view src/hotspot/cpu/s390/interp_masm_s390.cpp @ 54970:b776653628c5

8218991: s390: Add intrinsic for GHASH algorithm Reviewed-by: lucy, simonis
author mdoerr
date Mon, 18 Feb 2019 12:16:02 +0100
parents 6a880e576856
children 596ae6c3ef6f
line wrap: on
line source
/*
 * Copyright (c) 2016, 2018, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2016, 2018 SAP SE. 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.
 *
 */

// Major contributions by AHa, AS, JL, ML.

#include "precompiled.hpp"
#include "asm/macroAssembler.inline.hpp"
#include "gc/shared/barrierSet.hpp"
#include "gc/shared/barrierSetAssembler.hpp"
#include "interp_masm_s390.hpp"
#include "interpreter/interpreter.hpp"
#include "interpreter/interpreterRuntime.hpp"
#include "oops/arrayOop.hpp"
#include "oops/markOop.hpp"
#include "prims/jvmtiExport.hpp"
#include "prims/jvmtiThreadState.hpp"
#include "runtime/basicLock.hpp"
#include "runtime/biasedLocking.hpp"
#include "runtime/frame.inline.hpp"
#include "runtime/safepointMechanism.hpp"
#include "runtime/sharedRuntime.hpp"
#include "runtime/thread.inline.hpp"

// Implementation of InterpreterMacroAssembler.
// This file specializes the assembler with interpreter-specific macros.

#ifdef PRODUCT
#define BLOCK_COMMENT(str)
#define BIND(label)        bind(label);
#else
#define BLOCK_COMMENT(str) block_comment(str)
#define BIND(label)        bind(label); BLOCK_COMMENT(#label ":")
#endif

void InterpreterMacroAssembler::jump_to_entry(address entry, Register Rscratch) {
  assert(entry != NULL, "Entry must have been generated by now");
  assert(Rscratch != Z_R0, "Can't use R0 for addressing");
  branch_optimized(Assembler::bcondAlways, entry);
}

void InterpreterMacroAssembler::empty_expression_stack(void) {
  get_monitors(Z_R1_scratch);
  add2reg(Z_esp, -Interpreter::stackElementSize, Z_R1_scratch);
}

// Dispatch code executed in the prolog of a bytecode which does not do it's
// own dispatch.
void InterpreterMacroAssembler::dispatch_prolog(TosState state, int bcp_incr) {
  // On z/Architecture we are short on registers, therefore we do not preload the
  // dispatch address of the next bytecode.
}

// Dispatch code executed in the epilog of a bytecode which does not do it's
// own dispatch.
void InterpreterMacroAssembler::dispatch_epilog(TosState state, int step) {
  dispatch_next(state, step);
}

void InterpreterMacroAssembler::dispatch_next(TosState state, int bcp_incr, bool generate_poll) {
  z_llgc(Z_bytecode, bcp_incr, Z_R0, Z_bcp);  // Load next bytecode.
  add2reg(Z_bcp, bcp_incr);                   // Advance bcp. Add2reg produces optimal code.
  dispatch_base(state, Interpreter::dispatch_table(state), generate_poll);
}

// Common code to dispatch and dispatch_only.
// Dispatch value in Lbyte_code and increment Lbcp.

void InterpreterMacroAssembler::dispatch_base(TosState state, address* table, bool generate_poll) {
  verify_FPU(1, state);

#ifdef ASSERT
  address reentry = NULL;
  { Label OK;
    // Check if the frame pointer in Z_fp is correct.
    z_cg(Z_fp, 0, Z_SP);
    z_bre(OK);
    reentry = stop_chain_static(reentry, "invalid frame pointer Z_fp: " FILE_AND_LINE);
    bind(OK);
  }
  { Label OK;
    // check if the locals pointer in Z_locals is correct
    z_cg(Z_locals, _z_ijava_state_neg(locals), Z_fp);
    z_bre(OK);
    reentry = stop_chain_static(reentry, "invalid locals pointer Z_locals: " FILE_AND_LINE);
    bind(OK);
  }
#endif

  // TODO: Maybe implement +VerifyActivationFrameSize here.
  // verify_thread(); // Too slow. We will just verify on method entry & exit.
  verify_oop(Z_tos, state);

  // Dispatch table to use.
  load_absolute_address(Z_tmp_1, (address)table);  // Z_tmp_1 = table;

  if (SafepointMechanism::uses_thread_local_poll() && generate_poll) {
    address *sfpt_tbl = Interpreter::safept_table(state);
    if (table != sfpt_tbl) {
      Label dispatch;
      const Address poll_byte_addr(Z_thread, in_bytes(Thread::polling_page_offset()) + 7 /* Big Endian */);
      // Armed page has poll_bit set, if poll bit is cleared just continue.
      z_tm(poll_byte_addr, SafepointMechanism::poll_bit());
      z_braz(dispatch);
      load_absolute_address(Z_tmp_1, (address)sfpt_tbl);  // Z_tmp_1 = table;
      bind(dispatch);
    }
  }

  // 0 <= Z_bytecode < 256 => Use a 32 bit shift, because it is shorter than sllg.
  // Z_bytecode must have been loaded zero-extended for this approach to be correct.
  z_sll(Z_bytecode, LogBytesPerWord, Z_R0);   // Multiply by wordSize.
  z_lg(Z_tmp_1, 0, Z_bytecode, Z_tmp_1);      // Get entry addr.

  z_br(Z_tmp_1);
}

void InterpreterMacroAssembler::dispatch_only(TosState state, bool generate_poll) {
  dispatch_base(state, Interpreter::dispatch_table(state), generate_poll);
}

void InterpreterMacroAssembler::dispatch_only_normal(TosState state) {
  dispatch_base(state, Interpreter::normal_table(state));
}

void InterpreterMacroAssembler::dispatch_via(TosState state, address *table) {
  // Load current bytecode.
  z_llgc(Z_bytecode, Address(Z_bcp, (intptr_t)0));
  dispatch_base(state, table);
}

// The following call_VM*_base() methods overload and mask the respective
// declarations/definitions in class MacroAssembler. They are meant as a "detour"
// to perform additional, template interpreter specific tasks before actually
// calling their MacroAssembler counterparts.

void InterpreterMacroAssembler::call_VM_leaf_base(address entry_point) {
  bool allow_relocation = true; // Fenerally valid variant. Assume code is relocated.
  // interpreter specific
  // Note: No need to save/restore bcp (Z_R13) pointer since these are callee
  // saved registers and no blocking/ GC can happen in leaf calls.

  // super call
  MacroAssembler::call_VM_leaf_base(entry_point, allow_relocation);
}

void InterpreterMacroAssembler::call_VM_leaf_base(address entry_point, bool allow_relocation) {
  // interpreter specific
  // Note: No need to save/restore bcp (Z_R13) pointer since these are callee
  // saved registers and no blocking/ GC can happen in leaf calls.

  // super call
  MacroAssembler::call_VM_leaf_base(entry_point, allow_relocation);
}

void InterpreterMacroAssembler::call_VM_base(Register oop_result, Register last_java_sp,
                                             address entry_point, bool check_exceptions) {
  bool allow_relocation = true; // Fenerally valid variant. Assume code is relocated.
  // interpreter specific

  save_bcp();
  save_esp();
  // super call
  MacroAssembler::call_VM_base(oop_result, last_java_sp,
                               entry_point, allow_relocation, check_exceptions);
  restore_bcp();
}

void InterpreterMacroAssembler::call_VM_base(Register oop_result, Register last_java_sp,
                                             address entry_point, bool allow_relocation,
                                             bool check_exceptions) {
  // interpreter specific

  save_bcp();
  save_esp();
  // super call
  MacroAssembler::call_VM_base(oop_result, last_java_sp,
                               entry_point, allow_relocation, check_exceptions);
  restore_bcp();
}

void InterpreterMacroAssembler::check_and_handle_popframe(Register scratch_reg) {
  if (JvmtiExport::can_pop_frame()) {
    BLOCK_COMMENT("check_and_handle_popframe {");
    Label L;
    // Initiate popframe handling only if it is not already being
    // processed. If the flag has the popframe_processing bit set, it
    // means that this code is called *during* popframe handling - we
    // don't want to reenter.
    // TODO: Check if all four state combinations could be visible.
    // If (processing and !pending) is an invisible/impossible state,
    // there is optimization potential by testing both bits at once.
    // Then, All_Zeroes and All_Ones means skip, Mixed means doit.
    testbit(Address(Z_thread, JavaThread::popframe_condition_offset()),
            exact_log2(JavaThread::popframe_pending_bit));
    z_bfalse(L);
    testbit(Address(Z_thread, JavaThread::popframe_condition_offset()),
            exact_log2(JavaThread::popframe_processing_bit));
    z_btrue(L);

    // Call Interpreter::remove_activation_preserving_args_entry() to get the
    // address of the same-named entrypoint in the generated interpreter code.
    call_VM_leaf(CAST_FROM_FN_PTR(address, Interpreter::remove_activation_preserving_args_entry));
    // The above call should (as its only effect) return the contents of the field
    // _remove_activation_preserving_args_entry in Z_RET.
    // We just jump there to have the work done.
    z_br(Z_RET);
    // There is no way for control to fall thru here.

    bind(L);
    BLOCK_COMMENT("} check_and_handle_popframe");
  }
}


void InterpreterMacroAssembler::load_earlyret_value(TosState state) {
  Register RjvmtiState = Z_R1_scratch;
  int      tos_off     = in_bytes(JvmtiThreadState::earlyret_tos_offset());
  int      oop_off     = in_bytes(JvmtiThreadState::earlyret_oop_offset());
  int      val_off     = in_bytes(JvmtiThreadState::earlyret_value_offset());
  int      state_off   = in_bytes(JavaThread::jvmti_thread_state_offset());

  z_lg(RjvmtiState, state_off, Z_thread);

  switch (state) {
    case atos: z_lg(Z_tos, oop_off, RjvmtiState);
      store_const(Address(RjvmtiState, oop_off), 0L, 8, 8, Z_R0_scratch);
                                                    break;
    case ltos: z_lg(Z_tos, val_off, RjvmtiState);   break;
    case btos: // fall through
    case ztos: // fall through
    case ctos: // fall through
    case stos: // fall through
    case itos: z_llgf(Z_tos, val_off, RjvmtiState); break;
    case ftos: z_le(Z_ftos, val_off, RjvmtiState);  break;
    case dtos: z_ld(Z_ftos, val_off, RjvmtiState);  break;
    case vtos:   /* nothing to do */                break;
    default  : ShouldNotReachHere();
  }

  // Clean up tos value in the jvmti thread state.
  store_const(Address(RjvmtiState, val_off),   0L, 8, 8, Z_R0_scratch);
  // Set tos state field to illegal value.
  store_const(Address(RjvmtiState, tos_off), ilgl, 4, 1, Z_R0_scratch);
}

void InterpreterMacroAssembler::check_and_handle_earlyret(Register scratch_reg) {
  if (JvmtiExport::can_force_early_return()) {
    BLOCK_COMMENT("check_and_handle_earlyret {");
    Label L;
    // arg regs are save, because we are just behind the call in call_VM_base
    Register jvmti_thread_state = Z_ARG2;
    Register tmp                = Z_ARG3;
    load_and_test_long(jvmti_thread_state, Address(Z_thread, JavaThread::jvmti_thread_state_offset()));
    z_bre(L); // if (thread->jvmti_thread_state() == NULL) exit;

    // Initiate earlyret handling only if it is not already being processed.
    // If the flag has the earlyret_processing bit set, it means that this code
    // is called *during* earlyret handling - we don't want to reenter.

    assert((JvmtiThreadState::earlyret_pending != 0) && (JvmtiThreadState::earlyret_inactive == 0),
          "must fix this check, when changing the values of the earlyret enum");
    assert(JvmtiThreadState::earlyret_pending == 1, "must fix this check, when changing the values of the earlyret enum");

    load_and_test_int(tmp, Address(jvmti_thread_state, JvmtiThreadState::earlyret_state_offset()));
    z_brz(L); // if (thread->jvmti_thread_state()->_earlyret_state != JvmtiThreadState::earlyret_pending) exit;

    // Call Interpreter::remove_activation_early_entry() to get the address of the
    // same-named entrypoint in the generated interpreter code.
    assert(sizeof(TosState) == 4, "unexpected size");
    z_l(Z_ARG1, Address(jvmti_thread_state, JvmtiThreadState::earlyret_tos_offset()));
    call_VM_leaf(CAST_FROM_FN_PTR(address, Interpreter::remove_activation_early_entry), Z_ARG1);
    // The above call should (as its only effect) return the contents of the field
    // _remove_activation_preserving_args_entry in Z_RET.
    // We just jump there to have the work done.
    z_br(Z_RET);
    // There is no way for control to fall thru here.

    bind(L);
    BLOCK_COMMENT("} check_and_handle_earlyret");
  }
}

void InterpreterMacroAssembler::super_call_VM_leaf(address entry_point, Register arg_1, Register arg_2) {
  lgr_if_needed(Z_ARG1, arg_1);
  assert(arg_2 != Z_ARG1, "smashed argument");
  lgr_if_needed(Z_ARG2, arg_2);
  MacroAssembler::call_VM_leaf_base(entry_point, true);
}

void InterpreterMacroAssembler::get_cache_index_at_bcp(Register index, int bcp_offset, size_t index_size) {
  Address param(Z_bcp, bcp_offset);

  BLOCK_COMMENT("get_cache_index_at_bcp {");
  assert(bcp_offset > 0, "bcp is still pointing to start of bytecode");
  if (index_size == sizeof(u2)) {
    load_sized_value(index, param, 2, false /*signed*/);
  } else if (index_size == sizeof(u4)) {

    load_sized_value(index, param, 4, false);

    // Check if the secondary index definition is still ~x, otherwise
    // we have to change the following assembler code to calculate the
    // plain index.
    assert(ConstantPool::decode_invokedynamic_index(~123) == 123, "else change next line");
    not_(index);  // Convert to plain index.
  } else if (index_size == sizeof(u1)) {
    z_llgc(index, param);
  } else {
    ShouldNotReachHere();
  }
  BLOCK_COMMENT("}");
}


void InterpreterMacroAssembler::get_cache_and_index_at_bcp(Register cache, Register cpe_offset,
                                                           int bcp_offset, size_t index_size) {
  BLOCK_COMMENT("get_cache_and_index_at_bcp {");
  assert_different_registers(cache, cpe_offset);
  get_cache_index_at_bcp(cpe_offset, bcp_offset, index_size);
  z_lg(cache, Address(Z_fp, _z_ijava_state_neg(cpoolCache)));
  // Convert from field index to ConstantPoolCache offset in bytes.
  z_sllg(cpe_offset, cpe_offset, exact_log2(in_words(ConstantPoolCacheEntry::size()) * BytesPerWord));
  BLOCK_COMMENT("}");
}

// Kills Z_R0_scratch.
void InterpreterMacroAssembler::get_cache_and_index_and_bytecode_at_bcp(Register cache,
                                                                        Register cpe_offset,
                                                                        Register bytecode,
                                                                        int byte_no,
                                                                        int bcp_offset,
                                                                        size_t index_size) {
  BLOCK_COMMENT("get_cache_and_index_and_bytecode_at_bcp {");
  get_cache_and_index_at_bcp(cache, cpe_offset, bcp_offset, index_size);

  // We want to load (from CP cache) the bytecode that corresponds to the passed-in byte_no.
  // It is located at (cache + cpe_offset + base_offset + indices_offset + (8-1) (last byte in DW) - (byte_no+1).
  // Instead of loading, shifting and masking a DW, we just load that one byte of interest with z_llgc (unsigned).
  const int base_ix_off = in_bytes(ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset());
  const int off_in_DW   = (8-1) - (1+byte_no);
  assert(ConstantPoolCacheEntry::bytecode_1_mask == ConstantPoolCacheEntry::bytecode_2_mask, "common mask");
  assert(ConstantPoolCacheEntry::bytecode_1_mask == 0xff, "");
  load_sized_value(bytecode, Address(cache, cpe_offset, base_ix_off+off_in_DW), 1, false /*signed*/);

  BLOCK_COMMENT("}");
}

// Load object from cpool->resolved_references(index).
void InterpreterMacroAssembler::load_resolved_reference_at_index(Register result, Register index) {
  assert_different_registers(result, index);
  get_constant_pool(result);

  // Convert
  //  - from field index to resolved_references() index and
  //  - from word index to byte offset.
  // Since this is a java object, it is potentially compressed.
  Register tmp = index;  // reuse
  z_sllg(index, index, LogBytesPerHeapOop); // Offset into resolved references array.
  // Load pointer for resolved_references[] objArray.
  z_lg(result, ConstantPool::cache_offset_in_bytes(), result);
  z_lg(result, ConstantPoolCache::resolved_references_offset_in_bytes(), result);
  resolve_oop_handle(result); // Load resolved references array itself.
#ifdef ASSERT
  NearLabel index_ok;
  z_lgf(Z_R0, Address(result, arrayOopDesc::length_offset_in_bytes()));
  z_sllg(Z_R0, Z_R0, LogBytesPerHeapOop);
  compare64_and_branch(tmp, Z_R0, Assembler::bcondLow, index_ok);
  stop("resolved reference index out of bounds", 0x09256);
  bind(index_ok);
#endif
  z_agr(result, index);    // Address of indexed array element.
  load_heap_oop(result, Address(result, arrayOopDesc::base_offset_in_bytes(T_OBJECT)), tmp, noreg);
}

// load cpool->resolved_klass_at(index)
void InterpreterMacroAssembler::load_resolved_klass_at_offset(Register cpool, Register offset, Register iklass) {
  // int value = *(Rcpool->int_at_addr(which));
  // int resolved_klass_index = extract_low_short_from_int(value);
  z_llgh(offset, Address(cpool, offset, sizeof(ConstantPool) + 2)); // offset = resolved_klass_index (s390 is big-endian)
  z_sllg(offset, offset, LogBytesPerWord);                          // Convert 'index' to 'offset'
  z_lg(iklass, Address(cpool, ConstantPool::resolved_klasses_offset_in_bytes())); // iklass = cpool->_resolved_klasses
  z_lg(iklass, Address(iklass, offset, Array<Klass*>::base_offset_in_bytes()));
}

void InterpreterMacroAssembler::get_cache_entry_pointer_at_bcp(Register cache,
                                                               Register tmp,
                                                               int bcp_offset,
                                                               size_t index_size) {
  BLOCK_COMMENT("get_cache_entry_pointer_at_bcp {");
    get_cache_and_index_at_bcp(cache, tmp, bcp_offset, index_size);
    add2reg_with_index(cache, in_bytes(ConstantPoolCache::base_offset()), tmp, cache);
  BLOCK_COMMENT("}");
}

// Generate a subtype check: branch to ok_is_subtype if sub_klass is
// a subtype of super_klass. Blows registers Rsuper_klass, Rsub_klass, tmp1, tmp2.
void InterpreterMacroAssembler::gen_subtype_check(Register Rsub_klass,
                                                  Register Rsuper_klass,
                                                  Register Rtmp1,
                                                  Register Rtmp2,
                                                  Label &ok_is_subtype) {
  // Profile the not-null value's klass.
  profile_typecheck(Rtmp1, Rsub_klass, Rtmp2);

  // Do the check.
  check_klass_subtype(Rsub_klass, Rsuper_klass, Rtmp1, Rtmp2, ok_is_subtype);

  // Profile the failure of the check.
  profile_typecheck_failed(Rtmp1, Rtmp2);
}

// Pop topmost element from stack. It just disappears.
// Useful if consumed previously by access via stackTop().
void InterpreterMacroAssembler::popx(int len) {
  add2reg(Z_esp, len*Interpreter::stackElementSize);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

// Get Address object of stack top. No checks. No pop.
// Purpose: - Provide address of stack operand to exploit reg-mem operations.
//          - Avoid RISC-like mem2reg - reg-reg-op sequence.
Address InterpreterMacroAssembler::stackTop() {
  return Address(Z_esp, Interpreter::expr_offset_in_bytes(0));
}

void InterpreterMacroAssembler::pop_i(Register r) {
  z_l(r, Interpreter::expr_offset_in_bytes(0), Z_esp);
  add2reg(Z_esp, Interpreter::stackElementSize);
  assert_different_registers(r, Z_R1_scratch);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

void InterpreterMacroAssembler::pop_ptr(Register r) {
  z_lg(r, Interpreter::expr_offset_in_bytes(0), Z_esp);
  add2reg(Z_esp, Interpreter::stackElementSize);
  assert_different_registers(r, Z_R1_scratch);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

void InterpreterMacroAssembler::pop_l(Register r) {
  z_lg(r, Interpreter::expr_offset_in_bytes(0), Z_esp);
  add2reg(Z_esp, 2*Interpreter::stackElementSize);
  assert_different_registers(r, Z_R1_scratch);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

void InterpreterMacroAssembler::pop_f(FloatRegister f) {
  mem2freg_opt(f, Address(Z_esp, Interpreter::expr_offset_in_bytes(0)), false);
  add2reg(Z_esp, Interpreter::stackElementSize);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

void InterpreterMacroAssembler::pop_d(FloatRegister f) {
  mem2freg_opt(f, Address(Z_esp, Interpreter::expr_offset_in_bytes(0)), true);
  add2reg(Z_esp, 2*Interpreter::stackElementSize);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
}

void InterpreterMacroAssembler::push_i(Register r) {
  assert_different_registers(r, Z_R1_scratch);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
  z_st(r, Address(Z_esp));
  add2reg(Z_esp, -Interpreter::stackElementSize);
}

void InterpreterMacroAssembler::push_ptr(Register r) {
  z_stg(r, Address(Z_esp));
  add2reg(Z_esp, -Interpreter::stackElementSize);
}

void InterpreterMacroAssembler::push_l(Register r) {
  assert_different_registers(r, Z_R1_scratch);
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
  int offset = -Interpreter::stackElementSize;
  z_stg(r, Address(Z_esp, offset));
  clear_mem(Address(Z_esp), Interpreter::stackElementSize);
  add2reg(Z_esp, 2 * offset);
}

void InterpreterMacroAssembler::push_f(FloatRegister f) {
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
  freg2mem_opt(f, Address(Z_esp), false);
  add2reg(Z_esp, -Interpreter::stackElementSize);
}

void InterpreterMacroAssembler::push_d(FloatRegister d) {
  debug_only(verify_esp(Z_esp, Z_R1_scratch));
  int offset = -Interpreter::stackElementSize;
  freg2mem_opt(d, Address(Z_esp, offset));
  add2reg(Z_esp, 2 * offset);
}

void InterpreterMacroAssembler::push(TosState state) {
  verify_oop(Z_tos, state);
  switch (state) {
    case atos: push_ptr();           break;
    case btos: push_i();             break;
    case ztos:
    case ctos:
    case stos: push_i();             break;
    case itos: push_i();             break;
    case ltos: push_l();             break;
    case ftos: push_f();             break;
    case dtos: push_d();             break;
    case vtos: /* nothing to do */   break;
    default  : ShouldNotReachHere();
  }
}

void InterpreterMacroAssembler::pop(TosState state) {
  switch (state) {
    case atos: pop_ptr(Z_tos);       break;
    case btos: pop_i(Z_tos);         break;
    case ztos:
    case ctos:
    case stos: pop_i(Z_tos);         break;
    case itos: pop_i(Z_tos);         break;
    case ltos: pop_l(Z_tos);         break;
    case ftos: pop_f(Z_ftos);        break;
    case dtos: pop_d(Z_ftos);        break;
    case vtos: /* nothing to do */   break;
    default  : ShouldNotReachHere();
  }
  verify_oop(Z_tos, state);
}

// Helpers for swap and dup.
void InterpreterMacroAssembler::load_ptr(int n, Register val) {
  z_lg(val, Address(Z_esp, Interpreter::expr_offset_in_bytes(n)));
}

void InterpreterMacroAssembler::store_ptr(int n, Register val) {
  z_stg(val, Address(Z_esp, Interpreter::expr_offset_in_bytes(n)));
}

void InterpreterMacroAssembler::prepare_to_jump_from_interpreted(Register method) {
  // Satisfy interpreter calling convention (see generate_normal_entry()).
  z_lgr(Z_R10, Z_SP); // Set sender sp (aka initial caller sp, aka unextended sp).
  // Record top_frame_sp, because the callee might modify it, if it's compiled.
  z_stg(Z_SP, _z_ijava_state_neg(top_frame_sp), Z_fp);
  save_bcp();
  save_esp();
  z_lgr(Z_method, method); // Set Z_method (kills Z_fp!).
}

// Jump to from_interpreted entry of a call unless single stepping is possible
// in this thread in which case we must call the i2i entry.
void InterpreterMacroAssembler::jump_from_interpreted(Register method, Register temp) {
  assert_different_registers(method, Z_R10 /*used for initial_caller_sp*/, temp);
  prepare_to_jump_from_interpreted(method);

  if (JvmtiExport::can_post_interpreter_events()) {
    // JVMTI events, such as single-stepping, are implemented partly by avoiding running
    // compiled code in threads for which the event is enabled. Check here for
    // interp_only_mode if these events CAN be enabled.
    z_lg(Z_R1_scratch, Address(method, Method::from_interpreted_offset()));
    MacroAssembler::load_and_test_int(Z_R0_scratch, Address(Z_thread, JavaThread::interp_only_mode_offset()));
    z_bcr(bcondEqual, Z_R1_scratch); // Run compiled code if zero.
    // Run interpreted.
    z_lg(Z_R1_scratch, Address(method, Method::interpreter_entry_offset()));
    z_br(Z_R1_scratch);
  } else {
    // Run compiled code.
    z_lg(Z_R1_scratch, Address(method, Method::from_interpreted_offset()));
    z_br(Z_R1_scratch);
  }
}

#ifdef ASSERT
void InterpreterMacroAssembler::verify_esp(Register Resp, Register Rtemp) {
  // About to read or write Resp[0].
  // Make sure it is not in the monitors or the TOP_IJAVA_FRAME_ABI.
  address reentry = NULL;

  {
    // Check if the frame pointer in Z_fp is correct.
    NearLabel OK;
    z_cg(Z_fp, 0, Z_SP);
    z_bre(OK);
    reentry = stop_chain_static(reentry, "invalid frame pointer Z_fp");
    bind(OK);
  }
  {
    // Resp must not point into or below the operand stack,
    // i.e. IJAVA_STATE.monitors > Resp.
    NearLabel OK;
    Register Rmonitors = Rtemp;
    z_lg(Rmonitors, _z_ijava_state_neg(monitors), Z_fp);
    compareU64_and_branch(Rmonitors, Resp, bcondHigh, OK);
    reentry = stop_chain_static(reentry, "too many pops: Z_esp points into monitor area");
    bind(OK);
  }
  {
    // Resp may point to the last word of TOP_IJAVA_FRAME_ABI, but not below
    // i.e. !(Z_SP + frame::z_top_ijava_frame_abi_size - Interpreter::stackElementSize > Resp).
    NearLabel OK;
    Register Rabi_bottom = Rtemp;
    add2reg(Rabi_bottom, frame::z_top_ijava_frame_abi_size - Interpreter::stackElementSize, Z_SP);
    compareU64_and_branch(Rabi_bottom, Resp, bcondNotHigh, OK);
    reentry = stop_chain_static(reentry, "too many pushes: Z_esp points into TOP_IJAVA_FRAME_ABI");
    bind(OK);
  }
}

void InterpreterMacroAssembler::asm_assert_ijava_state_magic(Register tmp) {
  Label magic_ok;
  load_const_optimized(tmp, frame::z_istate_magic_number);
  z_cg(tmp, Address(Z_fp, _z_ijava_state_neg(magic)));
  z_bre(magic_ok);
  stop_static("error: wrong magic number in ijava_state access");
  bind(magic_ok);
}
#endif // ASSERT

void InterpreterMacroAssembler::save_bcp() {
  z_stg(Z_bcp, Address(Z_fp, _z_ijava_state_neg(bcp)));
  asm_assert_ijava_state_magic(Z_bcp);
  NOT_PRODUCT(z_lg(Z_bcp, Address(Z_fp, _z_ijava_state_neg(bcp))));
}

void InterpreterMacroAssembler::restore_bcp() {
  asm_assert_ijava_state_magic(Z_bcp);
  z_lg(Z_bcp, Address(Z_fp, _z_ijava_state_neg(bcp)));
}

void InterpreterMacroAssembler::save_esp() {
  z_stg(Z_esp, Address(Z_fp, _z_ijava_state_neg(esp)));
}

void InterpreterMacroAssembler::restore_esp() {
  asm_assert_ijava_state_magic(Z_esp);
  z_lg(Z_esp, Address(Z_fp, _z_ijava_state_neg(esp)));
}

void InterpreterMacroAssembler::get_monitors(Register reg) {
  asm_assert_ijava_state_magic(reg);
  mem2reg_opt(reg, Address(Z_fp, _z_ijava_state_neg(monitors)));
}

void InterpreterMacroAssembler::save_monitors(Register reg) {
  reg2mem_opt(reg, Address(Z_fp, _z_ijava_state_neg(monitors)));
}

void InterpreterMacroAssembler::get_mdp(Register mdp) {
  z_lg(mdp, _z_ijava_state_neg(mdx), Z_fp);
}

void InterpreterMacroAssembler::save_mdp(Register mdp) {
  z_stg(mdp, _z_ijava_state_neg(mdx), Z_fp);
}

// Values that are only read (besides initialization).
void InterpreterMacroAssembler::restore_locals() {
  asm_assert_ijava_state_magic(Z_locals);
  z_lg(Z_locals, Address(Z_fp, _z_ijava_state_neg(locals)));
}

void InterpreterMacroAssembler::get_method(Register reg) {
  asm_assert_ijava_state_magic(reg);
  z_lg(reg, Address(Z_fp, _z_ijava_state_neg(method)));
}

void InterpreterMacroAssembler::get_2_byte_integer_at_bcp(Register Rdst, int bcp_offset,
                                                          signedOrNot is_signed) {
  // Rdst is an 8-byte return value!!!

  // Unaligned loads incur only a small penalty on z/Architecture. The penalty
  // is a few (2..3) ticks, even when the load crosses a cache line
  // boundary. In case of a cache miss, the stall could, of course, be
  // much longer.

  switch (is_signed) {
    case Signed:
      z_lgh(Rdst, bcp_offset, Z_R0, Z_bcp);
     break;
   case Unsigned:
     z_llgh(Rdst, bcp_offset, Z_R0, Z_bcp);
     break;
   default:
     ShouldNotReachHere();
  }
}


void InterpreterMacroAssembler::get_4_byte_integer_at_bcp(Register Rdst, int bcp_offset,
                                                          setCCOrNot set_cc) {
  // Rdst is an 8-byte return value!!!

  // Unaligned loads incur only a small penalty on z/Architecture. The penalty
  // is a few (2..3) ticks, even when the load crosses a cache line
  // boundary. In case of a cache miss, the stall could, of course, be
  // much longer.

  // Both variants implement a sign-extending int2long load.
  if (set_cc == set_CC) {
    load_and_test_int2long(Rdst, Address(Z_bcp, (intptr_t)bcp_offset));
  } else {
    mem2reg_signed_opt(    Rdst, Address(Z_bcp, (intptr_t)bcp_offset));
  }
}

void InterpreterMacroAssembler::get_constant_pool(Register Rdst) {
  get_method(Rdst);
  mem2reg_opt(Rdst, Address(Rdst, Method::const_offset()));
  mem2reg_opt(Rdst, Address(Rdst, ConstMethod::constants_offset()));
}

void InterpreterMacroAssembler::get_cpool_and_tags(Register Rcpool, Register Rtags) {
  get_constant_pool(Rcpool);
  mem2reg_opt(Rtags, Address(Rcpool, ConstantPool::tags_offset_in_bytes()));
}

// Unlock if synchronized method.
//
// Unlock the receiver if this is a synchronized method.
// Unlock any Java monitors from syncronized blocks.
//
// If there are locked Java monitors
//   If throw_monitor_exception
//     throws IllegalMonitorStateException
//   Else if install_monitor_exception
//     installs IllegalMonitorStateException
//   Else
//     no error processing
void InterpreterMacroAssembler::unlock_if_synchronized_method(TosState state,
                                                              bool throw_monitor_exception,
                                                              bool install_monitor_exception) {
  NearLabel unlocked, unlock, no_unlock;

  {
    Register R_method = Z_ARG2;
    Register R_do_not_unlock_if_synchronized = Z_ARG3;

    // Get the value of _do_not_unlock_if_synchronized into G1_scratch.
    const Address do_not_unlock_if_synchronized(Z_thread,
                                                JavaThread::do_not_unlock_if_synchronized_offset());
    load_sized_value(R_do_not_unlock_if_synchronized, do_not_unlock_if_synchronized, 1, false /*unsigned*/);
    z_mvi(do_not_unlock_if_synchronized, false); // Reset the flag.

    // Check if synchronized method.
    get_method(R_method);
    verify_oop(Z_tos, state);
    push(state); // Save tos/result.
    testbit(method2_(R_method, access_flags), JVM_ACC_SYNCHRONIZED_BIT);
    z_bfalse(unlocked);

    // Don't unlock anything if the _do_not_unlock_if_synchronized flag
    // is set.
    compareU64_and_branch(R_do_not_unlock_if_synchronized, (intptr_t)0L, bcondNotEqual, no_unlock);
  }

  // unlock monitor

  // BasicObjectLock will be first in list, since this is a
  // synchronized method. However, need to check that the object has
  // not been unlocked by an explicit monitorexit bytecode.
  const Address monitor(Z_fp, -(frame::z_ijava_state_size + (int) sizeof(BasicObjectLock)));
  // We use Z_ARG2 so that if we go slow path it will be the correct
  // register for unlock_object to pass to VM directly.
  load_address(Z_ARG2, monitor); // Address of first monitor.
  z_lg(Z_ARG3, Address(Z_ARG2, BasicObjectLock::obj_offset_in_bytes()));
  compareU64_and_branch(Z_ARG3, (intptr_t)0L, bcondNotEqual, unlock);

  if (throw_monitor_exception) {
    // Entry already unlocked need to throw an exception.
    MacroAssembler::call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_illegal_monitor_state_exception));
    should_not_reach_here();
  } else {
    // Monitor already unlocked during a stack unroll.
    // If requested, install an illegal_monitor_state_exception.
    // Continue with stack unrolling.
    if (install_monitor_exception) {
      MacroAssembler::call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::new_illegal_monitor_state_exception));
    }
   z_bru(unlocked);
  }

  bind(unlock);

  unlock_object(Z_ARG2);

  bind(unlocked);

  // I0, I1: Might contain return value

  // Check that all monitors are unlocked.
  {
    NearLabel loop, exception, entry, restart;
    const int entry_size = frame::interpreter_frame_monitor_size() * wordSize;
    // We use Z_ARG2 so that if we go slow path it will be the correct
    // register for unlock_object to pass to VM directly.
    Register R_current_monitor = Z_ARG2;
    Register R_monitor_block_bot = Z_ARG1;
    const Address monitor_block_top(Z_fp, _z_ijava_state_neg(monitors));
    const Address monitor_block_bot(Z_fp, -frame::z_ijava_state_size);

    bind(restart);
    // Starting with top-most entry.
    z_lg(R_current_monitor, monitor_block_top);
    // Points to word before bottom of monitor block.
    load_address(R_monitor_block_bot, monitor_block_bot);
    z_bru(entry);

    // Entry already locked, need to throw exception.
    bind(exception);

    if (throw_monitor_exception) {
      // Throw exception.
      MacroAssembler::call_VM(noreg,
                              CAST_FROM_FN_PTR(address, InterpreterRuntime::
                                               throw_illegal_monitor_state_exception));
      should_not_reach_here();
    } else {
      // Stack unrolling. Unlock object and install illegal_monitor_exception.
      // Unlock does not block, so don't have to worry about the frame.
      // We don't have to preserve c_rarg1 since we are going to throw an exception.
      unlock_object(R_current_monitor);
      if (install_monitor_exception) {
        call_VM(noreg, CAST_FROM_FN_PTR(address,
                                        InterpreterRuntime::
                                        new_illegal_monitor_state_exception));
      }
      z_bru(restart);
    }

    bind(loop);
    // Check if current entry is used.
    load_and_test_long(Z_R0_scratch, Address(R_current_monitor, BasicObjectLock::obj_offset_in_bytes()));
    z_brne(exception);

    add2reg(R_current_monitor, entry_size); // Otherwise advance to next entry.
    bind(entry);
    compareU64_and_branch(R_current_monitor, R_monitor_block_bot, bcondNotEqual, loop);
  }

  bind(no_unlock);
  pop(state);
  verify_oop(Z_tos, state);
}

void InterpreterMacroAssembler::narrow(Register result, Register ret_type) {
  get_method(ret_type);
  z_lg(ret_type, Address(ret_type, in_bytes(Method::const_offset())));
  z_lb(ret_type, Address(ret_type, in_bytes(ConstMethod::result_type_offset())));

  Label notBool, notByte, notChar, done;

  // common case first
  compareU32_and_branch(ret_type, T_INT, bcondEqual, done);

  compareU32_and_branch(ret_type, T_BOOLEAN, bcondNotEqual, notBool);
  z_nilf(result, 0x1);
  z_bru(done);

  bind(notBool);
  compareU32_and_branch(ret_type, T_BYTE, bcondNotEqual, notByte);
  z_lbr(result, result);
  z_bru(done);

  bind(notByte);
  compareU32_and_branch(ret_type, T_CHAR, bcondNotEqual, notChar);
  z_nilf(result, 0xffff);
  z_bru(done);

  bind(notChar);
  // compareU32_and_branch(ret_type, T_SHORT, bcondNotEqual, notShort);
  z_lhr(result, result);

  // Nothing to do for T_INT
  bind(done);
}

// remove activation
//
// Unlock the receiver if this is a synchronized method.
// Unlock any Java monitors from syncronized blocks.
// Remove the activation from the stack.
//
// If there are locked Java monitors
//   If throw_monitor_exception
//     throws IllegalMonitorStateException
//   Else if install_monitor_exception
//     installs IllegalMonitorStateException
//   Else
//     no error processing
void InterpreterMacroAssembler::remove_activation(TosState state,
                                                  Register return_pc,
                                                  bool throw_monitor_exception,
                                                  bool install_monitor_exception,
                                                  bool notify_jvmti) {
  BLOCK_COMMENT("remove_activation {");
  unlock_if_synchronized_method(state, throw_monitor_exception, install_monitor_exception);

  // Save result (push state before jvmti call and pop it afterwards) and notify jvmti.
  notify_method_exit(false, state, notify_jvmti ? NotifyJVMTI : SkipNotifyJVMTI);

  if (StackReservedPages > 0) {
    BLOCK_COMMENT("reserved_stack_check:");
    // Test if reserved zone needs to be enabled.
    Label no_reserved_zone_enabling;

    // Compare frame pointers. There is no good stack pointer, as with stack
    // frame compression we can get different SPs when we do calls. A subsequent
    // call could have a smaller SP, so that this compare succeeds for an
    // inner call of the method annotated with ReservedStack.
    z_lg(Z_R0, Address(Z_SP, (intptr_t)_z_abi(callers_sp)));
    z_clg(Z_R0, Address(Z_thread, JavaThread::reserved_stack_activation_offset())); // Compare with frame pointer in memory.
    z_brl(no_reserved_zone_enabling);

    // Enable reserved zone again, throw stack overflow exception.
    call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::enable_stack_reserved_zone), Z_thread);
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::throw_delayed_StackOverflowError));

    should_not_reach_here();

    bind(no_reserved_zone_enabling);
  }

  verify_oop(Z_tos, state);
  verify_thread();

  pop_interpreter_frame(return_pc, Z_ARG2, Z_ARG3);
  BLOCK_COMMENT("} remove_activation");
}

// lock object
//
// Registers alive
//   monitor - Address of the BasicObjectLock to be used for locking,
//             which must be initialized with the object to lock.
//   object  - Address of the object to be locked.
void InterpreterMacroAssembler::lock_object(Register monitor, Register object) {

  if (UseHeavyMonitors) {
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorenter),
            monitor, /*check_for_exceptions=*/false);
    return;
  }

  // template code:
  //
  // markOop displaced_header = obj->mark().set_unlocked();
  // monitor->lock()->set_displaced_header(displaced_header);
  // if (Atomic::cmpxchg(/*ex=*/monitor, /*addr*/obj->mark_addr(), /*cmp*/displaced_header) == displaced_header) {
  //   // We stored the monitor address into the object's mark word.
  // } else if (THREAD->is_lock_owned((address)displaced_header))
  //   // Simple recursive case.
  //   monitor->lock()->set_displaced_header(NULL);
  // } else {
  //   // Slow path.
  //   InterpreterRuntime::monitorenter(THREAD, monitor);
  // }

  const Register displaced_header = Z_ARG5;
  const Register object_mark_addr = Z_ARG4;
  const Register current_header   = Z_ARG5;

  NearLabel done;
  NearLabel slow_case;

  // markOop displaced_header = obj->mark().set_unlocked();

  // Load markOop from object into displaced_header.
  z_lg(displaced_header, oopDesc::mark_offset_in_bytes(), object);

  if (UseBiasedLocking) {
    biased_locking_enter(object, displaced_header, Z_R1, Z_R0, done, &slow_case);
  }

  // Set displaced_header to be (markOop of object | UNLOCK_VALUE).
  z_oill(displaced_header, markOopDesc::unlocked_value);

  // monitor->lock()->set_displaced_header(displaced_header);

  // Initialize the box (Must happen before we update the object mark!).
  z_stg(displaced_header, BasicObjectLock::lock_offset_in_bytes() +
                          BasicLock::displaced_header_offset_in_bytes(), monitor);

  // if (Atomic::cmpxchg(/*ex=*/monitor, /*addr*/obj->mark_addr(), /*cmp*/displaced_header) == displaced_header) {

  // Store stack address of the BasicObjectLock (this is monitor) into object.
  add2reg(object_mark_addr, oopDesc::mark_offset_in_bytes(), object);

  z_csg(displaced_header, monitor, 0, object_mark_addr);
  assert(current_header==displaced_header, "must be same register"); // Identified two registers from z/Architecture.

  z_bre(done);

  // } else if (THREAD->is_lock_owned((address)displaced_header))
  //   // Simple recursive case.
  //   monitor->lock()->set_displaced_header(NULL);

  // We did not see an unlocked object so try the fast recursive case.

  // Check if owner is self by comparing the value in the markOop of object
  // (current_header) with the stack pointer.
  z_sgr(current_header, Z_SP);

  assert(os::vm_page_size() > 0xfff, "page size too small - change the constant");

  // The prior sequence "LGR, NGR, LTGR" can be done better
  // (Z_R1 is temp and not used after here).
  load_const_optimized(Z_R0, (~(os::vm_page_size()-1) | markOopDesc::lock_mask_in_place));
  z_ngr(Z_R0, current_header); // AND sets CC (result eq/ne 0)

  // If condition is true we are done and hence we can store 0 in the displaced
  // header indicating it is a recursive lock and be done.
  z_brne(slow_case);
  z_release();  // Membar unnecessary on zarch AND because the above csg does a sync before and after.
  z_stg(Z_R0/*==0!*/, BasicObjectLock::lock_offset_in_bytes() +
                      BasicLock::displaced_header_offset_in_bytes(), monitor);
  z_bru(done);

  // } else {
  //   // Slow path.
  //   InterpreterRuntime::monitorenter(THREAD, monitor);

  // None of the above fast optimizations worked so we have to get into the
  // slow case of monitor enter.
  bind(slow_case);

  call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorenter),
          monitor, /*check_for_exceptions=*/false);

  // }

  bind(done);
}

// Unlocks an object. Used in monitorexit bytecode and remove_activation.
//
// Registers alive
//   monitor - address of the BasicObjectLock to be used for locking,
//             which must be initialized with the object to lock.
//
// Throw IllegalMonitorException if object is not locked by current thread.
void InterpreterMacroAssembler::unlock_object(Register monitor, Register object) {

  if (UseHeavyMonitors) {
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorexit),
            monitor, /*check_for_exceptions=*/ true);
    return;
  }

// else {
  // template code:
  //
  // if ((displaced_header = monitor->displaced_header()) == NULL) {
  //   // Recursive unlock. Mark the monitor unlocked by setting the object field to NULL.
  //   monitor->set_obj(NULL);
  // } else if (Atomic::cmpxchg(displaced_header, obj->mark_addr(), monitor) == monitor) {
  //   // We swapped the unlocked mark in displaced_header into the object's mark word.
  //   monitor->set_obj(NULL);
  // } else {
  //   // Slow path.
  //   InterpreterRuntime::monitorexit(THREAD, monitor);
  // }

  const Register displaced_header = Z_ARG4;
  const Register current_header   = Z_R1;
  Address obj_entry(monitor, BasicObjectLock::obj_offset_in_bytes());
  Label done;

  if (object == noreg) {
    // In the template interpreter, we must assure that the object
    // entry in the monitor is cleared on all paths. Thus we move
    // loading up to here, and clear the entry afterwards.
    object = Z_ARG3; // Use Z_ARG3 if caller didn't pass object.
    z_lg(object, obj_entry);
  }

  assert_different_registers(monitor, object, displaced_header, current_header);

  // if ((displaced_header = monitor->displaced_header()) == NULL) {
  //   // Recursive unlock. Mark the monitor unlocked by setting the object field to NULL.
  //   monitor->set_obj(NULL);

  clear_mem(obj_entry, sizeof(oop));

  if (UseBiasedLocking) {
    // The object address from the monitor is in object.
    assert(oopDesc::mark_offset_in_bytes() == 0, "offset of _mark is not 0");
    biased_locking_exit(object, displaced_header, done);
  }

  // Test first if we are in the fast recursive case.
  MacroAssembler::load_and_test_long(displaced_header,
                                     Address(monitor, BasicObjectLock::lock_offset_in_bytes() +
                                                      BasicLock::displaced_header_offset_in_bytes()));
  z_bre(done); // displaced_header == 0 -> goto done

  // } else if (Atomic::cmpxchg(displaced_header, obj->mark_addr(), monitor) == monitor) {
  //   // We swapped the unlocked mark in displaced_header into the object's mark word.
  //   monitor->set_obj(NULL);

  // If we still have a lightweight lock, unlock the object and be done.

  // The markword is expected to be at offset 0.
  assert(oopDesc::mark_offset_in_bytes() == 0, "unlock_object: review code below");

  // We have the displaced header in displaced_header. If the lock is still
  // lightweight, it will contain the monitor address and we'll store the
  // displaced header back into the object's mark word.
  z_lgr(current_header, monitor);
  z_csg(current_header, displaced_header, 0, object);
  z_bre(done);

  // } else {
  //   // Slow path.
  //   InterpreterRuntime::monitorexit(THREAD, monitor);

  // The lock has been converted into a heavy lock and hence
  // we need to get into the slow case.
  z_stg(object, obj_entry);   // Restore object entry, has been cleared above.
  call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorexit),
          monitor,  /*check_for_exceptions=*/false);

  // }

  bind(done);
}

void InterpreterMacroAssembler::test_method_data_pointer(Register mdp, Label& zero_continue) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  load_and_test_long(mdp, Address(Z_fp, _z_ijava_state_neg(mdx)));
  z_brz(zero_continue);
}

// Set the method data pointer for the current bcp.
void InterpreterMacroAssembler::set_method_data_pointer_for_bcp() {
  assert(ProfileInterpreter, "must be profiling interpreter");
  Label    set_mdp;
  Register mdp    = Z_ARG4;
  Register method = Z_ARG5;

  get_method(method);
  // Test MDO to avoid the call if it is NULL.
  load_and_test_long(mdp, method2_(method, method_data));
  z_brz(set_mdp);

  call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::bcp_to_di), method, Z_bcp);
  // Z_RET: mdi
  // Mdo is guaranteed to be non-zero here, we checked for it before the call.
  assert(method->is_nonvolatile(), "choose nonvolatile reg or reload from frame");
  z_lg(mdp, method2_(method, method_data)); // Must reload, mdp is volatile reg.
  add2reg_with_index(mdp, in_bytes(MethodData::data_offset()), Z_RET, mdp);

  bind(set_mdp);
  save_mdp(mdp);
}

void InterpreterMacroAssembler::verify_method_data_pointer() {
  assert(ProfileInterpreter, "must be profiling interpreter");
#ifdef ASSERT
  NearLabel verify_continue;
  Register bcp_expected = Z_ARG3;
  Register mdp    = Z_ARG4;
  Register method = Z_ARG5;

  test_method_data_pointer(mdp, verify_continue); // If mdp is zero, continue
  get_method(method);

  // If the mdp is valid, it will point to a DataLayout header which is
  // consistent with the bcp. The converse is highly probable also.
  load_sized_value(bcp_expected, Address(mdp, DataLayout::bci_offset()), 2, false /*signed*/);
  z_ag(bcp_expected, Address(method, Method::const_offset()));
  load_address(bcp_expected, Address(bcp_expected, ConstMethod::codes_offset()));
  compareU64_and_branch(bcp_expected, Z_bcp, bcondEqual, verify_continue);
  call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::verify_mdp), method, Z_bcp, mdp);
  bind(verify_continue);
#endif // ASSERT
}

void InterpreterMacroAssembler::set_mdp_data_at(Register mdp_in, int constant, Register value) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  z_stg(value, constant, mdp_in);
}

void InterpreterMacroAssembler::increment_mdp_data_at(Register mdp_in,
                                                      int constant,
                                                      Register tmp,
                                                      bool decrement) {
  assert_different_registers(mdp_in, tmp);
  // counter address
  Address data(mdp_in, constant);
  const int delta = decrement ? -DataLayout::counter_increment : DataLayout::counter_increment;
  add2mem_64(Address(mdp_in, constant), delta, tmp);
}

void InterpreterMacroAssembler::set_mdp_flag_at(Register mdp_in,
                                                int flag_byte_constant) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  // Set the flag.
  z_oi(Address(mdp_in, DataLayout::flags_offset()), flag_byte_constant);
}

void InterpreterMacroAssembler::test_mdp_data_at(Register mdp_in,
                                                 int offset,
                                                 Register value,
                                                 Register test_value_out,
                                                 Label& not_equal_continue) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  if (test_value_out == noreg) {
    z_cg(value, Address(mdp_in, offset));
    z_brne(not_equal_continue);
  } else {
    // Put the test value into a register, so caller can use it:
    z_lg(test_value_out, Address(mdp_in, offset));
    compareU64_and_branch(test_value_out, value, bcondNotEqual, not_equal_continue);
  }
}

void InterpreterMacroAssembler::update_mdp_by_offset(Register mdp_in, int offset_of_disp) {
  update_mdp_by_offset(mdp_in, noreg, offset_of_disp);
}

void InterpreterMacroAssembler::update_mdp_by_offset(Register mdp_in,
                                                     Register dataidx,
                                                     int offset_of_disp) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  Address disp_address(mdp_in, dataidx, offset_of_disp);
  Assembler::z_ag(mdp_in, disp_address);
  save_mdp(mdp_in);
}

void InterpreterMacroAssembler::update_mdp_by_constant(Register mdp_in, int constant) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  add2reg(mdp_in, constant);
  save_mdp(mdp_in);
}

void InterpreterMacroAssembler::update_mdp_for_ret(Register return_bci) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  assert(return_bci->is_nonvolatile(), "choose nonvolatile reg or save/restore");
  call_VM(noreg,
          CAST_FROM_FN_PTR(address, InterpreterRuntime::update_mdp_for_ret),
          return_bci);
}

void InterpreterMacroAssembler::profile_taken_branch(Register mdp, Register bumped_count) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    // Otherwise, assign to mdp.
    test_method_data_pointer(mdp, profile_continue);

    // We are taking a branch. Increment the taken count.
    // We inline increment_mdp_data_at to return bumped_count in a register
    //increment_mdp_data_at(mdp, in_bytes(JumpData::taken_offset()));
    Address data(mdp, JumpData::taken_offset());
    z_lg(bumped_count, data);
    // 64-bit overflow is very unlikely. Saturation to 32-bit values is
    // performed when reading the counts.
    add2reg(bumped_count, DataLayout::counter_increment);
    z_stg(bumped_count, data); // Store back out

    // The method data pointer needs to be updated to reflect the new target.
    update_mdp_by_offset(mdp, in_bytes(JumpData::displacement_offset()));
    bind(profile_continue);
  }
}

// Kills Z_R1_scratch.
void InterpreterMacroAssembler::profile_not_taken_branch(Register mdp) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // We are taking a branch. Increment the not taken count.
    increment_mdp_data_at(mdp, in_bytes(BranchData::not_taken_offset()), Z_R1_scratch);

    // The method data pointer needs to be updated to correspond to
    // the next bytecode.
    update_mdp_by_constant(mdp, in_bytes(BranchData::branch_data_size()));
    bind(profile_continue);
  }
}

// Kills: Z_R1_scratch.
void InterpreterMacroAssembler::profile_call(Register mdp) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // We are making a call. Increment the count.
    increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));

    // The method data pointer needs to be updated to reflect the new target.
    update_mdp_by_constant(mdp, in_bytes(CounterData::counter_data_size()));
    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_final_call(Register mdp) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // We are making a call. Increment the count.
    increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));

    // The method data pointer needs to be updated to reflect the new target.
    update_mdp_by_constant(mdp, in_bytes(VirtualCallData::virtual_call_data_size()));
    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_virtual_call(Register receiver,
                                                     Register mdp,
                                                     Register reg2,
                                                     bool receiver_can_be_null) {
  if (ProfileInterpreter) {
    NearLabel profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    NearLabel skip_receiver_profile;
    if (receiver_can_be_null) {
      NearLabel not_null;
      compareU64_and_branch(receiver, (intptr_t)0L, bcondNotEqual, not_null);
      // We are making a call. Increment the count for null receiver.
      increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));
      z_bru(skip_receiver_profile);
      bind(not_null);
    }

    // Record the receiver type.
    record_klass_in_profile(receiver, mdp, reg2, true);
    bind(skip_receiver_profile);

    // The method data pointer needs to be updated to reflect the new target.
    update_mdp_by_constant(mdp, in_bytes(VirtualCallData::virtual_call_data_size()));
    bind(profile_continue);
  }
}

// This routine creates a state machine for updating the multi-row
// type profile at a virtual call site (or other type-sensitive bytecode).
// The machine visits each row (of receiver/count) until the receiver type
// is found, or until it runs out of rows. At the same time, it remembers
// the location of the first empty row. (An empty row records null for its
// receiver, and can be allocated for a newly-observed receiver type.)
// Because there are two degrees of freedom in the state, a simple linear
// search will not work; it must be a decision tree. Hence this helper
// function is recursive, to generate the required tree structured code.
// It's the interpreter, so we are trading off code space for speed.
// See below for example code.
void InterpreterMacroAssembler::record_klass_in_profile_helper(
                                        Register receiver, Register mdp,
                                        Register reg2, int start_row,
                                        Label& done, bool is_virtual_call) {
  if (TypeProfileWidth == 0) {
    if (is_virtual_call) {
      increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));
    }
    return;
  }

  int last_row = VirtualCallData::row_limit() - 1;
  assert(start_row <= last_row, "must be work left to do");
  // Test this row for both the receiver and for null.
  // Take any of three different outcomes:
  //   1. found receiver => increment count and goto done
  //   2. found null => keep looking for case 1, maybe allocate this cell
  //   3. found something else => keep looking for cases 1 and 2
  // Case 3 is handled by a recursive call.
  for (int row = start_row; row <= last_row; row++) {
    NearLabel next_test;
    bool test_for_null_also = (row == start_row);

    // See if the receiver is receiver[n].
    int recvr_offset = in_bytes(VirtualCallData::receiver_offset(row));
    test_mdp_data_at(mdp, recvr_offset, receiver,
                     (test_for_null_also ? reg2 : noreg),
                     next_test);
    // (Reg2 now contains the receiver from the CallData.)

    // The receiver is receiver[n]. Increment count[n].
    int count_offset = in_bytes(VirtualCallData::receiver_count_offset(row));
    increment_mdp_data_at(mdp, count_offset);
    z_bru(done);
    bind(next_test);

    if (test_for_null_also) {
      Label found_null;
      // Failed the equality check on receiver[n]... Test for null.
      z_ltgr(reg2, reg2);
      if (start_row == last_row) {
        // The only thing left to do is handle the null case.
        if (is_virtual_call) {
          z_brz(found_null);
          // Receiver did not match any saved receiver and there is no empty row for it.
          // Increment total counter to indicate polymorphic case.
          increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));
          z_bru(done);
          bind(found_null);
        } else {
          z_brnz(done);
        }
        break;
      }
      // Since null is rare, make it be the branch-taken case.
      z_brz(found_null);

      // Put all the "Case 3" tests here.
      record_klass_in_profile_helper(receiver, mdp, reg2, start_row + 1, done, is_virtual_call);

      // Found a null. Keep searching for a matching receiver,
      // but remember that this is an empty (unused) slot.
      bind(found_null);
    }
  }

  // In the fall-through case, we found no matching receiver, but we
  // observed the receiver[start_row] is NULL.

  // Fill in the receiver field and increment the count.
  int recvr_offset = in_bytes(VirtualCallData::receiver_offset(start_row));
  set_mdp_data_at(mdp, recvr_offset, receiver);
  int count_offset = in_bytes(VirtualCallData::receiver_count_offset(start_row));
  load_const_optimized(reg2, DataLayout::counter_increment);
  set_mdp_data_at(mdp, count_offset, reg2);
  if (start_row > 0) {
    z_bru(done);
  }
}

// Example state machine code for three profile rows:
//   // main copy of decision tree, rooted at row[1]
//   if (row[0].rec == rec) { row[0].incr(); goto done; }
//   if (row[0].rec != NULL) {
//     // inner copy of decision tree, rooted at row[1]
//     if (row[1].rec == rec) { row[1].incr(); goto done; }
//     if (row[1].rec != NULL) {
//       // degenerate decision tree, rooted at row[2]
//       if (row[2].rec == rec) { row[2].incr(); goto done; }
//       if (row[2].rec != NULL) { count.incr(); goto done; } // overflow
//       row[2].init(rec); goto done;
//     } else {
//       // remember row[1] is empty
//       if (row[2].rec == rec) { row[2].incr(); goto done; }
//       row[1].init(rec); goto done;
//     }
//   } else {
//     // remember row[0] is empty
//     if (row[1].rec == rec) { row[1].incr(); goto done; }
//     if (row[2].rec == rec) { row[2].incr(); goto done; }
//     row[0].init(rec); goto done;
//   }
//   done:

void InterpreterMacroAssembler::record_klass_in_profile(Register receiver,
                                                        Register mdp, Register reg2,
                                                        bool is_virtual_call) {
  assert(ProfileInterpreter, "must be profiling");
  Label done;

  record_klass_in_profile_helper(receiver, mdp, reg2, 0, done, is_virtual_call);

  bind (done);
}

void InterpreterMacroAssembler::profile_ret(Register return_bci, Register mdp) {
  if (ProfileInterpreter) {
    NearLabel profile_continue;
    uint row;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // Update the total ret count.
    increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));

    for (row = 0; row < RetData::row_limit(); row++) {
      NearLabel next_test;

      // See if return_bci is equal to bci[n]:
      test_mdp_data_at(mdp,
                       in_bytes(RetData::bci_offset(row)),
                       return_bci, noreg,
                       next_test);

      // Return_bci is equal to bci[n]. Increment the count.
      increment_mdp_data_at(mdp, in_bytes(RetData::bci_count_offset(row)));

      // The method data pointer needs to be updated to reflect the new target.
      update_mdp_by_offset(mdp, in_bytes(RetData::bci_displacement_offset(row)));
      z_bru(profile_continue);
      bind(next_test);
    }

    update_mdp_for_ret(return_bci);

    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_null_seen(Register mdp) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    set_mdp_flag_at(mdp, BitData::null_seen_byte_constant());

    // The method data pointer needs to be updated.
    int mdp_delta = in_bytes(BitData::bit_data_size());
    if (TypeProfileCasts) {
      mdp_delta = in_bytes(VirtualCallData::virtual_call_data_size());
    }
    update_mdp_by_constant(mdp, mdp_delta);

    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_typecheck_failed(Register mdp, Register tmp) {
  if (ProfileInterpreter && TypeProfileCasts) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    int count_offset = in_bytes(CounterData::count_offset());
    // Back up the address, since we have already bumped the mdp.
    count_offset -= in_bytes(VirtualCallData::virtual_call_data_size());

    // *Decrement* the counter. We expect to see zero or small negatives.
    increment_mdp_data_at(mdp, count_offset, tmp, true);

    bind (profile_continue);
  }
}

void InterpreterMacroAssembler::profile_typecheck(Register mdp, Register klass, Register reg2) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // The method data pointer needs to be updated.
    int mdp_delta = in_bytes(BitData::bit_data_size());
    if (TypeProfileCasts) {
      mdp_delta = in_bytes(VirtualCallData::virtual_call_data_size());

      // Record the object type.
      record_klass_in_profile(klass, mdp, reg2, false);
    }
    update_mdp_by_constant(mdp, mdp_delta);

    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_switch_default(Register mdp) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // Update the default case count.
    increment_mdp_data_at(mdp, in_bytes(MultiBranchData::default_count_offset()));

    // The method data pointer needs to be updated.
    update_mdp_by_offset(mdp, in_bytes(MultiBranchData::default_displacement_offset()));

    bind(profile_continue);
  }
}

// Kills: index, scratch1, scratch2.
void InterpreterMacroAssembler::profile_switch_case(Register index,
                                                    Register mdp,
                                                    Register scratch1,
                                                    Register scratch2) {
  if (ProfileInterpreter) {
    Label profile_continue;
    assert_different_registers(index, mdp, scratch1, scratch2);

    // If no method data exists, go to profile_continue.
    test_method_data_pointer(mdp, profile_continue);

    // Build the base (index * per_case_size_in_bytes()) +
    // case_array_offset_in_bytes().
    z_sllg(index, index, exact_log2(in_bytes(MultiBranchData::per_case_size())));
    add2reg(index, in_bytes(MultiBranchData::case_array_offset()));

    // Add the calculated base to the mdp -> address of the case' data.
    Address case_data_addr(mdp, index);
    Register case_data = scratch1;
    load_address(case_data, case_data_addr);

    // Update the case count.
    increment_mdp_data_at(case_data,
                          in_bytes(MultiBranchData::relative_count_offset()),
                          scratch2);

    // The method data pointer needs to be updated.
    update_mdp_by_offset(mdp,
                         index,
                         in_bytes(MultiBranchData::relative_displacement_offset()));

    bind(profile_continue);
  }
}

// kills: R0, R1, flags, loads klass from obj (if not null)
void InterpreterMacroAssembler::profile_obj_type(Register obj, Address mdo_addr, Register klass, bool cmp_done) {
  NearLabel null_seen, init_klass, do_nothing, do_update;

  // Klass = obj is allowed.
  const Register tmp = Z_R1;
  assert_different_registers(obj, mdo_addr.base(), tmp, Z_R0);
  assert_different_registers(klass, mdo_addr.base(), tmp, Z_R0);

  z_lg(tmp, mdo_addr);
  if (cmp_done) {
    z_brz(null_seen);
  } else {
    compareU64_and_branch(obj, (intptr_t)0, Assembler::bcondEqual, null_seen);
  }

  verify_oop(obj);
  load_klass(klass, obj);

  // Klass seen before, nothing to do (regardless of unknown bit).
  z_lgr(Z_R0, tmp);
  assert(Immediate::is_uimm(~TypeEntries::type_klass_mask, 16), "or change following instruction");
  z_nill(Z_R0, TypeEntries::type_klass_mask & 0xFFFF);
  compareU64_and_branch(Z_R0, klass, Assembler::bcondEqual, do_nothing);

  // Already unknown. Nothing to do anymore.
  z_tmll(tmp, TypeEntries::type_unknown);
  z_brc(Assembler::bcondAllOne, do_nothing);

  z_lgr(Z_R0, tmp);
  assert(Immediate::is_uimm(~TypeEntries::type_mask, 16), "or change following instruction");
  z_nill(Z_R0, TypeEntries::type_mask & 0xFFFF);
  compareU64_and_branch(Z_R0, (intptr_t)0, Assembler::bcondEqual, init_klass);

  // Different than before. Cannot keep accurate profile.
  z_oill(tmp, TypeEntries::type_unknown);
  z_bru(do_update);

  bind(init_klass);
  // Combine klass and null_seen bit (only used if (tmp & type_mask)==0).
  z_ogr(tmp, klass);
  z_bru(do_update);

  bind(null_seen);
  // Set null_seen if obj is 0.
  z_oill(tmp, TypeEntries::null_seen);
  // fallthru: z_bru(do_update);

  bind(do_update);
  z_stg(tmp, mdo_addr);

  bind(do_nothing);
}

void InterpreterMacroAssembler::profile_arguments_type(Register mdp, Register callee, Register tmp, bool is_virtual) {
  if (!ProfileInterpreter) {
    return;
  }

  assert_different_registers(mdp, callee, tmp);

  if (MethodData::profile_arguments() || MethodData::profile_return()) {
    Label profile_continue;

    test_method_data_pointer(mdp, profile_continue);

    int off_to_start = is_virtual ? in_bytes(VirtualCallData::virtual_call_data_size()) : in_bytes(CounterData::counter_data_size());

    z_cliy(in_bytes(DataLayout::tag_offset()) - off_to_start, mdp,
           is_virtual ? DataLayout::virtual_call_type_data_tag : DataLayout::call_type_data_tag);
    z_brne(profile_continue);

    if (MethodData::profile_arguments()) {
      NearLabel done;
      int off_to_args = in_bytes(TypeEntriesAtCall::args_data_offset());
      add2reg(mdp, off_to_args);

      for (int i = 0; i < TypeProfileArgsLimit; i++) {
        if (i > 0 || MethodData::profile_return()) {
          // If return value type is profiled we may have no argument to profile.
          z_lg(tmp, in_bytes(TypeEntriesAtCall::cell_count_offset())-off_to_args, mdp);
          add2reg(tmp, -i*TypeStackSlotEntries::per_arg_count());
          compare64_and_branch(tmp, TypeStackSlotEntries::per_arg_count(), Assembler::bcondLow, done);
        }
        z_lg(tmp, Address(callee, Method::const_offset()));
        z_lgh(tmp, Address(tmp, ConstMethod::size_of_parameters_offset()));
        // Stack offset o (zero based) from the start of the argument
        // list. For n arguments translates into offset n - o - 1 from
        // the end of the argument list. But there is an extra slot at
        // the top of the stack. So the offset is n - o from Lesp.
        z_sg(tmp, Address(mdp, in_bytes(TypeEntriesAtCall::stack_slot_offset(i))-off_to_args));
        z_sllg(tmp, tmp, Interpreter::logStackElementSize);
        Address stack_slot_addr(tmp, Z_esp);
        z_ltg(tmp, stack_slot_addr);

        Address mdo_arg_addr(mdp, in_bytes(TypeEntriesAtCall::argument_type_offset(i))-off_to_args);
        profile_obj_type(tmp, mdo_arg_addr, tmp, /*ltg did compare to 0*/ true);

        int to_add = in_bytes(TypeStackSlotEntries::per_arg_size());
        add2reg(mdp, to_add);
        off_to_args += to_add;
      }

      if (MethodData::profile_return()) {
        z_lg(tmp, in_bytes(TypeEntriesAtCall::cell_count_offset())-off_to_args, mdp);
        add2reg(tmp, -TypeProfileArgsLimit*TypeStackSlotEntries::per_arg_count());
      }

      bind(done);

      if (MethodData::profile_return()) {
        // We're right after the type profile for the last
        // argument. Tmp is the number of cells left in the
        // CallTypeData/VirtualCallTypeData to reach its end. Non null
        // if there's a return to profile.
        assert(ReturnTypeEntry::static_cell_count() < TypeStackSlotEntries::per_arg_count(), "can't move past ret type");
        z_sllg(tmp, tmp, exact_log2(DataLayout::cell_size));
        z_agr(mdp, tmp);
      }
      z_stg(mdp, _z_ijava_state_neg(mdx), Z_fp);
    } else {
      assert(MethodData::profile_return(), "either profile call args or call ret");
      update_mdp_by_constant(mdp, in_bytes(TypeEntriesAtCall::return_only_size()));
    }

    // Mdp points right after the end of the
    // CallTypeData/VirtualCallTypeData, right after the cells for the
    // return value type if there's one.
    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_return_type(Register mdp, Register ret, Register tmp) {
  assert_different_registers(mdp, ret, tmp);
  if (ProfileInterpreter && MethodData::profile_return()) {
    Label profile_continue;

    test_method_data_pointer(mdp, profile_continue);

    if (MethodData::profile_return_jsr292_only()) {
      // If we don't profile all invoke bytecodes we must make sure
      // it's a bytecode we indeed profile. We can't go back to the
      // beginning of the ProfileData we intend to update to check its
      // type because we're right after it and we don't known its
      // length.
      NearLabel do_profile;
      Address bc(Z_bcp);
      z_lb(tmp, bc);
      compare32_and_branch(tmp, Bytecodes::_invokedynamic, Assembler::bcondEqual, do_profile);
      compare32_and_branch(tmp, Bytecodes::_invokehandle, Assembler::bcondEqual, do_profile);
      get_method(tmp);
      // Supplement to 8139891: _intrinsic_id exceeded 1-byte size limit.
      if (Method::intrinsic_id_size_in_bytes() == 1) {
        z_cli(Method::intrinsic_id_offset_in_bytes(), tmp, vmIntrinsics::_compiledLambdaForm);
      } else {
        assert(Method::intrinsic_id_size_in_bytes() == 2, "size error: check Method::_intrinsic_id");
        z_lh(tmp, Method::intrinsic_id_offset_in_bytes(), Z_R0, tmp);
        z_chi(tmp, vmIntrinsics::_compiledLambdaForm);
      }
      z_brne(profile_continue);

      bind(do_profile);
    }

    Address mdo_ret_addr(mdp, -in_bytes(ReturnTypeEntry::size()));
    profile_obj_type(ret, mdo_ret_addr, tmp);

    bind(profile_continue);
  }
}

void InterpreterMacroAssembler::profile_parameters_type(Register mdp, Register tmp1, Register tmp2) {
  if (ProfileInterpreter && MethodData::profile_parameters()) {
    Label profile_continue, done;

    test_method_data_pointer(mdp, profile_continue);

    // Load the offset of the area within the MDO used for
    // parameters. If it's negative we're not profiling any parameters.
    Address parm_di_addr(mdp, in_bytes(MethodData::parameters_type_data_di_offset()) - in_bytes(MethodData::data_offset()));
    load_and_test_int2long(tmp1, parm_di_addr);
    z_brl(profile_continue);

    // Compute a pointer to the area for parameters from the offset
    // and move the pointer to the slot for the last
    // parameters. Collect profiling from last parameter down.
    // mdo start + parameters offset + array length - 1

    // Pointer to the parameter area in the MDO.
    z_agr(mdp, tmp1);

    // Offset of the current profile entry to update.
    const Register entry_offset = tmp1;
    // entry_offset = array len in number of cells.
    z_lg(entry_offset, Address(mdp, ArrayData::array_len_offset()));
    // entry_offset (number of cells) = array len - size of 1 entry
    add2reg(entry_offset, -TypeStackSlotEntries::per_arg_count());
    // entry_offset in bytes
    z_sllg(entry_offset, entry_offset, exact_log2(DataLayout::cell_size));

    Label loop;
    bind(loop);

    Address arg_off(mdp, entry_offset, ParametersTypeData::stack_slot_offset(0));
    Address arg_type(mdp, entry_offset, ParametersTypeData::type_offset(0));

    // Load offset on the stack from the slot for this parameter.
    z_lg(tmp2, arg_off);
    z_sllg(tmp2, tmp2, Interpreter::logStackElementSize);
    z_lcgr(tmp2); // Negate.

    // Profile the parameter.
    z_ltg(tmp2, Address(Z_locals, tmp2));
    profile_obj_type(tmp2, arg_type, tmp2, /*ltg did compare to 0*/ true);

    // Go to next parameter.
    z_aghi(entry_offset, -TypeStackSlotEntries::per_arg_count() * DataLayout::cell_size);
    z_brnl(loop);

    bind(profile_continue);
  }
}

// Jump if ((*counter_addr += increment) & mask) satisfies the condition.
void InterpreterMacroAssembler::increment_mask_and_jump(Address          counter_addr,
                                                        int              increment,
                                                        Address          mask,
                                                        Register         scratch,
                                                        bool             preloaded,
                                                        branch_condition cond,
                                                        Label           *where) {
  assert_different_registers(counter_addr.base(), scratch);
  if (preloaded) {
    add2reg(scratch, increment);
    reg2mem_opt(scratch, counter_addr, false);
  } else {
    if (VM_Version::has_MemWithImmALUOps() && Immediate::is_simm8(increment) && counter_addr.is_RSYform()) {
      z_alsi(counter_addr.disp20(), counter_addr.base(), increment);
      mem2reg_signed_opt(scratch, counter_addr);
    } else {
      mem2reg_signed_opt(scratch, counter_addr);
      add2reg(scratch, increment);
      reg2mem_opt(scratch, counter_addr, false);
    }
  }
  z_n(scratch, mask);
  if (where) { z_brc(cond, *where); }
}

// Get MethodCounters object for given method. Lazily allocated if necessary.
//   method    - Ptr to Method object.
//   Rcounters - Ptr to MethodCounters object associated with Method object.
//   skip      - Exit point if MethodCounters object can't be created (OOM condition).
void InterpreterMacroAssembler::get_method_counters(Register Rmethod,
                                                    Register Rcounters,
                                                    Label& skip) {
  assert_different_registers(Rmethod, Rcounters);

  BLOCK_COMMENT("get MethodCounters object {");

  Label has_counters;
  load_and_test_long(Rcounters, Address(Rmethod, Method::method_counters_offset()));
  z_brnz(has_counters);

  call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::build_method_counters), Rmethod, false);
  z_ltgr(Rcounters, Z_RET); // Runtime call returns MethodCounters object.
  z_brz(skip); // No MethodCounters, out of memory.

  bind(has_counters);

  BLOCK_COMMENT("} get MethodCounters object");
}

// Increment invocation counter in MethodCounters object.
// Return (invocation_counter+backedge_counter) as "result" in RctrSum.
// Counter values are all unsigned.
void InterpreterMacroAssembler::increment_invocation_counter(Register Rcounters, Register RctrSum) {
  assert(UseCompiler || LogTouchedMethods, "incrementing must be useful");
  assert_different_registers(Rcounters, RctrSum);

  int increment          = InvocationCounter::count_increment;
  int inv_counter_offset = in_bytes(MethodCounters::invocation_counter_offset() + InvocationCounter::counter_offset());
  int be_counter_offset  = in_bytes(MethodCounters::backedge_counter_offset()   + InvocationCounter::counter_offset());

  BLOCK_COMMENT("Increment invocation counter {");

  if (VM_Version::has_MemWithImmALUOps() && Immediate::is_simm8(increment)) {
    // Increment the invocation counter in place,
    // then add the incremented value to the backedge counter.
    z_l(RctrSum, be_counter_offset, Rcounters);
    z_alsi(inv_counter_offset, Rcounters, increment);     // Atomic increment @no extra cost!
    z_nilf(RctrSum, InvocationCounter::count_mask_value); // Mask off state bits.
    z_al(RctrSum, inv_counter_offset, Z_R0, Rcounters);
  } else {
    // This path is optimized for low register consumption
    // at the cost of somewhat higher operand delays.
    // It does not need an extra temp register.

    // Update the invocation counter.
    z_l(RctrSum, inv_counter_offset, Rcounters);
    if (RctrSum == Z_R0) {
      z_ahi(RctrSum, increment);
    } else {
      add2reg(RctrSum, increment);
    }
    z_st(RctrSum, inv_counter_offset, Rcounters);

    // Mask off the state bits.
    z_nilf(RctrSum, InvocationCounter::count_mask_value);

    // Add the backedge counter to the updated invocation counter to
    // form the result.
    z_al(RctrSum, be_counter_offset, Z_R0, Rcounters);
  }

  BLOCK_COMMENT("} Increment invocation counter");

  // Note that this macro must leave the backedge_count + invocation_count in Rtmp!
}


// increment backedge counter in MethodCounters object.
// return (invocation_counter+backedge_counter) as "result" in RctrSum
// counter values are all unsigned!
void InterpreterMacroAssembler::increment_backedge_counter(Register Rcounters, Register RctrSum) {
  assert(UseCompiler, "incrementing must be useful");
  assert_different_registers(Rcounters, RctrSum);

  int increment          = InvocationCounter::count_increment;
  int inv_counter_offset = in_bytes(MethodCounters::invocation_counter_offset() + InvocationCounter::counter_offset());
  int be_counter_offset  = in_bytes(MethodCounters::backedge_counter_offset()   + InvocationCounter::counter_offset());

  BLOCK_COMMENT("Increment backedge counter {");

  if (VM_Version::has_MemWithImmALUOps() && Immediate::is_simm8(increment)) {
    // Increment the invocation counter in place,
    // then add the incremented value to the backedge counter.
    z_l(RctrSum, inv_counter_offset, Rcounters);
    z_alsi(be_counter_offset, Rcounters, increment);      // Atomic increment @no extra cost!
    z_nilf(RctrSum, InvocationCounter::count_mask_value); // Mask off state bits.
    z_al(RctrSum, be_counter_offset, Z_R0, Rcounters);
  } else {
    // This path is optimized for low register consumption
    // at the cost of somewhat higher operand delays.
    // It does not need an extra temp register.

    // Update the invocation counter.
    z_l(RctrSum, be_counter_offset, Rcounters);
    if (RctrSum == Z_R0) {
      z_ahi(RctrSum, increment);
    } else {
      add2reg(RctrSum, increment);
    }
    z_st(RctrSum, be_counter_offset, Rcounters);

    // Mask off the state bits.
    z_nilf(RctrSum, InvocationCounter::count_mask_value);

    // Add the backedge counter to the updated invocation counter to
    // form the result.
    z_al(RctrSum, inv_counter_offset, Z_R0, Rcounters);
  }

  BLOCK_COMMENT("} Increment backedge counter");

  // Note that this macro must leave the backedge_count + invocation_count in Rtmp!
}

// Add an InterpMonitorElem to stack (see frame_s390.hpp).
void InterpreterMacroAssembler::add_monitor_to_stack(bool     stack_is_empty,
                                                     Register Rtemp1,
                                                     Register Rtemp2,
                                                     Register Rtemp3) {

  const Register Rcurr_slot = Rtemp1;
  const Register Rlimit     = Rtemp2;
  const jint delta = -frame::interpreter_frame_monitor_size() * wordSize;

  assert((delta & LongAlignmentMask) == 0,
         "sizeof BasicObjectLock must be even number of doublewords");
  assert(2 * wordSize == -delta, "this works only as long as delta == -2*wordSize");
  assert(Rcurr_slot != Z_R0, "Register must be usable as base register");
  assert_different_registers(Rlimit, Rcurr_slot, Rtemp3);

  get_monitors(Rlimit);

  // Adjust stack pointer for additional monitor entry.
  resize_frame(RegisterOrConstant((intptr_t) delta), Z_fp, false);

  if (!stack_is_empty) {
    // Must copy stack contents down.
    NearLabel next, done;

    // Rtemp := addr(Tos), Z_esp is pointing below it!
    add2reg(Rcurr_slot, wordSize, Z_esp);

    // Nothing to do, if already at monitor area.
    compareU64_and_branch(Rcurr_slot, Rlimit, bcondNotLow, done);

    bind(next);

    // Move one stack slot.
    mem2reg_opt(Rtemp3, Address(Rcurr_slot));
    reg2mem_opt(Rtemp3, Address(Rcurr_slot, delta));
    add2reg(Rcurr_slot, wordSize);
    compareU64_and_branch(Rcurr_slot, Rlimit, bcondLow, next); // Are we done?

    bind(done);
    // Done copying stack.
  }

  // Adjust expression stack and monitor pointers.
  add2reg(Z_esp, delta);
  add2reg(Rlimit, delta);
  save_monitors(Rlimit);
}

// Note: Index holds the offset in bytes afterwards.
// You can use this to store a new value (with Llocals as the base).
void InterpreterMacroAssembler::access_local_int(Register index, Register dst) {
  z_sllg(index, index, LogBytesPerWord);
  mem2reg_opt(dst, Address(Z_locals, index), false);
}

void InterpreterMacroAssembler::verify_oop(Register reg, TosState state) {
  if (state == atos) { MacroAssembler::verify_oop(reg); }
}

// Inline assembly for:
//
// if (thread is in interp_only_mode) {
//   InterpreterRuntime::post_method_entry();
// }

void InterpreterMacroAssembler::notify_method_entry() {

  // JVMTI
  // Whenever JVMTI puts a thread in interp_only_mode, method
  // entry/exit events are sent for that thread to track stack
  // depth. If it is possible to enter interp_only_mode we add
  // the code to check if the event should be sent.
  if (JvmtiExport::can_post_interpreter_events()) {
    Label jvmti_post_done;
    MacroAssembler::load_and_test_int(Z_R0, Address(Z_thread, JavaThread::interp_only_mode_offset()));
    z_bre(jvmti_post_done);
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::post_method_entry), /*check_exceptions=*/false);
    bind(jvmti_post_done);
  }
}

// Inline assembly for:
//
// if (thread is in interp_only_mode) {
//   if (!native_method) save result
//   InterpreterRuntime::post_method_exit();
//   if (!native_method) restore result
// }
// if (DTraceMethodProbes) {
//   SharedRuntime::dtrace_method_exit(thread, method);
// }
//
// For native methods their result is stored in z_ijava_state.lresult
// and z_ijava_state.fresult before coming here.
// Java methods have their result stored in the expression stack.
//
// Notice the dependency to frame::interpreter_frame_result().
void InterpreterMacroAssembler::notify_method_exit(bool native_method,
                                                   TosState state,
                                                   NotifyMethodExitMode mode) {
  // JVMTI
  // Whenever JVMTI puts a thread in interp_only_mode, method
  // entry/exit events are sent for that thread to track stack
  // depth. If it is possible to enter interp_only_mode we add
  // the code to check if the event should be sent.
  if (mode == NotifyJVMTI && JvmtiExport::can_post_interpreter_events()) {
    Label jvmti_post_done;
    MacroAssembler::load_and_test_int(Z_R0, Address(Z_thread, JavaThread::interp_only_mode_offset()));
    z_bre(jvmti_post_done);
    if (!native_method) push(state); // see frame::interpreter_frame_result()
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::post_method_exit), /*check_exceptions=*/false);
    if (!native_method) pop(state);
    bind(jvmti_post_done);
  }

#if 0
  // Dtrace currently not supported on z/Architecture.
  {
    SkipIfEqual skip(this, &DTraceMethodProbes, false);
    push(state);
    get_method(c_rarg1);
    call_VM_leaf(CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_method_exit),
                 r15_thread, c_rarg1);
    pop(state);
  }
#endif
}

void InterpreterMacroAssembler::skip_if_jvmti_mode(Label &Lskip, Register Rscratch) {
  if (!JvmtiExport::can_post_interpreter_events()) {
    return;
  }

  load_and_test_int(Rscratch, Address(Z_thread, JavaThread::interp_only_mode_offset()));
  z_brnz(Lskip);

}

// Pop the topmost TOP_IJAVA_FRAME and set it's sender_sp as new Z_SP.
// The return pc is loaded into the register return_pc.
//
// Registers updated:
//     return_pc  - The return pc of the calling frame.
//     tmp1, tmp2 - scratch
void InterpreterMacroAssembler::pop_interpreter_frame(Register return_pc, Register tmp1, Register tmp2) {
  // F0  Z_SP -> caller_sp (F1's)
  //             ...
  //             sender_sp (F1's)
  //             ...
  // F1  Z_fp -> caller_sp (F2's)
  //             return_pc (Continuation after return from F0.)
  //             ...
  // F2          caller_sp

  // Remove F0's activation. Restoring Z_SP to sender_sp reverts modifications
  // (a) by a c2i adapter and (b) by generate_fixed_frame().
  // In case (a) the new top frame F1 is an unextended compiled frame.
  // In case (b) F1 is converted from PARENT_IJAVA_FRAME to TOP_IJAVA_FRAME.

  // Case (b) seems to be redundant when returning to a interpreted caller,
  // because then the caller's top_frame_sp is installed as sp (see
  // TemplateInterpreterGenerator::generate_return_entry_for ()). But
  // pop_interpreter_frame() is also used in exception handling and there the
  // frame type of the caller is unknown, therefore top_frame_sp cannot be used,
  // so it is important that sender_sp is the caller's sp as TOP_IJAVA_FRAME.

  Register R_f1_sender_sp = tmp1;
  Register R_f2_sp = tmp2;

  // Tirst check the for the interpreter frame's magic.
  asm_assert_ijava_state_magic(R_f2_sp/*tmp*/);
  z_lg(R_f2_sp, _z_parent_ijava_frame_abi(callers_sp), Z_fp);
  z_lg(R_f1_sender_sp, _z_ijava_state_neg(sender_sp), Z_fp);
  if (return_pc->is_valid())
    z_lg(return_pc, _z_parent_ijava_frame_abi(return_pc), Z_fp);
  // Pop F0 by resizing to R_f1_sender_sp and using R_f2_sp as fp.
  resize_frame_absolute(R_f1_sender_sp, R_f2_sp, false/*load fp*/);

#ifdef ASSERT
  // The return_pc in the new top frame is dead... at least that's my
  // current understanding; to assert this I overwrite it.
  load_const_optimized(Z_ARG3, 0xb00b1);
  z_stg(Z_ARG3, _z_parent_ijava_frame_abi(return_pc), Z_SP);
#endif
}

void InterpreterMacroAssembler::verify_FPU(int stack_depth, TosState state) {
  if (VerifyFPU) {
    unimplemented("verfiyFPU");
  }
}