interp_masm_x86_32.cpp 52.2 KB
Newer Older
D
duke 已提交
1
/*
2
 * Copyright (c) 1997, 2018, Oracle and/or its affiliates. All rights reserved.
D
duke 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
 * 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.
 *
19 20 21
 * 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.
D
duke 已提交
22 23 24
 *
 */

25
#include "precompiled.hpp"
26
#include "interp_masm_x86.hpp"
27 28 29 30
#include "interpreter/interpreter.hpp"
#include "interpreter/interpreterRuntime.hpp"
#include "oops/arrayOop.hpp"
#include "oops/markOop.hpp"
31 32
#include "oops/methodData.hpp"
#include "oops/method.hpp"
33 34 35 36 37 38
#include "prims/jvmtiExport.hpp"
#include "prims/jvmtiRedefineClassesTrace.hpp"
#include "prims/jvmtiThreadState.hpp"
#include "runtime/basicLock.hpp"
#include "runtime/biasedLocking.hpp"
#include "runtime/sharedRuntime.hpp"
39
#include "runtime/thread.inline.hpp"
D
duke 已提交
40 41 42 43 44


// Implementation of InterpreterMacroAssembler
#ifdef CC_INTERP
void InterpreterMacroAssembler::get_method(Register reg) {
45 46
  movptr(reg, Address(rbp, -(sizeof(BytecodeInterpreter) + 2 * wordSize)));
  movptr(reg, Address(reg, byte_offset_of(BytecodeInterpreter, _method)));
D
duke 已提交
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
}
#endif // CC_INTERP


#ifndef CC_INTERP
void InterpreterMacroAssembler::call_VM_leaf_base(
  address entry_point,
  int     number_of_arguments
) {
  // interpreter specific
  //
  // Note: No need to save/restore bcp & locals (rsi & rdi) pointer
  //       since these are callee saved registers and no blocking/
  //       GC can happen in leaf calls.
  // Further Note: DO NOT save/restore bcp/locals. If a caller has
  // already saved them so that it can use rsi/rdi as temporaries
  // then a save/restore here will DESTROY the copy the caller
  // saved! There used to be a save_bcp() that only happened in
  // the ASSERT path (no restore_bcp). Which caused bizarre failures
  // when jvm built with ASSERTs.
#ifdef ASSERT
  { Label L;
69
    cmpptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
    jcc(Assembler::equal, L);
    stop("InterpreterMacroAssembler::call_VM_leaf_base: last_sp != NULL");
    bind(L);
  }
#endif
  // super call
  MacroAssembler::call_VM_leaf_base(entry_point, number_of_arguments);
  // interpreter specific

  // Used to ASSERT that rsi/rdi were equal to frame's bcp/locals
  // but since they may not have been saved (and we don't want to
  // save them here (see note above) the assert is invalid.
}


void InterpreterMacroAssembler::call_VM_base(
  Register oop_result,
  Register java_thread,
  Register last_java_sp,
  address  entry_point,
  int      number_of_arguments,
  bool     check_exceptions
) {
#ifdef ASSERT
  { Label L;
95
    cmpptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
    jcc(Assembler::equal, L);
    stop("InterpreterMacroAssembler::call_VM_base: last_sp != NULL");
    bind(L);
  }
#endif /* ASSERT */
  // interpreter specific
  //
  // Note: Could avoid restoring locals ptr (callee saved) - however doesn't
  //       really make a difference for these runtime calls, since they are
  //       slow anyway. Btw., bcp must be saved/restored since it may change
  //       due to GC.
  assert(java_thread == noreg , "not expecting a precomputed java thread");
  save_bcp();
  // super call
  MacroAssembler::call_VM_base(oop_result, java_thread, last_java_sp, entry_point, number_of_arguments, check_exceptions);
  // interpreter specific
  restore_bcp();
  restore_locals();
}


void InterpreterMacroAssembler::check_and_handle_popframe(Register java_thread) {
  if (JvmtiExport::can_pop_frame()) {
    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.
    Register pop_cond = java_thread;  // Not clear if any other register is available...
    movl(pop_cond, Address(java_thread, JavaThread::popframe_condition_offset()));
    testl(pop_cond, JavaThread::popframe_pending_bit);
    jcc(Assembler::zero, L);
    testl(pop_cond, JavaThread::popframe_processing_bit);
    jcc(Assembler::notZero, 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));
    jmp(rax);
    bind(L);
    get_thread(java_thread);
  }
}


void InterpreterMacroAssembler::load_earlyret_value(TosState state) {
  get_thread(rcx);
  movl(rcx, Address(rcx, JavaThread::jvmti_thread_state_offset()));
  const Address tos_addr (rcx, JvmtiThreadState::earlyret_tos_offset());
  const Address oop_addr (rcx, JvmtiThreadState::earlyret_oop_offset());
  const Address val_addr (rcx, JvmtiThreadState::earlyret_value_offset());
  const Address val_addr1(rcx, JvmtiThreadState::earlyret_value_offset()
                             + in_ByteSize(wordSize));
  switch (state) {
148
    case atos: movptr(rax, oop_addr);
149
               movptr(oop_addr, NULL_WORD);
D
duke 已提交
150
               verify_oop(rax, state);                break;
151 152
    case ltos:
               movl(rdx, val_addr1);               // fall through
D
duke 已提交
153
    case btos:                                     // fall through
K
kevinw 已提交
154
    case ztos:                                     // fall through
D
duke 已提交
155 156 157 158 159 160 161 162 163
    case ctos:                                     // fall through
    case stos:                                     // fall through
    case itos: movl(rax, val_addr);                   break;
    case ftos: fld_s(val_addr);                       break;
    case dtos: fld_d(val_addr);                       break;
    case vtos: /* nothing to do */                    break;
    default  : ShouldNotReachHere();
  }
  // Clean up tos value in the thread object
164
  movl(tos_addr,  (int32_t) ilgl);
165
  movptr(val_addr,  NULL_WORD);
166
  NOT_LP64(movptr(val_addr1, NULL_WORD));
D
duke 已提交
167 168 169 170 171 172 173
}


void InterpreterMacroAssembler::check_and_handle_earlyret(Register java_thread) {
  if (JvmtiExport::can_force_early_return()) {
    Label L;
    Register tmp = java_thread;
174 175
    movptr(tmp, Address(tmp, JavaThread::jvmti_thread_state_offset()));
    testptr(tmp, tmp);
D
duke 已提交
176 177 178 179 180 181 182 183 184 185 186 187
    jcc(Assembler::zero, 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.
    movl(tmp, Address(tmp, JvmtiThreadState::earlyret_state_offset()));
    cmpl(tmp, JvmtiThreadState::earlyret_pending);
    jcc(Assembler::notEqual, L);

    // Call Interpreter::remove_activation_early_entry() to get the address of the
    // same-named entrypoint in the generated interpreter code.
    get_thread(java_thread);
188
    movptr(tmp, Address(java_thread, JavaThread::jvmti_thread_state_offset()));
D
duke 已提交
189 190 191 192 193 194 195 196 197 198 199
    pushl(Address(tmp, JvmtiThreadState::earlyret_tos_offset()));
    call_VM_leaf(CAST_FROM_FN_PTR(address, Interpreter::remove_activation_early_entry), 1);
    jmp(rax);
    bind(L);
    get_thread(java_thread);
  }
}


void InterpreterMacroAssembler::get_unsigned_2_byte_index_at_bcp(Register reg, int bcp_offset) {
  assert(bcp_offset >= 0, "bcp is still pointing to start of bytecode");
200
  load_unsigned_short(reg, Address(rsi, bcp_offset));
201
  bswapl(reg);
D
duke 已提交
202 203 204 205
  shrl(reg, 16);
}


206
void InterpreterMacroAssembler::get_cache_index_at_bcp(Register reg, int bcp_offset, size_t index_size) {
D
duke 已提交
207
  assert(bcp_offset > 0, "bcp is still pointing to start of bytecode");
208
  if (index_size == sizeof(u2)) {
209
    load_unsigned_short(reg, Address(rsi, bcp_offset));
210
  } else if (index_size == sizeof(u4)) {
211
    assert(EnableInvokeDynamic, "giant index used only for JSR 292");
212
    movl(reg, Address(rsi, bcp_offset));
213 214 215
    // Check if the secondary index definition is still ~x, otherwise
    // we have to change the following assembler code to calculate the
    // plain index.
216
    assert(ConstantPool::decode_invokedynamic_index(~123) == 123, "else change next line");
217
    notl(reg);  // convert to plain index
218 219 220 221
  } else if (index_size == sizeof(u1)) {
    load_unsigned_byte(reg, Address(rsi, bcp_offset));
  } else {
    ShouldNotReachHere();
222 223 224 225 226
  }
}


void InterpreterMacroAssembler::get_cache_and_index_at_bcp(Register cache, Register index,
227
                                                           int bcp_offset, size_t index_size) {
228
  assert_different_registers(cache, index);
229
  get_cache_index_at_bcp(index, bcp_offset, index_size);
230
  movptr(cache, Address(rbp, frame::interpreter_frame_cache_offset * wordSize));
D
duke 已提交
231
  assert(sizeof(ConstantPoolCacheEntry) == 4*wordSize, "adjust code below");
232
  assert(exact_log2(in_words(ConstantPoolCacheEntry::size())) == 2, "else change next line");
233
  shlptr(index, 2); // convert from field index to ConstantPoolCacheEntry index
D
duke 已提交
234 235 236
}


237 238 239 240 241 242 243
void InterpreterMacroAssembler::get_cache_and_index_and_bytecode_at_bcp(Register cache,
                                                                        Register index,
                                                                        Register bytecode,
                                                                        int byte_no,
                                                                        int bcp_offset,
                                                                        size_t index_size) {
  get_cache_and_index_at_bcp(cache, index, bcp_offset, index_size);
244
  movptr(bytecode, Address(cache, index, Address::times_ptr, ConstantPoolCache::base_offset() + ConstantPoolCacheEntry::indices_offset()));
245
  const int shift_count = (1 + byte_no) * BitsPerByte;
246 247 248
  assert((byte_no == TemplateTable::f1_byte && shift_count == ConstantPoolCacheEntry::bytecode_1_shift) ||
         (byte_no == TemplateTable::f2_byte && shift_count == ConstantPoolCacheEntry::bytecode_2_shift),
         "correct shift count");
249
  shrptr(bytecode, shift_count);
250 251
  assert(ConstantPoolCacheEntry::bytecode_1_mask == ConstantPoolCacheEntry::bytecode_2_mask, "common mask");
  andptr(bytecode, ConstantPoolCacheEntry::bytecode_1_mask);
252 253 254
}


255
void InterpreterMacroAssembler::get_cache_entry_pointer_at_bcp(Register cache, Register tmp,
256
                                                               int bcp_offset, size_t index_size) {
D
duke 已提交
257
  assert(cache != tmp, "must use different register");
258
  get_cache_index_at_bcp(tmp, bcp_offset, index_size);
D
duke 已提交
259 260 261
  assert(sizeof(ConstantPoolCacheEntry) == 4*wordSize, "adjust code below");
                               // convert from field index to ConstantPoolCacheEntry index
                               // and from word offset to byte offset
262
  assert(exact_log2(in_bytes(ConstantPoolCacheEntry::size_in_bytes())) == 2 + LogBytesPerWord, "else change next line");
D
duke 已提交
263
  shll(tmp, 2 + LogBytesPerWord);
264
  movptr(cache, Address(rbp, frame::interpreter_frame_cache_offset * wordSize));
D
duke 已提交
265
                               // skip past the header
266
  addptr(cache, in_bytes(ConstantPoolCache::base_offset()));
267
  addptr(cache, tmp);            // construct pointer to cache entry
D
duke 已提交
268 269
}

270 271 272 273 274 275 276 277 278 279 280 281 282 283
void InterpreterMacroAssembler::get_method_counters(Register method,
                                                    Register mcs, Label& skip) {
  Label has_counters;
  movptr(mcs, Address(method, Method::method_counters_offset()));
  testptr(mcs, mcs);
  jcc(Assembler::notZero, has_counters);
  call_VM(noreg, CAST_FROM_FN_PTR(address,
          InterpreterRuntime::build_method_counters), method);
  movptr(mcs, Address(method,Method::method_counters_offset()));
  testptr(mcs, mcs);
  jcc(Assembler::zero, skip); // No MethodCounters allocated, OutOfMemory
  bind(has_counters);
}

284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301
// Load object from cpool->resolved_references(index)
void InterpreterMacroAssembler::load_resolved_reference_at_index(
                                           Register result, Register index) {
  assert_different_registers(result, index);
  // convert from field index to resolved_references() index and from
  // word index to byte offset. Since this is a java object, it can be compressed
  Register tmp = index;  // reuse
  shll(tmp, LogBytesPerHeapOop);

  get_constant_pool(result);
  // load pointer for resolved_references[] objArray
  movptr(result, Address(result, ConstantPool::resolved_references_offset_in_bytes()));
  // JNIHandles::resolve(obj);
  movptr(result, Address(result, 0));
  // Add in the index
  addptr(result, tmp);
  load_heap_oop(result, Address(result, arrayOopDesc::base_offset_in_bytes(T_OBJECT)));
}
D
duke 已提交
302 303 304 305 306 307

  // Generate a subtype check: branch to ok_is_subtype if sub_klass is
  // a subtype of super_klass.  EAX holds the super_klass.  Blows ECX.
  // Resets EDI to locals.  Register sub_klass cannot be any of the above.
void InterpreterMacroAssembler::gen_subtype_check( Register Rsub_klass, Label &ok_is_subtype ) {
  assert( Rsub_klass != rax, "rax, holds superklass" );
308 309
  assert( Rsub_klass != rcx, "used as a temp" );
  assert( Rsub_klass != rdi, "used as a temp, restored from locals" );
D
duke 已提交
310 311

  // Profile the not-null value's klass.
312 313 314 315 316 317
  profile_typecheck(rcx, Rsub_klass, rdi); // blows rcx, reloads rdi

  // Do the check.
  check_klass_subtype(Rsub_klass, rax, rcx, ok_is_subtype); // blows rcx

  // Profile the failure of the check.
D
duke 已提交
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338
  profile_typecheck_failed(rcx); // blows rcx
}

void InterpreterMacroAssembler::f2ieee() {
  if (IEEEPrecision) {
    fstp_s(Address(rsp, 0));
    fld_s(Address(rsp, 0));
  }
}


void InterpreterMacroAssembler::d2ieee() {
  if (IEEEPrecision) {
    fstp_d(Address(rsp, 0));
    fld_d(Address(rsp, 0));
  }
}

// Java Expression Stack

void InterpreterMacroAssembler::pop_ptr(Register r) {
339
  pop(r);
D
duke 已提交
340 341 342
}

void InterpreterMacroAssembler::pop_i(Register r) {
343
  pop(r);
D
duke 已提交
344 345 346
}

void InterpreterMacroAssembler::pop_l(Register lo, Register hi) {
347 348
  pop(lo);
  pop(hi);
D
duke 已提交
349 350 351 352
}

void InterpreterMacroAssembler::pop_f() {
  fld_s(Address(rsp, 0));
353
  addptr(rsp, 1 * wordSize);
D
duke 已提交
354 355 356 357
}

void InterpreterMacroAssembler::pop_d() {
  fld_d(Address(rsp, 0));
358
  addptr(rsp, 2 * wordSize);
D
duke 已提交
359 360 361 362 363 364 365
}


void InterpreterMacroAssembler::pop(TosState state) {
  switch (state) {
    case atos: pop_ptr(rax);                                 break;
    case btos:                                               // fall through
K
kevinw 已提交
366
    case ztos:                                               // fall through
D
duke 已提交
367 368 369 370 371 372 373 374 375 376 377 378 379
    case ctos:                                               // fall through
    case stos:                                               // fall through
    case itos: pop_i(rax);                                   break;
    case ltos: pop_l(rax, rdx);                              break;
    case ftos: pop_f();                                      break;
    case dtos: pop_d();                                      break;
    case vtos: /* nothing to do */                           break;
    default  : ShouldNotReachHere();
  }
  verify_oop(rax, state);
}

void InterpreterMacroAssembler::push_ptr(Register r) {
380
  push(r);
D
duke 已提交
381 382 383
}

void InterpreterMacroAssembler::push_i(Register r) {
384
  push(r);
D
duke 已提交
385 386 387
}

void InterpreterMacroAssembler::push_l(Register lo, Register hi) {
388 389
  push(hi);
  push(lo);
D
duke 已提交
390 391 392 393
}

void InterpreterMacroAssembler::push_f() {
  // Do not schedule for no AGI! Never write beyond rsp!
394
  subptr(rsp, 1 * wordSize);
D
duke 已提交
395 396 397 398
  fstp_s(Address(rsp, 0));
}

void InterpreterMacroAssembler::push_d(Register r) {
399 400 401
  // Do not schedule for no AGI! Never write beyond rsp!
  subptr(rsp, 2 * wordSize);
  fstp_d(Address(rsp, 0));
D
duke 已提交
402 403 404 405 406 407 408 409
}


void InterpreterMacroAssembler::push(TosState state) {
  verify_oop(rax, state);
  switch (state) {
    case atos: push_ptr(rax); break;
    case btos:                                               // fall through
K
kevinw 已提交
410
    case ztos:                                               // fall through
D
duke 已提交
411 412 413 414 415 416 417 418 419 420 421 422
    case ctos:                                               // fall through
    case stos:                                               // fall through
    case itos: push_i(rax);                                    break;
    case ltos: push_l(rax, rdx);                               break;
    case ftos: push_f();                                       break;
    case dtos: push_d(rax);                                    break;
    case vtos: /* nothing to do */                             break;
    default  : ShouldNotReachHere();
  }
}


423 424
// Helpers for swap and dup
void InterpreterMacroAssembler::load_ptr(int n, Register val) {
425
  movptr(val, Address(rsp, Interpreter::expr_offset_in_bytes(n)));
D
duke 已提交
426 427
}

428
void InterpreterMacroAssembler::store_ptr(int n, Register val) {
429
  movptr(Address(rsp, Interpreter::expr_offset_in_bytes(n)), val);
D
duke 已提交
430 431
}

432
void InterpreterMacroAssembler::prepare_to_jump_from_interpreted() {
D
duke 已提交
433
  // set sender sp
434
  lea(rsi, Address(rsp, wordSize));
D
duke 已提交
435
  // record last_sp
436
  movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), rsi);
437 438 439 440 441 442 443
}


// 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) {
  prepare_to_jump_from_interpreted();
D
duke 已提交
444 445 446 447 448 449 450 451

  if (JvmtiExport::can_post_interpreter_events()) {
    Label run_compiled_code;
    // 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.
    get_thread(temp);
    // interp_only is an int, on little endian it is sufficient to test the byte only
452
    // Is a cmpl faster?
D
duke 已提交
453
    cmpb(Address(temp, JavaThread::interp_only_mode_offset()), 0);
454
    jccb(Assembler::zero, run_compiled_code);
455
    jmp(Address(method, Method::interpreter_entry_offset()));
D
duke 已提交
456 457 458
    bind(run_compiled_code);
  }

459
  jmp(Address(method, Method::from_interpreted_offset()));
D
duke 已提交
460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478

}


// The following two routines provide a hook so that an implementation
// can schedule the dispatch in two parts.  Intel does not do this.
void InterpreterMacroAssembler::dispatch_prolog(TosState state, int step) {
  // Nothing Intel-specific to be done here.
}

void InterpreterMacroAssembler::dispatch_epilog(TosState state, int step) {
  dispatch_next(state, step);
}

void InterpreterMacroAssembler::dispatch_base(TosState state, address* table,
                                              bool verifyoop) {
  verify_FPU(1, state);
  if (VerifyActivationFrameSize) {
    Label L;
479 480
    mov(rcx, rbp);
    subptr(rcx, rsp);
D
duke 已提交
481
    int min_frame_size = (frame::link_offset - frame::interpreter_frame_initial_sp_offset) * wordSize;
482
    cmpptr(rcx, min_frame_size);
D
duke 已提交
483 484 485 486 487
    jcc(Assembler::greaterEqual, L);
    stop("broken stack frame");
    bind(L);
  }
  if (verifyoop) verify_oop(rax, state);
488
  Address index(noreg, rbx, Address::times_ptr);
D
duke 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549
  ExternalAddress tbl((address)table);
  ArrayAddress dispatch(tbl, index);
  jump(dispatch);
}


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


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

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


void InterpreterMacroAssembler::dispatch_next(TosState state, int step) {
  // load next bytecode (load before advancing rsi to prevent AGI)
  load_unsigned_byte(rbx, Address(rsi, step));
  // advance rsi
  increment(rsi, step);
  dispatch_base(state, Interpreter::dispatch_table(state));
}


void InterpreterMacroAssembler::dispatch_via(TosState state, address* table) {
  // load current bytecode
  load_unsigned_byte(rbx, Address(rsi, 0));
  dispatch_base(state, table);
}

// 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 ret_addr,
                                                  bool throw_monitor_exception,
                                                  bool install_monitor_exception,
                                                  bool notify_jvmdi) {
  // Note: Registers rax, rdx and FPU ST(0) may be in use for the result
  // check if synchronized method
  Label unlocked, unlock, no_unlock;

  get_thread(rcx);
  const Address do_not_unlock_if_synchronized(rcx,
    in_bytes(JavaThread::do_not_unlock_if_synchronized_offset()));

  movbool(rbx, do_not_unlock_if_synchronized);
550
  mov(rdi,rbx);
D
duke 已提交
551 552
  movbool(do_not_unlock_if_synchronized, false); // reset the flag

553
  movptr(rbx, Address(rbp, frame::interpreter_frame_method_offset * wordSize)); // get method access flags
554
  movl(rcx, Address(rbx, Method::access_flags_offset()));
D
duke 已提交
555 556 557 558 559 560

  testl(rcx, JVM_ACC_SYNCHRONIZED);
  jcc(Assembler::zero, unlocked);

  // Don't unlock anything if the _do_not_unlock_if_synchronized flag
  // is set.
561
  mov(rcx,rdi);
D
duke 已提交
562 563 564 565 566 567 568 569 570
  testbool(rcx);
  jcc(Assembler::notZero, no_unlock);

  // unlock monitor
  push(state);                                   // save result

  // 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(rbp, frame::interpreter_frame_initial_sp_offset * wordSize - (int)sizeof(BasicObjectLock));
571
  lea   (rdx, monitor);                          // address of first monitor
D
duke 已提交
572

573 574 575
  movptr (rax, Address(rdx, BasicObjectLock::obj_offset_in_bytes()));
  testptr(rax, rax);
  jcc    (Assembler::notZero, unlock);
D
duke 已提交
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611

  pop(state);
  if (throw_monitor_exception) {
    empty_FPU_stack();  // remove possible return value from FPU-stack, otherwise stack could overflow

    // Entry already unlocked, need to throw exception
    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) {
      empty_FPU_stack();  // remove possible return value from FPU-stack, otherwise stack could overflow
      call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::new_illegal_monitor_state_exception));
    }
    jmp(unlocked);
  }

  bind(unlock);
  unlock_object(rdx);
  pop(state);

  // Check that for block-structured locking (i.e., that all locked objects has been unlocked)
  bind(unlocked);

  // rax, rdx: Might contain return value

  // Check that all monitors are unlocked
  {
    Label loop, exception, entry, restart;
    const int entry_size               = frame::interpreter_frame_monitor_size()           * wordSize;
    const Address monitor_block_top(rbp, frame::interpreter_frame_monitor_block_top_offset * wordSize);
    const Address monitor_block_bot(rbp, frame::interpreter_frame_initial_sp_offset        * wordSize);

    bind(restart);
612 613
    movptr(rcx, monitor_block_top);           // points to current entry, starting with top-most entry
    lea(rbx, monitor_block_bot);              // points to word before bottom of monitor block
D
duke 已提交
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
    jmp(entry);

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

    if (throw_monitor_exception) {
      empty_FPU_stack();  // remove possible return value from FPU-stack, otherwise stack could overflow

      // Throw exception
      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

      push(state);
630
      mov(rdx, rcx);
D
duke 已提交
631 632 633 634 635 636 637 638 639 640 641 642
      unlock_object(rdx);
      pop(state);

      if (install_monitor_exception) {
        empty_FPU_stack();  // remove possible return value from FPU-stack, otherwise stack could overflow
        call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::new_illegal_monitor_state_exception));
      }

      jmp(restart);
    }

    bind(loop);
643
    cmpptr(Address(rcx, BasicObjectLock::obj_offset_in_bytes()), (int32_t)NULL_WORD);  // check if current entry is used
D
duke 已提交
644 645
    jcc(Assembler::notEqual, exception);

646
    addptr(rcx, entry_size);                     // otherwise advance to next entry
D
duke 已提交
647
    bind(entry);
648
    cmpptr(rcx, rbx);                            // check if bottom reached
D
duke 已提交
649 650 651 652 653 654 655 656 657 658 659 660 661
    jcc(Assembler::notEqual, loop);              // if not at bottom then check this entry
  }

  bind(no_unlock);

  // jvmti support
  if (notify_jvmdi) {
    notify_method_exit(state, NotifyJVMTI);     // preserve TOSCA
  } else {
    notify_method_exit(state, SkipNotifyJVMTI); // preserve TOSCA
  }

  // remove activation
662
  movptr(rbx, Address(rbp, frame::interpreter_frame_sender_sp_offset * wordSize)); // get sender sp
D
duke 已提交
663
  leave();                                     // remove frame anchor
664 665
  pop(ret_addr);                               // get return address
  mov(rsp, rbx);                               // set sp to sender sp
D
duke 已提交
666 667 668
  if (UseSSE) {
    // float and double are returned in xmm register in SSE-mode
    if (state == ftos && UseSSE >= 1) {
669
      subptr(rsp, wordSize);
D
duke 已提交
670 671
      fstp_s(Address(rsp, 0));
      movflt(xmm0, Address(rsp, 0));
672
      addptr(rsp, wordSize);
D
duke 已提交
673
    } else if (state == dtos && UseSSE >= 2) {
674
      subptr(rsp, 2*wordSize);
D
duke 已提交
675 676
      fstp_d(Address(rsp, 0));
      movdbl(xmm0, Address(rsp, 0));
677
      addptr(rsp, 2*wordSize);
D
duke 已提交
678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707
    }
  }
}

#endif /* !CC_INTERP */


// Lock object
//
// Argument: rdx : Points to BasicObjectLock to be used for locking. Must
// be initialized with object to lock
void InterpreterMacroAssembler::lock_object(Register lock_reg) {
  assert(lock_reg == rdx, "The argument is only for looks. It must be rdx");

  if (UseHeavyMonitors) {
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorenter), lock_reg);
  } else {

    Label done;

    const Register swap_reg = rax;  // Must use rax, for cmpxchg instruction
    const Register obj_reg  = rcx;  // Will contain the oop

    const int obj_offset = BasicObjectLock::obj_offset_in_bytes();
    const int lock_offset = BasicObjectLock::lock_offset_in_bytes ();
    const int mark_offset = lock_offset + BasicLock::displaced_header_offset_in_bytes();

    Label slow_case;

    // Load object pointer into obj_reg %rcx
708
    movptr(obj_reg, Address(lock_reg, obj_offset));
D
duke 已提交
709 710 711 712 713 714 715 716

    if (UseBiasedLocking) {
      // Note: we use noreg for the temporary register since it's hard
      // to come up with a free register on all incoming code paths
      biased_locking_enter(lock_reg, obj_reg, swap_reg, noreg, false, done, &slow_case);
    }

    // Load immediate 1 into swap_reg %rax,
717
    movptr(swap_reg, (int32_t)1);
D
duke 已提交
718 719

    // Load (object->mark() | 1) into swap_reg %rax,
720
    orptr(swap_reg, Address(obj_reg, 0));
D
duke 已提交
721 722

    // Save (object->mark() | 1) into BasicLock's displaced header
723
    movptr(Address(lock_reg, mark_offset), swap_reg);
D
duke 已提交
724 725 726 727 728

    assert(lock_offset == 0, "displached header must be first word in BasicObjectLock");
    if (os::is_MP()) {
      lock();
    }
729
    cmpxchgptr(lock_reg, Address(obj_reg, 0));
D
duke 已提交
730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
    if (PrintBiasedLockingStatistics) {
      cond_inc32(Assembler::zero,
                 ExternalAddress((address) BiasedLocking::fast_path_entry_count_addr()));
    }
    jcc(Assembler::zero, done);

    // Test if the oopMark is an obvious stack pointer, i.e.,
    //  1) (mark & 3) == 0, and
    //  2) rsp <= mark < mark + os::pagesize()
    //
    // These 3 tests can be done by evaluating the following
    // expression: ((mark - rsp) & (3 - os::vm_page_size())),
    // assuming both stack pointer and pagesize have their
    // least significant 2 bits clear.
    // NOTE: the oopMark is in swap_reg %rax, as the result of cmpxchg
745 746
    subptr(swap_reg, rsp);
    andptr(swap_reg, 3 - os::vm_page_size());
D
duke 已提交
747 748

    // Save the test result, for recursive case, the result is zero
749
    movptr(Address(lock_reg, mark_offset), swap_reg);
D
duke 已提交
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788

    if (PrintBiasedLockingStatistics) {
      cond_inc32(Assembler::zero,
                 ExternalAddress((address) BiasedLocking::fast_path_entry_count_addr()));
    }
    jcc(Assembler::zero, done);

    bind(slow_case);

    // Call the runtime routine for slow case
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorenter), lock_reg);

    bind(done);
  }
}


// Unlocks an object. Used in monitorexit bytecode and remove_activation.
//
// Argument: rdx : Points to BasicObjectLock structure for lock
// Throw an IllegalMonitorException if object is not locked by current thread
//
// Uses: rax, rbx, rcx, rdx
void InterpreterMacroAssembler::unlock_object(Register lock_reg) {
  assert(lock_reg == rdx, "The argument is only for looks. It must be rdx");

  if (UseHeavyMonitors) {
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorexit), lock_reg);
  } else {
    Label done;

    const Register swap_reg   = rax;  // Must use rax, for cmpxchg instruction
    const Register header_reg = rbx;  // Will contain the old oopMark
    const Register obj_reg    = rcx;  // Will contain the oop

    save_bcp(); // Save in case of exception

    // Convert from BasicObjectLock structure to object and BasicLock structure
    // Store the BasicLock address into %rax,
789
    lea(swap_reg, Address(lock_reg, BasicObjectLock::lock_offset_in_bytes()));
D
duke 已提交
790 791

    // Load oop into obj_reg(%rcx)
792
    movptr(obj_reg, Address(lock_reg, BasicObjectLock::obj_offset_in_bytes ()));
D
duke 已提交
793 794

    // Free entry
795
    movptr(Address(lock_reg, BasicObjectLock::obj_offset_in_bytes()), NULL_WORD);
D
duke 已提交
796 797 798 799 800 801

    if (UseBiasedLocking) {
      biased_locking_exit(obj_reg, header_reg, done);
    }

    // Load the old header from BasicLock structure
802
    movptr(header_reg, Address(swap_reg, BasicLock::displaced_header_offset_in_bytes()));
D
duke 已提交
803 804

    // Test for recursion
805
    testptr(header_reg, header_reg);
D
duke 已提交
806 807 808 809 810 811

    // zero for recursive case
    jcc(Assembler::zero, done);

    // Atomic swap back the old header
    if (os::is_MP()) lock();
812
    cmpxchgptr(header_reg, Address(obj_reg, 0));
D
duke 已提交
813 814 815 816 817

    // zero for recursive case
    jcc(Assembler::zero, done);

    // Call the runtime routine for slow case.
818
    movptr(Address(lock_reg, BasicObjectLock::obj_offset_in_bytes()), obj_reg); // restore obj
D
duke 已提交
819 820 821 822 823 824 825 826 827 828 829 830 831 832
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::monitorexit), lock_reg);

    bind(done);

    restore_bcp();
  }
}


#ifndef CC_INTERP

// Test ImethodDataPtr.  If it is null, continue at the specified label
void InterpreterMacroAssembler::test_method_data_pointer(Register mdp, Label& zero_continue) {
  assert(ProfileInterpreter, "must be profiling interpreter");
833 834
  movptr(mdp, Address(rbp, frame::interpreter_frame_mdx_offset * wordSize));
  testptr(mdp, mdp);
D
duke 已提交
835 836 837 838 839 840 841
  jcc(Assembler::zero, 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");
842
  Label set_mdp;
843 844
  push(rax);
  push(rbx);
D
duke 已提交
845 846 847

  get_method(rbx);
  // Test MDO to avoid the call if it is NULL.
848
  movptr(rax, Address(rbx, in_bytes(Method::method_data_offset())));
849
  testptr(rax, rax);
850
  jcc(Assembler::zero, set_mdp);
D
duke 已提交
851 852 853 854
  // rbx,: method
  // rsi: bcp
  call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::bcp_to_di), rbx, rsi);
  // rax,: mdi
855
  // mdo is guaranteed to be non-zero here, we checked for it before the call.
856 857
  movptr(rbx, Address(rbx, in_bytes(Method::method_data_offset())));
  addptr(rbx, in_bytes(MethodData::data_offset()));
858 859 860
  addptr(rax, rbx);
  bind(set_mdp);
  movptr(Address(rbp, frame::interpreter_frame_mdx_offset * wordSize), rax);
861 862
  pop(rbx);
  pop(rax);
D
duke 已提交
863 864 865 866 867 868
}

void InterpreterMacroAssembler::verify_method_data_pointer() {
  assert(ProfileInterpreter, "must be profiling interpreter");
#ifdef ASSERT
  Label verify_continue;
869 870 871 872
  push(rax);
  push(rbx);
  push(rcx);
  push(rdx);
D
duke 已提交
873 874 875 876 877
  test_method_data_pointer(rcx, verify_continue); // If mdp is zero, continue
  get_method(rbx);

  // If the mdp is valid, it will point to a DataLayout header which is
  // consistent with the bcp.  The converse is highly probable also.
878
  load_unsigned_short(rdx, Address(rcx, in_bytes(DataLayout::bci_offset())));
879 880
  addptr(rdx, Address(rbx, Method::const_offset()));
  lea(rdx, Address(rdx, ConstMethod::codes_offset()));
881
  cmpptr(rdx, rsi);
D
duke 已提交
882 883 884 885 886 887
  jcc(Assembler::equal, verify_continue);
  // rbx,: method
  // rsi: bcp
  // rcx: mdp
  call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::verify_mdp), rbx, rsi, rcx);
  bind(verify_continue);
888 889 890 891
  pop(rdx);
  pop(rcx);
  pop(rbx);
  pop(rax);
D
duke 已提交
892 893 894 895 896
#endif // ASSERT
}


void InterpreterMacroAssembler::set_mdp_data_at(Register mdp_in, int constant, Register value) {
897 898
  // %%% this seems to be used to store counter data which is surely 32bits
  // however 64bit side stores 64 bits which seems wrong
D
duke 已提交
899 900
  assert(ProfileInterpreter, "must be profiling interpreter");
  Address data(mdp_in, constant);
901
  movptr(data, value);
D
duke 已提交
902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920
}


void InterpreterMacroAssembler::increment_mdp_data_at(Register mdp_in,
                                                      int constant,
                                                      bool decrement) {
  // Counter address
  Address data(mdp_in, constant);

  increment_mdp_data_at(data, decrement);
}


void InterpreterMacroAssembler::increment_mdp_data_at(Address data,
                                                      bool decrement) {

  assert( DataLayout::counter_increment==1, "flow-free idiom only works with 1" );
  assert(ProfileInterpreter, "must be profiling interpreter");

921
  // %%% 64bit treats this as 64 bit which seems unlikely
D
duke 已提交
922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967
  if (decrement) {
    // Decrement the register.  Set condition codes.
    addl(data, -DataLayout::counter_increment);
    // If the decrement causes the counter to overflow, stay negative
    Label L;
    jcc(Assembler::negative, L);
    addl(data, DataLayout::counter_increment);
    bind(L);
  } else {
    assert(DataLayout::counter_increment == 1,
           "flow-free idiom only works with 1");
    // Increment the register.  Set carry flag.
    addl(data, DataLayout::counter_increment);
    // If the increment causes the counter to overflow, pull back by 1.
    sbbl(data, 0);
  }
}


void InterpreterMacroAssembler::increment_mdp_data_at(Register mdp_in,
                                                      Register reg,
                                                      int constant,
                                                      bool decrement) {
  Address data(mdp_in, reg, Address::times_1, constant);

  increment_mdp_data_at(data, decrement);
}


void InterpreterMacroAssembler::set_mdp_flag_at(Register mdp_in, int flag_byte_constant) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  int header_offset = in_bytes(DataLayout::header_offset());
  int header_bits = DataLayout::flag_mask_to_header_mask(flag_byte_constant);
  // Set the flag
  orl(Address(mdp_in, header_offset), header_bits);
}



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) {
968
    cmpptr(value, Address(mdp_in, offset));
D
duke 已提交
969 970
  } else {
    // Put the test value into a register, so caller can use it:
971 972
    movptr(test_value_out, Address(mdp_in, offset));
    cmpptr(test_value_out, value);
D
duke 已提交
973 974 975 976 977 978 979 980
  }
  jcc(Assembler::notEqual, not_equal_continue);
}


void InterpreterMacroAssembler::update_mdp_by_offset(Register mdp_in, int offset_of_disp) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  Address disp_address(mdp_in, offset_of_disp);
981 982
  addptr(mdp_in,disp_address);
  movptr(Address(rbp, frame::interpreter_frame_mdx_offset * wordSize), mdp_in);
D
duke 已提交
983 984 985 986 987 988
}


void InterpreterMacroAssembler::update_mdp_by_offset(Register mdp_in, Register reg, int offset_of_disp) {
  assert(ProfileInterpreter, "must be profiling interpreter");
  Address disp_address(mdp_in, reg, Address::times_1, offset_of_disp);
989 990
  addptr(mdp_in, disp_address);
  movptr(Address(rbp, frame::interpreter_frame_mdx_offset * wordSize), mdp_in);
D
duke 已提交
991 992 993 994 995
}


void InterpreterMacroAssembler::update_mdp_by_constant(Register mdp_in, int constant) {
  assert(ProfileInterpreter, "must be profiling interpreter");
996 997
  addptr(mdp_in, constant);
  movptr(Address(rbp, frame::interpreter_frame_mdx_offset * wordSize), mdp_in);
D
duke 已提交
998 999 1000 1001 1002
}


void InterpreterMacroAssembler::update_mdp_for_ret(Register return_bci) {
  assert(ProfileInterpreter, "must be profiling interpreter");
1003
  push(return_bci);             // save/restore across call_VM
D
duke 已提交
1004
  call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::update_mdp_for_ret), return_bci);
1005
  pop(return_bci);
D
duke 已提交
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
}


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, in_bytes(JumpData::taken_offset()));
1021 1022

    // %%% 64bit treats these cells as 64 bit but they seem to be 32 bit
D
duke 已提交
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
    movl(bumped_count,data);
    assert( DataLayout::counter_increment==1, "flow-free idiom only works with 1" );
    addl(bumped_count, DataLayout::counter_increment);
    sbbl(bumped_count, 0);
    movl(data,bumped_count);    // 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);
  }
}


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()));

    // 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);
  }
}

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);
  }
}


1086 1087 1088
void InterpreterMacroAssembler::profile_virtual_call(Register receiver, Register mdp,
                                                     Register reg2,
                                                     bool receiver_can_be_null) {
D
duke 已提交
1089 1090 1091 1092 1093 1094
  if (ProfileInterpreter) {
    Label profile_continue;

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

1095 1096
    Label skip_receiver_profile;
    if (receiver_can_be_null) {
1097
      Label not_null;
1098
      testptr(receiver, receiver);
1099 1100 1101 1102 1103
      jccb(Assembler::notZero, not_null);
      // We are making a call.  Increment the count for null receiver.
      increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));
      jmp(skip_receiver_profile);
      bind(not_null);
1104 1105
    }

D
duke 已提交
1106
    // Record the receiver type.
1107
    record_klass_in_profile(receiver, mdp, reg2, true);
1108
    bind(skip_receiver_profile);
D
duke 已提交
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120

    // 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::record_klass_in_profile_helper(
                                        Register receiver, Register mdp,
1121 1122 1123 1124 1125 1126
                                        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()));
    }
1127
    return;
1128
  }
1129

D
duke 已提交
1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
  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++) {
    Label 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);
    jmp(done);
    bind(next_test);

    if (row == start_row) {
1156
      Label found_null;
D
duke 已提交
1157
      // Failed the equality check on receiver[n]...  Test for null.
1158
      testptr(reg2, reg2);
D
duke 已提交
1159 1160
      if (start_row == last_row) {
        // The only thing left to do is handle the null case.
1161 1162 1163
        if (is_virtual_call) {
          jccb(Assembler::zero, found_null);
          // Receiver did not match any saved receiver and there is no empty row for it.
1164
          // Increment total counter to indicate polymorphic case.
1165 1166 1167 1168 1169 1170
          increment_mdp_data_at(mdp, in_bytes(CounterData::count_offset()));
          jmp(done);
          bind(found_null);
        } else {
          jcc(Assembler::notZero, done);
        }
D
duke 已提交
1171 1172 1173 1174 1175 1176
        break;
      }
      // Since null is rare, make it be the branch-taken case.
      jcc(Assembler::zero, found_null);

      // Put all the "Case 3" tests here.
1177
      record_klass_in_profile_helper(receiver, mdp, reg2, start_row + 1, done, is_virtual_call);
D
duke 已提交
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191

      // 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));
N
never 已提交
1192
  movptr(reg2, (intptr_t)DataLayout::counter_increment);
D
duke 已提交
1193
  set_mdp_data_at(mdp, count_offset, reg2);
1194 1195 1196
  if (start_row > 0) {
    jmp(done);
  }
D
duke 已提交
1197 1198 1199
}

void InterpreterMacroAssembler::record_klass_in_profile(Register receiver,
1200 1201
                                                        Register mdp, Register reg2,
                                                        bool is_virtual_call) {
D
duke 已提交
1202 1203 1204
  assert(ProfileInterpreter, "must be profiling");
  Label done;

1205
  record_klass_in_profile_helper(receiver, mdp, reg2, 0, done, is_virtual_call);
D
duke 已提交
1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250

  bind (done);
}

void InterpreterMacroAssembler::profile_ret(Register return_bci, Register mdp) {
  if (ProfileInterpreter) {
    Label 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++) {
      Label 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)));
      jmp(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);

1251 1252
    set_mdp_flag_at(mdp, BitData::null_seen_byte_constant());

D
duke 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
    // 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) {
  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, 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.
1298
      record_klass_in_profile(klass, mdp, reg2, false);
D
duke 已提交
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
      assert(reg2 == rdi, "we know how to fix this blown reg");
      restore_locals();         // Restore EDI
    }
    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);
  }
}


void InterpreterMacroAssembler::profile_switch_case(Register index, Register mdp, Register reg2) {
  if (ProfileInterpreter) {
    Label profile_continue;

    // 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()
N
never 已提交
1335
    movptr(reg2, (intptr_t)in_bytes(MultiBranchData::per_case_size()));
1336 1337 1338 1339
    // index is positive and so should have correct value if this code were
    // used on 64bits
    imulptr(index, reg2);
    addptr(index, in_bytes(MultiBranchData::case_array_offset()));
D
duke 已提交
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388

    // Update the case count
    increment_mdp_data_at(mdp, index, in_bytes(MultiBranchData::relative_count_offset()));

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

    bind (profile_continue);
  }
}

#endif // !CC_INTERP



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


#ifndef CC_INTERP
void InterpreterMacroAssembler::verify_FPU(int stack_depth, TosState state) {
  if (state == ftos || state == dtos) MacroAssembler::verify_FPU(stack_depth);
}

#endif /* CC_INTERP */


void InterpreterMacroAssembler::notify_method_entry() {
  // Whenever JVMTI is interp_only_mode, method entry/exit events are sent 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 L;
    get_thread(rcx);
    movl(rcx, Address(rcx, JavaThread::interp_only_mode_offset()));
    testl(rcx,rcx);
    jcc(Assembler::zero, L);
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::post_method_entry));
    bind(L);
  }

  {
    SkipIfEqual skip_if(this, &DTraceMethodProbes, 0);
    get_thread(rcx);
    get_method(rbx);
    call_VM_leaf(
      CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_method_entry), rcx, rbx);
  }
1389 1390 1391 1392 1393 1394 1395 1396 1397

  // RedefineClasses() tracing support for obsolete method entry
  if (RC_TRACE_IN_RANGE(0x00001000, 0x00002000)) {
    get_thread(rcx);
    get_method(rbx);
    call_VM_leaf(
      CAST_FROM_FN_PTR(address, SharedRuntime::rc_trace_method_entry),
      rcx, rbx);
  }
D
duke 已提交
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
}


void InterpreterMacroAssembler::notify_method_exit(
    TosState state, NotifyMethodExitMode mode) {
  // Whenever JVMTI is interp_only_mode, method entry/exit events are sent 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 L;
    // Note: frame::interpreter_frame_result has a dependency on how the
    // method result is saved across the call to post_method_exit. If this
    // is changed then the interpreter_frame_result implementation will
    // need to be updated too.

    // For c++ interpreter the result is always stored at a known location in the frame
    // template interpreter will leave it on the top of the stack.
    NOT_CC_INTERP(push(state);)
    get_thread(rcx);
    movl(rcx, Address(rcx, JavaThread::interp_only_mode_offset()));
    testl(rcx,rcx);
    jcc(Assembler::zero, L);
    call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::post_method_exit));
    bind(L);
    NOT_CC_INTERP(pop(state);)
  }

  {
    SkipIfEqual skip_if(this, &DTraceMethodProbes, 0);
1427
    NOT_CC_INTERP(push(state));
D
duke 已提交
1428 1429 1430 1431 1432
    get_thread(rbx);
    get_method(rcx);
    call_VM_leaf(
      CAST_FROM_FN_PTR(address, SharedRuntime::dtrace_method_exit),
      rbx, rcx);
1433
    NOT_CC_INTERP(pop(state));
D
duke 已提交
1434 1435
  }
}
I
iveresov 已提交
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447

// Jump if ((*counter_addr += increment) & mask) satisfies the condition.
void InterpreterMacroAssembler::increment_mask_and_jump(Address counter_addr,
                                                        int increment, int mask,
                                                        Register scratch, bool preloaded,
                                                        Condition cond, Label* where) {
  if (!preloaded) {
    movl(scratch, counter_addr);
  }
  incrementl(scratch, increment);
  movl(counter_addr, scratch);
  andl(scratch, mask);
1448 1449 1450
  if (where != NULL) {
    jcc(cond, *where);
  }
I
iveresov 已提交
1451
}