templateInterpreter_x86_64.cpp 70.4 KB
Newer Older
D
duke 已提交
1
/*
2
 * Copyright (c) 2003, 2013, 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 "asm/macroAssembler.hpp"
27 28 29 30 31 32
#include "interpreter/bytecodeHistogram.hpp"
#include "interpreter/interpreter.hpp"
#include "interpreter/interpreterGenerator.hpp"
#include "interpreter/interpreterRuntime.hpp"
#include "interpreter/templateTable.hpp"
#include "oops/arrayOop.hpp"
33 34
#include "oops/methodData.hpp"
#include "oops/method.hpp"
35 36 37 38 39 40 41 42 43 44 45 46
#include "oops/oop.inline.hpp"
#include "prims/jvmtiExport.hpp"
#include "prims/jvmtiThreadState.hpp"
#include "runtime/arguments.hpp"
#include "runtime/deoptimization.hpp"
#include "runtime/frame.inline.hpp"
#include "runtime/sharedRuntime.hpp"
#include "runtime/stubRoutines.hpp"
#include "runtime/synchronizer.hpp"
#include "runtime/timer.hpp"
#include "runtime/vframeArray.hpp"
#include "utilities/debug.hpp"
47
#include "utilities/macros.hpp"
D
duke 已提交
48 49 50

#define __ _masm->

51 52
#ifndef CC_INTERP

D
duke 已提交
53 54 55 56 57 58 59 60 61 62 63 64
const int method_offset = frame::interpreter_frame_method_offset * wordSize;
const int bci_offset    = frame::interpreter_frame_bcx_offset    * wordSize;
const int locals_offset = frame::interpreter_frame_locals_offset * wordSize;

//-----------------------------------------------------------------------------

address TemplateInterpreterGenerator::generate_StackOverflowError_handler() {
  address entry = __ pc();

#ifdef ASSERT
  {
    Label L;
65 66 67 68 69
    __ lea(rax, Address(rbp,
                        frame::interpreter_frame_monitor_block_top_offset *
                        wordSize));
    __ cmpptr(rax, rsp); // rax = maximal rsp for current rbp (stack
                         // grows negative)
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 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
    __ jcc(Assembler::aboveEqual, L); // check if frame is complete
    __ stop ("interpreter frame not set up");
    __ bind(L);
  }
#endif // ASSERT
  // Restore bcp under the assumption that the current frame is still
  // interpreted
  __ restore_bcp();

  // expression stack must be empty before entering the VM if an
  // exception happened
  __ empty_expression_stack();
  // throw exception
  __ call_VM(noreg,
             CAST_FROM_FN_PTR(address,
                              InterpreterRuntime::throw_StackOverflowError));
  return entry;
}

address TemplateInterpreterGenerator::generate_ArrayIndexOutOfBounds_handler(
        const char* name) {
  address entry = __ pc();
  // expression stack must be empty before entering the VM if an
  // exception happened
  __ empty_expression_stack();
  // setup parameters
  // ??? convention: expect aberrant index in register ebx
  __ lea(c_rarg1, ExternalAddress((address)name));
  __ call_VM(noreg,
             CAST_FROM_FN_PTR(address,
                              InterpreterRuntime::
                              throw_ArrayIndexOutOfBoundsException),
             c_rarg1, rbx);
  return entry;
}

address TemplateInterpreterGenerator::generate_ClassCastException_handler() {
  address entry = __ pc();

  // object is at TOS
110
  __ pop(c_rarg1);
D
duke 已提交
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129

  // expression stack must be empty before entering the VM if an
  // exception happened
  __ empty_expression_stack();

  __ call_VM(noreg,
             CAST_FROM_FN_PTR(address,
                              InterpreterRuntime::
                              throw_ClassCastException),
             c_rarg1);
  return entry;
}

address TemplateInterpreterGenerator::generate_exception_handler_common(
        const char* name, const char* message, bool pass_oop) {
  assert(!pass_oop || message == NULL, "either oop or message but not both");
  address entry = __ pc();
  if (pass_oop) {
    // object is at TOS
130
    __ pop(c_rarg2);
D
duke 已提交
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
  }
  // expression stack must be empty before entering the VM if an
  // exception happened
  __ empty_expression_stack();
  // setup parameters
  __ lea(c_rarg1, ExternalAddress((address)name));
  if (pass_oop) {
    __ call_VM(rax, CAST_FROM_FN_PTR(address,
                                     InterpreterRuntime::
                                     create_klass_exception),
               c_rarg1, c_rarg2);
  } else {
    // kind of lame ExternalAddress can't take NULL because
    // external_word_Relocation will assert.
    if (message != NULL) {
      __ lea(c_rarg2, ExternalAddress((address)message));
    } else {
      __ movptr(c_rarg2, NULL_WORD);
    }
    __ call_VM(rax,
               CAST_FROM_FN_PTR(address, InterpreterRuntime::create_exception),
               c_rarg1, c_rarg2);
  }
  // throw exception
  __ jump(ExternalAddress(Interpreter::throw_exception_entry()));
  return entry;
}


address TemplateInterpreterGenerator::generate_continuation_for(TosState state) {
  address entry = __ pc();
  // NULL last_sp until next java call
163
  __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
164 165 166 167 168
  __ dispatch_next(state);
  return entry;
}


169
address TemplateInterpreterGenerator::generate_return_entry_for(TosState state, int step) {
D
duke 已提交
170 171 172
  address entry = __ pc();

  // Restore stack bottom in case i2c adjusted stack
173
  __ movptr(rsp, Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize));
D
duke 已提交
174
  // and NULL it as marker that esp is now tos until next java call
175
  __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
176 177 178

  __ restore_bcp();
  __ restore_locals();
179

180 181 182 183 184
  Label L_got_cache, L_giant_index;
  if (EnableInvokeDynamic) {
    __ cmpb(Address(r13, 0), Bytecodes::_invokedynamic);
    __ jcc(Assembler::equal, L_giant_index);
  }
185
  __ get_cache_and_index_at_bcp(rbx, rcx, 1, sizeof(u2));
186
  __ bind(L_got_cache);
D
duke 已提交
187
  __ movl(rbx, Address(rbx, rcx,
188
                       Address::times_ptr,
189
                       in_bytes(ConstantPoolCache::base_offset()) +
D
duke 已提交
190 191
                       3 * wordSize));
  __ andl(rbx, 0xFF);
192
  __ lea(rsp, Address(rsp, rbx, Address::times_8));
D
duke 已提交
193
  __ dispatch_next(state, step);
194 195 196 197

  // out of the main line of code...
  if (EnableInvokeDynamic) {
    __ bind(L_giant_index);
198
    __ get_cache_and_index_at_bcp(rbx, rcx, 1, sizeof(u4));
199 200 201
    __ jmp(L_got_cache);
  }

D
duke 已提交
202 203 204 205 206 207 208 209
  return entry;
}


address TemplateInterpreterGenerator::generate_deopt_entry_for(TosState state,
                                                               int step) {
  address entry = __ pc();
  // NULL last_sp until next java call
210
  __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
211 212 213 214 215
  __ restore_bcp();
  __ restore_locals();
  // handle exceptions
  {
    Label L;
216
    __ cmpptr(Address(r15_thread, Thread::pending_exception_offset()), (int32_t) NULL_WORD);
D
duke 已提交
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
    __ jcc(Assembler::zero, L);
    __ call_VM(noreg,
               CAST_FROM_FN_PTR(address,
                                InterpreterRuntime::throw_pending_exception));
    __ should_not_reach_here();
    __ bind(L);
  }
  __ dispatch_next(state, step);
  return entry;
}

int AbstractInterpreter::BasicType_as_index(BasicType type) {
  int i = 0;
  switch (type) {
    case T_BOOLEAN: i = 0; break;
    case T_CHAR   : i = 1; break;
    case T_BYTE   : i = 2; break;
    case T_SHORT  : i = 3; break;
    case T_INT    : i = 4; break;
    case T_LONG   : i = 5; break;
    case T_VOID   : i = 6; break;
    case T_FLOAT  : i = 7; break;
    case T_DOUBLE : i = 8; break;
    case T_OBJECT : i = 9; break;
    case T_ARRAY  : i = 9; break;
    default       : ShouldNotReachHere();
  }
  assert(0 <= i && i < AbstractInterpreter::number_of_result_handlers,
         "index out of bounds");
  return i;
}


address TemplateInterpreterGenerator::generate_result_handler_for(
        BasicType type) {
  address entry = __ pc();
  switch (type) {
  case T_BOOLEAN: __ c2bool(rax);            break;
  case T_CHAR   : __ movzwl(rax, rax);       break;
  case T_BYTE   : __ sign_extend_byte(rax);  break;
  case T_SHORT  : __ sign_extend_short(rax); break;
  case T_INT    : /* nothing to do */        break;
  case T_LONG   : /* nothing to do */        break;
  case T_VOID   : /* nothing to do */        break;
  case T_FLOAT  : /* nothing to do */        break;
  case T_DOUBLE : /* nothing to do */        break;
  case T_OBJECT :
    // retrieve result from frame
265
    __ movptr(rax, Address(rbp, frame::interpreter_frame_oop_temp_offset*wordSize));
D
duke 已提交
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
    // and verify it
    __ verify_oop(rax);
    break;
  default       : ShouldNotReachHere();
  }
  __ ret(0);                                   // return from result handler
  return entry;
}

address TemplateInterpreterGenerator::generate_safept_entry_for(
        TosState state,
        address runtime_entry) {
  address entry = __ pc();
  __ push(state);
  __ call_VM(noreg, runtime_entry);
  __ dispatch_via(vtos, Interpreter::_normal_table.table_for(vtos));
  return entry;
}



// Helpers for commoning out cases in the various type of method entries.
//


// increment invocation count & check for overflow
//
// Note: checking for negative value instead of overflow
//       so we have a 'sticky' overflow test
//
// rbx: method
// ecx: invocation counter
//
void InterpreterGenerator::generate_counter_incr(
        Label* overflow,
        Label* profile_method,
        Label* profile_method_continue) {
303
  Label done;
304
  // Note: In tiered we increment either counters in Method* or in MDO depending if we're profiling or not.
I
iveresov 已提交
305 306 307
  if (TieredCompilation) {
    int increment = InvocationCounter::count_increment;
    int mask = ((1 << Tier0InvokeNotifyFreqLog)  - 1) << InvocationCounter::count_shift;
308
    Label no_mdo;
I
iveresov 已提交
309 310
    if (ProfileInterpreter) {
      // Are we profiling?
311
      __ movptr(rax, Address(rbx, Method::method_data_offset()));
I
iveresov 已提交
312 313 314
      __ testptr(rax, rax);
      __ jccb(Assembler::zero, no_mdo);
      // Increment counter in the MDO
315
      const Address mdo_invocation_counter(rax, in_bytes(MethodData::invocation_counter_offset()) +
I
iveresov 已提交
316 317
                                                in_bytes(InvocationCounter::counter_offset()));
      __ increment_mask_and_jump(mdo_invocation_counter, increment, mask, rcx, false, Assembler::zero, overflow);
318
      __ jmp(done);
I
iveresov 已提交
319 320
    }
    __ bind(no_mdo);
321 322 323 324 325 326 327
    // Increment counter in MethodCounters
    const Address invocation_counter(rax,
                  MethodCounters::invocation_counter_offset() +
                  InvocationCounter::counter_offset());
    __ get_method_counters(rbx, rax, done);
    __ increment_mask_and_jump(invocation_counter, increment, mask, rcx,
                               false, Assembler::zero, overflow);
I
iveresov 已提交
328 329
    __ bind(done);
  } else {
330 331 332 333 334 335
    const Address backedge_counter(rax,
                  MethodCounters::backedge_counter_offset() +
                  InvocationCounter::counter_offset());
    const Address invocation_counter(rax,
                  MethodCounters::invocation_counter_offset() +
                  InvocationCounter::counter_offset());
D
duke 已提交
336

337 338 339 340 341
    __ get_method_counters(rbx, rax, done);

    if (ProfileInterpreter) {
      __ incrementl(Address(rax,
              MethodCounters::interpreter_invocation_counter_offset()));
I
iveresov 已提交
342 343
    }
    // Update standard invocation counters
344
    __ movl(rcx, invocation_counter);
I
iveresov 已提交
345
    __ incrementl(rcx, InvocationCounter::count_increment);
346 347 348
    __ movl(invocation_counter, rcx); // save invocation count

    __ movl(rax, backedge_counter);   // load backedge counter
I
iveresov 已提交
349
    __ andl(rax, InvocationCounter::count_mask_value); // mask out the status bits
D
duke 已提交
350

I
iveresov 已提交
351
    __ addl(rcx, rax);                // add both counters
D
duke 已提交
352

I
iveresov 已提交
353 354
    // profile_method is non-null only for interpreted method so
    // profile_method != NULL == !native_call
D
duke 已提交
355

I
iveresov 已提交
356 357 358 359
    if (ProfileInterpreter && profile_method != NULL) {
      // Test to see if we should create a method data oop
      __ cmp32(rcx, ExternalAddress((address)&InvocationCounter::InterpreterProfileLimit));
      __ jcc(Assembler::less, *profile_method_continue);
D
duke 已提交
360

I
iveresov 已提交
361 362 363
      // if no method data exists, go to profile_method
      __ test_method_data_pointer(rax, *profile_method);
    }
D
duke 已提交
364

I
iveresov 已提交
365 366
    __ cmp32(rcx, ExternalAddress((address)&InvocationCounter::InterpreterInvocationLimit));
    __ jcc(Assembler::aboveEqual, *overflow);
367
    __ bind(done);
I
iveresov 已提交
368
  }
D
duke 已提交
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
}

void InterpreterGenerator::generate_counter_overflow(Label* do_continue) {

  // Asm interpreter on entry
  // r14 - locals
  // r13 - bcp
  // rbx - method
  // edx - cpool --- DOES NOT APPEAR TO BE TRUE
  // rbp - interpreter frame

  // On return (i.e. jump to entry_point) [ back to invocation of interpreter ]
  // Everything as it was on entry
  // rdx is not restored. Doesn't appear to really be set.

  // InterpreterRuntime::frequency_counter_overflow takes two
  // arguments, the first (thread) is passed by call_VM, the second
  // indicates if the counter overflow occurs at a backwards branch
  // (NULL bcp).  We pass zero for it.  The call returns the address
  // of the verified entry point for the method or NULL if the
  // compilation did not complete (either went background or bailed
  // out).
  __ movl(c_rarg1, 0);
  __ call_VM(noreg,
             CAST_FROM_FN_PTR(address,
                              InterpreterRuntime::frequency_counter_overflow),
             c_rarg1);

397
  __ movptr(rbx, Address(rbp, method_offset));   // restore Method*
D
duke 已提交
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
  // Preserve invariant that r13/r14 contain bcp/locals of sender frame
  // and jump to the interpreted entry.
  __ jmp(*do_continue, relocInfo::none);
}

// See if we've got enough room on the stack for locals plus overhead.
// The expression stack grows down incrementally, so the normal guard
// page mechanism will work for that.
//
// NOTE: Since the additional locals are also always pushed (wasn't
// obvious in generate_method_entry) so the guard should work for them
// too.
//
// Args:
//      rdx: number of additional locals this frame needs (what we must check)
413
//      rbx: Method*
D
duke 已提交
414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
//
// Kills:
//      rax
void InterpreterGenerator::generate_stack_overflow_check(void) {

  // monitor entry size: see picture of stack set
  // (generate_method_entry) and frame_amd64.hpp
  const int entry_size = frame::interpreter_frame_monitor_size() * wordSize;

  // total overhead size: entry_size + (saved rbp through expr stack
  // bottom).  be sure to change this if you add/subtract anything
  // to/from the overhead area
  const int overhead_size =
    -(frame::interpreter_frame_initial_sp_offset * wordSize) + entry_size;

  const int page_size = os::vm_page_size();

  Label after_frame_check;

  // see if the frame is greater than one page in size. If so,
  // then we need to verify there is enough stack space remaining
  // for the additional locals.
436
  __ cmpl(rdx, (page_size - overhead_size) / Interpreter::stackElementSize);
D
duke 已提交
437 438 439 440 441 442 443 444 445
  __ jcc(Assembler::belowEqual, after_frame_check);

  // compute rsp as if this were going to be the last frame on
  // the stack before the red zone

  const Address stack_base(r15_thread, Thread::stack_base_offset());
  const Address stack_size(r15_thread, Thread::stack_size_offset());

  // locals + overhead, in bytes
446
  __ mov(rax, rdx);
447
  __ shlptr(rax, Interpreter::logStackElementSize);  // 2 slots per parameter.
448
  __ addptr(rax, overhead_size);
D
duke 已提交
449 450 451 452

#ifdef ASSERT
  Label stack_base_okay, stack_size_okay;
  // verify that thread stack base is non-zero
453
  __ cmpptr(stack_base, (int32_t)NULL_WORD);
D
duke 已提交
454 455 456 457
  __ jcc(Assembler::notEqual, stack_base_okay);
  __ stop("stack base is zero");
  __ bind(stack_base_okay);
  // verify that thread stack size is non-zero
458
  __ cmpptr(stack_size, 0);
D
duke 已提交
459 460 461 462 463 464
  __ jcc(Assembler::notEqual, stack_size_okay);
  __ stop("stack size is zero");
  __ bind(stack_size_okay);
#endif

  // Add stack base to locals and subtract stack size
465 466
  __ addptr(rax, stack_base);
  __ subptr(rax, stack_size);
D
duke 已提交
467

468 469 470 471
  // Use the maximum number of pages we might bang.
  const int max_pages = StackShadowPages > (StackRedPages+StackYellowPages) ? StackShadowPages :
                                                                              (StackRedPages+StackYellowPages);

D
duke 已提交
472
  // add in the red and yellow zone sizes
473
  __ addptr(rax, max_pages * page_size);
D
duke 已提交
474 475

  // check against the current stack bottom
476
  __ cmpptr(rsp, rax);
D
duke 已提交
477 478
  __ jcc(Assembler::above, after_frame_check);

B
bdelsart 已提交
479 480 481 482 483 484 485 486 487 488 489 490
  // Restore sender's sp as SP. This is necessary if the sender's
  // frame is an extended compiled frame (see gen_c2i_adapter())
  // and safer anyway in case of JSR292 adaptations.

  __ pop(rax); // return address must be moved if SP is changed
  __ mov(rsp, r13);
  __ push(rax);

  // Note: the restored frame is not necessarily interpreted.
  // Use the shared runtime version of the StackOverflowError.
  assert(StubRoutines::throw_StackOverflowError_entry() != NULL, "stub not yet generated");
  __ jump(ExternalAddress(StubRoutines::throw_StackOverflowError_entry()));
D
duke 已提交
491 492 493 494 495 496 497 498

  // all done with frame size check
  __ bind(after_frame_check);
}

// Allocate monitor and lock method (asm interpreter)
//
// Args:
499
//      rbx: Method*
D
duke 已提交
500 501 502 503 504 505 506 507
//      r14: locals
//
// Kills:
//      rax
//      c_rarg0, c_rarg1, c_rarg2, c_rarg3, ...(param regs)
//      rscratch1, rscratch2 (scratch regs)
void InterpreterGenerator::lock_method(void) {
  // synchronize method
508
  const Address access_flags(rbx, Method::access_flags_offset());
D
duke 已提交
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
  const Address monitor_block_top(
        rbp,
        frame::interpreter_frame_monitor_block_top_offset * wordSize);
  const int entry_size = frame::interpreter_frame_monitor_size() * wordSize;

#ifdef ASSERT
  {
    Label L;
    __ movl(rax, access_flags);
    __ testl(rax, JVM_ACC_SYNCHRONIZED);
    __ jcc(Assembler::notZero, L);
    __ stop("method doesn't need synchronization");
    __ bind(L);
  }
#endif // ASSERT

  // get synchronization object
  {
527
    const int mirror_offset = in_bytes(Klass::java_mirror_offset());
D
duke 已提交
528 529 530 531
    Label done;
    __ movl(rax, access_flags);
    __ testl(rax, JVM_ACC_STATIC);
    // get receiver (assume this is frequent case)
532
    __ movptr(rax, Address(r14, Interpreter::local_offset_in_bytes(0)));
D
duke 已提交
533
    __ jcc(Assembler::zero, done);
534 535
    __ movptr(rax, Address(rbx, Method::const_offset()));
    __ movptr(rax, Address(rax, ConstMethod::constants_offset()));
536
    __ movptr(rax, Address(rax,
537
                           ConstantPool::pool_holder_offset_in_bytes()));
538
    __ movptr(rax, Address(rax, mirror_offset));
D
duke 已提交
539 540 541 542

#ifdef ASSERT
    {
      Label L;
543
      __ testptr(rax, rax);
D
duke 已提交
544 545 546 547 548 549 550 551 552 553
      __ jcc(Assembler::notZero, L);
      __ stop("synchronization object is NULL");
      __ bind(L);
    }
#endif // ASSERT

    __ bind(done);
  }

  // add space for monitor & lock
554 555
  __ subptr(rsp, entry_size); // add space for a monitor entry
  __ movptr(monitor_block_top, rsp);  // set new monitor block top
D
duke 已提交
556
  // store object
557 558
  __ movptr(Address(rsp, BasicObjectLock::obj_offset_in_bytes()), rax);
  __ movptr(c_rarg1, rsp); // object address
D
duke 已提交
559 560 561 562 563 564 565 566
  __ lock_object(c_rarg1);
}

// Generate a fixed interpreter frame. This is identical setup for
// interpreted methods and for native methods hence the shared code.
//
// Args:
//      rax: return address
567
//      rbx: Method*
D
duke 已提交
568 569 570 571 572
//      r14: pointer to locals
//      r13: sender sp
//      rdx: cp cache
void TemplateInterpreterGenerator::generate_fixed_frame(bool native_call) {
  // initialize fixed part of activation frame
573
  __ push(rax);        // save return address
D
duke 已提交
574
  __ enter();          // save old & set new rbp
575 576
  __ push(r13);        // set sender sp
  __ push((int)NULL_WORD); // leave last_sp as null
577 578 579
  __ movptr(r13, Address(rbx, Method::const_offset()));      // get ConstMethod*
  __ lea(r13, Address(r13, ConstMethod::codes_offset())); // get codebase
  __ push(rbx);        // save Method*
D
duke 已提交
580 581
  if (ProfileInterpreter) {
    Label method_data_continue;
582
    __ movptr(rdx, Address(rbx, in_bytes(Method::method_data_offset())));
583
    __ testptr(rdx, rdx);
D
duke 已提交
584
    __ jcc(Assembler::zero, method_data_continue);
585
    __ addptr(rdx, in_bytes(MethodData::data_offset()));
D
duke 已提交
586
    __ bind(method_data_continue);
587
    __ push(rdx);      // set the mdp (method data pointer)
D
duke 已提交
588
  } else {
589
    __ push(0);
D
duke 已提交
590 591
  }

592 593 594
  __ movptr(rdx, Address(rbx, Method::const_offset()));
  __ movptr(rdx, Address(rdx, ConstMethod::constants_offset()));
  __ movptr(rdx, Address(rdx, ConstantPool::cache_offset_in_bytes()));
595 596
  __ push(rdx); // set constant pool cache
  __ push(r14); // set locals pointer
D
duke 已提交
597
  if (native_call) {
598
    __ push(0); // no bcp
D
duke 已提交
599
  } else {
600
    __ push(r13); // set bcp
D
duke 已提交
601
  }
602 603
  __ push(0); // reserve word for pointer to expression stack bottom
  __ movptr(Address(rsp, 0), rsp); // set expression stack bottom
D
duke 已提交
604 605 606 607
}

// End of helpers

608 609 610 611 612 613 614 615
// Various method entries
//------------------------------------------------------------------------------------------------------------------------
//
//

// Call an accessor method (assuming it is resolved, otherwise drop
// into vanilla (slow path) entry
address InterpreterGenerator::generate_accessor_entry(void) {
616
  // rbx: Method*
617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643

  // r13: senderSP must preserver for slow path, set SP to it on fast path

  address entry_point = __ pc();
  Label xreturn_path;

  // do fastpath for resolved accessor methods
  if (UseFastAccessorMethods) {
    // Code: _aload_0, _(i|a)getfield, _(i|a)return or any rewrites
    //       thereof; parameter size = 1
    // Note: We can only use this code if the getfield has been resolved
    //       and if we don't have a null-pointer exception => check for
    //       these conditions first and use slow path if necessary.
    Label slow_path;
    // If we need a safepoint check, generate full interpreter entry.
    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
             SafepointSynchronize::_not_synchronized);

    __ jcc(Assembler::notEqual, slow_path);
    // rbx: method
    __ movptr(rax, Address(rsp, wordSize));

    // check if local 0 != NULL and read field
    __ testptr(rax, rax);
    __ jcc(Assembler::zero, slow_path);

    // read first instruction word and extract bytecode @ 1 and index @ 2
644 645 646
    __ movptr(rdx, Address(rbx, Method::const_offset()));
    __ movptr(rdi, Address(rdx, ConstMethod::constants_offset()));
    __ movl(rdx, Address(rdx, ConstMethod::codes_offset()));
647 648 649 650
    // Shift codes right to get the index on the right.
    // The bytecode fetched looks like <index><0xb4><0x2a>
    __ shrl(rdx, 2 * BitsPerByte);
    __ shll(rdx, exact_log2(in_words(ConstantPoolCacheEntry::size())));
651
    __ movptr(rdi, Address(rdi, ConstantPool::cache_offset_in_bytes()));
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666

    // rax: local 0
    // rbx: method
    // rdx: constant pool cache index
    // rdi: constant pool cache

    // check if getfield has been resolved and read constant pool cache entry
    // check the validity of the cache entry by testing whether _indices field
    // contains Bytecode::_getfield in b1 byte.
    assert(in_words(ConstantPoolCacheEntry::size()) == 4,
           "adjust shift below");
    __ movl(rcx,
            Address(rdi,
                    rdx,
                    Address::times_8,
667
                    ConstantPoolCache::base_offset() +
668 669 670 671 672 673 674 675 676 677 678
                    ConstantPoolCacheEntry::indices_offset()));
    __ shrl(rcx, 2 * BitsPerByte);
    __ andl(rcx, 0xFF);
    __ cmpl(rcx, Bytecodes::_getfield);
    __ jcc(Assembler::notEqual, slow_path);

    // Note: constant pool entry is not valid before bytecode is resolved
    __ movptr(rcx,
              Address(rdi,
                      rdx,
                      Address::times_8,
679
                      ConstantPoolCache::base_offset() +
680 681 682 683 684 685
                      ConstantPoolCacheEntry::f2_offset()));
    // edx: flags
    __ movl(rdx,
            Address(rdi,
                    rdx,
                    Address::times_8,
686
                    ConstantPoolCache::base_offset() +
687 688 689 690 691 692 693 694
                    ConstantPoolCacheEntry::flags_offset()));

    Label notObj, notInt, notByte, notShort;
    const Address field_address(rax, rcx, Address::times_1);

    // Need to differentiate between igetfield, agetfield, bgetfield etc.
    // because they are different sizes.
    // Use the type from the constant pool cache
695 696 697
    __ shrl(rdx, ConstantPoolCacheEntry::tos_state_shift);
    // Make sure we don't need to mask edx after the above shift
    ConstantPoolCacheEntry::verify_tos_state_shift();
698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722

    __ cmpl(rdx, atos);
    __ jcc(Assembler::notEqual, notObj);
    // atos
    __ load_heap_oop(rax, field_address);
    __ jmp(xreturn_path);

    __ bind(notObj);
    __ cmpl(rdx, itos);
    __ jcc(Assembler::notEqual, notInt);
    // itos
    __ movl(rax, field_address);
    __ jmp(xreturn_path);

    __ bind(notInt);
    __ cmpl(rdx, btos);
    __ jcc(Assembler::notEqual, notByte);
    // btos
    __ load_signed_byte(rax, field_address);
    __ jmp(xreturn_path);

    __ bind(notByte);
    __ cmpl(rdx, stos);
    __ jcc(Assembler::notEqual, notShort);
    // stos
723
    __ load_signed_short(rax, field_address);
724 725 726 727 728 729 730 731 732 733 734
    __ jmp(xreturn_path);

    __ bind(notShort);
#ifdef ASSERT
    Label okay;
    __ cmpl(rdx, ctos);
    __ jcc(Assembler::equal, okay);
    __ stop("what type is this?");
    __ bind(okay);
#endif
    // ctos
735
    __ load_unsigned_short(rax, field_address);
736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754

    __ bind(xreturn_path);

    // _ireturn/_areturn
    __ pop(rdi);
    __ mov(rsp, r13);
    __ jmp(rdi);
    __ ret(0);

    // generate a vanilla interpreter entry as the slow path
    __ bind(slow_path);
    (void) generate_normal_entry(false);
  } else {
    (void) generate_normal_entry(false);
  }

  return entry_point;
}

755 756
// Method entry for java.lang.ref.Reference.get.
address InterpreterGenerator::generate_Reference_get_entry(void) {
757
#if INCLUDE_ALL_GCS
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
  // Code: _aload_0, _getfield, _areturn
  // parameter size = 1
  //
  // The code that gets generated by this routine is split into 2 parts:
  //    1. The "intrinsified" code for G1 (or any SATB based GC),
  //    2. The slow path - which is an expansion of the regular method entry.
  //
  // Notes:-
  // * In the G1 code we do not check whether we need to block for
  //   a safepoint. If G1 is enabled then we must execute the specialized
  //   code for Reference.get (except when the Reference object is null)
  //   so that we can log the value in the referent field with an SATB
  //   update buffer.
  //   If the code for the getfield template is modified so that the
  //   G1 pre-barrier code is executed when the current method is
  //   Reference.get() then going through the normal method entry
  //   will be fine.
  // * The G1 code can, however, check the receiver object (the instance
  //   of java.lang.Reference) and jump to the slow path if null. If the
  //   Reference object is null then we obviously cannot fetch the referent
  //   and so we don't need to call the G1 pre-barrier. Thus we can use the
  //   regular method entry code to generate the NPE.
  //
  // This code is based on generate_accessor_enty.
  //
783
  // rbx: Method*
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835

  // r13: senderSP must preserve for slow path, set SP to it on fast path

  address entry = __ pc();

  const int referent_offset = java_lang_ref_Reference::referent_offset;
  guarantee(referent_offset > 0, "referent offset not initialized");

  if (UseG1GC) {
    Label slow_path;
    // rbx: method

    // Check if local 0 != NULL
    // If the receiver is null then it is OK to jump to the slow path.
    __ movptr(rax, Address(rsp, wordSize));

    __ testptr(rax, rax);
    __ jcc(Assembler::zero, slow_path);

    // rax: local 0
    // rbx: method (but can be used as scratch now)
    // rdx: scratch
    // rdi: scratch

    // Generate the G1 pre-barrier code to log the value of
    // the referent field in an SATB buffer.

    // Load the value of the referent field.
    const Address field_address(rax, referent_offset);
    __ load_heap_oop(rax, field_address);

    // Generate the G1 pre-barrier code to log the value of
    // the referent field in an SATB buffer.
    __ g1_write_barrier_pre(noreg /* obj */,
                            rax /* pre_val */,
                            r15_thread /* thread */,
                            rbx /* tmp */,
                            true /* tosca_live */,
                            true /* expand_call */);

    // _areturn
    __ pop(rdi);                // get return address
    __ mov(rsp, r13);           // set sp to sender sp
    __ jmp(rdi);
    __ ret(0);

    // generate a vanilla interpreter entry as the slow path
    __ bind(slow_path);
    (void) generate_normal_entry(false);

    return entry;
  }
836
#endif // INCLUDE_ALL_GCS
837 838 839 840 841 842 843

  // If G1 is not enabled then attempt to go through the accessor entry point
  // Reference.get is an accessor
  return generate_accessor_entry();
}


D
duke 已提交
844 845 846 847 848 849 850
// Interpreter stub for calling a native method. (asm interpreter)
// This sets up a somewhat different looking stack for calling the
// native method than the typical interpreter frame setup.
address InterpreterGenerator::generate_native_entry(bool synchronized) {
  // determine code generation flags
  bool inc_counter  = UseCompiler || CountCompiledCalls;

851
  // rbx: Method*
D
duke 已提交
852 853 854 855
  // r13: sender sp

  address entry_point = __ pc();

856
  const Address constMethod       (rbx, Method::const_offset());
857
  const Address access_flags      (rbx, Method::access_flags_offset());
858 859 860
  const Address size_of_parameters(rcx, ConstMethod::
                                        size_of_parameters_offset());

D
duke 已提交
861 862

  // get parameter size (always needed)
863
  __ movptr(rcx, constMethod);
864
  __ load_unsigned_short(rcx, size_of_parameters);
D
duke 已提交
865 866 867 868 869

  // native calls don't need the stack size check since they have no
  // expression stack and the arguments are already on the stack and
  // we only add a handful of words to the stack

870
  // rbx: Method*
D
duke 已提交
871 872
  // rcx: size of parameters
  // r13: sender sp
873
  __ pop(rax);                                       // get return address
D
duke 已提交
874 875 876 877

  // for natives the size of locals is zero

  // compute beginning of parameters (r14)
878
  __ lea(r14, Address(rsp, rcx, Address::times_8, -wordSize));
D
duke 已提交
879 880 881

  // add 2 zero-initialized slots for native calls
  // initialize result_handler slot
882
  __ push((int) NULL_WORD);
D
duke 已提交
883 884
  // slot for oop temp
  // (static native method holder mirror/jni oop result)
885
  __ push((int) NULL_WORD);
D
duke 已提交
886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957

  // initialize fixed part of activation frame
  generate_fixed_frame(true);

  // make sure method is native & not abstract
#ifdef ASSERT
  __ movl(rax, access_flags);
  {
    Label L;
    __ testl(rax, JVM_ACC_NATIVE);
    __ jcc(Assembler::notZero, L);
    __ stop("tried to execute non-native method as native");
    __ bind(L);
  }
  {
    Label L;
    __ testl(rax, JVM_ACC_ABSTRACT);
    __ jcc(Assembler::zero, L);
    __ stop("tried to execute abstract method in interpreter");
    __ bind(L);
  }
#endif

  // Since at this point in the method invocation the exception handler
  // would try to exit the monitor of synchronized methods which hasn't
  // been entered yet, we set the thread local variable
  // _do_not_unlock_if_synchronized to true. The remove_activation will
  // check this flag.

  const Address do_not_unlock_if_synchronized(r15_thread,
        in_bytes(JavaThread::do_not_unlock_if_synchronized_offset()));
  __ movbool(do_not_unlock_if_synchronized, true);

  // increment invocation count & check for overflow
  Label invocation_counter_overflow;
  if (inc_counter) {
    generate_counter_incr(&invocation_counter_overflow, NULL, NULL);
  }

  Label continue_after_compile;
  __ bind(continue_after_compile);

  bang_stack_shadow_pages(true);

  // reset the _do_not_unlock_if_synchronized flag
  __ movbool(do_not_unlock_if_synchronized, false);

  // check for synchronized methods
  // Must happen AFTER invocation_counter check and stack overflow check,
  // so method is not locked if overflows.
  if (synchronized) {
    lock_method();
  } else {
    // no synchronization necessary
#ifdef ASSERT
    {
      Label L;
      __ movl(rax, access_flags);
      __ testl(rax, JVM_ACC_SYNCHRONIZED);
      __ jcc(Assembler::zero, L);
      __ stop("method needs synchronization");
      __ bind(L);
    }
#endif
  }

  // start execution
#ifdef ASSERT
  {
    Label L;
    const Address monitor_block_top(rbp,
                 frame::interpreter_frame_monitor_block_top_offset * wordSize);
958 959
    __ movptr(rax, monitor_block_top);
    __ cmpptr(rax, rsp);
D
duke 已提交
960 961 962 963 964 965 966 967 968 969 970
    __ jcc(Assembler::equal, L);
    __ stop("broken stack frame setup in interpreter");
    __ bind(L);
  }
#endif

  // jvmti support
  __ notify_method_entry();

  // work registers
  const Register method = rbx;
971
  const Register t      = r11;
D
duke 已提交
972 973 974

  // allocate space for parameters
  __ get_method(method);
975 976
  __ movptr(t, Address(method, Method::const_offset()));
  __ load_unsigned_short(t, Address(t, ConstMethod::size_of_parameters_offset()));
977
  __ shll(t, Interpreter::logStackElementSize);
D
duke 已提交
978

979 980
  __ subptr(rsp, t);
  __ subptr(rsp, frame::arg_reg_save_area_bytes); // windows
T
twisti 已提交
981
  __ andptr(rsp, -16); // must be 16 byte boundary (see amd64 ABI)
D
duke 已提交
982 983 984 985

  // get signature handler
  {
    Label L;
986
    __ movptr(t, Address(method, Method::signature_handler_offset()));
987
    __ testptr(t, t);
D
duke 已提交
988 989 990 991 992 993
    __ jcc(Assembler::notZero, L);
    __ call_VM(noreg,
               CAST_FROM_FN_PTR(address,
                                InterpreterRuntime::prepare_native_call),
               method);
    __ get_method(method);
994
    __ movptr(t, Address(method, Method::signature_handler_offset()));
D
duke 已提交
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
    __ bind(L);
  }

  // call signature handler
  assert(InterpreterRuntime::SignatureHandlerGenerator::from() == r14,
         "adjust this code");
  assert(InterpreterRuntime::SignatureHandlerGenerator::to() == rsp,
         "adjust this code");
  assert(InterpreterRuntime::SignatureHandlerGenerator::temp() == rscratch1,
          "adjust this code");

  // The generated handlers do not touch RBX (the method oop).
  // However, large signatures cannot be cached and are generated
  // each time here.  The slow-path generator can do a GC on return,
  // so we must reload it after the call.
  __ call(t);
  __ get_method(method);        // slow path can do a GC, reload RBX


  // result handler is in rax
  // set result handler
1016 1017 1018
  __ movptr(Address(rbp,
                    (frame::interpreter_frame_result_handler_offset) * wordSize),
            rax);
D
duke 已提交
1019 1020 1021 1022

  // pass mirror handle if static call
  {
    Label L;
1023
    const int mirror_offset = in_bytes(Klass::java_mirror_offset());
1024
    __ movl(t, Address(method, Method::access_flags_offset()));
D
duke 已提交
1025 1026 1027
    __ testl(t, JVM_ACC_STATIC);
    __ jcc(Assembler::zero, L);
    // get mirror
1028 1029 1030
    __ movptr(t, Address(method, Method::const_offset()));
    __ movptr(t, Address(t, ConstMethod::constants_offset()));
    __ movptr(t, Address(t, ConstantPool::pool_holder_offset_in_bytes()));
1031
    __ movptr(t, Address(t, mirror_offset));
D
duke 已提交
1032
    // copy mirror into activation frame
1033
    __ movptr(Address(rbp, frame::interpreter_frame_oop_temp_offset * wordSize),
D
duke 已提交
1034 1035
            t);
    // pass handle to mirror
1036 1037
    __ lea(c_rarg1,
           Address(rbp, frame::interpreter_frame_oop_temp_offset * wordSize));
D
duke 已提交
1038 1039 1040 1041 1042 1043
    __ bind(L);
  }

  // get native function entry point
  {
    Label L;
1044
    __ movptr(rax, Address(method, Method::native_function_offset()));
D
duke 已提交
1045 1046
    ExternalAddress unsatisfied(SharedRuntime::native_method_throw_unsatisfied_link_error_entry());
    __ movptr(rscratch2, unsatisfied.addr());
1047
    __ cmpptr(rax, rscratch2);
D
duke 已提交
1048 1049 1050 1051 1052 1053
    __ jcc(Assembler::notEqual, L);
    __ call_VM(noreg,
               CAST_FROM_FN_PTR(address,
                                InterpreterRuntime::prepare_native_call),
               method);
    __ get_method(method);
1054
    __ movptr(rax, Address(method, Method::native_function_offset()));
D
duke 已提交
1055 1056 1057 1058
    __ bind(L);
  }

  // pass JNIEnv
1059
  __ lea(c_rarg0, Address(r15_thread, JavaThread::jni_environment_offset()));
D
duke 已提交
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

  // It is enough that the pc() points into the right code
  // segment. It does not have to be the correct return pc.
  __ set_last_Java_frame(rsp, rbp, (address) __ pc());

  // change thread state
#ifdef ASSERT
  {
    Label L;
    __ movl(t, Address(r15_thread, JavaThread::thread_state_offset()));
    __ cmpl(t, _thread_in_Java);
    __ jcc(Assembler::equal, L);
    __ stop("Wrong thread state in native stub");
    __ bind(L);
  }
#endif

  // Change state to native

  __ movl(Address(r15_thread, JavaThread::thread_state_offset()),
          _thread_in_native);

  // Call the native method.
  __ call(rax);
  // result potentially in rax or xmm0

1086 1087
  // Verify or restore cpu control state after JNI call
  __ restore_cpu_control_state_after_jni();
D
duke 已提交
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134

  // NOTE: The order of these pushes is known to frame::interpreter_frame_result
  // in order to extract the result of a method call. If the order of these
  // pushes change or anything else is added to the stack then the code in
  // interpreter_frame_result must also change.

  __ push(dtos);
  __ push(ltos);

  // change thread state
  __ movl(Address(r15_thread, JavaThread::thread_state_offset()),
          _thread_in_native_trans);

  if (os::is_MP()) {
    if (UseMembar) {
      // Force this write out before the read below
      __ membar(Assembler::Membar_mask_bits(
           Assembler::LoadLoad | Assembler::LoadStore |
           Assembler::StoreLoad | Assembler::StoreStore));
    } else {
      // Write serialization page so VM thread can do a pseudo remote membar.
      // We use the current thread pointer to calculate a thread specific
      // offset to write to within the page. This minimizes bus traffic
      // due to cache line collision.
      __ serialize_memory(r15_thread, rscratch2);
    }
  }

  // check for safepoint operation in progress and/or pending suspend requests
  {
    Label Continue;
    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
             SafepointSynchronize::_not_synchronized);

    Label L;
    __ jcc(Assembler::notEqual, L);
    __ cmpl(Address(r15_thread, JavaThread::suspend_flags_offset()), 0);
    __ jcc(Assembler::equal, Continue);
    __ bind(L);

    // Don't use call_VM as it will see a possible pending exception
    // and forward it and never return here preventing us from
    // clearing _last_native_pc down below.  Also can't use
    // call_VM_leaf either as it will check to see if r13 & r14 are
    // preserved and correspond to the bcp/locals pointers. So we do a
    // runtime call by hand.
    //
1135
    __ mov(c_rarg0, r15_thread);
1136
    __ mov(r12, rsp); // remember sp (can only use r12 if not using call_VM)
1137 1138
    __ subptr(rsp, frame::arg_reg_save_area_bytes); // windows
    __ andptr(rsp, -16); // align stack as required by ABI
D
duke 已提交
1139
    __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, JavaThread::check_special_condition_for_native_trans)));
1140
    __ mov(rsp, r12); // restore sp
1141
    __ reinit_heapbase();
D
duke 已提交
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151
    __ bind(Continue);
  }

  // change thread state
  __ movl(Address(r15_thread, JavaThread::thread_state_offset()), _thread_in_Java);

  // reset_last_Java_frame
  __ reset_last_Java_frame(true, true);

  // reset handle block
1152 1153
  __ movptr(t, Address(r15_thread, JavaThread::active_handles_offset()));
  __ movptr(Address(t, JNIHandleBlock::top_offset_in_bytes()), (int32_t)NULL_WORD);
D
duke 已提交
1154 1155 1156 1157 1158 1159 1160

  // If result is an oop unbox and store it in frame where gc will see it
  // and result handler will pick it up

  {
    Label no_oop, store_result;
    __ lea(t, ExternalAddress(AbstractInterpreter::result_handler(T_OBJECT)));
1161
    __ cmpptr(t, Address(rbp, frame::interpreter_frame_result_handler_offset*wordSize));
D
duke 已提交
1162 1163 1164
    __ jcc(Assembler::notEqual, no_oop);
    // retrieve result
    __ pop(ltos);
1165
    __ testptr(rax, rax);
D
duke 已提交
1166
    __ jcc(Assembler::zero, store_result);
1167
    __ movptr(rax, Address(rax, 0));
D
duke 已提交
1168
    __ bind(store_result);
1169
    __ movptr(Address(rbp, frame::interpreter_frame_oop_temp_offset*wordSize), rax);
D
duke 已提交
1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
    // keep stack depth as expected by pushing oop which will eventually be discarde
    __ push(ltos);
    __ bind(no_oop);
  }


  {
    Label no_reguard;
    __ cmpl(Address(r15_thread, JavaThread::stack_guard_state_offset()),
            JavaThread::stack_guard_yellow_disabled);
    __ jcc(Assembler::notEqual, no_reguard);

1182
    __ pusha(); // XXX only save smashed registers
1183
    __ mov(r12, rsp); // remember sp (can only use r12 if not using call_VM)
1184 1185
    __ subptr(rsp, frame::arg_reg_save_area_bytes); // windows
    __ andptr(rsp, -16); // align stack as required by ABI
D
duke 已提交
1186
    __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, SharedRuntime::reguard_yellow_pages)));
1187 1188
    __ mov(rsp, r12); // restore sp
    __ popa(); // XXX only restore smashed registers
1189
    __ reinit_heapbase();
D
duke 已提交
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201

    __ bind(no_reguard);
  }


  // The method register is junk from after the thread_in_native transition
  // until here.  Also can't call_VM until the bcp has been
  // restored.  Need bcp for throwing exception below so get it now.
  __ get_method(method);

  // restore r13 to have legal interpreter frame, i.e., bci == 0 <=>
  // r13 == code_base()
1202 1203
  __ movptr(r13, Address(method, Method::const_offset()));   // get ConstMethod*
  __ lea(r13, Address(r13, ConstMethod::codes_offset()));    // get codebase
D
duke 已提交
1204 1205 1206
  // handle exceptions (exception handling will handle unlocking!)
  {
    Label L;
1207
    __ cmpptr(Address(r15_thread, Thread::pending_exception_offset()), (int32_t) NULL_WORD);
D
duke 已提交
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222
    __ jcc(Assembler::zero, L);
    // Note: At some point we may want to unify this with the code
    // used in call_VM_base(); i.e., we should use the
    // StubRoutines::forward_exception code. For now this doesn't work
    // here because the rsp is not correctly set at this point.
    __ MacroAssembler::call_VM(noreg,
                               CAST_FROM_FN_PTR(address,
                               InterpreterRuntime::throw_pending_exception));
    __ should_not_reach_here();
    __ bind(L);
  }

  // do unlocking if necessary
  {
    Label L;
1223
    __ movl(t, Address(method, Method::access_flags_offset()));
D
duke 已提交
1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
    __ testl(t, JVM_ACC_SYNCHRONIZED);
    __ jcc(Assembler::zero, L);
    // the code below should be shared with interpreter macro
    // assembler implementation
    {
      Label unlock;
      // 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,
                            (intptr_t)(frame::interpreter_frame_initial_sp_offset *
                                       wordSize - sizeof(BasicObjectLock)));

      // monitor expect in c_rarg1 for slow unlock path
1238
      __ lea(c_rarg1, monitor); // address of first monitor
D
duke 已提交
1239

1240 1241
      __ movptr(t, Address(c_rarg1, BasicObjectLock::obj_offset_in_bytes()));
      __ testptr(t, t);
D
duke 已提交
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268
      __ jcc(Assembler::notZero, unlock);

      // Entry already unlocked, need to throw exception
      __ MacroAssembler::call_VM(noreg,
                                 CAST_FROM_FN_PTR(address,
                   InterpreterRuntime::throw_illegal_monitor_state_exception));
      __ should_not_reach_here();

      __ bind(unlock);
      __ unlock_object(c_rarg1);
    }
    __ bind(L);
  }

  // jvmti support
  // Note: This must happen _after_ handling/throwing any exceptions since
  //       the exception handler code notifies the runtime of method exits
  //       too. If this happens before, method entry/exit notifications are
  //       not properly paired (was bug - gri 11/22/99).
  __ notify_method_exit(vtos, InterpreterMacroAssembler::NotifyJVMTI);

  // restore potential result in edx:eax, call result handler to
  // restore potential result in ST0 & handle result

  __ pop(ltos);
  __ pop(dtos);

1269 1270
  __ movptr(t, Address(rbp,
                       (frame::interpreter_frame_result_handler_offset) * wordSize));
D
duke 已提交
1271 1272 1273
  __ call(t);

  // remove activation
1274 1275 1276
  __ movptr(t, Address(rbp,
                       frame::interpreter_frame_sender_sp_offset *
                       wordSize)); // get sender sp
D
duke 已提交
1277
  __ leave();                                // remove frame anchor
1278 1279
  __ pop(rdi);                               // get return address
  __ mov(rsp, t);                            // set sp to sender sp
D
duke 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
  __ jmp(rdi);

  if (inc_counter) {
    // Handle overflow of counter and compile method
    __ bind(invocation_counter_overflow);
    generate_counter_overflow(&continue_after_compile);
  }

  return entry_point;
}

//
// Generic interpreted method entry to (asm) interpreter
//
address InterpreterGenerator::generate_normal_entry(bool synchronized) {
  // determine code generation flags
  bool inc_counter  = UseCompiler || CountCompiledCalls;

1298
  // ebx: Method*
D
duke 已提交
1299 1300 1301
  // r13: sender sp
  address entry_point = __ pc();

1302
  const Address constMethod(rbx, Method::const_offset());
1303
  const Address access_flags(rbx, Method::access_flags_offset());
1304 1305 1306 1307
  const Address size_of_parameters(rdx,
                                   ConstMethod::size_of_parameters_offset());
  const Address size_of_locals(rdx, ConstMethod::size_of_locals_offset());

D
duke 已提交
1308 1309

  // get parameter size (always needed)
1310
  __ movptr(rdx, constMethod);
1311
  __ load_unsigned_short(rcx, size_of_parameters);
D
duke 已提交
1312

1313
  // rbx: Method*
D
duke 已提交
1314 1315 1316
  // rcx: size of parameters
  // r13: sender_sp (could differ from sp+wordSize if we were called via c2i )

1317
  __ load_unsigned_short(rdx, size_of_locals); // get size of locals in words
D
duke 已提交
1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
  __ subl(rdx, rcx); // rdx = no. of additional locals

  // YYY
//   __ incrementl(rdx);
//   __ andl(rdx, -2);

  // see if we've got enough room on the stack for locals plus overhead.
  generate_stack_overflow_check();

  // get return address
1328
  __ pop(rax);
D
duke 已提交
1329 1330

  // compute beginning of parameters (r14)
1331
  __ lea(r14, Address(rsp, rcx, Address::times_8, -wordSize));
D
duke 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340

  // rdx - # of additional locals
  // allocate space for locals
  // explicitly initialize locals
  {
    Label exit, loop;
    __ testl(rdx, rdx);
    __ jcc(Assembler::lessEqual, exit); // do nothing if rdx <= 0
    __ bind(loop);
1341
    __ push((int) NULL_WORD); // initialize local variables
D
duke 已提交
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 1389 1390 1391 1392 1393 1394 1395 1396 1397 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
    __ decrementl(rdx); // until everything initialized
    __ jcc(Assembler::greater, loop);
    __ bind(exit);
  }

  // initialize fixed part of activation frame
  generate_fixed_frame(false);

  // make sure method is not native & not abstract
#ifdef ASSERT
  __ movl(rax, access_flags);
  {
    Label L;
    __ testl(rax, JVM_ACC_NATIVE);
    __ jcc(Assembler::zero, L);
    __ stop("tried to execute native method as non-native");
    __ bind(L);
  }
  {
    Label L;
    __ testl(rax, JVM_ACC_ABSTRACT);
    __ jcc(Assembler::zero, L);
    __ stop("tried to execute abstract method in interpreter");
    __ bind(L);
  }
#endif

  // Since at this point in the method invocation the exception
  // handler would try to exit the monitor of synchronized methods
  // which hasn't been entered yet, we set the thread local variable
  // _do_not_unlock_if_synchronized to true. The remove_activation
  // will check this flag.

  const Address do_not_unlock_if_synchronized(r15_thread,
        in_bytes(JavaThread::do_not_unlock_if_synchronized_offset()));
  __ movbool(do_not_unlock_if_synchronized, true);

  // increment invocation count & check for overflow
  Label invocation_counter_overflow;
  Label profile_method;
  Label profile_method_continue;
  if (inc_counter) {
    generate_counter_incr(&invocation_counter_overflow,
                          &profile_method,
                          &profile_method_continue);
    if (ProfileInterpreter) {
      __ bind(profile_method_continue);
    }
  }

  Label continue_after_compile;
  __ bind(continue_after_compile);

  // check for synchronized interpreted methods
  bang_stack_shadow_pages(false);

  // reset the _do_not_unlock_if_synchronized flag
  __ movbool(do_not_unlock_if_synchronized, false);

  // check for synchronized methods
  // Must happen AFTER invocation_counter check and stack overflow check,
  // so method is not locked if overflows.
  if (synchronized) {
    // Allocate monitor and lock method
    lock_method();
  } else {
    // no synchronization necessary
#ifdef ASSERT
    {
      Label L;
      __ movl(rax, access_flags);
      __ testl(rax, JVM_ACC_SYNCHRONIZED);
      __ jcc(Assembler::zero, L);
      __ stop("method needs synchronization");
      __ bind(L);
    }
#endif
  }

  // start execution
#ifdef ASSERT
  {
    Label L;
     const Address monitor_block_top (rbp,
                 frame::interpreter_frame_monitor_block_top_offset * wordSize);
1427 1428
    __ movptr(rax, monitor_block_top);
    __ cmpptr(rax, rsp);
D
duke 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
    __ jcc(Assembler::equal, L);
    __ stop("broken stack frame setup in interpreter");
    __ bind(L);
  }
#endif

  // jvmti support
  __ notify_method_entry();

  __ dispatch_next(vtos);

  // invocation counter overflow
  if (inc_counter) {
    if (ProfileInterpreter) {
      // We have decided to profile this method in the interpreter
      __ bind(profile_method);
1445 1446
      __ call_VM(noreg, CAST_FROM_FN_PTR(address, InterpreterRuntime::profile_method));
      __ set_method_data_pointer_for_bcp();
1447
      __ get_method(rbx);
D
duke 已提交
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
      __ jmp(profile_method_continue);
    }
    // Handle overflow of counter and compile method
    __ bind(invocation_counter_overflow);
    generate_counter_overflow(&continue_after_compile);
  }

  return entry_point;
}

// Entry points
//
// Here we generate the various kind of entries into the interpreter.
// The two main entry type are generic bytecode methods and native
// call method.  These both come in synchronized and non-synchronized
// versions but the frame layout they create is very similar. The
// other method entry types are really just special purpose entries
// that are really entry and interpretation all in one. These are for
// trivial methods like accessor, empty, or special math methods.
//
// When control flow reaches any of the entry types for the interpreter
// the following holds ->
//
// Arguments:
//
1473
// rbx: Method*
D
duke 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
//
// Stack layout immediately at entry
//
// [ return address     ] <--- rsp
// [ parameter n        ]
//   ...
// [ parameter 1        ]
// [ expression stack   ] (caller's java expression stack)

// Assuming that we don't go to one of the trivial specialized entries
// the stack will look like below when we are ready to execute the
// first bytecode (or call the native routine). The register usage
// will be as the template based interpreter expects (see
// interpreter_amd64.hpp).
//
// local variables follow incoming parameters immediately; i.e.
// the return address is moved to the end of the locals).
//
// [ monitor entry      ] <--- rsp
//   ...
// [ monitor entry      ]
// [ expr. stack bottom ]
// [ saved r13          ]
// [ current r14        ]
1498
// [ Method*            ]
D
duke 已提交
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
// [ saved ebp          ] <--- rbp
// [ return address     ]
// [ local variable m   ]
//   ...
// [ local variable 1   ]
// [ parameter n        ]
//   ...
// [ parameter 1        ] <--- r14

address AbstractInterpreterGenerator::generate_method_entry(
                                        AbstractInterpreter::MethodKind kind) {
  // determine code generation flags
  bool synchronized = false;
  address entry_point = NULL;

  switch (kind) {
  case Interpreter::zerolocals             :                                                                             break;
  case Interpreter::zerolocals_synchronized: synchronized = true;                                                        break;
1517 1518 1519 1520 1521
  case Interpreter::native                 : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(false); break;
  case Interpreter::native_synchronized    : entry_point = ((InterpreterGenerator*)this)->generate_native_entry(true);  break;
  case Interpreter::empty                  : entry_point = ((InterpreterGenerator*)this)->generate_empty_entry();       break;
  case Interpreter::accessor               : entry_point = ((InterpreterGenerator*)this)->generate_accessor_entry();    break;
  case Interpreter::abstract               : entry_point = ((InterpreterGenerator*)this)->generate_abstract_entry();    break;
1522 1523 1524 1525 1526 1527 1528

  case Interpreter::java_lang_math_sin     : // fall thru
  case Interpreter::java_lang_math_cos     : // fall thru
  case Interpreter::java_lang_math_tan     : // fall thru
  case Interpreter::java_lang_math_abs     : // fall thru
  case Interpreter::java_lang_math_log     : // fall thru
  case Interpreter::java_lang_math_log10   : // fall thru
1529 1530
  case Interpreter::java_lang_math_sqrt    : // fall thru
  case Interpreter::java_lang_math_pow     : // fall thru
1531
  case Interpreter::java_lang_math_exp     : entry_point = ((InterpreterGenerator*)this)->generate_math_entry(kind);    break;
1532 1533
  case Interpreter::java_lang_ref_reference_get
                                           : entry_point = ((InterpreterGenerator*)this)->generate_Reference_get_entry(); break;
1534 1535 1536
  default:
    fatal(err_msg("unexpected method kind: %d", kind));
    break;
D
duke 已提交
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546
  }

  if (entry_point) {
    return entry_point;
  }

  return ((InterpreterGenerator*) this)->
                                generate_normal_entry(synchronized);
}

1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
// These should never be compiled since the interpreter will prefer
// the compiled version to the intrinsic version.
bool AbstractInterpreter::can_be_compiled(methodHandle m) {
  switch (method_kind(m)) {
    case Interpreter::java_lang_math_sin     : // fall thru
    case Interpreter::java_lang_math_cos     : // fall thru
    case Interpreter::java_lang_math_tan     : // fall thru
    case Interpreter::java_lang_math_abs     : // fall thru
    case Interpreter::java_lang_math_log     : // fall thru
    case Interpreter::java_lang_math_log10   : // fall thru
1557 1558 1559
    case Interpreter::java_lang_math_sqrt    : // fall thru
    case Interpreter::java_lang_math_pow     : // fall thru
    case Interpreter::java_lang_math_exp     :
1560 1561 1562 1563 1564 1565
      return false;
    default:
      return true;
  }
}

D
duke 已提交
1566
// How much stack a method activation needs in words.
1567
int AbstractInterpreter::size_top_interpreter_activation(Method* method) {
D
duke 已提交
1568 1569 1570 1571 1572 1573 1574 1575 1576
  const int entry_size = frame::interpreter_frame_monitor_size();

  // total overhead size: entry_size + (saved rbp thru expr stack
  // bottom).  be sure to change this if you add/subtract anything
  // to/from the overhead area
  const int overhead_size =
    -(frame::interpreter_frame_initial_sp_offset) + entry_size;

  const int stub_code = frame::entry_frame_after_call_words;
1577
  const int extra_stack = Method::extra_stack_entries();
1578
  const int method_stack = (method->max_locals() + method->max_stack() + extra_stack) *
1579
                           Interpreter::stackElementWords;
D
duke 已提交
1580 1581 1582
  return (overhead_size + method_stack + stub_code);
}

1583
int AbstractInterpreter::layout_activation(Method* method,
D
duke 已提交
1584 1585 1586
                                           int tempcount,
                                           int popframe_extra_args,
                                           int moncount,
1587
                                           int caller_actual_parameters,
D
duke 已提交
1588 1589 1590 1591
                                           int callee_param_count,
                                           int callee_locals,
                                           frame* caller,
                                           frame* interpreter_frame,
1592 1593
                                           bool is_top_frame,
                                           bool is_bottom_frame) {
D
duke 已提交
1594 1595 1596 1597 1598 1599 1600 1601
  // Note: This calculation must exactly parallel the frame setup
  // in AbstractInterpreterGenerator::generate_method_entry.
  // If interpreter_frame!=NULL, set up the method, locals, and monitors.
  // The frame interpreter_frame, if not NULL, is guaranteed to be the
  // right size, as determined by a previous call to this method.
  // It is also guaranteed to be walkable even though it is in a skeletal state

  // fixed size of an interpreter frame:
1602
  int max_locals = method->max_locals() * Interpreter::stackElementWords;
D
duke 已提交
1603
  int extra_locals = (method->max_locals() - method->size_of_parameters()) *
1604
                     Interpreter::stackElementWords;
D
duke 已提交
1605 1606 1607 1608 1609 1610 1611 1612

  int overhead = frame::sender_sp_offset -
                 frame::interpreter_frame_initial_sp_offset;
  // Our locals were accounted for by the caller (or last_frame_adjust
  // on the transistion) Since the callee parameters already account
  // for the callee's params we only need to account for the extra
  // locals.
  int size = overhead +
1613
         (callee_locals - callee_param_count)*Interpreter::stackElementWords +
D
duke 已提交
1614
         moncount * frame::interpreter_frame_monitor_size() +
1615
         tempcount* Interpreter::stackElementWords + popframe_extra_args;
D
duke 已提交
1616 1617
  if (interpreter_frame != NULL) {
#ifdef ASSERT
1618
    if (!EnableInvokeDynamic)
1619 1620 1621
      // @@@ FIXME: Should we correct interpreter_frame_sender_sp in the calling sequences?
      // Probably, since deoptimization doesn't work yet.
      assert(caller->unextended_sp() == interpreter_frame->interpreter_frame_sender_sp(), "Frame not properly walkable");
D
duke 已提交
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
    assert(caller->sp() == interpreter_frame->sender_sp(), "Frame not properly walkable(2)");
#endif

    interpreter_frame->interpreter_frame_set_method(method);
    // NOTE the difference in using sender_sp and
    // interpreter_frame_sender_sp interpreter_frame_sender_sp is
    // the original sp of the caller (the unextended_sp) and
    // sender_sp is fp+16 XXX
    intptr_t* locals = interpreter_frame->sender_sp() + max_locals - 1;

1632 1633 1634 1635 1636 1637
#ifdef ASSERT
    if (caller->is_interpreted_frame()) {
      assert(locals < caller->fp() + frame::interpreter_frame_initial_sp_offset, "bad placement");
    }
#endif

D
duke 已提交
1638 1639 1640 1641 1642 1643 1644
    interpreter_frame->interpreter_frame_set_locals(locals);
    BasicObjectLock* montop = interpreter_frame->interpreter_frame_monitor_begin();
    BasicObjectLock* monbot = montop - moncount;
    interpreter_frame->interpreter_frame_set_monitor_end(monbot);

    // Set last_sp
    intptr_t*  esp = (intptr_t*) monbot -
1645
                     tempcount*Interpreter::stackElementWords -
D
duke 已提交
1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672
                     popframe_extra_args;
    interpreter_frame->interpreter_frame_set_last_sp(esp);

    // All frames but the initial (oldest) interpreter frame we fill in have
    // a value for sender_sp that allows walking the stack but isn't
    // truly correct. Correct the value here.
    if (extra_locals != 0 &&
        interpreter_frame->sender_sp() ==
        interpreter_frame->interpreter_frame_sender_sp()) {
      interpreter_frame->set_interpreter_frame_sender_sp(caller->sp() +
                                                         extra_locals);
    }
    *interpreter_frame->interpreter_frame_cache_addr() =
      method->constants()->cache();
  }
  return size;
}

//-----------------------------------------------------------------------------
// Exceptions

void TemplateInterpreterGenerator::generate_throw_exception() {
  // Entry point in previous activation (i.e., if the caller was
  // interpreted)
  Interpreter::_rethrow_exception_entry = __ pc();
  // Restore sp to interpreter_frame_last_sp even though we are going
  // to empty the expression stack for the exception processing.
1673
  __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
1674 1675 1676 1677
  // rax: exception
  // rdx: return address/pc that threw exception
  __ restore_bcp();    // r13 points to call/send
  __ restore_locals();
1678
  __ reinit_heapbase();  // restore r12 as heapbase.
D
duke 已提交
1679 1680 1681 1682 1683 1684
  // Entry point for exceptions thrown within interpreter code
  Interpreter::_throw_exception_entry = __ pc();
  // expression stack is undefined here
  // rax: exception
  // r13: exception bcp
  __ verify_oop(rax);
1685
  __ mov(c_rarg1, rax);
D
duke 已提交
1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739

  // expression stack must be empty before entering the VM in case of
  // an exception
  __ empty_expression_stack();
  // find exception handler address and preserve exception oop
  __ call_VM(rdx,
             CAST_FROM_FN_PTR(address,
                          InterpreterRuntime::exception_handler_for_exception),
             c_rarg1);
  // rax: exception handler entry point
  // rdx: preserved exception oop
  // r13: bcp for exception handler
  __ push_ptr(rdx); // push exception which is now the only value on the stack
  __ jmp(rax); // jump to exception handler (may be _remove_activation_entry!)

  // If the exception is not handled in the current frame the frame is
  // removed and the exception is rethrown (i.e. exception
  // continuation is _rethrow_exception).
  //
  // Note: At this point the bci is still the bxi for the instruction
  // which caused the exception and the expression stack is
  // empty. Thus, for any VM calls at this point, GC will find a legal
  // oop map (with empty expression stack).

  // In current activation
  // tos: exception
  // esi: exception bcp

  //
  // JVMTI PopFrame support
  //

  Interpreter::_remove_activation_preserving_args_entry = __ pc();
  __ empty_expression_stack();
  // Set the popframe_processing bit in pending_popframe_condition
  // indicating that we are currently handling popframe, so that
  // call_VMs that may happen later do not trigger new popframe
  // handling cycles.
  __ movl(rdx, Address(r15_thread, JavaThread::popframe_condition_offset()));
  __ orl(rdx, JavaThread::popframe_processing_bit);
  __ movl(Address(r15_thread, JavaThread::popframe_condition_offset()), rdx);

  {
    // Check to see whether we are returning to a deoptimized frame.
    // (The PopFrame call ensures that the caller of the popped frame is
    // either interpreted or compiled and deoptimizes it if compiled.)
    // In this case, we can't call dispatch_next() after the frame is
    // popped, but instead must save the incoming arguments and restore
    // them after deoptimization has occurred.
    //
    // Note that we don't compare the return PC against the
    // deoptimization blob's unpack entry because of the presence of
    // adapter frames in C2.
    Label caller_not_deoptimized;
1740
    __ movptr(c_rarg1, Address(rbp, frame::return_addr_offset * wordSize));
D
duke 已提交
1741 1742 1743 1744 1745 1746 1747 1748
    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address,
                               InterpreterRuntime::interpreter_contains), c_rarg1);
    __ testl(rax, rax);
    __ jcc(Assembler::notZero, caller_not_deoptimized);

    // Compute size of arguments for saving when returning to
    // deoptimized caller
    __ get_method(rax);
1749 1750
    __ movptr(rax, Address(rax, Method::const_offset()));
    __ load_unsigned_short(rax, Address(rax, in_bytes(ConstMethod::
D
duke 已提交
1751
                                                size_of_parameters_offset())));
1752
    __ shll(rax, Interpreter::logStackElementSize);
D
duke 已提交
1753
    __ restore_locals(); // XXX do we need this?
1754 1755
    __ subptr(r14, rax);
    __ addptr(r14, wordSize);
D
duke 已提交
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
    // Save these arguments
    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address,
                                           Deoptimization::
                                           popframe_preserve_args),
                          r15_thread, rax, r14);

    __ remove_activation(vtos, rdx,
                         /* throw_monitor_exception */ false,
                         /* install_monitor_exception */ false,
                         /* notify_jvmdi */ false);

    // Inform deoptimization that it is responsible for restoring
    // these arguments
    __ movl(Address(r15_thread, JavaThread::popframe_condition_offset()),
            JavaThread::popframe_force_deopt_reexecution_bit);

    // Continue in deoptimization handler
    __ jmp(rdx);

    __ bind(caller_not_deoptimized);
  }

  __ remove_activation(vtos, rdx, /* rdx result (retaddr) is not used */
                       /* throw_monitor_exception */ false,
                       /* install_monitor_exception */ false,
                       /* notify_jvmdi */ false);

  // Finish with popframe handling
  // A previous I2C followed by a deoptimization might have moved the
  // outgoing arguments further up the stack. PopFrame expects the
  // mutations to those outgoing arguments to be preserved and other
  // constraints basically require this frame to look exactly as
  // though it had previously invoked an interpreted activation with
  // no space between the top of the expression stack (current
  // last_sp) and the top of stack. Rather than force deopt to
  // maintain this kind of invariant all the time we call a small
  // fixup routine to move the mutated arguments onto the top of our
  // expression stack if necessary.
1794 1795
  __ mov(c_rarg1, rsp);
  __ movptr(c_rarg2, Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize));
D
duke 已提交
1796 1797 1798 1799 1800
  // PC must point into interpreter here
  __ set_last_Java_frame(noreg, rbp, __ pc());
  __ super_call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::popframe_move_outgoing_args), r15_thread, c_rarg1, c_rarg2);
  __ reset_last_Java_frame(true, true);
  // Restore the last_sp and null it out
1801 1802
  __ movptr(rsp, Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize));
  __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
D
duke 已提交
1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822

  __ restore_bcp();  // XXX do we need this?
  __ restore_locals(); // XXX do we need this?
  // The method data pointer was incremented already during
  // call profiling. We have to restore the mdp for the current bcp.
  if (ProfileInterpreter) {
    __ set_method_data_pointer_for_bcp();
  }

  // Clear the popframe condition flag
  __ movl(Address(r15_thread, JavaThread::popframe_condition_offset()),
          JavaThread::popframe_inactive);

  __ dispatch_next(vtos);
  // end of PopFrame support

  Interpreter::_remove_activation_entry = __ pc();

  // preserve exception over this code sequence
  __ pop_ptr(rax);
1823
  __ movptr(Address(r15_thread, JavaThread::vm_result_offset()), rax);
D
duke 已提交
1824 1825 1826
  // remove the activation (without doing throws on illegalMonitorExceptions)
  __ remove_activation(vtos, rdx, false, true, false);
  // restore exception
1827
  __ get_vm_result(rax, r15_thread);
D
duke 已提交
1828 1829 1830 1831 1832 1833 1834 1835 1836

  // In between activations - previous activation type unknown yet
  // compute continuation point - the continuation point expects the
  // following registers set up:
  //
  // rax: exception
  // rdx: return address/pc that threw exception
  // rsp: expression stack of caller
  // rbp: ebp of caller
1837 1838
  __ push(rax);                                  // save exception
  __ push(rdx);                                  // save return address
D
duke 已提交
1839 1840
  __ super_call_VM_leaf(CAST_FROM_FN_PTR(address,
                          SharedRuntime::exception_handler_for_return_address),
1841
                        r15_thread, rdx);
1842 1843 1844
  __ mov(rbx, rax);                              // save exception handler
  __ pop(rdx);                                   // restore return address
  __ pop(rax);                                   // restore exception
D
duke 已提交
1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861
  // Note that an "issuing PC" is actually the next PC after the call
  __ jmp(rbx);                                   // jump to exception
                                                 // handler of caller
}


//
// JVMTI ForceEarlyReturn support
//
address TemplateInterpreterGenerator::generate_earlyret_entry_for(TosState state) {
  address entry = __ pc();

  __ restore_bcp();
  __ restore_locals();
  __ empty_expression_stack();
  __ load_earlyret_value(state);

1862
  __ movptr(rdx, Address(r15_thread, JavaThread::jvmti_thread_state_offset()));
D
duke 已提交
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
  Address cond_addr(rdx, JvmtiThreadState::earlyret_state_offset());

  // Clear the earlyret state
  __ movl(cond_addr, JvmtiThreadState::earlyret_inactive);

  __ remove_activation(state, rsi,
                       false, /* throw_monitor_exception */
                       false, /* install_monitor_exception */
                       true); /* notify_jvmdi */
  __ jmp(rsi);

  return entry;
} // end of ForceEarlyReturn support


//-----------------------------------------------------------------------------
// Helper for vtos entry point generation

void TemplateInterpreterGenerator::set_vtos_entry_points(Template* t,
                                                         address& bep,
                                                         address& cep,
                                                         address& sep,
                                                         address& aep,
                                                         address& iep,
                                                         address& lep,
                                                         address& fep,
                                                         address& dep,
                                                         address& vep) {
  assert(t->is_valid() && t->tos_in() == vtos, "illegal template");
  Label L;
  aep = __ pc();  __ push_ptr();  __ jmp(L);
  fep = __ pc();  __ push_f();    __ jmp(L);
  dep = __ pc();  __ push_d();    __ jmp(L);
  lep = __ pc();  __ push_l();    __ jmp(L);
  bep = cep = sep =
  iep = __ pc();  __ push_i();
  vep = __ pc();
  __ bind(L);
  generate_and_dispatch(t);
}


//-----------------------------------------------------------------------------
// Generation of individual instructions

// helpers for generate_and_dispatch


InterpreterGenerator::InterpreterGenerator(StubQueue* code)
  : TemplateInterpreterGenerator(code) {
   generate_all(); // down here so it can be "virtual"
}

//-----------------------------------------------------------------------------

// Non-product code
#ifndef PRODUCT
address TemplateInterpreterGenerator::generate_trace_code(TosState state) {
  address entry = __ pc();

  __ push(state);
1924 1925 1926 1927 1928
  __ push(c_rarg0);
  __ push(c_rarg1);
  __ push(c_rarg2);
  __ push(c_rarg3);
  __ mov(c_rarg2, rax);  // Pass itos
D
duke 已提交
1929 1930 1931 1932 1933 1934
#ifdef _WIN64
  __ movflt(xmm3, xmm0); // Pass ftos
#endif
  __ call_VM(noreg,
             CAST_FROM_FN_PTR(address, SharedRuntime::trace_bytecode),
             c_rarg1, c_rarg2, c_rarg3);
1935 1936 1937 1938
  __ pop(c_rarg3);
  __ pop(c_rarg2);
  __ pop(c_rarg1);
  __ pop(c_rarg0);
D
duke 已提交
1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971
  __ pop(state);
  __ ret(0);                                   // return from result handler

  return entry;
}

void TemplateInterpreterGenerator::count_bytecode() {
  __ incrementl(ExternalAddress((address) &BytecodeCounter::_counter_value));
}

void TemplateInterpreterGenerator::histogram_bytecode(Template* t) {
  __ incrementl(ExternalAddress((address) &BytecodeHistogram::_counters[t->bytecode()]));
}

void TemplateInterpreterGenerator::histogram_bytecode_pair(Template* t) {
  __ mov32(rbx, ExternalAddress((address) &BytecodePairHistogram::_index));
  __ shrl(rbx, BytecodePairHistogram::log2_number_of_codes);
  __ orl(rbx,
         ((int) t->bytecode()) <<
         BytecodePairHistogram::log2_number_of_codes);
  __ mov32(ExternalAddress((address) &BytecodePairHistogram::_index), rbx);
  __ lea(rscratch1, ExternalAddress((address) BytecodePairHistogram::_counters));
  __ incrementl(Address(rscratch1, rbx, Address::times_4));
}


void TemplateInterpreterGenerator::trace_bytecode(Template* t) {
  // Call a little run-time stub to avoid blow-up for each bytecode.
  // The run-time runtime saves the right registers, depending on
  // the tosca in-state for the given template.

  assert(Interpreter::trace_code(t->tos_in()) != NULL,
         "entry must have been generated");
1972
  __ mov(r12, rsp); // remember sp (can only use r12 if not using call_VM)
1973
  __ andptr(rsp, -16); // align stack as required by ABI
D
duke 已提交
1974
  __ call(RuntimeAddress(Interpreter::trace_code(t->tos_in())));
1975
  __ mov(rsp, r12); // restore sp
1976
  __ reinit_heapbase();
D
duke 已提交
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
}


void TemplateInterpreterGenerator::stop_interpreter_at() {
  Label L;
  __ cmp32(ExternalAddress((address) &BytecodeCounter::_counter_value),
           StopInterpreterAt);
  __ jcc(Assembler::notEqual, L);
  __ int3();
  __ bind(L);
}
#endif // !PRODUCT
1989
#endif // ! CC_INTERP