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