1 /*
   2  * Copyright (c) 1997, 2014, 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     alignaddr_op3  = 0x36,
 127     faligndata_op3 = 0x36,
 128     flog3_op3    = 0x36,
 129     edge_op3     = 0x36,
 130     fsrc_op3     = 0x36,
 131     impdep2_op3  = 0x37,
 132     stpartialf_op3 = 0x37,
 133     jmpl_op3     = 0x38,
 134     rett_op3     = 0x39,
 135     trap_op3     = 0x3a,
 136     flush_op3    = 0x3b,
 137     save_op3     = 0x3c,
 138     restore_op3  = 0x3d,
 139     done_op3     = 0x3e,
 140     retry_op3    = 0x3e,
 141 
 142     lduw_op3     = 0x00,
 143     ldub_op3     = 0x01,
 144     lduh_op3     = 0x02,
 145     ldd_op3      = 0x03,
 146     stw_op3      = 0x04,
 147     stb_op3      = 0x05,
 148     sth_op3      = 0x06,
 149     std_op3      = 0x07,
 150     ldsw_op3     = 0x08,
 151     ldsb_op3     = 0x09,
 152     ldsh_op3     = 0x0a,
 153     ldx_op3      = 0x0b,
 154 
 155     stx_op3      = 0x0e,
 156     swap_op3     = 0x0f,
 157 
 158     stwa_op3     = 0x14,
 159     stxa_op3     = 0x1e,
 160 
 161     ldf_op3      = 0x20,
 162     ldfsr_op3    = 0x21,
 163     ldqf_op3     = 0x22,
 164     lddf_op3     = 0x23,
 165     stf_op3      = 0x24,
 166     stfsr_op3    = 0x25,
 167     stqf_op3     = 0x26,
 168     stdf_op3     = 0x27,
 169 
 170     prefetch_op3 = 0x2d,
 171 
 172     casa_op3     = 0x3c,
 173     casxa_op3    = 0x3e,
 174 
 175     mftoi_op3    = 0x36,
 176 
 177     alt_bit_op3  = 0x10,
 178      cc_bit_op3  = 0x10
 179   };
 180 
 181   enum opfs {
 182     // selected opfs
 183     edge8n_opf         = 0x01,
 184 
 185     fmovs_opf          = 0x01,
 186     fmovd_opf          = 0x02,
 187 
 188     fnegs_opf          = 0x05,
 189     fnegd_opf          = 0x06,
 190 
 191     alignaddr_opf      = 0x18,
 192 
 193     fadds_opf          = 0x41,
 194     faddd_opf          = 0x42,
 195     fsubs_opf          = 0x45,
 196     fsubd_opf          = 0x46,
 197 
 198     faligndata_opf     = 0x48,
 199 
 200     fmuls_opf          = 0x49,
 201     fmuld_opf          = 0x4a,
 202     fdivs_opf          = 0x4d,
 203     fdivd_opf          = 0x4e,
 204 
 205     fcmps_opf          = 0x51,
 206     fcmpd_opf          = 0x52,
 207 
 208     fstox_opf          = 0x81,
 209     fdtox_opf          = 0x82,
 210     fxtos_opf          = 0x84,
 211     fxtod_opf          = 0x88,
 212     fitos_opf          = 0xc4,
 213     fdtos_opf          = 0xc6,
 214     fitod_opf          = 0xc8,
 215     fstod_opf          = 0xc9,
 216     fstoi_opf          = 0xd1,
 217     fdtoi_opf          = 0xd2,
 218 
 219     mdtox_opf          = 0x110,
 220     mstouw_opf         = 0x111,
 221     mstosw_opf         = 0x113,
 222     mxtod_opf          = 0x118,
 223     mwtos_opf          = 0x119,
 224 
 225     aes_kexpand0_opf   = 0x130,
 226     aes_kexpand2_opf   = 0x131
 227   };
 228 
 229   enum op5s {
 230     aes_eround01_op5     = 0x00,
 231     aes_eround23_op5     = 0x01,
 232     aes_dround01_op5     = 0x02,
 233     aes_dround23_op5     = 0x03,
 234     aes_eround01_l_op5   = 0x04,
 235     aes_eround23_l_op5   = 0x05,
 236     aes_dround01_l_op5   = 0x06,
 237     aes_dround23_l_op5   = 0x07,
 238     aes_kexpand1_op5     = 0x08
 239   };
 240 
 241   enum RCondition {  rc_z = 1,  rc_lez = 2,  rc_lz = 3, rc_nz = 5, rc_gz = 6, rc_gez = 7, rc_last = rc_gez  };
 242 
 243   enum Condition {
 244      // for FBfcc & FBPfcc instruction
 245     f_never                     = 0,
 246     f_notEqual                  = 1,
 247     f_notZero                   = 1,
 248     f_lessOrGreater             = 2,
 249     f_unorderedOrLess           = 3,
 250     f_less                      = 4,
 251     f_unorderedOrGreater        = 5,
 252     f_greater                   = 6,
 253     f_unordered                 = 7,
 254     f_always                    = 8,
 255     f_equal                     = 9,
 256     f_zero                      = 9,
 257     f_unorderedOrEqual          = 10,
 258     f_greaterOrEqual            = 11,
 259     f_unorderedOrGreaterOrEqual = 12,
 260     f_lessOrEqual               = 13,
 261     f_unorderedOrLessOrEqual    = 14,
 262     f_ordered                   = 15,
 263 
 264     // V8 coproc, pp 123 v8 manual
 265 
 266     cp_always  = 8,
 267     cp_never   = 0,
 268     cp_3       = 7,
 269     cp_2       = 6,
 270     cp_2or3    = 5,
 271     cp_1       = 4,
 272     cp_1or3    = 3,
 273     cp_1or2    = 2,
 274     cp_1or2or3 = 1,
 275     cp_0       = 9,
 276     cp_0or3    = 10,
 277     cp_0or2    = 11,
 278     cp_0or2or3 = 12,
 279     cp_0or1    = 13,
 280     cp_0or1or3 = 14,
 281     cp_0or1or2 = 15,
 282 
 283 
 284     // for integers
 285 
 286     never                 =  0,
 287     equal                 =  1,
 288     zero                  =  1,
 289     lessEqual             =  2,
 290     less                  =  3,
 291     lessEqualUnsigned     =  4,
 292     lessUnsigned          =  5,
 293     carrySet              =  5,
 294     negative              =  6,
 295     overflowSet           =  7,
 296     always                =  8,
 297     notEqual              =  9,
 298     notZero               =  9,
 299     greater               =  10,
 300     greaterEqual          =  11,
 301     greaterUnsigned       =  12,
 302     greaterEqualUnsigned  =  13,
 303     carryClear            =  13,
 304     positive              =  14,
 305     overflowClear         =  15
 306   };
 307 
 308   enum CC {
 309     icc  = 0,  xcc  = 2,
 310     // ptr_cc is the correct condition code for a pointer or intptr_t:
 311     ptr_cc = NOT_LP64(icc) LP64_ONLY(xcc),
 312     fcc0 = 0,  fcc1 = 1, fcc2 = 2, fcc3 = 3
 313   };
 314 
 315   enum PrefetchFcn {
 316     severalReads = 0,  oneRead = 1,  severalWritesAndPossiblyReads = 2, oneWrite = 3, page = 4
 317   };
 318 
 319  public:
 320   // Helper functions for groups of instructions
 321 
 322   enum Predict { pt = 1, pn = 0 }; // pt = predict taken
 323 
 324   enum Membar_mask_bits { // page 184, v9
 325     StoreStore = 1 << 3,
 326     LoadStore  = 1 << 2,
 327     StoreLoad  = 1 << 1,
 328     LoadLoad   = 1 << 0,
 329 
 330     Sync       = 1 << 6,
 331     MemIssue   = 1 << 5,
 332     Lookaside  = 1 << 4
 333   };
 334 
 335   static bool is_in_wdisp_range(address a, address b, int nbits) {
 336     intptr_t d = intptr_t(b) - intptr_t(a);
 337     return is_simm(d, nbits + 2);
 338   }
 339 
 340   address target_distance(Label& L) {
 341     // Assembler::target(L) should be called only when
 342     // a branch instruction is emitted since non-bound
 343     // labels record current pc() as a branch address.
 344     if (L.is_bound()) return target(L);
 345     // Return current address for non-bound labels.
 346     return pc();
 347   }
 348 
 349   // test if label is in simm16 range in words (wdisp16).
 350   bool is_in_wdisp16_range(Label& L) {
 351     return is_in_wdisp_range(target_distance(L), pc(), 16);
 352   }
 353   // test if the distance between two addresses fits in simm30 range in words
 354   static bool is_in_wdisp30_range(address a, address b) {
 355     return is_in_wdisp_range(a, b, 30);
 356   }
 357 
 358   enum ASIs { // page 72, v9
 359     ASI_PRIMARY            = 0x80,
 360     ASI_PRIMARY_NOFAULT    = 0x82,
 361     ASI_PRIMARY_LITTLE     = 0x88,
 362     // 8x8-bit partial store
 363     ASI_PST8_PRIMARY       = 0xC0,
 364     // Block initializing store
 365     ASI_ST_BLKINIT_PRIMARY = 0xE2,
 366     // Most-Recently-Used (MRU) BIS variant
 367     ASI_ST_BLKINIT_MRU_PRIMARY = 0xF2
 368     // add more from book as needed
 369   };
 370 
 371  protected:
 372   // helpers
 373 
 374   // x is supposed to fit in a field "nbits" wide
 375   // and be sign-extended. Check the range.
 376 
 377   static void assert_signed_range(intptr_t x, int nbits) {
 378     assert(nbits == 32 || (-(1 << nbits-1) <= x  &&  x < ( 1 << nbits-1)),
 379            err_msg("value out of range: x=" INTPTR_FORMAT ", nbits=%d", x, nbits));
 380   }
 381 
 382   static void assert_signed_word_disp_range(intptr_t x, int nbits) {
 383     assert( (x & 3) == 0, "not word aligned");
 384     assert_signed_range(x, nbits + 2);
 385   }
 386 
 387   static void assert_unsigned_const(int x, int nbits) {
 388     assert( juint(x)  <  juint(1 << nbits), "unsigned constant out of range");
 389   }
 390 
 391   // fields: note bits numbered from LSB = 0,
 392   //  fields known by inclusive bit range
 393 
 394   static int fmask(juint hi_bit, juint lo_bit) {
 395     assert( hi_bit >= lo_bit  &&  0 <= lo_bit  &&  hi_bit < 32, "bad bits");
 396     return (1 << ( hi_bit-lo_bit + 1 )) - 1;
 397   }
 398 
 399   // inverse of u_field
 400 
 401   static int inv_u_field(int x, int hi_bit, int lo_bit) {
 402     juint r = juint(x) >> lo_bit;
 403     r &= fmask( hi_bit, lo_bit);
 404     return int(r);
 405   }
 406 
 407 
 408   // signed version: extract from field and sign-extend
 409 
 410   static int inv_s_field(int x, int hi_bit, int lo_bit) {
 411     int sign_shift = 31 - hi_bit;
 412     return inv_u_field( ((x << sign_shift) >> sign_shift), hi_bit, lo_bit);
 413   }
 414 
 415   // given a field that ranges from hi_bit to lo_bit (inclusive,
 416   // LSB = 0), and an unsigned value for the field,
 417   // shift it into the field
 418 
 419 #ifdef ASSERT
 420   static int u_field(int x, int hi_bit, int lo_bit) {
 421     assert( ( x & ~fmask(hi_bit, lo_bit))  == 0,
 422             "value out of range");
 423     int r = x << lo_bit;
 424     assert( inv_u_field(r, hi_bit, lo_bit) == x, "just checking");
 425     return r;
 426   }
 427 #else
 428   // make sure this is inlined as it will reduce code size significantly
 429   #define u_field(x, hi_bit, lo_bit)   ((x) << (lo_bit))
 430 #endif
 431 
 432   static int inv_op(  int x ) { return inv_u_field(x, 31, 30); }
 433   static int inv_op2( int x ) { return inv_u_field(x, 24, 22); }
 434   static int inv_op3( int x ) { return inv_u_field(x, 24, 19); }
 435   static int inv_cond( int x ){ return inv_u_field(x, 28, 25); }
 436 
 437   static bool inv_immed( int x ) { return (x & Assembler::immed(true)) != 0; }
 438 
 439   static Register inv_rd(  int x ) { return as_Register(inv_u_field(x, 29, 25)); }
 440   static Register inv_rs1( int x ) { return as_Register(inv_u_field(x, 18, 14)); }
 441   static Register inv_rs2( int x ) { return as_Register(inv_u_field(x,  4,  0)); }
 442 
 443   static int op(       int         x)  { return  u_field(x,             31, 30); }
 444   static int rd(       Register    r)  { return  u_field(r->encoding(), 29, 25); }
 445   static int fcn(      int         x)  { return  u_field(x,             29, 25); }
 446   static int op3(      int         x)  { return  u_field(x,             24, 19); }
 447   static int rs1(      Register    r)  { return  u_field(r->encoding(), 18, 14); }
 448   static int rs2(      Register    r)  { return  u_field(r->encoding(),  4,  0); }
 449   static int annul(    bool        a)  { return  u_field(a ? 1 : 0,     29, 29); }
 450   static int cond(     int         x)  { return  u_field(x,             28, 25); }
 451   static int cond_mov( int         x)  { return  u_field(x,             17, 14); }
 452   static int rcond(    RCondition  x)  { return  u_field(x,             12, 10); }
 453   static int op2(      int         x)  { return  u_field(x,             24, 22); }
 454   static int predict(  bool        p)  { return  u_field(p ? 1 : 0,     19, 19); }
 455   static int branchcc( CC       fcca)  { return  u_field(fcca,          21, 20); }
 456   static int cmpcc(    CC       fcca)  { return  u_field(fcca,          26, 25); }
 457   static int imm_asi(  int         x)  { return  u_field(x,             12,  5); }
 458   static int immed(    bool        i)  { return  u_field(i ? 1 : 0,     13, 13); }
 459   static int opf_low6( int         w)  { return  u_field(w,             10,  5); }
 460   static int opf_low5( int         w)  { return  u_field(w,              9,  5); }
 461   static int op5(      int         x)  { return  u_field(x,              8,  5); }
 462   static int trapcc(   CC         cc)  { return  u_field(cc,            12, 11); }
 463   static int sx(       int         i)  { return  u_field(i,             12, 12); } // shift x=1 means 64-bit
 464   static int opf(      int         x)  { return  u_field(x,             13,  5); }
 465 
 466   static bool is_cbcond( int x ) {
 467     return (VM_Version::has_cbcond() && (inv_cond(x) > rc_last) &&
 468             inv_op(x) == branch_op && inv_op2(x) == bpr_op2);
 469   }
 470   static bool is_cxb( int x ) {
 471     assert(is_cbcond(x), "wrong instruction");
 472     return (x & (1<<21)) != 0;
 473   }
 474   static int cond_cbcond( int         x)  { return  u_field((((x & 8)<<1) + 8 + (x & 7)), 29, 25); }
 475   static int inv_cond_cbcond(int      x)  {
 476     assert(is_cbcond(x), "wrong instruction");
 477     return inv_u_field(x, 27, 25) | (inv_u_field(x, 29, 29)<<3);
 478   }
 479 
 480   static int opf_cc(   CC          c, bool useFloat ) { return u_field((useFloat ? 0 : 4) + c, 13, 11); }
 481   static int mov_cc(   CC          c, bool useFloat ) { return u_field(useFloat ? 0 : 1,  18, 18) | u_field(c, 12, 11); }
 482 
 483   static int fd( FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 29, 25); };
 484   static int fs1(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 18, 14); };
 485   static int fs2(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa),  4,  0); };
 486   static int fs3(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 13,  9); };
 487 
 488   // some float instructions use this encoding on the op3 field
 489   static int alt_op3(int op, FloatRegisterImpl::Width w) {
 490     int r;
 491     switch(w) {
 492      case FloatRegisterImpl::S: r = op + 0;  break;
 493      case FloatRegisterImpl::D: r = op + 3;  break;
 494      case FloatRegisterImpl::Q: r = op + 2;  break;
 495      default: ShouldNotReachHere(); break;
 496     }
 497     return op3(r);
 498   }
 499 
 500 
 501   // compute inverse of simm
 502   static int inv_simm(int x, int nbits) {
 503     return (int)(x << (32 - nbits)) >> (32 - nbits);
 504   }
 505 
 506   static int inv_simm13( int x ) { return inv_simm(x, 13); }
 507 
 508   // signed immediate, in low bits, nbits long
 509   static int simm(int x, int nbits) {
 510     assert_signed_range(x, nbits);
 511     return x  &  (( 1 << nbits ) - 1);
 512   }
 513 
 514   // compute inverse of wdisp16
 515   static intptr_t inv_wdisp16(int x, intptr_t pos) {
 516     int lo = x & (( 1 << 14 ) - 1);
 517     int hi = (x >> 20) & 3;
 518     if (hi >= 2) hi |= ~1;
 519     return (((hi << 14) | lo) << 2) + pos;
 520   }
 521 
 522   // word offset, 14 bits at LSend, 2 bits at B21, B20
 523   static int wdisp16(intptr_t x, intptr_t off) {
 524     intptr_t xx = x - off;
 525     assert_signed_word_disp_range(xx, 16);
 526     int r =  (xx >> 2) & ((1 << 14) - 1)
 527            |  (  ( (xx>>(2+14)) & 3 )  <<  20 );
 528     assert( inv_wdisp16(r, off) == x,  "inverse is not inverse");
 529     return r;
 530   }
 531 
 532   // compute inverse of wdisp10
 533   static intptr_t inv_wdisp10(int x, intptr_t pos) {
 534     assert(is_cbcond(x), "wrong instruction");
 535     int lo = inv_u_field(x, 12, 5);
 536     int hi = (x >> 19) & 3;
 537     if (hi >= 2) hi |= ~1;
 538     return (((hi << 8) | lo) << 2) + pos;
 539   }
 540 
 541   // word offset for cbcond, 8 bits at [B12,B5], 2 bits at [B20,B19]
 542   static int wdisp10(intptr_t x, intptr_t off) {
 543     assert(VM_Version::has_cbcond(), "This CPU does not have CBCOND instruction");
 544     intptr_t xx = x - off;
 545     assert_signed_word_disp_range(xx, 10);
 546     int r =  ( ( (xx >>  2   ) & ((1 << 8) - 1) ) <<  5 )
 547            | ( ( (xx >> (2+8)) & 3              ) << 19 );
 548     // Have to fake cbcond instruction to pass assert in inv_wdisp10()
 549     assert(inv_wdisp10((r | op(branch_op) | cond_cbcond(rc_last+1) | op2(bpr_op2)), off) == x,  "inverse is not inverse");
 550     return r;
 551   }
 552 
 553   // word displacement in low-order nbits bits
 554 
 555   static intptr_t inv_wdisp( int x, intptr_t pos, int nbits ) {
 556     int pre_sign_extend = x & (( 1 << nbits ) - 1);
 557     int r =  pre_sign_extend >= ( 1 << (nbits-1) )
 558        ?   pre_sign_extend | ~(( 1 << nbits ) - 1)
 559        :   pre_sign_extend;
 560     return (r << 2) + pos;
 561   }
 562 
 563   static int wdisp( intptr_t x, intptr_t off, int nbits ) {
 564     intptr_t xx = x - off;
 565     assert_signed_word_disp_range(xx, nbits);
 566     int r =  (xx >> 2) & (( 1 << nbits ) - 1);
 567     assert( inv_wdisp( r, off, nbits )  ==  x, "inverse not inverse");
 568     return r;
 569   }
 570 
 571 
 572   // Extract the top 32 bits in a 64 bit word
 573   static int32_t hi32( int64_t x ) {
 574     int32_t r = int32_t( (uint64_t)x >> 32 );
 575     return r;
 576   }
 577 
 578   // given a sethi instruction, extract the constant, left-justified
 579   static int inv_hi22( int x ) {
 580     return x << 10;
 581   }
 582 
 583   // create an imm22 field, given a 32-bit left-justified constant
 584   static int hi22( int x ) {
 585     int r = int( juint(x) >> 10 );
 586     assert( (r & ~((1 << 22) - 1))  ==  0, "just checkin'");
 587     return r;
 588   }
 589 
 590   // create a low10 __value__ (not a field) for a given a 32-bit constant
 591   static int low10( int x ) {
 592     return x & ((1 << 10) - 1);
 593   }
 594 
 595   // AES crypto instructions supported only on certain processors
 596   static void aes_only() { assert( VM_Version::has_aes(), "This instruction only works on SPARC with AES instructions support"); }
 597 
 598   // instruction only in VIS1
 599   static void vis1_only() { assert( VM_Version::has_vis1(), "This instruction only works on SPARC with VIS1"); }
 600 
 601   // instruction only in VIS2
 602   static void vis2_only() { assert( VM_Version::has_vis2(), "This instruction only works on SPARC with VIS2"); }
 603 
 604   // instruction only in VIS3
 605   static void vis3_only() { assert( VM_Version::has_vis3(), "This instruction only works on SPARC with VIS3"); }
 606 
 607   // instruction only in v9
 608   static void v9_only() { } // do nothing
 609 
 610   // instruction deprecated in v9
 611   static void v9_dep()  { } // do nothing for now
 612 
 613   // v8 has no CC field
 614   static void v8_no_cc(CC cc)  { if (cc)  v9_only(); }
 615 
 616  protected:
 617   // Simple delay-slot scheme:
 618   // In order to check the programmer, the assembler keeps track of deley slots.
 619   // It forbids CTIs in delay slots (conservative, but should be OK).
 620   // Also, when putting an instruction into a delay slot, you must say
 621   // asm->delayed()->add(...), in order to check that you don't omit
 622   // delay-slot instructions.
 623   // To implement this, we use a simple FSA
 624 
 625 #ifdef ASSERT
 626   #define CHECK_DELAY
 627 #endif
 628 #ifdef CHECK_DELAY
 629   enum Delay_state { no_delay, at_delay_slot, filling_delay_slot } delay_state;
 630 #endif
 631 
 632  public:
 633   // Tells assembler next instruction must NOT be in delay slot.
 634   // Use at start of multinstruction macros.
 635   void assert_not_delayed() {
 636     // This is a separate overloading to avoid creation of string constants
 637     // in non-asserted code--with some compilers this pollutes the object code.
 638 #ifdef CHECK_DELAY
 639     assert_not_delayed("next instruction should not be a delay slot");
 640 #endif
 641   }
 642   void assert_not_delayed(const char* msg) {
 643 #ifdef CHECK_DELAY
 644     assert(delay_state == no_delay, msg);
 645 #endif
 646   }
 647 
 648  protected:
 649   // Insert a nop if the previous is cbcond
 650   void insert_nop_after_cbcond() {
 651     if (UseCBCond && cbcond_before()) {
 652       nop();
 653     }
 654   }
 655   // Delay slot helpers
 656   // cti is called when emitting control-transfer instruction,
 657   // BEFORE doing the emitting.
 658   // Only effective when assertion-checking is enabled.
 659   void cti() {
 660     // A cbcond instruction immediately followed by a CTI
 661     // instruction introduces pipeline stalls, we need to avoid that.
 662     no_cbcond_before();
 663 #ifdef CHECK_DELAY
 664     assert_not_delayed("cti should not be in delay slot");
 665 #endif
 666   }
 667 
 668   // called when emitting cti with a delay slot, AFTER emitting
 669   void has_delay_slot() {
 670 #ifdef CHECK_DELAY
 671     assert_not_delayed("just checking");
 672     delay_state = at_delay_slot;
 673 #endif
 674   }
 675 
 676   // cbcond instruction should not be generated one after an other
 677   bool cbcond_before() {
 678     if (offset() == 0) return false; // it is first instruction
 679     int x = *(int*)(intptr_t(pc()) - 4); // previous instruction
 680     return is_cbcond(x);
 681   }
 682 
 683   void no_cbcond_before() {
 684     assert(offset() == 0 || !cbcond_before(), "cbcond should not follow an other cbcond");
 685   }
 686 public:
 687 
 688   bool use_cbcond(Label& L) {
 689     if (!UseCBCond || cbcond_before()) return false;
 690     intptr_t x = intptr_t(target_distance(L)) - intptr_t(pc());
 691     assert( (x & 3) == 0, "not word aligned");
 692     return is_simm12(x);
 693   }
 694 
 695   // Tells assembler you know that next instruction is delayed
 696   Assembler* delayed() {
 697 #ifdef CHECK_DELAY
 698     assert ( delay_state == at_delay_slot, "delayed instruction is not in delay slot");
 699     delay_state = filling_delay_slot;
 700 #endif
 701     return this;
 702   }
 703 
 704   void flush() {
 705 #ifdef CHECK_DELAY
 706     assert ( delay_state == no_delay, "ending code with a delay slot");
 707 #endif
 708     AbstractAssembler::flush();
 709   }
 710 
 711   inline void emit_int32(int);  // shadows AbstractAssembler::emit_int32
 712   inline void emit_data(int x) { emit_int32(x); }
 713   inline void emit_data(int, RelocationHolder const&);
 714   inline void emit_data(int, relocInfo::relocType rtype);
 715   // helper for above fcns
 716   inline void check_delay();
 717 
 718 
 719  public:
 720   // instructions, refer to page numbers in the SPARC Architecture Manual, V9
 721 
 722   // pp 135 (addc was addx in v8)
 723 
 724   inline void add(Register s1, Register s2, Register d );
 725   inline void add(Register s1, int simm13a, Register d );
 726 
 727   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) ); }
 728   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) ); }
 729   void addc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(addc_op3             ) | rs1(s1) | rs2(s2) ); }
 730   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) ); }
 731   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) ); }
 732   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) ); }
 733 
 734 
 735   // 4-operand AES instructions
 736 
 737   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) ); }
 738   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) ); }
 739   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) ); }
 740   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) ); }
 741   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) ); }
 742   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) ); }
 743   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) ); }
 744   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) ); }
 745   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) ); }
 746 
 747 
 748   // 3-operand AES instructions
 749 
 750   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) ); }
 751   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) ); }
 752 
 753   // pp 136
 754 
 755   inline void bpr(RCondition c, bool a, Predict p, Register s1, address d, relocInfo::relocType rt = relocInfo::none);
 756   inline void bpr(RCondition c, bool a, Predict p, Register s1, Label& L);
 757 
 758   // compare and branch
 759   inline void cbcond(Condition c, CC cc, Register s1, Register s2, Label& L);
 760   inline void cbcond(Condition c, CC cc, Register s1, int simm5, Label& L);
 761 
 762  protected: // use MacroAssembler::br instead
 763 
 764   // pp 138
 765 
 766   inline void fb( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 767   inline void fb( Condition c, bool a, Label& L );
 768 
 769   // pp 141
 770 
 771   inline void fbp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 772   inline void fbp( Condition c, bool a, CC cc, Predict p, Label& L );
 773 
 774   // pp 144
 775 
 776   inline void br( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 777   inline void br( Condition c, bool a, Label& L );
 778 
 779   // pp 146
 780 
 781   inline void bp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 782   inline void bp( Condition c, bool a, CC cc, Predict p, Label& L );
 783 
 784   // pp 149
 785 
 786   inline void call( address d,  relocInfo::relocType rt = relocInfo::runtime_call_type );
 787   inline void call( Label& L,   relocInfo::relocType rt = relocInfo::runtime_call_type );
 788 
 789  public:
 790 
 791   // pp 150
 792 
 793   // These instructions compare the contents of s2 with the contents of
 794   // memory at address in s1. If the values are equal, the contents of memory
 795   // at address s1 is swapped with the data in d. If the values are not equal,
 796   // the the contents of memory at s1 is loaded into d, without the swap.
 797 
 798   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)); }
 799   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)); }
 800 
 801   // pp 152
 802 
 803   void udiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(udiv_op3             ) | rs1(s1) | rs2(s2)); }
 804   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) ); }
 805   void sdiv(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sdiv_op3             ) | rs1(s1) | rs2(s2)); }
 806   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) ); }
 807   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)); }
 808   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) ); }
 809   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)); }
 810   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) ); }
 811 
 812   // pp 155
 813 
 814   void done()  { v9_only();  cti();  emit_int32( op(arith_op) | fcn(0) | op3(done_op3) ); }
 815   void retry() { v9_only();  cti();  emit_int32( op(arith_op) | fcn(1) | op3(retry_op3) ); }
 816 
 817   // pp 156
 818 
 819   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)); }
 820   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)); }
 821 
 822   // pp 157
 823 
 824   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)); }
 825   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)); }
 826 
 827   // pp 159
 828 
 829   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)); }
 830   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)); }
 831 
 832   // pp 160
 833 
 834   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)); }
 835 
 836   // pp 161
 837 
 838   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)); }
 839   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)); }
 840 
 841   // pp 162
 842 
 843   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)); }
 844 
 845   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)); }
 846 
 847   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)); }
 848 
 849   // pp 163
 850 
 851   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)); }
 852   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)); }
 853   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)); }
 854 
 855   // FXORs/FXORd instructions
 856 
 857   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)); }
 858 
 859   // pp 164
 860 
 861   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)); }
 862 
 863   // pp 165
 864 
 865   inline void flush( Register s1, Register s2 );
 866   inline void flush( Register s1, int simm13a);
 867 
 868   // pp 167
 869 
 870   void flushw() { v9_only();  emit_int32( op(arith_op) | op3(flushw_op3) ); }
 871 
 872   // pp 168
 873 
 874   void illtrap( int const22a) { if (const22a != 0) v9_only();  emit_int32( op(branch_op) | u_field(const22a, 21, 0) ); }
 875   // v8 unimp == illtrap(0)
 876 
 877   // pp 169
 878 
 879   void impdep1( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep1_op3) | u_field(const19a, 18, 0)); }
 880   void impdep2( int id1, int const19a ) { v9_only();  emit_int32( op(arith_op) | fcn(id1) | op3(impdep2_op3) | u_field(const19a, 18, 0)); }
 881 
 882   // pp 170
 883 
 884   void jmpl( Register s1, Register s2, Register d );
 885   void jmpl( Register s1, int simm13a, Register d, RelocationHolder const& rspec = RelocationHolder() );
 886 
 887   // 171
 888 
 889   inline void ldf(FloatRegisterImpl::Width w, Register s1, Register s2, FloatRegister d);
 890   inline void ldf(FloatRegisterImpl::Width w, Register s1, int simm13a, FloatRegister d, RelocationHolder const& rspec = RelocationHolder());
 891 
 892 
 893   inline void ldfsr(  Register s1, Register s2 );
 894   inline void ldfsr(  Register s1, int simm13a);
 895   inline void ldxfsr( Register s1, Register s2 );
 896   inline void ldxfsr( Register s1, int simm13a);
 897 
 898   // 173
 899 
 900   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) ); }
 901   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) ); }
 902 
 903   // pp 175, lduw is ld on v8
 904 
 905   inline void ldsb(  Register s1, Register s2, Register d );
 906   inline void ldsb(  Register s1, int simm13a, Register d);
 907   inline void ldsh(  Register s1, Register s2, Register d );
 908   inline void ldsh(  Register s1, int simm13a, Register d);
 909   inline void ldsw(  Register s1, Register s2, Register d );
 910   inline void ldsw(  Register s1, int simm13a, Register d);
 911   inline void ldub(  Register s1, Register s2, Register d );
 912   inline void ldub(  Register s1, int simm13a, Register d);
 913   inline void lduh(  Register s1, Register s2, Register d );
 914   inline void lduh(  Register s1, int simm13a, Register d);
 915   inline void lduw(  Register s1, Register s2, Register d );
 916   inline void lduw(  Register s1, int simm13a, Register d);
 917   inline void ldx(   Register s1, Register s2, Register d );
 918   inline void ldx(   Register s1, int simm13a, Register d);
 919   inline void ldd(   Register s1, Register s2, Register d );
 920   inline void ldd(   Register s1, int simm13a, Register d);
 921 
 922   // pp 177
 923 
 924   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) ); }
 925   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) ); }
 926   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) ); }
 927   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) ); }
 928   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) ); }
 929   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) ); }
 930   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) ); }
 931   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) ); }
 932   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) ); }
 933   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) ); }
 934   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) ); }
 935   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) ); }
 936   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) ); }
 937   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) ); }
 938 
 939   // pp 181
 940 
 941   void and3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(and_op3              ) | rs1(s1) | rs2(s2) ); }
 942   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) ); }
 943   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) ); }
 944   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) ); }
 945   void andn(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(andn_op3             ) | rs1(s1) | rs2(s2) ); }
 946   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) ); }
 947   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) ); }
 948   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) ); }
 949   void or3(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(or_op3               ) | rs1(s1) | rs2(s2) ); }
 950   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) ); }
 951   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) ); }
 952   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) ); }
 953   void orn(     Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(orn_op3) | rs1(s1) | rs2(s2) ); }
 954   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) ); }
 955   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) ); }
 956   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) ); }
 957   void xor3(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xor_op3              ) | rs1(s1) | rs2(s2) ); }
 958   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) ); }
 959   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) ); }
 960   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) ); }
 961   void xnor(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(xnor_op3             ) | rs1(s1) | rs2(s2) ); }
 962   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) ); }
 963   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) ); }
 964   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) ); }
 965 
 966   // pp 183
 967 
 968   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)); }
 969 
 970   // pp 185
 971 
 972   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)); }
 973 
 974   // pp 189
 975 
 976   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)); }
 977 
 978   // pp 191
 979 
 980   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) ); }
 981   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) ); }
 982 
 983   // pp 195
 984 
 985   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) ); }
 986   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) ); }
 987 
 988   // pp 196
 989 
 990   void mulx(  Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(mulx_op3 ) | rs1(s1) | rs2(s2) ); }
 991   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) ); }
 992   void sdivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(sdivx_op3) | rs1(s1) | rs2(s2) ); }
 993   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) ); }
 994   void udivx( Register s1, Register s2, Register d ) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(udivx_op3) | rs1(s1) | rs2(s2) ); }
 995   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) ); }
 996 
 997   // pp 197
 998 
 999   void umul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(umul_op3             ) | rs1(s1) | rs2(s2) ); }
1000   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) ); }
1001   void smul(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(smul_op3             ) | rs1(s1) | rs2(s2) ); }
1002   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) ); }
1003   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) ); }
1004   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) ); }
1005   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) ); }
1006   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) ); }
1007 
1008   // pp 201
1009 
1010   void nop() { emit_int32( op(branch_op) | op2(sethi_op2) ); }
1011 
1012 
1013   // pp 202
1014 
1015   void popc( Register s,  Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | rs2(s)); }
1016   void popc( int simm13a, Register d) { v9_only();  emit_int32( op(arith_op) | rd(d) | op3(popc_op3) | immed(true) | simm(simm13a, 13)); }
1017 
1018   // pp 203
1019 
1020   void prefetch(   Register s1, Register s2, PrefetchFcn f) { v9_only();  emit_int32( op(ldst_op) | fcn(f) | op3(prefetch_op3) | rs1(s1) | rs2(s2) ); }
1021   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)); }
1022 
1023   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) ); }
1024   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) ); }
1025 
1026   // pp 208
1027 
1028   // not implementing read privileged register
1029 
1030   inline void rdy(    Register d) { v9_dep();  emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(0, 18, 14)); }
1031   inline void rdccr(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(2, 18, 14)); }
1032   inline void rdasi(  Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(3, 18, 14)); }
1033   inline void rdtick( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(4, 18, 14)); } // Spoon!
1034   inline void rdpc(   Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(5, 18, 14)); }
1035   inline void rdfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rd(d) | op3(rdreg_op3) | u_field(6, 18, 14)); }
1036 
1037   // pp 213
1038 
1039   inline void rett( Register s1, Register s2);
1040   inline void rett( Register s1, int simm13a, relocInfo::relocType rt = relocInfo::none);
1041 
1042   // pp 214
1043 
1044   void save(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | rs2(s2) ); }
1045   void save(    Register s1, int simm13a, Register d ) {
1046     // make sure frame is at least large enough for the register save area
1047     assert(-simm13a >= 16 * wordSize, "frame too small");
1048     emit_int32( op(arith_op) | rd(d) | op3(save_op3) | rs1(s1) | immed(true) | simm(simm13a, 13) );
1049   }
1050 
1051   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) ); }
1052   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) ); }
1053 
1054   // pp 216
1055 
1056   void saved()    { v9_only();  emit_int32( op(arith_op) | fcn(0) | op3(saved_op3)); }
1057   void restored() { v9_only();  emit_int32( op(arith_op) | fcn(1) | op3(saved_op3)); }
1058 
1059   // pp 217
1060 
1061   inline void sethi( int imm22a, Register d, RelocationHolder const& rspec = RelocationHolder() );
1062   // pp 218
1063 
1064   void sll(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sll_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1065   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) ); }
1066   void srl(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(srl_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1067   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) ); }
1068   void sra(  Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sra_op3) | rs1(s1) | sx(0) | rs2(s2) ); }
1069   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) ); }
1070 
1071   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) ); }
1072   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) ); }
1073   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) ); }
1074   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) ); }
1075   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) ); }
1076   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) ); }
1077 
1078   // pp 220
1079 
1080   void sir( int simm13a ) { emit_int32( op(arith_op) | fcn(15) | op3(sir_op3) | immed(true) | simm(simm13a, 13)); }
1081 
1082   // pp 221
1083 
1084   void stbar() { emit_int32( op(arith_op) | op3(membar_op3) | u_field(15, 18, 14)); }
1085 
1086   // pp 222
1087 
1088   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2);
1089   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a);
1090 
1091   inline void stfsr(  Register s1, Register s2 );
1092   inline void stfsr(  Register s1, int simm13a);
1093   inline void stxfsr( Register s1, Register s2 );
1094   inline void stxfsr( Register s1, int simm13a);
1095 
1096   //  pp 224
1097 
1098   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) ); }
1099   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) ); }
1100 
1101   // p 226
1102 
1103   inline void stb(  Register d, Register s1, Register s2 );
1104   inline void stb(  Register d, Register s1, int simm13a);
1105   inline void sth(  Register d, Register s1, Register s2 );
1106   inline void sth(  Register d, Register s1, int simm13a);
1107   inline void stw(  Register d, Register s1, Register s2 );
1108   inline void stw(  Register d, Register s1, int simm13a);
1109   inline void stx(  Register d, Register s1, Register s2 );
1110   inline void stx(  Register d, Register s1, int simm13a);
1111   inline void std(  Register d, Register s1, Register s2 );
1112   inline void std(  Register d, Register s1, int simm13a);
1113 
1114   // pp 177
1115 
1116   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) ); }
1117   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) ); }
1118   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) ); }
1119   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) ); }
1120   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) ); }
1121   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) ); }
1122   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) ); }
1123   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) ); }
1124   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) ); }
1125   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) ); }
1126 
1127   // pp 230
1128 
1129   void sub(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(sub_op3              ) | rs1(s1) | rs2(s2) ); }
1130   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) ); }
1131 
1132   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) ); }
1133   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) ); }
1134   void subc(   Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(subc_op3             ) | rs1(s1) | rs2(s2) ); }
1135   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) ); }
1136   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) ); }
1137   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) ); }
1138 
1139   // pp 231
1140 
1141   inline void swap( Register s1, Register s2, Register d );
1142   inline void swap( Register s1, int simm13a, Register d);
1143 
1144   // pp 232
1145 
1146   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) ); }
1147   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) ); }
1148 
1149   // pp 234, note op in book is wrong, see pp 268
1150 
1151   void taddcc(    Register s1, Register s2, Register d ) {            emit_int32( op(arith_op) | rd(d) | op3(taddcc_op3  ) | rs1(s1) | rs2(s2) ); }
1152   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) ); }
1153 
1154   // pp 235
1155 
1156   void tsubcc(    Register s1, Register s2, Register d ) { emit_int32( op(arith_op) | rd(d) | op3(tsubcc_op3  ) | rs1(s1) | rs2(s2) ); }
1157   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) ); }
1158 
1159   // pp 237
1160 
1161   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)); }
1162   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)); }
1163   // simple uncond. trap
1164   void trap( int trapa ) { trap( always, icc, G0, trapa ); }
1165 
1166   // pp 239 omit write priv register for now
1167 
1168   inline void wry(    Register d) { v9_dep();  emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(0, 29, 25)); }
1169   inline void wrccr(Register s) { v9_only(); emit_int32( op(arith_op) | rs1(s) | op3(wrreg_op3) | u_field(2, 29, 25)); }
1170   inline void wrccr(Register s, int simm13a) { v9_only(); emit_int32( op(arith_op) |
1171                                                                            rs1(s) |
1172                                                                            op3(wrreg_op3) |
1173                                                                            u_field(2, 29, 25) |
1174                                                                            immed(true) |
1175                                                                            simm(simm13a, 13)); }
1176   inline void wrasi(Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(3, 29, 25)); }
1177   // wrasi(d, imm) stores (d xor imm) to asi
1178   inline void wrasi(Register d, int simm13a) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) |
1179                                                u_field(3, 29, 25) | immed(true) | simm(simm13a, 13)); }
1180   inline void wrfprs( Register d) { v9_only(); emit_int32( op(arith_op) | rs1(d) | op3(wrreg_op3) | u_field(6, 29, 25)); }
1181 
1182 
1183   //  VIS1 instructions
1184 
1185   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)); }
1186 
1187   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)); }
1188 
1189   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)); }
1190 
1191   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)); }
1192 
1193   //  VIS2 instructions
1194 
1195   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)); }
1196 
1197   // VIS3 instructions
1198 
1199   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)); }
1200   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)); }
1201   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)); }
1202 
1203   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)); }
1204   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)); }
1205 
1206   // Creation
1207   Assembler(CodeBuffer* code) : AbstractAssembler(code) {
1208 #ifdef CHECK_DELAY
1209     delay_state = no_delay;
1210 #endif
1211   }
1212 };
1213 
1214 #endif // CPU_SPARC_VM_ASSEMBLER_SPARC_HPP