macroAssembler_sparc.inline.hpp 26.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765
/*
 * Copyright (c) 1997, 2012, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 *
 */

#ifndef CPU_SPARC_VM_MACROASSEMBLER_SPARC_INLINE_HPP
#define CPU_SPARC_VM_MACROASSEMBLER_SPARC_INLINE_HPP

#include "asm/assembler.inline.hpp"
#include "asm/macroAssembler.hpp"
#include "asm/codeBuffer.hpp"
#include "code/codeCache.hpp"

inline bool Address::is_simm13(int offset) { return Assembler::is_simm13(disp() + offset); }


inline int AddressLiteral::low10() const {
  return Assembler::low10(value());
}


inline void MacroAssembler::pd_patch_instruction(address branch, address target) {
  jint& stub_inst = *(jint*) branch;
  stub_inst = patched_branch(target - branch, stub_inst, 0);
}

#ifndef PRODUCT
inline void MacroAssembler::pd_print_patched_instruction(address branch) {
  jint stub_inst = *(jint*) branch;
  print_instruction(stub_inst);
  ::tty->print("%s", " (unresolved)");
}
#endif // PRODUCT

// Use the right loads/stores for the platform
inline void MacroAssembler::ld_ptr( Register s1, Register s2, Register d ) {
#ifdef _LP64
  Assembler::ldx(s1, s2, d);
#else
             ld( s1, s2, d);
#endif
}

inline void MacroAssembler::ld_ptr( Register s1, int simm13a, Register d ) {
#ifdef _LP64
  Assembler::ldx(s1, simm13a, d);
#else
             ld( s1, simm13a, d);
#endif
}

#ifdef ASSERT
// ByteSize is only a class when ASSERT is defined, otherwise it's an int.
inline void MacroAssembler::ld_ptr( Register s1, ByteSize simm13a, Register d ) {
  ld_ptr(s1, in_bytes(simm13a), d);
}
#endif

inline void MacroAssembler::ld_ptr( Register s1, RegisterOrConstant s2, Register d ) {
#ifdef _LP64
  ldx(s1, s2, d);
#else
  ld( s1, s2, d);
#endif
}

inline void MacroAssembler::ld_ptr(const Address& a, Register d, int offset) {
#ifdef _LP64
  ldx(a, d, offset);
#else
  ld( a, d, offset);
#endif
}

inline void MacroAssembler::st_ptr( Register d, Register s1, Register s2 ) {
#ifdef _LP64
  Assembler::stx(d, s1, s2);
#else
             st( d, s1, s2);
#endif
}

inline void MacroAssembler::st_ptr( Register d, Register s1, int simm13a ) {
#ifdef _LP64
  Assembler::stx(d, s1, simm13a);
#else
             st( d, s1, simm13a);
#endif
}

#ifdef ASSERT
// ByteSize is only a class when ASSERT is defined, otherwise it's an int.
inline void MacroAssembler::st_ptr( Register d, Register s1, ByteSize simm13a ) {
  st_ptr(d, s1, in_bytes(simm13a));
}
#endif

inline void MacroAssembler::st_ptr( Register d, Register s1, RegisterOrConstant s2 ) {
#ifdef _LP64
  stx(d, s1, s2);
#else
  st( d, s1, s2);
#endif
}

inline void MacroAssembler::st_ptr(Register d, const Address& a, int offset) {
#ifdef _LP64
  stx(d, a, offset);
#else
  st( d, a, offset);
#endif
}

// Use the right loads/stores for the platform
inline void MacroAssembler::ld_long( Register s1, Register s2, Register d ) {
#ifdef _LP64
  Assembler::ldx(s1, s2, d);
#else
  Assembler::ldd(s1, s2, d);
#endif
}

inline void MacroAssembler::ld_long( Register s1, int simm13a, Register d ) {
#ifdef _LP64
  Assembler::ldx(s1, simm13a, d);
#else
  Assembler::ldd(s1, simm13a, d);
#endif
}

inline void MacroAssembler::ld_long( Register s1, RegisterOrConstant s2, Register d ) {
#ifdef _LP64
  ldx(s1, s2, d);
#else
  ldd(s1, s2, d);
#endif
}

inline void MacroAssembler::ld_long(const Address& a, Register d, int offset) {
#ifdef _LP64
  ldx(a, d, offset);
#else
  ldd(a, d, offset);
#endif
}

inline void MacroAssembler::st_long( Register d, Register s1, Register s2 ) {
#ifdef _LP64
  Assembler::stx(d, s1, s2);
#else
  Assembler::std(d, s1, s2);
#endif
}

inline void MacroAssembler::st_long( Register d, Register s1, int simm13a ) {
#ifdef _LP64
  Assembler::stx(d, s1, simm13a);
#else
  Assembler::std(d, s1, simm13a);
#endif
}

inline void MacroAssembler::st_long( Register d, Register s1, RegisterOrConstant s2 ) {
#ifdef _LP64
  stx(d, s1, s2);
#else
  std(d, s1, s2);
#endif
}

inline void MacroAssembler::st_long( Register d, const Address& a, int offset ) {
#ifdef _LP64
  stx(d, a, offset);
#else
  std(d, a, offset);
#endif
}

// Functions for isolating 64 bit shifts for LP64

inline void MacroAssembler::sll_ptr( Register s1, Register s2, Register d ) {
#ifdef _LP64
  Assembler::sllx(s1, s2, d);
#else
  Assembler::sll( s1, s2, d);
#endif
}

inline void MacroAssembler::sll_ptr( Register s1, int imm6a,   Register d ) {
#ifdef _LP64
  Assembler::sllx(s1, imm6a, d);
#else
  Assembler::sll( s1, imm6a, d);
#endif
}

inline void MacroAssembler::srl_ptr( Register s1, Register s2, Register d ) {
#ifdef _LP64
  Assembler::srlx(s1, s2, d);
#else
  Assembler::srl( s1, s2, d);
#endif
}

inline void MacroAssembler::srl_ptr( Register s1, int imm6a,   Register d ) {
#ifdef _LP64
  Assembler::srlx(s1, imm6a, d);
#else
  Assembler::srl( s1, imm6a, d);
#endif
}

inline void MacroAssembler::sll_ptr( Register s1, RegisterOrConstant s2, Register d ) {
  if (s2.is_register())  sll_ptr(s1, s2.as_register(), d);
  else                   sll_ptr(s1, s2.as_constant(), d);
}

// Use the right branch for the platform

inline void MacroAssembler::br( Condition c, bool a, Predict p, address d, relocInfo::relocType rt ) {
  if (VM_Version::v9_instructions_work())
    Assembler::bp(c, a, icc, p, d, rt);
  else
    Assembler::br(c, a, d, rt);
}

inline void MacroAssembler::br( Condition c, bool a, Predict p, Label& L ) {
  br(c, a, p, target(L));
}


// Branch that tests either xcc or icc depending on the
// architecture compiled (LP64 or not)
inline void MacroAssembler::brx( Condition c, bool a, Predict p, address d, relocInfo::relocType rt ) {
#ifdef _LP64
    Assembler::bp(c, a, xcc, p, d, rt);
#else
    MacroAssembler::br(c, a, p, d, rt);
#endif
}

inline void MacroAssembler::brx( Condition c, bool a, Predict p, Label& L ) {
  brx(c, a, p, target(L));
}

inline void MacroAssembler::ba( Label& L ) {
  br(always, false, pt, L);
}

// Warning: V9 only functions
inline void MacroAssembler::bp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt ) {
  Assembler::bp(c, a, cc, p, d, rt);
}

inline void MacroAssembler::bp( Condition c, bool a, CC cc, Predict p, Label& L ) {
  Assembler::bp(c, a, cc, p, L);
}

inline void MacroAssembler::fb( Condition c, bool a, Predict p, address d, relocInfo::relocType rt ) {
  if (VM_Version::v9_instructions_work())
    fbp(c, a, fcc0, p, d, rt);
  else
    Assembler::fb(c, a, d, rt);
}

inline void MacroAssembler::fb( Condition c, bool a, Predict p, Label& L ) {
  fb(c, a, p, target(L));
}

inline void MacroAssembler::fbp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt ) {
  Assembler::fbp(c, a, cc, p, d, rt);
}

inline void MacroAssembler::fbp( Condition c, bool a, CC cc, Predict p, Label& L ) {
  Assembler::fbp(c, a, cc, p, L);
}

inline void MacroAssembler::jmp( Register s1, Register s2 ) { jmpl( s1, s2, G0 ); }
inline void MacroAssembler::jmp( Register s1, int simm13a, RelocationHolder const& rspec ) { jmpl( s1, simm13a, G0, rspec); }

inline bool MacroAssembler::is_far_target(address d) {
  if (ForceUnreachable) {
    // References outside the code cache should be treated as far
    return d < CodeCache::low_bound() || d > CodeCache::high_bound();
  }
  return !is_in_wdisp30_range(d, CodeCache::low_bound()) || !is_in_wdisp30_range(d, CodeCache::high_bound());
}

// Call with a check to see if we need to deal with the added
// expense of relocation and if we overflow the displacement
// of the quick call instruction.
inline void MacroAssembler::call( address d, relocInfo::relocType rt ) {
#ifdef _LP64
  intptr_t disp;
  // NULL is ok because it will be relocated later.
  // Must change NULL to a reachable address in order to
  // pass asserts here and in wdisp.
  if ( d == NULL )
    d = pc();

  // Is this address within range of the call instruction?
  // If not, use the expensive instruction sequence
  if (is_far_target(d)) {
    relocate(rt);
    AddressLiteral dest(d);
    jumpl_to(dest, O7, O7);
  } else {
    Assembler::call(d, rt);
  }
#else
  Assembler::call( d, rt );
#endif
}

inline void MacroAssembler::call( Label& L,   relocInfo::relocType rt ) {
  MacroAssembler::call( target(L), rt);
}



inline void MacroAssembler::callr( Register s1, Register s2 ) { jmpl( s1, s2, O7 ); }
inline void MacroAssembler::callr( Register s1, int simm13a, RelocationHolder const& rspec ) { jmpl( s1, simm13a, O7, rspec); }

// prefetch instruction
inline void MacroAssembler::iprefetch( address d, relocInfo::relocType rt ) {
  if (VM_Version::v9_instructions_work())
    Assembler::bp( never, true, xcc, pt, d, rt );
}
inline void MacroAssembler::iprefetch( Label& L) { iprefetch( target(L) ); }


// clobbers o7 on V8!!
// returns delta from gotten pc to addr after
inline int MacroAssembler::get_pc( Register d ) {
  int x = offset();
  if (VM_Version::v9_instructions_work())
    rdpc(d);
  else {
    Label lbl;
    Assembler::call(lbl, relocInfo::none);  // No relocation as this is call to pc+0x8
    if (d == O7)  delayed()->nop();
    else          delayed()->mov(O7, d);
    bind(lbl);
  }
  return offset() - x;
}


// Note:  All MacroAssembler::set_foo functions are defined out-of-line.


// Loads the current PC of the following instruction as an immediate value in
// 2 instructions.  All PCs in the CodeCache are within 2 Gig of each other.
inline intptr_t MacroAssembler::load_pc_address( Register reg, int bytes_to_skip ) {
  intptr_t thepc = (intptr_t)pc() + 2*BytesPerInstWord + bytes_to_skip;
#ifdef _LP64
  Unimplemented();
#else
  Assembler::sethi(   thepc & ~0x3ff, reg, internal_word_Relocation::spec((address)thepc));
             add(reg, thepc &  0x3ff, reg, internal_word_Relocation::spec((address)thepc));
#endif
  return thepc;
}


inline void MacroAssembler::load_contents(const AddressLiteral& addrlit, Register d, int offset) {
  assert_not_delayed();
  if (ForceUnreachable) {
    patchable_sethi(addrlit, d);
  } else {
    sethi(addrlit, d);
  }
  ld(d, addrlit.low10() + offset, d);
}


inline void MacroAssembler::load_bool_contents(const AddressLiteral& addrlit, Register d, int offset) {
  assert_not_delayed();
  if (ForceUnreachable) {
    patchable_sethi(addrlit, d);
  } else {
    sethi(addrlit, d);
  }
  ldub(d, addrlit.low10() + offset, d);
}


inline void MacroAssembler::load_ptr_contents(const AddressLiteral& addrlit, Register d, int offset) {
  assert_not_delayed();
  if (ForceUnreachable) {
    patchable_sethi(addrlit, d);
  } else {
    sethi(addrlit, d);
  }
  ld_ptr(d, addrlit.low10() + offset, d);
}


inline void MacroAssembler::store_contents(Register s, const AddressLiteral& addrlit, Register temp, int offset) {
  assert_not_delayed();
  if (ForceUnreachable) {
    patchable_sethi(addrlit, temp);
  } else {
    sethi(addrlit, temp);
  }
  st(s, temp, addrlit.low10() + offset);
}


inline void MacroAssembler::store_ptr_contents(Register s, const AddressLiteral& addrlit, Register temp, int offset) {
  assert_not_delayed();
  if (ForceUnreachable) {
    patchable_sethi(addrlit, temp);
  } else {
    sethi(addrlit, temp);
  }
  st_ptr(s, temp, addrlit.low10() + offset);
}


// This code sequence is relocatable to any address, even on LP64.
inline void MacroAssembler::jumpl_to(const AddressLiteral& addrlit, Register temp, Register d, int offset) {
  assert_not_delayed();
  // Force fixed length sethi because NativeJump and NativeFarCall don't handle
  // variable length instruction streams.
  patchable_sethi(addrlit, temp);
  jmpl(temp, addrlit.low10() + offset, d);
}


inline void MacroAssembler::jump_to(const AddressLiteral& addrlit, Register temp, int offset) {
  jumpl_to(addrlit, temp, G0, offset);
}


inline void MacroAssembler::jump_indirect_to(Address& a, Register temp,
                                             int ld_offset, int jmp_offset) {
  assert_not_delayed();
  //sethi(al);                   // sethi is caller responsibility for this one
  ld_ptr(a, temp, ld_offset);
  jmp(temp, jmp_offset);
}


inline void MacroAssembler::set_metadata(Metadata* obj, Register d) {
  set_metadata(allocate_metadata_address(obj), d);
}

inline void MacroAssembler::set_metadata_constant(Metadata* obj, Register d) {
  set_metadata(constant_metadata_address(obj), d);
}

inline void MacroAssembler::set_metadata(const AddressLiteral& obj_addr, Register d) {
  assert(obj_addr.rspec().type() == relocInfo::metadata_type, "must be a metadata reloc");
  set(obj_addr, d);
}

inline void MacroAssembler::set_oop(jobject obj, Register d) {
  set_oop(allocate_oop_address(obj), d);
}


inline void MacroAssembler::set_oop_constant(jobject obj, Register d) {
  set_oop(constant_oop_address(obj), d);
}


inline void MacroAssembler::set_oop(const AddressLiteral& obj_addr, Register d) {
  assert(obj_addr.rspec().type() == relocInfo::oop_type, "must be an oop reloc");
  set(obj_addr, d);
}


inline void MacroAssembler::load_argument( Argument& a, Register  d ) {
  if (a.is_register())
    mov(a.as_register(), d);
  else
    ld (a.as_address(),  d);
}

inline void MacroAssembler::store_argument( Register s, Argument& a ) {
  if (a.is_register())
    mov(s, a.as_register());
  else
    st_ptr (s, a.as_address());         // ABI says everything is right justified.
}

inline void MacroAssembler::store_ptr_argument( Register s, Argument& a ) {
  if (a.is_register())
    mov(s, a.as_register());
  else
    st_ptr (s, a.as_address());
}


#ifdef _LP64
inline void MacroAssembler::store_float_argument( FloatRegister s, Argument& a ) {
  if (a.is_float_register())
// V9 ABI has F1, F3, F5 are used to pass instead of O0, O1, O2
    fmov(FloatRegisterImpl::S, s, a.as_float_register() );
  else
    // Floats are stored in the high half of the stack entry
    // The low half is undefined per the ABI.
    stf(FloatRegisterImpl::S, s, a.as_address(), sizeof(jfloat));
}

inline void MacroAssembler::store_double_argument( FloatRegister s, Argument& a ) {
  if (a.is_float_register())
// V9 ABI has D0, D2, D4 are used to pass instead of O0, O1, O2
    fmov(FloatRegisterImpl::D, s, a.as_double_register() );
  else
    stf(FloatRegisterImpl::D, s, a.as_address());
}

inline void MacroAssembler::store_long_argument( Register s, Argument& a ) {
  if (a.is_register())
    mov(s, a.as_register());
  else
    stx(s, a.as_address());
}
#endif

inline void MacroAssembler::add(Register s1, int simm13a, Register d, relocInfo::relocType rtype) {
  relocate(rtype);
  add(s1, simm13a, d);
}
inline void MacroAssembler::add(Register s1, int simm13a, Register d, RelocationHolder const& rspec) {
  relocate(rspec);
  add(s1, simm13a, d);
}

// form effective addresses this way:
inline void MacroAssembler::add(const Address& a, Register d, int offset) {
  if (a.has_index())   add(a.base(), a.index(),         d);
  else               { add(a.base(), a.disp() + offset, d, a.rspec(offset)); offset = 0; }
  if (offset != 0)     add(d,        offset,            d);
}
inline void MacroAssembler::add(Register s1, RegisterOrConstant s2, Register d, int offset) {
  if (s2.is_register())  add(s1, s2.as_register(),          d);
  else                 { add(s1, s2.as_constant() + offset, d); offset = 0; }
  if (offset != 0)       add(d,  offset,                    d);
}

inline void MacroAssembler::andn(Register s1, RegisterOrConstant s2, Register d) {
  if (s2.is_register())  andn(s1, s2.as_register(), d);
  else                   andn(s1, s2.as_constant(), d);
}

inline void MacroAssembler::clrb( Register s1, Register s2) { stb( G0, s1, s2 ); }
inline void MacroAssembler::clrh( Register s1, Register s2) { sth( G0, s1, s2 ); }
inline void MacroAssembler::clr(  Register s1, Register s2) { stw( G0, s1, s2 ); }
inline void MacroAssembler::clrx( Register s1, Register s2) { stx( G0, s1, s2 ); }

inline void MacroAssembler::clrb( Register s1, int simm13a) { stb( G0, s1, simm13a); }
inline void MacroAssembler::clrh( Register s1, int simm13a) { sth( G0, s1, simm13a); }
inline void MacroAssembler::clr(  Register s1, int simm13a) { stw( G0, s1, simm13a); }
inline void MacroAssembler::clrx( Register s1, int simm13a) { stx( G0, s1, simm13a); }

#ifdef _LP64
// Make all 32 bit loads signed so 64 bit registers maintain proper sign
inline void MacroAssembler::ld(  Register s1, Register s2, Register d)      { ldsw( s1, s2, d); }
inline void MacroAssembler::ld(  Register s1, int simm13a, Register d)      { ldsw( s1, simm13a, d); }
#else
inline void MacroAssembler::ld(  Register s1, Register s2, Register d)      { lduw( s1, s2, d); }
inline void MacroAssembler::ld(  Register s1, int simm13a, Register d)      { lduw( s1, simm13a, d); }
#endif

#ifdef ASSERT
  // ByteSize is only a class when ASSERT is defined, otherwise it's an int.
# ifdef _LP64
inline void MacroAssembler::ld(Register s1, ByteSize simm13a, Register d) { ldsw( s1, in_bytes(simm13a), d); }
# else
inline void MacroAssembler::ld(Register s1, ByteSize simm13a, Register d) { lduw( s1, in_bytes(simm13a), d); }
# endif
#endif

inline void MacroAssembler::ld(  const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ld(  a.base(), a.index(),         d); }
  else               {                          ld(  a.base(), a.disp() + offset, d); }
}

inline void MacroAssembler::ldsb(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldsb(a.base(), a.index(),         d); }
  else               {                          ldsb(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::ldsh(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldsh(a.base(), a.index(),         d); }
  else               {                          ldsh(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::ldsw(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldsw(a.base(), a.index(),         d); }
  else               {                          ldsw(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::ldub(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldub(a.base(), a.index(),         d); }
  else               {                          ldub(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::lduh(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); lduh(a.base(), a.index(),         d); }
  else               {                          lduh(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::lduw(const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); lduw(a.base(), a.index(),         d); }
  else               {                          lduw(a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::ldd( const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldd( a.base(), a.index(),         d); }
  else               {                          ldd( a.base(), a.disp() + offset, d); }
}
inline void MacroAssembler::ldx( const Address& a, Register d, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); ldx( a.base(), a.index(),         d); }
  else               {                          ldx( a.base(), a.disp() + offset, d); }
}

inline void MacroAssembler::ldub(Register s1, RegisterOrConstant s2, Register d) { ldub(Address(s1, s2), d); }
inline void MacroAssembler::ldsb(Register s1, RegisterOrConstant s2, Register d) { ldsb(Address(s1, s2), d); }
inline void MacroAssembler::lduh(Register s1, RegisterOrConstant s2, Register d) { lduh(Address(s1, s2), d); }
inline void MacroAssembler::ldsh(Register s1, RegisterOrConstant s2, Register d) { ldsh(Address(s1, s2), d); }
inline void MacroAssembler::lduw(Register s1, RegisterOrConstant s2, Register d) { lduw(Address(s1, s2), d); }
inline void MacroAssembler::ldsw(Register s1, RegisterOrConstant s2, Register d) { ldsw(Address(s1, s2), d); }
inline void MacroAssembler::ldx( Register s1, RegisterOrConstant s2, Register d) { ldx( Address(s1, s2), d); }
inline void MacroAssembler::ld(  Register s1, RegisterOrConstant s2, Register d) { ld(  Address(s1, s2), d); }
inline void MacroAssembler::ldd( Register s1, RegisterOrConstant s2, Register d) { ldd( Address(s1, s2), d); }

inline void MacroAssembler::ldf(FloatRegisterImpl::Width w, Register s1, RegisterOrConstant s2, FloatRegister d) {
  if (s2.is_register())  ldf(w, s1, s2.as_register(), d);
  else                   ldf(w, s1, s2.as_constant(), d);
}

inline void MacroAssembler::ldf(FloatRegisterImpl::Width w, const Address& a, FloatRegister d, int offset) {
  relocate(a.rspec(offset));
  ldf(w, a.base(), a.disp() + offset, d);
}

// returns if membar generates anything, obviously this code should mirror
// membar below.
inline bool MacroAssembler::membar_has_effect( Membar_mask_bits const7a ) {
  if( !os::is_MP() ) return false;  // Not needed on single CPU
  if( VM_Version::v9_instructions_work() ) {
    const Membar_mask_bits effective_mask =
        Membar_mask_bits(const7a & ~(LoadLoad | LoadStore | StoreStore));
    return (effective_mask != 0);
  } else {
    return true;
  }
}

inline void MacroAssembler::membar( Membar_mask_bits const7a ) {
  // Uniprocessors do not need memory barriers
  if (!os::is_MP()) return;
  // Weakened for current Sparcs and TSO.  See the v9 manual, sections 8.4.3,
  // 8.4.4.3, a.31 and a.50.
  if( VM_Version::v9_instructions_work() ) {
    // Under TSO, setting bit 3, 2, or 0 is redundant, so the only value
    // of the mmask subfield of const7a that does anything that isn't done
    // implicitly is StoreLoad.
    const Membar_mask_bits effective_mask =
        Membar_mask_bits(const7a & ~(LoadLoad | LoadStore | StoreStore));
    if ( effective_mask != 0 ) {
      Assembler::membar( effective_mask );
    }
  } else {
    // stbar is the closest there is on v8.  Equivalent to membar(StoreStore).  We
    // do not issue the stbar because to my knowledge all v8 machines implement TSO,
    // which guarantees that all stores behave as if an stbar were issued just after
    // each one of them.  On these machines, stbar ought to be a nop.  There doesn't
    // appear to be an equivalent of membar(StoreLoad) on v8: TSO doesn't require it,
    // it can't be specified by stbar, nor have I come up with a way to simulate it.
    //
    // Addendum.  Dave says that ldstub guarantees a write buffer flush to coherent
    // space.  Put one here to be on the safe side.
    Assembler::ldstub(SP, 0, G0);
  }
}

inline void MacroAssembler::prefetch(const Address& a, PrefetchFcn f, int offset) {
  relocate(a.rspec(offset));
  assert(!a.has_index(), "");
  prefetch(a.base(), a.disp() + offset, f);
}

inline void MacroAssembler::st(Register d, Register s1, Register s2)      { stw(d, s1, s2); }
inline void MacroAssembler::st(Register d, Register s1, int simm13a)      { stw(d, s1, simm13a); }

#ifdef ASSERT
// ByteSize is only a class when ASSERT is defined, otherwise it's an int.
inline void MacroAssembler::st(Register d, Register s1, ByteSize simm13a) { stw(d, s1, in_bytes(simm13a)); }
#endif

inline void MacroAssembler::st(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); st( d, a.base(), a.index()        ); }
  else               {                          st( d, a.base(), a.disp() + offset); }
}

inline void MacroAssembler::stb(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); stb(d, a.base(), a.index()        ); }
  else               {                          stb(d, a.base(), a.disp() + offset); }
}
inline void MacroAssembler::sth(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); sth(d, a.base(), a.index()        ); }
  else               {                          sth(d, a.base(), a.disp() + offset); }
}
inline void MacroAssembler::stw(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); stw(d, a.base(), a.index()        ); }
  else               {                          stw(d, a.base(), a.disp() + offset); }
}
inline void MacroAssembler::std(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); std(d, a.base(), a.index()        ); }
  else               {                          std(d, a.base(), a.disp() + offset); }
}
inline void MacroAssembler::stx(Register d, const Address& a, int offset) {
  if (a.has_index()) { assert(offset == 0, ""); stx(d, a.base(), a.index()        ); }
  else               {                          stx(d, a.base(), a.disp() + offset); }
}

inline void MacroAssembler::stb(Register d, Register s1, RegisterOrConstant s2) { stb(d, Address(s1, s2)); }
inline void MacroAssembler::sth(Register d, Register s1, RegisterOrConstant s2) { sth(d, Address(s1, s2)); }
inline void MacroAssembler::stw(Register d, Register s1, RegisterOrConstant s2) { stw(d, Address(s1, s2)); }
inline void MacroAssembler::stx(Register d, Register s1, RegisterOrConstant s2) { stx(d, Address(s1, s2)); }
inline void MacroAssembler::std(Register d, Register s1, RegisterOrConstant s2) { std(d, Address(s1, s2)); }
inline void MacroAssembler::st( Register d, Register s1, RegisterOrConstant s2) { st( d, Address(s1, s2)); }

inline void MacroAssembler::stf(FloatRegisterImpl::Width w, FloatRegister d, Register s1, RegisterOrConstant s2) {
  if (s2.is_register())  stf(w, d, s1, s2.as_register());
  else                   stf(w, d, s1, s2.as_constant());
}

inline void MacroAssembler::stf(FloatRegisterImpl::Width w, FloatRegister d, const Address& a, int offset) {
  relocate(a.rspec(offset));
  if (a.has_index()) { assert(offset == 0, ""); stf(w, d, a.base(), a.index()        ); }
  else               {                          stf(w, d, a.base(), a.disp() + offset); }
}

inline void MacroAssembler::sub(Register s1, RegisterOrConstant s2, Register d, int offset) {
  if (s2.is_register())  sub(s1, s2.as_register(),          d);
  else                 { sub(s1, s2.as_constant() + offset, d); offset = 0; }
  if (offset != 0)       sub(d,  offset,                    d);
}

inline void MacroAssembler::swap(Address& a, Register d, int offset) {
  relocate(a.rspec(offset));
  if (a.has_index()) { assert(offset == 0, ""); swap(a.base(), a.index(), d        ); }
  else               {                          swap(a.base(), a.disp() + offset, d); }
}

#endif // CPU_SPARC_VM_MACROASSEMBLER_SPARC_INLINE_HPP