1 /*
   2  * Copyright (c) 1997, 2015, 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_SPARC_VM_ASSEMBLER_SPARC_HPP
  26 #define CPU_SPARC_VM_ASSEMBLER_SPARC_HPP
  27 
  28 #include "asm/register.hpp"
  29 
  30 // The SPARC Assembler: Pure assembler doing NO optimizations on the instruction
  31 // level; i.e., what you write
  32 // is what you get. The Assembler is generating code into a CodeBuffer.
  33 
  34 class Assembler : public AbstractAssembler  {
  35   friend class AbstractAssembler;
  36   friend class AddressLiteral;
  37 
  38   // code patchers need various routines like inv_wdisp()
  39   friend class NativeInstruction;
  40   friend class NativeGeneralJump;
  41   friend class Relocation;
  42   friend class Label;
  43 
  44  public:
  45   // op carries format info; see page 62 & 267
  46 
  47   enum ops {
  48     call_op   = 1, // fmt 1
  49     branch_op = 0, // also sethi (fmt2)
  50     arith_op  = 2, // fmt 3, arith & misc
  51     ldst_op   = 3  // fmt 3, load/store
  52   };
  53 
  54   enum op2s {
  55     bpr_op2   = 3,
  56     fb_op2    = 6,
  57     fbp_op2   = 5,
  58     br_op2    = 2,
  59     bp_op2    = 1,
  60     sethi_op2 = 4
  61   };
  62 
  63   enum op3s {
  64     // selected op3s
  65     add_op3      = 0x00,
  66     and_op3      = 0x01,
  67     or_op3       = 0x02,
  68     xor_op3      = 0x03,
  69     sub_op3      = 0x04,
  70     andn_op3     = 0x05,
  71     orn_op3      = 0x06,
  72     xnor_op3     = 0x07,
  73     addc_op3     = 0x08,
  74     mulx_op3     = 0x09,
  75     umul_op3     = 0x0a,
  76     smul_op3     = 0x0b,
  77     subc_op3     = 0x0c,
  78     udivx_op3    = 0x0d,
  79     udiv_op3     = 0x0e,
  80     sdiv_op3     = 0x0f,
  81 
  82     addcc_op3    = 0x10,
  83     andcc_op3    = 0x11,
  84     orcc_op3     = 0x12,
  85     xorcc_op3    = 0x13,
  86     subcc_op3    = 0x14,
  87     andncc_op3   = 0x15,
  88     orncc_op3    = 0x16,
  89     xnorcc_op3   = 0x17,
  90     addccc_op3   = 0x18,
  91     aes4_op3     = 0x19,
  92     umulcc_op3   = 0x1a,
  93     smulcc_op3   = 0x1b,
  94     subccc_op3   = 0x1c,
  95     udivcc_op3   = 0x1e,
  96     sdivcc_op3   = 0x1f,
  97 
  98     taddcc_op3   = 0x20,
  99     tsubcc_op3   = 0x21,
 100     taddcctv_op3 = 0x22,
 101     tsubcctv_op3 = 0x23,
 102     mulscc_op3   = 0x24,
 103     sll_op3      = 0x25,
 104     sllx_op3     = 0x25,
 105     srl_op3      = 0x26,
 106     srlx_op3     = 0x26,
 107     sra_op3      = 0x27,
 108     srax_op3     = 0x27,
 109     rdreg_op3    = 0x28,
 110     membar_op3   = 0x28,
 111 
 112     flushw_op3   = 0x2b,
 113     movcc_op3    = 0x2c,
 114     sdivx_op3    = 0x2d,
 115     popc_op3     = 0x2e,
 116     movr_op3     = 0x2f,
 117 
 118     sir_op3      = 0x30,
 119     wrreg_op3    = 0x30,
 120     saved_op3    = 0x31,
 121 
 122     fpop1_op3    = 0x34,
 123     fpop2_op3    = 0x35,
 124     impdep1_op3  = 0x36,
 125     aes3_op3     = 0x36,
 126     sha_op3      = 0x36,
 127     alignaddr_op3  = 0x36,
 128     faligndata_op3 = 0x36,
 129     flog3_op3    = 0x36,
 130     edge_op3     = 0x36,
 131     fsrc_op3     = 0x36,
 132     xmulx_op3    = 0x36,
 133     impdep2_op3  = 0x37,
 134     stpartialf_op3 = 0x37,
 135     jmpl_op3     = 0x38,
 136     rett_op3     = 0x39,
 137     trap_op3     = 0x3a,
 138     flush_op3    = 0x3b,
 139     save_op3     = 0x3c,
 140     restore_op3  = 0x3d,
 141     done_op3     = 0x3e,
 142     retry_op3    = 0x3e,
 143 
 144     lduw_op3     = 0x00,
 145     ldub_op3     = 0x01,
 146     lduh_op3     = 0x02,
 147     ldd_op3      = 0x03,
 148     stw_op3      = 0x04,
 149     stb_op3      = 0x05,
 150     sth_op3      = 0x06,
 151     std_op3      = 0x07,
 152     ldsw_op3     = 0x08,
 153     ldsb_op3     = 0x09,
 154     ldsh_op3     = 0x0a,
 155     ldx_op3      = 0x0b,
 156 
 157     stx_op3      = 0x0e,
 158     swap_op3     = 0x0f,
 159 
 160     stwa_op3     = 0x14,
 161     stxa_op3     = 0x1e,
 162 
 163     ldf_op3      = 0x20,
 164     ldfsr_op3    = 0x21,
 165     ldqf_op3     = 0x22,
 166     lddf_op3     = 0x23,
 167     stf_op3      = 0x24,
 168     stfsr_op3    = 0x25,
 169     stqf_op3     = 0x26,
 170     stdf_op3     = 0x27,
 171 
 172     prefetch_op3 = 0x2d,
 173 
 174     casa_op3     = 0x3c,
 175     casxa_op3    = 0x3e,
 176 
 177     mftoi_op3    = 0x36,
 178 
 179     alt_bit_op3  = 0x10,
 180      cc_bit_op3  = 0x10
 181   };
 182 
 183   enum opfs {
 184     // selected opfs
 185     edge8n_opf         = 0x01,
 186 
 187     fmovs_opf          = 0x01,
 188     fmovd_opf          = 0x02,
 189 
 190     fnegs_opf          = 0x05,
 191     fnegd_opf          = 0x06,
 192 
 193     alignaddr_opf      = 0x18,
 194 
 195     fadds_opf          = 0x41,
 196     faddd_opf          = 0x42,
 197     fsubs_opf          = 0x45,
 198     fsubd_opf          = 0x46,
 199 
 200     faligndata_opf     = 0x48,
 201 
 202     fmuls_opf          = 0x49,
 203     fmuld_opf          = 0x4a,
 204     fdivs_opf          = 0x4d,
 205     fdivd_opf          = 0x4e,
 206 
 207     fcmps_opf          = 0x51,
 208     fcmpd_opf          = 0x52,
 209 
 210     fstox_opf          = 0x81,
 211     fdtox_opf          = 0x82,
 212     fxtos_opf          = 0x84,
 213     fxtod_opf          = 0x88,
 214     fitos_opf          = 0xc4,
 215     fdtos_opf          = 0xc6,
 216     fitod_opf          = 0xc8,
 217     fstod_opf          = 0xc9,
 218     fstoi_opf          = 0xd1,
 219     fdtoi_opf          = 0xd2,
 220 
 221     mdtox_opf          = 0x110,
 222     mstouw_opf         = 0x111,
 223     mstosw_opf         = 0x113,
 224     xmulx_opf          = 0x115,
 225     xmulxhi_opf        = 0x116,
 226     mxtod_opf          = 0x118,
 227     mwtos_opf          = 0x119,
 228 
 229     aes_kexpand0_opf   = 0x130,
 230     aes_kexpand2_opf   = 0x131,
 231 
 232     sha1_opf           = 0x141,
 233     sha256_opf         = 0x142,
 234     sha512_opf         = 0x143
 235   };
 236 
 237   enum op5s {
 238     aes_eround01_op5     = 0x00,
 239     aes_eround23_op5     = 0x01,
 240     aes_dround01_op5     = 0x02,
 241     aes_dround23_op5     = 0x03,
 242     aes_eround01_l_op5   = 0x04,
 243     aes_eround23_l_op5   = 0x05,
 244     aes_dround01_l_op5   = 0x06,
 245     aes_dround23_l_op5   = 0x07,
 246     aes_kexpand1_op5     = 0x08
 247   };
 248 
 249   enum RCondition {  rc_z = 1,  rc_lez = 2,  rc_lz = 3, rc_nz = 5, rc_gz = 6, rc_gez = 7, rc_last = rc_gez  };
 250 
 251   enum Condition {
 252      // for FBfcc & FBPfcc instruction
 253     f_never                     = 0,
 254     f_notEqual                  = 1,
 255     f_notZero                   = 1,
 256     f_lessOrGreater             = 2,
 257     f_unorderedOrLess           = 3,
 258     f_less                      = 4,
 259     f_unorderedOrGreater        = 5,
 260     f_greater                   = 6,
 261     f_unordered                 = 7,
 262     f_always                    = 8,
 263     f_equal                     = 9,
 264     f_zero                      = 9,
 265     f_unorderedOrEqual          = 10,
 266     f_greaterOrEqual            = 11,
 267     f_unorderedOrGreaterOrEqual = 12,
 268     f_lessOrEqual               = 13,
 269     f_unorderedOrLessOrEqual    = 14,
 270     f_ordered                   = 15,
 271 
 272     // V8 coproc, pp 123 v8 manual
 273 
 274     cp_always  = 8,
 275     cp_never   = 0,
 276     cp_3       = 7,
 277     cp_2       = 6,
 278     cp_2or3    = 5,
 279     cp_1       = 4,
 280     cp_1or3    = 3,
 281     cp_1or2    = 2,
 282     cp_1or2or3 = 1,
 283     cp_0       = 9,
 284     cp_0or3    = 10,
 285     cp_0or2    = 11,
 286     cp_0or2or3 = 12,
 287     cp_0or1    = 13,
 288     cp_0or1or3 = 14,
 289     cp_0or1or2 = 15,
 290 
 291 
 292     // for integers
 293 
 294     never                 =  0,
 295     equal                 =  1,
 296     zero                  =  1,
 297     lessEqual             =  2,
 298     less                  =  3,
 299     lessEqualUnsigned     =  4,
 300     lessUnsigned          =  5,
 301     carrySet              =  5,
 302     negative              =  6,
 303     overflowSet           =  7,
 304     always                =  8,
 305     notEqual              =  9,
 306     notZero               =  9,
 307     greater               =  10,
 308     greaterEqual          =  11,
 309     greaterUnsigned       =  12,
 310     greaterEqualUnsigned  =  13,
 311     carryClear            =  13,
 312     positive              =  14,
 313     overflowClear         =  15
 314   };
 315 
 316   enum CC {
 317     icc  = 0,  xcc  = 2,
 318     // ptr_cc is the correct condition code for a pointer or intptr_t:
 319     ptr_cc = NOT_LP64(icc) LP64_ONLY(xcc),
 320     fcc0 = 0,  fcc1 = 1, fcc2 = 2, fcc3 = 3
 321   };
 322 
 323   enum PrefetchFcn {
 324     severalReads = 0,  oneRead = 1,  severalWritesAndPossiblyReads = 2, oneWrite = 3, page = 4
 325   };
 326 
 327  public:
 328   // Helper functions for groups of instructions
 329 
 330   enum Predict { pt = 1, pn = 0 }; // pt = predict taken
 331 
 332   enum Membar_mask_bits { // page 184, v9
 333     StoreStore = 1 << 3,
 334     LoadStore  = 1 << 2,
 335     StoreLoad  = 1 << 1,
 336     LoadLoad   = 1 << 0,
 337 
 338     Sync       = 1 << 6,
 339     MemIssue   = 1 << 5,
 340     Lookaside  = 1 << 4
 341   };
 342 
 343   static bool is_in_wdisp_range(address a, address b, int nbits) {
 344     intptr_t d = intptr_t(b) - intptr_t(a);
 345     return is_simm(d, nbits + 2);
 346   }
 347 
 348   address target_distance(Label& L) {
 349     // Assembler::target(L) should be called only when
 350     // a branch instruction is emitted since non-bound
 351     // labels record current pc() as a branch address.
 352     if (L.is_bound()) return target(L);
 353     // Return current address for non-bound labels.
 354     return pc();
 355   }
 356 
 357   // test if label is in simm16 range in words (wdisp16).
 358   bool is_in_wdisp16_range(Label& L) {
 359     return is_in_wdisp_range(target_distance(L), pc(), 16);
 360   }
 361   // test if the distance between two addresses fits in simm30 range in words
 362   static bool is_in_wdisp30_range(address a, address b) {
 363     return is_in_wdisp_range(a, b, 30);
 364   }
 365 
 366   enum ASIs { // page 72, v9
 367     ASI_PRIMARY            = 0x80,
 368     ASI_PRIMARY_NOFAULT    = 0x82,
 369     ASI_PRIMARY_LITTLE     = 0x88,
 370     // 8x8-bit partial store
 371     ASI_PST8_PRIMARY       = 0xC0,
 372     // Block initializing store
 373     ASI_ST_BLKINIT_PRIMARY = 0xE2,
 374     // Most-Recently-Used (MRU) BIS variant
 375     ASI_ST_BLKINIT_MRU_PRIMARY = 0xF2
 376     // add more from book as needed
 377   };
 378 
 379  protected:
 380   // helpers
 381 
 382   // x is supposed to fit in a field "nbits" wide
 383   // and be sign-extended. Check the range.
 384 
 385   static void assert_signed_range(intptr_t x, int nbits) {
 386     assert(nbits == 32 || (-(1 << nbits-1) <= x  &&  x < ( 1 << nbits-1)),
 387            err_msg("value out of range: x=" INTPTR_FORMAT ", nbits=%d", x, nbits));
 388   }
 389 
 390   static void assert_signed_word_disp_range(intptr_t x, int nbits) {
 391     assert( (x & 3) == 0, "not word aligned");
 392     assert_signed_range(x, nbits + 2);
 393   }
 394 
 395   static void assert_unsigned_const(int x, int nbits) {
 396     assert( juint(x)  <  juint(1 << nbits), "unsigned constant out of range");
 397   }
 398 
 399   // fields: note bits numbered from LSB = 0,
 400   //  fields known by inclusive bit range
 401 
 402   static int fmask(juint hi_bit, juint lo_bit) {
 403     assert( hi_bit >= lo_bit  &&  0 <= lo_bit  &&  hi_bit < 32, "bad bits");
 404     return (1 << ( hi_bit-lo_bit + 1 )) - 1;
 405   }
 406 
 407   // inverse of u_field
 408 
 409   static int inv_u_field(int x, int hi_bit, int lo_bit) {
 410     juint r = juint(x) >> lo_bit;
 411     r &= fmask( hi_bit, lo_bit);
 412     return int(r);
 413   }
 414 
 415 
 416   // signed version: extract from field and sign-extend
 417 
 418   static int inv_s_field(int x, int hi_bit, int lo_bit) {
 419     int sign_shift = 31 - hi_bit;
 420     return inv_u_field( ((x << sign_shift) >> sign_shift), hi_bit, lo_bit);
 421   }
 422 
 423   // given a field that ranges from hi_bit to lo_bit (inclusive,
 424   // LSB = 0), and an unsigned value for the field,
 425   // shift it into the field
 426 
 427 #ifdef ASSERT
 428   static int u_field(int x, int hi_bit, int lo_bit) {
 429     assert( ( x & ~fmask(hi_bit, lo_bit))  == 0,
 430             "value out of range");
 431     int r = x << lo_bit;
 432     assert( inv_u_field(r, hi_bit, lo_bit) == x, "just checking");
 433     return r;
 434   }
 435 #else
 436   // make sure this is inlined as it will reduce code size significantly
 437   #define u_field(x, hi_bit, lo_bit)   ((x) << (lo_bit))
 438 #endif
 439 
 440   static int inv_op(  int x ) { return inv_u_field(x, 31, 30); }
 441   static int inv_op2( int x ) { return inv_u_field(x, 24, 22); }
 442   static int inv_op3( int x ) { return inv_u_field(x, 24, 19); }
 443   static int inv_cond( int x ){ return inv_u_field(x, 28, 25); }
 444 
 445   static bool inv_immed( int x ) { return (x & Assembler::immed(true)) != 0; }
 446 
 447   static Register inv_rd(  int x ) { return as_Register(inv_u_field(x, 29, 25)); }
 448   static Register inv_rs1( int x ) { return as_Register(inv_u_field(x, 18, 14)); }
 449   static Register inv_rs2( int x ) { return as_Register(inv_u_field(x,  4,  0)); }
 450 
 451   static int op(       int         x)  { return  u_field(x,             31, 30); }
 452   static int rd(       Register    r)  { return  u_field(r->encoding(), 29, 25); }
 453   static int fcn(      int         x)  { return  u_field(x,             29, 25); }
 454   static int op3(      int         x)  { return  u_field(x,             24, 19); }
 455   static int rs1(      Register    r)  { return  u_field(r->encoding(), 18, 14); }
 456   static int rs2(      Register    r)  { return  u_field(r->encoding(),  4,  0); }
 457   static int annul(    bool        a)  { return  u_field(a ? 1 : 0,     29, 29); }
 458   static int cond(     int         x)  { return  u_field(x,             28, 25); }
 459   static int cond_mov( int         x)  { return  u_field(x,             17, 14); }
 460   static int rcond(    RCondition  x)  { return  u_field(x,             12, 10); }
 461   static int op2(      int         x)  { return  u_field(x,             24, 22); }
 462   static int predict(  bool        p)  { return  u_field(p ? 1 : 0,     19, 19); }
 463   static int branchcc( CC       fcca)  { return  u_field(fcca,          21, 20); }
 464   static int cmpcc(    CC       fcca)  { return  u_field(fcca,          26, 25); }
 465   static int imm_asi(  int         x)  { return  u_field(x,             12,  5); }
 466   static int immed(    bool        i)  { return  u_field(i ? 1 : 0,     13, 13); }
 467   static int opf_low6( int         w)  { return  u_field(w,             10,  5); }
 468   static int opf_low5( int         w)  { return  u_field(w,              9,  5); }
 469   static int op5(      int         x)  { return  u_field(x,              8,  5); }
 470   static int trapcc(   CC         cc)  { return  u_field(cc,            12, 11); }
 471   static int sx(       int         i)  { return  u_field(i,             12, 12); } // shift x=1 means 64-bit
 472   static int opf(      int         x)  { return  u_field(x,             13,  5); }
 473 
 474   static bool is_cbcond( int x ) {
 475     return (VM_Version::has_cbcond() && (inv_cond(x) > rc_last) &&
 476             inv_op(x) == branch_op && inv_op2(x) == bpr_op2);
 477   }
 478   static bool is_cxb( int x ) {
 479     assert(is_cbcond(x), "wrong instruction");
 480     return (x & (1<<21)) != 0;
 481   }
 482   static int cond_cbcond( int         x)  { return  u_field((((x & 8)<<1) + 8 + (x & 7)), 29, 25); }
 483   static int inv_cond_cbcond(int      x)  {
 484     assert(is_cbcond(x), "wrong instruction");
 485     return inv_u_field(x, 27, 25) | (inv_u_field(x, 29, 29)<<3);
 486   }
 487 
 488   static int opf_cc(   CC          c, bool useFloat ) { return u_field((useFloat ? 0 : 4) + c, 13, 11); }
 489   static int mov_cc(   CC          c, bool useFloat ) { return u_field(useFloat ? 0 : 1,  18, 18) | u_field(c, 12, 11); }
 490 
 491   static int fd( FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 29, 25); };
 492   static int fs1(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 18, 14); };
 493   static int fs2(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa),  4,  0); };
 494   static int fs3(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 13,  9); };
 495 
 496   // some float instructions use this encoding on the op3 field
 497   static int alt_op3(int op, FloatRegisterImpl::Width w) {
 498     int r;
 499     switch(w) {
 500      case FloatRegisterImpl::S: r = op + 0;  break;
 501      case FloatRegisterImpl::D: r = op + 3;  break;
 502      case FloatRegisterImpl::Q: r = op + 2;  break;
 503      default: ShouldNotReachHere(); break;
 504     }
 505     return op3(r);
 506   }
 507 
 508 
 509   // compute inverse of simm
 510   static int inv_simm(int x, int nbits) {
 511     return (int)(x << (32 - nbits)) >> (32 - nbits);
 512   }
 513 
 514   static int inv_simm13( int x ) { return inv_simm(x, 13); }
 515 
 516   // signed immediate, in low bits, nbits long
 517   static int simm(int x, int nbits) {
 518     assert_signed_range(x, nbits);
 519     return x  &  (( 1 << nbits ) - 1);
 520   }
 521 
 522   // compute inverse of wdisp16
 523   static intptr_t inv_wdisp16(int x, intptr_t pos) {
 524     int lo = x & (( 1 << 14 ) - 1);
 525     int hi = (x >> 20) & 3;
 526     if (hi >= 2) hi |= ~1;
 527     return (((hi << 14) | lo) << 2) + pos;
 528   }
 529 
 530   // word offset, 14 bits at LSend, 2 bits at B21, B20
 531   static int wdisp16(intptr_t x, intptr_t off) {
 532     intptr_t xx = x - off;
 533     assert_signed_word_disp_range(xx, 16);
 534     int r =  (xx >> 2) & ((1 << 14) - 1)
 535            |  (  ( (xx>>(2+14)) & 3 )  <<  20 );
 536     assert( inv_wdisp16(r, off) == x,  "inverse is not inverse");
 537     return r;
 538   }
 539 
 540   // compute inverse of wdisp10
 541   static intptr_t inv_wdisp10(int x, intptr_t pos) {
 542     assert(is_cbcond(x), "wrong instruction");
 543     int lo = inv_u_field(x, 12, 5);
 544     int hi = (x >> 19) & 3;
 545     if (hi >= 2) hi |= ~1;
 546     return (((hi << 8) | lo) << 2) + pos;
 547   }
 548 
 549   // word offset for cbcond, 8 bits at [B12,B5], 2 bits at [B20,B19]
 550   static int wdisp10(intptr_t x, intptr_t off) {
 551     assert(VM_Version::has_cbcond(), "This CPU does not have CBCOND instruction");
 552     intptr_t xx = x - off;
 553     assert_signed_word_disp_range(xx, 10);
 554     int r =  ( ( (xx >>  2   ) & ((1 << 8) - 1) ) <<  5 )
 555            | ( ( (xx >> (2+8)) & 3              ) << 19 );
 556     // Have to fake cbcond instruction to pass assert in inv_wdisp10()
 557     assert(inv_wdisp10((r | op(branch_op) | cond_cbcond(rc_last+1) | op2(bpr_op2)), off) == x,  "inverse is not inverse");
 558     return r;
 559   }
 560 
 561   // word displacement in low-order nbits bits
 562 
 563   static intptr_t inv_wdisp( int x, intptr_t pos, int nbits ) {
 564     int pre_sign_extend = x & (( 1 << nbits ) - 1);
 565     int r =  pre_sign_extend >= ( 1 << (nbits-1) )
 566        ?   pre_sign_extend | ~(( 1 << nbits ) - 1)
 567        :   pre_sign_extend;
 568     return (r << 2) + pos;
 569   }
 570 
 571   static int wdisp( intptr_t x, intptr_t off, int nbits ) {
 572     intptr_t xx = x - off;
 573     assert_signed_word_disp_range(xx, nbits);
 574     int r =  (xx >> 2) & (( 1 << nbits ) - 1);
 575     assert( inv_wdisp( r, off, nbits )  ==  x, "inverse not inverse");
 576     return r;
 577   }
 578 
 579 
 580   // Extract the top 32 bits in a 64 bit word
 581   static int32_t hi32( int64_t x ) {
 582     int32_t r = int32_t( (uint64_t)x >> 32 );
 583     return r;
 584   }
 585 
 586   // given a sethi instruction, extract the constant, left-justified
 587   static int inv_hi22( int x ) {
 588     return x << 10;
 589   }
 590 
 591   // create an imm22 field, given a 32-bit left-justified constant
 592   static int hi22( int x ) {
 593     int r = int( juint(x) >> 10 );
 594     assert( (r & ~((1 << 22) - 1))  ==  0, "just checkin'");
 595     return r;
 596   }
 597 
 598   // create a low10 __value__ (not a field) for a given a 32-bit constant
 599   static int low10( int x ) {
 600     return x & ((1 << 10) - 1);
 601   }
 602 
 603   // AES crypto instructions supported only on certain processors
 604   static void aes_only() { assert( VM_Version::has_aes(), "This instruction only works on SPARC with AES instructions support"); }
 605 
 606   // SHA crypto instructions supported only on certain processors
 607   static void sha1_only()   { assert( VM_Version::has_sha1(),   "This instruction only works on SPARC with SHA1"); }
 608   static void sha256_only() { assert( VM_Version::has_sha256(), "This instruction only works on SPARC with SHA256"); }
 609   static void sha512_only() { assert( VM_Version::has_sha512(), "This instruction only works on SPARC with SHA512"); }
 610 
 611   // instruction only in VIS1
 612   static void vis1_only() { assert( VM_Version::has_vis1(), "This instruction only works on SPARC with VIS1"); }
 613 
 614   // instruction only in VIS2
 615   static void vis2_only() { assert( VM_Version::has_vis2(), "This instruction only works on SPARC with VIS2"); }
 616 
 617   // instruction only in VIS3
 618   static void vis3_only() { assert( VM_Version::has_vis3(), "This instruction only works on SPARC with VIS3"); }
 619 
 620   // instruction only in v9
 621   static void v9_only() { } // do nothing
 622 
 623   // instruction deprecated in v9
 624   static void v9_dep()  { } // do nothing for now
 625 
 626   // v8 has no CC field
 627   static void v8_no_cc(CC cc)  { if (cc)  v9_only(); }
 628 
 629  protected:
 630   // Simple delay-slot scheme:
 631   // In order to check the programmer, the assembler keeps track of deley slots.
 632   // It forbids CTIs in delay slots (conservative, but should be OK).
 633   // Also, when putting an instruction into a delay slot, you must say
 634   // asm->delayed()->add(...), in order to check that you don't omit
 635   // delay-slot instructions.
 636   // To implement this, we use a simple FSA
 637 
 638 #ifdef ASSERT
 639   #define CHECK_DELAY
 640 #endif
 641 #ifdef CHECK_DELAY
 642   enum Delay_state { no_delay, at_delay_slot, filling_delay_slot } delay_state;
 643 #endif
 644 
 645  public:
 646   // Tells assembler next instruction must NOT be in delay slot.
 647   // Use at start of multinstruction macros.
 648   void assert_not_delayed() {
 649     // This is a separate overloading to avoid creation of string constants
 650     // in non-asserted code--with some compilers this pollutes the object code.
 651 #ifdef CHECK_DELAY
 652     assert_not_delayed("next instruction should not be a delay slot");
 653 #endif
 654   }
 655   void assert_not_delayed(const char* msg) {
 656 #ifdef CHECK_DELAY
 657     assert(delay_state == no_delay, msg);
 658 #endif
 659   }
 660 
 661  protected:
 662   // Insert a nop if the previous is cbcond
 663   void insert_nop_after_cbcond() {
 664     if (UseCBCond && cbcond_before()) {
 665       nop();
 666     }
 667   }
 668   // Delay slot helpers
 669   // cti is called when emitting control-transfer instruction,
 670   // BEFORE doing the emitting.
 671   // Only effective when assertion-checking is enabled.
 672   void cti() {
 673     // A cbcond instruction immediately followed by a CTI
 674     // instruction introduces pipeline stalls, we need to avoid that.
 675     no_cbcond_before();
 676 #ifdef CHECK_DELAY
 677     assert_not_delayed("cti should not be in delay slot");
 678 #endif
 679   }
 680 
 681   // called when emitting cti with a delay slot, AFTER emitting
 682   void has_delay_slot() {
 683 #ifdef CHECK_DELAY
 684     assert_not_delayed("just checking");
 685     delay_state = at_delay_slot;
 686 #endif
 687   }
 688 
 689   // cbcond instruction should not be generated one after an other
 690   bool cbcond_before() {
 691     if (offset() == 0) return false; // it is first instruction
 692     int x = *(int*)(intptr_t(pc()) - 4); // previous instruction
 693     return is_cbcond(x);
 694   }
 695 
 696   void no_cbcond_before() {
 697     assert(offset() == 0 || !cbcond_before(), "cbcond should not follow an other cbcond");
 698   }
 699 public:
 700 
 701   bool use_cbcond(Label& L) {
 702     if (!UseCBCond || cbcond_before()) return false;
 703     intptr_t x = intptr_t(target_distance(L)) - intptr_t(pc());
 704     assert( (x & 3) == 0, "not word aligned");
 705     return is_simm12(x);
 706   }
 707 
 708   // Tells assembler you know that next instruction is delayed
 709   Assembler* delayed() {
 710 #ifdef CHECK_DELAY
 711     assert ( delay_state == at_delay_slot, "delayed instruction is not in delay slot");
 712     delay_state = filling_delay_slot;
 713 #endif
 714     return this;
 715   }
 716 
 717   void flush() {
 718 #ifdef CHECK_DELAY
 719     assert ( delay_state == no_delay, "ending code with a delay slot");
 720 #endif
 721     AbstractAssembler::flush();
 722   }
 723 
 724   inline void emit_int32(int);  // shadows AbstractAssembler::emit_int32
 725   inline void emit_data(int x) { emit_int32(x); }
 726   inline void emit_data(int, RelocationHolder const&);
 727   inline void emit_data(int, relocInfo::relocType rtype);
 728   // helper for above fcns
 729   inline void check_delay();
 730 
 731 
 732  public:
 733   // instructions, refer to page numbers in the SPARC Architecture Manual, V9
 734 
 735   // pp 135 (addc was addx in v8)
 736 
 737   inline void add(Register s1, Register s2, Register d );
 738   inline void add(Register s1, int simm13a, Register d );
 739 
 740   void addcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 741   void addcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(add_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 742   void addc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | rs2(s2) ); }
 743   void addc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 744   void addccc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 745   void addccc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 746 
 747 
 748   // 4-operand AES instructions
 749 
 750   void aes_eround01(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_eround01_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 751   void aes_eround23(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_eround23_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 752   void aes_dround01(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_dround01_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 753   void aes_dround23(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_dround23_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 754   void aes_eround01_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_eround01_l_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 755   void aes_eround23_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_eround23_l_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 756   void aes_dround01_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_dround01_l_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 757   void aes_dround23_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | fs3(s3, FloatRegisterImpl::D) | op5(aes_dround23_l_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 758   void aes_kexpand1(  FloatRegister s1, FloatRegister s2, int imm5a, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes4_op3) | fs1(s1, FloatRegisterImpl::D) | u_field(imm5a, 13, 9) | op5(aes_kexpand1_op5) | fs2(s2, FloatRegisterImpl::D) ); }
 759 
 760 
 761   // 3-operand AES instructions
 762 
 763   void aes_kexpand0(  FloatRegister s1, FloatRegister s2, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes3_op3) | fs1(s1, FloatRegisterImpl::D) | opf(aes_kexpand0_opf) | fs2(s2, FloatRegisterImpl::D) ); }
 764   void aes_kexpand2(  FloatRegister s1, FloatRegister s2, FloatRegister d ) { aes_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(aes3_op3) | fs1(s1, FloatRegisterImpl::D) | opf(aes_kexpand2_opf) | fs2(s2, FloatRegisterImpl::D) ); }
 765 
 766   // pp 136
 767 
 768   inline void bpr(RCondition c, bool a, Predict p, Register s1, address d, relocInfo::relocType rt = relocInfo::none);
 769   inline void bpr(RCondition c, bool a, Predict p, Register s1, Label& L);
 770 
 771   // compare and branch
 772   inline void cbcond(Condition c, CC cc, Register s1, Register s2, Label& L);
 773   inline void cbcond(Condition c, CC cc, Register s1, int simm5, Label& L);
 774 
 775  protected: // use MacroAssembler::br instead
 776 
 777   // pp 138
 778 
 779   inline void fb( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 780   inline void fb( Condition c, bool a, Label& L );
 781 
 782   // pp 141
 783 
 784   inline void fbp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 785   inline void fbp( Condition c, bool a, CC cc, Predict p, Label& L );
 786 
 787   // pp 144
 788 
 789   inline void br( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 790   inline void br( Condition c, bool a, Label& L );
 791 
 792   // pp 146
 793 
 794   inline void bp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 795   inline void bp( Condition c, bool a, CC cc, Predict p, Label& L );
 796 
 797   // pp 149
 798 
 799   inline void call( address d,  relocInfo::relocType rt = relocInfo::runtime_call_type );
 800   inline void call( Label& L,   relocInfo::relocType rt = relocInfo::runtime_call_type );
 801 
 802  public:
 803 
 804   // pp 150
 805 
 806   // These instructions compare the contents of s2 with the contents of
 807   // memory at address in s1. If the values are equal, the contents of memory
 808   // at address s1 is swapped with the data in d. If the values are not equal,
 809   // the the contents of memory at s1 is loaded into d, without the swap.
 810 
 811   void casa(  Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(casa_op3 ) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
 812   void casxa( Register s1, Register s2, Register d, int ia = -1 ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(casxa_op3) | rs1(s1) | (ia == -1  ? immed(true) : imm_asi(ia)) | rs2(s2)); }
 813 
 814   // pp 152
 815 
 816   void udiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | rs2(s2)); }
 817   void udiv(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 818   void sdiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | rs2(s2)); }
 819   void sdiv(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 820   void udivcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
 821   void udivcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 822   void sdivcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | rs2(s2)); }
 823   void sdivcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 824 
 825   // pp 155
 826 
 827   void done()  { v9_only();  cti();  emit_int32( op(arith_op) | fcn(0) | op3(done_op3) ); }
 828   void retry() { v9_only();  cti();  emit_int32( op(arith_op) | fcn(1) | op3(retry_op3) ); }
 829 
 830   // pp 156
 831 
 832   void fadd( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x40 + w) | fs2(s2, w)); }
 833   void fsub( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | fs1(s1, w) | opf(0x44 + w) | fs2(s2, w)); }
 834 
 835   // pp 157
 836 
 837   void fcmp(  FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { emit_int32( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x50 + w) | fs2(s2, w)); }
 838   void fcmpe( FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2) { emit_int32( op(arith_op) | cmpcc(cc) | op3(fpop2_op3) | fs1(s1, w) | opf(0x54 + w) | fs2(s2, w)); }
 839 
 840   // pp 159
 841 
 842   void ftox( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(fpop1_op3) | opf(0x80 + w) | fs2(s, w)); }
 843   void ftoi( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(fpop1_op3) | opf(0xd0 + w) | fs2(s, w)); }
 844 
 845   // pp 160
 846 
 847   void ftof( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | opf(0xc0 + sw + dw*4) | fs2(s, sw)); }
 848 
 849   // pp 161
 850 
 851   void fxtof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x80 + w*4) | fs2(s, FloatRegisterImpl::D)); }
 852   void fitof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) {             emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0xc0 + w*4) | fs2(s, FloatRegisterImpl::S)); }
 853 
 854   // pp 162
 855 
 856   void fmov( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x00 + w) | fs2(s, w)); }
 857 
 858   void fneg( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x04 + w) | fs2(s, w)); }
 859 
 860   void fabs( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x08 + w) | fs2(s, w)); }
 861 
 862   // pp 163
 863 
 864   void fmul( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x48 + w)         | fs2(s2, w)); }
 865   void fmul( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw,  FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, dw) | op3(fpop1_op3) | fs1(s1, sw) | opf(0x60 + sw + dw*4) | fs2(s2, sw)); }
 866   void fdiv( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w)  | op3(fpop1_op3) | fs1(s1, w)  | opf(0x4c + w)         | fs2(s2, w)); }
 867 
 868   // FXORs/FXORd instructions
 869 
 870   void fxor( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d ) { vis1_only(); emit_int32( op(arith_op) | fd(d, w) | op3(flog3_op3) | fs1(s1, w) | opf(0x6E - w) | fs2(s2, w)); }
 871 
 872   // pp 164
 873 
 874   void fsqrt( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d ) { emit_int32( op(arith_op) | fd(d, w) | op3(fpop1_op3) | opf(0x28 + w) | fs2(s, w)); }
 875 
 876   // pp 165
 877 
 878   inline void flush( Register s1, Register s2 );
 879   inline void flush( Register s1, int simm13a);
 880 
 881   // pp 167
 882 
 883   void flushw() { v9_only();  emit_int32( op(arith_op) | op3(flushw_op3) ); }
 884 
 885   // pp 168
 886 
 887   void illtrap( int const22a) { if (const22a != 0) v9_only();  emit_int32( op(branch_op) | u_field(const22a, 21, 0) ); }
 888   // v8 unimp == illtrap(0)
 889 
 890   // pp 169
 891 
 892   void impdep1( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep1_op3) | u_field(const19a, 18, 0)); }
 893   void impdep2( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep2_op3) | u_field(const19a, 18, 0)); }
 894 
 895   // pp 170
 896 
 897   void jmpl( Register s1, Register s2, Register d );
 898   void jmpl( Register s1, int simm13a, Register d, RelocationHolder const& rspec = RelocationHolder() );
 899 
 900   // 171
 901 
 902   inline void ldf(FloatRegisterImpl::Width w, Register s1, Register s2, FloatRegister d);
 903   inline void ldf(FloatRegisterImpl::Width w, Register s1, int simm13a, FloatRegister d, RelocationHolder const& rspec = RelocationHolder());
 904 
 905 
 906   inline void ldfsr(  Register s1, Register s2 );
 907   inline void ldfsr(  Register s1, int simm13a);
 908   inline void ldxfsr( Register s1, Register s2 );
 909   inline void ldxfsr( Register s1, int simm13a);
 910 
 911   // 173
 912 
 913   void ldfa(  FloatRegisterImpl::Width w, Register s1, Register s2, int ia, FloatRegister d ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 914   void ldfa(  FloatRegisterImpl::Width w, Register s1, int simm13a,         FloatRegister d ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(ldf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 915 
 916   // pp 175, lduw is ld on v8
 917 
 918   inline void ldsb(  Register s1, Register s2, Register d );
 919   inline void ldsb(  Register s1, int simm13a, Register d);
 920   inline void ldsh(  Register s1, Register s2, Register d );
 921   inline void ldsh(  Register s1, int simm13a, Register d);
 922   inline void ldsw(  Register s1, Register s2, Register d );
 923   inline void ldsw(  Register s1, int simm13a, Register d);
 924   inline void ldub(  Register s1, Register s2, Register d );
 925   inline void ldub(  Register s1, int simm13a, Register d);
 926   inline void lduh(  Register s1, Register s2, Register d );
 927   inline void lduh(  Register s1, int simm13a, Register d);
 928   inline void lduw(  Register s1, Register s2, Register d );
 929   inline void lduw(  Register s1, int simm13a, Register d);
 930   inline void ldx(   Register s1, Register s2, Register d );
 931   inline void ldx(   Register s1, int simm13a, Register d);
 932   inline void ldd(   Register s1, Register s2, Register d );
 933   inline void ldd(   Register s1, int simm13a, Register d);
 934 
 935   // pp 177
 936 
 937   void ldsba(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 938   void ldsba(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 939   void ldsha(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 940   void ldsha(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldsh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 941   void ldswa(  Register s1, Register s2, int ia, Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 942   void ldswa(  Register s1, int simm13a,         Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldsw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 943   void lduba(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 944   void lduba(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(ldub_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 945   void lduha(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 946   void lduha(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduh_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 947   void lduwa(  Register s1, Register s2, int ia, Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 948   void lduwa(  Register s1, int simm13a,         Register d ) {             emit_int32( op(ldst_op) | rd(d) | op3(lduw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 949   void ldxa(   Register s1, Register s2, int ia, Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
 950   void ldxa(   Register s1, int simm13a,         Register d ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(ldx_op3  | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 951 
 952   // pp 181
 953 
 954   void and3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | rs2(s2) ); }
 955   void and3(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 956   void andcc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 957   void andcc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 958   void andn(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | rs2(s2) ); }
 959   void andn(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 960   void andncc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 961   void andncc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 962   void or3(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | rs2(s2) ); }
 963   void or3(     Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 964   void orcc(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 965   void orcc(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3   | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 966   void orn(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | rs2(s2) ); }
 967   void orn(     Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 968   void orncc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 969   void orncc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 970   void xor3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | rs2(s2) ); }
 971   void xor3(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 972   void xorcc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 973   void xorcc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3  | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 974   void xnor(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | rs2(s2) ); }
 975   void xnor(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 976   void xnorcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
 977   void xnorcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
 978 
 979   // pp 183
 980 
 981   void membar( Membar_mask_bits const7a ) { v9_only(); emit_int32( op(arith_op) | op3(membar_op3) | rs1(O7) | immed(true) | u_field( int(const7a), 6, 0)); }
 982 
 983   // pp 185
 984 
 985   void fmov( FloatRegisterImpl::Width w, Condition c,  bool floatCC, CC cca, FloatRegister s2, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop2_op3) | cond_mov(c) | opf_cc(cca, floatCC) | opf_low6(w) | fs2(s2, w)); }
 986 
 987   // pp 189
 988 
 989   void fmov( FloatRegisterImpl::Width w, RCondition c, Register s1,  FloatRegister s2, FloatRegister d ) { v9_only();  emit_int32( op(arith_op) | fd(d, w) | op3(fpop2_op3) | rs1(s1) | rcond(c) | opf_low5(4 + w) | fs2(s2, w)); }
 990 
 991   // pp 191
 992 
 993   void movcc( Condition c, bool floatCC, CC cca, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | rs2(s2) ); }
 994   void movcc( Condition c, bool floatCC, CC cca, int simm11a, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movcc_op3) | mov_cc(cca, floatCC) | cond_mov(c) | immed(true) | simm(simm11a, 11) ); }
 995 
 996   // pp 195
 997 
 998   void movr( RCondition c, Register s1, Register s2,  Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | rs2(s2) ); }
 999   void movr( RCondition c, Register s1, int simm10a,  Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(movr_op3) | rs1(s1) | rcond(c) | immed(true) | simm(simm10a, 10) ); }
1000 
1001   // pp 196
1002 
1003   void mulx(  Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | rs2(s2) ); }
1004   void mulx(  Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1005   void sdivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | rs2(s2) ); }
1006   void sdivx( Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1007   void udivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | rs2(s2) ); }
1008   void udivx( Register s1, int simm13a, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1009 
1010   // pp 197
1011 
1012   void umul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | rs2(s2) ); }
1013   void umul(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1014   void smul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | rs2(s2) ); }
1015   void smul(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1016   void umulcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
1017   void umulcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1018   void smulcc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
1019   void smulcc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1020 
1021   // pp 201
1022 
1023   void nop() { emit_int32( op(branch_op) | op2(sethi_op2) ); }
1024 
1025 
1026   // pp 202
1027 
1028   void popc( Register s,  Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | rs2(s)); }
1029   void popc( int simm13a, Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | immed(true) | simm(simm13a, 13)); }
1030 
1031   // pp 203
1032 
1033   void prefetch(   Register s1, Register s2, PrefetchFcn f) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | rs2(s2) ); }
1034   void prefetch(   Register s1, int simm13a, PrefetchFcn f) { v9_only();  emit_data( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | immed(true) | simm(simm13a, 13)); }
1035 
1036   void prefetcha(  Register s1, Register s2, int ia, PrefetchFcn f ) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1037   void prefetcha(  Register s1, int simm13a,         PrefetchFcn f ) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1038 
1039   // pp 208
1040 
1041   // not implementing read privileged register
1042 
1043   inline void rdy(    Register d) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(0, 18, 14)); }
1044   inline void rdccr(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(2, 18, 14)); }
1045   inline void rdasi(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(3, 18, 14)); }
1046   inline void rdtick( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(4, 18, 14)); } // Spoon!
1047   inline void rdpc(   Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(5, 18, 14)); }
1048   inline void rdfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(6, 18, 14)); }
1049 
1050   // pp 213
1051 
1052   inline void rett( Register s1, Register s2);
1053   inline void rett( Register s1, int simm13a, relocInfo::relocType rt = relocInfo::none);
1054 
1055   // pp 214
1056 
1057   void save(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | rs2(s2) ); }
1058   void save(    Register s1, int simm13a, Register d ) {
1059     // make sure frame is at least large enough for the register save area
1060     assert(-simm13a >= 16 * wordSize, "frame too small");
1061     emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) );
1062   }
1063 
1064   void restore( Register s1 = G0,  Register s2 = G0, Register d = G0 ) { emit_int32( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | rs2(s2) ); }
1065   void restore( Register s1,       int simm13a,      Register d      ) { emit_int32( op(arith_op) | rd(d) | op3(restore_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1066 
1067   // pp 216
1068 
1069   void saved()    { v9_only();  emit_int32( op(arith_op) | fcn(0) | op3(saved_op3)); }
1070   void restored() { v9_only();  emit_int32( op(arith_op) | fcn(1) | op3(saved_op3)); }
1071 
1072   // pp 217
1073 
1074   inline void sethi( int imm22a, Register d, RelocationHolder const& rspec = RelocationHolder() );
1075   // pp 218
1076 
1077   void sll(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1078   void sll(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
1079   void srl(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1080   void srl(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
1081   void sra(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1082   void sra(  Register s1, int imm5a,   Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | immed(true) | u_field(imm5a, 4, 0) ); }
1083 
1084   void sllx( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
1085   void sllx( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
1086   void srlx( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
1087   void srlx( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
1088   void srax( Register s1, Register s2, Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | rs2(s2) ); }
1089   void srax( Register s1, int imm6a,   Register d ) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(1) | immed(true) | u_field(imm6a, 5, 0) ); }
1090 
1091   // pp 220
1092 
1093   void sir( int simm13a ) { emit_int32( op(arith_op) | fcn(15) | op3(sir_op3) | immed(true) | simm(simm13a, 13)); }
1094 
1095   // pp 221
1096 
1097   void stbar() { emit_int32( op(arith_op) | op3(membar_op3) | u_field(15, 18, 14)); }
1098 
1099   // pp 222
1100 
1101   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2);
1102   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a);
1103 
1104   inline void stfsr(  Register s1, Register s2 );
1105   inline void stfsr(  Register s1, int simm13a);
1106   inline void stxfsr( Register s1, Register s2 );
1107   inline void stxfsr( Register s1, int simm13a);
1108 
1109   //  pp 224
1110 
1111   void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2, int ia ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1112   void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a         ) { v9_only();  emit_int32( op(ldst_op) | fd(d, w) | alt_op3(stf_op3 | alt_bit_op3, w) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1113 
1114   // p 226
1115 
1116   inline void stb(  Register d, Register s1, Register s2 );
1117   inline void stb(  Register d, Register s1, int simm13a);
1118   inline void sth(  Register d, Register s1, Register s2 );
1119   inline void sth(  Register d, Register s1, int simm13a);
1120   inline void stw(  Register d, Register s1, Register s2 );
1121   inline void stw(  Register d, Register s1, int simm13a);
1122   inline void stx(  Register d, Register s1, Register s2 );
1123   inline void stx(  Register d, Register s1, int simm13a);
1124   inline void std(  Register d, Register s1, Register s2 );
1125   inline void std(  Register d, Register s1, int simm13a);
1126 
1127   // pp 177
1128 
1129   void stba(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1130   void stba(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(stb_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1131   void stha(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1132   void stha(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(sth_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1133   void stwa(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1134   void stwa(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(stw_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1135   void stxa(  Register d, Register s1, Register s2, int ia ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1136   void stxa(  Register d, Register s1, int simm13a         ) { v9_only();  emit_int32( op(ldst_op) | rd(d) | op3(stx_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1137   void stda(  Register d, Register s1, Register s2, int ia ) {             emit_int32( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1138   void stda(  Register d, Register s1, int simm13a         ) {             emit_int32( op(ldst_op) | rd(d) | op3(std_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1139 
1140   // pp 230
1141 
1142   void sub(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | rs2(s2) ); }
1143   void sub(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1144 
1145   void subcc(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | rs2(s2) ); }
1146   void subcc(  Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3 | cc_bit_op3 ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1147   void subc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | rs2(s2) ); }
1148   void subc(   Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1149   void subccc( Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | rs2(s2) ); }
1150   void subccc( Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3 | cc_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1151 
1152   // pp 231
1153 
1154   inline void swap( Register s1, Register s2, Register d );
1155   inline void swap( Register s1, int simm13a, Register d);
1156 
1157   // pp 232
1158 
1159   void swapa(   Register s1, Register s2, int ia, Register d ) { v9_dep();  emit_int32( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | imm_asi(ia) | rs2(s2) ); }
1160   void swapa(   Register s1, int simm13a,         Register d ) { v9_dep();  emit_int32( op(ldst_op) | rd(d) | op3(swap_op3 | alt_bit_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1161 
1162   // pp 234, note op in book is wrong, see pp 268
1163 
1164   void taddcc(    Register s1, Register s2, Register d ) {            emit_int32( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | rs2(s2) ); }
1165   void taddcc(    Register s1, int simm13a, Register d ) {            emit_int32( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1166 
1167   // pp 235
1168 
1169   void tsubcc(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | rs2(s2) ); }
1170   void tsubcc(    Register s1, int simm13a, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | immed(true) | simm(simm13a, 13) ); }
1171 
1172   // pp 237
1173 
1174   void trap( Condition c, CC cc, Register s1, Register s2 ) { emit_int32( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | rs2(s2)); }
1175   void trap( Condition c, CC cc, Register s1, int trapa   ) { emit_int32( op(arith_op) | cond(c) | op3(trap_op3) | rs1(s1) | trapcc(cc) | immed(true) | u_field(trapa, 6, 0)); }
1176   // simple uncond. trap
1177   void trap( int trapa ) { trap( always, icc, G0, trapa ); }
1178 
1179   // pp 239 omit write priv register for now
1180 
1181   inline void wry(    Register d) { v9_dep();  emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(0, 29, 25)); }
1182   inline void wrccr(Register s) { v9_only(); emit_int32( op(arith_op) | rs1(s) | op3(wrreg_op3) | u_field(2, 29, 25)); }
1183   inline void wrccr(Register s, int simm13a) { v9_only(); emit_int32( op(arith_op) |
1184                                                                            rs1(s) |
1185                                                                            op3(wrreg_op3) |
1186                                                                            u_field(2, 29, 25) |
1187                                                                            immed(true) |
1188                                                                            simm(simm13a, 13)); }
1189   inline void wrasi(Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(3, 29, 25)); }
1190   // wrasi(d, imm) stores (d xor imm) to asi
1191   inline void wrasi(Register d, int simm13a) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) |
1192                                                u_field(3, 29, 25) | immed(true) | simm(simm13a, 13)); }
1193   inline void wrfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(6, 29, 25)); }
1194 
1195   //  VIS1 instructions
1196 
1197   void alignaddr( Register s1, Register s2, Register d ) { vis1_only(); emit_int32( op(arith_op) | rd(d) | op3(alignaddr_op3) | rs1(s1) | opf(alignaddr_opf) | rs2(s2)); }
1198 
1199   void faligndata( FloatRegister s1, FloatRegister s2, FloatRegister d ) { vis1_only(); emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(faligndata_op3) | fs1(s1, FloatRegisterImpl::D) | opf(faligndata_opf) | fs2(s2, FloatRegisterImpl::D)); }
1200 
1201   void fsrc2( FloatRegisterImpl::Width w, FloatRegister s2, FloatRegister d ) { vis1_only(); emit_int32( op(arith_op) | fd(d, w) | op3(fsrc_op3) | opf(0x7A - w) | fs2(s2, w)); }
1202 
1203   void stpartialf( Register s1, Register s2, FloatRegister d, int ia = -1 ) { vis1_only(); emit_int32( op(ldst_op) | fd(d, FloatRegisterImpl::D) | op3(stpartialf_op3) | rs1(s1) | imm_asi(ia) | rs2(s2)); }
1204 
1205   //  VIS2 instructions
1206 
1207   void edge8n( Register s1, Register s2, Register d ) { vis2_only(); emit_int32( op(arith_op) | rd(d) | op3(edge_op3) | rs1(s1) | opf(edge8n_opf) | rs2(s2)); }
1208 
1209   // VIS3 instructions
1210 
1211   void movstosw( FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstosw_opf) | fs2(s, FloatRegisterImpl::S)); }
1212   void movstouw( FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mstouw_opf) | fs2(s, FloatRegisterImpl::S)); }
1213   void movdtox(  FloatRegister s, Register d ) { vis3_only();  emit_int32( op(arith_op) | rd(d) | op3(mftoi_op3) | opf(mdtox_opf) | fs2(s, FloatRegisterImpl::D)); }
1214 
1215   void movwtos( Register s, FloatRegister d ) { vis3_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::S) | op3(mftoi_op3) | opf(mwtos_opf) | rs2(s)); }
1216   void movxtod( Register s, FloatRegister d ) { vis3_only();  emit_int32( op(arith_op) | fd(d, FloatRegisterImpl::D) | op3(mftoi_op3) | opf(mxtod_opf) | rs2(s)); }
1217 
1218   void xmulx(Register s1, Register s2, Register d) { vis3_only(); emit_int32( op(arith_op) | rd(d) | op3(xmulx_op3) | rs1(s1) | opf(xmulx_opf) | rs2(s2)); }
1219   void xmulxhi(Register s1, Register s2, Register d) { vis3_only(); emit_int32( op(arith_op) | rd(d) | op3(xmulx_op3) | rs1(s1) | opf(xmulxhi_opf) | rs2(s2)); }
1220 
1221   // Crypto SHA instructions
1222 
1223   void sha1()   { sha1_only();    emit_int32( op(arith_op) | op3(sha_op3) | opf(sha1_opf)); }
1224   void sha256() { sha256_only();  emit_int32( op(arith_op) | op3(sha_op3) | opf(sha256_opf)); }
1225   void sha512() { sha512_only();  emit_int32( op(arith_op) | op3(sha_op3) | opf(sha512_opf)); }
1226 
1227   // Creation
1228   Assembler(CodeBuffer* code) : AbstractAssembler(code) {
1229 #ifdef CHECK_DELAY
1230     delay_state = no_delay;
1231 #endif
1232   }
1233 };
1234 
1235 #endif // CPU_SPARC_VM_ASSEMBLER_SPARC_HPP