1 /*
   2  * Copyright (c) 1997, 2015, Oracle and/or its affiliates. All rights reserved.
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * This code is free software; you can redistribute it and/or modify it
   6  * under the terms of the GNU General Public License version 2 only, as
   7  * published by the Free Software Foundation.
   8  *
   9  * This code is distributed in the hope that it will be useful, but WITHOUT
  10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12  * version 2 for more details (a copy is included in the LICENSE file that
  13  * accompanied this code).
  14  *
  15  * You should have received a copy of the GNU General Public License version
  16  * 2 along with this work; if not, write to the Free Software Foundation,
  17  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
  18  *
  19  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
  20  * or visit www.oracle.com if you need additional information or have any
  21  * questions.
  22  *
  23  */
  24 
  25 #ifndef CPU_SPARC_VM_ASSEMBLER_SPARC_HPP
  26 #define CPU_SPARC_VM_ASSEMBLER_SPARC_HPP
  27 
  28 #include "asm/register.hpp"
  29 
  30 // The SPARC Assembler: Pure assembler doing NO optimizations on the instruction
  31 // level; i.e., what you write
  32 // is what you get. The Assembler is generating code into a CodeBuffer.
  33 
  34 class Assembler : public AbstractAssembler  {
  35   friend class AbstractAssembler;
  36   friend class AddressLiteral;
  37 
  38   // code patchers need various routines like inv_wdisp()
  39   friend class NativeInstruction;
  40   friend class NativeGeneralJump;
  41   friend class Relocation;
  42   friend class Label;
  43 
  44  public:
  45   // op carries format info; see page 62 & 267
  46 
  47   enum ops {
  48     call_op   = 1, // fmt 1
  49     branch_op = 0, // also sethi (fmt2)
  50     arith_op  = 2, // fmt 3, arith & misc
  51     ldst_op   = 3  // fmt 3, load/store
  52   };
  53 
  54   enum op2s {
  55     bpr_op2   = 3,
  56     fb_op2    = 6,
  57     fbp_op2   = 5,
  58     br_op2    = 2,
  59     bp_op2    = 1,
  60     sethi_op2 = 4
  61   };
  62 
  63   enum op3s {
  64     // selected op3s
  65     add_op3      = 0x00,
  66     and_op3      = 0x01,
  67     or_op3       = 0x02,
  68     xor_op3      = 0x03,
  69     sub_op3      = 0x04,
  70     andn_op3     = 0x05,
  71     orn_op3      = 0x06,
  72     xnor_op3     = 0x07,
  73     addc_op3     = 0x08,
  74     mulx_op3     = 0x09,
  75     umul_op3     = 0x0a,
  76     smul_op3     = 0x0b,
  77     subc_op3     = 0x0c,
  78     udivx_op3    = 0x0d,
  79     udiv_op3     = 0x0e,
  80     sdiv_op3     = 0x0f,
  81 
  82     addcc_op3    = 0x10,
  83     andcc_op3    = 0x11,
  84     orcc_op3     = 0x12,
  85     xorcc_op3    = 0x13,
  86     subcc_op3    = 0x14,
  87     andncc_op3   = 0x15,
  88     orncc_op3    = 0x16,
  89     xnorcc_op3   = 0x17,
  90     addccc_op3   = 0x18,
  91     aes4_op3     = 0x19,
  92     umulcc_op3   = 0x1a,
  93     smulcc_op3   = 0x1b,
  94     subccc_op3   = 0x1c,
  95     udivcc_op3   = 0x1e,
  96     sdivcc_op3   = 0x1f,
  97 
  98     taddcc_op3   = 0x20,
  99     tsubcc_op3   = 0x21,
 100     taddcctv_op3 = 0x22,
 101     tsubcctv_op3 = 0x23,
 102     mulscc_op3   = 0x24,
 103     sll_op3      = 0x25,
 104     sllx_op3     = 0x25,
 105     srl_op3      = 0x26,
 106     srlx_op3     = 0x26,
 107     sra_op3      = 0x27,
 108     srax_op3     = 0x27,
 109     rdreg_op3    = 0x28,
 110     membar_op3   = 0x28,
 111 
 112     flushw_op3   = 0x2b,
 113     movcc_op3    = 0x2c,
 114     sdivx_op3    = 0x2d,
 115     popc_op3     = 0x2e,
 116     movr_op3     = 0x2f,
 117 
 118     sir_op3      = 0x30,
 119     wrreg_op3    = 0x30,
 120     saved_op3    = 0x31,
 121 
 122     fpop1_op3    = 0x34,
 123     fpop2_op3    = 0x35,
 124     impdep1_op3  = 0x36,
 125     aes3_op3     = 0x36,
 126     sha_op3      = 0x36,
 127     bmask_op3    = 0x36,
 128     bshuffle_op3   = 0x36,
 129     alignaddr_op3  = 0x36,
 130     faligndata_op3 = 0x36,
 131     flog3_op3    = 0x36,
 132     edge_op3     = 0x36,
 133     fzero_op3    = 0x36,
 134     fsrc_op3     = 0x36,
 135     fnot_op3     = 0x36,
 136     lzcnt_op3    = 0x36,
 137     xmulx_op3    = 0x36,
 138     crc32c_op3   = 0x36,
 139     impdep2_op3  = 0x37,
 140     stpartialf_op3 = 0x37,
 141     jmpl_op3     = 0x38,
 142     rett_op3     = 0x39,
 143     trap_op3     = 0x3a,
 144     flush_op3    = 0x3b,
 145     save_op3     = 0x3c,
 146     restore_op3  = 0x3d,
 147     done_op3     = 0x3e,
 148     retry_op3    = 0x3e,
 149 
 150     lduw_op3     = 0x00,
 151     ldub_op3     = 0x01,
 152     lduh_op3     = 0x02,
 153     ldd_op3      = 0x03,
 154     stw_op3      = 0x04,
 155     stb_op3      = 0x05,
 156     sth_op3      = 0x06,
 157     std_op3      = 0x07,
 158     ldsw_op3     = 0x08,
 159     ldsb_op3     = 0x09,
 160     ldsh_op3     = 0x0a,
 161     ldx_op3      = 0x0b,
 162 
 163     stx_op3      = 0x0e,
 164     swap_op3     = 0x0f,
 165 
 166     stwa_op3     = 0x14,
 167     stxa_op3     = 0x1e,
 168 
 169     ldf_op3      = 0x20,
 170     ldfsr_op3    = 0x21,
 171     ldqf_op3     = 0x22,
 172     lddf_op3     = 0x23,
 173     stf_op3      = 0x24,
 174     stfsr_op3    = 0x25,
 175     stqf_op3     = 0x26,
 176     stdf_op3     = 0x27,
 177 
 178     prefetch_op3 = 0x2d,
 179 
 180     casa_op3     = 0x3c,
 181     casxa_op3    = 0x3e,
 182 
 183     mftoi_op3    = 0x36,
 184 
 185     alt_bit_op3  = 0x10,
 186      cc_bit_op3  = 0x10
 187   };
 188 
 189   enum opfs {
 190     // selected opfs
 191     edge8n_opf         = 0x01,
 192 
 193     fmovs_opf          = 0x01,
 194     fmovd_opf          = 0x02,
 195 
 196     fnegs_opf          = 0x05,
 197     fnegd_opf          = 0x06,
 198 
 199     lzcnt_opf          = 0x17,
 200     alignaddr_opf      = 0x18,
 201     bmask_opf          = 0x19,
 202 
 203     fadds_opf          = 0x41,
 204     faddd_opf          = 0x42,
 205     fsubs_opf          = 0x45,
 206     fsubd_opf          = 0x46,
 207 
 208     faligndata_opf     = 0x48,
 209 
 210     fmuls_opf          = 0x49,
 211     fmuld_opf          = 0x4a,
 212     bshuffle_opf       = 0x4c,
 213     fdivs_opf          = 0x4d,
 214     fdivd_opf          = 0x4e,
 215 
 216     fcmps_opf          = 0x51,
 217     fcmpd_opf          = 0x52,
 218 
 219     fstox_opf          = 0x81,
 220     fdtox_opf          = 0x82,
 221     fxtos_opf          = 0x84,
 222     fxtod_opf          = 0x88,
 223     fitos_opf          = 0xc4,
 224     fdtos_opf          = 0xc6,
 225     fitod_opf          = 0xc8,
 226     fstod_opf          = 0xc9,
 227     fstoi_opf          = 0xd1,
 228     fdtoi_opf          = 0xd2,
 229 
 230     mdtox_opf          = 0x110,
 231     mstouw_opf         = 0x111,
 232     mstosw_opf         = 0x113,
 233     xmulx_opf          = 0x115,
 234     xmulxhi_opf        = 0x116,
 235     mxtod_opf          = 0x118,
 236     mwtos_opf          = 0x119,
 237 
 238     aes_kexpand0_opf   = 0x130,
 239     aes_kexpand2_opf   = 0x131,
 240 
 241     sha1_opf           = 0x141,
 242     sha256_opf         = 0x142,
 243     sha512_opf         = 0x143,
 244 
 245     crc32c_opf         = 0x147
 246   };
 247 
 248   enum op5s {
 249     aes_eround01_op5     = 0x00,
 250     aes_eround23_op5     = 0x01,
 251     aes_dround01_op5     = 0x02,
 252     aes_dround23_op5     = 0x03,
 253     aes_eround01_l_op5   = 0x04,
 254     aes_eround23_l_op5   = 0x05,
 255     aes_dround01_l_op5   = 0x06,
 256     aes_dround23_l_op5   = 0x07,
 257     aes_kexpand1_op5     = 0x08
 258   };
 259 
 260   enum RCondition {  rc_z = 1,  rc_lez = 2,  rc_lz = 3, rc_nz = 5, rc_gz = 6, rc_gez = 7, rc_last = rc_gez  };
 261 
 262   enum Condition {
 263      // for FBfcc & FBPfcc instruction
 264     f_never                     = 0,
 265     f_notEqual                  = 1,
 266     f_notZero                   = 1,
 267     f_lessOrGreater             = 2,
 268     f_unorderedOrLess           = 3,
 269     f_less                      = 4,
 270     f_unorderedOrGreater        = 5,
 271     f_greater                   = 6,
 272     f_unordered                 = 7,
 273     f_always                    = 8,
 274     f_equal                     = 9,
 275     f_zero                      = 9,
 276     f_unorderedOrEqual          = 10,
 277     f_greaterOrEqual            = 11,
 278     f_unorderedOrGreaterOrEqual = 12,
 279     f_lessOrEqual               = 13,
 280     f_unorderedOrLessOrEqual    = 14,
 281     f_ordered                   = 15,
 282 
 283     // V8 coproc, pp 123 v8 manual
 284 
 285     cp_always  = 8,
 286     cp_never   = 0,
 287     cp_3       = 7,
 288     cp_2       = 6,
 289     cp_2or3    = 5,
 290     cp_1       = 4,
 291     cp_1or3    = 3,
 292     cp_1or2    = 2,
 293     cp_1or2or3 = 1,
 294     cp_0       = 9,
 295     cp_0or3    = 10,
 296     cp_0or2    = 11,
 297     cp_0or2or3 = 12,
 298     cp_0or1    = 13,
 299     cp_0or1or3 = 14,
 300     cp_0or1or2 = 15,
 301 
 302 
 303     // for integers
 304 
 305     never                 =  0,
 306     equal                 =  1,
 307     zero                  =  1,
 308     lessEqual             =  2,
 309     less                  =  3,
 310     lessEqualUnsigned     =  4,
 311     lessUnsigned          =  5,
 312     carrySet              =  5,
 313     negative              =  6,
 314     overflowSet           =  7,
 315     always                =  8,
 316     notEqual              =  9,
 317     notZero               =  9,
 318     greater               =  10,
 319     greaterEqual          =  11,
 320     greaterUnsigned       =  12,
 321     greaterEqualUnsigned  =  13,
 322     carryClear            =  13,
 323     positive              =  14,
 324     overflowClear         =  15
 325   };
 326 
 327   enum CC {
 328     icc  = 0,  xcc  = 2,
 329     // ptr_cc is the correct condition code for a pointer or intptr_t:
 330     ptr_cc = NOT_LP64(icc) LP64_ONLY(xcc),
 331     fcc0 = 0,  fcc1 = 1, fcc2 = 2, fcc3 = 3
 332   };
 333 
 334   enum PrefetchFcn {
 335     severalReads = 0,  oneRead = 1,  severalWritesAndPossiblyReads = 2, oneWrite = 3, page = 4
 336   };
 337 
 338  public:
 339   // Helper functions for groups of instructions
 340 
 341   enum Predict { pt = 1, pn = 0 }; // pt = predict taken
 342 
 343   enum Membar_mask_bits { // page 184, v9
 344     StoreStore = 1 << 3,
 345     LoadStore  = 1 << 2,
 346     StoreLoad  = 1 << 1,
 347     LoadLoad   = 1 << 0,
 348 
 349     Sync       = 1 << 6,
 350     MemIssue   = 1 << 5,
 351     Lookaside  = 1 << 4
 352   };
 353 
 354   static bool is_in_wdisp_range(address a, address b, int nbits) {
 355     intptr_t d = intptr_t(b) - intptr_t(a);
 356     return is_simm(d, nbits + 2);
 357   }
 358 
 359   address target_distance(Label& L) {
 360     // Assembler::target(L) should be called only when
 361     // a branch instruction is emitted since non-bound
 362     // labels record current pc() as a branch address.
 363     if (L.is_bound()) return target(L);
 364     // Return current address for non-bound labels.
 365     return pc();
 366   }
 367 
 368   // test if label is in simm16 range in words (wdisp16).
 369   bool is_in_wdisp16_range(Label& L) {
 370     return is_in_wdisp_range(target_distance(L), pc(), 16);
 371   }
 372   // test if the distance between two addresses fits in simm30 range in words
 373   static bool is_in_wdisp30_range(address a, address b) {
 374     return is_in_wdisp_range(a, b, 30);
 375   }
 376 
 377   enum ASIs { // page 72, v9
 378     ASI_PRIMARY            = 0x80,
 379     ASI_PRIMARY_NOFAULT    = 0x82,
 380     ASI_PRIMARY_LITTLE     = 0x88,
 381     // 8x8-bit partial store
 382     ASI_PST8_PRIMARY       = 0xC0,
 383     // Block initializing store
 384     ASI_ST_BLKINIT_PRIMARY = 0xE2,
 385     // Most-Recently-Used (MRU) BIS variant
 386     ASI_ST_BLKINIT_MRU_PRIMARY = 0xF2
 387     // add more from book as needed
 388   };
 389 
 390  protected:
 391   // helpers
 392 
 393   // x is supposed to fit in a field "nbits" wide
 394   // and be sign-extended. Check the range.
 395 
 396   static void assert_signed_range(intptr_t x, int nbits) {
 397     assert(nbits == 32 || (-(1 << nbits-1) <= x  &&  x < ( 1 << nbits-1)),
 398            "value out of range: x=" INTPTR_FORMAT ", nbits=%d", x, nbits);
 399   }
 400 
 401   static void assert_signed_word_disp_range(intptr_t x, int nbits) {
 402     assert( (x & 3) == 0, "not word aligned");
 403     assert_signed_range(x, nbits + 2);
 404   }
 405 
 406   static void assert_unsigned_const(int x, int nbits) {
 407     assert( juint(x)  <  juint(1 << nbits), "unsigned constant out of range");
 408   }
 409 
 410   // fields: note bits numbered from LSB = 0,
 411   //  fields known by inclusive bit range
 412 
 413   static int fmask(juint hi_bit, juint lo_bit) {
 414     assert( hi_bit >= lo_bit  &&  0 <= lo_bit  &&  hi_bit < 32, "bad bits");
 415     return (1 << ( hi_bit-lo_bit + 1 )) - 1;
 416   }
 417 
 418   // inverse of u_field
 419 
 420   static int inv_u_field(int x, int hi_bit, int lo_bit) {
 421     juint r = juint(x) >> lo_bit;
 422     r &= fmask( hi_bit, lo_bit);
 423     return int(r);
 424   }
 425 
 426 
 427   // signed version: extract from field and sign-extend
 428 
 429   static int inv_s_field(int x, int hi_bit, int lo_bit) {
 430     int sign_shift = 31 - hi_bit;
 431     return inv_u_field( ((x << sign_shift) >> sign_shift), hi_bit, lo_bit);
 432   }
 433 
 434   // given a field that ranges from hi_bit to lo_bit (inclusive,
 435   // LSB = 0), and an unsigned value for the field,
 436   // shift it into the field
 437 
 438 #ifdef ASSERT
 439   static int u_field(int x, int hi_bit, int lo_bit) {
 440     assert( ( x & ~fmask(hi_bit, lo_bit))  == 0,
 441             "value out of range");
 442     int r = x << lo_bit;
 443     assert( inv_u_field(r, hi_bit, lo_bit) == x, "just checking");
 444     return r;
 445   }
 446 #else
 447   // make sure this is inlined as it will reduce code size significantly
 448   #define u_field(x, hi_bit, lo_bit)   ((x) << (lo_bit))
 449 #endif
 450 
 451   static int inv_op(  int x ) { return inv_u_field(x, 31, 30); }
 452   static int inv_op2( int x ) { return inv_u_field(x, 24, 22); }
 453   static int inv_op3( int x ) { return inv_u_field(x, 24, 19); }
 454   static int inv_cond( int x ){ return inv_u_field(x, 28, 25); }
 455 
 456   static bool inv_immed( int x ) { return (x & Assembler::immed(true)) != 0; }
 457 
 458   static Register inv_rd(  int x ) { return as_Register(inv_u_field(x, 29, 25)); }
 459   static Register inv_rs1( int x ) { return as_Register(inv_u_field(x, 18, 14)); }
 460   static Register inv_rs2( int x ) { return as_Register(inv_u_field(x,  4,  0)); }
 461 
 462   static int op(       int         x)  { return  u_field(x,             31, 30); }
 463   static int rd(       Register    r)  { return  u_field(r->encoding(), 29, 25); }
 464   static int fcn(      int         x)  { return  u_field(x,             29, 25); }
 465   static int op3(      int         x)  { return  u_field(x,             24, 19); }
 466   static int rs1(      Register    r)  { return  u_field(r->encoding(), 18, 14); }
 467   static int rs2(      Register    r)  { return  u_field(r->encoding(),  4,  0); }
 468   static int annul(    bool        a)  { return  u_field(a ? 1 : 0,     29, 29); }
 469   static int cond(     int         x)  { return  u_field(x,             28, 25); }
 470   static int cond_mov( int         x)  { return  u_field(x,             17, 14); }
 471   static int rcond(    RCondition  x)  { return  u_field(x,             12, 10); }
 472   static int op2(      int         x)  { return  u_field(x,             24, 22); }
 473   static int predict(  bool        p)  { return  u_field(p ? 1 : 0,     19, 19); }
 474   static int branchcc( CC       fcca)  { return  u_field(fcca,          21, 20); }
 475   static int cmpcc(    CC       fcca)  { return  u_field(fcca,          26, 25); }
 476   static int imm_asi(  int         x)  { return  u_field(x,             12,  5); }
 477   static int immed(    bool        i)  { return  u_field(i ? 1 : 0,     13, 13); }
 478   static int opf_low6( int         w)  { return  u_field(w,             10,  5); }
 479   static int opf_low5( int         w)  { return  u_field(w,              9,  5); }
 480   static int op5(      int         x)  { return  u_field(x,              8,  5); }
 481   static int trapcc(   CC         cc)  { return  u_field(cc,            12, 11); }
 482   static int sx(       int         i)  { return  u_field(i,             12, 12); } // shift x=1 means 64-bit
 483   static int opf(      int         x)  { return  u_field(x,             13,  5); }
 484 
 485   static bool is_cbcond( int x ) {
 486     return (VM_Version::has_cbcond() && (inv_cond(x) > rc_last) &&
 487             inv_op(x) == branch_op && inv_op2(x) == bpr_op2);
 488   }
 489   static bool is_cxb( int x ) {
 490     assert(is_cbcond(x), "wrong instruction");
 491     return (x & (1<<21)) != 0;
 492   }
 493   static int cond_cbcond( int         x)  { return  u_field((((x & 8)<<1) + 8 + (x & 7)), 29, 25); }
 494   static int inv_cond_cbcond(int      x)  {
 495     assert(is_cbcond(x), "wrong instruction");
 496     return inv_u_field(x, 27, 25) | (inv_u_field(x, 29, 29)<<3);
 497   }
 498 
 499   static int opf_cc(   CC          c, bool useFloat ) { return u_field((useFloat ? 0 : 4) + c, 13, 11); }
 500   static int mov_cc(   CC          c, bool useFloat ) { return u_field(useFloat ? 0 : 1,  18, 18) | u_field(c, 12, 11); }
 501 
 502   static int fd( FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 29, 25); };
 503   static int fs1(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 18, 14); };
 504   static int fs2(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa),  4,  0); };
 505   static int fs3(FloatRegister r,  FloatRegisterImpl::Width fwa) { return u_field(r->encoding(fwa), 13,  9); };
 506 
 507   // some float instructions use this encoding on the op3 field
 508   static int alt_op3(int op, FloatRegisterImpl::Width w) {
 509     int r;
 510     switch(w) {
 511      case FloatRegisterImpl::S: r = op + 0;  break;
 512      case FloatRegisterImpl::D: r = op + 3;  break;
 513      case FloatRegisterImpl::Q: r = op + 2;  break;
 514      default: ShouldNotReachHere(); break;
 515     }
 516     return op3(r);
 517   }
 518 
 519 
 520   // compute inverse of simm
 521   static int inv_simm(int x, int nbits) {
 522     return (int)(x << (32 - nbits)) >> (32 - nbits);
 523   }
 524 
 525   static int inv_simm13( int x ) { return inv_simm(x, 13); }
 526 
 527   // signed immediate, in low bits, nbits long
 528   static int simm(int x, int nbits) {
 529     assert_signed_range(x, nbits);
 530     return x  &  (( 1 << nbits ) - 1);
 531   }
 532 
 533   // compute inverse of wdisp16
 534   static intptr_t inv_wdisp16(int x, intptr_t pos) {
 535     int lo = x & (( 1 << 14 ) - 1);
 536     int hi = (x >> 20) & 3;
 537     if (hi >= 2) hi |= ~1;
 538     return (((hi << 14) | lo) << 2) + pos;
 539   }
 540 
 541   // word offset, 14 bits at LSend, 2 bits at B21, B20
 542   static int wdisp16(intptr_t x, intptr_t off) {
 543     intptr_t xx = x - off;
 544     assert_signed_word_disp_range(xx, 16);
 545     int r =  (xx >> 2) & ((1 << 14) - 1)
 546            |  (  ( (xx>>(2+14)) & 3 )  <<  20 );
 547     assert( inv_wdisp16(r, off) == x,  "inverse is not inverse");
 548     return r;
 549   }
 550 
 551   // compute inverse of wdisp10
 552   static intptr_t inv_wdisp10(int x, intptr_t pos) {
 553     assert(is_cbcond(x), "wrong instruction");
 554     int lo = inv_u_field(x, 12, 5);
 555     int hi = (x >> 19) & 3;
 556     if (hi >= 2) hi |= ~1;
 557     return (((hi << 8) | lo) << 2) + pos;
 558   }
 559 
 560   // word offset for cbcond, 8 bits at [B12,B5], 2 bits at [B20,B19]
 561   static int wdisp10(intptr_t x, intptr_t off) {
 562     assert(VM_Version::has_cbcond(), "This CPU does not have CBCOND instruction");
 563     intptr_t xx = x - off;
 564     assert_signed_word_disp_range(xx, 10);
 565     int r =  ( ( (xx >>  2   ) & ((1 << 8) - 1) ) <<  5 )
 566            | ( ( (xx >> (2+8)) & 3              ) << 19 );
 567     // Have to fake cbcond instruction to pass assert in inv_wdisp10()
 568     assert(inv_wdisp10((r | op(branch_op) | cond_cbcond(rc_last+1) | op2(bpr_op2)), off) == x,  "inverse is not inverse");
 569     return r;
 570   }
 571 
 572   // word displacement in low-order nbits bits
 573 
 574   static intptr_t inv_wdisp( int x, intptr_t pos, int nbits ) {
 575     int pre_sign_extend = x & (( 1 << nbits ) - 1);
 576     int r =  pre_sign_extend >= ( 1 << (nbits-1) )
 577        ?   pre_sign_extend | ~(( 1 << nbits ) - 1)
 578        :   pre_sign_extend;
 579     return (r << 2) + pos;
 580   }
 581 
 582   static int wdisp( intptr_t x, intptr_t off, int nbits ) {
 583     intptr_t xx = x - off;
 584     assert_signed_word_disp_range(xx, nbits);
 585     int r =  (xx >> 2) & (( 1 << nbits ) - 1);
 586     assert( inv_wdisp( r, off, nbits )  ==  x, "inverse not inverse");
 587     return r;
 588   }
 589 
 590 
 591   // Extract the top 32 bits in a 64 bit word
 592   static int32_t hi32( int64_t x ) {
 593     int32_t r = int32_t( (uint64_t)x >> 32 );
 594     return r;
 595   }
 596 
 597   // given a sethi instruction, extract the constant, left-justified
 598   static int inv_hi22( int x ) {
 599     return x << 10;
 600   }
 601 
 602   // create an imm22 field, given a 32-bit left-justified constant
 603   static int hi22( int x ) {
 604     int r = int( juint(x) >> 10 );
 605     assert( (r & ~((1 << 22) - 1))  ==  0, "just checkin'");
 606     return r;
 607   }
 608 
 609   // create a low10 __value__ (not a field) for a given a 32-bit constant
 610   static int low10( int x ) {
 611     return x & ((1 << 10) - 1);
 612   }
 613 
 614   // create a low12 __value__ (not a field) for a given a 32-bit constant
 615   static int low12( int x ) {
 616     return x & ((1 << 12) - 1);
 617   }
 618 
 619   // AES crypto instructions supported only on certain processors
 620   static void aes_only() { assert( VM_Version::has_aes(), "This instruction only works on SPARC with AES instructions support"); }
 621 
 622   // SHA crypto instructions supported only on certain processors
 623   static void sha1_only()   { assert( VM_Version::has_sha1(),   "This instruction only works on SPARC with SHA1"); }
 624   static void sha256_only() { assert( VM_Version::has_sha256(), "This instruction only works on SPARC with SHA256"); }
 625   static void sha512_only() { assert( VM_Version::has_sha512(), "This instruction only works on SPARC with SHA512"); }
 626 
 627   // CRC32C instruction supported only on certain processors
 628   static void crc32c_only() { assert( VM_Version::has_crc32c(), "This instruction only works on SPARC with CRC32C"); }
 629 
 630   // instruction only in VIS1
 631   static void vis1_only() { assert( VM_Version::has_vis1(), "This instruction only works on SPARC with VIS1"); }
 632 
 633   // instruction only in VIS2
 634   static void vis2_only() { assert( VM_Version::has_vis2(), "This instruction only works on SPARC with VIS2"); }
 635 
 636   // instruction only in VIS3
 637   static void vis3_only() { assert( VM_Version::has_vis3(), "This instruction only works on SPARC with VIS3"); }
 638 
 639   // instruction only in v9
 640   static void v9_only() { } // do nothing
 641 
 642   // instruction deprecated in v9
 643   static void v9_dep()  { } // do nothing for now
 644 
 645   // v8 has no CC field
 646   static void v8_no_cc(CC cc)  { if (cc)  v9_only(); }
 647 
 648  protected:
 649   // Simple delay-slot scheme:
 650   // In order to check the programmer, the assembler keeps track of deley slots.
 651   // It forbids CTIs in delay slots (conservative, but should be OK).
 652   // Also, when putting an instruction into a delay slot, you must say
 653   // asm->delayed()->add(...), in order to check that you don't omit
 654   // delay-slot instructions.
 655   // To implement this, we use a simple FSA
 656 
 657 #ifdef ASSERT
 658   #define CHECK_DELAY
 659 #endif
 660 #ifdef CHECK_DELAY
 661   enum Delay_state { no_delay, at_delay_slot, filling_delay_slot } delay_state;
 662 #endif
 663 
 664  public:
 665   // Tells assembler next instruction must NOT be in delay slot.
 666   // Use at start of multinstruction macros.
 667   void assert_not_delayed() {
 668     // This is a separate overloading to avoid creation of string constants
 669     // in non-asserted code--with some compilers this pollutes the object code.
 670 #ifdef CHECK_DELAY
 671     assert_not_delayed("next instruction should not be a delay slot");
 672 #endif
 673   }
 674   void assert_not_delayed(const char* msg) {
 675 #ifdef CHECK_DELAY
 676     assert(delay_state == no_delay, msg);
 677 #endif
 678   }
 679 
 680  protected:
 681   // Insert a nop if the previous is cbcond
 682   inline void insert_nop_after_cbcond();
 683 
 684   // Delay slot helpers
 685   // cti is called when emitting control-transfer instruction,
 686   // BEFORE doing the emitting.
 687   // Only effective when assertion-checking is enabled.
 688   void cti() {
 689     // A cbcond instruction immediately followed by a CTI
 690     // instruction introduces pipeline stalls, we need to avoid that.
 691     no_cbcond_before();
 692 #ifdef CHECK_DELAY
 693     assert_not_delayed("cti should not be in delay slot");
 694 #endif
 695   }
 696 
 697   // called when emitting cti with a delay slot, AFTER emitting
 698   void has_delay_slot() {
 699 #ifdef CHECK_DELAY
 700     assert_not_delayed("just checking");
 701     delay_state = at_delay_slot;
 702 #endif
 703   }
 704 
 705   // cbcond instruction should not be generated one after an other
 706   bool cbcond_before() {
 707     if (offset() == 0) return false; // it is first instruction
 708     int x = *(int*)(intptr_t(pc()) - 4); // previous instruction
 709     return is_cbcond(x);
 710   }
 711 
 712   void no_cbcond_before() {
 713     assert(offset() == 0 || !cbcond_before(), "cbcond should not follow an other cbcond");
 714   }
 715 public:
 716 
 717   bool use_cbcond(Label& L) {
 718     if (!UseCBCond || cbcond_before()) return false;
 719     intptr_t x = intptr_t(target_distance(L)) - intptr_t(pc());
 720     assert( (x & 3) == 0, "not word aligned");
 721     return is_simm12(x);
 722   }
 723 
 724   // Tells assembler you know that next instruction is delayed
 725   Assembler* delayed() {
 726 #ifdef CHECK_DELAY
 727     assert ( delay_state == at_delay_slot, "delayed instruction is not in delay slot");
 728     delay_state = filling_delay_slot;
 729 #endif
 730     return this;
 731   }
 732 
 733   void flush() {
 734 #ifdef CHECK_DELAY
 735     assert ( delay_state == no_delay, "ending code with a delay slot");
 736 #endif
 737     AbstractAssembler::flush();
 738   }
 739 
 740   inline void emit_int32(int);  // shadows AbstractAssembler::emit_int32
 741   inline void emit_data(int x);
 742   inline void emit_data(int, RelocationHolder const&);
 743   inline void emit_data(int, relocInfo::relocType rtype);
 744   // helper for above fcns
 745   inline void check_delay();
 746 
 747 
 748  public:
 749   // instructions, refer to page numbers in the SPARC Architecture Manual, V9
 750 
 751   // pp 135 (addc was addx in v8)
 752 
 753   inline void add(Register s1, Register s2, Register d );
 754   inline void add(Register s1, int simm13a, Register d );
 755 
 756   inline void addcc(  Register s1, Register s2, Register d );
 757   inline void addcc(  Register s1, int simm13a, Register d );
 758   inline void addc(   Register s1, Register s2, Register d );
 759   inline void addc(   Register s1, int simm13a, Register d );
 760   inline void addccc( Register s1, Register s2, Register d );
 761   inline void addccc( Register s1, int simm13a, Register d );
 762 
 763 
 764   // 4-operand AES instructions
 765 
 766   inline void aes_eround01(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 767   inline void aes_eround23(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 768   inline void aes_dround01(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 769   inline void aes_dround23(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 770   inline void aes_eround01_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 771   inline void aes_eround23_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 772   inline void aes_dround01_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 773   inline void aes_dround23_l(  FloatRegister s1, FloatRegister s2, FloatRegister s3, FloatRegister d );
 774   inline void aes_kexpand1(  FloatRegister s1, FloatRegister s2, int imm5a, FloatRegister d );
 775 
 776 
 777   // 3-operand AES instructions
 778 
 779   inline void aes_kexpand0(  FloatRegister s1, FloatRegister s2, FloatRegister d );
 780   inline void aes_kexpand2(  FloatRegister s1, FloatRegister s2, FloatRegister d );
 781 
 782   // pp 136
 783 
 784   inline void bpr(RCondition c, bool a, Predict p, Register s1, address d, relocInfo::relocType rt = relocInfo::none);
 785   inline void bpr(RCondition c, bool a, Predict p, Register s1, Label& L);
 786 
 787   // compare and branch
 788   inline void cbcond(Condition c, CC cc, Register s1, Register s2, Label& L);
 789   inline void cbcond(Condition c, CC cc, Register s1, int simm5, Label& L);
 790 
 791  protected: // use MacroAssembler::br instead
 792 
 793   // pp 138
 794 
 795   inline void fb( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 796   inline void fb( Condition c, bool a, Label& L );
 797 
 798   // pp 141
 799 
 800   inline void fbp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 801   inline void fbp( Condition c, bool a, CC cc, Predict p, Label& L );
 802 
 803   // pp 144
 804 
 805   inline void br( Condition c, bool a, address d, relocInfo::relocType rt = relocInfo::none );
 806   inline void br( Condition c, bool a, Label& L );
 807 
 808   // pp 146
 809 
 810   inline void bp( Condition c, bool a, CC cc, Predict p, address d, relocInfo::relocType rt = relocInfo::none );
 811   inline void bp( Condition c, bool a, CC cc, Predict p, Label& L );
 812 
 813   // pp 149
 814 
 815   inline void call( address d,  relocInfo::relocType rt = relocInfo::runtime_call_type );
 816   inline void call( Label& L,   relocInfo::relocType rt = relocInfo::runtime_call_type );
 817 
 818   inline void call( address d,  RelocationHolder const& rspec );
 819 
 820  public:
 821 
 822   // pp 150
 823 
 824   // These instructions compare the contents of s2 with the contents of
 825   // memory at address in s1. If the values are equal, the contents of memory
 826   // at address s1 is swapped with the data in d. If the values are not equal,
 827   // the the contents of memory at s1 is loaded into d, without the swap.
 828 
 829   inline void casa(  Register s1, Register s2, Register d, int ia = -1 );
 830   inline void casxa( Register s1, Register s2, Register d, int ia = -1 );
 831 
 832   // pp 152
 833 
 834   inline void udiv(   Register s1, Register s2, Register d );
 835   inline void udiv(   Register s1, int simm13a, Register d );
 836   inline void sdiv(   Register s1, Register s2, Register d );
 837   inline void sdiv(   Register s1, int simm13a, Register d );
 838   inline void udivcc( Register s1, Register s2, Register d );
 839   inline void udivcc( Register s1, int simm13a, Register d );
 840   inline void sdivcc( Register s1, Register s2, Register d );
 841   inline void sdivcc( Register s1, int simm13a, Register d );
 842 
 843   // pp 155
 844 
 845   inline void done();
 846   inline void retry();
 847 
 848   // pp 156
 849 
 850   inline void fadd( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d );
 851   inline void fsub( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d );
 852 
 853   // pp 157
 854 
 855   inline void fcmp(  FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2);
 856   inline void fcmpe( FloatRegisterImpl::Width w, CC cc, FloatRegister s1, FloatRegister s2);
 857 
 858   // pp 159
 859 
 860   inline void ftox( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 861   inline void ftoi( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 862 
 863   // pp 160
 864 
 865   inline void ftof( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw, FloatRegister s, FloatRegister d );
 866 
 867   // pp 161
 868 
 869   inline void fxtof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 870   inline void fitof( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 871 
 872   // pp 162
 873 
 874   inline void fmov( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 875 
 876   inline void fneg( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 877 
 878   inline void fabs( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 879 
 880   // pp 163
 881 
 882   inline void fmul( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d );
 883   inline void fmul( FloatRegisterImpl::Width sw, FloatRegisterImpl::Width dw,  FloatRegister s1, FloatRegister s2, FloatRegister d );
 884   inline void fdiv( FloatRegisterImpl::Width w,                            FloatRegister s1, FloatRegister s2, FloatRegister d );
 885 
 886   // FXORs/FXORd instructions
 887 
 888   inline void fxor( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister s2, FloatRegister d );
 889 
 890   // pp 164
 891 
 892   inline void fsqrt( FloatRegisterImpl::Width w, FloatRegister s, FloatRegister d );
 893 
 894   // pp 165
 895 
 896   inline void flush( Register s1, Register s2 );
 897   inline void flush( Register s1, int simm13a);
 898 
 899   // pp 167
 900 
 901   void flushw();
 902 
 903   // pp 168
 904 
 905   void illtrap( int const22a);
 906   // v8 unimp == illtrap(0)
 907 
 908   // pp 169
 909 
 910   void impdep1( int id1, int const19a );
 911   void impdep2( int id1, int const19a );
 912 
 913   // pp 170
 914 
 915   void jmpl( Register s1, Register s2, Register d );
 916   void jmpl( Register s1, int simm13a, Register d, RelocationHolder const& rspec = RelocationHolder() );
 917 
 918   // 171
 919 
 920   inline void ldf(FloatRegisterImpl::Width w, Register s1, Register s2, FloatRegister d);
 921   inline void ldf(FloatRegisterImpl::Width w, Register s1, int simm13a, FloatRegister d, RelocationHolder const& rspec = RelocationHolder());
 922 
 923 
 924   inline void ldfsr(  Register s1, Register s2 );
 925   inline void ldfsr(  Register s1, int simm13a);
 926   inline void ldxfsr( Register s1, Register s2 );
 927   inline void ldxfsr( Register s1, int simm13a);
 928 
 929   // 173
 930 
 931   inline void ldfa(  FloatRegisterImpl::Width w, Register s1, Register s2, int ia, FloatRegister d );
 932   inline void ldfa(  FloatRegisterImpl::Width w, Register s1, int simm13a,         FloatRegister d );
 933 
 934   // pp 175, lduw is ld on v8
 935 
 936   inline void ldsb(  Register s1, Register s2, Register d );
 937   inline void ldsb(  Register s1, int simm13a, Register d);
 938   inline void ldsh(  Register s1, Register s2, Register d );
 939   inline void ldsh(  Register s1, int simm13a, Register d);
 940   inline void ldsw(  Register s1, Register s2, Register d );
 941   inline void ldsw(  Register s1, int simm13a, Register d);
 942   inline void ldub(  Register s1, Register s2, Register d );
 943   inline void ldub(  Register s1, int simm13a, Register d);
 944   inline void lduh(  Register s1, Register s2, Register d );
 945   inline void lduh(  Register s1, int simm13a, Register d);
 946   inline void lduw(  Register s1, Register s2, Register d );
 947   inline void lduw(  Register s1, int simm13a, Register d);
 948   inline void ldx(   Register s1, Register s2, Register d );
 949   inline void ldx(   Register s1, int simm13a, Register d);
 950   inline void ldd(   Register s1, Register s2, Register d );
 951   inline void ldd(   Register s1, int simm13a, Register d);
 952 
 953   // pp 177
 954 
 955   inline void ldsba(  Register s1, Register s2, int ia, Register d );
 956   inline void ldsba(  Register s1, int simm13a,         Register d );
 957   inline void ldsha(  Register s1, Register s2, int ia, Register d );
 958   inline void ldsha(  Register s1, int simm13a,         Register d );
 959   inline void ldswa(  Register s1, Register s2, int ia, Register d );
 960   inline void ldswa(  Register s1, int simm13a,         Register d );
 961   inline void lduba(  Register s1, Register s2, int ia, Register d );
 962   inline void lduba(  Register s1, int simm13a,         Register d );
 963   inline void lduha(  Register s1, Register s2, int ia, Register d );
 964   inline void lduha(  Register s1, int simm13a,         Register d );
 965   inline void lduwa(  Register s1, Register s2, int ia, Register d );
 966   inline void lduwa(  Register s1, int simm13a,         Register d );
 967   inline void ldxa(   Register s1, Register s2, int ia, Register d );
 968   inline void ldxa(   Register s1, int simm13a,         Register d );
 969 
 970   // pp 181
 971 
 972   inline void and3(    Register s1, Register s2, Register d );
 973   inline void and3(    Register s1, int simm13a, Register d );
 974   inline void andcc(   Register s1, Register s2, Register d );
 975   inline void andcc(   Register s1, int simm13a, Register d );
 976   inline void andn(    Register s1, Register s2, Register d );
 977   inline void andn(    Register s1, int simm13a, Register d );
 978   inline void andncc(  Register s1, Register s2, Register d );
 979   inline void andncc(  Register s1, int simm13a, Register d );
 980   inline void or3(     Register s1, Register s2, Register d );
 981   inline void or3(     Register s1, int simm13a, Register d );
 982   inline void orcc(    Register s1, Register s2, Register d );
 983   inline void orcc(    Register s1, int simm13a, Register d );
 984   inline void orn(     Register s1, Register s2, Register d );
 985   inline void orn(     Register s1, int simm13a, Register d );
 986   inline void orncc(   Register s1, Register s2, Register d );
 987   inline void orncc(   Register s1, int simm13a, Register d );
 988   inline void xor3(    Register s1, Register s2, Register d );
 989   inline void xor3(    Register s1, int simm13a, Register d );
 990   inline void xorcc(   Register s1, Register s2, Register d );
 991   inline void xorcc(   Register s1, int simm13a, Register d );
 992   inline void xnor(    Register s1, Register s2, Register d );
 993   inline void xnor(    Register s1, int simm13a, Register d );
 994   inline void xnorcc(  Register s1, Register s2, Register d );
 995   inline void xnorcc(  Register s1, int simm13a, Register d );
 996 
 997   // pp 183
 998 
 999   inline void membar( Membar_mask_bits const7a );
1000 
1001   // pp 185
1002 
1003   inline void fmov( FloatRegisterImpl::Width w, Condition c,  bool floatCC, CC cca, FloatRegister s2, FloatRegister d );
1004 
1005   // pp 189
1006 
1007   inline void fmov( FloatRegisterImpl::Width w, RCondition c, Register s1,  FloatRegister s2, FloatRegister d );
1008 
1009   // pp 191
1010 
1011   inline void movcc( Condition c, bool floatCC, CC cca, Register s2, Register d );
1012   inline void movcc( Condition c, bool floatCC, CC cca, int simm11a, Register d );
1013 
1014   // pp 195
1015 
1016   inline void movr( RCondition c, Register s1, Register s2,  Register d );
1017   inline void movr( RCondition c, Register s1, int simm10a,  Register d );
1018 
1019   // pp 196
1020 
1021   inline void mulx(  Register s1, Register s2, Register d );
1022   inline void mulx(  Register s1, int simm13a, Register d );
1023   inline void sdivx( Register s1, Register s2, Register d );
1024   inline void sdivx( Register s1, int simm13a, Register d );
1025   inline void udivx( Register s1, Register s2, Register d );
1026   inline void udivx( Register s1, int simm13a, Register d );
1027 
1028   // pp 197
1029 
1030   inline void umul(   Register s1, Register s2, Register d );
1031   inline void umul(   Register s1, int simm13a, Register d );
1032   inline void smul(   Register s1, Register s2, Register d );
1033   inline void smul(   Register s1, int simm13a, Register d );
1034   inline void umulcc( Register s1, Register s2, Register d );
1035   inline void umulcc( Register s1, int simm13a, Register d );
1036   inline void smulcc( Register s1, Register s2, Register d );
1037   inline void smulcc( Register s1, int simm13a, Register d );
1038 
1039   // pp 201
1040 
1041   inline void nop();
1042 
1043   inline void sw_count();
1044 
1045   // pp 202
1046 
1047   inline void popc( Register s,  Register d);
1048   inline void popc( int simm13a, Register d);
1049 
1050   // pp 203
1051 
1052   inline void prefetch(   Register s1, Register s2, PrefetchFcn f);
1053   inline void prefetch(   Register s1, int simm13a, PrefetchFcn f);
1054 
1055   inline void prefetcha(  Register s1, Register s2, int ia, PrefetchFcn f );
1056   inline void prefetcha(  Register s1, int simm13a,         PrefetchFcn f );
1057 
1058   // pp 208
1059 
1060   // not implementing read privileged register
1061 
1062   inline void rdy(    Register d);
1063   inline void rdccr(  Register d);
1064   inline void rdasi(  Register d);
1065   inline void rdtick( Register d);
1066   inline void rdpc(   Register d);
1067   inline void rdfprs( Register d);
1068 
1069   // pp 213
1070 
1071   inline void rett( Register s1, Register s2);
1072   inline void rett( Register s1, int simm13a, relocInfo::relocType rt = relocInfo::none);
1073 
1074   // pp 214
1075 
1076   inline void save(    Register s1, Register s2, Register d );
1077   inline void save(    Register s1, int simm13a, Register d );
1078 
1079   inline void restore( Register s1 = G0,  Register s2 = G0, Register d = G0 );
1080   inline void restore( Register s1,       int simm13a,      Register d      );
1081 
1082   // pp 216
1083 
1084   inline void saved();
1085   inline void restored();
1086 
1087   // pp 217
1088 
1089   inline void sethi( int imm22a, Register d, RelocationHolder const& rspec = RelocationHolder() );
1090   // pp 218
1091 
1092   inline void sll(  Register s1, Register s2, Register d );
1093   inline void sll(  Register s1, int imm5a,   Register d );
1094   inline void srl(  Register s1, Register s2, Register d );
1095   inline void srl(  Register s1, int imm5a,   Register d );
1096   inline void sra(  Register s1, Register s2, Register d );
1097   inline void sra(  Register s1, int imm5a,   Register d );
1098 
1099   inline void sllx( Register s1, Register s2, Register d );
1100   inline void sllx( Register s1, int imm6a,   Register d );
1101   inline void srlx( Register s1, Register s2, Register d );
1102   inline void srlx( Register s1, int imm6a,   Register d );
1103   inline void srax( Register s1, Register s2, Register d );
1104   inline void srax( Register s1, int imm6a,   Register d );
1105 
1106   // pp 220
1107 
1108   inline void sir( int simm13a );
1109 
1110   // pp 221
1111 
1112   inline void stbar();
1113 
1114   // pp 222
1115 
1116   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2);
1117   inline void stf(    FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a);
1118 
1119   inline void stfsr(  Register s1, Register s2 );
1120   inline void stfsr(  Register s1, int simm13a);
1121   inline void stxfsr( Register s1, Register s2 );
1122   inline void stxfsr( Register s1, int simm13a);
1123 
1124   //  pp 224
1125 
1126   inline void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, Register s2, int ia );
1127   inline void stfa(  FloatRegisterImpl::Width w, FloatRegister d, Register s1, int simm13a         );
1128 
1129   // p 226
1130 
1131   inline void stb(  Register d, Register s1, Register s2 );
1132   inline void stb(  Register d, Register s1, int simm13a);
1133   inline void sth(  Register d, Register s1, Register s2 );
1134   inline void sth(  Register d, Register s1, int simm13a);
1135   inline void stw(  Register d, Register s1, Register s2 );
1136   inline void stw(  Register d, Register s1, int simm13a);
1137   inline void stx(  Register d, Register s1, Register s2 );
1138   inline void stx(  Register d, Register s1, int simm13a);
1139   inline void std(  Register d, Register s1, Register s2 );
1140   inline void std(  Register d, Register s1, int simm13a);
1141 
1142   // pp 177
1143 
1144   inline void stba(  Register d, Register s1, Register s2, int ia );
1145   inline void stba(  Register d, Register s1, int simm13a         );
1146   inline void stha(  Register d, Register s1, Register s2, int ia );
1147   inline void stha(  Register d, Register s1, int simm13a         );
1148   inline void stwa(  Register d, Register s1, Register s2, int ia );
1149   inline void stwa(  Register d, Register s1, int simm13a         );
1150   inline void stxa(  Register d, Register s1, Register s2, int ia );
1151   inline void stxa(  Register d, Register s1, int simm13a         );
1152   inline void stda(  Register d, Register s1, Register s2, int ia );
1153   inline void stda(  Register d, Register s1, int simm13a         );
1154 
1155   // pp 230
1156 
1157   inline void sub(    Register s1, Register s2, Register d );
1158   inline void sub(    Register s1, int simm13a, Register d );
1159 
1160   inline void subcc(  Register s1, Register s2, Register d );
1161   inline void subcc(  Register s1, int simm13a, Register d );
1162   inline void subc(   Register s1, Register s2, Register d );
1163   inline void subc(   Register s1, int simm13a, Register d );
1164   inline void subccc( Register s1, Register s2, Register d );
1165   inline void subccc( Register s1, int simm13a, Register d );
1166 
1167   // pp 231
1168 
1169   inline void swap( Register s1, Register s2, Register d );
1170   inline void swap( Register s1, int simm13a, Register d);
1171 
1172   // pp 232
1173 
1174   inline void swapa(   Register s1, Register s2, int ia, Register d );
1175   inline void swapa(   Register s1, int simm13a,         Register d );
1176 
1177   // pp 234, note op in book is wrong, see pp 268
1178 
1179   inline void taddcc(    Register s1, Register s2, Register d );
1180   inline void taddcc(    Register s1, int simm13a, Register d );
1181 
1182   // pp 235
1183 
1184   inline void tsubcc(    Register s1, Register s2, Register d );
1185   inline void tsubcc(    Register s1, int simm13a, Register d );
1186 
1187   // pp 237
1188 
1189   inline void trap( Condition c, CC cc, Register s1, Register s2 );
1190   inline void trap( Condition c, CC cc, Register s1, int trapa   );
1191   // simple uncond. trap
1192   inline void trap( int trapa );
1193 
1194   // pp 239 omit write priv register for now
1195 
1196   inline void wry(    Register d);
1197   inline void wrccr(Register s);
1198   inline void wrccr(Register s, int simm13a);
1199   inline void wrasi(Register d);
1200   // wrasi(d, imm) stores (d xor imm) to asi
1201   inline void wrasi(Register d, int simm13a);
1202   inline void wrfprs( Register d);
1203 
1204   //  VIS1 instructions
1205 
1206   inline void alignaddr( Register s1, Register s2, Register d );
1207 
1208   inline void faligndata( FloatRegister s1, FloatRegister s2, FloatRegister d );
1209 
1210   inline void fzero( FloatRegisterImpl::Width w, FloatRegister d );
1211 
1212   inline void fsrc2( FloatRegisterImpl::Width w, FloatRegister s2, FloatRegister d );
1213 
1214   inline void fnot1( FloatRegisterImpl::Width w, FloatRegister s1, FloatRegister d );
1215 
1216   inline void fpmerge( FloatRegister s1, FloatRegister s2, FloatRegister d );
1217 
1218   inline void stpartialf( Register s1, Register s2, FloatRegister d, int ia = -1 );
1219 
1220   //  VIS2 instructions
1221 
1222   inline void edge8n( Register s1, Register s2, Register d );
1223 
1224   inline void bmask( Register s1, Register s2, Register d );
1225   inline void bshuffle( FloatRegister s1, FloatRegister s2, FloatRegister d );
1226 
1227   // VIS3 instructions
1228 
1229   inline void lzcnt( Register s, Register d );
1230 
1231   inline void movstosw( FloatRegister s, Register d );
1232   inline void movstouw( FloatRegister s, Register d );
1233   inline void movdtox(  FloatRegister s, Register d );
1234 
1235   inline void movwtos( Register s, FloatRegister d );
1236   inline void movxtod( Register s, FloatRegister d );
1237 
1238   inline void xmulx(Register s1, Register s2, Register d);
1239   inline void xmulxhi(Register s1, Register s2, Register d);
1240 
1241   // Crypto SHA instructions
1242 
1243   inline void sha1();
1244   inline void sha256();
1245   inline void sha512();
1246 
1247   // CRC32C instruction
1248 
1249   inline void crc32c( FloatRegister s1, FloatRegister s2, FloatRegister d );
1250 
1251   // Creation
1252   Assembler(CodeBuffer* code) : AbstractAssembler(code) {
1253 #ifdef CHECK_DELAY
1254     delay_state = no_delay;
1255 #endif
1256   }
1257 };
1258 
1259 #endif // CPU_SPARC_VM_ASSEMBLER_SPARC_HPP