1 /*
   2  * Copyright (c) 1997, 2023, Oracle and/or its affiliates. All rights reserved.
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * This code is free software; you can redistribute it and/or modify it
   6  * under the terms of the GNU General Public License version 2 only, as
   7  * published by the Free Software Foundation.
   8  *
   9  * This code is distributed in the hope that it will be useful, but WITHOUT
  10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12  * version 2 for more details (a copy is included in the LICENSE file that
  13  * accompanied this code).
  14  *
  15  * You should have received a copy of the GNU General Public License version
  16  * 2 along with this work; if not, write to the Free Software Foundation,
  17  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
  18  *
  19  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
  20  * or visit www.oracle.com if you need additional information or have any
  21  * questions.
  22  *
  23  */
  24 
  25 #ifndef CPU_X86_ASSEMBLER_X86_HPP
  26 #define CPU_X86_ASSEMBLER_X86_HPP
  27 
  28 #include "asm/register.hpp"
  29 #include "utilities/checkedCast.hpp"
  30 #include "utilities/powerOfTwo.hpp"
  31 
  32 // Contains all the definitions needed for x86 assembly code generation.
  33 
  34 // Calling convention
  35 class Argument {
  36  public:
  37   enum {
  38 #ifdef _LP64
  39 #ifdef _WIN64
  40     n_int_register_parameters_c   = 4, // rcx, rdx, r8, r9 (c_rarg0, c_rarg1, ...)
  41     n_float_register_parameters_c = 4,  // xmm0 - xmm3 (c_farg0, c_farg1, ... )
  42     n_int_register_returns_c = 1, // rax
  43     n_float_register_returns_c = 1, // xmm0
  44 #else
  45     n_int_register_parameters_c   = 6, // rdi, rsi, rdx, rcx, r8, r9 (c_rarg0, c_rarg1, ...)
  46     n_float_register_parameters_c = 8,  // xmm0 - xmm7 (c_farg0, c_farg1, ... )
  47     n_int_register_returns_c = 2, // rax, rdx
  48     n_float_register_returns_c = 2, // xmm0, xmm1
  49 #endif // _WIN64
  50     n_int_register_parameters_j   = 6, // j_rarg0, j_rarg1, ...
  51     n_float_register_parameters_j = 8  // j_farg0, j_farg1, ...
  52 #else
  53     n_register_parameters = 0,   // 0 registers used to pass arguments
  54     n_int_register_parameters_j   = 0,
  55     n_float_register_parameters_j = 0
  56 #endif // _LP64
  57   };
  58 };
  59 
  60 
  61 #ifdef _LP64
  62 // Symbolically name the register arguments used by the c calling convention.
  63 // Windows is different from linux/solaris. So much for standards...
  64 
  65 #ifdef _WIN64
  66 
  67 constexpr Register c_rarg0 = rcx;
  68 constexpr Register c_rarg1 = rdx;
  69 constexpr Register c_rarg2 =  r8;
  70 constexpr Register c_rarg3 =  r9;
  71 
  72 constexpr XMMRegister c_farg0 = xmm0;
  73 constexpr XMMRegister c_farg1 = xmm1;
  74 constexpr XMMRegister c_farg2 = xmm2;
  75 constexpr XMMRegister c_farg3 = xmm3;
  76 
  77 #else
  78 
  79 constexpr Register c_rarg0 = rdi;
  80 constexpr Register c_rarg1 = rsi;
  81 constexpr Register c_rarg2 = rdx;
  82 constexpr Register c_rarg3 = rcx;
  83 constexpr Register c_rarg4 =  r8;
  84 constexpr Register c_rarg5 =  r9;
  85 
  86 constexpr XMMRegister c_farg0 = xmm0;
  87 constexpr XMMRegister c_farg1 = xmm1;
  88 constexpr XMMRegister c_farg2 = xmm2;
  89 constexpr XMMRegister c_farg3 = xmm3;
  90 constexpr XMMRegister c_farg4 = xmm4;
  91 constexpr XMMRegister c_farg5 = xmm5;
  92 constexpr XMMRegister c_farg6 = xmm6;
  93 constexpr XMMRegister c_farg7 = xmm7;
  94 
  95 #endif // _WIN64
  96 
  97 // Symbolically name the register arguments used by the Java calling convention.
  98 // We have control over the convention for java so we can do what we please.
  99 // What pleases us is to offset the java calling convention so that when
 100 // we call a suitable jni method the arguments are lined up and we don't
 101 // have to do little shuffling. A suitable jni method is non-static and a
 102 // small number of arguments (two fewer args on windows)
 103 //
 104 //        |-------------------------------------------------------|
 105 //        | c_rarg0   c_rarg1  c_rarg2 c_rarg3 c_rarg4 c_rarg5    |
 106 //        |-------------------------------------------------------|
 107 //        | rcx       rdx      r8      r9      rdi*    rsi*       | windows (* not a c_rarg)
 108 //        | rdi       rsi      rdx     rcx     r8      r9         | solaris/linux
 109 //        |-------------------------------------------------------|
 110 //        | j_rarg5   j_rarg0  j_rarg1 j_rarg2 j_rarg3 j_rarg4    |
 111 //        |-------------------------------------------------------|
 112 
 113 constexpr Register j_rarg0 = c_rarg1;
 114 constexpr Register j_rarg1 = c_rarg2;
 115 constexpr Register j_rarg2 = c_rarg3;
 116 // Windows runs out of register args here
 117 #ifdef _WIN64
 118 constexpr Register j_rarg3 = rdi;
 119 constexpr Register j_rarg4 = rsi;
 120 #else
 121 constexpr Register j_rarg3 = c_rarg4;
 122 constexpr Register j_rarg4 = c_rarg5;
 123 #endif /* _WIN64 */
 124 constexpr Register j_rarg5 = c_rarg0;
 125 
 126 constexpr XMMRegister j_farg0 = xmm0;
 127 constexpr XMMRegister j_farg1 = xmm1;
 128 constexpr XMMRegister j_farg2 = xmm2;
 129 constexpr XMMRegister j_farg3 = xmm3;
 130 constexpr XMMRegister j_farg4 = xmm4;
 131 constexpr XMMRegister j_farg5 = xmm5;
 132 constexpr XMMRegister j_farg6 = xmm6;
 133 constexpr XMMRegister j_farg7 = xmm7;
 134 
 135 constexpr Register rscratch1 = r10;  // volatile
 136 constexpr Register rscratch2 = r11;  // volatile
 137 
 138 constexpr Register r12_heapbase = r12; // callee-saved
 139 constexpr Register r15_thread   = r15; // callee-saved
 140 
 141 #else
 142 // rscratch1 will appear in 32bit code that is dead but of course must compile
 143 // Using noreg ensures if the dead code is incorrectly live and executed it
 144 // will cause an assertion failure
 145 #define rscratch1 noreg
 146 #define rscratch2 noreg
 147 
 148 #endif // _LP64
 149 
 150 // JSR 292
 151 // On x86, the SP does not have to be saved when invoking method handle intrinsics
 152 // or compiled lambda forms. We indicate that by setting rbp_mh_SP_save to noreg.
 153 constexpr Register rbp_mh_SP_save = noreg;
 154 
 155 // Address is an abstraction used to represent a memory location
 156 // using any of the amd64 addressing modes with one object.
 157 //
 158 // Note: A register location is represented via a Register, not
 159 //       via an address for efficiency & simplicity reasons.
 160 
 161 class ArrayAddress;
 162 
 163 class Address {
 164  public:
 165   enum ScaleFactor {
 166     no_scale = -1,
 167     times_1  =  0,
 168     times_2  =  1,
 169     times_4  =  2,
 170     times_8  =  3,
 171     times_ptr = LP64_ONLY(times_8) NOT_LP64(times_4)
 172   };
 173   static ScaleFactor times(int size) {
 174     assert(size >= 1 && size <= 8 && is_power_of_2(size), "bad scale size");
 175     if (size == 8)  return times_8;
 176     if (size == 4)  return times_4;
 177     if (size == 2)  return times_2;
 178     return times_1;
 179   }
 180   static int scale_size(ScaleFactor scale) {
 181     assert(scale != no_scale, "");
 182     assert(((1 << (int)times_1) == 1 &&
 183             (1 << (int)times_2) == 2 &&
 184             (1 << (int)times_4) == 4 &&
 185             (1 << (int)times_8) == 8), "");
 186     return (1 << (int)scale);
 187   }
 188 
 189  private:
 190   Register         _base;
 191   Register         _index;
 192   XMMRegister      _xmmindex;
 193   ScaleFactor      _scale;
 194   int              _disp;
 195   bool             _isxmmindex;
 196   RelocationHolder _rspec;
 197 
 198   // Easily misused constructors make them private
 199   // %%% can we make these go away?
 200   NOT_LP64(Address(address loc, RelocationHolder spec);)
 201   Address(int disp, address loc, relocInfo::relocType rtype);
 202   Address(int disp, address loc, RelocationHolder spec);
 203 
 204  public:
 205 
 206  int disp() { return _disp; }
 207   // creation
 208   Address()
 209     : _base(noreg),
 210       _index(noreg),
 211       _xmmindex(xnoreg),
 212       _scale(no_scale),
 213       _disp(0),
 214       _isxmmindex(false){
 215   }
 216 
 217   explicit Address(Register base, int disp = 0)
 218     : _base(base),
 219       _index(noreg),
 220       _xmmindex(xnoreg),
 221       _scale(no_scale),
 222       _disp(disp),
 223       _isxmmindex(false){
 224   }
 225 
 226   Address(Register base, Register index, ScaleFactor scale, int disp = 0)
 227     : _base (base),
 228       _index(index),
 229       _xmmindex(xnoreg),
 230       _scale(scale),
 231       _disp (disp),
 232       _isxmmindex(false) {
 233     assert(!index->is_valid() == (scale == Address::no_scale),
 234            "inconsistent address");
 235   }
 236 
 237   Address(Register base, RegisterOrConstant index, ScaleFactor scale = times_1, int disp = 0)
 238     : _base (base),
 239       _index(index.register_or_noreg()),
 240       _xmmindex(xnoreg),
 241       _scale(scale),
 242       _disp (disp + checked_cast<int>(index.constant_or_zero() * scale_size(scale))),
 243       _isxmmindex(false){
 244     if (!index.is_register())  scale = Address::no_scale;
 245     assert(!_index->is_valid() == (scale == Address::no_scale),
 246            "inconsistent address");
 247   }
 248 
 249   Address(Register base, XMMRegister index, ScaleFactor scale, int disp = 0)
 250     : _base (base),
 251       _index(noreg),
 252       _xmmindex(index),
 253       _scale(scale),
 254       _disp(disp),
 255       _isxmmindex(true) {
 256       assert(!index->is_valid() == (scale == Address::no_scale),
 257              "inconsistent address");
 258   }
 259 
 260   // The following overloads are used in connection with the
 261   // ByteSize type (see sizes.hpp).  They simplify the use of
 262   // ByteSize'd arguments in assembly code.
 263 
 264   Address(Register base, ByteSize disp)
 265     : Address(base, in_bytes(disp)) {}
 266 
 267   Address(Register base, Register index, ScaleFactor scale, ByteSize disp)
 268     : Address(base, index, scale, in_bytes(disp)) {}
 269 
 270   Address(Register base, RegisterOrConstant index, ScaleFactor scale, ByteSize disp)
 271     : Address(base, index, scale, in_bytes(disp)) {}
 272 
 273   Address plus_disp(int disp) const {
 274     Address a = (*this);
 275     a._disp += disp;
 276     return a;
 277   }
 278   Address plus_disp(RegisterOrConstant disp, ScaleFactor scale = times_1) const {
 279     Address a = (*this);
 280     a._disp += checked_cast<int>(disp.constant_or_zero() * scale_size(scale));
 281     if (disp.is_register()) {
 282       assert(!a.index()->is_valid(), "competing indexes");
 283       a._index = disp.as_register();
 284       a._scale = scale;
 285     }
 286     return a;
 287   }
 288   bool is_same_address(Address a) const {
 289     // disregard _rspec
 290     return _base == a._base && _disp == a._disp && _index == a._index && _scale == a._scale;
 291   }
 292 
 293   // accessors
 294   bool        uses(Register reg) const { return _base == reg || _index == reg; }
 295   Register    base()             const { return _base;  }
 296   Register    index()            const { return _index; }
 297   XMMRegister xmmindex()         const { return _xmmindex; }
 298   ScaleFactor scale()            const { return _scale; }
 299   int         disp()             const { return _disp;  }
 300   bool        isxmmindex()       const { return _isxmmindex; }
 301 
 302   // Convert the raw encoding form into the form expected by the constructor for
 303   // Address.  An index of 4 (rsp) corresponds to having no index, so convert
 304   // that to noreg for the Address constructor.
 305   static Address make_raw(int base, int index, int scale, int disp, relocInfo::relocType disp_reloc);
 306 
 307   static Address make_array(ArrayAddress);
 308 
 309  private:
 310   bool base_needs_rex() const {
 311     return _base->is_valid() && _base->encoding() >= 8;
 312   }
 313 
 314   bool index_needs_rex() const {
 315     return _index->is_valid() &&_index->encoding() >= 8;
 316   }
 317 
 318   bool xmmindex_needs_rex() const {
 319     return _xmmindex->is_valid() && _xmmindex->encoding() >= 8;
 320   }
 321 
 322   relocInfo::relocType reloc() const { return _rspec.type(); }
 323 
 324   friend class Assembler;
 325   friend class MacroAssembler;
 326   friend class LIR_Assembler; // base/index/scale/disp
 327 };
 328 
 329 //
 330 // AddressLiteral has been split out from Address because operands of this type
 331 // need to be treated specially on 32bit vs. 64bit platforms. By splitting it out
 332 // the few instructions that need to deal with address literals are unique and the
 333 // MacroAssembler does not have to implement every instruction in the Assembler
 334 // in order to search for address literals that may need special handling depending
 335 // on the instruction and the platform. As small step on the way to merging i486/amd64
 336 // directories.
 337 //
 338 class AddressLiteral {
 339   friend class ArrayAddress;
 340   RelocationHolder _rspec;
 341   // Typically we use AddressLiterals we want to use their rval
 342   // However in some situations we want the lval (effect address) of the item.
 343   // We provide a special factory for making those lvals.
 344   bool _is_lval;
 345 
 346   // If the target is far we'll need to load the ea of this to
 347   // a register to reach it. Otherwise if near we can do rip
 348   // relative addressing.
 349 
 350   address          _target;
 351 
 352  protected:
 353   // creation
 354   AddressLiteral()
 355     : _is_lval(false),
 356       _target(nullptr)
 357   {}
 358 
 359   public:
 360 
 361 
 362   AddressLiteral(address target, relocInfo::relocType rtype);
 363 
 364   AddressLiteral(address target, RelocationHolder const& rspec)
 365     : _rspec(rspec),
 366       _is_lval(false),
 367       _target(target)
 368   {}
 369 
 370   AddressLiteral addr() {
 371     AddressLiteral ret = *this;
 372     ret._is_lval = true;
 373     return ret;
 374   }
 375 
 376 
 377  private:
 378 
 379   address target() { return _target; }
 380   bool is_lval() const { return _is_lval; }
 381 
 382   relocInfo::relocType reloc() const { return _rspec.type(); }
 383   const RelocationHolder& rspec() const { return _rspec; }
 384 
 385   friend class Assembler;
 386   friend class MacroAssembler;
 387   friend class Address;
 388   friend class LIR_Assembler;
 389 };
 390 
 391 // Convenience classes
 392 class RuntimeAddress: public AddressLiteral {
 393 
 394   public:
 395 
 396   RuntimeAddress(address target) : AddressLiteral(target, relocInfo::runtime_call_type) {}
 397 
 398 };
 399 
 400 class ExternalAddress: public AddressLiteral {
 401  private:
 402   static relocInfo::relocType reloc_for_target(address target) {
 403     // Sometimes ExternalAddress is used for values which aren't
 404     // exactly addresses, like the card table base.
 405     // external_word_type can't be used for values in the first page
 406     // so just skip the reloc in that case.
 407     return external_word_Relocation::can_be_relocated(target) ? relocInfo::external_word_type : relocInfo::none;
 408   }
 409 
 410  public:
 411 
 412   ExternalAddress(address target) : AddressLiteral(target, reloc_for_target(target)) {}
 413 
 414 };
 415 
 416 class InternalAddress: public AddressLiteral {
 417 
 418   public:
 419 
 420   InternalAddress(address target) : AddressLiteral(target, relocInfo::internal_word_type) {}
 421 
 422 };
 423 
 424 // x86 can do array addressing as a single operation since disp can be an absolute
 425 // address amd64 can't. We create a class that expresses the concept but does extra
 426 // magic on amd64 to get the final result
 427 
 428 class ArrayAddress {
 429   private:
 430 
 431   AddressLiteral _base;
 432   Address        _index;
 433 
 434   public:
 435 
 436   ArrayAddress() {};
 437   ArrayAddress(AddressLiteral base, Address index): _base(base), _index(index) {};
 438   AddressLiteral base() { return _base; }
 439   Address index() { return _index; }
 440 
 441 };
 442 
 443 class InstructionAttr;
 444 
 445 // 64-bit reflect the fxsave size which is 512 bytes and the new xsave area on EVEX which is another 2176 bytes
 446 // See fxsave and xsave(EVEX enabled) documentation for layout
 447 const int FPUStateSizeInWords = NOT_LP64(27) LP64_ONLY(2688 / wordSize);
 448 
 449 // The Intel x86/Amd64 Assembler: Pure assembler doing NO optimizations on the instruction
 450 // level (e.g. mov rax, 0 is not translated into xor rax, rax!); i.e., what you write
 451 // is what you get. The Assembler is generating code into a CodeBuffer.
 452 
 453 class Assembler : public AbstractAssembler  {
 454   friend class AbstractAssembler; // for the non-virtual hack
 455   friend class LIR_Assembler; // as_Address()
 456   friend class StubGenerator;
 457 
 458  public:
 459   enum Condition {                     // The x86 condition codes used for conditional jumps/moves.
 460     zero          = 0x4,
 461     notZero       = 0x5,
 462     equal         = 0x4,
 463     notEqual      = 0x5,
 464     less          = 0xc,
 465     lessEqual     = 0xe,
 466     greater       = 0xf,
 467     greaterEqual  = 0xd,
 468     below         = 0x2,
 469     belowEqual    = 0x6,
 470     above         = 0x7,
 471     aboveEqual    = 0x3,
 472     overflow      = 0x0,
 473     noOverflow    = 0x1,
 474     carrySet      = 0x2,
 475     carryClear    = 0x3,
 476     negative      = 0x8,
 477     positive      = 0x9,
 478     parity        = 0xa,
 479     noParity      = 0xb
 480   };
 481 
 482   enum Prefix {
 483     // segment overrides
 484     CS_segment = 0x2e,
 485     SS_segment = 0x36,
 486     DS_segment = 0x3e,
 487     ES_segment = 0x26,
 488     FS_segment = 0x64,
 489     GS_segment = 0x65,
 490 
 491     REX        = 0x40,
 492 
 493     REX_B      = 0x41,
 494     REX_X      = 0x42,
 495     REX_XB     = 0x43,
 496     REX_R      = 0x44,
 497     REX_RB     = 0x45,
 498     REX_RX     = 0x46,
 499     REX_RXB    = 0x47,
 500 
 501     REX_W      = 0x48,
 502 
 503     REX_WB     = 0x49,
 504     REX_WX     = 0x4A,
 505     REX_WXB    = 0x4B,
 506     REX_WR     = 0x4C,
 507     REX_WRB    = 0x4D,
 508     REX_WRX    = 0x4E,
 509     REX_WRXB   = 0x4F,
 510 
 511     VEX_3bytes = 0xC4,
 512     VEX_2bytes = 0xC5,
 513     EVEX_4bytes = 0x62,
 514     Prefix_EMPTY = 0x0
 515   };
 516 
 517   enum VexPrefix {
 518     VEX_B = 0x20,
 519     VEX_X = 0x40,
 520     VEX_R = 0x80,
 521     VEX_W = 0x80
 522   };
 523 
 524   enum ExexPrefix {
 525     EVEX_F  = 0x04,
 526     EVEX_V  = 0x08,
 527     EVEX_Rb = 0x10,
 528     EVEX_X  = 0x40,
 529     EVEX_Z  = 0x80
 530   };
 531 
 532   enum EvexRoundPrefix {
 533     EVEX_RNE = 0x0,
 534     EVEX_RD  = 0x1,
 535     EVEX_RU  = 0x2,
 536     EVEX_RZ  = 0x3
 537   };
 538 
 539   enum VexSimdPrefix {
 540     VEX_SIMD_NONE = 0x0,
 541     VEX_SIMD_66   = 0x1,
 542     VEX_SIMD_F3   = 0x2,
 543     VEX_SIMD_F2   = 0x3
 544   };
 545 
 546   enum VexOpcode {
 547     VEX_OPCODE_NONE  = 0x0,
 548     VEX_OPCODE_0F    = 0x1,
 549     VEX_OPCODE_0F_38 = 0x2,
 550     VEX_OPCODE_0F_3A = 0x3,
 551     VEX_OPCODE_MASK  = 0x1F
 552   };
 553 
 554   enum AvxVectorLen {
 555     AVX_128bit = 0x0,
 556     AVX_256bit = 0x1,
 557     AVX_512bit = 0x2,
 558     AVX_NoVec  = 0x4
 559   };
 560 
 561   enum EvexTupleType {
 562     EVEX_FV   = 0,
 563     EVEX_HV   = 4,
 564     EVEX_FVM  = 6,
 565     EVEX_T1S  = 7,
 566     EVEX_T1F  = 11,
 567     EVEX_T2   = 13,
 568     EVEX_T4   = 15,
 569     EVEX_T8   = 17,
 570     EVEX_HVM  = 18,
 571     EVEX_QVM  = 19,
 572     EVEX_OVM  = 20,
 573     EVEX_M128 = 21,
 574     EVEX_DUP  = 22,
 575     EVEX_ETUP = 23
 576   };
 577 
 578   enum EvexInputSizeInBits {
 579     EVEX_8bit  = 0,
 580     EVEX_16bit = 1,
 581     EVEX_32bit = 2,
 582     EVEX_64bit = 3,
 583     EVEX_NObit = 4
 584   };
 585 
 586   enum WhichOperand {
 587     // input to locate_operand, and format code for relocations
 588     imm_operand  = 0,            // embedded 32-bit|64-bit immediate operand
 589     disp32_operand = 1,          // embedded 32-bit displacement or address
 590     call32_operand = 2,          // embedded 32-bit self-relative displacement
 591 #ifndef _LP64
 592     _WhichOperand_limit = 3
 593 #else
 594      narrow_oop_operand = 3,     // embedded 32-bit immediate narrow oop
 595     _WhichOperand_limit = 4
 596 #endif
 597   };
 598 
 599   // Comparison predicates for integral types & FP types when using SSE
 600   enum ComparisonPredicate {
 601     eq = 0,
 602     lt = 1,
 603     le = 2,
 604     _false = 3,
 605     neq = 4,
 606     nlt = 5,
 607     nle = 6,
 608     _true = 7
 609   };
 610 
 611   // Comparison predicates for FP types when using AVX
 612   // O means ordered. U is unordered. When using ordered, any NaN comparison is false. Otherwise, it is true.
 613   // S means signaling. Q means non-signaling. When signaling is true, instruction signals #IA on NaN.
 614   enum ComparisonPredicateFP {
 615     EQ_OQ = 0,
 616     LT_OS = 1,
 617     LE_OS = 2,
 618     UNORD_Q = 3,
 619     NEQ_UQ = 4,
 620     NLT_US = 5,
 621     NLE_US = 6,
 622     ORD_Q = 7,
 623     EQ_UQ = 8,
 624     NGE_US = 9,
 625     NGT_US = 0xA,
 626     FALSE_OQ = 0XB,
 627     NEQ_OQ = 0xC,
 628     GE_OS = 0xD,
 629     GT_OS = 0xE,
 630     TRUE_UQ = 0xF,
 631     EQ_OS = 0x10,
 632     LT_OQ = 0x11,
 633     LE_OQ = 0x12,
 634     UNORD_S = 0x13,
 635     NEQ_US = 0x14,
 636     NLT_UQ = 0x15,
 637     NLE_UQ = 0x16,
 638     ORD_S = 0x17,
 639     EQ_US = 0x18,
 640     NGE_UQ = 0x19,
 641     NGT_UQ = 0x1A,
 642     FALSE_OS = 0x1B,
 643     NEQ_OS = 0x1C,
 644     GE_OQ = 0x1D,
 645     GT_OQ = 0x1E,
 646     TRUE_US =0x1F
 647   };
 648 
 649   enum Width {
 650     B = 0,
 651     W = 1,
 652     D = 2,
 653     Q = 3
 654   };
 655 
 656   //---<  calculate length of instruction  >---
 657   // As instruction size can't be found out easily on x86/x64,
 658   // we just use '4' for len and maxlen.
 659   // instruction must start at passed address
 660   static unsigned int instr_len(unsigned char *instr) { return 4; }
 661 
 662   //---<  longest instructions  >---
 663   // Max instruction length is not specified in architecture documentation.
 664   // We could use a "safe enough" estimate (15), but just default to
 665   // instruction length guess from above.
 666   static unsigned int instr_maxlen() { return 4; }
 667 
 668   // NOTE: The general philopsophy of the declarations here is that 64bit versions
 669   // of instructions are freely declared without the need for wrapping them an ifdef.
 670   // (Some dangerous instructions are ifdef's out of inappropriate jvm's.)
 671   // In the .cpp file the implementations are wrapped so that they are dropped out
 672   // of the resulting jvm. This is done mostly to keep the footprint of MINIMAL
 673   // to the size it was prior to merging up the 32bit and 64bit assemblers.
 674   //
 675   // This does mean you'll get a linker/runtime error if you use a 64bit only instruction
 676   // in a 32bit vm. This is somewhat unfortunate but keeps the ifdef noise down.
 677 
 678 private:
 679 
 680   bool _legacy_mode_bw;
 681   bool _legacy_mode_dq;
 682   bool _legacy_mode_vl;
 683   bool _legacy_mode_vlbw;
 684   NOT_LP64(bool _is_managed;)
 685 
 686   InstructionAttr *_attributes;
 687   void set_attributes(InstructionAttr* attributes);
 688 
 689   // 64bit prefixes
 690   void prefix(Register reg);
 691   void prefix(Register dst, Register src, Prefix p);
 692   void prefix(Register dst, Address adr, Prefix p);
 693 
 694   void prefix(Address adr);
 695   void prefix(Address adr, Register reg,  bool byteinst = false);
 696   void prefix(Address adr, XMMRegister reg);
 697 
 698   int prefix_and_encode(int reg_enc, bool byteinst = false);
 699   int prefix_and_encode(int dst_enc, int src_enc) {
 700     return prefix_and_encode(dst_enc, false, src_enc, false);
 701   }
 702   int prefix_and_encode(int dst_enc, bool dst_is_byte, int src_enc, bool src_is_byte);
 703 
 704   // Some prefixq variants always emit exactly one prefix byte, so besides a
 705   // prefix-emitting method we provide a method to get the prefix byte to emit,
 706   // which can then be folded into a byte stream.
 707   int8_t get_prefixq(Address adr);
 708   int8_t get_prefixq(Address adr, Register reg);
 709 
 710   void prefixq(Address adr);
 711   void prefixq(Address adr, Register reg);
 712   void prefixq(Address adr, XMMRegister reg);
 713 
 714   int prefixq_and_encode(int reg_enc);
 715   int prefixq_and_encode(int dst_enc, int src_enc);
 716 
 717   void rex_prefix(Address adr, XMMRegister xreg,
 718                   VexSimdPrefix pre, VexOpcode opc, bool rex_w);
 719   int  rex_prefix_and_encode(int dst_enc, int src_enc,
 720                              VexSimdPrefix pre, VexOpcode opc, bool rex_w);
 721 
 722   void vex_prefix(bool vex_r, bool vex_b, bool vex_x, int nds_enc, VexSimdPrefix pre, VexOpcode opc);
 723 
 724   void evex_prefix(bool vex_r, bool vex_b, bool vex_x, bool evex_r, bool evex_v,
 725                    int nds_enc, VexSimdPrefix pre, VexOpcode opc);
 726 
 727   void vex_prefix(Address adr, int nds_enc, int xreg_enc,
 728                   VexSimdPrefix pre, VexOpcode opc,
 729                   InstructionAttr *attributes);
 730 
 731   int  vex_prefix_and_encode(int dst_enc, int nds_enc, int src_enc,
 732                              VexSimdPrefix pre, VexOpcode opc,
 733                              InstructionAttr *attributes);
 734 
 735   void simd_prefix(XMMRegister xreg, XMMRegister nds, Address adr, VexSimdPrefix pre,
 736                    VexOpcode opc, InstructionAttr *attributes);
 737 
 738   int simd_prefix_and_encode(XMMRegister dst, XMMRegister nds, XMMRegister src, VexSimdPrefix pre,
 739                              VexOpcode opc, InstructionAttr *attributes);
 740 
 741   // Helper functions for groups of instructions
 742   void emit_arith_b(int op1, int op2, Register dst, int imm8);
 743 
 744   void emit_arith(int op1, int op2, Register dst, int32_t imm32);
 745   // Force generation of a 4 byte immediate value even if it fits into 8bit
 746   void emit_arith_imm32(int op1, int op2, Register dst, int32_t imm32);
 747   void emit_arith(int op1, int op2, Register dst, Register src);
 748 
 749   bool emit_compressed_disp_byte(int &disp);
 750 
 751   void emit_modrm(int mod, int dst_enc, int src_enc);
 752   void emit_modrm_disp8(int mod, int dst_enc, int src_enc,
 753                         int disp);
 754   void emit_modrm_sib(int mod, int dst_enc, int src_enc,
 755                       Address::ScaleFactor scale, int index_enc, int base_enc);
 756   void emit_modrm_sib_disp8(int mod, int dst_enc, int src_enc,
 757                             Address::ScaleFactor scale, int index_enc, int base_enc,
 758                             int disp);
 759 
 760   void emit_operand_helper(int reg_enc,
 761                            int base_enc, int index_enc, Address::ScaleFactor scale,
 762                            int disp,
 763                            RelocationHolder const& rspec,
 764                            int post_addr_length);
 765 
 766   void emit_operand(Register reg,
 767                     Register base, Register index, Address::ScaleFactor scale,
 768                     int disp,
 769                     RelocationHolder const& rspec,
 770                     int post_addr_length);
 771 
 772   void emit_operand(Register reg,
 773                     Register base, XMMRegister index, Address::ScaleFactor scale,
 774                     int disp,
 775                     RelocationHolder const& rspec,
 776                     int post_addr_length);
 777 
 778   void emit_operand(XMMRegister xreg,
 779                     Register base, XMMRegister xindex, Address::ScaleFactor scale,
 780                     int disp,
 781                     RelocationHolder const& rspec,
 782                     int post_addr_length);
 783 
 784   void emit_operand(Register reg, Address adr,
 785                     int post_addr_length);
 786 
 787   void emit_operand(XMMRegister reg,
 788                     Register base, Register index, Address::ScaleFactor scale,
 789                     int disp,
 790                     RelocationHolder const& rspec,
 791                     int post_addr_length);
 792 
 793   void emit_operand_helper(KRegister kreg,
 794                            int base_enc, int index_enc, Address::ScaleFactor scale,
 795                            int disp,
 796                            RelocationHolder const& rspec,
 797                            int post_addr_length);
 798 
 799   void emit_operand(KRegister kreg, Address adr,
 800                     int post_addr_length);
 801 
 802   void emit_operand(KRegister kreg,
 803                     Register base, Register index, Address::ScaleFactor scale,
 804                     int disp,
 805                     RelocationHolder const& rspec,
 806                     int post_addr_length);
 807 
 808   void emit_operand(XMMRegister reg, Address adr, int post_addr_length);
 809 
 810   // Immediate-to-memory forms
 811   void emit_arith_operand(int op1, Register rm, Address adr, int32_t imm32);
 812   void emit_arith_operand_imm32(int op1, Register rm, Address adr, int32_t imm32);
 813 
 814  protected:
 815 #ifdef ASSERT
 816   void check_relocation(RelocationHolder const& rspec, int format);
 817 #endif
 818 
 819   void emit_data(jint data, relocInfo::relocType    rtype, int format);
 820   void emit_data(jint data, RelocationHolder const& rspec, int format);
 821   void emit_data64(jlong data, relocInfo::relocType rtype, int format = 0);
 822   void emit_data64(jlong data, RelocationHolder const& rspec, int format = 0);
 823 
 824   bool always_reachable(AddressLiteral adr) NOT_LP64( { return true; } );
 825   bool        reachable(AddressLiteral adr) NOT_LP64( { return true; } );
 826 
 827 
 828   // These are all easily abused and hence protected
 829 
 830  public:
 831   // 32BIT ONLY SECTION
 832 #ifndef _LP64
 833   // Make these disappear in 64bit mode since they would never be correct
 834   void cmp_literal32(Register src1, int32_t imm32, RelocationHolder const& rspec);   // 32BIT ONLY
 835   void cmp_literal32(Address src1, int32_t imm32, RelocationHolder const& rspec);    // 32BIT ONLY
 836 
 837   void mov_literal32(Register dst, int32_t imm32, RelocationHolder const& rspec);    // 32BIT ONLY
 838   void mov_literal32(Address dst, int32_t imm32, RelocationHolder const& rspec);     // 32BIT ONLY
 839 
 840   void push_literal32(int32_t imm32, RelocationHolder const& rspec);                 // 32BIT ONLY
 841 #else
 842   // 64BIT ONLY SECTION
 843   void mov_literal64(Register dst, intptr_t imm64, RelocationHolder const& rspec);   // 64BIT ONLY
 844 
 845   void cmp_narrow_oop(Register src1, int32_t imm32, RelocationHolder const& rspec);
 846   void cmp_narrow_oop(Address src1, int32_t imm32, RelocationHolder const& rspec);
 847 
 848   void mov_narrow_oop(Register dst, int32_t imm32, RelocationHolder const& rspec);
 849   void mov_narrow_oop(Address dst, int32_t imm32, RelocationHolder const& rspec);
 850 #endif // _LP64
 851 
 852  protected:
 853   // These are unique in that we are ensured by the caller that the 32bit
 854   // relative in these instructions will always be able to reach the potentially
 855   // 64bit address described by entry. Since they can take a 64bit address they
 856   // don't have the 32 suffix like the other instructions in this class.
 857 
 858   void call_literal(address entry, RelocationHolder const& rspec);
 859   void jmp_literal(address entry, RelocationHolder const& rspec);
 860 
 861   // Avoid using directly section
 862   // Instructions in this section are actually usable by anyone without danger
 863   // of failure but have performance issues that are addressed my enhanced
 864   // instructions which will do the proper thing base on the particular cpu.
 865   // We protect them because we don't trust you...
 866 
 867   // Don't use next inc() and dec() methods directly. INC & DEC instructions
 868   // could cause a partial flag stall since they don't set CF flag.
 869   // Use MacroAssembler::decrement() & MacroAssembler::increment() methods
 870   // which call inc() & dec() or add() & sub() in accordance with
 871   // the product flag UseIncDec value.
 872 
 873   void decl(Register dst);
 874   void decl(Address dst);
 875   void decq(Address dst);
 876 
 877   void incl(Register dst);
 878   void incl(Address dst);
 879   void incq(Register dst);
 880   void incq(Address dst);
 881 
 882   // New cpus require use of movsd and movss to avoid partial register stall
 883   // when loading from memory. But for old Opteron use movlpd instead of movsd.
 884   // The selection is done in MacroAssembler::movdbl() and movflt().
 885 
 886   // Move Scalar Single-Precision Floating-Point Values
 887   void movss(XMMRegister dst, Address src);
 888   void movss(XMMRegister dst, XMMRegister src);
 889   void movss(Address dst, XMMRegister src);
 890 
 891   // Move Scalar Double-Precision Floating-Point Values
 892   void movsd(XMMRegister dst, Address src);
 893   void movsd(XMMRegister dst, XMMRegister src);
 894   void movsd(Address dst, XMMRegister src);
 895   void movlpd(XMMRegister dst, Address src);
 896 
 897   void vmovsd(XMMRegister dst, XMMRegister src, XMMRegister src2);
 898 
 899   // New cpus require use of movaps and movapd to avoid partial register stall
 900   // when moving between registers.
 901   void movaps(XMMRegister dst, XMMRegister src);
 902   void movapd(XMMRegister dst, XMMRegister src);
 903 
 904   // End avoid using directly
 905 
 906 
 907   // Instruction prefixes
 908   void prefix(Prefix p);
 909 
 910   public:
 911 
 912   // Creation
 913   Assembler(CodeBuffer* code) : AbstractAssembler(code) {
 914     init_attributes();
 915   }
 916 
 917   // Decoding
 918   static address locate_operand(address inst, WhichOperand which);
 919   static address locate_next_instruction(address inst);
 920 
 921   // Utilities
 922   static bool query_compressed_disp_byte(int disp, bool is_evex_inst, int vector_len,
 923                                          int cur_tuple_type, int in_size_in_bits, int cur_encoding);
 924 
 925   // Generic instructions
 926   // Does 32bit or 64bit as needed for the platform. In some sense these
 927   // belong in macro assembler but there is no need for both varieties to exist
 928 
 929   void init_attributes(void);
 930   void clear_attributes(void) { _attributes = nullptr; }
 931 
 932   void set_managed(void) { NOT_LP64(_is_managed = true;) }
 933   void clear_managed(void) { NOT_LP64(_is_managed = false;) }
 934   bool is_managed(void) {
 935     NOT_LP64(return _is_managed;)
 936     LP64_ONLY(return false;) }
 937 
 938   void lea(Register dst, Address src);
 939 
 940   void mov(Register dst, Register src);
 941 
 942 #ifdef _LP64
 943   // support caching the result of some routines
 944 
 945   // must be called before pusha(), popa(), vzeroupper() - checked with asserts
 946   static void precompute_instructions();
 947 
 948   void pusha_uncached();
 949   void popa_uncached();
 950 #endif
 951   void vzeroupper_uncached();
 952   void decq(Register dst);
 953 
 954   void pusha();
 955   void popa();
 956 
 957   void pushf();
 958   void popf();
 959 
 960   void push(int32_t imm32);
 961 
 962   void push(Register src);
 963 
 964   void pop(Register dst);
 965 
 966   // These do register sized moves/scans
 967   void rep_mov();
 968   void rep_stos();
 969   void rep_stosb();
 970   void repne_scan();
 971 #ifdef _LP64
 972   void repne_scanl();
 973 #endif
 974 
 975   // Vanilla instructions in lexical order
 976 
 977   void adcl(Address dst, int32_t imm32);
 978   void adcl(Address dst, Register src);
 979   void adcl(Register dst, int32_t imm32);
 980   void adcl(Register dst, Address src);
 981   void adcl(Register dst, Register src);
 982 
 983   void adcq(Register dst, int32_t imm32);
 984   void adcq(Register dst, Address src);
 985   void adcq(Register dst, Register src);
 986 
 987   void addb(Address dst, int imm8);
 988   void addb(Address dst, Register src);
 989   void addb(Register dst, int imm8);
 990   void addw(Register dst, Register src);
 991   void addw(Address dst, int imm16);
 992   void addw(Address dst, Register src);
 993 
 994   void addl(Address dst, int32_t imm32);
 995   void addl(Address dst, Register src);
 996   void addl(Register dst, int32_t imm32);
 997   void addl(Register dst, Address src);
 998   void addl(Register dst, Register src);
 999 
1000   void addq(Address dst, int32_t imm32);
1001   void addq(Address dst, Register src);
1002   void addq(Register dst, int32_t imm32);
1003   void addq(Register dst, Address src);
1004   void addq(Register dst, Register src);
1005 
1006 #ifdef _LP64
1007  //Add Unsigned Integers with Carry Flag
1008   void adcxq(Register dst, Register src);
1009 
1010  //Add Unsigned Integers with Overflow Flag
1011   void adoxq(Register dst, Register src);
1012 #endif
1013 
1014   void addr_nop_4();
1015   void addr_nop_5();
1016   void addr_nop_7();
1017   void addr_nop_8();
1018 
1019   // Add Scalar Double-Precision Floating-Point Values
1020   void addsd(XMMRegister dst, Address src);
1021   void addsd(XMMRegister dst, XMMRegister src);
1022 
1023   // Add Scalar Single-Precision Floating-Point Values
1024   void addss(XMMRegister dst, Address src);
1025   void addss(XMMRegister dst, XMMRegister src);
1026 
1027   // AES instructions
1028   void aesdec(XMMRegister dst, Address src);
1029   void aesdec(XMMRegister dst, XMMRegister src);
1030   void aesdeclast(XMMRegister dst, Address src);
1031   void aesdeclast(XMMRegister dst, XMMRegister src);
1032   void aesenc(XMMRegister dst, Address src);
1033   void aesenc(XMMRegister dst, XMMRegister src);
1034   void aesenclast(XMMRegister dst, Address src);
1035   void aesenclast(XMMRegister dst, XMMRegister src);
1036   // Vector AES instructions
1037   void vaesenc(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1038   void vaesenclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1039   void vaesdec(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1040   void vaesdeclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1041 
1042   void andw(Register dst, Register src);
1043   void andb(Address dst, Register src);
1044 
1045   void andl(Address  dst, int32_t imm32);
1046   void andl(Register dst, int32_t imm32);
1047   void andl(Register dst, Address src);
1048   void andl(Register dst, Register src);
1049   void andl(Address dst, Register src);
1050 
1051   void andq(Address  dst, int32_t imm32);
1052   void andq(Register dst, int32_t imm32);
1053   void andq(Register dst, Address src);
1054   void andq(Register dst, Register src);
1055   void andq(Address dst, Register src);
1056 
1057   // BMI instructions
1058   void andnl(Register dst, Register src1, Register src2);
1059   void andnl(Register dst, Register src1, Address src2);
1060   void andnq(Register dst, Register src1, Register src2);
1061   void andnq(Register dst, Register src1, Address src2);
1062 
1063   void blsil(Register dst, Register src);
1064   void blsil(Register dst, Address src);
1065   void blsiq(Register dst, Register src);
1066   void blsiq(Register dst, Address src);
1067 
1068   void blsmskl(Register dst, Register src);
1069   void blsmskl(Register dst, Address src);
1070   void blsmskq(Register dst, Register src);
1071   void blsmskq(Register dst, Address src);
1072 
1073   void blsrl(Register dst, Register src);
1074   void blsrl(Register dst, Address src);
1075   void blsrq(Register dst, Register src);
1076   void blsrq(Register dst, Address src);
1077 
1078   void bsfl(Register dst, Register src);
1079   void bsrl(Register dst, Register src);
1080 
1081 #ifdef _LP64
1082   void bsfq(Register dst, Register src);
1083   void bsrq(Register dst, Register src);
1084 #endif
1085 
1086   void bswapl(Register reg);
1087 
1088   void bswapq(Register reg);
1089 
1090   void call(Label& L, relocInfo::relocType rtype);
1091   void call(Register reg);  // push pc; pc <- reg
1092   void call(Address adr);   // push pc; pc <- adr
1093 
1094   void cdql();
1095 
1096   void cdqq();
1097 
1098   void cld();
1099 
1100   void clflush(Address adr);
1101   void clflushopt(Address adr);
1102   void clwb(Address adr);
1103 
1104   void cmovl(Condition cc, Register dst, Register src);
1105   void cmovl(Condition cc, Register dst, Address src);
1106 
1107   void cmovq(Condition cc, Register dst, Register src);
1108   void cmovq(Condition cc, Register dst, Address src);
1109 
1110 
1111   void cmpb(Address dst, int imm8);
1112 
1113   void cmpl(Address dst, int32_t imm32);
1114   void cmpl(Register dst, int32_t imm32);
1115   void cmpl(Register dst, Register src);
1116   void cmpl(Register dst, Address src);
1117   void cmpl_imm32(Address dst, int32_t imm32);
1118 
1119   void cmpq(Address dst, int32_t imm32);
1120   void cmpq(Address dst, Register src);
1121   void cmpq(Register dst, int32_t imm32);
1122   void cmpq(Register dst, Register src);
1123   void cmpq(Register dst, Address src);
1124 
1125   void cmpw(Address dst, int imm16);
1126 
1127   void cmpxchg8 (Address adr);
1128 
1129   void cmpxchgb(Register reg, Address adr);
1130   void cmpxchgl(Register reg, Address adr);
1131 
1132   void cmpxchgq(Register reg, Address adr);
1133   void cmpxchgw(Register reg, Address adr);
1134 
1135   // Ordered Compare Scalar Double-Precision Floating-Point Values and set EFLAGS
1136   void comisd(XMMRegister dst, Address src);
1137   void comisd(XMMRegister dst, XMMRegister src);
1138 
1139   // Ordered Compare Scalar Single-Precision Floating-Point Values and set EFLAGS
1140   void comiss(XMMRegister dst, Address src);
1141   void comiss(XMMRegister dst, XMMRegister src);
1142 
1143   // Identify processor type and features
1144   void cpuid();
1145 
1146   // CRC32C
1147   void crc32(Register crc, Register v, int8_t sizeInBytes);
1148   void crc32(Register crc, Address adr, int8_t sizeInBytes);
1149 
1150   // Convert Scalar Double-Precision Floating-Point Value to Scalar Single-Precision Floating-Point Value
1151   void cvtsd2ss(XMMRegister dst, XMMRegister src);
1152   void cvtsd2ss(XMMRegister dst, Address src);
1153 
1154   // Convert Doubleword Integer to Scalar Double-Precision Floating-Point Value
1155   void cvtsi2sdl(XMMRegister dst, Register src);
1156   void cvtsi2sdl(XMMRegister dst, Address src);
1157   void cvtsi2sdq(XMMRegister dst, Register src);
1158   void cvtsi2sdq(XMMRegister dst, Address src);
1159 
1160   // Convert Doubleword Integer to Scalar Single-Precision Floating-Point Value
1161   void cvtsi2ssl(XMMRegister dst, Register src);
1162   void cvtsi2ssl(XMMRegister dst, Address src);
1163   void cvtsi2ssq(XMMRegister dst, Register src);
1164   void cvtsi2ssq(XMMRegister dst, Address src);
1165 
1166   // Convert Packed Signed Doubleword Integers to Packed Double-Precision Floating-Point Value
1167   void cvtdq2pd(XMMRegister dst, XMMRegister src);
1168   void vcvtdq2pd(XMMRegister dst, XMMRegister src, int vector_len);
1169 
1170   // Convert Halffloat to Single Precision Floating-Point value
1171   void vcvtps2ph(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1172   void vcvtph2ps(XMMRegister dst, XMMRegister src, int vector_len);
1173   void evcvtps2ph(Address dst, KRegister mask, XMMRegister src, int imm8, int vector_len);
1174   void vcvtps2ph(Address dst, XMMRegister src, int imm8, int vector_len);
1175   void vcvtph2ps(XMMRegister dst, Address src, int vector_len);
1176 
1177   // Convert Packed Signed Doubleword Integers to Packed Single-Precision Floating-Point Value
1178   void cvtdq2ps(XMMRegister dst, XMMRegister src);
1179   void vcvtdq2ps(XMMRegister dst, XMMRegister src, int vector_len);
1180 
1181   // Convert Scalar Single-Precision Floating-Point Value to Scalar Double-Precision Floating-Point Value
1182   void cvtss2sd(XMMRegister dst, XMMRegister src);
1183   void cvtss2sd(XMMRegister dst, Address src);
1184 
1185   // Convert with Truncation Scalar Double-Precision Floating-Point Value to Doubleword Integer
1186   void cvtsd2siq(Register dst, XMMRegister src);
1187   void cvttsd2sil(Register dst, Address src);
1188   void cvttsd2sil(Register dst, XMMRegister src);
1189   void cvttsd2siq(Register dst, Address src);
1190   void cvttsd2siq(Register dst, XMMRegister src);
1191 
1192   // Convert with Truncation Scalar Single-Precision Floating-Point Value to Doubleword Integer
1193   void cvttss2sil(Register dst, XMMRegister src);
1194   void cvttss2siq(Register dst, XMMRegister src);
1195   void cvtss2sil(Register dst, XMMRegister src);
1196 
1197   // Convert vector double to int
1198   void cvttpd2dq(XMMRegister dst, XMMRegister src);
1199 
1200   // Convert vector float and double
1201   void vcvtps2pd(XMMRegister dst, XMMRegister src, int vector_len);
1202   void vcvtpd2ps(XMMRegister dst, XMMRegister src, int vector_len);
1203 
1204   // Convert vector float to int/long
1205   void vcvtps2dq(XMMRegister dst, XMMRegister src, int vector_len);
1206   void vcvttps2dq(XMMRegister dst, XMMRegister src, int vector_len);
1207   void evcvttps2qq(XMMRegister dst, XMMRegister src, int vector_len);
1208 
1209   // Convert vector long to vector FP
1210   void evcvtqq2ps(XMMRegister dst, XMMRegister src, int vector_len);
1211   void evcvtqq2pd(XMMRegister dst, XMMRegister src, int vector_len);
1212 
1213   // Convert vector double to long
1214   void evcvtpd2qq(XMMRegister dst, XMMRegister src, int vector_len);
1215   void evcvttpd2qq(XMMRegister dst, XMMRegister src, int vector_len);
1216 
1217   // Convert vector double to int
1218   void vcvttpd2dq(XMMRegister dst, XMMRegister src, int vector_len);
1219 
1220   // Evex casts with truncation
1221   void evpmovwb(XMMRegister dst, XMMRegister src, int vector_len);
1222   void evpmovdw(XMMRegister dst, XMMRegister src, int vector_len);
1223   void evpmovdb(XMMRegister dst, XMMRegister src, int vector_len);
1224   void evpmovqd(XMMRegister dst, XMMRegister src, int vector_len);
1225   void evpmovqb(XMMRegister dst, XMMRegister src, int vector_len);
1226   void evpmovqw(XMMRegister dst, XMMRegister src, int vector_len);
1227 
1228   // Evex casts with signed saturation
1229   void evpmovsqd(XMMRegister dst, XMMRegister src, int vector_len);
1230 
1231   //Abs of packed Integer values
1232   void pabsb(XMMRegister dst, XMMRegister src);
1233   void pabsw(XMMRegister dst, XMMRegister src);
1234   void pabsd(XMMRegister dst, XMMRegister src);
1235   void vpabsb(XMMRegister dst, XMMRegister src, int vector_len);
1236   void vpabsw(XMMRegister dst, XMMRegister src, int vector_len);
1237   void vpabsd(XMMRegister dst, XMMRegister src, int vector_len);
1238   void evpabsq(XMMRegister dst, XMMRegister src, int vector_len);
1239 
1240   // Divide Scalar Double-Precision Floating-Point Values
1241   void divsd(XMMRegister dst, Address src);
1242   void divsd(XMMRegister dst, XMMRegister src);
1243 
1244   // Divide Scalar Single-Precision Floating-Point Values
1245   void divss(XMMRegister dst, Address src);
1246   void divss(XMMRegister dst, XMMRegister src);
1247 
1248 
1249   void fnstsw_ax();
1250   void fprem();
1251   void fld_d(Address adr);
1252   void fstp_d(Address adr);
1253   void fstp_d(int index);
1254 
1255  private:
1256 
1257   void emit_farith(int b1, int b2, int i);
1258 
1259  public:
1260 #ifndef _LP64
1261   void emms();
1262 
1263   void fabs();
1264 
1265   void fadd(int i);
1266 
1267   void fadd_d(Address src);
1268   void fadd_s(Address src);
1269 
1270   // "Alternate" versions of x87 instructions place result down in FPU
1271   // stack instead of on TOS
1272 
1273   void fadda(int i); // "alternate" fadd
1274   void faddp(int i = 1);
1275 
1276   void fchs();
1277 
1278   void fcom(int i);
1279 
1280   void fcomp(int i = 1);
1281   void fcomp_d(Address src);
1282   void fcomp_s(Address src);
1283 
1284   void fcompp();
1285 
1286   void fcos();
1287 
1288   void fdecstp();
1289 
1290   void fdiv(int i);
1291   void fdiv_d(Address src);
1292   void fdivr_s(Address src);
1293   void fdiva(int i);  // "alternate" fdiv
1294   void fdivp(int i = 1);
1295 
1296   void fdivr(int i);
1297   void fdivr_d(Address src);
1298   void fdiv_s(Address src);
1299 
1300   void fdivra(int i); // "alternate" reversed fdiv
1301 
1302   void fdivrp(int i = 1);
1303 
1304   void ffree(int i = 0);
1305 
1306   void fild_d(Address adr);
1307   void fild_s(Address adr);
1308 
1309   void fincstp();
1310 
1311   void finit();
1312 
1313   void fist_s (Address adr);
1314   void fistp_d(Address adr);
1315   void fistp_s(Address adr);
1316 
1317   void fld1();
1318 
1319   void fld_s(Address adr);
1320   void fld_s(int index);
1321 
1322   void fldcw(Address src);
1323 
1324   void fldenv(Address src);
1325 
1326   void fldlg2();
1327 
1328   void fldln2();
1329 
1330   void fldz();
1331 
1332   void flog();
1333   void flog10();
1334 
1335   void fmul(int i);
1336 
1337   void fmul_d(Address src);
1338   void fmul_s(Address src);
1339 
1340   void fmula(int i);  // "alternate" fmul
1341 
1342   void fmulp(int i = 1);
1343 
1344   void fnsave(Address dst);
1345 
1346   void fnstcw(Address src);
1347   void fprem1();
1348 
1349   void frstor(Address src);
1350 
1351   void fsin();
1352 
1353   void fsqrt();
1354 
1355   void fst_d(Address adr);
1356   void fst_s(Address adr);
1357 
1358   void fstp_s(Address adr);
1359 
1360   void fsub(int i);
1361   void fsub_d(Address src);
1362   void fsub_s(Address src);
1363 
1364   void fsuba(int i);  // "alternate" fsub
1365 
1366   void fsubp(int i = 1);
1367 
1368   void fsubr(int i);
1369   void fsubr_d(Address src);
1370   void fsubr_s(Address src);
1371 
1372   void fsubra(int i); // "alternate" reversed fsub
1373 
1374   void fsubrp(int i = 1);
1375 
1376   void ftan();
1377 
1378   void ftst();
1379 
1380   void fucomi(int i = 1);
1381   void fucomip(int i = 1);
1382 
1383   void fwait();
1384 
1385   void fxch(int i = 1);
1386 
1387   void fyl2x();
1388   void frndint();
1389   void f2xm1();
1390   void fldl2e();
1391 #endif // !_LP64
1392 
1393   // operands that only take the original 32bit registers
1394   void emit_operand32(Register reg, Address adr, int post_addr_length);
1395 
1396   void fld_x(Address adr);  // extended-precision (80-bit) format
1397   void fstp_x(Address adr); // extended-precision (80-bit) format
1398   void fxrstor(Address src);
1399   void xrstor(Address src);
1400 
1401   void fxsave(Address dst);
1402   void xsave(Address dst);
1403 
1404   void hlt();
1405 
1406   void idivl(Register src);
1407   void divl(Register src); // Unsigned division
1408 
1409 #ifdef _LP64
1410   void idivq(Register src);
1411   void divq(Register src); // Unsigned division
1412 #endif
1413 
1414   void imull(Register src);
1415   void imull(Register dst, Register src);
1416   void imull(Register dst, Register src, int value);
1417   void imull(Register dst, Address src, int value);
1418   void imull(Register dst, Address src);
1419 
1420 #ifdef _LP64
1421   void imulq(Register dst, Register src);
1422   void imulq(Register dst, Register src, int value);
1423   void imulq(Register dst, Address src, int value);
1424   void imulq(Register dst, Address src);
1425   void imulq(Register dst);
1426 #endif
1427 
1428   // jcc is the generic conditional branch generator to run-
1429   // time routines, jcc is used for branches to labels. jcc
1430   // takes a branch opcode (cc) and a label (L) and generates
1431   // either a backward branch or a forward branch and links it
1432   // to the label fixup chain. Usage:
1433   //
1434   // Label L;      // unbound label
1435   // jcc(cc, L);   // forward branch to unbound label
1436   // bind(L);      // bind label to the current pc
1437   // jcc(cc, L);   // backward branch to bound label
1438   // bind(L);      // illegal: a label may be bound only once
1439   //
1440   // Note: The same Label can be used for forward and backward branches
1441   // but it may be bound only once.
1442 
1443   void jcc(Condition cc, Label& L, bool maybe_short = true);
1444 
1445   // Conditional jump to a 8-bit offset to L.
1446   // WARNING: be very careful using this for forward jumps.  If the label is
1447   // not bound within an 8-bit offset of this instruction, a run-time error
1448   // will occur.
1449 
1450   // Use macro to record file and line number.
1451   #define jccb(cc, L) jccb_0(cc, L, __FILE__, __LINE__)
1452 
1453   void jccb_0(Condition cc, Label& L, const char* file, int line);
1454 
1455   void jmp(Address entry);    // pc <- entry
1456 
1457   // Label operations & relative jumps (PPUM Appendix D)
1458   void jmp(Label& L, bool maybe_short = true);   // unconditional jump to L
1459 
1460   void jmp(Register entry); // pc <- entry
1461 
1462   // Unconditional 8-bit offset jump to L.
1463   // WARNING: be very careful using this for forward jumps.  If the label is
1464   // not bound within an 8-bit offset of this instruction, a run-time error
1465   // will occur.
1466 
1467   // Use macro to record file and line number.
1468   #define jmpb(L) jmpb_0(L, __FILE__, __LINE__)
1469 
1470   void jmpb_0(Label& L, const char* file, int line);
1471 
1472   void ldmxcsr( Address src );
1473 
1474   void leal(Register dst, Address src);
1475 
1476   void leaq(Register dst, Address src);
1477 
1478   void lfence();
1479 
1480   void lock();
1481   void size_prefix();
1482 
1483   void lzcntl(Register dst, Register src);
1484   void lzcntl(Register dst, Address src);
1485 
1486 #ifdef _LP64
1487   void lzcntq(Register dst, Register src);
1488   void lzcntq(Register dst, Address src);
1489 #endif
1490 
1491   enum Membar_mask_bits {
1492     StoreStore = 1 << 3,
1493     LoadStore  = 1 << 2,
1494     StoreLoad  = 1 << 1,
1495     LoadLoad   = 1 << 0
1496   };
1497 
1498   // Serializes memory and blows flags
1499   void membar(Membar_mask_bits order_constraint);
1500 
1501   void mfence();
1502   void sfence();
1503 
1504   // Moves
1505 
1506   void mov64(Register dst, int64_t imm64);
1507   void mov64(Register dst, int64_t imm64, relocInfo::relocType rtype, int format);
1508 
1509   void movb(Address dst, Register src);
1510   void movb(Address dst, int imm8);
1511   void movb(Register dst, Address src);
1512 
1513   void movddup(XMMRegister dst, XMMRegister src);
1514   void movddup(XMMRegister dst, Address src);
1515   void vmovddup(XMMRegister dst, Address src, int vector_len);
1516 
1517   void kandbl(KRegister dst, KRegister src1, KRegister src2);
1518   void kandwl(KRegister dst, KRegister src1, KRegister src2);
1519   void kanddl(KRegister dst, KRegister src1, KRegister src2);
1520   void kandql(KRegister dst, KRegister src1, KRegister src2);
1521 
1522   void korbl(KRegister dst, KRegister src1, KRegister src2);
1523   void korwl(KRegister dst, KRegister src1, KRegister src2);
1524   void kordl(KRegister dst, KRegister src1, KRegister src2);
1525   void korql(KRegister dst, KRegister src1, KRegister src2);
1526 
1527   void kxorbl(KRegister dst, KRegister src1, KRegister src2);
1528   void kxorwl(KRegister dst, KRegister src1, KRegister src2);
1529   void kxordl(KRegister dst, KRegister src1, KRegister src2);
1530   void kxorql(KRegister dst, KRegister src1, KRegister src2);
1531   void kmovbl(KRegister dst, Register src);
1532   void kmovbl(Register dst, KRegister src);
1533   void kmovbl(KRegister dst, KRegister src);
1534   void kmovwl(KRegister dst, Register src);
1535   void kmovwl(KRegister dst, Address src);
1536   void kmovwl(Register dst, KRegister src);
1537   void kmovwl(Address dst, KRegister src);
1538   void kmovwl(KRegister dst, KRegister src);
1539   void kmovdl(KRegister dst, Register src);
1540   void kmovdl(Register dst, KRegister src);
1541   void kmovql(KRegister dst, KRegister src);
1542   void kmovql(Address dst, KRegister src);
1543   void kmovql(KRegister dst, Address src);
1544   void kmovql(KRegister dst, Register src);
1545   void kmovql(Register dst, KRegister src);
1546 
1547   void knotbl(KRegister dst, KRegister src);
1548   void knotwl(KRegister dst, KRegister src);
1549   void knotdl(KRegister dst, KRegister src);
1550   void knotql(KRegister dst, KRegister src);
1551 
1552   void kortestbl(KRegister dst, KRegister src);
1553   void kortestwl(KRegister dst, KRegister src);
1554   void kortestdl(KRegister dst, KRegister src);
1555   void kortestql(KRegister dst, KRegister src);
1556 
1557   void kxnorbl(KRegister dst, KRegister src1, KRegister src2);
1558   void kshiftlbl(KRegister dst, KRegister src, int imm8);
1559   void kshiftlql(KRegister dst, KRegister src, int imm8);
1560   void kshiftrbl(KRegister dst, KRegister src, int imm8);
1561   void kshiftrwl(KRegister dst, KRegister src, int imm8);
1562   void kshiftrdl(KRegister dst, KRegister src, int imm8);
1563   void kshiftrql(KRegister dst, KRegister src, int imm8);
1564   void ktestq(KRegister src1, KRegister src2);
1565   void ktestd(KRegister src1, KRegister src2);
1566   void kunpckdql(KRegister dst, KRegister src1, KRegister src2);
1567 
1568 
1569   void ktestql(KRegister dst, KRegister src);
1570   void ktestdl(KRegister dst, KRegister src);
1571   void ktestwl(KRegister dst, KRegister src);
1572   void ktestbl(KRegister dst, KRegister src);
1573 
1574   void movdl(XMMRegister dst, Register src);
1575   void movdl(Register dst, XMMRegister src);
1576   void movdl(XMMRegister dst, Address src);
1577   void movdl(Address dst, XMMRegister src);
1578 
1579   // Move Double Quadword
1580   void movdq(XMMRegister dst, Register src);
1581   void movdq(Register dst, XMMRegister src);
1582 
1583   // Move Aligned Double Quadword
1584   void movdqa(XMMRegister dst, XMMRegister src);
1585   void movdqa(XMMRegister dst, Address src);
1586 
1587   // Move Unaligned Double Quadword
1588   void movdqu(Address     dst, XMMRegister src);
1589   void movdqu(XMMRegister dst, Address src);
1590   void movdqu(XMMRegister dst, XMMRegister src);
1591 
1592   // Move Unaligned 256bit Vector
1593   void vmovdqu(Address dst, XMMRegister src);
1594   void vmovdqu(XMMRegister dst, Address src);
1595   void vmovdqu(XMMRegister dst, XMMRegister src);
1596 
1597    // Move Unaligned 512bit Vector
1598   void evmovdqub(XMMRegister dst, XMMRegister src, int vector_len);
1599   void evmovdqub(XMMRegister dst, Address src, int vector_len);
1600   void evmovdqub(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1601   void evmovdqub(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1602   void evmovdqub(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1603 
1604   void evmovdquw(XMMRegister dst, Address src, int vector_len);
1605   void evmovdquw(Address dst, XMMRegister src, int vector_len);
1606   void evmovdquw(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1607   void evmovdquw(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1608   void evmovdquw(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1609 
1610   void evmovdqul(XMMRegister dst, XMMRegister src, int vector_len);
1611   void evmovdqul(XMMRegister dst, Address src, int vector_len);
1612   void evmovdqul(Address dst, XMMRegister src, int vector_len);
1613 
1614   void evmovdqul(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1615   void evmovdqul(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1616   void evmovdqul(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1617 
1618   void evmovntdquq(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1619   void evmovntdquq(Address dst, XMMRegister src, int vector_len);
1620 
1621   void evmovdquq(Address dst, XMMRegister src, int vector_len);
1622   void evmovdquq(XMMRegister dst, Address src, int vector_len);
1623   void evmovdquq(XMMRegister dst, XMMRegister src, int vector_len);
1624 
1625   void evmovdquq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1626   void evmovdquq(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1627   void evmovdquq(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1628 
1629   // Move lower 64bit to high 64bit in 128bit register
1630   void movlhps(XMMRegister dst, XMMRegister src);
1631 
1632   void movl(Register dst, int32_t imm32);
1633   void movl(Address dst, int32_t imm32);
1634   void movl(Register dst, Register src);
1635   void movl(Register dst, Address src);
1636   void movl(Address dst, Register src);
1637 
1638 #ifdef _LP64
1639   void movq(Register dst, Register src);
1640   void movq(Register dst, Address src);
1641   void movq(Address  dst, Register src);
1642   void movq(Address  dst, int32_t imm32);
1643   void movq(Register  dst, int32_t imm32);
1644 #endif
1645 
1646   // Move Quadword
1647   void movq(Address     dst, XMMRegister src);
1648   void movq(XMMRegister dst, Address src);
1649   void movq(XMMRegister dst, XMMRegister src);
1650   void movq(Register dst, XMMRegister src);
1651   void movq(XMMRegister dst, Register src);
1652 
1653   void movsbl(Register dst, Address src);
1654   void movsbl(Register dst, Register src);
1655 
1656 #ifdef _LP64
1657   void movsbq(Register dst, Address src);
1658   void movsbq(Register dst, Register src);
1659 
1660   // Move signed 32bit immediate to 64bit extending sign
1661   void movslq(Address  dst, int32_t imm64);
1662 
1663   void movslq(Register dst, Address src);
1664   void movslq(Register dst, Register src);
1665 #endif
1666 
1667   void movswl(Register dst, Address src);
1668   void movswl(Register dst, Register src);
1669 
1670 #ifdef _LP64
1671   void movswq(Register dst, Address src);
1672   void movswq(Register dst, Register src);
1673 #endif
1674 
1675   void movups(XMMRegister dst, Address src);
1676   void vmovups(XMMRegister dst, Address src, int vector_len);
1677   void movups(Address dst, XMMRegister src);
1678   void vmovups(Address dst, XMMRegister src, int vector_len);
1679 
1680   void movw(Address dst, int imm16);
1681   void movw(Register dst, Address src);
1682   void movw(Address dst, Register src);
1683 
1684   void movzbl(Register dst, Address src);
1685   void movzbl(Register dst, Register src);
1686 
1687 #ifdef _LP64
1688   void movzbq(Register dst, Address src);
1689   void movzbq(Register dst, Register src);
1690 #endif
1691 
1692   void movzwl(Register dst, Address src);
1693   void movzwl(Register dst, Register src);
1694 
1695 #ifdef _LP64
1696   void movzwq(Register dst, Address src);
1697   void movzwq(Register dst, Register src);
1698 #endif
1699 
1700   // Unsigned multiply with RAX destination register
1701   void mull(Address src);
1702   void mull(Register src);
1703 
1704 #ifdef _LP64
1705   void mulq(Address src);
1706   void mulq(Register src);
1707   void mulxq(Register dst1, Register dst2, Register src);
1708 #endif
1709 
1710   // Multiply Scalar Double-Precision Floating-Point Values
1711   void mulsd(XMMRegister dst, Address src);
1712   void mulsd(XMMRegister dst, XMMRegister src);
1713 
1714   // Multiply Scalar Single-Precision Floating-Point Values
1715   void mulss(XMMRegister dst, Address src);
1716   void mulss(XMMRegister dst, XMMRegister src);
1717 
1718   void negl(Register dst);
1719   void negl(Address dst);
1720 
1721 #ifdef _LP64
1722   void negq(Register dst);
1723   void negq(Address dst);
1724 #endif
1725 
1726   void nop(int i = 1);
1727 
1728   void notl(Register dst);
1729 
1730 #ifdef _LP64
1731   void notq(Register dst);
1732 
1733   void btsq(Address dst, int imm8);
1734   void btrq(Address dst, int imm8);
1735 #endif
1736 
1737   void orw(Register dst, Register src);
1738 
1739   void orl(Address dst, int32_t imm32);
1740   void orl(Register dst, int32_t imm32);
1741   void orl(Register dst, Address src);
1742   void orl(Register dst, Register src);
1743   void orl(Address dst, Register src);
1744 
1745   void orb(Address dst, int imm8);
1746   void orb(Address dst, Register src);
1747 
1748   void orq(Address dst, int32_t imm32);
1749   void orq(Address dst, Register src);
1750   void orq(Register dst, int32_t imm32);
1751   void orq_imm32(Register dst, int32_t imm32);
1752   void orq(Register dst, Address src);
1753   void orq(Register dst, Register src);
1754 
1755   // Pack with signed saturation
1756   void packsswb(XMMRegister dst, XMMRegister src);
1757   void vpacksswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1758   void packssdw(XMMRegister dst, XMMRegister src);
1759   void vpackssdw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1760 
1761   // Pack with unsigned saturation
1762   void packuswb(XMMRegister dst, XMMRegister src);
1763   void packuswb(XMMRegister dst, Address src);
1764   void packusdw(XMMRegister dst, XMMRegister src);
1765   void vpackuswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1766   void vpackusdw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1767 
1768   // Permutations
1769   void vpermq(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1770   void vpermq(XMMRegister dst, XMMRegister src, int imm8);
1771   void vpermq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1772   void vpermb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1773   void vpermb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
1774   void vpermw(XMMRegister dst,  XMMRegister nds, XMMRegister src, int vector_len);
1775   void vpermd(XMMRegister dst,  XMMRegister nds, Address src, int vector_len);
1776   void vpermd(XMMRegister dst,  XMMRegister nds, XMMRegister src, int vector_len);
1777   void vpermps(XMMRegister dst,  XMMRegister nds, XMMRegister src, int vector_len);
1778   void vperm2i128(XMMRegister dst,  XMMRegister nds, XMMRegister src, int imm8);
1779   void vperm2f128(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8);
1780   void vpermilps(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1781   void vpermilps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1782   void vpermilpd(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1783   void vpermpd(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1784   void evpermi2q(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1785   void evpermt2b(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1786   void evpmultishiftqb(XMMRegister dst, XMMRegister ctl, XMMRegister src, int vector_len);
1787 
1788   void pause();
1789 
1790   // Undefined Instruction
1791   void ud2();
1792 
1793   // SSE4.2 string instructions
1794   void pcmpestri(XMMRegister xmm1, XMMRegister xmm2, int imm8);
1795   void pcmpestri(XMMRegister xmm1, Address src, int imm8);
1796 
1797   void pcmpeqb(XMMRegister dst, XMMRegister src);
1798   void vpcmpCCbwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cond_encoding, int vector_len);
1799 
1800   void vpcmpeqb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1801   void evpcmpeqb(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1802   void evpcmpeqb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1803   void evpcmpeqb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1804 
1805   void vpcmpgtb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1806   void evpcmpgtb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1807   void evpcmpgtb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1808 
1809   void evpcmpuw(KRegister kdst, XMMRegister nds, XMMRegister src, ComparisonPredicate vcc, int vector_len);
1810   void evpcmpuw(KRegister kdst, XMMRegister nds, Address src, ComparisonPredicate vcc, int vector_len);
1811 
1812   void evpcmpuq(KRegister kdst, XMMRegister nds, XMMRegister src, ComparisonPredicate vcc, int vector_len);
1813 
1814   void pcmpeqw(XMMRegister dst, XMMRegister src);
1815   void vpcmpeqw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1816   void evpcmpeqw(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1817   void evpcmpeqw(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1818 
1819   void vpcmpgtw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1820 
1821   void pcmpeqd(XMMRegister dst, XMMRegister src);
1822   void vpcmpeqd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1823   void evpcmpeqd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src, int vector_len);
1824   void evpcmpeqd(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1825 
1826   void pcmpeqq(XMMRegister dst, XMMRegister src);
1827   void evpcmpeqq(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src, int vector_len);
1828   void vpcmpCCq(XMMRegister dst, XMMRegister nds, XMMRegister src, int cond_encoding, int vector_len);
1829   void vpcmpeqq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1830   void evpcmpeqq(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1831   void evpcmpeqq(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1832 
1833   void pcmpgtq(XMMRegister dst, XMMRegister src);
1834   void vpcmpgtq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1835 
1836   void pmovmskb(Register dst, XMMRegister src);
1837   void vpmovmskb(Register dst, XMMRegister src, int vec_enc);
1838   void vmovmskps(Register dst, XMMRegister src, int vec_enc);
1839   void vmovmskpd(Register dst, XMMRegister src, int vec_enc);
1840   void vpmaskmovd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
1841   void vpmaskmovq(XMMRegister dst, XMMRegister mask, Address src, int vector_len);
1842 
1843 
1844   void vmaskmovps(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
1845   void vmaskmovpd(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
1846   void vmaskmovps(Address dst, XMMRegister src, XMMRegister mask, int vector_len);
1847   void vmaskmovpd(Address dst, XMMRegister src, XMMRegister mask, int vector_len);
1848 
1849   // SSE 4.1 extract
1850   void pextrd(Register dst, XMMRegister src, int imm8);
1851   void pextrq(Register dst, XMMRegister src, int imm8);
1852   void pextrd(Address dst, XMMRegister src, int imm8);
1853   void pextrq(Address dst, XMMRegister src, int imm8);
1854   void pextrb(Register dst, XMMRegister src, int imm8);
1855   void pextrb(Address dst, XMMRegister src, int imm8);
1856   // SSE 2 extract
1857   void pextrw(Register dst, XMMRegister src, int imm8);
1858   void pextrw(Address dst, XMMRegister src, int imm8);
1859 
1860   // SSE 4.1 insert
1861   void pinsrd(XMMRegister dst, Register src, int imm8);
1862   void pinsrq(XMMRegister dst, Register src, int imm8);
1863   void pinsrb(XMMRegister dst, Register src, int imm8);
1864   void pinsrd(XMMRegister dst, Address src, int imm8);
1865   void pinsrq(XMMRegister dst, Address src, int imm8);
1866   void pinsrb(XMMRegister dst, Address src, int imm8);
1867   void insertps(XMMRegister dst, XMMRegister src, int imm8);
1868   // SSE 2 insert
1869   void pinsrw(XMMRegister dst, Register src, int imm8);
1870   void pinsrw(XMMRegister dst, Address src, int imm8);
1871 
1872   // AVX insert
1873   void vpinsrd(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1874   void vpinsrb(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1875   void vpinsrq(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1876   void vpinsrw(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1877   void vinsertps(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8);
1878 
1879   // Zero extend moves
1880   void pmovzxbw(XMMRegister dst, XMMRegister src);
1881   void pmovzxbw(XMMRegister dst, Address src);
1882   void pmovzxbd(XMMRegister dst, XMMRegister src);
1883   void vpmovzxbw(XMMRegister dst, Address src, int vector_len);
1884   void vpmovzxbw(XMMRegister dst, XMMRegister src, int vector_len);
1885   void vpmovzxbd(XMMRegister dst, XMMRegister src, int vector_len);
1886   void vpmovzxbq(XMMRegister dst, XMMRegister src, int vector_len);
1887   void vpmovzxwd(XMMRegister dst, XMMRegister src, int vector_len);
1888   void vpmovzxwq(XMMRegister dst, XMMRegister src, int vector_len);
1889   void pmovzxdq(XMMRegister dst, XMMRegister src);
1890   void vpmovzxdq(XMMRegister dst, XMMRegister src, int vector_len);
1891   void evpmovzxbw(XMMRegister dst, KRegister mask, Address src, int vector_len);
1892   void evpmovzxbd(XMMRegister dst, KRegister mask, Address src, int vector_len);
1893   void evpmovzxbd(XMMRegister dst, Address src, int vector_len);
1894 
1895   // Sign extend moves
1896   void pmovsxbd(XMMRegister dst, XMMRegister src);
1897   void pmovsxbq(XMMRegister dst, XMMRegister src);
1898   void pmovsxbw(XMMRegister dst, XMMRegister src);
1899   void pmovsxwd(XMMRegister dst, XMMRegister src);
1900   void vpmovsxbd(XMMRegister dst, XMMRegister src, int vector_len);
1901   void vpmovsxbq(XMMRegister dst, XMMRegister src, int vector_len);
1902   void vpmovsxbw(XMMRegister dst, XMMRegister src, int vector_len);
1903   void vpmovsxwd(XMMRegister dst, XMMRegister src, int vector_len);
1904   void vpmovsxwq(XMMRegister dst, XMMRegister src, int vector_len);
1905   void vpmovsxdq(XMMRegister dst, XMMRegister src, int vector_len);
1906 
1907   void evpmovwb(Address dst, XMMRegister src, int vector_len);
1908   void evpmovwb(Address dst, KRegister mask, XMMRegister src, int vector_len);
1909   void evpmovdb(Address dst, XMMRegister src, int vector_len);
1910 
1911   // Multiply add
1912   void pmaddwd(XMMRegister dst, XMMRegister src);
1913   void vpmaddwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1914   void vpmaddubsw(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
1915   void evpmadd52luq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
1916   void evpmadd52luq(XMMRegister dst, KRegister mask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
1917   void evpmadd52huq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
1918   void evpmadd52huq(XMMRegister dst, KRegister mask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
1919 
1920   // Multiply add accumulate
1921   void evpdpwssd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1922 
1923 #ifndef _LP64 // no 32bit push/pop on amd64
1924   void popl(Address dst);
1925 #endif
1926 
1927 #ifdef _LP64
1928   void popq(Address dst);
1929   void popq(Register dst);
1930 #endif
1931 
1932   void popcntl(Register dst, Address src);
1933   void popcntl(Register dst, Register src);
1934 
1935   void evpopcntb(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1936   void evpopcntw(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1937   void evpopcntd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1938   void evpopcntq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1939 
1940 #ifdef _LP64
1941   void popcntq(Register dst, Address src);
1942   void popcntq(Register dst, Register src);
1943 #endif
1944 
1945   // Prefetches (SSE, SSE2, 3DNOW only)
1946 
1947   void prefetchnta(Address src);
1948   void prefetchr(Address src);
1949   void prefetcht0(Address src);
1950   void prefetcht1(Address src);
1951   void prefetcht2(Address src);
1952   void prefetchw(Address src);
1953 
1954   // Shuffle Bytes
1955   void pshufb(XMMRegister dst, XMMRegister src);
1956   void pshufb(XMMRegister dst, Address src);
1957   void vpshufb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1958   void vpshufb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
1959   void evpshufb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
1960 
1961 
1962   // Shuffle Packed Doublewords
1963   void pshufd(XMMRegister dst, XMMRegister src, int mode);
1964   void pshufd(XMMRegister dst, Address src,     int mode);
1965   void vpshufd(XMMRegister dst, XMMRegister src, int mode, int vector_len);
1966 
1967   // Shuffle Packed High/Low Words
1968   void pshufhw(XMMRegister dst, XMMRegister src, int mode);
1969   void pshuflw(XMMRegister dst, XMMRegister src, int mode);
1970   void pshuflw(XMMRegister dst, Address src,     int mode);
1971   void vpshufhw(XMMRegister dst, XMMRegister src, int mode, int vector_len);
1972   void vpshuflw(XMMRegister dst, XMMRegister src, int mode, int vector_len);
1973 
1974   //shuffle floats and doubles
1975   void shufps(XMMRegister, XMMRegister, int);
1976   void shufpd(XMMRegister, XMMRegister, int);
1977   void vshufps(XMMRegister, XMMRegister, XMMRegister, int, int);
1978   void vshufpd(XMMRegister, XMMRegister, XMMRegister, int, int);
1979 
1980   // Shuffle packed values at 128 bit granularity
1981   void evshufi64x2(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
1982 
1983   // Shift Right by bytes Logical DoubleQuadword Immediate
1984   void psrldq(XMMRegister dst, int shift);
1985   // Shift Left by bytes Logical DoubleQuadword Immediate
1986   void pslldq(XMMRegister dst, int shift);
1987 
1988   // Logical Compare 128bit
1989   void ptest(XMMRegister dst, XMMRegister src);
1990   void ptest(XMMRegister dst, Address src);
1991   // Logical Compare 256bit
1992   void vptest(XMMRegister dst, XMMRegister src);
1993   void vptest(XMMRegister dst, Address src);
1994 
1995   void evptestmb(KRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1996   void evptestmd(KRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1997   void evptestnmd(KRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1998 
1999   // Vector compare
2000   void vptest(XMMRegister dst, XMMRegister src, int vector_len);
2001   void vtestps(XMMRegister dst, XMMRegister src, int vector_len);
2002 
2003   // Interleave Low Bytes
2004   void punpcklbw(XMMRegister dst, XMMRegister src);
2005   void punpcklbw(XMMRegister dst, Address src);
2006 
2007   // Interleave Low Doublewords
2008   void punpckldq(XMMRegister dst, XMMRegister src);
2009   void punpckldq(XMMRegister dst, Address src);
2010   void vpunpckldq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2011 
2012   // Interleave High Word
2013   void vpunpckhwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2014 
2015   // Interleave Low Word
2016   void vpunpcklwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2017 
2018   // Interleave High Doublewords
2019   void vpunpckhdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2020 
2021   // Interleave Low Quadwords
2022   void punpcklqdq(XMMRegister dst, XMMRegister src);
2023 
2024   void evpunpcklqdq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2025   void evpunpcklqdq(XMMRegister dst, KRegister mask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
2026   void evpunpckhqdq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2027   void evpunpckhqdq(XMMRegister dst, KRegister mask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
2028 
2029   // Vector sum of absolute difference.
2030   void vpsadbw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2031 
2032 #ifndef _LP64 // no 32bit push/pop on amd64
2033   void pushl(Address src);
2034 #endif
2035 
2036   void pushq(Address src);
2037 
2038   void rcll(Register dst, int imm8);
2039 
2040   void rclq(Register dst, int imm8);
2041 
2042   void rcrq(Register dst, int imm8);
2043 
2044   void rcpps(XMMRegister dst, XMMRegister src);
2045 
2046   void rcpss(XMMRegister dst, XMMRegister src);
2047 
2048   void rdtsc();
2049   void rdtscp();
2050 
2051   void ret(int imm16);
2052 
2053   void roll(Register dst);
2054 
2055   void roll(Register dst, int imm8);
2056 
2057   void rorl(Register dst);
2058 
2059   void rorl(Register dst, int imm8);
2060 
2061 #ifdef _LP64
2062   void rolq(Register dst);
2063   void rolq(Register dst, int imm8);
2064   void rorq(Register dst);
2065   void rorq(Register dst, int imm8);
2066   void rorxl(Register dst, Register src, int imm8);
2067   void rorxl(Register dst, Address src, int imm8);
2068   void rorxq(Register dst, Register src, int imm8);
2069   void rorxq(Register dst, Address src, int imm8);
2070 #endif
2071 
2072   void sahf();
2073 
2074   void sall(Register dst, int imm8);
2075   void sall(Register dst);
2076   void sall(Address dst, int imm8);
2077   void sall(Address dst);
2078 
2079   void sarl(Address dst, int imm8);
2080   void sarl(Address dst);
2081   void sarl(Register dst, int imm8);
2082   void sarl(Register dst);
2083 
2084 #ifdef _LP64
2085   void salq(Register dst, int imm8);
2086   void salq(Register dst);
2087   void salq(Address dst, int imm8);
2088   void salq(Address dst);
2089 
2090   void sarq(Address dst, int imm8);
2091   void sarq(Address dst);
2092   void sarq(Register dst, int imm8);
2093   void sarq(Register dst);
2094 #endif
2095 
2096   void sbbl(Address dst, int32_t imm32);
2097   void sbbl(Register dst, int32_t imm32);
2098   void sbbl(Register dst, Address src);
2099   void sbbl(Register dst, Register src);
2100 
2101   void sbbq(Address dst, int32_t imm32);
2102   void sbbq(Register dst, int32_t imm32);
2103   void sbbq(Register dst, Address src);
2104   void sbbq(Register dst, Register src);
2105 
2106   void setb(Condition cc, Register dst);
2107 
2108   void palignr(XMMRegister dst, XMMRegister src, int imm8);
2109   void vpalignr(XMMRegister dst, XMMRegister src1, XMMRegister src2, int imm8, int vector_len);
2110   void evalignq(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2111 
2112   void pblendw(XMMRegister dst, XMMRegister src, int imm8);
2113   void vblendps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int imm8, int vector_len);
2114 
2115   void sha1rnds4(XMMRegister dst, XMMRegister src, int imm8);
2116   void sha1nexte(XMMRegister dst, XMMRegister src);
2117   void sha1msg1(XMMRegister dst, XMMRegister src);
2118   void sha1msg2(XMMRegister dst, XMMRegister src);
2119   // xmm0 is implicit additional source to the following instruction.
2120   void sha256rnds2(XMMRegister dst, XMMRegister src);
2121   void sha256msg1(XMMRegister dst, XMMRegister src);
2122   void sha256msg2(XMMRegister dst, XMMRegister src);
2123 
2124   void shldl(Register dst, Register src);
2125   void shldl(Register dst, Register src, int8_t imm8);
2126   void shrdl(Register dst, Register src);
2127   void shrdl(Register dst, Register src, int8_t imm8);
2128 #ifdef _LP64
2129   void shldq(Register dst, Register src, int8_t imm8);
2130   void shrdq(Register dst, Register src, int8_t imm8);
2131 #endif
2132 
2133   void shll(Register dst, int imm8);
2134   void shll(Register dst);
2135 
2136   void shlq(Register dst, int imm8);
2137   void shlq(Register dst);
2138 
2139   void shrl(Register dst, int imm8);
2140   void shrl(Register dst);
2141   void shrl(Address dst);
2142   void shrl(Address dst, int imm8);
2143 
2144   void shrq(Register dst, int imm8);
2145   void shrq(Register dst);
2146   void shrq(Address dst);
2147   void shrq(Address dst, int imm8);
2148 
2149   void smovl(); // QQQ generic?
2150 
2151   // Compute Square Root of Scalar Double-Precision Floating-Point Value
2152   void sqrtsd(XMMRegister dst, Address src);
2153   void sqrtsd(XMMRegister dst, XMMRegister src);
2154 
2155   void roundsd(XMMRegister dst, Address src, int32_t rmode);
2156   void roundsd(XMMRegister dst, XMMRegister src, int32_t rmode);
2157 
2158   // Compute Square Root of Scalar Single-Precision Floating-Point Value
2159   void sqrtss(XMMRegister dst, Address src);
2160   void sqrtss(XMMRegister dst, XMMRegister src);
2161 
2162   void std();
2163 
2164   void stmxcsr( Address dst );
2165 
2166   void subl(Address dst, int32_t imm32);
2167   void subl(Address dst, Register src);
2168   void subl(Register dst, int32_t imm32);
2169   void subl(Register dst, Address src);
2170   void subl(Register dst, Register src);
2171 
2172   void subq(Address dst, int32_t imm32);
2173   void subq(Address dst, Register src);
2174   void subq(Register dst, int32_t imm32);
2175   void subq(Register dst, Address src);
2176   void subq(Register dst, Register src);
2177 
2178   // Force generation of a 4 byte immediate value even if it fits into 8bit
2179   void subl_imm32(Register dst, int32_t imm32);
2180   void subq_imm32(Register dst, int32_t imm32);
2181 
2182   // Subtract Scalar Double-Precision Floating-Point Values
2183   void subsd(XMMRegister dst, Address src);
2184   void subsd(XMMRegister dst, XMMRegister src);
2185 
2186   // Subtract Scalar Single-Precision Floating-Point Values
2187   void subss(XMMRegister dst, Address src);
2188   void subss(XMMRegister dst, XMMRegister src);
2189 
2190   void testb(Address dst, int imm8);
2191   void testb(Register dst, int imm8, bool use_ral = true);
2192 
2193   void testl(Address dst, int32_t imm32);
2194   void testl(Register dst, int32_t imm32);
2195   void testl(Register dst, Register src);
2196   void testl(Register dst, Address src);
2197 
2198   void testq(Address dst, int32_t imm32);
2199   void testq(Register dst, int32_t imm32);
2200   void testq(Register dst, Register src);
2201   void testq(Register dst, Address src);
2202 
2203   // BMI - count trailing zeros
2204   void tzcntl(Register dst, Register src);
2205   void tzcntl(Register dst, Address src);
2206   void tzcntq(Register dst, Register src);
2207   void tzcntq(Register dst, Address src);
2208 
2209   // Unordered Compare Scalar Double-Precision Floating-Point Values and set EFLAGS
2210   void ucomisd(XMMRegister dst, Address src);
2211   void ucomisd(XMMRegister dst, XMMRegister src);
2212 
2213   // Unordered Compare Scalar Single-Precision Floating-Point Values and set EFLAGS
2214   void ucomiss(XMMRegister dst, Address src);
2215   void ucomiss(XMMRegister dst, XMMRegister src);
2216 
2217   void xabort(int8_t imm8);
2218 
2219   void xaddb(Address dst, Register src);
2220   void xaddw(Address dst, Register src);
2221   void xaddl(Address dst, Register src);
2222   void xaddq(Address dst, Register src);
2223 
2224   void xbegin(Label& abort, relocInfo::relocType rtype = relocInfo::none);
2225 
2226   void xchgb(Register reg, Address adr);
2227   void xchgw(Register reg, Address adr);
2228   void xchgl(Register reg, Address adr);
2229   void xchgl(Register dst, Register src);
2230 
2231   void xchgq(Register reg, Address adr);
2232   void xchgq(Register dst, Register src);
2233 
2234   void xend();
2235 
2236   // Get Value of Extended Control Register
2237   void xgetbv();
2238 
2239   void xorl(Register dst, int32_t imm32);
2240   void xorl(Address dst, int32_t imm32);
2241   void xorl(Register dst, Address src);
2242   void xorl(Register dst, Register src);
2243   void xorl(Address dst, Register src);
2244 
2245   void xorb(Address dst, Register src);
2246   void xorb(Register dst, Address src);
2247   void xorw(Register dst, Register src);
2248 
2249   void xorq(Register dst, Address src);
2250   void xorq(Address dst, int32_t imm32);
2251   void xorq(Register dst, Register src);
2252   void xorq(Register dst, int32_t imm32);
2253   void xorq(Address dst, Register src);
2254 
2255   // AVX 3-operands scalar instructions (encoded with VEX prefix)
2256 
2257   void vaddsd(XMMRegister dst, XMMRegister nds, Address src);
2258   void vaddsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2259   void vaddss(XMMRegister dst, XMMRegister nds, Address src);
2260   void vaddss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2261   void vdivsd(XMMRegister dst, XMMRegister nds, Address src);
2262   void vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2263   void evdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src, EvexRoundPrefix rmode);
2264   void vdivss(XMMRegister dst, XMMRegister nds, Address src);
2265   void vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2266   void vfmadd231sd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2267   void vfnmadd213sd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2268   void evfnmadd213sd(XMMRegister dst, XMMRegister nds, XMMRegister src, EvexRoundPrefix rmode);
2269   void vfnmadd231sd(XMMRegister dst, XMMRegister src1, XMMRegister src2);
2270   void vfmadd231ss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2271   void vmulsd(XMMRegister dst, XMMRegister nds, Address src);
2272   void vmulsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2273   void vmulss(XMMRegister dst, XMMRegister nds, Address src);
2274   void vmulss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2275   void vsubsd(XMMRegister dst, XMMRegister nds, Address src);
2276   void vsubsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2277   void vsubss(XMMRegister dst, XMMRegister nds, Address src);
2278   void vsubss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2279 
2280   void vmaxss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2281   void vmaxsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2282   void vminss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2283   void vminsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2284 
2285   void sarxl(Register dst, Register src1, Register src2);
2286   void sarxl(Register dst, Address src1, Register src2);
2287   void sarxq(Register dst, Register src1, Register src2);
2288   void sarxq(Register dst, Address src1, Register src2);
2289   void shlxl(Register dst, Register src1, Register src2);
2290   void shlxl(Register dst, Address src1, Register src2);
2291   void shlxq(Register dst, Register src1, Register src2);
2292   void shlxq(Register dst, Address src1, Register src2);
2293   void shrxl(Register dst, Register src1, Register src2);
2294   void shrxl(Register dst, Address src1, Register src2);
2295   void shrxq(Register dst, Register src1, Register src2);
2296   void shrxq(Register dst, Address src1, Register src2);
2297 
2298   void bzhiq(Register dst, Register src1, Register src2);
2299 
2300   void pextl(Register dst, Register src1, Register src2);
2301   void pdepl(Register dst, Register src1, Register src2);
2302   void pextq(Register dst, Register src1, Register src2);
2303   void pdepq(Register dst, Register src1, Register src2);
2304   void pextl(Register dst, Register src1, Address src2);
2305   void pdepl(Register dst, Register src1, Address src2);
2306   void pextq(Register dst, Register src1, Address src2);
2307   void pdepq(Register dst, Register src1, Address src2);
2308 
2309 
2310   //====================VECTOR ARITHMETIC=====================================
2311   // Add Packed Floating-Point Values
2312   void addpd(XMMRegister dst, XMMRegister src);
2313   void addpd(XMMRegister dst, Address src);
2314   void addps(XMMRegister dst, XMMRegister src);
2315   void vaddpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2316   void vaddps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2317   void vaddpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2318   void vaddps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2319 
2320   // Subtract Packed Floating-Point Values
2321   void subpd(XMMRegister dst, XMMRegister src);
2322   void subps(XMMRegister dst, XMMRegister src);
2323   void vsubpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2324   void vsubps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2325   void vsubpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2326   void vsubps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2327 
2328   // Multiply Packed Floating-Point Values
2329   void mulpd(XMMRegister dst, XMMRegister src);
2330   void mulpd(XMMRegister dst, Address src);
2331   void mulps(XMMRegister dst, XMMRegister src);
2332   void vmulpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2333   void vmulps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2334   void vmulpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2335   void vmulps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2336 
2337   void vfmadd231pd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2338   void vfmadd231ps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2339   void vfmadd231pd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2340   void vfmadd231ps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2341 
2342   // Divide Packed Floating-Point Values
2343   void divpd(XMMRegister dst, XMMRegister src);
2344   void divps(XMMRegister dst, XMMRegister src);
2345   void vdivpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2346   void vdivps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2347   void vdivpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2348   void vdivps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2349 
2350   // Sqrt Packed Floating-Point Values
2351   void vsqrtpd(XMMRegister dst, XMMRegister src, int vector_len);
2352   void vsqrtpd(XMMRegister dst, Address src, int vector_len);
2353   void vsqrtps(XMMRegister dst, XMMRegister src, int vector_len);
2354   void vsqrtps(XMMRegister dst, Address src, int vector_len);
2355 
2356   // Round Packed Double precision value.
2357   void vroundpd(XMMRegister dst, XMMRegister src, int32_t rmode, int vector_len);
2358   void vroundpd(XMMRegister dst, Address src, int32_t rmode, int vector_len);
2359   void vrndscalesd(XMMRegister dst,  XMMRegister src1,  XMMRegister src2, int32_t rmode);
2360   void vrndscalepd(XMMRegister dst,  XMMRegister src,  int32_t rmode, int vector_len);
2361   void vrndscalepd(XMMRegister dst, Address src, int32_t rmode, int vector_len);
2362   void vroundsd(XMMRegister dst, XMMRegister src, XMMRegister src2, int32_t rmode);
2363   void vroundsd(XMMRegister dst, XMMRegister src, Address src2, int32_t rmode);
2364 
2365   // Bitwise Logical AND of Packed Floating-Point Values
2366   void andpd(XMMRegister dst, XMMRegister src);
2367   void andps(XMMRegister dst, XMMRegister src);
2368   void vandpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2369   void vandps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2370   void vandpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2371   void vandps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2372 
2373   void unpckhpd(XMMRegister dst, XMMRegister src);
2374   void unpcklpd(XMMRegister dst, XMMRegister src);
2375 
2376   // Bitwise Logical XOR of Packed Floating-Point Values
2377   void xorpd(XMMRegister dst, XMMRegister src);
2378   void xorps(XMMRegister dst, XMMRegister src);
2379   void vxorpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2380   void vxorps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2381   void vxorpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2382   void vxorps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2383 
2384   // Add horizontal packed integers
2385   void vphaddw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2386   void vphaddd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2387   void phaddw(XMMRegister dst, XMMRegister src);
2388   void phaddd(XMMRegister dst, XMMRegister src);
2389 
2390   // Add packed integers
2391   void paddb(XMMRegister dst, XMMRegister src);
2392   void paddw(XMMRegister dst, XMMRegister src);
2393   void paddd(XMMRegister dst, XMMRegister src);
2394   void paddd(XMMRegister dst, Address src);
2395   void paddq(XMMRegister dst, XMMRegister src);
2396   void vpaddb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2397   void vpaddw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2398   void vpaddd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2399   void vpaddq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2400   void vpaddb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2401   void vpaddw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2402   void vpaddd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2403   void vpaddq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2404 
2405   // Leaf level assembler routines for masked operations.
2406   void evpaddb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2407   void evpaddb(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2408   void evpaddw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2409   void evpaddw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2410   void evpaddd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2411   void evpaddd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2412   void evpaddq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2413   void evpaddq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2414   void evaddps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2415   void evaddps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2416   void evaddpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2417   void evaddpd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2418   void evpsubb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2419   void evpsubb(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2420   void evpsubw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2421   void evpsubw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2422   void evpsubd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2423   void evpsubd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2424   void evpsubq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2425   void evpsubq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2426   void evsubps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2427   void evsubps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2428   void evsubpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2429   void evsubpd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2430   void evpmullw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2431   void evpmullw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2432   void evpmulld(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2433   void evpmulld(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2434   void evpmullq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2435   void evpmullq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2436   void evmulps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2437   void evmulps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2438   void evmulpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2439   void evmulpd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2440   void evdivps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2441   void evdivps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2442   void evdivpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2443   void evdivpd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2444   void evpabsb(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2445   void evpabsb(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
2446   void evpabsw(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2447   void evpabsw(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
2448   void evpabsd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2449   void evpabsd(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
2450   void evpabsq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2451   void evpabsq(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
2452   void evpfma213ps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2453   void evpfma213ps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2454   void evpfma213pd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2455   void evpfma213pd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2456   void evpermb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2457   void evpermb(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2458   void evpermw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2459   void evpermw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2460   void evpermd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2461   void evpermd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2462   void evpermq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2463   void evpermq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2464   void evpsllw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2465   void evpslld(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2466   void evpsllq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2467   void evpsrlw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2468   void evpsrld(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2469   void evpsrlq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2470   void evpsraw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2471   void evpsrad(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2472   void evpsraq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2473   void evsqrtps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2474   void evsqrtps(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2475   void evsqrtpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2476   void evsqrtpd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2477 
2478   void evpsllw(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2479   void evpslld(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2480   void evpsllq(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2481   void evpsrlw(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2482   void evpsrld(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2483   void evpsrlq(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2484   void evpsraw(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2485   void evpsrad(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2486   void evpsraq(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2487 
2488   void evpsllvw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2489   void evpsllvd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2490   void evpsllvq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2491   void evpsrlvw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2492   void evpsrlvd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2493   void evpsrlvq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2494   void evpsravw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2495   void evpsravd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2496   void evpsravq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2497   void evpmaxsb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2498   void evpmaxsw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2499   void evpmaxsd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2500   void evpmaxsq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2501   void evpminsb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2502   void evpminsw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2503   void evpminsd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2504   void evpminsq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2505   void evpmaxsb(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2506   void evpmaxsw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2507   void evpmaxsd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2508   void evpmaxsq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2509   void evpminsb(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2510   void evpminsw(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2511   void evpminsd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2512   void evpminsq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2513   void evpord(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2514   void evpord(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2515   void evporq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2516   void evporq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2517   void evpandd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2518   void evpandd(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2519   void evpandq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2520   void evpandq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2521   void evpxord(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2522   void evpxord(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2523   void evpxorq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2524   void evpxorq(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2525 
2526   void evprold(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2527   void evprolq(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2528   void evprolvd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2529   void evprolvq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2530   void evprord(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2531   void evprorq(XMMRegister dst, KRegister mask, XMMRegister src, int shift, bool merge, int vector_len);
2532   void evprorvd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2533   void evprorvq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2534 
2535   void evpternlogd(XMMRegister dst, int imm8, KRegister mask, XMMRegister src2, XMMRegister src3, bool merge, int vector_len);
2536   void evpternlogd(XMMRegister dst, int imm8, KRegister mask, XMMRegister src2, Address src3, bool merge, int vector_len);
2537   void evpternlogq(XMMRegister dst, int imm8, KRegister mask, XMMRegister src2, XMMRegister src3, bool merge, int vector_len);
2538   void evpternlogq(XMMRegister dst, int imm8, KRegister mask, XMMRegister src2, Address src3, bool merge, int vector_len);
2539 
2540   void evplzcntd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2541   void evplzcntq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2542 
2543   // Sub packed integers
2544   void psubb(XMMRegister dst, XMMRegister src);
2545   void psubw(XMMRegister dst, XMMRegister src);
2546   void psubd(XMMRegister dst, XMMRegister src);
2547   void psubq(XMMRegister dst, XMMRegister src);
2548   void vpsubusb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2549   void vpsubb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2550   void vpsubw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2551   void vpsubd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2552   void vpsubq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2553   void vpsubb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2554   void vpsubw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2555   void vpsubd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2556   void vpsubq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2557 
2558   // Multiply packed integers (only shorts and ints)
2559   void pmullw(XMMRegister dst, XMMRegister src);
2560   void pmulld(XMMRegister dst, XMMRegister src);
2561   void pmuludq(XMMRegister dst, XMMRegister src);
2562   void vpmullw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2563   void vpmulld(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2564   void evpmullq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2565   void vpmuludq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2566   void vpmullw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2567   void vpmulld(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2568   void evpmullq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2569   void vpmulhuw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2570 
2571   // Minimum of packed integers
2572   void pminsb(XMMRegister dst, XMMRegister src);
2573   void vpminsb(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2574   void pminsw(XMMRegister dst, XMMRegister src);
2575   void vpminsw(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2576   void pminsd(XMMRegister dst, XMMRegister src);
2577   void vpminsd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2578   void vpminsq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2579   void minps(XMMRegister dst, XMMRegister src);
2580   void vminps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2581   void minpd(XMMRegister dst, XMMRegister src);
2582   void vminpd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2583 
2584   // Maximum of packed integers
2585   void pmaxsb(XMMRegister dst, XMMRegister src);
2586   void vpmaxsb(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2587   void pmaxsw(XMMRegister dst, XMMRegister src);
2588   void vpmaxsw(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2589   void pmaxsd(XMMRegister dst, XMMRegister src);
2590   void vpmaxsd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2591   void vpmaxsq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2592   void maxps(XMMRegister dst, XMMRegister src);
2593   void vmaxps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2594   void maxpd(XMMRegister dst, XMMRegister src);
2595   void vmaxpd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2596 
2597   // Shift left packed integers
2598   void psllw(XMMRegister dst, int shift);
2599   void pslld(XMMRegister dst, int shift);
2600   void psllq(XMMRegister dst, int shift);
2601   void psllw(XMMRegister dst, XMMRegister shift);
2602   void pslld(XMMRegister dst, XMMRegister shift);
2603   void psllq(XMMRegister dst, XMMRegister shift);
2604   void vpsllw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2605   void vpslld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2606   void vpsllq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2607   void vpsllw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2608   void vpslld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2609   void vpsllq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2610   void vpslldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2611 
2612   // Logical shift right packed integers
2613   void psrlw(XMMRegister dst, int shift);
2614   void psrld(XMMRegister dst, int shift);
2615   void psrlq(XMMRegister dst, int shift);
2616   void psrlw(XMMRegister dst, XMMRegister shift);
2617   void psrld(XMMRegister dst, XMMRegister shift);
2618   void psrlq(XMMRegister dst, XMMRegister shift);
2619   void vpsrlw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2620   void vpsrld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2621   void vpsrlq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2622   void vpsrlw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2623   void vpsrld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2624   void vpsrlq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2625   void vpsrldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2626   void evpsrlvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2627   void evpsllvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2628 
2629   // Arithmetic shift right packed integers (only shorts and ints, no instructions for longs)
2630   void psraw(XMMRegister dst, int shift);
2631   void psrad(XMMRegister dst, int shift);
2632   void psraw(XMMRegister dst, XMMRegister shift);
2633   void psrad(XMMRegister dst, XMMRegister shift);
2634   void vpsraw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2635   void vpsrad(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2636   void vpsraw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2637   void vpsrad(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2638   void evpsravw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2639   void evpsraq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2640   void evpsraq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2641 
2642   // Variable shift left packed integers
2643   void vpsllvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2644   void vpsllvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2645 
2646   // Variable shift right packed integers
2647   void vpsrlvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2648   void vpsrlvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2649 
2650   // Variable shift right arithmetic packed integers
2651   void vpsravd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2652   void evpsravq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2653 
2654   void vpshldvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2655   void vpshrdvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2656 
2657   // And packed integers
2658   void pand(XMMRegister dst, XMMRegister src);
2659   void vpand(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2660   void vpand(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2661   void evpandq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2662   void evpandq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2663 
2664   // Andn packed integers
2665   void pandn(XMMRegister dst, XMMRegister src);
2666   void vpandn(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2667 
2668   // Or packed integers
2669   void por(XMMRegister dst, XMMRegister src);
2670   void vpor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2671   void vpor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2672   void evporq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2673   void evporq(XMMRegister dst, XMMRegister nds, Address     src, int vector_len);
2674 
2675   // Xor packed integers
2676   void pxor(XMMRegister dst, XMMRegister src);
2677   void vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2678   void vpxor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2679   void vpxorq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2680   void evpxorq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2681   void evpxorq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2682 
2683   // Ternary logic instruction.
2684   void vpternlogd(XMMRegister dst, int imm8, XMMRegister src2, XMMRegister src3, int vector_len);
2685   void vpternlogd(XMMRegister dst, int imm8, XMMRegister src2, Address     src3, int vector_len);
2686   void vpternlogq(XMMRegister dst, int imm8, XMMRegister src2, XMMRegister src3, int vector_len);
2687   void vpternlogq(XMMRegister dst, int imm8, XMMRegister src2, Address     src3, int vector_len);
2688 
2689   // Vector compress/expand instructions.
2690   void evpcompressb(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2691   void evpcompressw(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2692   void evpcompressd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2693   void evpcompressq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2694   void evcompressps(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2695   void evcompresspd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2696 
2697   void evpexpandb(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2698   void evpexpandw(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2699   void evpexpandd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2700   void evpexpandq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2701   void evexpandps(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2702   void evexpandpd(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
2703 
2704   // Vector Rotate Left/Right instruction.
2705   void evprolvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2706   void evprolvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2707   void evprorvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2708   void evprorvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2709   void evprold(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2710   void evprolq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2711   void evprord(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2712   void evprorq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2713 
2714   // vinserti forms
2715   void vinserti128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2716   void vinserti128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2717   void vinserti32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2718   void vinserti32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2719   void vinserti64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2720 
2721   // vinsertf forms
2722   void vinsertf128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2723   void vinsertf128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2724   void vinsertf32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2725   void vinsertf32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2726   void vinsertf64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2727   void vinsertf64x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2728 
2729   // vextracti forms
2730   void vextracti128(XMMRegister dst, XMMRegister src, uint8_t imm8);
2731   void vextracti128(Address dst, XMMRegister src, uint8_t imm8);
2732   void vextracti32x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2733   void vextracti32x4(Address dst, XMMRegister src, uint8_t imm8);
2734   void vextracti64x2(XMMRegister dst, XMMRegister src, uint8_t imm8);
2735   void vextracti64x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2736   void vextracti64x4(Address dst, XMMRegister src, uint8_t imm8);
2737 
2738   // vextractf forms
2739   void vextractf128(XMMRegister dst, XMMRegister src, uint8_t imm8);
2740   void vextractf128(Address dst, XMMRegister src, uint8_t imm8);
2741   void vextractf32x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2742   void vextractf32x4(Address dst, XMMRegister src, uint8_t imm8);
2743   void vextractf64x2(XMMRegister dst, XMMRegister src, uint8_t imm8);
2744   void vextractf64x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2745   void vextractf64x4(Address dst, XMMRegister src, uint8_t imm8);
2746 
2747   void extractps(Register dst, XMMRegister src, uint8_t imm8);
2748 
2749   // xmm/mem sourced byte/word/dword/qword replicate
2750   void vpbroadcastb(XMMRegister dst, XMMRegister src, int vector_len);
2751   void vpbroadcastb(XMMRegister dst, Address src, int vector_len);
2752   void vpbroadcastw(XMMRegister dst, XMMRegister src, int vector_len);
2753   void vpbroadcastw(XMMRegister dst, Address src, int vector_len);
2754   void vpbroadcastd(XMMRegister dst, XMMRegister src, int vector_len);
2755   void vpbroadcastd(XMMRegister dst, Address src, int vector_len);
2756   void vpbroadcastq(XMMRegister dst, XMMRegister src, int vector_len);
2757   void vpbroadcastq(XMMRegister dst, Address src, int vector_len);
2758 
2759   void evbroadcasti32x4(XMMRegister dst, Address src, int vector_len);
2760   void evbroadcasti64x2(XMMRegister dst, XMMRegister src, int vector_len);
2761   void evbroadcasti64x2(XMMRegister dst, Address src, int vector_len);
2762 
2763   // scalar single/double/128bit precision replicate
2764   void vbroadcastss(XMMRegister dst, XMMRegister src, int vector_len);
2765   void vbroadcastss(XMMRegister dst, Address src, int vector_len);
2766   void vbroadcastsd(XMMRegister dst, XMMRegister src, int vector_len);
2767   void vbroadcastsd(XMMRegister dst, Address src, int vector_len);
2768   void vbroadcastf128(XMMRegister dst, Address src, int vector_len);
2769 
2770   // gpr sourced byte/word/dword/qword replicate
2771   void evpbroadcastb(XMMRegister dst, Register src, int vector_len);
2772   void evpbroadcastw(XMMRegister dst, Register src, int vector_len);
2773   void evpbroadcastd(XMMRegister dst, Register src, int vector_len);
2774   void evpbroadcastq(XMMRegister dst, Register src, int vector_len);
2775 
2776   // Gather AVX2 and AVX3
2777   void vpgatherdd(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2778   void vpgatherdq(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2779   void vgatherdpd(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2780   void vgatherdps(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2781   void evpgatherdd(XMMRegister dst, KRegister mask, Address src, int vector_len);
2782   void evpgatherdq(XMMRegister dst, KRegister mask, Address src, int vector_len);
2783   void evgatherdpd(XMMRegister dst, KRegister mask, Address src, int vector_len);
2784   void evgatherdps(XMMRegister dst, KRegister mask, Address src, int vector_len);
2785 
2786   //Scatter AVX3 only
2787   void evpscatterdd(Address dst, KRegister mask, XMMRegister src, int vector_len);
2788   void evpscatterdq(Address dst, KRegister mask, XMMRegister src, int vector_len);
2789   void evscatterdps(Address dst, KRegister mask, XMMRegister src, int vector_len);
2790   void evscatterdpd(Address dst, KRegister mask, XMMRegister src, int vector_len);
2791 
2792   // Carry-Less Multiplication Quadword
2793   void pclmulqdq(XMMRegister dst, XMMRegister src, int mask);
2794   void vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask);
2795   void evpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask, int vector_len);
2796   // AVX instruction which is used to clear upper 128 bits of YMM registers and
2797   // to avoid transaction penalty between AVX and SSE states. There is no
2798   // penalty if legacy SSE instructions are encoded using VEX prefix because
2799   // they always clear upper 128 bits. It should be used before calling
2800   // runtime code and native libraries.
2801   void vzeroupper();
2802 
2803   void vzeroall();
2804 
2805   // Vector double compares
2806   void vcmppd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cop, int vector_len);
2807   void evcmppd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2808                ComparisonPredicateFP comparison, int vector_len);
2809 
2810   // Vector float compares
2811   void vcmpps(XMMRegister dst, XMMRegister nds, XMMRegister src, int comparison, int vector_len);
2812   void evcmpps(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2813                ComparisonPredicateFP comparison, int vector_len);
2814 
2815   // Vector integer compares
2816   void vpcmpgtd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2817   void evpcmpd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2818                int comparison, bool is_signed, int vector_len);
2819   void evpcmpd(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2820                int comparison, bool is_signed, int vector_len);
2821 
2822   // Vector long compares
2823   void evpcmpq(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2824                int comparison, bool is_signed, int vector_len);
2825   void evpcmpq(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2826                int comparison, bool is_signed, int vector_len);
2827 
2828   // Vector byte compares
2829   void evpcmpb(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2830                int comparison, bool is_signed, int vector_len);
2831   void evpcmpb(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2832                int comparison, bool is_signed, int vector_len);
2833 
2834   // Vector short compares
2835   void evpcmpw(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2836                int comparison, bool is_signed, int vector_len);
2837   void evpcmpw(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2838                int comparison, bool is_signed, int vector_len);
2839 
2840   void evpmovb2m(KRegister dst, XMMRegister src, int vector_len);
2841   void evpmovw2m(KRegister dst, XMMRegister src, int vector_len);
2842   void evpmovd2m(KRegister dst, XMMRegister src, int vector_len);
2843   void evpmovq2m(KRegister dst, XMMRegister src, int vector_len);
2844   void evpmovm2b(XMMRegister dst, KRegister src, int vector_len);
2845   void evpmovm2w(XMMRegister dst, KRegister src, int vector_len);
2846   void evpmovm2d(XMMRegister dst, KRegister src, int vector_len);
2847   void evpmovm2q(XMMRegister dst, KRegister src, int vector_len);
2848 
2849   // floating point class tests
2850   void vfpclassss(KRegister kdst, XMMRegister src, uint8_t imm8);
2851   void vfpclasssd(KRegister kdst, XMMRegister src, uint8_t imm8);
2852 
2853   // Vector blends
2854   void blendvps(XMMRegister dst, XMMRegister src);
2855   void blendvpd(XMMRegister dst, XMMRegister src);
2856   void pblendvb(XMMRegister dst, XMMRegister src);
2857   void blendvpb(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len);
2858   void vblendvps(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len);
2859   void vblendvpd(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len);
2860   void vpblendvb(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len);
2861   void vpblendd(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
2862   void evblendmpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2863   void evblendmps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2864   void evpblendmb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2865   void evpblendmw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2866   void evpblendmd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2867   void evpblendmq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2868 
2869   // Galois field affine transformation instructions.
2870   void gf2p8affineqb(XMMRegister dst, XMMRegister src, int imm8);
2871   void vgf2p8affineqb(XMMRegister dst, XMMRegister src2, XMMRegister src3, int imm8, int vector_len);
2872 
2873  protected:
2874   // Next instructions require address alignment 16 bytes SSE mode.
2875   // They should be called only from corresponding MacroAssembler instructions.
2876   void andpd(XMMRegister dst, Address src);
2877   void andps(XMMRegister dst, Address src);
2878   void xorpd(XMMRegister dst, Address src);
2879   void xorps(XMMRegister dst, Address src);
2880 
2881 };
2882 
2883 // The Intel x86/Amd64 Assembler attributes: All fields enclosed here are to guide encoding level decisions.
2884 // Specific set functions are for specialized use, else defaults or whatever was supplied to object construction
2885 // are applied.
2886 class InstructionAttr {
2887 public:
2888   InstructionAttr(
2889     int vector_len,     // The length of vector to be applied in encoding - for both AVX and EVEX
2890     bool rex_vex_w,     // Width of data: if 32-bits or less, false, else if 64-bit or specially defined, true
2891     bool legacy_mode,   // Details if either this instruction is conditionally encoded to AVX or earlier if true else possibly EVEX
2892     bool no_reg_mask,   // when true, k0 is used when EVEX encoding is chosen, else embedded_opmask_register_specifier is used
2893     bool uses_vl)       // This instruction may have legacy constraints based on vector length for EVEX
2894     :
2895       _rex_vex_w(rex_vex_w),
2896       _legacy_mode(legacy_mode || UseAVX < 3),
2897       _no_reg_mask(no_reg_mask),
2898       _uses_vl(uses_vl),
2899       _rex_vex_w_reverted(false),
2900       _is_evex_instruction(false),
2901       _is_clear_context(true),
2902       _is_extended_context(false),
2903       _avx_vector_len(vector_len),
2904       _tuple_type(Assembler::EVEX_ETUP),
2905       _input_size_in_bits(Assembler::EVEX_NObit),
2906       _evex_encoding(0),
2907       _embedded_opmask_register_specifier(0), // hard code k0
2908       _current_assembler(nullptr) { }
2909 
2910   ~InstructionAttr() {
2911     if (_current_assembler != nullptr) {
2912       _current_assembler->clear_attributes();
2913     }
2914   }
2915 
2916 private:
2917   bool _rex_vex_w;
2918   bool _legacy_mode;
2919   bool _no_reg_mask;
2920   bool _uses_vl;
2921   bool _rex_vex_w_reverted;
2922   bool _is_evex_instruction;
2923   bool _is_clear_context;
2924   bool _is_extended_context;
2925   int  _avx_vector_len;
2926   int  _tuple_type;
2927   int  _input_size_in_bits;
2928   int  _evex_encoding;
2929   int _embedded_opmask_register_specifier;
2930 
2931   Assembler *_current_assembler;
2932 
2933 public:
2934   // query functions for field accessors
2935   bool is_rex_vex_w(void) const { return _rex_vex_w; }
2936   bool is_legacy_mode(void) const { return _legacy_mode; }
2937   bool is_no_reg_mask(void) const { return _no_reg_mask; }
2938   bool uses_vl(void) const { return _uses_vl; }
2939   bool is_rex_vex_w_reverted(void) { return _rex_vex_w_reverted; }
2940   bool is_evex_instruction(void) const { return _is_evex_instruction; }
2941   bool is_clear_context(void) const { return _is_clear_context; }
2942   bool is_extended_context(void) const { return _is_extended_context; }
2943   int  get_vector_len(void) const { return _avx_vector_len; }
2944   int  get_tuple_type(void) const { return _tuple_type; }
2945   int  get_input_size(void) const { return _input_size_in_bits; }
2946   int  get_evex_encoding(void) const { return _evex_encoding; }
2947   int  get_embedded_opmask_register_specifier(void) const { return _embedded_opmask_register_specifier; }
2948 
2949   // Set the vector len manually
2950   void set_vector_len(int vector_len) { _avx_vector_len = vector_len; }
2951 
2952   // Set revert rex_vex_w for avx encoding
2953   void set_rex_vex_w_reverted(void) { _rex_vex_w_reverted = true; }
2954 
2955   // Set rex_vex_w based on state
2956   void set_rex_vex_w(bool state) { _rex_vex_w = state; }
2957 
2958   // Set the instruction to be encoded in AVX mode
2959   void set_is_legacy_mode(void) { _legacy_mode = true; }
2960 
2961   // Set the current instruction to be encoded as an EVEX instruction
2962   void set_is_evex_instruction(void) { _is_evex_instruction = true; }
2963 
2964   // Internal encoding data used in compressed immediate offset programming
2965   void set_evex_encoding(int value) { _evex_encoding = value; }
2966 
2967   // When the Evex.Z field is set (true), it is used to clear all non directed XMM/YMM/ZMM components.
2968   // This method unsets it so that merge semantics are used instead.
2969   void reset_is_clear_context(void) { _is_clear_context = false; }
2970 
2971   // Map back to current assembler so that we can manage object level association
2972   void set_current_assembler(Assembler *current_assembler) { _current_assembler = current_assembler; }
2973 
2974   // Address modifiers used for compressed displacement calculation
2975   void set_address_attributes(int tuple_type, int input_size_in_bits);
2976 
2977   // Set embedded opmask register specifier.
2978   void set_embedded_opmask_register_specifier(KRegister mask) {
2979     _embedded_opmask_register_specifier = mask->encoding() & 0x7;
2980   }
2981 
2982   void set_extended_context(void) { _is_extended_context = true; }
2983 
2984 };
2985 
2986 #endif // CPU_X86_ASSEMBLER_X86_HPP