField("_size_of_parameters"), 0);
 
     // start of byte code
     bytecodeOffset = type.getSize();
@@ -96,6 +98,8 @@ public class ConstMethod extends VMObject {
   private static CIntField signatureIndex;
   private static CIntField idnum;
   private static CIntField maxStack;
+  private static CIntField maxLocals;
+  private static CIntField sizeOfParameters;
 
   // start of bytecode
   private static long bytecodeOffset;
@@ -151,6 +155,14 @@ public class ConstMethod extends VMObject {
     return maxStack.getValue(this);
   }
 
+  public long getMaxLocals() {
+    return maxLocals.getValue(this);
+  }
+
+  public long getSizeOfParameters() {
+    return sizeOfParameters.getValue(this);
+  }
+
   public Symbol getName() {
     return getMethod().getName();
   }
@@ -247,6 +259,8 @@ public class ConstMethod extends VMObject {
       visitor.doCInt(signatureIndex, true);
       visitor.doCInt(codeSize, true);
       visitor.doCInt(maxStack, true);
+      visitor.doCInt(maxLocals, true);
+      visitor.doCInt(sizeOfParameters, true);
     }
 
   // Accessors

agent/src/share/classes/sun/jvm/hotspot/oops/Method.java

@@ -50,8 +50,6 @@ public class Method extends Metadata {
     constMethod                = type.getAddressField("_constMethod");
     methodData                 = type.getAddressField("_method_data");
     methodSize                 = new CIntField(type.getCIntegerField("_method_size"), 0);
-    maxLocals                  = new CIntField(type.getCIntegerField("_max_locals"), 0);
-    sizeOfParameters           = new CIntField(type.getCIntegerField("_size_of_parameters"), 0);
     accessFlags                = new CIntField(type.getCIntegerField("_access_flags"), 0);
     code                       = type.getAddressField("_code");
     vtableIndex                = new CIntField(type.getCIntegerField("_vtable_index"), 0);
@@ -83,8 +81,6 @@ public class Method extends Metadata {
   private static AddressField  constMethod;
   private static AddressField  methodData;
   private static CIntField methodSize;
-  private static CIntField maxLocals;
-  private static CIntField sizeOfParameters;
   private static CIntField accessFlags;
   private static CIntField vtableIndex;
   private static CIntField invocationCounter;
@@ -134,8 +130,8 @@ public class Method extends Metadata {
   /** WARNING: this is in words, not useful in this system; use getObjectSize() instead */
   public long         getMethodSize()                 { return                methodSize.getValue(this);        }
   public long         getMaxStack()                   { return                getConstMethod().getMaxStack();   }
-  public long         getMaxLocals()                  { return                maxLocals.getValue(this);         }
-  public long         getSizeOfParameters()           { return                sizeOfParameters.getValue(this);  }
+  public long         getMaxLocals()                  { return                getConstMethod().getMaxLocals();         }
+  public long         getSizeOfParameters()           { return                getConstMethod().getSizeOfParameters();  }
   public long         getNameIndex()                  { return                getConstMethod().getNameIndex();  }
   public long         getSignatureIndex()             { return            getConstMethod().getSignatureIndex(); }
   public long         getGenericSignatureIndex()      { return     getConstMethod().getGenericSignatureIndex(); }
@@ -282,8 +278,6 @@ public class Method extends Metadata {
 
   public void iterateFields(MetadataVisitor visitor) {
       visitor.doCInt(methodSize, true);
-      visitor.doCInt(maxLocals, true);
-      visitor.doCInt(sizeOfParameters, true);
       visitor.doCInt(accessFlags, true);
     }
 

make/hotspot_version

@@ -35,7 +35,7 @@ HOTSPOT_VM_COPYRIGHT=Copyright 2012
 
 HS_MAJOR_VER=25
 HS_MINOR_VER=0
-HS_BUILD_NUMBER=13
+HS_BUILD_NUMBER=15
 
 JDK_MAJOR_VER=1
 JDK_MINOR_VER=8

src/cpu/sparc/vm/c1_CodeStubs_sparc.cpp

@@ -298,7 +298,7 @@ void PatchingStub::emit_code(LIR_Assembler* ce) {
     for (int i = 0; i < _bytes_to_copy; i++) {
       address ptr = (address)(_pc_start + i);
       int a_byte = (*ptr) & 0xFF;
-      __ a_byte (a_byte);
+      __ emit_int8 (a_byte);
     }
   }
 
@@ -340,10 +340,10 @@ void PatchingStub::emit_code(LIR_Assembler* ce) {
   int being_initialized_entry_offset = __ offset() - being_initialized_entry + sizeof_patch_record;
 
   // Emit the patch record.  We need to emit a full word, so emit an extra empty byte
-  __ a_byte(0);
-  __ a_byte(being_initialized_entry_offset);
-  __ a_byte(bytes_to_skip);
-  __ a_byte(_bytes_to_copy);
+  __ emit_int8(0);
+  __ emit_int8(being_initialized_entry_offset);
+  __ emit_int8(bytes_to_skip);
+  __ emit_int8(_bytes_to_copy);
   address patch_info_pc = __ pc();
   assert(patch_info_pc - end_of_patch == bytes_to_skip, "incorrect patch info");
 

src/cpu/sparc/vm/cppInterpreter_sparc.cpp

@@ -582,7 +582,9 @@ address InterpreterGenerator::generate_native_entry(bool synchronized) {
   // the following temporary registers are used during frame creation
   const Register Gtmp1 = G3_scratch ;
   const Register Gtmp2 = G1_scratch;
-  const Address size_of_parameters(G5_method, 0, in_bytes(Method::size_of_parameters_offset()));
+  const Register RconstMethod = Gtmp1;
+  const Address constMethod(G5_method, 0, in_bytes(Method::const_offset()));
+  const Address size_of_parameters(RconstMethod, 0, in_bytes(ConstMethod::size_of_parameters_offset()));
 
   bool inc_counter  = UseCompiler || CountCompiledCalls;
 
@@ -618,6 +620,7 @@ address InterpreterGenerator::generate_native_entry(bool synchronized) {
   }
 #endif // ASSERT
 
+  __ ld_ptr(constMethod, RconstMethod);
   __ lduh(size_of_parameters, Gtmp1);
   __ sll(Gtmp1, LogBytesPerWord, Gtmp2);       // parameter size in bytes
   __ add(Gargs, Gtmp2, Gargs);                 // points to first local + BytesPerWord
@@ -1047,8 +1050,6 @@ void CppInterpreterGenerator::generate_compute_interpreter_state(const Register
   const Register Gtmp = G3_scratch;
   const Address constMethod       (G5_method, 0, in_bytes(Method::const_offset()));
   const Address access_flags      (G5_method, 0, in_bytes(Method::access_flags_offset()));
-  const Address size_of_parameters(G5_method, 0, in_bytes(Method::size_of_parameters_offset()));
-  const Address size_of_locals    (G5_method, 0, in_bytes(Method::size_of_locals_offset()));
 
   // slop factor is two extra slots on the expression stack so that
   // we always have room to store a result when returning from a call without parameters
@@ -1066,6 +1067,9 @@ void CppInterpreterGenerator::generate_compute_interpreter_state(const Register
   // Now compute new frame size
 
   if (native) {
+    const Register RconstMethod = Gtmp;
+    const Address size_of_parameters(RconstMethod, 0, in_bytes(ConstMethod::size_of_parameters_offset()));
+    __ ld_ptr(constMethod, RconstMethod);
     __ lduh( size_of_parameters, Gtmp );
     __ calc_mem_param_words(Gtmp, Gtmp);     // space for native call parameters passed on the stack in words
   } else {
@@ -1236,9 +1240,13 @@ void CppInterpreterGenerator::generate_compute_interpreter_state(const Register
     }
     if (init_value != noreg) {
       Label clear_loop;
+      const Register RconstMethod = O1;
+      const Address size_of_parameters(RconstMethod, 0, in_bytes(ConstMethod::size_of_parameters_offset()));
+      const Address size_of_locals    (RconstMethod, 0, in_bytes(ConstMethod::size_of_locals_offset()));
 
       // NOTE: If you change the frame layout, this code will need to
       // be updated!
+      __ ld_ptr( constMethod, RconstMethod );
       __ lduh( size_of_locals, O2 );
       __ lduh( size_of_parameters, O1 );
       __ sll( O2, LogBytesPerWord, O2);
@@ -1483,13 +1491,16 @@ void CppInterpreterGenerator::adjust_callers_stack(Register args) {
 //
 //  assert_different_registers(state, prev_state);
   const Register Gtmp = G3_scratch;
+  const RconstMethod = G3_scratch;
   const Register tmp = O2;
-  const Address size_of_parameters(G5_method, 0, in_bytes(Method::size_of_parameters_offset()));
-  const Address size_of_locals    (G5_method, 0, in_bytes(Method::size_of_locals_offset()));
+  const Address constMethod(G5_method, 0, in_bytes(Method::const_offset()));
+  const Address size_of_parameters(RconstMethod, 0, in_bytes(ConstMethod::size_of_parameters_offset()));
+  const Address size_of_locals    (RconstMethod, 0, in_bytes(ConstMethod::size_of_locals_offset()));
 
+  __ ld_ptr(constMethod, RconstMethod);
   __ lduh(size_of_parameters, tmp);
-  __ sll(tmp, LogBytesPerWord, Gtmp);       // parameter size in bytes
-  __ add(args, Gtmp, Gargs);                // points to first local + BytesPerWord
+  __ sll(tmp, LogBytesPerWord, Gargs);       // parameter size in bytes
+  __ add(args, Gargs, Gargs);                // points to first local + BytesPerWord
   // NEW
   __ add(Gargs, -wordSize, Gargs);             // points to first local[0]
   // determine extra space for non-argument locals & adjust caller's SP
@@ -1541,8 +1552,6 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
 
   const Address constMethod       (G5_method, 0, in_bytes(Method::const_offset()));
   const Address access_flags      (G5_method, 0, in_bytes(Method::access_flags_offset()));
-  const Address size_of_parameters(G5_method, 0, in_bytes(Method::size_of_parameters_offset()));
-  const Address size_of_locals    (G5_method, 0, in_bytes(Method::size_of_locals_offset()));
 
   address entry_point = __ pc();
   __ mov(G0, prevState);                                                 // no current activation
@@ -1750,7 +1759,9 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
 
   __ ld_ptr(STATE(_result._to_call._callee), L4_scratch);                        // called method
   __ ld_ptr(STATE(_stack), L1_scratch);                                          // get top of java expr stack
-  __ lduh(L4_scratch, in_bytes(Method::size_of_parameters_offset()), L2_scratch); // get parameter size
+  // get parameter size
+  __ ld_ptr(L4_scratch, in_bytes(Method::const_offset()), L2_scratch);
+  __ lduh(L2_scratch, in_bytes(ConstMethod::size_of_parameters_offset()), L2_scratch);
   __ sll(L2_scratch, LogBytesPerWord, L2_scratch     );                           // parameter size in bytes
   __ add(L1_scratch, L2_scratch, L1_scratch);                                      // stack destination for result
   __ ld(L4_scratch, in_bytes(Method::result_index_offset()), L3_scratch); // called method result type index

src/cpu/sparc/vm/macroAssembler_sparc.cpp

@@ -100,34 +100,6 @@ const char* Argument::name() const {
 bool AbstractAssembler::pd_check_instruction_mark() { return false; }
 #endif
 
-
-void MacroAssembler::print_instruction(int inst) {
-  const char* s;
-  switch (inv_op(inst)) {
-  default:         s = "????"; break;
-  case call_op:    s = "call"; break;
-  case branch_op:
-    switch (inv_op2(inst)) {
-      case fb_op2:     s = "fb";   break;
-      case fbp_op2:    s = "fbp";  break;
-      case br_op2:     s = "br";   break;
-      case bp_op2:     s = "bp";   break;
-      case cb_op2:     s = "cb";   break;
-      case bpr_op2: {
-        if (is_cbcond(inst)) {
-          s = is_cxb(inst) ? "cxb" : "cwb";
-        } else {
-          s = "bpr";
-        }
-        break;
-      }
-      default:         s = "????"; break;
-    }
-  }
-  ::tty->print("%s", s);
-}
-
-
 // Patch instruction inst at offset inst_pos to refer to dest_pos
 // and return the resulting instruction.
 // We should have pcs, not offsets, but since all is relative, it will work out

src/cpu/sparc/vm/macroAssembler_sparc.hpp

@@ -603,7 +603,6 @@ class MacroAssembler : public Assembler {
   friend class Label;
 
  protected:
-  static void print_instruction(int inst);
   static int  patched_branch(int dest_pos, int inst, int inst_pos);
   static int  branch_destination(int inst, int pos);
 
@@ -759,9 +758,6 @@ class MacroAssembler : public Assembler {
   // Required platform-specific helpers for Label::patch_instructions.
   // They _shadow_ the declarations in AbstractAssembler, which are undefined.
   void pd_patch_instruction(address branch, address target);
-#ifndef PRODUCT
-  static void pd_print_patched_instruction(address branch);
-#endif
 
   // sethi Macro handles optimizations and relocations
 private:

src/cpu/sparc/vm/macroAssembler_sparc.inline.hpp

@@ -43,14 +43,6 @@ inline void MacroAssembler::pd_patch_instruction(address branch, address target)
   stub_inst = patched_branch(target - branch, stub_inst, 0);
 }
 
-#ifndef PRODUCT
-inline void MacroAssembler::pd_print_patched_instruction(address branch) {
-  jint stub_inst = *(jint*) branch;
-  print_instruction(stub_inst);
-  ::tty->print("%s", " (unresolved)");
-}
-#endif // PRODUCT
-
 // Use the right loads/stores for the platform
 inline void MacroAssembler::ld_ptr( Register s1, Register s2, Register d ) {
 #ifdef _LP64

src/cpu/sparc/vm/methodHandles_sparc.cpp

@@ -171,7 +171,8 @@ void MethodHandles::jump_to_lambda_form(MacroAssembler* _masm,
 
   if (VerifyMethodHandles && !for_compiler_entry) {
     // make sure recv is already on stack
-    __ load_sized_value(Address(method_temp, Method::size_of_parameters_offset()),
+    __ ld_ptr(method_temp, in_bytes(Method::const_offset()), temp2);
+    __ load_sized_value(Address(temp2, ConstMethod::size_of_parameters_offset()),
                         temp2,
                         sizeof(u2), /*is_signed*/ false);
     // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), "");
@@ -233,7 +234,8 @@ address MethodHandles::generate_method_handle_interpreter_entry(MacroAssembler*
   int ref_kind = signature_polymorphic_intrinsic_ref_kind(iid);
   assert(ref_kind != 0 || iid == vmIntrinsics::_invokeBasic, "must be _invokeBasic or a linkTo intrinsic");
   if (ref_kind == 0 || MethodHandles::ref_kind_has_receiver(ref_kind)) {
-    __ load_sized_value(Address(G5_method, Method::size_of_parameters_offset()),
+    __ ld_ptr(G5_method, in_bytes(Method::const_offset()), O4_param_size);
+    __ load_sized_value(Address(O4_param_size, ConstMethod::size_of_parameters_offset()),
                         O4_param_size,
                         sizeof(u2), /*is_signed*/ false);
     // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), "");

src/cpu/sparc/vm/sparc.ad

@@ -10224,7 +10224,7 @@ instruct array_equals(o0RegP ary1, o1RegP ary2, g3RegI tmp1, notemp_iRegI result
 
 //---------- Zeros Count Instructions ------------------------------------------
 
-instruct countLeadingZerosI(iRegI dst, iRegI src, iRegI tmp, flagsReg cr) %{
+instruct countLeadingZerosI(iRegIsafe dst, iRegI src, iRegI tmp, flagsReg cr) %{
   predicate(UsePopCountInstruction);  // See Matcher::match_rule_supported
   match(Set dst (CountLeadingZerosI src));
   effect(TEMP dst, TEMP tmp, KILL cr);
@@ -10321,7 +10321,7 @@ instruct countLeadingZerosL(iRegIsafe dst, iRegL src, iRegL tmp, flagsReg cr) %{
   ins_pipe(ialu_reg);
 %}
 
-instruct countTrailingZerosI(iRegI dst, iRegI src, flagsReg cr) %{
+instruct countTrailingZerosI(iRegIsafe dst, iRegI src, flagsReg cr) %{
   predicate(UsePopCountInstruction);  // See Matcher::match_rule_supported
   match(Set dst (CountTrailingZerosI src));
   effect(TEMP dst, KILL cr);
@@ -10364,19 +10364,21 @@ instruct countTrailingZerosL(iRegIsafe dst, iRegL src, flagsReg cr) %{
 
 //---------- Population Count Instructions -------------------------------------
 
-instruct popCountI(iRegI dst, iRegI src) %{
+instruct popCountI(iRegIsafe dst, iRegI src) %{
   predicate(UsePopCountInstruction);
   match(Set dst (PopCountI src));
 
-  format %{ "POPC   $src, $dst" %}
+  format %{ "SRL    $src, G0, $dst\t! clear upper word for 64 bit POPC\n\t"
+            "POPC   $dst, $dst" %}
   ins_encode %{
-    __ popc($src$$Register, $dst$$Register);
+    __ srl($src$$Register, G0, $dst$$Register);
+    __ popc($dst$$Register, $dst$$Register);
   %}
   ins_pipe(ialu_reg);
 %}
 
 // Note: Long.bitCount(long) returns an int.
-instruct popCountL(iRegI dst, iRegL src) %{
+instruct popCountL(iRegIsafe dst, iRegL src) %{
   predicate(UsePopCountInstruction);
   match(Set dst (PopCountL src));
 

src/cpu/sparc/vm/templateInterpreter_sparc.cpp

@@ -434,7 +434,7 @@ void TemplateInterpreterGenerator::generate_stack_overflow_check(Register Rframe
 
   // the frame is greater than one page in size, so check against
   // the bottom of the stack
-  __ cmp_and_brx_short(SP, Rscratch, Assembler::greater, Assembler::pt, after_frame_check);
+  __ cmp_and_brx_short(SP, Rscratch, Assembler::greaterUnsigned, Assembler::pt, after_frame_check);
 
   // the stack will overflow, throw an exception
 
@@ -494,9 +494,6 @@ void TemplateInterpreterGenerator::generate_fixed_frame(bool native_call) {
   // (gri - 2/25/2000)
 
 
-  const Address size_of_parameters(G5_method, Method::size_of_parameters_offset());
-  const Address size_of_locals    (G5_method, Method::size_of_locals_offset());
-  const Address constMethod       (G5_method, Method::const_offset());
   int rounded_vm_local_words = round_to( frame::interpreter_frame_vm_local_words, WordsPerLong );
 
   const int extra_space =
@@ -506,11 +503,15 @@ void TemplateInterpreterGenerator::generate_fixed_frame(bool native_call) {
     (native_call ? frame::interpreter_frame_extra_outgoing_argument_words : 0);
 
   const Register Glocals_size = G3;
+  const Register RconstMethod = Glocals_size;
   const Register Otmp1 = O3;
   const Register Otmp2 = O4;
   // Lscratch can't be used as a temporary because the call_stub uses
   // it to assert that the stack frame was setup correctly.
+  const Address constMethod       (G5_method, Method::const_offset());
+  const Address size_of_parameters(RconstMethod, ConstMethod::size_of_parameters_offset());
 
+  __ ld_ptr( constMethod, RconstMethod );
   __ lduh( size_of_parameters, Glocals_size);
 
   // Gargs points to first local + BytesPerWord
@@ -530,6 +531,8 @@ void TemplateInterpreterGenerator::generate_fixed_frame(bool native_call) {
     //
     // Compute number of locals in method apart from incoming parameters
     //
+    const Address size_of_locals    (Otmp1, ConstMethod::size_of_locals_offset());
+    __ ld_ptr( constMethod, Otmp1 );
     __ lduh( size_of_locals, Otmp1 );
     __ sub( Otmp1, Glocals_size, Glocals_size );
     __ round_to( Glocals_size, WordsPerLong );
@@ -1256,8 +1259,7 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   // make sure registers are different!
   assert_different_registers(G2_thread, G5_method, Gargs, Gtmp1, Gtmp2);
 
-  const Address size_of_parameters(G5_method, Method::size_of_parameters_offset());
-  const Address size_of_locals    (G5_method, Method::size_of_locals_offset());
+  const Address constMethod       (G5_method, Method::const_offset());
   // Seems like G5_method is live at the point this is used. So we could make this look consistent
   // and use in the asserts.
   const Address access_flags      (Lmethod,   Method::access_flags_offset());
@@ -1307,8 +1309,13 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   init_value = G0;
   Label clear_loop;
 
+  const Register RconstMethod = O1;
+  const Address size_of_parameters(RconstMethod, ConstMethod::size_of_parameters_offset());
+  const Address size_of_locals    (RconstMethod, ConstMethod::size_of_locals_offset());
+
   // NOTE: If you change the frame layout, this code will need to
   // be updated!
+  __ ld_ptr( constMethod, RconstMethod );
   __ lduh( size_of_locals, O2 );
   __ lduh( size_of_parameters, O1 );
   __ sll( O2, Interpreter::logStackElementSize, O2);
@@ -1823,9 +1830,13 @@ void TemplateInterpreterGenerator::generate_throw_exception() {
 
     const Register Gtmp1 = G3_scratch;
     const Register Gtmp2 = G1_scratch;
+    const Register RconstMethod = Gtmp1;
+    const Address constMethod(Lmethod, Method::const_offset());
+    const Address size_of_parameters(RconstMethod, ConstMethod::size_of_parameters_offset());
 
     // Compute size of arguments for saving when returning to deoptimized caller
-    __ lduh(Lmethod, in_bytes(Method::size_of_parameters_offset()), Gtmp1);
+    __ ld_ptr(constMethod, RconstMethod);
+    __ lduh(size_of_parameters, Gtmp1);
     __ sll(Gtmp1, Interpreter::logStackElementSize, Gtmp1);
     __ sub(Llocals, Gtmp1, Gtmp2);
     __ add(Gtmp2, wordSize, Gtmp2);

src/cpu/sparc/vm/templateTable_sparc.cpp

@@ -3040,7 +3040,8 @@ void TemplateTable::invokevfinal_helper(Register Rscratch, Register Rret) {
   Register Rtemp = G4_scratch;
 
   // Load receiver from stack slot
-  __ lduh(G5_method, in_bytes(Method::size_of_parameters_offset()), G4_scratch);
+  __ ld_ptr(G5_method, in_bytes(Method::const_offset()), G4_scratch);
+  __ lduh(G4_scratch, in_bytes(ConstMethod::size_of_parameters_offset()), G4_scratch);
   __ load_receiver(G4_scratch, O0);
 
   // receiver NULL check

src/cpu/x86/vm/c1_CodeStubs_x86.cpp

@@ -313,10 +313,10 @@ void PatchingStub::emit_code(LIR_Assembler* ce) {
 #endif
   } else {
     // make a copy the code which is going to be patched.
-    for ( int i = 0; i < _bytes_to_copy; i++) {
+    for (int i = 0; i < _bytes_to_copy; i++) {
       address ptr = (address)(_pc_start + i);
       int a_byte = (*ptr) & 0xFF;
-      __ a_byte (a_byte);
+      __ emit_int8(a_byte);
       *ptr = 0x90; // make the site look like a nop
     }
   }
@@ -363,11 +363,11 @@ void PatchingStub::emit_code(LIR_Assembler* ce) {
   // emit the offsets needed to find the code to patch
   int being_initialized_entry_offset = __ pc() - being_initialized_entry + sizeof_patch_record;
 
-  __ a_byte(0xB8);
-  __ a_byte(0);
-  __ a_byte(being_initialized_entry_offset);
-  __ a_byte(bytes_to_skip);
-  __ a_byte(_bytes_to_copy);
+  __ emit_int8((unsigned char)0xB8);
+  __ emit_int8(0);
+  __ emit_int8(being_initialized_entry_offset);
+  __ emit_int8(bytes_to_skip);
+  __ emit_int8(_bytes_to_copy);
   address patch_info_pc = __ pc();
   assert(patch_info_pc - end_of_patch == bytes_to_skip, "incorrect patch info");
 

src/cpu/x86/vm/cppInterpreter_x86.cpp

@@ -611,8 +611,6 @@ void InterpreterGenerator::generate_counter_overflow(Label* do_continue) {
   // C++ interpreter only
   // rsi/r13 - previous interpreter state pointer
 
-  const Address size_of_parameters(rbx, Method::size_of_parameters_offset());
-
   // InterpreterRuntime::frequency_counter_overflow takes one argument
   // indicating if the counter overflow occurs at a backwards branch (non-NULL bcp).
   // The call returns the address of the verified entry point for the method or NULL
@@ -977,15 +975,16 @@ address InterpreterGenerator::generate_native_entry(bool synchronized) {
   //      to save/restore.
   address entry_point = __ pc();
 
-  const Address size_of_parameters(rbx, Method::size_of_parameters_offset());
-  const Address size_of_locals    (rbx, Method::size_of_locals_offset());
+  const Address constMethod       (rbx, Method::const_offset());
   const Address invocation_counter(rbx, Method::invocation_counter_offset() + InvocationCounter::counter_offset());
   const Address access_flags      (rbx, Method::access_flags_offset());
+  const Address size_of_parameters(rcx, ConstMethod::size_of_parameters_offset());
 
   // rsi/r13 == state/locals rdi == prevstate
   const Register locals = rdi;
 
   // get parameter size (always needed)
+  __ movptr(rcx, constMethod);
   __ load_unsigned_short(rcx, size_of_parameters);
 
   // rbx: Method*
@@ -994,6 +993,7 @@ address InterpreterGenerator::generate_native_entry(bool synchronized) {
   // for natives the size of locals is zero
 
   // compute beginning of parameters /locals
+
   __ lea(locals, Address(rsp, rcx, Address::times_ptr, -wordSize));
 
   // initialize fixed part of activation frame
@@ -1107,11 +1107,14 @@ address InterpreterGenerator::generate_native_entry(bool synchronized) {
   const Register method = rbx;
   const Register thread = LP64_ONLY(r15_thread) NOT_LP64(rdi);
   const Register t      = InterpreterRuntime::SignatureHandlerGenerator::temp();    // rcx|rscratch1
+  const Address constMethod       (method, Method::const_offset());
+  const Address size_of_parameters(t, ConstMethod::size_of_parameters_offset());
 
   // allocate space for parameters
   __ movptr(method, STATE(_method));
   __ verify_method_ptr(method);
-  __ load_unsigned_short(t, Address(method, Method::size_of_parameters_offset()));
+  __ movptr(t, constMethod);
+  __ load_unsigned_short(t, size_of_parameters);
   __ shll(t, 2);
 #ifdef _LP64
   __ subptr(rsp, t);
@@ -1700,15 +1703,17 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   // save sender sp
   __ push(rcx);
 
-  const Address size_of_parameters(rbx, Method::size_of_parameters_offset());
-  const Address size_of_locals    (rbx, Method::size_of_locals_offset());
+  const Address constMethod       (rbx, Method::const_offset());
   const Address access_flags      (rbx, Method::access_flags_offset());
+  const Address size_of_parameters(rdx, ConstMethod::size_of_parameters_offset());
+  const Address size_of_locals    (rdx, ConstMethod::size_of_locals_offset());
 
   // const Address monitor_block_top (rbp, frame::interpreter_frame_monitor_block_top_offset * wordSize);
   // const Address monitor_block_bot (rbp, frame::interpreter_frame_initial_sp_offset        * wordSize);
   // const Address monitor(rbp, frame::interpreter_frame_initial_sp_offset * wordSize - (int)sizeof(BasicObjectLock));
 
   // get parameter size (always needed)
+  __ movptr(rdx, constMethod);
   __ load_unsigned_short(rcx, size_of_parameters);
 
   // rbx: Method*
@@ -1989,7 +1994,9 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   __ movptr(rbx, STATE(_result._to_call._callee));
 
   // callee left args on top of expression stack, remove them
-  __ load_unsigned_short(rcx, Address(rbx, Method::size_of_parameters_offset()));
+  __ movptr(rcx, constMethod);
+  __ load_unsigned_short(rcx, Address(rcx, ConstMethod::size_of_parameters_offset()));
+
   __ lea(rsp, Address(rsp, rcx, Address::times_ptr));
 
   __ movl(rcx, Address(rbx, Method::result_index_offset()));
@@ -2159,7 +2166,9 @@ address InterpreterGenerator::generate_normal_entry(bool synchronized) {
   // Make it look like call_stub calling conventions
 
   // Get (potential) receiver
-  __ load_unsigned_short(rcx, size_of_parameters);                   // get size of parameters in words
+  // get size of parameters in words
+  __ movptr(rcx, constMethod);
+  __ load_unsigned_short(rcx, Address(rcx, ConstMethod::size_of_parameters_offset()));
 
   ExternalAddress recursive(CAST_FROM_FN_PTR(address, RecursiveInterpreterActivation));
   __ pushptr(recursive.addr());                                      // make it look good in the debugger

src/cpu/x86/vm/macroAssembler_x86.cpp

@@ -1023,7 +1023,7 @@ void MacroAssembler::lea(Address dst, AddressLiteral adr) {
 
 void MacroAssembler::leave() {
   // %%% is this really better? Why not on 32bit too?
-  emit_byte(0xC9); // LEAVE
+  emit_int8((unsigned char)0xC9); // LEAVE
 }
 
 void MacroAssembler::lneg(Register hi, Register lo) {
@@ -2112,11 +2112,11 @@ void MacroAssembler::fat_nop() {
   if (UseAddressNop) {
     addr_nop_5();
   } else {
-    emit_byte(0x26); // es:
-    emit_byte(0x2e); // cs:
-    emit_byte(0x64); // fs:
-    emit_byte(0x65); // gs:
-    emit_byte(0x90);
+    emit_int8(0x26); // es:
+    emit_int8(0x2e); // cs:
+    emit_int8(0x64); // fs:
+    emit_int8(0x65); // gs:
+    emit_int8((unsigned char)0x90);
   }
 }
 
@@ -2534,12 +2534,12 @@ void MacroAssembler::jump_cc(Condition cc, AddressLiteral dst) {
     int offs = (intptr_t)dst.target() - ((intptr_t)pc());
     if (dst.reloc() == relocInfo::none && is8bit(offs - short_size)) {
       // 0111 tttn #8-bit disp
-      emit_byte(0x70 | cc);
-      emit_byte((offs - short_size) & 0xFF);
+      emit_int8(0x70 | cc);
+      emit_int8((offs - short_size) & 0xFF);
     } else {
       // 0000 1111 1000 tttn #32-bit disp
-      emit_byte(0x0F);
-      emit_byte(0x80 | cc);
+      emit_int8(0x0F);
+      emit_int8((unsigned char)(0x80 | cc));
       emit_long(offs - long_size);
     }
   } else {
@@ -3085,7 +3085,8 @@ void MacroAssembler::xorps(XMMRegister dst, AddressLiteral src) {
 
 void MacroAssembler::pshufb(XMMRegister dst, AddressLiteral src) {
   // Used in sign-bit flipping with aligned address.
-  assert((UseAVX > 0) || (((intptr_t)src.target() & 15) == 0), "SSE mode requires address alignment 16 bytes");
+  bool aligned_adr = (((intptr_t)src.target() & 15) == 0);
+  assert((UseAVX > 0) || aligned_adr, "SSE mode requires address alignment 16 bytes");
   if (reachable(src)) {
     Assembler::pshufb(dst, as_Address(src));
   } else {

src/cpu/x86/vm/macroAssembler_x86.hpp

@@ -126,25 +126,6 @@ class MacroAssembler: public Assembler {
     }
   }
 
-#ifndef PRODUCT
-  static void pd_print_patched_instruction(address branch) {
-    const char* s;
-    unsigned char op = branch[0];
-    if (op == 0xE8) {
-      s = "call";
-    } else if (op == 0xE9 || op == 0xEB) {
-      s = "jmp";
-    } else if ((op & 0xF0) == 0x70) {
-      s = "jcc";
-    } else if (op == 0x0F) {
-      s = "jcc";
-    } else {
-      s = "????";
-    }
-    tty->print("%s (unresolved)", s);
-  }
-#endif
-
   // The following 4 methods return the offset of the appropriate move instruction
 
   // Support for fast byte/short loading with zero extension (depending on particular CPU)

src/cpu/x86/vm/methodHandles_x86.cpp

@@ -169,8 +169,9 @@ void MethodHandles::jump_to_lambda_form(MacroAssembler* _masm,
 
   if (VerifyMethodHandles && !for_compiler_entry) {
     // make sure recv is already on stack
+    __ movptr(temp2, Address(method_temp, Method::const_offset()));
     __ load_sized_value(temp2,
-                        Address(method_temp, Method::size_of_parameters_offset()),
+                        Address(temp2, ConstMethod::size_of_parameters_offset()),
                         sizeof(u2), /*is_signed*/ false);
     // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), "");
     Label L;
@@ -234,8 +235,9 @@ address MethodHandles::generate_method_handle_interpreter_entry(MacroAssembler*
   int ref_kind = signature_polymorphic_intrinsic_ref_kind(iid);
   assert(ref_kind != 0 || iid == vmIntrinsics::_invokeBasic, "must be _invokeBasic or a linkTo intrinsic");
   if (ref_kind == 0 || MethodHandles::ref_kind_has_receiver(ref_kind)) {
+    __ movptr(rdx_argp, Address(rbx_method, Method::const_offset()));
     __ load_sized_value(rdx_argp,
-                        Address(rbx_method, Method::size_of_parameters_offset()),
+                        Address(rdx_argp, ConstMethod::size_of_parameters_offset()),
                         sizeof(u2), /*is_signed*/ false);
     // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), "");
     rdx_first_arg_addr = __ argument_address(rdx_argp, -1);

src/cpu/x86/vm/stubGenerator_x86_32.cpp

@@ -2174,13 +2174,13 @@ class StubGenerator: public StubCodeGenerator {
   //   c_rarg2   - K (key) in little endian int array
   //
   address generate_aescrypt_encryptBlock() {
-    assert(UseAES && (UseAVX > 0), "need AES instructions and misaligned SSE support");
+    assert(UseAES, "need AES instructions and misaligned SSE support");
     __ align(CodeEntryAlignment);
     StubCodeMark mark(this, "StubRoutines", "aescrypt_encryptBlock");
     Label L_doLast;
     address start = __ pc();
 
-    const Register from        = rsi;      // source array address
+    const Register from        = rdx;      // source array address
     const Register to          = rdx;      // destination array address
     const Register key         = rcx;      // key array address
     const Register keylen      = rax;
@@ -2189,47 +2189,74 @@ class StubGenerator: public StubCodeGenerator {
     const Address  key_param (rbp, 8+8);
 
     const XMMRegister xmm_result = xmm0;
-    const XMMRegister xmm_temp   = xmm1;
-    const XMMRegister xmm_key_shuf_mask = xmm2;
+    const XMMRegister xmm_key_shuf_mask = xmm1;
+    const XMMRegister xmm_temp1  = xmm2;
+    const XMMRegister xmm_temp2  = xmm3;
+    const XMMRegister xmm_temp3  = xmm4;
+    const XMMRegister xmm_temp4  = xmm5;
 
-    __ enter(); // required for proper stackwalking of RuntimeStub frame
-    __ push(rsi);
-    __ movptr(from , from_param);
-    __ movptr(to   , to_param);
-    __ movptr(key  , key_param);
+    __ enter();   // required for proper stackwalking of RuntimeStub frame
+    __ movptr(from, from_param);
+    __ movptr(key, key_param);
 
+    // keylen could be only {11, 13, 15} * 4 = {44, 52, 60}
     __ movl(keylen, Address(key, arrayOopDesc::length_offset_in_bytes() - arrayOopDesc::base_offset_in_bytes(T_INT)));
-    // keylen = # of 32-bit words, convert to 128-bit words
-    __ shrl(keylen, 2);
-    __ subl(keylen, 11);   // every key has at least 11 128-bit words, some have more
 
     __ movdqu(xmm_key_shuf_mask, ExternalAddress(StubRoutines::x86::key_shuffle_mask_addr()));
     __ movdqu(xmm_result, Address(from, 0));  // get 16 bytes of input
+    __ movptr(to, to_param);
 
     // For encryption, the java expanded key ordering is just what we need
 
-    load_key(xmm_temp, key, 0x00, xmm_key_shuf_mask);
-    __ pxor(xmm_result, xmm_temp);
-    for (int offset = 0x10; offset <= 0x90; offset += 0x10) {
-      aes_enc_key(xmm_result, xmm_temp, key, offset, xmm_key_shuf_mask);
-    }
-    load_key  (xmm_temp, key, 0xa0, xmm_key_shuf_mask);
-    __ cmpl(keylen, 0);
-    __ jcc(Assembler::equal, L_doLast);
-    __ aesenc(xmm_result, xmm_temp);                   // only in 192 and 256 bit keys
-    aes_enc_key(xmm_result, xmm_temp, key, 0xb0, xmm_key_shuf_mask);
-    load_key(xmm_temp, key, 0xc0, xmm_key_shuf_mask);
-    __ subl(keylen, 2);
-    __ jcc(Assembler::equal, L_doLast);
-    __ aesenc(xmm_result, xmm_temp);                   // only in 256 bit keys
-    aes_enc_key(xmm_result, xmm_temp, key, 0xd0, xmm_key_shuf_mask);
-    load_key(xmm_temp, key, 0xe0, xmm_key_shuf_mask);
+    load_key(xmm_temp1, key, 0x00, xmm_key_shuf_mask);
+    __ pxor(xmm_result, xmm_temp1);
+
+    load_key(xmm_temp1, key, 0x10, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0x20, xmm_key_shuf_mask);
+    load_key(xmm_temp3, key, 0x30, xmm_key_shuf_mask);
+    load_key(xmm_temp4, key, 0x40, xmm_key_shuf_mask);
+
+    __ aesenc(xmm_result, xmm_temp1);
+    __ aesenc(xmm_result, xmm_temp2);
+    __ aesenc(xmm_result, xmm_temp3);
+    __ aesenc(xmm_result, xmm_temp4);
+
+    load_key(xmm_temp1, key, 0x50, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0x60, xmm_key_shuf_mask);
+    load_key(xmm_temp3, key, 0x70, xmm_key_shuf_mask);
+    load_key(xmm_temp4, key, 0x80, xmm_key_shuf_mask);
+
+    __ aesenc(xmm_result, xmm_temp1);
+    __ aesenc(xmm_result, xmm_temp2);
+    __ aesenc(xmm_result, xmm_temp3);
+    __ aesenc(xmm_result, xmm_temp4);
+
+    load_key(xmm_temp1, key, 0x90, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xa0, xmm_key_shuf_mask);
+
+    __ cmpl(keylen, 44);
+    __ jccb(Assembler::equal, L_doLast);
+
+    __ aesenc(xmm_result, xmm_temp1);
+    __ aesenc(xmm_result, xmm_temp2);
+
+    load_key(xmm_temp1, key, 0xb0, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xc0, xmm_key_shuf_mask);
+
+    __ cmpl(keylen, 52);
+    __ jccb(Assembler::equal, L_doLast);
+
+    __ aesenc(xmm_result, xmm_temp1);
+    __ aesenc(xmm_result, xmm_temp2);
+
+    load_key(xmm_temp1, key, 0xd0, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xe0, xmm_key_shuf_mask);
 
     __ BIND(L_doLast);
-    __ aesenclast(xmm_result, xmm_temp);
+    __ aesenc(xmm_result, xmm_temp1);
+    __ aesenclast(xmm_result, xmm_temp2);
     __ movdqu(Address(to, 0), xmm_result);        // store the result
     __ xorptr(rax, rax); // return 0
-    __ pop(rsi);
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
@@ -2245,13 +2272,13 @@ class StubGenerator: public StubCodeGenerator {
   //   c_rarg2   - K (key) in little endian int array
   //
   address generate_aescrypt_decryptBlock() {
-    assert(UseAES && (UseAVX > 0), "need AES instructions and misaligned SSE support");
+    assert(UseAES, "need AES instructions and misaligned SSE support");
     __ align(CodeEntryAlignment);
     StubCodeMark mark(this, "StubRoutines", "aescrypt_decryptBlock");
     Label L_doLast;
     address start = __ pc();
 
-    const Register from        = rsi;      // source array address
+    const Register from        = rdx;      // source array address
     const Register to          = rdx;      // destination array address
     const Register key         = rcx;      // key array address
     const Register keylen      = rax;
@@ -2260,51 +2287,76 @@ class StubGenerator: public StubCodeGenerator {
     const Address  key_param (rbp, 8+8);
 
     const XMMRegister xmm_result = xmm0;
-    const XMMRegister xmm_temp   = xmm1;
-    const XMMRegister xmm_key_shuf_mask = xmm2;
+    const XMMRegister xmm_key_shuf_mask = xmm1;
+    const XMMRegister xmm_temp1  = xmm2;
+    const XMMRegister xmm_temp2  = xmm3;
+    const XMMRegister xmm_temp3  = xmm4;
+    const XMMRegister xmm_temp4  = xmm5;
 
     __ enter(); // required for proper stackwalking of RuntimeStub frame
-    __ push(rsi);
-    __ movptr(from , from_param);
-    __ movptr(to   , to_param);
-    __ movptr(key  , key_param);
+    __ movptr(from, from_param);
+    __ movptr(key, key_param);
 
+    // keylen could be only {11, 13, 15} * 4 = {44, 52, 60}
     __ movl(keylen, Address(key, arrayOopDesc::length_offset_in_bytes() - arrayOopDesc::base_offset_in_bytes(T_INT)));
-    // keylen = # of 32-bit words, convert to 128-bit words
-    __ shrl(keylen, 2);
-    __ subl(keylen, 11);   // every key has at least 11 128-bit words, some have more
 
     __ movdqu(xmm_key_shuf_mask, ExternalAddress(StubRoutines::x86::key_shuffle_mask_addr()));
     __ movdqu(xmm_result, Address(from, 0));
+    __ movptr(to, to_param);
 
     // for decryption java expanded key ordering is rotated one position from what we want
     // so we start from 0x10 here and hit 0x00 last
     // we don't know if the key is aligned, hence not using load-execute form
-    load_key(xmm_temp, key, 0x10, xmm_key_shuf_mask);
-    __ pxor  (xmm_result, xmm_temp);
-    for (int offset = 0x20; offset <= 0xa0; offset += 0x10) {
-      aes_dec_key(xmm_result, xmm_temp, key, offset, xmm_key_shuf_mask);
-    }
-    __ cmpl(keylen, 0);
-    __ jcc(Assembler::equal, L_doLast);
-    // only in 192 and 256 bit keys
-    aes_dec_key(xmm_result, xmm_temp, key, 0xb0, xmm_key_shuf_mask);
-    aes_dec_key(xmm_result, xmm_temp, key, 0xc0, xmm_key_shuf_mask);
-    __ subl(keylen, 2);
-    __ jcc(Assembler::equal, L_doLast);
-    // only in 256 bit keys
-    aes_dec_key(xmm_result, xmm_temp, key, 0xd0, xmm_key_shuf_mask);
-    aes_dec_key(xmm_result, xmm_temp, key, 0xe0, xmm_key_shuf_mask);
+    load_key(xmm_temp1, key, 0x10, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0x20, xmm_key_shuf_mask);
+    load_key(xmm_temp3, key, 0x30, xmm_key_shuf_mask);
+    load_key(xmm_temp4, key, 0x40, xmm_key_shuf_mask);
+
+    __ pxor  (xmm_result, xmm_temp1);
+    __ aesdec(xmm_result, xmm_temp2);
+    __ aesdec(xmm_result, xmm_temp3);
+    __ aesdec(xmm_result, xmm_temp4);
+
+    load_key(xmm_temp1, key, 0x50, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0x60, xmm_key_shuf_mask);
+    load_key(xmm_temp3, key, 0x70, xmm_key_shuf_mask);
+    load_key(xmm_temp4, key, 0x80, xmm_key_shuf_mask);
+
+    __ aesdec(xmm_result, xmm_temp1);
+    __ aesdec(xmm_result, xmm_temp2);
+    __ aesdec(xmm_result, xmm_temp3);
+    __ aesdec(xmm_result, xmm_temp4);
+
+    load_key(xmm_temp1, key, 0x90, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xa0, xmm_key_shuf_mask);
+    load_key(xmm_temp3, key, 0x00, xmm_key_shuf_mask);
+
+    __ cmpl(keylen, 44);
+    __ jccb(Assembler::equal, L_doLast);
+
+    __ aesdec(xmm_result, xmm_temp1);
+    __ aesdec(xmm_result, xmm_temp2);
+
+    load_key(xmm_temp1, key, 0xb0, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xc0, xmm_key_shuf_mask);
+
+    __ cmpl(keylen, 52);
+    __ jccb(Assembler::equal, L_doLast);
+
+    __ aesdec(xmm_result, xmm_temp1);
+    __ aesdec(xmm_result, xmm_temp2);
+
+    load_key(xmm_temp1, key, 0xd0, xmm_key_shuf_mask);
+    load_key(xmm_temp2, key, 0xe0, xmm_key_shuf_mask);
 
     __ BIND(L_doLast);
-    // for decryption the aesdeclast operation is always on key+0x00
-    load_key(xmm_temp, key, 0x00, xmm_key_shuf_mask);
-    __ aesdeclast(xmm_result, xmm_temp);
+    __ aesdec(xmm_result, xmm_temp1);
+    __ aesdec(xmm_result, xmm_temp2);
 
+    // for decryption the aesdeclast operation is always on key+0x00
+    __ aesdeclast(xmm_result, xmm_temp3);
     __ movdqu(Address(to, 0), xmm_result);  // store the result
-
     __ xorptr(rax, rax); // return 0
-    __ pop(rsi);
     __ leave(); // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
@@ -2340,7 +2392,7 @@ class StubGenerator: public StubCodeGenerator {
   //   c_rarg4   - input length
   //
   address generate_cipherBlockChaining_encryptAESCrypt() {
-    assert(UseAES && (UseAVX > 0), "need AES instructions and misaligned SSE support");
+    assert(UseAES, "need AES instructions and misaligned SSE support");
     __ align(CodeEntryAlignment);
     StubCodeMark mark(this, "StubRoutines", "cipherBlockChaining_encryptAESCrypt");
     address start = __ pc();
@@ -2393,7 +2445,7 @@ class StubGenerator: public StubCodeGenerator {
     __ jcc(Assembler::notEqual, L_key_192_256);
 
     // 128 bit code follows here
-    __ movptr(pos, 0);
+    __ movl(pos, 0);
     __ align(OptoLoopAlignment);
     __ BIND(L_loopTop_128);
     __ movdqu(xmm_temp, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of input
@@ -2423,15 +2475,15 @@ class StubGenerator: public StubCodeGenerator {
     __ leave();                                  // required for proper stackwalking of RuntimeStub frame
     __ ret(0);
 
-  __ BIND(L_key_192_256);
-  // here rax = len in ints of AESCrypt.KLE array (52=192, or 60=256)
+    __ BIND(L_key_192_256);
+    // here rax = len in ints of AESCrypt.KLE array (52=192, or 60=256)
     __ cmpl(rax, 52);
     __ jcc(Assembler::notEqual, L_key_256);
 
     // 192-bit code follows here (could be changed to use more xmm registers)
-    __ movptr(pos, 0);
-  __ align(OptoLoopAlignment);
-  __ BIND(L_loopTop_192);
+    __ movl(pos, 0);
+    __ align(OptoLoopAlignment);
+    __ BIND(L_loopTop_192);
     __ movdqu(xmm_temp, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of input
     __ pxor  (xmm_result, xmm_temp);                                // xor with the current r vector
 
@@ -2452,11 +2504,11 @@ class StubGenerator: public StubCodeGenerator {
     __ jcc(Assembler::notEqual, L_loopTop_192);
     __ jmp(L_exit);
 
-  __ BIND(L_key_256);
+    __ BIND(L_key_256);
     // 256-bit code follows here (could be changed to use more xmm registers)
-    __ movptr(pos, 0);
-  __ align(OptoLoopAlignment);
-  __ BIND(L_loopTop_256);
+    __ movl(pos, 0);
+    __ align(OptoLoopAlignment);
+    __ BIND(L_loopTop_256);
     __ movdqu(xmm_temp, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of input
     __ pxor  (xmm_result, xmm_temp);                                // xor with the current r vector
 
@@ -2495,7 +2547,7 @@ class StubGenerator: public StubCodeGenerator {
   //
 
   address generate_cipherBlockChaining_decryptAESCrypt() {
-    assert(UseAES && (UseAVX > 0), "need AES instructions and misaligned SSE support");
+    assert(UseAES, "need AES instructions and misaligned SSE support");
     __ align(CodeEntryAlignment);
     StubCodeMark mark(this, "StubRoutines", "cipherBlockChaining_decryptAESCrypt");
     address start = __ pc();
@@ -2556,9 +2608,9 @@ class StubGenerator: public StubCodeGenerator {
 
 
     // 128-bit code follows here, parallelized
-    __ movptr(pos, 0);
-  __ align(OptoLoopAlignment);
-  __ BIND(L_singleBlock_loopTop_128);
+    __ movl(pos, 0);
+    __ align(OptoLoopAlignment);
+    __ BIND(L_singleBlock_loopTop_128);
     __ cmpptr(len_reg, 0);           // any blocks left??
     __ jcc(Assembler::equal, L_exit);
     __ movdqu(xmm_result, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of cipher input
@@ -2597,7 +2649,7 @@ class StubGenerator: public StubCodeGenerator {
     __ jcc(Assembler::notEqual, L_key_256);
 
     // 192-bit code follows here (could be optimized to use parallelism)
-    __ movptr(pos, 0);
+    __ movl(pos, 0);
     __ align(OptoLoopAlignment);
     __ BIND(L_singleBlock_loopTop_192);
     __ movdqu(xmm_result, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of cipher input
@@ -2622,7 +2674,7 @@ class StubGenerator: public StubCodeGenerator {
 
     __ BIND(L_key_256);
     // 256-bit code follows here (could be optimized to use parallelism)
-    __ movptr(pos, 0);
+    __ movl(pos, 0);
     __ align(OptoLoopAlignment);
     __ BIND(L_singleBlock_loopTop_256);
     __ movdqu(xmm_result, Address(from, pos, Address::times_1, 0));   // get next 16 bytes of cipher input

src/os_cpu/windows_x86/vm/assembler_windows_x86.cpp

@@ -30,7 +30,7 @@
 
 
 void MacroAssembler::int3() {
-  emit_byte(0xCC);
+  emit_int8((unsigned char)0xCC);
 }
 
 #ifndef _LP64