jdk8-mips64-public/hotspot: src/cpu/mips/vm/templateTable_mips

src/cpu/mips/vm/templateTable_mips_64.cpp@179d203c9b2b (annotated)

src/cpu/mips/vm/templateTable_mips_64.cpp

Thu, 25 Aug 2016 22:31:58 +0800

author: jiangshaofeng
date: Thu, 25 Aug 2016 22:31:58 +0800
changeset 89: 179d203c9b2b
parent 88: 02ae7081a1a7
child 104: 36f7453a6977
permissions: -rw-r--r--

#4428: Use gsdmult to optimize the lmul template of the interpreter.
I have run the test program on 3A2000 computer

Effects:
[loongson@localhost project]$ ./LmulTest.sh
before
time:3225ms
after
time:3115ms

The test java program:
public class LmulTest{
public static void main(String args[]){
int count = 10000000;
long startTime = System.currentTimeMillis();
//long startTime = System.nanoTime();
for(int i = 0; i < count; i++){
long a, b, c;
a = 12345678;
a++;
b = 87654321;
b++;
c = a * b;
//System.out.println(c);
}
long endTime = System.currentTimeMillis();
//long endTime = System.nanoTime();
System.out.println("time:" + (endTime - startTime) + "ms");
}
}

 /*
  * Copyright (c) 2003, 2013, Oracle and/or its affiliates. All rights reserved.
  * Copyright (c) 2015, 2016, Loongson Technology. 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.
  *
  */
 #include "precompiled.hpp"
 #include "asm/macroAssembler.hpp"
 #include "interpreter/interpreter.hpp"
 #include "interpreter/interpreterRuntime.hpp"
 #include "interpreter/templateTable.hpp"
 #include "memory/universe.inline.hpp"
 #include "oops/methodData.hpp"
 #include "oops/objArrayKlass.hpp"
 #include "oops/oop.inline.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
 #include "runtime/synchronizer.hpp"
 #ifndef CC_INTERP
 #define __ _masm->
 // Platform-dependent initialization
 void TemplateTable::pd_initialize() {
   // No mips specific initialization
 }
 // Address computation: local variables
 // we use t8 as the local variables pointer register, by yjl 6/27/2005
 static inline Address iaddress(int n) {
   return Address(LVP, Interpreter::local_offset_in_bytes(n));
 }
 static inline Address laddress(int n) {
   return iaddress(n + 1);
 }
 static inline Address faddress(int n) {
   return iaddress(n);
 }
 static inline Address daddress(int n) {
   return laddress(n);
 }
 static inline Address aaddress(int n) {
   return iaddress(n);
 }
 static inline Address haddress(int n)            { return iaddress(n + 0); }
 //FIXME , can not use dadd and dsll
 /*
 static inline Address iaddress(Register r) {
   return Address(r14, r, Address::times_8, Interpreter::value_offset_in_bytes());
 }
 static inline Address laddress(Register r) {
   return Address(r14, r, Address::times_8, Interpreter::local_offset_in_bytes(1));
 }
 static inline Address faddress(Register r) {
   return iaddress(r);
 }
 static inline Address daddress(Register r) {
   return laddress(r);
 }
 static inline Address aaddress(Register r) {
   return iaddress(r);
 }
 */
 static inline Address at_sp() 						{	return Address(SP, 	0); }
 static inline Address at_sp_p1()          { return Address(SP,  1 * wordSize); }
 static inline Address at_sp_p2()          { return Address(SP,  2 * wordSize); }
 // At top of Java expression stack which may be different than esp().  It
 // isn't for category 1 objects.
 static inline Address at_tos   () {
   Address tos = Address(SP,  Interpreter::expr_offset_in_bytes(0));
   return tos;
 }
 static inline Address at_tos_p1() {
   return Address(SP,  Interpreter::expr_offset_in_bytes(1));
 }
 static inline Address at_tos_p2() {
   return Address(SP,  Interpreter::expr_offset_in_bytes(2));
 }
 static inline Address at_tos_p3() {
   return Address(SP,  Interpreter::expr_offset_in_bytes(3));
 }
 // we use S0 as bcp, be sure you have bcp in S0 before you call any of the Template generator
 Address TemplateTable::at_bcp(int offset) {
   assert(_desc->uses_bcp(), "inconsistent uses_bcp information");
   return Address(BCP, offset);
 }
 #define callee_saved_register(R) assert((R>=S0 && R<=S7), "should use callee saved registers!")
 // bytecode folding
 void TemplateTable::patch_bytecode(Bytecodes::Code bc, Register bc_reg,
                                    Register tmp_reg,
                                    bool load_bc_into_bc_reg,/*=true*/
                                    int byte_no) {
   if (!RewriteBytecodes) {
     return;
   }
   Label L_patch_done;
   switch (bc) {
   case Bytecodes::_fast_aputfield:
   case Bytecodes::_fast_bputfield:
   case Bytecodes::_fast_cputfield:
   case Bytecodes::_fast_dputfield:
   case Bytecodes::_fast_fputfield:
   case Bytecodes::_fast_iputfield:
   case Bytecodes::_fast_lputfield:
   case Bytecodes::_fast_sputfield:
     {
     // We skip bytecode quickening for putfield instructions when the put_code written to the constant pool cache
     // is zero. This is required so that every execution of this instruction calls out to
     // InterpreterRuntime::resolve_get_put to do additional, required work.
     assert(byte_no == f1_byte || byte_no == f2_byte, "byte_no out of range");
     assert(load_bc_into_bc_reg, "we use bc_reg as temp");
     __ get_cache_and_index_and_bytecode_at_bcp(tmp_reg, bc_reg, tmp_reg, byte_no, 1);
     __ daddi(bc_reg, R0, bc);
     __ beq(tmp_reg, R0, L_patch_done);
     __ delayed()->nop();
     }
     break;
   default:
     assert(byte_no == -1, "sanity");
  // the pair bytecodes have already done the load.
   if (load_bc_into_bc_reg) {
     __ move(bc_reg, bc);
   }
   }
   if (JvmtiExport::can_post_breakpoint()) {
     Label L_fast_patch;
     // if a breakpoint is present we can't rewrite the stream directly
     __ lbu(tmp_reg, at_bcp(0));
     __ move(AT, Bytecodes::_breakpoint);
     __ bne(tmp_reg, AT, L_fast_patch);
     __ delayed()->nop();
     __ get_method(tmp_reg);
     // Let breakpoint table handling rewrite to quicker bytecode
     __ call_VM(NOREG, CAST_FROM_FN_PTR(address,
 	  InterpreterRuntime::set_original_bytecode_at), tmp_reg, BCP, bc_reg);
     __ b(L_patch_done);
     __ delayed()->nop();
     __ bind(L_fast_patch);
   }
 #ifdef ASSERT
   Label L_okay;
   __ lbu(tmp_reg, at_bcp(0));
   __ move(AT, (int)Bytecodes::java_code(bc));
   __ beq(tmp_reg, AT, L_okay);
   __ delayed()->nop();
   __ beq(tmp_reg, bc_reg, L_patch_done);
   __ delayed()->nop();
   __ stop("patching the wrong bytecode");
   __ bind(L_okay);
 #endif
   // patch bytecode
   __ sb(bc_reg, at_bcp(0));
   __ bind(L_patch_done);
 }
 // Individual instructions
 void TemplateTable::nop() {
   transition(vtos, vtos);
   // nothing to do
 }
 void TemplateTable::shouldnotreachhere() {
   transition(vtos, vtos);
   __ stop("shouldnotreachhere bytecode");
 }
 void TemplateTable::aconst_null() {
   transition(vtos, atos);
   __ move(FSR, R0);
 }
 void TemplateTable::iconst(int value) {
   transition(vtos, itos);
   if (value == 0) {
     __ move(FSR, R0);
   } else {
     __ move(FSR, value);
   }
 }
 void TemplateTable::lconst(int value) {
   transition(vtos, ltos);
   if (value == 0) {
     __ move(FSR, R0);
   } else {
     __ move(FSR, value);
   }
   assert(value >= 0, "check this code");
   //__ move(SSR, R0);
 }
 void TemplateTable::fconst(int value) {
   static float  _f1 = 1.0, _f2 = 2.0;
   transition(vtos, ftos);
   float* p;
   switch( value ) {
     default: ShouldNotReachHere();
     case 0:  __ dmtc1(R0, FSF);  return;
     case 1:  p = &_f1;   break;
     case 2:  p = &_f2;   break;
   }
   __ li(AT, (address)p);
   __ lwc1(FSF, AT, 0);
 }
 void TemplateTable::dconst(int value) {
   static double _d1 = 1.0;
   transition(vtos, dtos);
   double* p;
   switch( value ) {
     default: ShouldNotReachHere();
     case 0:  __ dmtc1(R0, FSF);  return;
     case 1:  p = &_d1;   break;
   }
   __ li(AT, (address)p);
   __ ldc1(FSF, AT, 0);
 }
 void TemplateTable::bipush() {
   transition(vtos, itos);
   __ lb(FSR, at_bcp(1));
 }
 void TemplateTable::sipush() {
 	transition(vtos, itos);
 	__ get_2_byte_integer_at_bcp(FSR, AT, 1);
 	__ hswap(FSR);
 }
 // T1 : tags
 // T2 : index
 // T3 : cpool
 // T8 : tag
 void TemplateTable::ldc(bool wide) {
   transition(vtos, vtos);
   Label call_ldc, notFloat, notClass, Done;
   // get index in cpool
   if (wide) {
     __ get_2_byte_integer_at_bcp(T2, AT, 1);
     __ huswap(T2);
   } else {
     __ lbu(T2, at_bcp(1));
   }
   __ get_cpool_and_tags(T3, T1);
   const int base_offset = ConstantPool::header_size() * wordSize;
   const int tags_offset = Array<u1>::base_offset_in_bytes();
   // get type
   __ dadd(AT, T1, T2);
   __ lb(T1, AT, tags_offset);
   //now T1 is the tag
   // unresolved string - get the resolved string
   /*__ daddiu(AT, T1, - JVM_CONSTANT_UnresolvedString);
   __ beq(AT, R0, call_ldc);
   __ delayed()->nop();*/
   // unresolved class - get the resolved class
   __ daddiu(AT, T1, - JVM_CONSTANT_UnresolvedClass);
   __ beq(AT, R0, call_ldc);
   __ delayed()->nop();
   // unresolved class in error (resolution failed) - call into runtime
   // so that the same error from first resolution attempt is thrown.
   __ daddiu(AT, T1, -JVM_CONSTANT_UnresolvedClassInError);
   __ beq(AT, R0, call_ldc);
   __ delayed()->nop();
   // resolved class - need to call vm to get java mirror of the class
   __ daddiu(AT, T1, - JVM_CONSTANT_Class);
   __ bne(AT, R0, notClass);
   __ delayed()->dsll(T2, T2, Address::times_8);
   __ bind(call_ldc);
   __ move(A1, wide);
   call_VM(FSR, CAST_FROM_FN_PTR(address, InterpreterRuntime::ldc), A1);
   //	__ sw(FSR, SP, - 1 * wordSize);
   __ push(atos);
   __ b(Done);
   //	__ delayed()->daddi(SP, SP, - 1 * wordSize);
   __ delayed()->nop();
   __ bind(notClass);
   __ daddiu(AT, T1, -JVM_CONSTANT_Float);
   __ bne(AT, R0, notFloat);
   __ delayed()->nop();
   // ftos
   __ dadd(AT, T3, T2);
   __ lwc1(FSF, AT, base_offset);
   __ push_f();
   __ b(Done);
   __ delayed()->nop();
   __ bind(notFloat);
 #ifdef ASSERT
   {
     Label L;
     __ daddiu(AT, T1, -JVM_CONSTANT_Integer);
     __ beq(AT, R0, L);
     __ delayed()->nop();
     __ stop("unexpected tag type in ldc");
     __ bind(L);
   }
 #endif
   // atos and itos
   __ dadd(T0, T3, T2);
   __ lw(FSR, T0, base_offset);
   __ push(itos);
   __ b(Done);
   __ delayed()->nop();
   if (VerifyOops) {
     __ verify_oop(FSR);
   }
   __ bind(Done);
 }
 // Fast path for caching oop constants.
 void TemplateTable::fast_aldc(bool wide) {
   transition(vtos, atos);
   Register result = FSR;
   Register tmp = SSR;
   int index_size = wide ? sizeof(u2) : sizeof(u1);
   Label resolved;
  // We are resolved if the resolved reference cache entry contains a
  // non-null object (String, MethodType, etc.)
   assert_different_registers(result, tmp);
   __ get_cache_index_at_bcp(tmp, 1, index_size);
   __ load_resolved_reference_at_index(result, tmp);
   __ bne(result, R0, resolved);
   __ delayed()->nop();
   address entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_ldc);
   // first time invocation - must resolve first
   int i = (int)bytecode();
   __ move(tmp, i);
   __ call_VM(result, entry, tmp);
   __ bind(resolved);
   if (VerifyOops) {
     __ verify_oop(result);
   }
 }
 // used register: T2, T3, T1
 // T2 : index
 // T3 : cpool
 // T1 : tag
 void TemplateTable::ldc2_w() {
   transition(vtos, vtos);
   Label Long, Done;
   // get index in cpool
   __ get_2_byte_integer_at_bcp(T2, AT, 1);
   __ huswap(T2);
   __ get_cpool_and_tags(T3, T1);
   const int base_offset = ConstantPool::header_size() * wordSize;
   const int tags_offset = Array<u1>::base_offset_in_bytes();
   // get type in T1
   __ dadd(AT, T1, T2);
   __ lb(T1, AT, tags_offset);
   __ daddiu(AT, T1, - JVM_CONSTANT_Double);
   __ bne(AT, R0, Long);
   __ delayed()->dsll(T2, T2, Address::times_8);
   // dtos
   __ daddu(AT, T3, T2);
   __ ldc1(FSF, AT, base_offset + 0 * wordSize);
   __ sdc1(FSF, SP, - 2 * wordSize);
   __ b(Done);
   __ delayed()->daddi(SP, SP, - 2 * wordSize);
   // ltos
   __ bind(Long);
   __ dadd(AT, T3, T2);
   __ ld(FSR, AT, base_offset + 0 * wordSize);
   __ push(ltos);
   __ bind(Done);
 }
 // we compute the actual local variable address here
 // the x86 dont do so for it has scaled index memory access model, we dont have, so do here
 void TemplateTable::locals_index(Register reg, int offset) {
   __ lbu(reg, at_bcp(offset));
   __ dsll(reg, reg, Address::times_8);
   __ dsub(reg, LVP, reg);
 }
 // this method will do bytecode folding of the two form:
 // iload iload			iload caload
 // used register : T2, T3
 // T2 : bytecode
 // T3 : folded code
 void TemplateTable::iload() {
   transition(vtos, itos);
   if (RewriteFrequentPairs) {
     Label rewrite, done;
     // get the next bytecode in T2
     __ lbu(T2, at_bcp(Bytecodes::length_for(Bytecodes::_iload)));
     // if _iload, wait to rewrite to iload2.  We only want to rewrite the
     // last two iloads in a pair.  Comparing against fast_iload means that
     // the next bytecode is neither an iload or a caload, and therefore
     // an iload pair.
     __ move(AT, Bytecodes::_iload);
     __ beq(AT, T2, done);
     __ delayed()->nop();
     __ move(T3, Bytecodes::_fast_iload2);
     __ move(AT, Bytecodes::_fast_iload);
     __ beq(AT, T2, rewrite);
     __ delayed()->nop();
     // if _caload, rewrite to fast_icaload
     __ move(T3, Bytecodes::_fast_icaload);
     __ move(AT, Bytecodes::_caload);
     __ beq(AT, T2, rewrite);
     __ delayed()->nop();
     // rewrite so iload doesn't check again.
     __ move(T3, Bytecodes::_fast_iload);
     // rewrite
     // T3 : fast bytecode
     __ bind(rewrite);
     patch_bytecode(Bytecodes::_iload, T3, T2, false);
     __ bind(done);
   }
   // Get the local value into tos
   locals_index(T2);
   __ lw(FSR, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::fast_iload2() {
 	transition(vtos, itos);
 	locals_index(T2);
 	__ lw(FSR, T2, 0);
 	__ push(itos);
 	locals_index(T2, 3);
 	__ lw(FSR, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::fast_iload() {
   transition(vtos, itos);
   locals_index(T2);
   __ lw(FSR, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::lload() {
   transition(vtos, ltos);
   locals_index(T2);
   __ ld(FSR, T2, -wordSize);
   __ ld(SSR, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::fload() {
   transition(vtos, ftos);
   locals_index(T2);
 //FIXME, aoqi. How should the high 32bits be when store a single float into a 64bits register.
   //__ mtc1(R0, FSF);
   __ lwc1(FSF, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::dload() {
   transition(vtos, dtos);
   locals_index(T2);
 /*  if (TaggedStackInterpreter) {
     // Get double out of locals array, onto temp stack and load with
     // float instruction into ST0
     __ dsll(AT,T2,Interpreter::stackElementScale());
     __ dadd(AT, LVP, AT);
     __ ldc1(FSF, AT, Interpreter::local_offset_in_bytes(1));
   } else {*/
     __ ldc1(FSF, T2, -wordSize);
     __ ldc1(SSF, T2, 0);
  // }
 }
 // used register T2
 // T2 : index
 void TemplateTable::aload()
 {
   transition(vtos, atos);
   locals_index(T2);
   __ ld(FSR, T2, 0);
 }
 void TemplateTable::locals_index_wide(Register reg) {
   __ get_2_byte_integer_at_bcp(reg, AT, 2);
   __ huswap(reg);
   __ dsll(reg, reg, Address::times_8);
   __ dsub(reg, LVP, reg);
 }
 // used register T2
 // T2 : index
 void TemplateTable::wide_iload() {
 	transition(vtos, itos);
 	locals_index_wide(T2);
 	__ ld(FSR, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::wide_lload() {
 	transition(vtos, ltos);
 	locals_index_wide(T2);
 	__ ld(FSR, T2, -4);
 }
 // used register T2
 // T2 : index
 void TemplateTable::wide_fload() {
 	transition(vtos, ftos);
 	locals_index_wide(T2);
 	__ lwc1(FSF, T2, 0);
 }
 // used register T2
 // T2 : index
 void TemplateTable::wide_dload() {
 	transition(vtos, dtos);
 	locals_index_wide(T2);
 /*	if (TaggedStackInterpreter) {
 		// Get double out of locals array, onto temp stack and load with
 		// float instruction into ST0
 		//   __ movl(eax, laddress(ebx));
 		//  __ movl(edx, haddress(ebx));
 		__ dsll(AT,T2,Interpreter::stackElementScale());
 		__ dadd(AT, LVP, AT);
 		__ ldc1(FSF, AT, Interpreter::local_offset_in_bytes(1));
 		//  __ pushl(edx);  // push hi first
 		//  __ pushl(eax);
 		//  __ fld_d(Address(esp));
 		//  __ addl(esp, 2*wordSize);
 	} else {*/
 		__ ldc1(FSF, T2, -4);
 	//}
 }
 // used register T2
 // T2 : index
 void TemplateTable::wide_aload() {
 	transition(vtos, atos);
 	locals_index_wide(T2);
 	__ ld(FSR, T2, 0);
 }
 // we use A2 as the regiser for index, BE CAREFUL!
 // we dont use our tge 29 now, for later optimization
 void TemplateTable::index_check(Register array, Register index) {
   // Pop ptr into array
   __ pop_ptr(array);
   index_check_without_pop(array, index);
 }
 void TemplateTable::index_check_without_pop(Register array, Register index) {
   // destroys ebx
   // check array
   __ null_check(array, arrayOopDesc::length_offset_in_bytes());
   // check index
   Label ok;
   __ lw(AT, array, arrayOopDesc::length_offset_in_bytes());
 #ifndef OPT_RANGECHECK
   __ sltu(AT, index, AT);
   __ bne(AT, R0, ok);
   __ delayed()->nop();
   //throw_ArrayIndexOutOfBoundsException assume abberrant index in A2
   if (A2 != index) __ move(A2, index);
   __ jmp(Interpreter::_throw_ArrayIndexOutOfBoundsException_entry);
   __ delayed()->nop();
   __ bind(ok);
 #else
   __ lw(AT, array, arrayOopDesc::length_offset_in_bytes());
   __ move(A2, index);
   __ tgeu(A2, AT, 29);
 #endif
 }
 void TemplateTable::iaload() {
   transition(itos, itos);
   //  __ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(FSR, FSR, 2);
   __ dadd(FSR, SSR, FSR);
   //FSR: index
   __ lw(FSR, FSR, arrayOopDesc::base_offset_in_bytes(T_INT));
 }
 void TemplateTable::laload() {
   transition(itos, ltos);
   //  __ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(AT, FSR, Address::times_8);
   __ dadd(AT, SSR, AT);
   __ ld(FSR, AT, arrayOopDesc::base_offset_in_bytes(T_LONG) + 0 * wordSize);
 }
 void TemplateTable::faload() {
 	transition(itos, ftos);
 	// __ pop(SSR);
 	index_check(SSR, FSR);
 	__ shl(FSR, 2);
 	__ dadd(FSR, SSR, FSR);
 	__ lwc1(FSF, FSR, arrayOopDesc::base_offset_in_bytes(T_FLOAT));
 }
 void TemplateTable::daload() {
 	transition(itos, dtos);
 	//__ pop(SSR);
 	index_check(SSR, FSR);
 	__ dsll(AT, FSR, 3);
 	__ dadd(AT, SSR, AT);
 	__ ldc1(FSF, AT, arrayOopDesc::base_offset_in_bytes(T_DOUBLE) + 0 * wordSize);
 }
 void TemplateTable::aaload() {
   transition(itos, atos);
   //__ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(FSR, FSR, UseCompressedOops ? Address::times_4 : Address::times_8);
   __ dadd(FSR, SSR, FSR);
   //add for compressedoops
   __ load_heap_oop(FSR, Address(FSR, arrayOopDesc::base_offset_in_bytes(T_OBJECT)));
 }
 void TemplateTable::baload() {
   transition(itos, itos);
   //__ pop(SSR);
   index_check(SSR, FSR);
   __ dadd(FSR, SSR, FSR);
   __ lb(FSR, FSR, arrayOopDesc::base_offset_in_bytes(T_BYTE));
 }
 void TemplateTable::caload() {
   transition(itos, itos);
   // __ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(FSR, FSR, Address::times_2);
   __ dadd(FSR, SSR, FSR);
   __ lhu(FSR, FSR,  arrayOopDesc::base_offset_in_bytes(T_CHAR));
 }
 // iload followed by caload frequent pair
 // used register : T2
 // T2 : index
 void TemplateTable::fast_icaload() {
   transition(vtos, itos);
   // load index out of locals
   locals_index(T2);
   __ lw(FSR, T2, 0);
   //	__ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(FSR, FSR, 1);
   __ dadd(FSR, SSR, FSR);
   __ lhu(FSR, FSR,  arrayOopDesc::base_offset_in_bytes(T_CHAR));
 }
 void TemplateTable::saload() {
   transition(itos, itos);
   // __ pop(SSR);
   index_check(SSR, FSR);
   __ dsll(FSR, FSR, Address::times_2);
   __ dadd(FSR, SSR, FSR);
   __ lh(FSR, FSR,  arrayOopDesc::base_offset_in_bytes(T_SHORT));
 }
 void TemplateTable::iload(int n) {
 	transition(vtos, itos);
 	__ lw(FSR, iaddress(n));
 }
 void TemplateTable::lload(int n) {
 	transition(vtos, ltos);
 	__ ld(FSR, laddress(n));
 }
 void TemplateTable::fload(int n) {
   transition(vtos, ftos);
   //__ mtc1(R0, FSF);
   __ lwc1(FSF, faddress(n));
 }
 //FIXME here
 void TemplateTable::dload(int n) {
 	transition(vtos, dtos);
 	__ ldc1(FSF, laddress(n));
 }
 void TemplateTable::aload(int n) {
   transition(vtos, atos);
   __ ld(FSR, aaddress(n));
 }
 // used register : T2, T3
 // T2 : bytecode
 // T3 : folded code
 void TemplateTable::aload_0() {
 	transition(vtos, atos);
 	// According to bytecode histograms, the pairs:
 	//
 	// _aload_0, _fast_igetfield
 	// _aload_0, _fast_agetfield
 	// _aload_0, _fast_fgetfield
 	//
 	// occur frequently. If RewriteFrequentPairs is set, the (slow) _aload_0
 	// bytecode checks if the next bytecode is either _fast_igetfield,
 	// _fast_agetfield or _fast_fgetfield and then rewrites the
 	// current bytecode into a pair bytecode; otherwise it rewrites the current
 	// bytecode into _fast_aload_0 that doesn't do the pair check anymore.
 	//
 	// Note: If the next bytecode is _getfield, the rewrite must be delayed,
 	//       otherwise we may miss an opportunity for a pair.
 	//
 	// Also rewrite frequent pairs
 	//   aload_0, aload_1
 	//   aload_0, iload_1
 	// These bytecodes with a small amount of code are most profitable to rewrite
 	if (RewriteFrequentPairs) {
 		Label rewrite, done;
 		// get the next bytecode in T2
 		__ lbu(T2, at_bcp(Bytecodes::length_for(Bytecodes::_aload_0)));
 		// do actual aload_0
 		aload(0);
 		// if _getfield then wait with rewrite
 		__ move(AT, Bytecodes::_getfield);
 		__ beq(AT, T2, done);
 		__ delayed()->nop();
 		// if _igetfield then reqrite to _fast_iaccess_0
 		assert(Bytecodes::java_code(Bytecodes::_fast_iaccess_0) ==
 				Bytecodes::_aload_0, "fix bytecode definition");
 		__ move(T3, Bytecodes::_fast_iaccess_0);
 		__ move(AT, Bytecodes::_fast_igetfield);
 		__ beq(AT, T2, rewrite);
 		__ delayed()->nop();
 		// if _agetfield then reqrite to _fast_aaccess_0
 		assert(Bytecodes::java_code(Bytecodes::_fast_aaccess_0) ==
 				Bytecodes::_aload_0, "fix bytecode definition");
 		__ move(T3, Bytecodes::_fast_aaccess_0);
 		__ move(AT, Bytecodes::_fast_agetfield);
 		__ beq(AT, T2, rewrite);
 		__ delayed()->nop();
 		// if _fgetfield then reqrite to _fast_faccess_0
 		assert(Bytecodes::java_code(Bytecodes::_fast_faccess_0) ==
 				Bytecodes::_aload_0, "fix bytecode definition");
 		__ move(T3, Bytecodes::_fast_faccess_0);
 		__ move(AT, Bytecodes::_fast_fgetfield);
 		__ beq(AT, T2, rewrite);
 		__ delayed()->nop();
 		// else rewrite to _fast_aload0
 		assert(Bytecodes::java_code(Bytecodes::_fast_aload_0) ==
 				Bytecodes::_aload_0, "fix bytecode definition");
 		__ move(T3, Bytecodes::_fast_aload_0);
 		// rewrite
 		__ bind(rewrite);
 		patch_bytecode(Bytecodes::_aload_0, T3, T2, false);
 		__ bind(done);
 	} else {
 		aload(0);
 	}
 }
 void TemplateTable::istore() {
 	transition(itos, vtos);
 	locals_index(T2);
 	__ sw(FSR, T2, 0);
 }
 void TemplateTable::lstore() {
   transition(ltos, vtos);
   locals_index(T2);
   __ sd(FSR, T2, -wordSize);
 }
 void TemplateTable::fstore() {
 	transition(ftos, vtos);
 	locals_index(T2);
 	__ swc1(FSF, T2, 0);
 }
 void TemplateTable::dstore() {
   transition(dtos, vtos);
   locals_index(T2);
   __ sdc1(FSF, T2, -wordSize);
 }
 void TemplateTable::astore() {
   transition(vtos, vtos);
   //  __ pop(FSR);
   __ pop_ptr(FSR);
   locals_index(T2);
   __ sd(FSR, T2, 0);
 }
 void TemplateTable::wide_istore() {
 	transition(vtos, vtos);
 	//  __ pop(FSR);
 	__ pop_i(FSR);
 	locals_index_wide(T2);
 	__ sd(FSR, T2, 0);
 }
 void TemplateTable::wide_lstore() {
 	transition(vtos, vtos);
 	//__ pop2(FSR, SSR);
 	//__ pop_l(FSR, SSR);
 	__ pop_l(FSR); //aoqi:FIXME Is this right?
 	locals_index_wide(T2);
 	__ sd(FSR, T2, -4);
 }
 void TemplateTable::wide_fstore() {
 	wide_istore();
 }
 void TemplateTable::wide_dstore() {
 	wide_lstore();
 }
 void TemplateTable::wide_astore() {
 	transition(vtos, vtos);
 	__ pop_ptr(FSR);
 	locals_index_wide(T2);
 	__ sd(FSR, T2, 0);
 }
 // used register : T2
 void TemplateTable::iastore() {
   transition(itos, vtos);
   __ pop_i(SSR);
   index_check(T2, SSR);  // prefer index in ebx
   __ dsll(SSR, SSR, Address::times_4);
   __ dadd(T2, T2, SSR);
   __ sw(FSR, T2, arrayOopDesc::base_offset_in_bytes(T_INT));
 }
 // used register T2, T3
 void TemplateTable::lastore() {
   transition(ltos, vtos);
   __ pop_i (T2);
   index_check(T3, T2);
   __ dsll(T2, T2, Address::times_8);
   __ dadd(T3, T3, T2);
   __ sd(FSR, T3, arrayOopDesc::base_offset_in_bytes(T_LONG) + 0 * wordSize);
 }
 // used register T2
 void TemplateTable::fastore() {
   transition(ftos, vtos);
   __ pop_i(SSR);
   index_check(T2, SSR);
   __ dsll(SSR, SSR, Address::times_4);
   __ dadd(T2, T2, SSR);
   __ swc1(FSF, T2, arrayOopDesc::base_offset_in_bytes(T_FLOAT));
 }
 // used register T2, T3
 void TemplateTable::dastore() {
   transition(dtos, vtos);
   __ pop_i (T2);
   index_check(T3, T2);
   __ dsll(T2, T2, Address::times_8);
   __ daddu(T3, T3, T2);
   __ sdc1(FSF, T3, arrayOopDesc::base_offset_in_bytes(T_DOUBLE) + 0 * wordSize);
 }
 // used register : T2, T3, T8
 // T2 : array
 // T3 : subklass
 // T8 : supklass
 void TemplateTable::aastore() {
   Label is_null, ok_is_subtype, done;
   transition(vtos, vtos);
   // stack: ..., array, index, value
   __ ld(FSR, at_tos());     // Value
   __ lw(SSR, at_tos_p1());  // Index
   __ ld(T2, at_tos_p2());  // Array
   // index_check(T2, SSR);
   index_check_without_pop(T2, SSR);
   // do array store check - check for NULL value first
   __ beq(FSR, R0, is_null);
   __ delayed()->nop();
   // Move subklass into T3
   //__ ld(T3,  Address(FSR, oopDesc::klass_offset_in_bytes()));
   //add for compressedoops
   __ load_klass(T3, FSR);
   // Move superklass into T8
   //__ ld(T8, Address(T2, oopDesc::klass_offset_in_bytes()));
   //add for compressedoops
   __ load_klass(T8, T2);
   __ ld(T8, Address(T8,  ObjArrayKlass::element_klass_offset()));
   // Compress array+index*4+12 into a single register. T2
   __ dsll(AT, SSR, UseCompressedOops? Address::times_4 : Address::times_8);
   __ dadd(T2, T2, AT);
   __ daddi(T2, T2, arrayOopDesc::base_offset_in_bytes(T_OBJECT));
   // Generate subtype check.
   // Superklass in T8.  Subklass in T3.
   __ gen_subtype_check(T8, T3, ok_is_subtype);				// <-- Jin
   // Come here on failure
   // object is at FSR
   __ jmp(Interpreter::_throw_ArrayStoreException_entry);    // <-- Jin
   __ delayed()->nop();
   // Come here on success
   __ bind(ok_is_subtype);
   //replace with do_oop_store->store_heap_oop
   //__ sd(FSR, T2, 0);
   __ store_heap_oop(Address(T2, 0), FSR);					// <-- Jin
   __ sync();
   __ store_check(T2);
   __ b(done);
   __ delayed()->nop();
   // Have a NULL in FSR, EDX=T2, SSR=index.  Store NULL at ary[idx]
   __ bind(is_null);
   __ profile_null_seen(T9);
   __ dsll(AT, SSR, UseCompressedOops? Address::times_4 : Address::times_8);
   __ dadd(T2, T2, AT);
   //__ sd(FSR, T2, arrayOopDesc::base_offset_in_bytes(T_OBJECT));
   __ store_heap_oop(Address(T2, arrayOopDesc::base_offset_in_bytes(T_OBJECT)), FSR);	/* FSR is null here */
   __ sync();
   __ bind(done);
   __ daddi(SP, SP, 3 * Interpreter::stackElementSize);
 }
 void TemplateTable::bastore() {
   transition(itos, vtos);
   __ pop_i (SSR);
   index_check(T2, SSR);
   __ dadd(SSR, T2, SSR);
   __ sb(FSR, SSR, arrayOopDesc::base_offset_in_bytes(T_BYTE));
 }
 void TemplateTable::castore() {
   transition(itos, vtos);
   __ pop_i(SSR);
   index_check(T2, SSR);
   __ dsll(SSR, SSR, Address::times_2);
   __ dadd(SSR, T2, SSR);
   __ sh(FSR, SSR, arrayOopDesc::base_offset_in_bytes(T_CHAR));
 }
 void TemplateTable::sastore() {
   castore();
 }
 void TemplateTable::istore(int n) {
   transition(itos, vtos);
   __ sw(FSR, iaddress(n));
 }
 void TemplateTable::lstore(int n) {
   transition(ltos, vtos);
   __ sd(FSR, laddress(n));
 }
 void TemplateTable::fstore(int n) {
   transition(ftos, vtos);
   __ swc1(FSF, faddress(n));
 }
 void TemplateTable::dstore(int n) {
   transition(dtos, vtos);
   __ sdc1(FSF, laddress(n));
 }
 void TemplateTable::astore(int n) {
   transition(vtos, vtos);
   __ pop_ptr(FSR);
   __ sd(FSR, aaddress(n));
 }
 void TemplateTable::pop() {
   transition(vtos, vtos);
   __ daddi(SP, SP, Interpreter::stackElementSize);
 }
 void TemplateTable::pop2() {
   transition(vtos, vtos);
   __ daddi(SP, SP, 2 * Interpreter::stackElementSize);
 }
 void TemplateTable::dup() {
   transition(vtos, vtos);
   // stack: ..., a
   __ load_ptr(0, FSR);
   __ push_ptr(FSR);
   // stack: ..., a, a
 }
 // blows FSR
 void TemplateTable::dup_x1() {
 	transition(vtos, vtos);
 	// stack: ..., a, b
 	__ load_ptr(0, FSR);  // load b
 	__ load_ptr(1, A5);  // load a
 	__ store_ptr(1, FSR); // store b
 	__ store_ptr(0, A5); // store a
 	__ push_ptr(FSR);             // push b
 	// stack: ..., b, a, b
 }
 // blows FSR
 void TemplateTable::dup_x2() {
 	transition(vtos, vtos);
 	// stack: ..., a, b, c
 	__ load_ptr(0, FSR);  // load c
 	__ load_ptr(2, A5);  // load a
 	__ store_ptr(2, FSR); // store c in a
 	__ push_ptr(FSR);             // push c
 	// stack: ..., c, b, c, c
 	__ load_ptr(2, FSR);  // load b
 	__ store_ptr(2, A5); // store a in b
 	// stack: ..., c, a, c, c
 	__ store_ptr(1, FSR); // store b in c
 	// stack: ..., c, a, b, c
 }
 // blows FSR
 void TemplateTable::dup2() {
 	transition(vtos, vtos);
 	// stack: ..., a, b
 	__ load_ptr(1, FSR);  // load a
 	__ push_ptr(FSR);             // push a
 	__ load_ptr(1, FSR);  // load b
 	__ push_ptr(FSR);             // push b
 	// stack: ..., a, b, a, b
 }
 // blows FSR
 void TemplateTable::dup2_x1() {
 	transition(vtos, vtos);
 	// stack: ..., a, b, c
 	__ load_ptr(0, T2);  // load c
 	__ load_ptr(1, FSR);  // load b
 	__ push_ptr(FSR);             // push b
 	__ push_ptr(T2);             // push c
 	// stack: ..., a, b, c, b, c
 	__ store_ptr(3, T2); // store c in b
 	// stack: ..., a, c, c, b, c
 	__ load_ptr(4, T2);  // load a
 	__ store_ptr(2, T2); // store a in 2nd c
 	// stack: ..., a, c, a, b, c
 	__ store_ptr(4, FSR); // store b in a
 	// stack: ..., b, c, a, b, c
 	// stack: ..., b, c, a, b, c
 }
 // blows FSR, SSR
 void TemplateTable::dup2_x2() {
 	transition(vtos, vtos);
 	// stack: ..., a, b, c, d
 	// stack: ..., a, b, c, d
 	__ load_ptr(0, T2);  // load d
 	__ load_ptr(1, FSR);  // load c
 	__ push_ptr(FSR);             // push c
 	__ push_ptr(T2);             // push d
 	// stack: ..., a, b, c, d, c, d
 	__ load_ptr(4, FSR);  // load b
 	__ store_ptr(2, FSR); // store b in d
 	__ store_ptr(4, T2); // store d in b
 	// stack: ..., a, d, c, b, c, d
 	__ load_ptr(5, T2);  // load a
 	__ load_ptr(3, FSR);  // load c
 	__ store_ptr(3, T2); // store a in c
 	__ store_ptr(5, FSR); // store c in a
 	// stack: ..., c, d, a, b, c, d
 	// stack: ..., c, d, a, b, c, d
 }
 // blows FSR
 void TemplateTable::swap() {
 	transition(vtos, vtos);
 	// stack: ..., a, b
 	__ load_ptr(1, A5);  // load a
 	__ load_ptr(0, FSR);  // load b
 	__ store_ptr(0, A5); // store a in b
 	__ store_ptr(1, FSR); // store b in a
 	// stack: ..., b, a
 }
 void TemplateTable::iop2(Operation op) {
 	transition(itos, itos);
 	switch (op) {
 		case add  :
 			__ pop_i(SSR);
 			__ addu32(FSR, SSR, FSR);
 			break;
 		case sub  :
 			__ pop_i(SSR);
 			__ subu32(FSR, SSR, FSR);
 			break;
 		case mul  :
 			__ lw(SSR, SP, 0);
 			__ daddi(SP, SP, wordSize);
                         __ mul(FSR, SSR, FSR);
 			break;
 		case _and :
 			__ pop_i(SSR);
 			__ andr(FSR, SSR, FSR);
 			break;
 		case _or  :
 			__ pop_i(SSR);
 			__ orr(FSR, SSR, FSR);
 			break;
 		case _xor :
 			__ pop_i(SSR);
 			__ xorr(FSR, SSR, FSR);
 			break;
 		case shl  :
 			__ pop_i(SSR);
 			__ sllv(FSR, SSR, FSR);
 			break; // implicit masking of lower 5 bits by Intel shift instr. mips also
 		case shr  :
 			__ pop_i(SSR);
 			__ srav(FSR, SSR, FSR);
 			break; // implicit masking of lower 5 bits by Intel shift instr. mips also
 		case ushr :
 			__ pop_i(SSR);
 			__ srlv(FSR, SSR, FSR);
 			break; // implicit masking of lower 5 bits by Intel shift instr. mips also
 		default   : ShouldNotReachHere();
 	}
 }
 // the result stored in FSR, SSR,
 // used registers : T2, T3
 //FIXME, aoqi
 void TemplateTable::lop2(Operation op) {
   transition(ltos, ltos);
   //__ pop2(T2, T3);
   __ pop_l(T2, T3);
 #ifdef ASSERT
   {
     Label  L;
     __ beq(T3, R0, L);
     __ delayed()->nop();
     // FIXME: stack verification required
 //    __ stop("lop2, wrong stack");  // <--- Fu 20130930
     __ bind(L);
   }
 #endif
   switch (op) {
     case add :
       __ daddu(FSR, T2, FSR);
       //__ sltu(AT, FSR, T2);
       //__ daddu(SSR, T3, SSR);
       //__ daddu(SSR, SSR, AT);
       break;
     case sub :
       __ dsubu(FSR, T2, FSR);
       //__ sltu(AT, T2, FSR);
       //__ dsubu(SSR, T3, SSR);
       //__ dsubu(SSR, SSR, AT);
       break;
     case _and:
       __ andr(FSR, T2, FSR);
       //__ andr(SSR, T3, SSR);
       break;
     case _or :
       __ orr(FSR, T2, FSR);
       //__ orr(SSR, T3, SSR);
       break;
     case _xor:
       __ xorr(FSR, T2, FSR);
       //__ xorr(SSR, T3, SSR);
       break;
     default : ShouldNotReachHere();
   }
 }
 // java require this bytecode could handle 0x80000000/-1, dont cause a overflow exception,
 // the result is 0x80000000
 // the godson2 cpu do the same, so we need not handle this specially like x86
 void TemplateTable::idiv() {
 	transition(itos, itos);
 	Label not_zero;
 	__ bne(FSR, R0, not_zero);
 	__ delayed()->nop();
 	__ jmp(Interpreter::_throw_ArithmeticException_entry);
 	__ delayed()->nop();
 	__ bind(not_zero);
 	__ pop_i(SSR);
         if (UseLoongsonISA) {
           __ gsdiv(FSR, SSR, FSR);
         } else {
 	  __ div(SSR, FSR);
 	  __ mflo(FSR);
         }
 }
 void TemplateTable::irem() {
 	transition(itos, itos);
 	Label not_zero;
 	//__ pop(SSR);
 	__ pop_i(SSR);
 	__ div(SSR, FSR);
 	__ bne(FSR, R0, not_zero);
 	__ delayed()->nop();
 	//__ brk(7);
 	__ jmp(Interpreter::_throw_ArithmeticException_entry);
 	__ delayed()->nop();
 	__ bind(not_zero);
 	__ mfhi(FSR);
 }
 void TemplateTable::lmul() {
   transition(ltos, ltos);
   __ pop_l(T2);
   if(UseLoongsonISA){
     __ gsdmult(FSR, T2, FSR);
   } else {
       __ dmult(T2, FSR);
       __ mflo(FSR);
   }
 }
 // NOTE: i DONT use the Interpreter::_throw_ArithmeticException_entry
 void TemplateTable::ldiv() {
   transition(ltos, ltos);
   Label normal;
   __ bne(FSR, R0, normal);
   __ delayed()->nop();
   //__ brk(7);		//generate FPE
   __ jmp(Interpreter::_throw_ArithmeticException_entry);
   __ delayed()->nop();
   __ bind(normal);
   __ pop_l(A2, A3);
   if (UseLoongsonISA) {
     __ gsddiv(FSR, A2, FSR);
   } else {
     __ ddiv(A2, FSR);
     __ mflo(FSR);
   }
 }
 // NOTE: i DONT use the Interpreter::_throw_ArithmeticException_entry
 void TemplateTable::lrem() {
   transition(ltos, ltos);
   Label normal;
   __ bne(FSR, R0, normal);
   __ delayed()->nop();
   __ jmp(Interpreter::_throw_ArithmeticException_entry);
   __ delayed()->nop();
   __ bind(normal);
   __ pop_l (A2, A3);
   if(UseLoongsonISA){
     __ gsdmod(FSR, A2, FSR);
   } else {
     __ ddiv(A2, FSR);
     __ mfhi(FSR);
   }
 }
 // result in FSR
 // used registers : T0
 void TemplateTable::lshl() {
   transition(itos, ltos);
   __ pop_l(T0, T1);
 #ifdef ASSERT
   {
     Label  L;
     __ beq(T1, R0, L);
     __ delayed()->nop();
     //__ stop("lshl, wrong stack");  // <-- Fu 20130930
     __ bind(L);
   }
 #endif
   __ andi(FSR, FSR, 0x3f);	      // the bit to be shifted
   __ dsllv(FSR, T0, FSR);
 }
 // used registers : T0
 void TemplateTable::lshr() {
   transition(itos, ltos);
   __ pop_l(T0, T1);
 #ifdef ASSERT
   {
     Label  L;
     __ beq(T1, R0, L);
     __ delayed()->nop();
     __ stop("lshr, wrong stack");
     __ bind(L);
   }
 #endif
   __ andi(FSR, FSR, 0x3f);				// the bit to be shifted
   __ dsrav(FSR, T0, FSR);
 }
 // used registers : T0
 void TemplateTable::lushr() {
   transition(itos, ltos);
   __ pop_l(T0, T1);
 #ifdef ASSERT
   {
     Label  L;
     __ beq(T1, R0, L);
     __ delayed()->nop();
     __ stop("lushr, wrong stack");
     __ bind(L);
   }
 #endif
   __ andi(FSR, FSR, 0x3f);				// the bit to be shifted
   __ dsrlv(FSR, T0, FSR);
 }
 // result in FSF
 void TemplateTable::fop2(Operation op) {
 	transition(ftos, ftos);
 	__ pop_ftos_to_esp();  // pop ftos into esp
 	switch (op) {
 		case add:
 			__ lwc1(FTF, at_sp());
 			__ add_s(FSF, FTF, FSF);
 			break;
 		case sub:
 			__ lwc1(FTF, at_sp());
 			__ sub_s(FSF, FTF, FSF);
 			break;
 		case mul:
 			__ lwc1(FTF, at_sp());
 			__ mul_s(FSF, FTF, FSF);
 			break;
 		case div:
 			__ lwc1(FTF, at_sp());
 			__ div_s(FSF, FTF, FSF);
 			break;
 		case rem:
 			__ mfc1(FSR, FSF);
 			__ mtc1(FSR, F12);
 			__ lwc1(FTF, at_sp());
 			__ rem_s(FSF, FTF, F12, FSF);
 			break;
 		default : ShouldNotReachHere();
 	}
 	__ daddi(SP, SP, 1 * wordSize);
 }
 // result in SSF||FSF
 // i dont handle the strict flags
 void TemplateTable::dop2(Operation op) {
 	transition(dtos, dtos);
 	__ pop_dtos_to_esp();  // pop dtos into esp
 	switch (op) {
 		case add:
 			__ ldc1(FTF, at_sp());
 			__ add_d(FSF, FTF, FSF);
 			break;
 		case sub:
 			__ ldc1(FTF, at_sp());
 			__ sub_d(FSF, FTF, FSF);
 			break;
 		case mul:
 			__ ldc1(FTF, at_sp());
 			__ mul_d(FSF, FTF, FSF);
 			break;
 		case div:
 			__ ldc1(FTF, at_sp());
 			__ div_d(FSF, FTF, FSF);
 			break;
 		case rem:
 			__ dmfc1(FSR, FSF);
 			__ dmtc1(FSR, F12);
 			__ ldc1(FTF, at_sp());
 			__ rem_d(FSF, FTF, F12, FSF);
 			break;
 		default : ShouldNotReachHere();
 	}
 	__ daddi(SP, SP, 2 * wordSize);
 }
 void TemplateTable::ineg() {
 	transition(itos, itos);
 	__ neg(FSR);
 }
 void TemplateTable::lneg() {
 	transition(ltos, ltos);
 	__ dsubu(FSR, R0, FSR);
 }
 /*
 // Note: 'double' and 'long long' have 32-bits alignment on x86.
 static jlong* double_quadword(jlong *adr, jlong lo, jlong hi) {
   // Use the expression (adr)&(~0xF) to provide 128-bits aligned address
   // of 128-bits operands for SSE instructions.
   jlong *operand = (jlong*)(((intptr_t)adr)&((intptr_t)(~0xF)));
   // Store the value to a 128-bits operand.
   operand[0] = lo;
   operand[1] = hi;
   return operand;
 }
 // Buffer for 128-bits masks used by SSE instructions.
 static jlong float_signflip_pool[2*2];
 static jlong double_signflip_pool[2*2];
 */
 void TemplateTable::fneg() {
 	transition(ftos, ftos);
 	__ neg_s(FSF, FSF);
 }
 void TemplateTable::dneg() {
 	transition(dtos, dtos);
 	__ neg_d(FSF, FSF);
 }
 // used registers : T2
 void TemplateTable::iinc() {
 	transition(vtos, vtos);
 	locals_index(T2);
 	__ lw(FSR, T2, 0);
 	__ lb(AT, at_bcp(2));           // get constant
 	__ daddu(FSR, FSR, AT);
 	__ sw(FSR, T2, 0);
 }
 // used register : T2
 void TemplateTable::wide_iinc() {
 	transition(vtos, vtos);
 	locals_index_wide(T2);
 	__ get_2_byte_integer_at_bcp(FSR, AT, 4);
 	__ hswap(FSR);
 	__ lw(AT, T2, 0);
 	__ daddu(FSR, AT, FSR);
 	__ sw(FSR, T2, 0);
 }
 void TemplateTable::convert() {
   // Checking
 #ifdef ASSERT
   { TosState tos_in  = ilgl;
     TosState tos_out = ilgl;
     switch (bytecode()) {
       case Bytecodes::_i2l: // fall through
       case Bytecodes::_i2f: // fall through
       case Bytecodes::_i2d: // fall through
       case Bytecodes::_i2b: // fall through
       case Bytecodes::_i2c: // fall through
       case Bytecodes::_i2s: tos_in = itos; break;
       case Bytecodes::_l2i: // fall through
       case Bytecodes::_l2f: // fall through
       case Bytecodes::_l2d: tos_in = ltos; break;
       case Bytecodes::_f2i: // fall through
       case Bytecodes::_f2l: // fall through
       case Bytecodes::_f2d: tos_in = ftos; break;
       case Bytecodes::_d2i: // fall through
       case Bytecodes::_d2l: // fall through
       case Bytecodes::_d2f: tos_in = dtos; break;
       default             : ShouldNotReachHere();
     }
     switch (bytecode()) {
       case Bytecodes::_l2i: // fall through
       case Bytecodes::_f2i: // fall through
       case Bytecodes::_d2i: // fall through
       case Bytecodes::_i2b: // fall through
       case Bytecodes::_i2c: // fall through
       case Bytecodes::_i2s: tos_out = itos; break;
       case Bytecodes::_i2l: // fall through
       case Bytecodes::_f2l: // fall through
       case Bytecodes::_d2l: tos_out = ltos; break;
       case Bytecodes::_i2f: // fall through
       case Bytecodes::_l2f: // fall through
       case Bytecodes::_d2f: tos_out = ftos; break;
       case Bytecodes::_i2d: // fall through
       case Bytecodes::_l2d: // fall through
       case Bytecodes::_f2d: tos_out = dtos; break;
       default             : ShouldNotReachHere();
     }
     transition(tos_in, tos_out);
   }
 #endif // ASSERT
   // Conversion
   // (Note: use pushl(ecx)/popl(ecx) for 1/2-word stack-ptr manipulation)
   switch (bytecode()) {
     case Bytecodes::_i2l:
       //__ extend_sign(SSR, FSR);
       __ sll(FSR, FSR, 0);
       break;
     case Bytecodes::_i2f:
       __ mtc1(FSR, FSF);
       __ cvt_s_w(FSF, FSF);
       break;
     case Bytecodes::_i2d:
       __ mtc1(FSR, FSF);
       __ cvt_d_w(FSF, FSF);
       break;
     case Bytecodes::_i2b:
       __ dsll32(FSR, FSR, 24);
       __ dsra32(FSR, FSR, 24);
       break;
     case Bytecodes::_i2c:
       __ andi(FSR, FSR, 0xFFFF);  // truncate upper 56 bits
       break;
     case Bytecodes::_i2s:
       __ dsll32(FSR, FSR, 16);
       __ dsra32(FSR, FSR, 16);
       break;
     case Bytecodes::_l2i:
       __ dsll32(FSR, FSR, 0);
       __ dsra32(FSR, FSR, 0);
       break;
     case Bytecodes::_l2f:
       __ dmtc1(FSR, FSF);
       //__ mtc1(SSR, SSF);
       __ cvt_s_l(FSF, FSF);
       break;
     case Bytecodes::_l2d:
       __ dmtc1(FSR, FSF);
       //__ mtc1(SSR, SSF);
       __ cvt_d_l(FSF, FSF);
       break;
     case Bytecodes::_f2i:
       {
 	Label L;
 	/*
 	__ c_un_s(FSF, FSF);		//NaN?
 	__ bc1t(L);
 	__ delayed(); __ move(FSR, R0);
 	*/
 	__ trunc_w_s(F12, FSF);
 	__ cfc1(AT, 31);
 	__ li(T0, 0x10000);
 	__ andr(AT, AT, T0);
 	__ beq(AT, R0, L);
 	__ delayed()->mfc1(FSR, F12);
 	__ mov_s(F12, FSF);
 	__ call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::f2i), 1);
 	__ bind(L);
       }
       break;
     case Bytecodes::_f2l:
       {
 	Label L;
 	/*
 	__ move(SSR, R0);
 	__ c_un_s(FSF, FSF);		//NaN?
 	__ bc1t(L);
 	__ delayed();
 	__ move(FSR, R0);
 	*/
 	__ trunc_l_s(F12, FSF);
 	__ cfc1(AT, 31);
 	__ li(T0, 0x10000);
 	__ andr(AT, AT, T0);
 	__ beq(AT, R0, L);
 	__ delayed()->dmfc1(FSR, F12);
 	__ mov_s(F12, FSF);
 	__ call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::f2l), 1);
 	__ bind(L);
       }
       break;
     case Bytecodes::_f2d:
       __ cvt_d_s(FSF, FSF);
       break;
     case Bytecodes::_d2i:
       {
 	Label L;
 	/*
 	__ c_un_d(FSF, FSF);		//NaN?
 	__ bc1t(L);
 	__ delayed(); __ move(FSR, R0);
 	*/
 	__ trunc_w_d(F12, FSF);
 	__ cfc1(AT, 31);
 	__ li(T0, 0x10000);
 	__ andr(AT, AT, T0);
 	__ beq(AT, R0, L);
 	__ delayed()->mfc1(FSR, F12);
 	__ mov_d(F12, FSF);
 	__ call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::d2i), 1);
 	__ bind(L);
       }
       break;
     case Bytecodes::_d2l:
       {
 	Label L;
 	/*
 	__ move(SSR, R0);
 	__ c_un_d(FSF, FSF);		//NaN?
 	__ bc1t(L);
 	__ delayed(); __ move(FSR, R0);
 	*/
 	__ trunc_l_d(F12, FSF);
 	__ cfc1(AT, 31);
 	__ li(T0, 0x10000);
 	__ andr(AT, AT, T0);
 	__ beq(AT, R0, L);
 	__ delayed()->dmfc1(FSR, F12);
 	__ mov_d(F12, FSF);
 	__ call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::d2l), 1);
 	__ bind(L);
       }
       break;
     case Bytecodes::_d2f:
       __ cvt_s_d(FSF, FSF);
       break;
     default             :
       ShouldNotReachHere();
   }
 }
 void TemplateTable::lcmp() {
   transition(ltos, itos);
   Label low, high, done;
   __ pop(T0);
   __ pop(R0);
   __ slt(AT, T0, FSR);
   __ bne(AT, R0, low);
   __ delayed()->nop();
   __ bne(T0, FSR, high);
   __ delayed()->nop();
   __ li(FSR, (long)0);
   __ b(done);
   __ delayed()->nop();
   __ bind(low);
   __ li(FSR, (long)-1);
   __ b(done);
   __ delayed()->nop();
   __ bind(high);
   __ li(FSR, (long)1);
   __ b(done);
   __ delayed()->nop();
   __ bind(done);
 }
 void TemplateTable::float_cmp(bool is_float, int unordered_result) {
 	Label less, done;
 	__ move(FSR, R0);
 	if (is_float) {
 		__ pop_ftos_to_esp();
 		__ lwc1(FTF, at_sp());
 		__ c_eq_s(FTF, FSF);
 		__ bc1t(done);
 		__ delayed()->daddi(SP, SP, 1 * wordSize);
 		if (unordered_result<0)
 			__ c_ult_s(FTF, FSF);
 		else
 			__ c_olt_s(FTF, FSF);
 	} else {
 		__ pop_dtos_to_esp();
 		__ ldc1(FTF, at_sp());
 		__ c_eq_d(FTF, FSF);
 		__ bc1t(done);
 		__ delayed()->daddi(SP, SP, 2 * wordSize);
 		if (unordered_result<0)
 			__ c_ult_d(FTF, FSF);
 		else
 			__ c_olt_d(FTF, FSF);
 	}
 	__ bc1t(less);
 	__ delayed()->nop();
 	__ move(FSR, 1);
 	__ b(done);
 	__ delayed()->nop();
 	__ bind(less);
 	__ move(FSR, -1);
 	__ bind(done);
 }
 // used registers : T3, A7, Rnext
 // FSR : return bci, this is defined by the vm specification
 // T2 : MDO taken count
 // T3 : method
 // A7 : offset
 // Rnext : next bytecode, this is required by dispatch_base
 void TemplateTable::branch(bool is_jsr, bool is_wide) {
   __ get_method(T3);
   __ profile_taken_branch(A7, T2);		// only C2 meaningful
 #ifndef CORE
   const ByteSize be_offset = MethodCounters::backedge_counter_offset()
     + InvocationCounter::counter_offset();
   const ByteSize inv_offset = MethodCounters::invocation_counter_offset()
     + InvocationCounter::counter_offset();
   const int method_offset = frame::interpreter_frame_method_offset * wordSize;
 #endif // CORE
   // Load up T4 with the branch displacement
   if (!is_wide) {
     __ get_2_byte_integer_at_bcp(A7, AT, 1);
     __ hswap(A7);
   } else {
     __ get_4_byte_integer_at_bcp(A7, AT, 1);
     __ swap(A7);
   }
   // Handle all the JSR stuff here, then exit.
   // It's much shorter and cleaner than intermingling with the
   // non-JSR normal-branch stuff occuring below.
   if (is_jsr) {
     // Pre-load the next target bytecode into Rnext
     __ dadd(AT, BCP, A7);
     __ lbu(Rnext, AT, 0);
     // compute return address as bci in FSR
     __ daddi(FSR, BCP, (is_wide?5:3) - in_bytes(ConstMethod::codes_offset()));
     __ ld(AT, T3, in_bytes(Method::const_offset()));
     __ dsub(FSR, FSR, AT);
     // Adjust the bcp in BCP by the displacement in A7
     __ dadd(BCP, BCP, A7);
     // jsr returns atos that is not an oop
     // __ dispatch_only_noverify(atos);
     // Push return address
     __ push_i(FSR);
     // jsr returns vtos
     __ dispatch_only_noverify(vtos);
     return;
   }
   // Normal (non-jsr) branch handling
   // Adjust the bcp in S0 by the displacement in T4
   __ dadd(BCP, BCP, A7);
 #ifdef CORE
   // Pre-load the next target bytecode into EBX
   __ lbu(Rnext, BCP, 0);
   // continue with the bytecode @ target
   __ dispatch_only(vtos);
 #else
   assert(UseLoopCounter || !UseOnStackReplacement, "on-stack-replacement requires loop counters");
   Label backedge_counter_overflow;
   Label profile_method;
   Label dispatch;
   if (UseLoopCounter) {
     // increment backedge counter for backward branches
     // eax: MDO
     // ebx: MDO bumped taken-count
     // T3: method
     // T4: target offset
     // BCP: target bcp
     // LVP: locals pointer
     __ bgtz(A7, dispatch);	// check if forward or backward branch
     __ delayed()->nop();
     // check if MethodCounters exists
     Label has_counters;
     __ ld(AT, T3, in_bytes(Method::method_counters_offset()));  // use AT as MDO, TEMP
     __ bne(AT, R0, has_counters);
     __ nop();
     //__ push(T3);
     //__ push(A7);
     __ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::build_method_counters),
                T3);
     //__ pop(A7);
     //__ pop(T3);
     __ ld(AT, T3, in_bytes(Method::method_counters_offset()));  // use AT as MDO, TEMP
     __ beq(AT, R0, dispatch);
     __ nop();
     __ bind(has_counters);
     // increment back edge counter
     __ ld(T1, T3, in_bytes(Method::method_counters_offset()));
     __ lw(T0, T1, in_bytes(be_offset));
     __ increment(T0, InvocationCounter::count_increment);
     __ sw(T0, T1, in_bytes(be_offset));
     // load invocation counter
     __ lw(T1, T1, in_bytes(inv_offset));
     // buffer bit added, mask no needed
     // by yjl 10/24/2005
     //__ move(AT, InvocationCounter::count_mask_value);
     //__ andr(T1, T1, AT);
     // dadd backedge counter & invocation counter
     __ dadd(T1, T1, T0);
     if (ProfileInterpreter) {
       // Test to see if we should create a method data oop
       //__ lui(AT, Assembler::split_high(int(&InvocationCounter::InterpreterProfileLimit)));
       //__ lw(AT, AT, Assembler::split_low(int(&InvocationCounter::InterpreterProfileLimit)));
       // T1 : backedge counter & invocation counter
       __ li(AT, (long)&InvocationCounter::InterpreterProfileLimit);
       __ lw(AT, AT, 0);
       __ slt(AT, T1, AT);
       __ bne(AT, R0, dispatch);
       __ delayed()->nop();
       // if no method data exists, go to profile method
       __ test_method_data_pointer(T1, profile_method);
       if (UseOnStackReplacement) {
 	// check for overflow against ebx which is the MDO taken count
 	//__ lui(AT, Assembler::split_high(int(&InvocationCounter::InterpreterBackwardBranchLimit)));
 	//__ lw(AT, AT, Assembler::split_low(int(&InvocationCounter::InterpreterBackwardBranchLimit)));
 	__ li(AT, (long)&InvocationCounter::InterpreterBackwardBranchLimit);
 	__ lw(AT, AT, 0);
 	// the value Rnext Is get from the beginning profile_taken_branch
 	__ slt(AT, T2, AT);
 	__ bne(AT, R0, dispatch);
 	__ delayed()->nop();
 	// When ProfileInterpreter is on, the backedge_count comes
 	// from the methodDataOop, which value does not get reset on
 	// the call to  frequency_counter_overflow().
 	// To avoid excessive calls to the overflow routine while
 	// the method is being compiled, dadd a second test to make
 	// sure the overflow function is called only once every
 	// overflow_frequency.
 	const int overflow_frequency = 1024;
 	__ andi(AT, T2, overflow_frequency-1);
 	__ beq(AT, R0, backedge_counter_overflow);
 	__ delayed()->nop();
       }
     } else {
       if (UseOnStackReplacement) {
 	// check for overflow against eax, which is the sum of the counters
 	//__ lui(AT, Assembler::split_high(int(&InvocationCounter::InterpreterBackwardBranchLimit)));
 	//__ lw(AT, AT, Assembler::split_low(int(&InvocationCounter::InterpreterBackwardBranchLimit)));
 	__ li(AT, (long)&InvocationCounter::InterpreterBackwardBranchLimit);
 	__ lw(AT, AT, 0);
 	__ slt(AT, T1, AT);
 	__ beq(AT, R0, backedge_counter_overflow);
 	__ delayed()->nop();
       }
     }
     __ bind(dispatch);
   }
   // Pre-load the next target bytecode into Rnext
   __ lbu(Rnext, BCP, 0);
   // continue with the bytecode @ target
   // FSR: return bci for jsr's, unused otherwise
   // Rnext: target bytecode
   // BCP: target bcp
   __ dispatch_only(vtos);
   if (UseLoopCounter) {
     if (ProfileInterpreter) {
       // Out-of-line code to allocate method data oop.
       __ bind(profile_method);
       __ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::profile_method));
       __ lbu(Rnext, BCP, 0);
       __ set_method_data_pointer_for_bcp();
 /*
       __ ld(T3, FP, method_offset);
       __ lw(T3, T3, in_bytes(Method::method_data_offset()));
       __ sw(T3, FP, frame::interpreter_frame_mdx_offset * wordSize);
       __ test_method_data_pointer(T3, dispatch);
       // offset non-null mdp by MDO::data_offset() + IR::profile_method()
       __ daddi(T3, T3, in_bytes(MethodData::data_offset()));
       __ dadd(T3, T3, T1);
       __ sw(T3, FP, frame::interpreter_frame_mdx_offset * wordSize);
 */
       __ b(dispatch);
       __ delayed()->nop();
     }
     if (UseOnStackReplacement) {
       // invocation counter overflow
       __ bind(backedge_counter_overflow);
       __ sub(A7, BCP, A7);	// branch bcp
       call_VM(NOREG, CAST_FROM_FN_PTR(address,
 	    InterpreterRuntime::frequency_counter_overflow), A7);
       __ lbu(Rnext, BCP, 0);
       // V0: osr nmethod (osr ok) or NULL (osr not possible)
       // V1: osr adapter frame return address
       // Rnext: target bytecode
       // LVP: locals pointer
       // BCP: bcp
       __ beq(V0, R0, dispatch);
       __ delayed()->nop();
       // nmethod may have been invalidated (VM may block upon call_VM return)
       __ lw(T3, V0, nmethod::entry_bci_offset());
       __ move(AT, InvalidOSREntryBci);
       __ beq(AT, T3, dispatch);
       __ delayed()->nop();
       // We need to prepare to execute the OSR method. First we must
       // migrate the locals and monitors off of the stack.
       //eax V0: osr nmethod (osr ok) or NULL (osr not possible)
       //ebx V1: osr adapter frame return address
       //edx  Rnext: target bytecode
       //edi  LVP: locals pointer
       //esi  BCP: bcp
       __ move(BCP, V0);
       // const Register thread = ecx;
       const Register thread = TREG;
 #ifndef OPT_THREAD
       __ get_thread(thread);
 #endif
       call_VM(noreg, CAST_FROM_FN_PTR(address,
 	    SharedRuntime::OSR_migration_begin));
       // eax is OSR buffer, move it to expected parameter location
       //refer to osrBufferPointer in c1_LIRAssembler_mips.cpp
       __ move(T0, V0);
       // pop the interpreter frame
       //  __ movl(edx, Address(ebp, frame::interpreter_frame_sender_sp_offset
       //  * wordSize)); // get sender sp
       __ ld(A7, Address(FP,
 	    frame::interpreter_frame_sender_sp_offset * wordSize));
       //FIXME, shall we keep the return address on the stack?
       __ leave();                                // remove frame anchor
       // __ popl(edi);                         // get return address
       //__ daddi(SP, SP, wordSize);               // get return address
       //   __ pop(LVP);
       __ move(LVP, RA);
       // __ movl(esp, edx);                         // set sp to sender sp
       __ move(SP, A7);
       Label skip;
       Label chkint;
       // The interpreter frame we have removed may be returning to
       // either the callstub or the interpreter. Since we will
       // now be returning from a compiled (OSR) nmethod we must
       // adjust the return to the return were it can handler compiled
       // results and clean the fpu stack. This is very similar to
       // what a i2c adapter must do.
       // Are we returning to the call stub?
 #if 0
       // __ cmpl(edi, (int)StubRoutines::_call_stub_return_address);
       __ daddi(AT, LVP, -(int)StubRoutines::_call_stub_return_address);
       //  __ jcc(Assembler::notEqual, chkint);
       __ bne(AT, R0, chkint);
       __ delayed()->nop();
       // yes adjust to the specialized call stub  return.
       // assert(StubRoutines::i486::get_call_stub_compiled_return() != NULL,
       // "must be set");
       assert(StubRoutines::gs2::get_call_stub_compiled_return() != NULL,
 	  "must be set");
       // __ movl(edi, (intptr_t) StubRoutines::i486::get_call_stub_compiled_return());
       __ move(LVP, (intptr_t) StubRoutines::gs2::get_call_stub_compiled_return());
       //  __ jmp(skip);
       __ b(skip);
       __ delayed()->nop();
       __ bind(chkint);
       // Are we returning to the interpreter? Look for sentinel
       //__ cmpl(Address(edi, -8), Interpreter::return_sentinel);
       __ lw(AT, LVP , -8);
       __ daddi(AT, AT, -Interpreter::return_sentinel);
       //__ jcc(Assembler::notEqual, skip);
       __ bne(AT, R0, skip);
       __ delayed()->nop();
       // Adjust to compiled return back to interpreter
       // __ movl(edi, Address(edi, -4));
       __ lw(LVP, LVP, -4);
       __ bind(skip);
 #endif
       // Align stack pointer for compiled code (note that caller is
       // responsible for undoing this fixup by remembering the old SP
       // in an ebp-relative location)
       //  __ andl(esp, -(StackAlignmentInBytes));
       __ move(AT, -(StackAlignmentInBytes));
       __ andr(SP , SP , AT);
       // push the (possibly adjusted) return address
       //  __ pushl(edi);
       //__ push(LVP);
       //			__ move(RA, LVP);
       // and begin the OSR nmethod
       //  __ jmp(Address(esi, nmethod::osr_entry_point_offset()));
       //refer to osr_entry in c1_LIRAssembler_mips.cpp
       __ ld(AT, BCP, nmethod::osr_entry_point_offset());
       __ jr(AT);
       __ delayed()->nop();
     }
   }
 #endif // not CORE
 }
 void TemplateTable::if_0cmp(Condition cc) {
   transition(itos, vtos);
   // assume branch is more often taken than not (loops use backward branches)
   Label not_taken;
   switch(cc) {
     case not_equal:
       __ beq(FSR, R0, not_taken);
       break;
     case equal:
       __ bne(FSR, R0, not_taken);
       break;
     case less:
       __ bgez(FSR, not_taken);
       break;
     case less_equal:
       __ bgtz(FSR, not_taken);
       break;
     case greater:
       __ blez(FSR, not_taken);
       break;
     case greater_equal:
       __ bltz(FSR, not_taken);
       break;
   }
   __ delayed()->nop();
   branch(false, false);
   __ bind(not_taken);
   __ profile_not_taken_branch(FSR);
 }
 void TemplateTable::if_icmp(Condition cc) {
   transition(itos, vtos);
   // assume branch is more often taken than not (loops use backward branches)
   Label not_taken;
   __ pop_i(SSR);
   switch(cc) {
     case not_equal:
       __ beq(SSR, FSR, not_taken);
       break;
     case equal:
       __ bne(SSR, FSR, not_taken);
       break;
     case less:
       __ slt(AT, SSR, FSR);
       __ beq(AT, R0, not_taken);
       break;
     case less_equal:
       __ slt(AT, FSR, SSR);
       __ bne(AT, R0, not_taken);
       break;
     case greater:
       __ slt(AT, FSR, SSR);
       __ beq(AT, R0, not_taken);
       break;
     case greater_equal:
       __ slt(AT, SSR, FSR);
       __ bne(AT, R0, not_taken);
       break;
   }
   __ delayed()->nop();
   branch(false, false);
   __ bind(not_taken);
   __ profile_not_taken_branch(FSR);
 }
 void TemplateTable::if_nullcmp(Condition cc) {
   transition(atos, vtos);
   // assume branch is more often taken than not (loops use backward branches)
   Label not_taken;
   switch(cc) {
     case not_equal:
       __ beq(FSR, R0, not_taken);
       break;
     case equal:
       __ bne(FSR, R0, not_taken);
       break;
     default:
       ShouldNotReachHere();
   }
   __ delayed()->nop();
   branch(false, false);
   __ bind(not_taken);
   __ profile_not_taken_branch(FSR);
 }
 void TemplateTable::if_acmp(Condition cc) {
 	transition(atos, vtos);
 	// assume branch is more often taken than not (loops use backward branches)
 	Label not_taken;
 	//	__ lw(SSR, SP, 0);
 	__ pop_ptr(SSR);
 	switch(cc) {
 		case not_equal:
 			__ beq(SSR, FSR, not_taken);
 			break;
 		case equal:
 			__ bne(SSR, FSR, not_taken);
 			break;
 		default:
 			ShouldNotReachHere();
 	}
 	//	__ delayed()->daddi(SP, SP, 4);
 	__ delayed()->nop();
 	branch(false, false);
 	__ bind(not_taken);
 	__ profile_not_taken_branch(FSR);
 }
 // used registers : T1, T2, T3
 // T1 : method
 // T2 : returb bci
 void TemplateTable::ret() {
 	transition(vtos, vtos);
 	locals_index(T2);
 	__ ld(T2, T2, 0);
 	__ profile_ret(T2, T3);
 	__ get_method(T1);
 	__ ld(BCP, T1, in_bytes(Method::const_offset()));
 	__ dadd(BCP, BCP, T2);
 	__ daddi(BCP, BCP, in_bytes(ConstMethod::codes_offset()));
 	__ dispatch_next(vtos);
 }
 // used registers : T1, T2, T3
 // T1 : method
 // T2 : returb bci
 void TemplateTable::wide_ret() {
 	transition(vtos, vtos);
 	locals_index_wide(T2);
 	__ ld(T2, T2, 0);                   // get return bci, compute return bcp
 	__ profile_ret(T2, T3);
 	__ get_method(T1);
 	__ ld(BCP, T1, in_bytes(Method::const_offset()));
 	__ dadd(BCP, BCP, T2);
 	__ daddi(BCP, BCP, in_bytes(ConstMethod::codes_offset()));
 	__ dispatch_next(vtos);
 }
 // used register T2, T3, A7, Rnext
 // T2 : bytecode pointer
 // T3 : low
 // A7 : high
 // Rnext : dest bytecode, required by dispatch_base
 void TemplateTable::tableswitch() {
 	Label default_case, continue_execution;
 	transition(itos, vtos);
 	// align BCP
 	__ daddi(T2, BCP, BytesPerInt);
 	__ li(AT, -BytesPerInt);
 	__ andr(T2, T2, AT);
 	// load lo & hi
 	__ lw(T3, T2, 1 * BytesPerInt);
 	__ swap(T3);
 	__ lw(A7, T2, 2 * BytesPerInt);
 	__ swap(A7);
 	// check against lo & hi
 	__ slt(AT, FSR, T3);
 	__ bne(AT, R0, default_case);
 	__ delayed()->nop();
 	__ slt(AT, A7, FSR);
 	__ bne(AT, R0, default_case);
 	__ delayed()->nop();
 	// lookup dispatch offset, in A7 big endian
 	__ dsub(FSR, FSR, T3);
 	__ dsll(AT, FSR, Address::times_4);
 	__ dadd(AT, T2, AT);
 	__ lw(A7, AT, 3 * BytesPerInt);
 	__ profile_switch_case(FSR, T9, T3);
 	__ bind(continue_execution);
 	__ swap(A7);
 	__ dadd(BCP, BCP, A7);
 	__ lbu(Rnext, BCP, 0);
 	__ dispatch_only(vtos);
 	// handle default
 	__ bind(default_case);
 	__ profile_switch_default(FSR);
 	__ lw(A7, T2, 0);
 	__ b(continue_execution);
 	__ delayed()->nop();
 }
 void TemplateTable::lookupswitch() {
 	transition(itos, itos);
 	__ stop("lookupswitch bytecode should have been rewritten");
 }
 // used registers : T2, T3, A7, Rnext
 // T2 : bytecode pointer
 // T3 : pair index
 // A7 : offset
 // Rnext : dest bytecode
 // the data after the opcode is the same as lookupswitch
 // see Rewriter::rewrite_method for more information
 void TemplateTable::fast_linearswitch() {
   transition(itos, vtos);
   Label loop_entry, loop, found, continue_execution;
   // swap eax so we can avoid swapping the table entries
   __ swap(FSR);
   // align BCP
   __ daddi(T2, BCP, BytesPerInt);
   __ li(AT, -BytesPerInt);
   __ andr(T2, T2, AT);
   // set counter
   __ lw(T3, T2, BytesPerInt);
   __ swap(T3);
   __ b(loop_entry);
   __ delayed()->nop();
   // table search
   __ bind(loop);
   // get the entry value
   __ dsll(AT, T3, Address::times_8);
   __ dadd(AT, T2, AT);
   __ lw(AT, AT, 2 * BytesPerInt);
   // found?
   __ beq(FSR, AT, found);
   __ delayed()->nop();
   __ bind(loop_entry);
   __ bgtz(T3, loop);
   __ delayed()->daddiu(T3, T3, -1);
   // default case
   __ profile_switch_default(FSR);
   __ lw(A7, T2, 0);
   __ b(continue_execution);
   __ delayed()->nop();
   // entry found -> get offset
   __ bind(found);
   __ dsll(AT, T3, Address::times_8);
   __ dadd(AT, T2, AT);
   __ lw(A7, AT, 3 * BytesPerInt);
   __ profile_switch_case(T3, FSR, T2);
   // continue execution
   __ bind(continue_execution);
   __ swap(A7);
   __ dadd(BCP, BCP, A7);
   __ lbu(Rnext, BCP, 0);
   __ dispatch_only(vtos);
 }
 // used registers : T0, T1, T2, T3, A7, Rnext
 // T2 : pairs address(array)
 // Rnext : dest bytecode
 // the data after the opcode is the same as lookupswitch
 // see Rewriter::rewrite_method for more information
 void TemplateTable::fast_binaryswitch() {
   transition(itos, vtos);
   // Implementation using the following core algorithm:
   //
   // int binary_search(int key, LookupswitchPair* array, int n) {
   //   // Binary search according to "Methodik des Programmierens" by
   //   // Edsger W. Dijkstra and W.H.J. Feijen, Addison Wesley Germany 1985.
   //   int i = 0;
   //   int j = n;
   //   while (i+1 < j) {
   //     // invariant P: 0 <= i < j <= n and (a[i] <= key < a[j] or Q)
   //     // with      Q: for all i: 0 <= i < n: key < a[i]
   //     // where a stands for the array and assuming that the (inexisting)
   //     // element a[n] is infinitely big.
   //     int h = (i + j) >> 1;
   //     // i < h < j
   //     if (key < array[h].fast_match()) {
   //       j = h;
   //     } else {
   //       i = h;
   //     }
   //   }
   //   // R: a[i] <= key < a[i+1] or Q
   //   // (i.e., if key is within array, i is the correct index)
   //   return i;
   // }
   // register allocation
   const Register array = T2;
   const Register i = T3, j = A7;
   const Register h = T1;
   const Register temp = T0;
   const Register key = FSR;
   // setup array
   __ daddi(array, BCP, 3*BytesPerInt);
   __ li(AT, -BytesPerInt);
   __ andr(array, array, AT);
   // initialize i & j
   __ move(i, R0);
   __ lw(j, array, - 1 * BytesPerInt);
   // Convert j into native byteordering
   __ swap(j);
   // and start
   Label entry;
   __ b(entry);
   __ delayed()->nop();
   // binary search loop
   {
     Label loop;
     __ bind(loop);
     // int h = (i + j) >> 1;
     __ dadd(h, i, j);
     __ dsrl(h, h, 1);
     // if (key < array[h].fast_match()) {
     //   j = h;
     // } else {
     //   i = h;
     // }
     // Convert array[h].match to native byte-ordering before compare
     __ dsll(AT, h, Address::times_8);
     __ dadd(AT, array, AT);
     __ lw(temp, AT, 0 * BytesPerInt);
     __ swap(temp);
     {
       Label set_i, end_of_if;
       __ slt(AT, key, temp);
       __ beq(AT, R0, set_i);
       __ delayed()->nop();
       __ b(end_of_if);
       __ delayed(); __ move(j, h);
       __ bind(set_i);
       __ move(i, h);
       __ bind(end_of_if);
     }
     // while (i+1 < j)
     __ bind(entry);
     __ daddi(h, i, 1);
     __ slt(AT, h, j);
     __ bne(AT, R0, loop);
     __ delayed()->nop();
   }
   // end of binary search, result index is i (must check again!)
   Label default_case;
   // Convert array[i].match to native byte-ordering before compare
   __ dsll(AT, i, Address::times_8);
   __ dadd(AT, array, AT);
   __ lw(temp, AT, 0 * BytesPerInt);
   __ swap(temp);
   __ bne(key, temp, default_case);
   __ delayed()->nop();
   // entry found -> j = offset
   __ dsll(AT, i, Address::times_8);
   __ dadd(AT, array, AT);
   __ lw(j, AT, 1 * BytesPerInt);
   __ profile_switch_case(i, key, array);
   __ swap(j);
   __ dadd(BCP, BCP, j);
   __ lbu(Rnext, BCP, 0);
   __ dispatch_only(vtos);
   // default case -> j = default offset
   __ bind(default_case);
   __ profile_switch_default(i);
   __ lw(j, array, - 2 * BytesPerInt);
   __ swap(j);
   __ dadd(BCP, BCP, j);
   __ lbu(Rnext, BCP, 0);
   __ dispatch_only(vtos);
 }
 void TemplateTable::_return(TosState state) {
   transition(state, state);
   assert(_desc->calls_vm(), "inconsistent calls_vm information"); // call in remove_activation
   if (_desc->bytecode() == Bytecodes::_return_register_finalizer) {
     assert(state == vtos, "only valid state");
     __ ld(T1, aaddress(0));
     //__ ld(LVP, T1, oopDesc::klass_offset_in_bytes());
     __ load_klass(LVP, T1);
     __ lw(LVP, LVP, in_bytes(Klass::access_flags_offset()));
     __ move(AT, JVM_ACC_HAS_FINALIZER);
     __ andr(AT, AT, LVP);//by_css
     Label skip_register_finalizer;
     __ beq(AT, R0, skip_register_finalizer);
     __ delayed()->nop();
     __ call_VM(noreg, CAST_FROM_FN_PTR(address,
 	  InterpreterRuntime::register_finalizer), T1);
     __ bind(skip_register_finalizer);
   }
   __ remove_activation(state, T9);
   __ sync();
   __ jr(T9);
   __ delayed()->nop();
 }
 // ----------------------------------------------------------------------------
 // Volatile variables demand their effects be made known to all CPU's
 // in order.  Store buffers on most chips allow reads & writes to
 // reorder; the JMM's ReadAfterWrite.java test fails in -Xint mode
 // without some kind of memory barrier (i.e., it's not sufficient that
 // the interpreter does not reorder volatile references, the hardware
 // also must not reorder them).
 //
 // According to the new Java Memory Model (JMM):
 // (1) All volatiles are serialized wrt to each other.  ALSO reads &
 //     writes act as aquire & release, so:
 // (2) A read cannot let unrelated NON-volatile memory refs that
 //     happen after the read float up to before the read.  It's OK for
 //     non-volatile memory refs that happen before the volatile read to
 //     float down below it.
 // (3) Similar a volatile write cannot let unrelated NON-volatile
 //     memory refs that happen BEFORE the write float down to after the
 //     write.  It's OK for non-volatile memory refs that happen after the
 //     volatile write to float up before it.
 //
 // We only put in barriers around volatile refs (they are expensive),
 // not _between_ memory refs (that would require us to track the
 // flavor of the previous memory refs).  Requirements (2) and (3)
 // require some barriers before volatile stores and after volatile
 // loads.  These nearly cover requirement (1) but miss the
 // volatile-store-volatile-load case.  This final case is placed after
 // volatile-stores although it could just as well go before
 // volatile-loads.
 //void TemplateTable::volatile_barrier(Assembler::Membar_mask_bits
 //                                     order_constraint) {
 void TemplateTable::volatile_barrier( ) {
   // Helper function to insert a is-volatile test and memory barrier
   //if (os::is_MP()) { // Not needed on single CPU
   //  __ membar(order_constraint);
   //}
 	if( !os::is_MP() ) return;	// Not needed on single CPU
 	__ sync();
 }
 // we dont shift left 2 bits in get_cache_and_index_at_bcp
 // for we always need shift the index we use it. the ConstantPoolCacheEntry
 // is 16-byte long, index is the index in
 // ConstantPoolCache, so cache + base_offset() + index * 16 is
 // the corresponding ConstantPoolCacheEntry
 // used registers : T2
 // NOTE : the returned index need also shift left 4 to get the address!
 void TemplateTable::resolve_cache_and_index(int byte_no,
                                             Register Rcache,
 					    Register index,
                                             size_t index_size) {
   assert(byte_no == f1_byte || byte_no == f2_byte, "byte_no out of range");
   const Register temp = A1;
   assert_different_registers(Rcache, index);
   const int shift_count = (1 + byte_no)*BitsPerByte;
   Label resolved;
   __ get_cache_and_index_and_bytecode_at_bcp(Rcache, index, temp, byte_no, 1, index_size);
   // is resolved?
   int i = (int)bytecode();
   __ addi(temp, temp, -i);
   __ beq(temp, R0, resolved);
   __ delayed()->nop();
   // resolve first time through
   address entry;
   switch (bytecode()) {
     case Bytecodes::_getstatic      : // fall through
     case Bytecodes::_putstatic      : // fall through
     case Bytecodes::_getfield       : // fall through
     case Bytecodes::_putfield       :
       entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_get_put);
       break;
     case Bytecodes::_invokevirtual  : // fall through
     case Bytecodes::_invokespecial  : // fall through
     case Bytecodes::_invokestatic   : // fall through
     case Bytecodes::_invokeinterface:
       entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invoke);
       break;
     case Bytecodes::_invokehandle:
       entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokehandle);
       break;
     case Bytecodes::_invokedynamic:
       entry = CAST_FROM_FN_PTR(address, InterpreterRuntime::resolve_invokedynamic);
       break;
     default                      		:
       fatal(err_msg("unexpected bytecode: %s", Bytecodes::name(bytecode())));
   }
   __ move(temp, i);
   __ call_VM(NOREG, entry, temp);
   // Update registers with resolved info
   __ get_cache_and_index_at_bcp(Rcache, index, 1, index_size);
   __ bind(resolved);
 }
 // The Rcache and index registers must be set before call
 void TemplateTable::load_field_cp_cache_entry(Register obj,
                                               Register cache,
                                               Register index,
                                               Register off,
                                               Register flags,
                                               bool is_static = false) {
   assert_different_registers(cache, index, flags, off);
   ByteSize cp_base_offset = ConstantPoolCache::base_offset();
   // Field offset
   __ dsll(AT, index, Address::times_ptr);
   __ dadd(AT, cache, AT);
   __ ld(off, AT, in_bytes(cp_base_offset + ConstantPoolCacheEntry::f2_offset()));
   // Flags
   __ ld(flags, AT, in_bytes(cp_base_offset + ConstantPoolCacheEntry::flags_offset()));
   // klass     overwrite register
   if (is_static) {
     __ ld(obj, AT, in_bytes(cp_base_offset + ConstantPoolCacheEntry::f1_offset()));
     const int mirror_offset = in_bytes(Klass::java_mirror_offset());
     __ ld(obj, Address(obj, mirror_offset));
     __ verify_oop(obj);
   }
 }
 // get the method, itable_index and flags of the current invoke
 void TemplateTable::load_invoke_cp_cache_entry(int byte_no,
                                                Register method,
                                                Register itable_index,
                                                Register flags,
                                                bool is_invokevirtual,
                                                bool is_invokevfinal, /*unused*/
                                                bool is_invokedynamic) {
   // setup registers
   const Register cache = T3;
   const Register index = T1;
   assert_different_registers(method, flags);
   assert_different_registers(method, cache, index);
   assert_different_registers(itable_index, flags);
   assert_different_registers(itable_index, cache, index);
   assert(is_invokevirtual == (byte_no == f2_byte), "is invokevirtual flag redundant");
   // determine constant pool cache field offsets
   const int method_offset = in_bytes(
       ConstantPoolCache::base_offset() +
       ((byte_no == f2_byte)
        ? ConstantPoolCacheEntry::f2_offset()
        : ConstantPoolCacheEntry::f1_offset()
       )
       );
   const int flags_offset = in_bytes(ConstantPoolCache::base_offset() +
       ConstantPoolCacheEntry::flags_offset());
   // access constant pool cache fields
   const int index_offset = in_bytes(ConstantPoolCache::base_offset() +
       ConstantPoolCacheEntry::f2_offset());
   size_t index_size = (is_invokedynamic ? sizeof(u4): sizeof(u2));
   resolve_cache_and_index(byte_no, cache, index, index_size);
   //assert(wordSize == 8, "adjust code below");
   // note we shift 4 not 2, for we get is the true inde
   // of ConstantPoolCacheEntry, not the shifted 2-bit index as x86 version
   __ dsll(AT, index, Address::times_ptr);
   __ dadd(AT, cache, AT);
   __ ld(method, AT, method_offset);
   if (itable_index != NOREG) {
     __ ld(itable_index, AT, index_offset);
   }
   __ ld(flags, AT, flags_offset);
 }
 // The registers cache and index expected to be set before call.
 // Correct values of the cache and index registers are preserved.
 void TemplateTable::jvmti_post_field_access(Register cache, Register index,
                                             bool is_static, bool has_tos) {
   // do the JVMTI work here to avoid disturbing the register state below
   // We use c_rarg registers here because we want to use the register used in
   // the call to the VM
 	if (JvmtiExport::can_post_field_access()) {
 		// Check to see if a field access watch has been set before we take
 		// the time to call into the VM.
 		Label L1;
 		assert_different_registers(cache, index, FSR);
 		__ li(AT, (intptr_t)JvmtiExport::get_field_access_count_addr());
 		__ lw(FSR, AT, 0);
 		__ beq(FSR, R0, L1);
 		__ delayed()->nop();
 		// We rely on the bytecode being resolved and the cpCache entry filled in.
 		// cache entry pointer
 		//__ get_cache_and_index_at_bcp(c_rarg2, c_rarg3, 1);
 		__ daddi(cache, cache, in_bytes(ConstantPoolCache::base_offset()));
 		__ shl(index, 4);
 		__ dadd(cache, cache, index);
 		if (is_static) {
 			__ move(FSR, R0);
 		} else {
 			__ lw(FSR, SP, 0);
 			__ verify_oop(FSR);
 		}
 		// FSR: object pointer or NULL
 		// cache: cache entry pointer
 		__ call_VM(NOREG, CAST_FROM_FN_PTR(address,
 					InterpreterRuntime::post_field_access), FSR, cache);
 		__ get_cache_and_index_at_bcp(cache, index, 1);
 		__ bind(L1);
 	}
 }
 void TemplateTable::pop_and_check_object(Register r) {
   __ pop_ptr(r);
   __ null_check(r);  // for field access must check obj.
   __ verify_oop(r);
 }
 // used registers : T1, T2, T3, T1
 // T1 : flags
 // T2 : off
 // T3 : obj
 // T1 : field address
 // The flags 31, 30, 29, 28 together build a 4 bit number 0 to 8 with the
 // following mapping to the TosState states:
 // btos: 0
 // ctos: 1
 // stos: 2
 // itos: 3
 // ltos: 4
 // ftos: 5
 // dtos: 6
 // atos: 7
 // vtos: 8
 // see ConstantPoolCacheEntry::set_field for more info
 void TemplateTable::getfield_or_static(int byte_no, bool is_static) {
   transition(vtos, vtos);
   const Register cache = T3;
   const Register index = T0;
   const Register obj   = T3;
   const Register off   = T2;
   const Register flags = T1;
   resolve_cache_and_index(byte_no, cache, index, sizeof(u2));
   //jvmti_post_field_access(cache, index, is_static, false);
   load_field_cp_cache_entry(obj, cache, index, off, flags, is_static);
   if (!is_static) pop_and_check_object(obj);
   __ dadd(index, obj, off);
   Label Done, notByte, notInt, notShort, notChar, notLong, notFloat, notObj, notDouble;
   assert(btos == 0, "change code, btos != 0");
   __ dsrl(flags, flags, ConstantPoolCacheEntry::tos_state_shift);
   __ andi(flags, flags, 0xf);
   __ bne(flags, R0, notByte);
   __ delayed()->nop();
   // btos
   __ sync();
   __ lb(FSR, index, 0);
   __ sd(FSR, SP, - wordSize);
   // Rewrite bytecode to be faster
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_bgetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notByte);
   __ move(AT, itos);
   __ bne(flags, AT, notInt);
   __ delayed()->nop();
   // itos
   __ sync();
   __ lw(FSR, index, 0);
   __ sd(FSR, SP, - wordSize);
   // Rewrite bytecode to be faster
   if (!is_static) {
     // patch_bytecode(Bytecodes::_fast_igetfield, T3, T2);
     patch_bytecode(Bytecodes::_fast_igetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notInt);
   __ move(AT, atos);
   __ bne(flags, AT, notObj);
   __ delayed()->nop();
   // atos
   //add for compressedoops
   __ sync();
   __ load_heap_oop(FSR, Address(index, 0));
   __ sd(FSR, SP, - wordSize);
   if (!is_static) {
     //patch_bytecode(Bytecodes::_fast_agetfield, T3, T2);
     patch_bytecode(Bytecodes::_fast_agetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notObj);
   __ move(AT, ctos);
   __ bne(flags, AT, notChar);
   __ delayed()->nop();
   // ctos
   __ sync();
   __ lhu(FSR, index, 0);
   __ sd(FSR, SP, - wordSize);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_cgetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notChar);
   __ move(AT, stos);
   __ bne(flags, AT, notShort);
   __ delayed()->nop();
   // stos
   __ sync();
   __ lh(FSR, index, 0);
   __ sd(FSR, SP, - wordSize);
   if (!is_static) {
     // patch_bytecode(Bytecodes::_fast_sgetfield, T3, T2);
     patch_bytecode(Bytecodes::_fast_sgetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notShort);
   __ move(AT, ltos);
   __ bne(flags, AT, notLong);
   __ delayed()->nop();
   // FIXME : the load/store should be atomic, we have no simple method to do this in mips32
   // ltos
   __ sync();
   __ ld(FSR, index, 0 * wordSize);
   __ sd(FSR, SP, -2 * wordSize);
   __ sd(R0, SP, -1 * wordSize);
   // Don't rewrite to _fast_lgetfield for potential volatile case.
   __ b(Done);
   __ delayed()->daddi(SP, SP, - 2 * wordSize);
   __ bind(notLong);
   __ move(AT, ftos);
   __ bne(flags, AT, notFloat);
   __ delayed()->nop();
   // ftos
   __ sync();
   __ lwc1(FSF, index, 0);
   __ sdc1(FSF, SP, - wordSize);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_fgetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - wordSize);
   __ bind(notFloat);
   __ move(AT, dtos);
   __ bne(flags, AT, notDouble);
   __ delayed()->nop();
   // dtos
   __ sync();
   __ ldc1(FSF, index, 0 * wordSize);
   __ sdc1(FSF, SP, - 2 * wordSize);
   __ sd(R0, SP, - 1 * wordSize);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_dgetfield, T3, T2);
   }
   __ b(Done);
   __ delayed()->daddi(SP, SP, - 2 * wordSize);
   __ bind(notDouble);
   __ stop("Bad state");
   __ bind(Done);
 }
 void TemplateTable::getfield(int byte_no) {
   getfield_or_static(byte_no, false);
 }
 void TemplateTable::getstatic(int byte_no) {
   getfield_or_static(byte_no, true);
 }
 /*
 // used registers : T1, T2, T3, T1
 // T1 : cache & cp entry
 // T2 : obj
 // T3 : flags & value pointer
 // T1 : index
 // see ConstantPoolCacheEntry::set_field for more info
 void TemplateTable::jvmti_post_field_mod(int byte_no, bool is_static) {
  */
 // The registers cache and index expected to be set before call.
 // The function may destroy various registers, just not the cache and index registers.
 void TemplateTable::jvmti_post_field_mod(Register cache, Register index, bool is_static) {
 	ByteSize cp_base_offset = ConstantPoolCache::base_offset();
 	if (JvmtiExport::can_post_field_modification()) {
 		// Check to see if a field modification watch has been set before we take
 		// the time to call into the VM.
 		Label L1;
 		assert_different_registers(cache, index, AT);
 		//__ lui(AT, Assembler::split_high((int)JvmtiExport::get_field_modification_count_addr()));
 		//__ lw(FSR, AT, Assembler::split_low((int)JvmtiExport::get_field_modification_count_addr()));
 		__ li(AT, JvmtiExport::get_field_modification_count_addr());
 		__ lw(FSR, AT, 0);
 		__ beq(FSR, R0, L1);
 		__ delayed()->nop();
 		/* // We rely on the bytecode being resolved and the cpCache entry filled in.
 		   resolve_cache_and_index(byte_no, T1, T1);
 		   */
 		// The cache and index registers have been already set.
 		// This allows to eliminate this call but the cache and index
 		// registers have to be correspondingly used after this line.
 		// __ get_cache_and_index_at_bcp(eax, edx, 1);
 		__ get_cache_and_index_at_bcp(T1, T9, 1);
 		if (is_static) {
 			__ move(T2, R0);
 		} else {
 			// Life is harder. The stack holds the value on top,
 			// followed by the object.
 			// We don't know the size of the value, though;
 			// it could be one or two words
 			// depending on its type. As a result, we must find
 			// the type to determine where the object is.
 			Label two_word, valsize_known;
 			__ dsll(AT, T1, 4);
 			__ dadd(AT, T1, AT);
 			__ lw(T3, AT, in_bytes(cp_base_offset
 						+ ConstantPoolCacheEntry::flags_offset()));
 			__ move(T2, SP);
 			__ shr(T3, ConstantPoolCacheEntry::tos_state_shift);
 			// Make sure we don't need to mask ecx for tos_state_shift
 			// after the above shift
 			ConstantPoolCacheEntry::verify_tos_state_shift();
 			__ move(AT, ltos);
 			__ beq(T3, AT, two_word);
 			__ delayed()->nop();
 			__ move(AT, dtos);
 			__ beq(T3, AT, two_word);
 			__ delayed()->nop();
 			__ b(valsize_known);
 			//__ delayed()->daddi(T2, T2, wordSize*1);
 			__ delayed()->daddi(T2, T2,Interpreter::expr_offset_in_bytes(1) );
 			__ bind(two_word);
 			//	__ daddi(T2, T2, wordSize*2);
 			__ daddi(T2, T2,Interpreter::expr_offset_in_bytes(2));
 			__ bind(valsize_known);
 			// setup object pointer
 			__ lw(T2, T2, 0*wordSize);
 		}
 		// cache entry pointer
 		__ daddi(T1, T1, in_bytes(cp_base_offset));
 		__ shl(T1, 4);
 		__ daddu(T1, T1, T1);
 		// object (tos)
 		__ move(T3, SP);
 		// T2: object pointer set up above (NULL if static)
 		// T1: cache entry pointer
 		// T3: jvalue object on the stack
 		__ call_VM(NOREG, CAST_FROM_FN_PTR(address,
 				InterpreterRuntime::post_field_modification), T2, T1, T3);
 		__ get_cache_and_index_at_bcp(cache, index, 1);
 		__ bind(L1);
 	}
 }
 // used registers : T0, T1, T2, T3, T8
 // T1 : flags
 // T2 : off
 // T3 : obj
 // T8 : volatile bit
 // see ConstantPoolCacheEntry::set_field for more info
 void TemplateTable::putfield_or_static(int byte_no, bool is_static) {
   transition(vtos, vtos);
   const Register cache = T3;
   const Register index = T0;
   const Register obj   = T3;
   const Register off   = T2;
   const Register flags = T1;
   const Register bc    = T3;
   resolve_cache_and_index(byte_no, cache, index, sizeof(u2));
   //TODO: LEE
   //jvmti_post_field_mod(cache, index, is_static);
   load_field_cp_cache_entry(obj, cache, index, off, flags, is_static);
   // Doug Lea believes this is not needed with current Sparcs (TSO) and Intel (PSO).
   // volatile_barrier( );
   Label notVolatile, Done;
   __ move(AT, 1<<ConstantPoolCacheEntry::is_volatile_shift);
   __ andr(T8, flags, AT);
   Label notByte, notInt, notShort, notChar, notLong, notFloat, notObj, notDouble;
   assert(btos == 0, "change code, btos != 0");
   // btos
   __ dsrl(flags, flags, ConstantPoolCacheEntry::tos_state_shift);
   __ andi(flags, flags, ConstantPoolCacheEntry::tos_state_mask);
   __ bne(flags, R0, notByte);
   __ delayed()->nop();
   __ pop(btos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sb(FSR, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_bputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notByte);
   // itos
   __ move(AT, itos);
   __ bne(flags, AT, notInt);
   __ delayed()->nop();
   __ pop(itos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sw(FSR, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_iputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notInt);
   // atos
   __ move(AT, atos);
   __ bne(flags, AT, notObj);
   __ delayed()->nop();
   __ pop(atos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   //__ sd(FSR, AT, 0);
   __ store_heap_oop(Address(AT, 0), FSR);
   __ sync();
   __ store_check(obj);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_aputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notObj);
   // ctos
   __ move(AT, ctos);
   __ bne(flags, AT, notChar);
   __ delayed()->nop();
   __ pop(ctos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sh(FSR, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_cputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notChar);
   // stos
   __ move(AT, stos);
   __ bne(flags, AT, notShort);
   __ delayed()->nop();
   __ pop(stos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sh(FSR, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_sputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notShort);
   // ltos
   __ move(AT, ltos);
   __ bne(flags, AT, notLong);
   __ delayed()->nop();
   // FIXME: there is no simple method to load/store 64-bit data in a atomic operation
   // we just ignore the volatile flag.
   //Label notVolatileLong;
   //__ beq(T1, R0, notVolatileLong);
   //__ delayed()->nop();
   //addent = 2 * wordSize;
   // no need
   //__ lw(FSR, SP, 0);
   //__ lw(SSR, SP, 1 * wordSize);
   //if (!is_static) {
   //	__ lw(T3, SP, addent);
   //	addent += 1 * wordSize;
   //	__ verify_oop(T3);
   //}
   //__ daddu(AT, T3, T2);
   // Replace with real volatile test
   // NOTE : we assume that sdc1&ldc1 operate in 32-bit, this is true for Godson2 even in 64-bit kernel
   // last modified by yjl 7/12/2005
   //__ ldc1(FSF, SP, 0);
   //__ sdc1(FSF, AT, 0);
   //volatile_barrier();
   // Don't rewrite volatile version
   //__ b(notVolatile);
   //__ delayed()->addiu(SP, SP, addent);
   //__ bind(notVolatileLong);
   //__ pop(ltos);  // overwrites edx
   //	__ lw(FSR, SP, 0 * wordSize);
   //	__ lw(SSR, SP, 1 * wordSize);
   //	__ daddi(SP, SP, 2*wordSize);
   __ pop(ltos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sd(FSR, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_lputfield, bc, off, true, byte_no);
   }
   __ b(notVolatile);
   __ delayed()->nop();
   __ bind(notLong);
   // ftos
   __ move(AT, ftos);
   __ bne(flags, AT, notFloat);
   __ delayed()->nop();
   __ pop(ftos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ swc1(FSF, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_fputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notFloat);
   // dtos
   __ move(AT, dtos);
   __ bne(flags, AT, notDouble);
   __ delayed()->nop();
   __ pop(dtos);
   if (!is_static) {
     pop_and_check_object(obj);
   }
   __ dadd(AT, obj, off);
   __ sdc1(FSF, AT, 0);
   if (!is_static) {
     patch_bytecode(Bytecodes::_fast_dputfield, bc, off, true, byte_no);
   }
   __ b(Done);
   __ delayed()->nop();
   __ bind(notDouble);
   __ stop("Bad state");
   __ bind(Done);
   // Check for volatile store
   __ beq(T8, R0, notVolatile);
   __ delayed()->nop();
   volatile_barrier( );
   __ bind(notVolatile);
 }
 void TemplateTable::putfield(int byte_no) {
   putfield_or_static(byte_no, false);
 }
 void TemplateTable::putstatic(int byte_no) {
   putfield_or_static(byte_no, true);
 }
 // used registers : T1, T2, T3
 // T1 : cp_entry
 // T2 : obj
 // T3 : value pointer
 void TemplateTable::jvmti_post_fast_field_mod() {
 	if (JvmtiExport::can_post_field_modification()) {
 		// Check to see if a field modification watch has been set before we take
 		// the time to call into the VM.
 		Label L2;
 		//__ lui(AT, Assembler::split_high((intptr_t)JvmtiExport::get_field_modification_count_addr()));
 		//__ lw(T3, AT, Assembler::split_low((intptr_t)JvmtiExport::get_field_modification_count_addr()));
 		__ li(AT, JvmtiExport::get_field_modification_count_addr());
 		__ lw(T3, AT, 0);
 		__ beq(T3, R0, L2);
 		__ delayed()->nop();
 		//__ pop(T2);
 		__ pop_ptr(T2);
 		//__ lw(T2, SP, 0);
 		__ verify_oop(T2);
 		__ push_ptr(T2);
 		__ li(AT, -sizeof(jvalue));
 		__ daddu(SP, SP, AT);
 		__ move(T3, SP);
 		//__ push(T2);
 		//__ move(T2, R0);
 		switch (bytecode()) {          // load values into the jvalue object
 			case Bytecodes::_fast_bputfield:
 				__ sb(FSR, SP, 0);
 				break;
 			case Bytecodes::_fast_sputfield:
 				__ sh(FSR, SP, 0);
 				break;
 			case Bytecodes::_fast_cputfield:
 				__ sh(FSR, SP, 0);
 				break;
 			case Bytecodes::_fast_iputfield:
 				__ sw(FSR, SP, 0);
 				break;
 			case Bytecodes::_fast_lputfield:
 				__ sd(FSR, SP, 0);
 				break;
 			case Bytecodes::_fast_fputfield:
 				__ swc1(FSF, SP, 0);
 				break;
 			case Bytecodes::_fast_dputfield:
 				__ sdc1(FSF, SP, 0);
 				break;
 			case Bytecodes::_fast_aputfield:
 				__ sd(FSR, SP, 0);
 				break;
 			default:  ShouldNotReachHere();
 		}
 		//__ pop(T2);  // restore copy of object pointer
 		// Save eax and sometimes edx because call_VM() will clobber them,
 		// then use them for JVM/DI purposes
 		__ push(FSR);
 		if (bytecode() == Bytecodes::_fast_lputfield) __ push(SSR);
 		// access constant pool cache entry
 		__ get_cache_entry_pointer_at_bcp(T1, T2, 1);
 		// no need, verified ahead
 		__ verify_oop(T2);
 		// ebx: object pointer copied above
 		// eax: cache entry pointer
 		// ecx: jvalue object on the stack
 		__ call_VM(NOREG, CAST_FROM_FN_PTR(address,
 					InterpreterRuntime::post_field_modification), T2, T1, T3);
 		if (bytecode() == Bytecodes::_fast_lputfield) __ pop(SSR);  // restore high value
 		//__ pop(FSR);     // restore lower value
 		//__ daddi(SP, SP, sizeof(jvalue));  // release jvalue object space
 		__ lw(FSR, SP, 0);
 		__ daddiu(SP, SP, sizeof(jvalue) + 1 * wordSize);
 		__ bind(L2);
 	}
 }
 // used registers : T2, T3, T1
 // T2 : index & off & field address
 // T3 : cache & obj
 // T1 : flags
 void TemplateTable::fast_storefield(TosState state) {
   transition(state, vtos);
   ByteSize base = ConstantPoolCache::base_offset();
   jvmti_post_fast_field_mod();
   // access constant pool cache
   __ get_cache_and_index_at_bcp(T3, T2, 1);
   // test for volatile with edx but edx is tos register for lputfield.
   __ dsll(AT, T2, Address::times_8);
   __ dadd(AT, T3, AT);
   __ ld(T1, AT, in_bytes(base + ConstantPoolCacheEntry::flags_offset()));
   // replace index with field offset from cache entry
   __ ld(T2, AT, in_bytes(base + ConstantPoolCacheEntry::f2_offset()));
   // Doug Lea believes this is not needed with current Sparcs (TSO) and Intel (PSO).
   // volatile_barrier( );
   Label notVolatile, Done;
   // Check for volatile store
   __ move(AT, 1<<ConstantPoolCacheEntry::is_volatile_shift);
   __ andr(AT, T1, AT);
   __ beq(AT, R0, notVolatile);
   __ delayed()->nop();
   // Get object from stack
   // NOTE : the value in FSR/FSF now
   //	__ pop(T3);
   //	__ verify_oop(T3);
   pop_and_check_object(T3);
   // field addresses
   __ dadd(T2, T3, T2);
   // access field
   switch (bytecode()) {
     case Bytecodes::_fast_bputfield:
       __ sb(FSR, T2, 0);
       break;
     case Bytecodes::_fast_sputfield: // fall through
     case Bytecodes::_fast_cputfield:
       __ sh(FSR, T2, 0);
       break;
     case Bytecodes::_fast_iputfield:
       __ sw(FSR, T2, 0);
       break;
     case Bytecodes::_fast_lputfield:
       __ sd(FSR, T2, 0 * wordSize);
       break;
     case Bytecodes::_fast_fputfield:
       __ swc1(FSF, T2, 0);
       break;
     case Bytecodes::_fast_dputfield:
       __ sdc1(FSF, T2, 0 * wordSize);
       break;
     case Bytecodes::_fast_aputfield:
       __ store_heap_oop(Address(T2, 0), FSR);
       __ sync();
       __ store_check(T3);
       break;
     default:
       ShouldNotReachHere();
   }
   Label done;
   volatile_barrier( );
   __ b(done);
   __ delayed()->nop();
   // Same code as above, but don't need edx to test for volatile.
   __ bind(notVolatile);
   // Get object from stack
   //	__ pop(T3);
   //	__ verify_oop(T3);
   pop_and_check_object(T3);
   //get the field address
   __ dadd(T2, T3, T2);
   // access field
   switch (bytecode()) {
     case Bytecodes::_fast_bputfield:
       __ sb(FSR, T2, 0);
       break;
     case Bytecodes::_fast_sputfield: // fall through
     case Bytecodes::_fast_cputfield:
       __ sh(FSR, T2, 0);
       break;
     case Bytecodes::_fast_iputfield:
       __ sw(FSR, T2, 0);
       break;
     case Bytecodes::_fast_lputfield:
       __ sd(FSR, T2, 0 * wordSize);
       break;
     case Bytecodes::_fast_fputfield:
       __ swc1(FSF, T2, 0);
       break;
     case Bytecodes::_fast_dputfield:
       __ sdc1(FSF, T2, 0 * wordSize);
       break;
     case Bytecodes::_fast_aputfield:
       //add for compressedoops
       __ store_heap_oop(Address(T2, 0), FSR);
       __ sync();
       __ store_check(T3);
       break;
     default:
       ShouldNotReachHere();
   }
   __ bind(done);
 }
 // used registers : T2, T3, T1
 // T3 : cp_entry & cache
 // T2 : index & offset
 void TemplateTable::fast_accessfield(TosState state) {
   transition(atos, state);
   // do the JVMTI work here to avoid disturbing the register state below
   if (JvmtiExport::can_post_field_access()) {
     // Check to see if a field access watch has been set before we take
     // the time to call into the VM.
     Label L1;
     __ li(AT, (intptr_t)JvmtiExport::get_field_access_count_addr());
     __ lw(T3, AT, 0);
     __ beq(T3, R0, L1);
     __ delayed()->nop();
     // access constant pool cache entry
     __ get_cache_entry_pointer_at_bcp(T3, T1, 1);
     __ move(TSR, FSR);
     __ verify_oop(FSR);
     // FSR: object pointer copied above
     // T3: cache entry pointer
     __ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::post_field_access),
 	FSR, T3);
     __ move(FSR, TSR);
     __ bind(L1);
   }
   // access constant pool cache
   __ get_cache_and_index_at_bcp(T3, T2, 1);
   // replace index with field offset from cache entry
   __ dsll(AT, T2, Address::times_8);
   //__ dsll(AT, T2, 4);
   __ dadd(AT, T3, AT);
   __ ld(T2, AT, in_bytes(ConstantPoolCache::base_offset()
 	+ ConstantPoolCacheEntry::f2_offset()));
   // eax: object
   __ verify_oop(FSR);
   // __ null_check(FSR, 0);
   __ null_check(FSR);
   // field addresses
   __ dadd(FSR, FSR, T2);
   // access field
   switch (bytecode()) {
     case Bytecodes::_fast_bgetfield:
       __ lb(FSR, FSR, 0);
       break;
     case Bytecodes::_fast_sgetfield:
       __ lh(FSR, FSR, 0);
       break;
     case Bytecodes::_fast_cgetfield:
       __ lhu(FSR, FSR, 0);
       break;
     case Bytecodes::_fast_igetfield:
       __ lw(FSR, FSR, 0);
       break;
     case Bytecodes::_fast_lgetfield:
       __ stop("should not be rewritten");
       break;
     case Bytecodes::_fast_fgetfield:
       __ lwc1(FSF, FSR, 0);
       break;
     case Bytecodes::_fast_dgetfield:
       __ ldc1(FSF, FSR, 0);
       break;
     case Bytecodes::_fast_agetfield:
       //add for compressedoops
       __ load_heap_oop(FSR, Address(FSR, 0));
       __ verify_oop(FSR);
       break;
     default:
       ShouldNotReachHere();
   }
   // Doug Lea believes this is not needed with current Sparcs(TSO) and Intel(PSO)
   // volatile_barrier( );
 }
 // generator for _fast_iaccess_0, _fast_aaccess_0, _fast_faccess_0
 // used registers : T1, T2, T3, T1
 // T1 : obj & field address
 // T2 : off
 // T3 : cache
 // T1 : index
 void TemplateTable::fast_xaccess(TosState state) {
   transition(vtos, state);
   // get receiver
   __ ld(T1, aaddress(0));
   // access constant pool cache
   __ get_cache_and_index_at_bcp(T3, T2, 2);
   __ dsll(AT, T2, Address::times_8);
   __ dadd(AT, T3, AT);
   __ ld(T2, AT, in_bytes(ConstantPoolCache::base_offset()
 	+ ConstantPoolCacheEntry::f2_offset()));
   // make sure exception is reported in correct bcp range (getfield is next instruction)
   __ daddi(BCP, BCP, 1);
   //	__ null_check(T1, 0);
   __ null_check(T1);
   __ dadd(T1, T1, T2);
   if (state == itos) {
     __ lw(FSR, T1, 0);
   } else if (state == atos) {
     //__ ld(FSR, T1, 0);
     __ load_heap_oop(FSR, Address(T1, 0));
     __ verify_oop(FSR);
   } else if (state == ftos) {
     __ lwc1(FSF, T1, 0);
   } else {
     ShouldNotReachHere();
   }
   __ daddi(BCP, BCP, -1);
 }
 //---------------------------------------------------
 //-------------------------------------------------
 // Calls
 void TemplateTable::count_calls(Register method, Register temp) {
 	// implemented elsewhere
 	ShouldNotReachHere();
 }
 // method, index, recv, flags: T1, T2, T3, T1
 // byte_no = 2 for _invokevirtual, 1 else
 // T0 : return address
 // get the method & index of the invoke, and push the return address of
 // the invoke(first word in the frame)
 // this address is where the return code jmp to.
 // NOTE : this method will set T3&T1 as recv&flags
 void TemplateTable::prepare_invoke(int byte_no,
                                    Register method, //linked method (or i-klass)
                                    Register index, //itable index, MethodType ,etc.
                                    Register recv, // if caller wants to see it
                                    Register flags // if caller wants to test it
 		                   ) {
   // determine flags
   const Bytecodes::Code code = bytecode();
   const bool is_invokeinterface  = code == Bytecodes::_invokeinterface;
   const bool is_invokedynamic    = code == Bytecodes::_invokedynamic;
   const bool is_invokehandle     = code == Bytecodes::_invokehandle;
   const bool is_invokevirtual    = code == Bytecodes::_invokevirtual;
   const bool is_invokespecial    = code == Bytecodes::_invokespecial;
   const bool load_receiver       = (recv  != noreg);
   const bool save_flags          = (flags != noreg);
   assert(load_receiver == (code != Bytecodes::_invokestatic && code != Bytecodes::_invokedynamic),"");
   assert(save_flags    == (is_invokeinterface || is_invokevirtual), "need flags for vfinal");
   assert(flags == noreg || flags == T1, "error flags reg.");
   assert(recv  == noreg || recv  == T3, "error recv reg.");
   // setup registers & access constant pool cache
   if(recv == noreg) recv  = T3;
   if(flags == noreg) flags  = T1;
   assert_different_registers(method, index, recv, flags);
   // save 'interpreter return address'
   __ save_bcp();
   load_invoke_cp_cache_entry(byte_no, method, index, flags, is_invokevirtual, false, is_invokedynamic);
   if (is_invokedynamic || is_invokehandle) {
    Label L_no_push;
      __ move(AT, (1 << ConstantPoolCacheEntry::has_appendix_shift));
      __ andr(AT, AT, flags);
      __ beq(AT, R0, L_no_push);
      __ delayed()->nop();
      // Push the appendix as a trailing parameter.
      // This must be done before we get the receiver,
      // since the parameter_size includes it.
      Register tmp = SSR;
      __ push(tmp);
      __ move(tmp, index);
      assert(ConstantPoolCacheEntry::_indy_resolved_references_appendix_offset == 0, "appendix expected at index+0");
      __ load_resolved_reference_at_index(index, tmp);
      __ pop(tmp);
      __ push(index);  // push appendix (MethodType, CallSite, etc.)
      __ bind(L_no_push);
   }
 // load receiver if needed (after appendix is pushed so parameter size is correct)
 // Note: no return address pushed yet
   if (load_receiver) {
 	 __ move(AT, ConstantPoolCacheEntry::parameter_size_mask);
 	 __ andr(recv, flags, AT);
          // 2014/07/31 Fu: Since we won't push RA on stack, no_return_pc_pushed_yet should be 0.
 	 const int no_return_pc_pushed_yet = 0;  // argument slot correction before we push return address
 	 const int receiver_is_at_end      = -1;  // back off one slot to get receiver
 	 Address recv_addr = __ argument_address(recv, no_return_pc_pushed_yet + receiver_is_at_end);
 	 __ ld(recv, recv_addr);
 	 __ verify_oop(recv);
   }
   if(save_flags) {
     //__ movl(r13, flags);
     __ move(BCP, flags);
   }
   // compute return type
   __ dsrl(flags, flags, ConstantPoolCacheEntry::tos_state_shift);
   __ andi(flags, flags, 0xf);
   // Make sure we don't need to mask flags for tos_state_shift after the above shift
   ConstantPoolCacheEntry::verify_tos_state_shift();
   // load return address
   {
     const address table = (address) Interpreter::invoke_return_entry_table_for(code);
     __ li(AT, (long)table);
     __ dsll(flags, flags, LogBytesPerWord);
     __ dadd(AT, AT, flags);
     __ ld(RA, AT, 0);
   }
   if (save_flags) {
     __ move(flags, BCP);
     __ restore_bcp();
   }
 }
 // used registers : T0, T3, T1, T2
 // T3 : recv, this two register using convention is by prepare_invoke
 // T1 : flags, klass
 // Rmethod : method, index must be Rmethod
 void TemplateTable::invokevirtual_helper(Register index, Register recv,
 		Register flags) {
   assert_different_registers(index, recv, flags, T2);
   // Test for an invoke of a final method
   Label notFinal;
   __ move(AT, (1 << ConstantPoolCacheEntry::is_vfinal_shift));
   __ andr(AT, flags, AT);
   __ beq(AT, R0, notFinal);
   __ delayed()->nop();
   Register method = index;  // method must be Rmethod
   assert(method == Rmethod, "methodOop must be Rmethod for interpreter calling convention");
   // do the call - the index is actually the method to call
   // the index is indeed methodOop, for this is vfinal,
   // see ConstantPoolCacheEntry::set_method for more info
   __ verify_oop(method);
   // It's final, need a null check here!
   __ null_check(recv);
   // profile this call
   __ profile_final_call(T2);
   // 2014/11/24 Fu
   // T2: tmp, used for mdp
   // method: callee
   // T9: tmp
   // is_virtual: true
   __ profile_arguments_type(T2, method, T9, true);
 //  __ move(T0, recv);
   __ jump_from_interpreted(method, T2);
   __ bind(notFinal);
   // get receiver klass
   __ null_check(recv, oopDesc::klass_offset_in_bytes());
   // Keep recv in ecx for callee expects it there
   __ load_klass(T2, recv);
   __ verify_oop(T2);
   // profile this call
   __ profile_virtual_call(T2, T0, T1);
   // get target methodOop & entry point
   const int base = InstanceKlass::vtable_start_offset() * wordSize;
   assert(vtableEntry::size() * wordSize == 8, "adjust the scaling in the code below");
   __ dsll(AT, index, Address::times_8);
   __ dadd(AT, T2, AT);
   //this is a ualign read
   __ ld(method, AT, base + vtableEntry::method_offset_in_bytes());
   __ jump_from_interpreted(method, T2);
 }
 void TemplateTable::invokevirtual(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f2_byte, "use this argument");
   prepare_invoke(byte_no, Rmethod, NOREG, T3, T1);
   // now recv & flags in T3, T1
   invokevirtual_helper(Rmethod, T3, T1);
 }
 // T9 : entry
 // Rmethod : method
 void TemplateTable::invokespecial(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
   prepare_invoke(byte_no, Rmethod, NOREG, T3);
   // now recv & flags in T3, T1
   __ verify_oop(T3);
   __ null_check(T3);
   __ profile_call(T9);
   // 2014/11/24 Fu
   // T8: tmp, used for mdp
   // Rmethod: callee
   // T9: tmp
   // is_virtual: false
   __ profile_arguments_type(T8, Rmethod, T9, false);
   __ jump_from_interpreted(Rmethod, T9);
   __ move(T0, T3);//aoqi ?
 }
 void TemplateTable::invokestatic(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
   prepare_invoke(byte_no, Rmethod, NOREG);
   __ verify_oop(Rmethod);
   __ profile_call(T9);
   // 2014/11/24 Fu
   // T8: tmp, used for mdp
   // Rmethod: callee
   // T9: tmp
   // is_virtual: false
   __ profile_arguments_type(T8, Rmethod, T9, false);
   __ jump_from_interpreted(Rmethod, T9);
 }
 // i have no idea what to do here, now. for future change. FIXME.
 void TemplateTable::fast_invokevfinal(int byte_no) {
 	transition(vtos, vtos);
 	assert(byte_no == f2_byte, "use this argument");
 	__ stop("fast_invokevfinal not used on x86");
 }
 // used registers : T0, T1, T2, T3, T1, A7
 // T0 : itable, vtable, entry
 // T1 : interface
 // T3 : receiver
 // T1 : flags, klass
 // Rmethod : index, method, this is required by interpreter_entry
 void TemplateTable::invokeinterface(int byte_no) {
   transition(vtos, vtos);
   //this method will use T1-T4 and T0
   assert(byte_no == f1_byte, "use this argument");
   prepare_invoke(byte_no, T2, Rmethod, T3, T1);
   // T2: Interface
   // Rmethod: index
   // T3: receiver
   // T1: flags
   Label notMethod;
   __ move(AT, (1 << ConstantPoolCacheEntry::is_forced_virtual_shift));
   __ andr(AT, T1, AT);
   __ beq(AT, R0, notMethod);
   __ delayed()->nop();
   // Special case of invokeinterface called for virtual method of
   // java.lang.Object.  See cpCacheOop.cpp for details.
   // This code isn't produced by javac, but could be produced by
   // another compliant java compiler.
   invokevirtual_helper(Rmethod, T3, T1);
   __ bind(notMethod);
   // Get receiver klass into T1 - also a null check
   //__ ld(T1, T3, oopDesc::klass_offset_in_bytes());
   //add for compressedoops
   //__ restore_locals();
   //__ null_check(T3, oopDesc::klass_offset_in_bytes());
   __ load_klass(T1, T3);
   __ verify_oop(T1);
   // profile this call
   __ profile_virtual_call(T1, T0, FSR);
   // Compute start of first itableOffsetEntry (which is at the end of the vtable)
   // TODO: x86 add a new method lookup_interface_method  // LEE
   const int base = InstanceKlass::vtable_start_offset() * wordSize;
   assert(vtableEntry::size() * wordSize == 8, "adjust the scaling in the code below");
   __ lw(AT, T1, InstanceKlass::vtable_length_offset() * wordSize);
   __ dsll(AT, AT, Address::times_8);
   __ dadd(T0, T1, AT);
   __ daddi(T0, T0, base);
   if (HeapWordsPerLong > 1) {
     // Round up to align_object_offset boundary
     __ round_to(T0, BytesPerLong);
   }
   // now T0 is the begin of the itable
   Label entry, search, interface_ok;
   ///__ jmp(entry);
   __ b(entry);
   __ delayed()->nop();
   __ bind(search);
   __ increment(T0, itableOffsetEntry::size() * wordSize);
   __ bind(entry);
   // Check that the entry is non-null.  A null entry means that the receiver
   // class doesn't implement the interface, and wasn't the same as the
   // receiver class checked when the interface was resolved.
   __ ld(AT, T0, itableOffsetEntry::interface_offset_in_bytes());
   __ bne(AT, R0, interface_ok);
   __ delayed()->nop();
   // throw exception
   // the call_VM checks for exception, so we should never return here.
   //__ pop();//FIXME here,
   // pop return address (pushed by prepare_invoke).
   // no need now, we just save the value in RA now
   __ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_IncompatibleClassChangeError));
   __ should_not_reach_here();
   __ bind(interface_ok);
   //NOTICE here, no pop as x86 do
   //__ lw(AT, T0, itableOffsetEntry::interface_offset_in_bytes());
   __ bne(AT, T2, search);
   __ delayed()->nop();
   // now we get vtable of the interface
   __ ld(T0, T0, itableOffsetEntry::offset_offset_in_bytes());
   __ daddu(T0, T1, T0);
   assert(itableMethodEntry::size() * wordSize == 8, "adjust the scaling in the code below");
   __ dsll(AT, Rmethod, Address::times_8);
   __ daddu(AT, T0, AT);
   // now we get the method
   __ ld(Rmethod, AT, 0);
   // Rnext: methodOop to call
   // T3: receiver
   // Check for abstract method error
   // Note: This should be done more efficiently via a throw_abstract_method_error
   //       interpreter entry point and a conditional jump to it in case of a null
   //       method.
   {
     Label L;
     ///__ testl(ebx, ebx);
     ///__ jcc(Assembler::notZero, L);
     __ bne(Rmethod, R0, L);
     __ delayed()->nop();
     // throw exception
     // note: must restore interpreter registers to canonical
     //       state for exception handling to work correctly!
     ///__ popl(ebx);          // pop return address (pushed by prepare_invoke)
     //__ restore_bcp();      // esi must be correct for exception handler
     //(was destroyed)
     //__ restore_locals();   // make sure locals pointer
     //is correct as well (was destroyed)
     ///__ call_VM(noreg, CAST_FROM_FN_PTR(address,
     //InterpreterRuntime::throw_AbstractMethodError));
     __ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_AbstractMethodError));
     // the call_VM checks for exception, so we should never return here.
     __ should_not_reach_here();
     __ bind(L);
   }
   // 2014/11/24 Fu
   // T8: tmp, used for mdp
   // Rmethod: callee
   // T9: tmp
   // is_virtual: true
   __ profile_arguments_type(T8, Rmethod, T9, true);
   __ jump_from_interpreted(Rmethod, T9);
 }
 void TemplateTable::invokehandle(int byte_no) {
   transition(vtos, vtos);
   assert(byte_no == f1_byte, "use this argument");
   const Register T2_method = Rmethod;
   const Register FSR_mtype  = FSR;
   const Register T3_recv   = T3;
   if (!EnableInvokeDynamic) {
      // rewriter does not generate this bytecode
      __ should_not_reach_here();
      return;
    }
    prepare_invoke(byte_no, T2_method, FSR_mtype, T3_recv);
    //??__ verify_method_ptr(T2_method);
    __ verify_oop(T3_recv);
    __ null_check(T3_recv);
    // rax: MethodType object (from cpool->resolved_references[f1], if necessary)
    // rbx: MH.invokeExact_MT method (from f2)
    // Note:  rax_mtype is already pushed (if necessary) by prepare_invoke
    // FIXME: profile the LambdaForm also
    __ profile_final_call(T9);
    // 2014/11/24 Fu
    // T8: tmp, used for mdp
    // T2_method: callee
    // T9: tmp
    // is_virtual: true
    __ profile_arguments_type(T8, T2_method, T9, true);
   __ jump_from_interpreted(T2_method, T9);
 }
  void TemplateTable::invokedynamic(int byte_no) {
    transition(vtos, vtos);
    assert(byte_no == f1_byte, "use this argument");
    if (!EnableInvokeDynamic) {
      // We should not encounter this bytecode if !EnableInvokeDynamic.
      // The verifier will stop it.  However, if we get past the verifier,
      // this will stop the thread in a reasonable way, without crashing the JVM.
      __ call_VM(noreg, CAST_FROM_FN_PTR(address,
                       InterpreterRuntime::throw_IncompatibleClassChangeError));
      // the call_VM checks for exception, so we should never return here.
      __ should_not_reach_here();
      return;
    }
    //const Register Rmethod   = T2;
    const Register T2_callsite = T2;
    prepare_invoke(byte_no, Rmethod, T2_callsite);
    // rax: CallSite object (from cpool->resolved_references[f1])
    // rbx: MH.linkToCallSite method (from f2)
    // Note:  rax_callsite is already pushed by prepare_invoke
    // %%% should make a type profile for any invokedynamic that takes a ref argument
    // profile this call
    __ profile_call(T9);
    // 2014/11/24 Fu
    // T8: tmp, used for mdp
    // Rmethod: callee
    // T9: tmp
    // is_virtual: false
    __ profile_arguments_type(T8, Rmethod, T9, false);
    __ verify_oop(T2_callsite);
    __ jump_from_interpreted(Rmethod, T9);
  }
 //----------------------------------------------------------------------------------------------------
 // Allocation
 // T1 : tags & buffer end & thread
 // T2 : object end
 // T3 : klass
 // T1 : object size
 // A1 : cpool
 // A2 : cp index
 // return object in FSR
 void TemplateTable::_new() {
   transition(vtos, atos);
   __ get_2_byte_integer_at_bcp(A2, AT, 1);
   __ huswap(A2);
   Label slow_case;
   Label done;
   Label initialize_header;
   Label initialize_object;  // including clearing the fields
   Label allocate_shared;
   // get InstanceKlass in T3
   __ get_cpool_and_tags(A1, T1);
   __ dsll(AT, A2, Address::times_8);
   __ dadd(AT, A1, AT);
   __ ld(T3, AT, sizeof(ConstantPool));
   // make sure the class we're about to instantiate has been resolved.
   // Note: slow_case does a pop of stack, which is why we loaded class/pushed above
   const int tags_offset = Array<u1>::base_offset_in_bytes();
   __ dadd(T1, T1, A2);
   __ lb(AT, T1, tags_offset);
   //__ addiu(AT, AT, - (int)JVM_CONSTANT_UnresolvedClass);
   __ daddiu(AT, AT, - (int)JVM_CONSTANT_Class);
   //__ beq(AT, R0, slow_case);
   __ bne(AT, R0, slow_case);
   __ delayed()->nop();
   /*make sure klass is initialized & doesn't have finalizer*/
   // make sure klass is fully initialized
   __ lhu(T1, T3, in_bytes(InstanceKlass::init_state_offset()));
   __ daddiu(AT, T1, - (int)InstanceKlass::fully_initialized);
   __ bne(AT, R0, slow_case);
   __ delayed()->nop();
   // has_finalizer
   //__ lw(T1, T3, Klass::access_flags_offset() + sizeof(oopDesc));
   //__ move(AT, JVM_ACC_CAN_BE_FASTPATH_ALLOCATED);
   //__ andr(AT, T1, AT);
   __ lw(T1, T3, in_bytes(Klass::layout_helper_offset()) );
   __ andi(AT, T1, Klass::_lh_instance_slow_path_bit);
   __ bne(AT, R0, slow_case);
   __ delayed()->nop();
   // get instance_size in InstanceKlass (already aligned) in T0,
   // be sure to preserve this value
   //__ lw(T0, T3, Klass::size_helper_offset_in_bytes() + sizeof(oopDesc));
   //Klass::_size_helper is renamed Klass::_layout_helper. aoqi
   __ lw(T0, T3, in_bytes(Klass::layout_helper_offset()) );
   //
   // Allocate the instance
   // 1) Try to allocate in the TLAB
   // 2) if fail and the object is large allocate in the shared Eden
   // 3) if the above fails (or is not applicable), go to a slow case
   // (creates a new TLAB, etc.)
   const bool allow_shared_alloc =
     Universe::heap()->supports_inline_contig_alloc() && !CMSIncrementalMode;
   if (UseTLAB) {
 #ifndef OPT_THREAD
     const Register thread = T8;
     __ get_thread(thread);
 #else
     const Register thread = TREG;
 #endif
     // get tlab_top
     __ ld(FSR, thread, in_bytes(JavaThread::tlab_top_offset()));
     __ dadd(T2, FSR, T0);
     // get tlab_end
     __ ld(AT, thread, in_bytes(JavaThread::tlab_end_offset()));
     __ slt(AT, AT, T2);
     //		__ bne(AT, R0, allocate_shared);
     __ bne(AT, R0, allow_shared_alloc ? allocate_shared : slow_case);
     __ delayed()->nop();
     __ sd(T2, thread, in_bytes(JavaThread::tlab_top_offset()));
     if (ZeroTLAB) {
       // the fields have been already cleared
       __ b_far(initialize_header);
     } else {
       // initialize both the header and fields
       __ b_far(initialize_object);
     }
     __ delayed()->nop();
     /*
        if (CMSIncrementalMode) {
     // No allocation in shared eden.
     ///__ jmp(slow_case);
     __ b(slow_case);
     __ delayed()->nop();
     }
      */
   }
   // Allocation in the shared Eden , if allowed
   // T0 : instance size in words
   if(allow_shared_alloc){
     __ bind(allocate_shared);
     Label retry;
     //Address heap_top(T1, (int)Universe::heap()->top_addr());
     Address heap_top(T1);
     //__ lui(T1, Assembler::split_high((int)Universe::heap()->top_addr()));
     __ li(T1, (long)Universe::heap()->top_addr());
     __ ld(FSR, heap_top);
     __ bind(retry);
     __ dadd(T2, FSR, T0);
     //__ lui(AT, Assembler::split_high((int)Universe::heap()->end_addr()));
     //__ lw(AT, AT, Assembler::split_low((int)Universe::heap()->end_addr()));
     __ li(AT, (long)Universe::heap()->end_addr());
     __ ld(AT, AT, 0);
     __ slt(AT, AT, T2);
     __ bne(AT, R0, slow_case);
     __ delayed()->nop();
     // Compare FSR with the top addr, and if still equal, store the new
     // top addr in ebx at the address of the top addr pointer. Sets ZF if was
     // equal, and clears it otherwise. Use lock prefix for atomicity on MPs.
     //
     // FSR: object begin
     // T2: object end
     // T0: instance size in words
     // if someone beat us on the allocation, try again, otherwise continue
     //__ lui(T1, Assembler::split_high((int)Universe::heap()->top_addr()));
     __ cmpxchg(T2, heap_top, FSR);
     __ beq(AT, R0, retry);
     __ delayed()->nop();
   }
   if (UseTLAB || Universe::heap()->supports_inline_contig_alloc()) {
     // The object is initialized before the header.  If the object size is
     // zero, go directly to the header initialization.
     __ bind(initialize_object);
     __ li(AT, - sizeof(oopDesc));
     __ daddu(T0, T0, AT);
     __ beq_far(T0, R0, initialize_header);
     __ delayed()->nop();
     // T0 must have been multiple of 2
 #ifdef ASSERT
     // make sure T0 was multiple of 2
     Label L;
     __ andi(AT, T0, 1);
     __ beq(AT, R0, L);
     __ delayed()->nop();
     __ stop("object size is not multiple of 2 - adjust this code");
     __ bind(L);
     // edx must be > 0, no extra check needed here
 #endif
     // initialize remaining object fields: T0 is a multiple of 2
     {
       Label loop;
       __ dadd(T1, FSR, T0);
       __ daddi(T1, T1, -oopSize);
       __ bind(loop);
       __ sd(R0, T1, sizeof(oopDesc) + 0 * oopSize);
 //      __ sd(R0, T1, sizeof(oopDesc) + 1 * oopSize);
       __ bne(T1, FSR, loop); //dont clear header
       __ delayed()->daddi(T1, T1, -oopSize);
       // actually sizeof(oopDesc)==8, so we can move
       // __ addiu(AT, AT, -8) to delay slot, and compare FSR with T1
     }
     //klass in T3,
     // initialize object header only.
     __ bind(initialize_header);
     if (UseBiasedLocking) {
       // __ popl(ecx);   // get saved klass back in the register.
       // __ movl(ebx, Address(ecx, Klass::prototype_header_offset_in_bytes()
       // + klassOopDesc::klass_part_offset_in_bytes()));
       __ ld(AT, T3, in_bytes(Klass::prototype_header_offset()));
       // __ movl(Address(eax, oopDesc::mark_offset_in_bytes ()), ebx);
       __ sd(AT, FSR, oopDesc::mark_offset_in_bytes ());
     } else {
       __ li(AT, (long)markOopDesc::prototype());
       __ sd(AT, FSR, oopDesc::mark_offset_in_bytes());
     }
     //__ sd(T3, FSR, oopDesc::klass_offset_in_bytes());
     __ store_klass_gap(FSR, R0);
     __ store_klass(FSR, T3);
     {
       SkipIfEqual skip_if(_masm, &DTraceAllocProbes, 0);
       // Trigger dtrace event for fastpath
       __ push(atos);
       __ call_VM_leaf(
 	  CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_object_alloc), FSR);
       __ pop(atos);
     }
     __ b(done);
     __ delayed()->nop();
   }
   // slow case
   __ bind(slow_case);
   call_VM(FSR, CAST_FROM_FN_PTR(address, InterpreterRuntime::_new), A1, A2);
   // continue
   __ bind(done);
   __ sync();
 }
 void TemplateTable::newarray() {
 	transition(itos, atos);
 	__ lbu(A1, at_bcp(1));
 	//type, count
 	call_VM(FSR, CAST_FROM_FN_PTR(address, InterpreterRuntime::newarray), A1, FSR);
         __ sync();
 }
 void TemplateTable::anewarray() {
   transition(itos, atos);
   __ get_2_byte_integer_at_bcp(A2, AT, 1);
   __ huswap(A2);
   __ get_constant_pool(A1);
   // cp, index, count
   call_VM(FSR, CAST_FROM_FN_PTR(address, InterpreterRuntime::anewarray), A1, A2, FSR);
   __ sync();
 }
 void TemplateTable::arraylength() {
   transition(atos, itos);
   __ null_check(FSR, arrayOopDesc::length_offset_in_bytes());
   __ lw(FSR, FSR, arrayOopDesc::length_offset_in_bytes());
 }
 // i use T2 as ebx, T3 as ecx, T1 as edx
 // when invoke gen_subtype_check, super in T3, sub in T2, object in FSR(it's always)
 // T2 : sub klass
 // T3 : cpool
 // T3 : super klass
 void TemplateTable::checkcast() {
   transition(atos, atos);
   Label done, is_null, ok_is_subtype, quicked, resolved;
   __ beq(FSR, R0, is_null);
   __ delayed()->nop();
   // Get cpool & tags index
   __ get_cpool_and_tags(T3, T1);
   __ get_2_byte_integer_at_bcp(T2, AT, 1);
   __ huswap(T2);
   // See if bytecode has already been quicked
   __ dadd(AT, T1, T2);
   __ lb(AT, AT, Array<u1>::base_offset_in_bytes());
   __ daddiu(AT, AT, - (int)JVM_CONSTANT_Class);
   __ beq(AT, R0, quicked);
   __ delayed()->nop();
   /* 2012/6/2 Jin: In InterpreterRuntime::quicken_io_cc, lots of new classes may be loaded.
    *  Then, GC will move the object in V0 to another places in heap.
    *  Therefore, We should never save such an object in register.
    *  Instead, we should save it in the stack. It can be modified automatically by the GC thread.
    *  After GC, the object address in FSR is changed to a new place.
    */
   __ push(atos);
   const Register thread = TREG;
 #ifndef OPT_THREAD
   __ get_thread(thread);
 #endif
   call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::quicken_io_cc));
   __ get_vm_result_2(T3, thread);
   __ pop_ptr(FSR);
   __ b(resolved);
   __ delayed()->nop();
   // klass already in cp, get superklass in T3
   __ bind(quicked);
   __ dsll(AT, T2, Address::times_8);
   __ dadd(AT, T3, AT);
   __ ld(T3, AT, sizeof(ConstantPool));
   __ bind(resolved);
   // get subklass in T2
   //__ ld(T2, FSR, oopDesc::klass_offset_in_bytes());
   //add for compressedoops
   __ load_klass(T2, FSR);
   // Superklass in T3.  Subklass in T2.
   __ gen_subtype_check(T3, T2, ok_is_subtype);
   // Come here on failure
   // object is at FSR
   __ jmp(Interpreter::_throw_ClassCastException_entry);
   __ delayed()->nop();
   // Come here on success
   __ bind(ok_is_subtype);
   // Collect counts on whether this check-cast sees NULLs a lot or not.
   if (ProfileInterpreter) {
 	__ b(done);
 	__ delayed()->nop();
 	__ bind(is_null);
 	__ profile_null_seen(T3);
   } else {
 	__ bind(is_null);
   }
   __ bind(done);
 }
 // i use T3 as cpool, T1 as tags, T2 as index
 // object always in FSR, superklass in T3, subklass in T2
 void TemplateTable::instanceof() {
   transition(atos, itos);
   Label done, is_null, ok_is_subtype, quicked, resolved;
   __ beq(FSR, R0, is_null);
   __ delayed()->nop();
   // Get cpool & tags index
   __ get_cpool_and_tags(T3, T1);
   // get index
   __ get_2_byte_integer_at_bcp(T2, AT, 1);
   __ hswap(T2);
   // See if bytecode has already been quicked
   // quicked
   __ daddu(AT, T1, T2);
   __ lb(AT, AT, Array<u1>::base_offset_in_bytes());
   __ daddiu(AT, AT, - (int)JVM_CONSTANT_Class);
   __ beq(AT, R0, quicked);
   __ delayed()->nop();
   // get superklass in T3
   //__ move(TSR, FSR);
   // sometimes S2 may be changed during the call,
   // be careful if u use TSR as a saving place
   //__ push(FSR);
   __ push(atos);
   const Register thread = TREG;
 #ifndef OPT_THREAD
   __ get_thread(thread);
 #endif
   call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::quicken_io_cc));
   __ get_vm_result_2(T3, thread);
   //__ lw(FSR, SP, 0);
   __ pop_ptr(FSR);
   __ b(resolved);
   __ delayed()->nop();
   //__ move(FSR, TSR);
   // get superklass in T3, subklass in T2
   __ bind(quicked);
   __ dsll(AT, T2, Address::times_8);
   __ daddu(AT, T3, AT);
   __ ld(T3, AT, sizeof(ConstantPool));
   __ bind(resolved);
   // get subklass in T2
   //__ ld(T2, FSR, oopDesc::klass_offset_in_bytes());
   //add for compressedoops
   __ load_klass(T2, FSR);
   // Superklass in T3.  Subklass in T2.
   __ gen_subtype_check(T3, T2, ok_is_subtype);
   // Come here on failure
   __ b(done);
   __ delayed(); __ move(FSR, R0);
   // Come here on success
   __ bind(ok_is_subtype);
   __ move(FSR, 1);
   // Collect counts on whether this test sees NULLs a lot or not.
   if (ProfileInterpreter) {
      __ beq(R0, R0, done);
      __ nop();
      __ bind(is_null);
      __ profile_null_seen(T3);
   } else {
      __ bind(is_null);   // same as 'done'
   }
   __ bind(done);
   // FSR = 0: obj == NULL or  obj is not an instanceof the specified klass
   // FSR = 1: obj != NULL and obj is     an instanceof the specified klass
 }
 //--------------------------------------------------------
 //--------------------------------------------
 // Breakpoints
 void TemplateTable::_breakpoint() {
 	// Note: We get here even if we are single stepping..
 	// jbug inists on setting breakpoints at every bytecode
 	// even if we are in single step mode.
 	transition(vtos, vtos);
 	// get the unpatched byte code
 	///__ get_method(ecx);
 	///__ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::get_original_bytecode_at)
 	//, ecx, esi);
 	///__ movl(ebx, eax);
 	__ get_method(A1);
 	__ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::get_original_bytecode_at),
 			A1, BCP);
 	__ move(Rnext, V0); // Jin: Rnext will be used in dispatch_only_normal
 	// post the breakpoint event
 	///__ get_method(ecx);
 	///__ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::_breakpoint), ecx, esi);
 	__ get_method(A1);
 	__ call_VM(NOREG, CAST_FROM_FN_PTR(address, InterpreterRuntime::_breakpoint), A1, BCP);
 	// complete the execution of original bytecode
 	__ dispatch_only_normal(vtos);
 }
 //----------------------------------------------------------------------------------------------------
 // Exceptions
 void TemplateTable::athrow() {
 	transition(atos, vtos);
 	__ null_check(FSR);
 	__ jmp(Interpreter::throw_exception_entry());
 	__ delayed()->nop();
 }
 //----------------------------------------------------------------------------------------------------
 // Synchronization
 //
 // Note: monitorenter & exit are symmetric routines; which is reflected
 //       in the assembly code structure as well
 //
 // Stack layout:
 //
 // [expressions  ] <--- SP               = expression stack top
 // ..
 // [expressions  ]
 // [monitor entry] <--- monitor block top = expression stack bot
 // ..
 // [monitor entry]
 // [frame data   ] <--- monitor block bot
 // ...
 // [return addr  ] <--- FP
 // we use T2 as monitor entry pointer, T3 as monitor top pointer, c_rarg0 as free slot pointer
 // object always in FSR
 void TemplateTable::monitorenter() {
   transition(atos, vtos);
   // check for NULL object
   __ null_check(FSR);
   const Address monitor_block_top(FP, frame::interpreter_frame_monitor_block_top_offset
       * wordSize);
   const int entry_size = (frame::interpreter_frame_monitor_size()* wordSize);
   Label allocated;
   // initialize entry pointer
   __ move(c_rarg0, R0);
   // find a free slot in the monitor block (result in edx)
   {
     Label entry, loop, exit, next;
     __ ld(T2, monitor_block_top);
     __ b(entry);
     __ delayed()->daddi(T3, FP, frame::interpreter_frame_initial_sp_offset * wordSize);
     // free slot?
     __ bind(loop);
     __ ld(AT, T2, BasicObjectLock::obj_offset_in_bytes());
     __ bne(AT, R0, next);
     __ delayed()->nop();
     __ move(c_rarg0, T2);
     __ bind(next);
     __ beq(FSR, AT, exit);
     __ delayed()->nop();
     __ daddi(T2, T2, entry_size);
     __ bind(entry);
     __ bne(T3, T2, loop);
     __ delayed()->nop();
     __ bind(exit);
   }
   __ bne(c_rarg0, R0, allocated);
   __ delayed()->nop();
   // allocate one if there's no free slot
   {
     Label entry, loop;
     // 1. compute new pointers                   // SP: old expression stack top
     __ ld(c_rarg0, monitor_block_top);
     __ daddi(SP, SP, - entry_size);
     __ daddi(c_rarg0, c_rarg0, - entry_size);
     __ sd(c_rarg0, monitor_block_top);
     __ b(entry);
     __ delayed(); __ move(T3, SP);
     // 2. move expression stack contents
     __ bind(loop);
     __ ld(AT, T3, entry_size);
     __ sd(AT, T3, 0);
     __ daddi(T3, T3, wordSize);
     __ bind(entry);
     __ bne(T3, c_rarg0, loop);
     __ delayed()->nop();
   }
   __ bind(allocated);
   // Increment bcp to point to the next bytecode,
   // so exception handling for async. exceptions work correctly.
   // The object has already been poped from the stack, so the
   // expression stack looks correct.
   __ daddi(BCP, BCP, 1);
   __ sd(FSR, c_rarg0, BasicObjectLock::obj_offset_in_bytes());
   __ lock_object(c_rarg0);
   // check to make sure this monitor doesn't cause stack overflow after locking
   __ save_bcp();  // in case of exception
   __ generate_stack_overflow_check(0);
   // The bcp has already been incremented. Just need to dispatch to next instruction.
   __ dispatch_next(vtos);
 }
 // T2 : top
 // c_rarg0 : entry
 void TemplateTable::monitorexit() {
   transition(atos, vtos);
   __ null_check(FSR);
   const int entry_size =(frame::interpreter_frame_monitor_size()* wordSize);
   Label found;
   // find matching slot
   {
     Label entry, loop;
     __ ld(c_rarg0, FP, frame::interpreter_frame_monitor_block_top_offset * wordSize);
     __ b(entry);
     __ delayed()->daddiu(T2, FP, frame::interpreter_frame_initial_sp_offset * wordSize);
     __ bind(loop);
     __ ld(AT, c_rarg0, BasicObjectLock::obj_offset_in_bytes());
     __ beq(FSR, AT, found);
     __ delayed()->nop();
     __ daddiu(c_rarg0, c_rarg0, entry_size);
     __ bind(entry);
     __ bne(T2, c_rarg0, loop);
     __ delayed()->nop();
   }
   // error handling. Unlocking was not block-structured
   Label end;
   __ call_VM(NOREG, CAST_FROM_FN_PTR(address,
 	InterpreterRuntime::throw_illegal_monitor_state_exception));
   __ should_not_reach_here();
   // call run-time routine
   // c_rarg0: points to monitor entry
   __ bind(found);
   __ move(TSR, FSR);
   __ unlock_object(c_rarg0);
   __ move(FSR, TSR);
   __ bind(end);
 }
 //--------------------------------------------------------------------------------------------------// Wide instructions
 void TemplateTable::wide() {
   transition(vtos, vtos);
   // Note: the esi increment step is part of the individual wide bytecode implementations
   __ lbu(Rnext, at_bcp(1));
   __ dsll(T9, Rnext, Address::times_8);
   __ li(AT, (long)Interpreter::_wentry_point);
   __ dadd(AT, T9, AT);
   __ ld(T9, AT, 0);
   __ jr(T9);
   __ delayed()->nop();
 }
 //--------------------------------------------------------------------------------------------------// Multi arrays
 void TemplateTable::multianewarray() {
   transition(vtos, atos);
   // last dim is on top of stack; we want address of first one:
   // first_addr = last_addr + (ndims - 1) * wordSize
   __ lbu(A1, at_bcp(3));	// dimension
   __ daddi(A1, A1, -1);
   __ dsll(A1, A1, Address::times_8);
   __ dadd(A1, SP, A1);		// now A1 pointer to the count array on the stack
   call_VM(FSR, CAST_FROM_FN_PTR(address, InterpreterRuntime::multianewarray), A1);
   __ lbu(AT, at_bcp(3));
   __ dsll(AT, AT, Address::times_8);
   __ dadd(SP, SP, AT);
   __ sync();
 }
 #endif // !CC_INTERP

Mercurial > jdk8-mips64-public > hotspot / annotate

src/cpu/mips/vm/templateTable_mips_64.cpp@179d203c9b2b (annotated)

src/cpu/mips/vm/templateTable_mips_64.cpp