opcodes/or1k-asm.c

   1 /* DO NOT EDIT!  -*- buffer-read-only: t -*- vi:set ro:  */
   2 /* Assembler interface for targets using CGEN. -*- C -*-
   3    CGEN: Cpu tools GENerator
   4
   5    THIS FILE IS MACHINE GENERATED WITH CGEN.
   6    - the resultant file is machine generated, cgen-asm.in isn't
   7
   8    Copyright (C) 1996-2019 Free Software Foundation, Inc.
   9
  10    This file is part of libopcodes.
  11
  12    This library is free software; you can redistribute it and/or modify
  13    it under the terms of the GNU General Public License as published by
  14    the Free Software Foundation; either version 3, or (at your option)
  15    any later version.
  16
  17    It is distributed in the hope that it will be useful, but WITHOUT
  18    ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
  19    or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
  20    License for more details.
  21
  22    You should have received a copy of the GNU General Public License
  23    along with this program; if not, write to the Free Software Foundation, Inc.,
  24    51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
  25
  26
  27 /* ??? Eventually more and more of this stuff can go to cpu-independent files.
  28    Keep that in mind.  */
  29
  30 #include "sysdep.h"
  31 #include <stdio.h>
  32 #include "ansidecl.h"
  33 #include "bfd.h"
  34 #include "symcat.h"
  35 #include "or1k-desc.h"
  36 #include "or1k-opc.h"
  37 #include "opintl.h"
  38 #include "xregex.h"
  39 #include "libiberty.h"
  40 #include "safe-ctype.h"
  41
  42 #undef  min
  43 #define min(a,b) ((a) < (b) ? (a) : (b))
  44 #undef  max
  45 #define max(a,b) ((a) > (b) ? (a) : (b))
  46
  47 static const char * parse_insn_normal
  48   (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *);
  49 \f
  50 /* -- assembler routines inserted here.  */
  51
  52 /* -- asm.c */
  53
  54 static const char * MISSING_CLOSING_PARENTHESIS = N_("missing `)'");
  55 static const char * INVALID_STORE_RELOC = N_("relocation invalid for store");
  56 static const char * INVALID_RELOC_TYPE = N_("internal relocation type invalid");
  57
  58 #define CGEN_VERBOSE_ASSEMBLER_ERRORS
  59
  60 static const char *
  61 parse_disp26 (CGEN_CPU_DESC cd,
  62               const char ** strp,
  63               int opindex,
  64               int opinfo ATTRIBUTE_UNUSED,
  65               enum cgen_parse_operand_result * resultp,
  66               bfd_vma * valuep)
  67 {
  68   const char *str = *strp;
  69   const char *errmsg = NULL;
  70   bfd_reloc_code_real_type reloc = BFD_RELOC_OR1K_REL_26;
  71
  72   if (strncasecmp (str, "plta(", 5) == 0)
  73     {
  74       *strp = str + 5;
  75       reloc = BFD_RELOC_OR1K_PLTA26;
  76     }
  77   else if (strncasecmp (str, "plt(", 4) == 0)
  78     {
  79       *strp = str + 4;
  80       reloc = BFD_RELOC_OR1K_PLT26;
  81     }
  82
  83   errmsg = cgen_parse_address (cd, strp, opindex, reloc, resultp, valuep);
  84
  85   if (reloc != BFD_RELOC_OR1K_REL_26)
  86     {
  87       if (**strp != ')')
  88         errmsg = MISSING_CLOSING_PARENTHESIS;
  89       else
  90         ++*strp;
  91     }
  92
  93   return errmsg;
  94 }
  95
  96 static const char *
  97 parse_disp21 (CGEN_CPU_DESC cd,
  98               const char ** strp,
  99               int opindex,
 100               int opinfo ATTRIBUTE_UNUSED,
 101               enum cgen_parse_operand_result * resultp,
 102               bfd_vma * valuep)
 103 {
 104   const char *str = *strp;
 105   const char *errmsg = NULL;
 106   bfd_reloc_code_real_type reloc = BFD_RELOC_OR1K_PCREL_PG21;
 107
 108   if (strncasecmp (str, "got(", 4) == 0)
 109     {
 110       *strp = str + 4;
 111       reloc = BFD_RELOC_OR1K_GOT_PG21;
 112     }
 113   else if (strncasecmp (str, "tlsgd(", 6) == 0)
 114     {
 115       *strp = str + 6;
 116       reloc = BFD_RELOC_OR1K_TLS_GD_PG21;
 117     }
 118   else if (strncasecmp (str, "tlsldm(", 7) == 0)
 119     {
 120       *strp = str + 7;
 121       reloc = BFD_RELOC_OR1K_TLS_LDM_PG21;
 122     }
 123   else if (strncasecmp (str, "gottp(", 6) == 0)
 124     {
 125       *strp = str + 6;
 126       reloc = BFD_RELOC_OR1K_TLS_IE_PG21;
 127     }
 128
 129   errmsg = cgen_parse_address (cd, strp, opindex, reloc, resultp, valuep);
 130
 131   if (reloc != BFD_RELOC_OR1K_PCREL_PG21)
 132     {
 133       if (**strp != ')')
 134         errmsg = MISSING_CLOSING_PARENTHESIS;
 135       else
 136         ++*strp;
 137     }
 138
 139   return errmsg;
 140 }
 141
 142 enum or1k_rclass
 143 {
 144   RCLASS_DIRECT   = 0,
 145   RCLASS_GOT      = 1,
 146   RCLASS_GOTPC    = 2,
 147   RCLASS_GOTOFF   = 3,
 148   RCLASS_TLSGD    = 4,
 149   RCLASS_TLSLDM   = 5,
 150   RCLASS_DTPOFF   = 6,
 151   RCLASS_GOTTPOFF = 7,
 152   RCLASS_TPOFF    = 8,
 153 };
 154
 155 enum or1k_rtype
 156 {
 157   RTYPE_LO = 0,
 158   RTYPE_SLO = 1,
 159   RTYPE_PO = 2,
 160   RTYPE_SPO = 3,
 161   RTYPE_HI = 4,
 162   RTYPE_AHI = 5,
 163 };
 164
 165 #define RCLASS_SHIFT 3
 166 #define RTYPE_MASK   7
 167
 168 static const bfd_reloc_code_real_type or1k_imm16_relocs[][6] = {
 169   { BFD_RELOC_LO16,
 170     BFD_RELOC_OR1K_SLO16,
 171     BFD_RELOC_OR1K_LO13,
 172     BFD_RELOC_OR1K_SLO13,
 173     BFD_RELOC_HI16,
 174     BFD_RELOC_HI16_S, },
 175   { BFD_RELOC_OR1K_GOT16,
 176     BFD_RELOC_UNUSED,
 177     BFD_RELOC_OR1K_GOT_LO13,
 178     BFD_RELOC_UNUSED,
 179     BFD_RELOC_UNUSED,
 180     BFD_RELOC_UNUSED },
 181   { BFD_RELOC_OR1K_GOTPC_LO16,
 182     BFD_RELOC_UNUSED,
 183     BFD_RELOC_UNUSED,
 184     BFD_RELOC_UNUSED,
 185     BFD_RELOC_OR1K_GOTPC_HI16,
 186     BFD_RELOC_UNUSED },
 187   { BFD_RELOC_LO16_GOTOFF,
 188     BFD_RELOC_OR1K_GOTOFF_SLO16,
 189     BFD_RELOC_UNUSED,
 190     BFD_RELOC_UNUSED,
 191     BFD_RELOC_HI16_GOTOFF,
 192     BFD_RELOC_HI16_S_GOTOFF },
 193   { BFD_RELOC_OR1K_TLS_GD_LO16,
 194     BFD_RELOC_UNUSED,
 195     BFD_RELOC_OR1K_TLS_GD_LO13,
 196     BFD_RELOC_UNUSED,
 197     BFD_RELOC_OR1K_TLS_GD_HI16,
 198     BFD_RELOC_UNUSED },
 199   { BFD_RELOC_OR1K_TLS_LDM_LO16,
 200     BFD_RELOC_UNUSED,
 201     BFD_RELOC_OR1K_TLS_LDM_LO13,
 202     BFD_RELOC_UNUSED,
 203     BFD_RELOC_OR1K_TLS_LDM_HI16,
 204     BFD_RELOC_UNUSED },
 205   { BFD_RELOC_OR1K_TLS_LDO_LO16,
 206     BFD_RELOC_UNUSED,
 207     BFD_RELOC_UNUSED,
 208     BFD_RELOC_UNUSED,
 209     BFD_RELOC_OR1K_TLS_LDO_HI16,
 210     BFD_RELOC_UNUSED },
 211   { BFD_RELOC_OR1K_TLS_IE_LO16,
 212     BFD_RELOC_UNUSED,
 213     BFD_RELOC_OR1K_TLS_IE_LO13,
 214     BFD_RELOC_UNUSED,
 215     BFD_RELOC_OR1K_TLS_IE_HI16,
 216     BFD_RELOC_OR1K_TLS_IE_AHI16 },
 217   { BFD_RELOC_OR1K_TLS_LE_LO16,
 218     BFD_RELOC_OR1K_TLS_LE_SLO16,
 219     BFD_RELOC_UNUSED,
 220     BFD_RELOC_UNUSED,
 221     BFD_RELOC_OR1K_TLS_LE_HI16,
 222     BFD_RELOC_OR1K_TLS_LE_AHI16 },
 223 };
 224
 225 static int
 226 parse_reloc (const char **strp)
 227 {
 228     const char *str = *strp;
 229     enum or1k_rclass cls = RCLASS_DIRECT;
 230     enum or1k_rtype typ;
 231
 232     if (strncasecmp (str, "got(", 4) == 0)
 233       {
 234         *strp = str + 4;
 235         return (RCLASS_GOT << RCLASS_SHIFT) | RTYPE_LO;
 236       }
 237     if (strncasecmp (str, "gotpo(", 6) == 0)
 238       {
 239         *strp = str + 6;
 240         return (RCLASS_GOT << RCLASS_SHIFT) | RTYPE_PO;
 241       }
 242     if (strncasecmp (str, "gottppo(", 8) == 0)
 243       {
 244         *strp = str + 8;
 245         return (RCLASS_GOTTPOFF << RCLASS_SHIFT) | RTYPE_PO;
 246       }
 247
 248     if (strncasecmp (str, "gotpc", 5) == 0)
 249       {
 250         str += 5;
 251         cls = RCLASS_GOTPC;
 252       }
 253     else if (strncasecmp (str, "gotoff", 6) == 0)
 254       {
 255         str += 6;
 256         cls = RCLASS_GOTOFF;
 257       }
 258     else if (strncasecmp (str, "tlsgd", 5) == 0)
 259       {
 260         str += 5;
 261         cls = RCLASS_TLSGD;
 262       }
 263     else if (strncasecmp (str, "tlsldm", 6) == 0)
 264       {
 265         str += 6;
 266         cls = RCLASS_TLSLDM;
 267       }
 268     else if (strncasecmp (str, "dtpoff", 6) == 0)
 269       {
 270         str += 6;
 271         cls = RCLASS_DTPOFF;
 272       }
 273     else if (strncasecmp (str, "gottpoff", 8) == 0)
 274       {
 275         str += 8;
 276         cls = RCLASS_GOTTPOFF;
 277       }
 278     else if (strncasecmp (str, "tpoff", 5) == 0)
 279       {
 280         str += 5;
 281         cls = RCLASS_TPOFF;
 282       }
 283
 284     if (strncasecmp (str, "hi(", 3) == 0)
 285       {
 286         str += 3;
 287         typ = RTYPE_HI;
 288       }
 289     else if (strncasecmp (str, "lo(", 3) == 0)
 290       {
 291         str += 3;
 292         typ = RTYPE_LO;
 293       }
 294     else if (strncasecmp (str, "ha(", 3) == 0)
 295       {
 296         str += 3;
 297         typ = RTYPE_AHI;
 298       }
 299     else if (strncasecmp (str, "po(", 3) == 0 && cls != RCLASS_GOTTPOFF)
 300       {
 301         str += 3;
 302         typ = RTYPE_PO;
 303       }
 304     else
 305       return -1;
 306
 307     *strp = str;
 308     return (cls << RCLASS_SHIFT) | typ;
 309 }
 310
 311 static const char *
 312 parse_imm16 (CGEN_CPU_DESC cd, const char **strp, int opindex,
 313              long *valuep, int splitp)
 314 {
 315   const char *errmsg;
 316   enum cgen_parse_operand_result result_type;
 317   bfd_reloc_code_real_type reloc = BFD_RELOC_UNUSED;
 318   enum or1k_rtype reloc_type;
 319   int reloc_code;
 320   bfd_vma ret;
 321
 322   if (**strp == '#')
 323     ++*strp;
 324
 325   reloc_code = parse_reloc (strp);
 326   reloc_type = reloc_code & RTYPE_MASK;
 327   if (reloc_code >= 0)
 328     {
 329       enum or1k_rclass reloc_class = reloc_code >> RCLASS_SHIFT;
 330       if (splitp)
 331         {
 332           if ((reloc_type == RTYPE_LO || reloc_type == RTYPE_PO)
 333               && reloc_class != RCLASS_GOT)
 334             /* If split we or up the type to RTYPE_SLO or RTYPE_SPO.  */
 335             reloc_type |= 1;
 336           else
 337             return INVALID_STORE_RELOC;
 338         }
 339       reloc = or1k_imm16_relocs[reloc_class][reloc_type];
 340     }
 341
 342   if (reloc != BFD_RELOC_UNUSED)
 343     {
 344       bfd_vma value;
 345
 346       errmsg = cgen_parse_address (cd, strp, opindex, reloc,
 347                                    &result_type, &value);
 348       if (**strp != ')')
 349         errmsg = MISSING_CLOSING_PARENTHESIS;
 350       ++*strp;
 351
 352       ret = value;
 353
 354       if (errmsg == NULL && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 355         switch (reloc_type)
 356           {
 357           case RTYPE_AHI:
 358             ret += 0x8000;
 359             /* FALLTHRU */
 360           case RTYPE_HI:
 361             ret >>= 16;
 362             /* FALLTHRU */
 363           case RTYPE_LO:
 364           case RTYPE_SLO:
 365             ret &= 0xffff;
 366             ret = (ret ^ 0x8000) - 0x8000;
 367             break;
 368           case RTYPE_PO:
 369           case RTYPE_SPO:
 370             ret &= 0x1fff;
 371             break;
 372           default:
 373             errmsg = INVALID_RELOC_TYPE;
 374           }
 375     }
 376   else
 377     {
 378       long value;
 379       errmsg = cgen_parse_signed_integer (cd, strp, opindex, &value);
 380       ret = value;
 381     }
 382
 383   if (errmsg == NULL)
 384     *valuep = ret;
 385
 386   return errmsg;
 387 }
 388
 389 static const char *
 390 parse_simm16 (CGEN_CPU_DESC cd, const char **strp, int opindex, long *valuep)
 391 {
 392   return parse_imm16(cd, strp, opindex, (long *) valuep, 0);
 393 }
 394
 395 static const char *
 396 parse_simm16_split (CGEN_CPU_DESC cd, const char **strp, int opindex,
 397                     long *valuep)
 398 {
 399   return parse_imm16(cd, strp, opindex, (long *) valuep, 1);
 400 }
 401
 402 static const char *
 403 parse_uimm16 (CGEN_CPU_DESC cd, const char **strp, int opindex,
 404               unsigned long *valuep)
 405 {
 406   const char *errmsg = parse_imm16(cd, strp, opindex, (long *) valuep, 0);
 407   if (errmsg == NULL)
 408     *valuep &= 0xffff;
 409   return errmsg;
 410 }
 411
 412 static const char *
 413 parse_uimm16_split (CGEN_CPU_DESC cd, const char **strp, int opindex,
 414                     unsigned long *valuep)
 415 {
 416   const char *errmsg = parse_imm16(cd, strp, opindex, (long *) valuep, 1);
 417   if (errmsg == NULL)
 418     *valuep &= 0xffff;
 419   return errmsg;
 420 }
 421
 422 /* Parse register pairs with syntax rA,rB to a flag + rA value.  */
 423
 424 static const char *
 425 parse_regpair (CGEN_CPU_DESC cd, const char **strp,
 426                int opindex ATTRIBUTE_UNUSED, unsigned long *valuep)
 427 {
 428   long reg1_index;
 429   long reg2_index;
 430   const char *errmsg;
 431
 432   /* The first part should just be a register.  */
 433   errmsg = cgen_parse_keyword (cd, strp, &or1k_cgen_opval_h_gpr,
 434                                &reg1_index);
 435
 436   /* If that worked skip the comma separator.  */
 437   if (errmsg == NULL)
 438     {
 439       if (**strp == ',')
 440         ++*strp;
 441       else
 442         errmsg = "Unexpected character, expected ','";
 443     }
 444
 445   /* If that worked the next part is just another register.  */
 446   if (errmsg == NULL)
 447     errmsg = cgen_parse_keyword (cd, strp, &or1k_cgen_opval_h_gpr,
 448                                  &reg2_index);
 449
 450   /* Validate the register pair is valid and create the output value.  */
 451   if (errmsg == NULL)
 452     {
 453       int regoffset = reg2_index - reg1_index;
 454
 455       if (regoffset == 1 || regoffset == 2)
 456         {
 457           unsigned short offsetmask;
 458           unsigned short value;
 459
 460           offsetmask = ((regoffset == 2 ? 1 : 0) << 5);
 461           value = offsetmask | reg1_index;
 462
 463           *valuep = value;
 464         }
 465       else
 466         errmsg = "Invalid register pair, offset not 1 or 2.";
 467     }
 468
 469   return errmsg;
 470 }
 471
 472 /* -- */
 473
 474 const char * or1k_cgen_parse_operand
 475   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
 476
 477 /* Main entry point for operand parsing.
 478
 479    This function is basically just a big switch statement.  Earlier versions
 480    used tables to look up the function to use, but
 481    - if the table contains both assembler and disassembler functions then
 482      the disassembler contains much of the assembler and vice-versa,
 483    - there's a lot of inlining possibilities as things grow,
 484    - using a switch statement avoids the function call overhead.
 485
 486    This function could be moved into `parse_insn_normal', but keeping it
 487    separate makes clear the interface between `parse_insn_normal' and each of
 488    the handlers.  */
 489
 490 const char *
 491 or1k_cgen_parse_operand (CGEN_CPU_DESC cd,
 492                            int opindex,
 493                            const char ** strp,
 494                            CGEN_FIELDS * fields)
 495 {
 496   const char * errmsg = NULL;
 497   /* Used by scalar operands that still need to be parsed.  */
 498   long junk ATTRIBUTE_UNUSED;
 499
 500   switch (opindex)
 501     {
 502     case OR1K_OPERAND_DISP21 :
 503       {
 504         bfd_vma value = 0;
 505         errmsg = parse_disp21 (cd, strp, OR1K_OPERAND_DISP21, 0, NULL,  & value);
 506         fields->f_disp21 = value;
 507       }
 508       break;
 509     case OR1K_OPERAND_DISP26 :
 510       {
 511         bfd_vma value = 0;
 512         errmsg = parse_disp26 (cd, strp, OR1K_OPERAND_DISP26, 0, NULL,  & value);
 513         fields->f_disp26 = value;
 514       }
 515       break;
 516     case OR1K_OPERAND_RA :
 517       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r2);
 518       break;
 519     case OR1K_OPERAND_RAD32F :
 520       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RAD32F, (unsigned long *) (& fields->f_rad32));
 521       break;
 522     case OR1K_OPERAND_RADF :
 523       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r2);
 524       break;
 525     case OR1K_OPERAND_RADI :
 526       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RADI, (unsigned long *) (& fields->f_rad32));
 527       break;
 528     case OR1K_OPERAND_RASF :
 529       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r2);
 530       break;
 531     case OR1K_OPERAND_RB :
 532       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r3);
 533       break;
 534     case OR1K_OPERAND_RBD32F :
 535       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RBD32F, (unsigned long *) (& fields->f_rbd32));
 536       break;
 537     case OR1K_OPERAND_RBDF :
 538       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r3);
 539       break;
 540     case OR1K_OPERAND_RBDI :
 541       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RBDI, (unsigned long *) (& fields->f_rbd32));
 542       break;
 543     case OR1K_OPERAND_RBSF :
 544       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r3);
 545       break;
 546     case OR1K_OPERAND_RD :
 547       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r1);
 548       break;
 549     case OR1K_OPERAND_RDD32F :
 550       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RDD32F, (unsigned long *) (& fields->f_rdd32));
 551       break;
 552     case OR1K_OPERAND_RDDF :
 553       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r1);
 554       break;
 555     case OR1K_OPERAND_RDDI :
 556       errmsg = parse_regpair (cd, strp, OR1K_OPERAND_RDDI, (unsigned long *) (& fields->f_rdd32));
 557       break;
 558     case OR1K_OPERAND_RDSF :
 559       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r1);
 560       break;
 561     case OR1K_OPERAND_SIMM16 :
 562       errmsg = parse_simm16 (cd, strp, OR1K_OPERAND_SIMM16, (long *) (& fields->f_simm16));
 563       break;
 564     case OR1K_OPERAND_SIMM16_SPLIT :
 565       errmsg = parse_simm16_split (cd, strp, OR1K_OPERAND_SIMM16_SPLIT, (long *) (& fields->f_simm16_split));
 566       break;
 567     case OR1K_OPERAND_UIMM16 :
 568       errmsg = parse_uimm16 (cd, strp, OR1K_OPERAND_UIMM16, (unsigned long *) (& fields->f_uimm16));
 569       break;
 570     case OR1K_OPERAND_UIMM16_SPLIT :
 571       errmsg = parse_uimm16_split (cd, strp, OR1K_OPERAND_UIMM16_SPLIT, (unsigned long *) (& fields->f_uimm16_split));
 572       break;
 573     case OR1K_OPERAND_UIMM6 :
 574       errmsg = cgen_parse_unsigned_integer (cd, strp, OR1K_OPERAND_UIMM6, (unsigned long *) (& fields->f_uimm6));
 575       break;
 576
 577     default :
 578       /* xgettext:c-format */
 579       opcodes_error_handler
 580         (_("internal error: unrecognized field %d while parsing"),
 581          opindex);
 582       abort ();
 583   }
 584
 585   return errmsg;
 586 }
 587
 588 cgen_parse_fn * const or1k_cgen_parse_handlers[] =
 589 {
 590   parse_insn_normal,
 591 };
 592
 593 void
 594 or1k_cgen_init_asm (CGEN_CPU_DESC cd)
 595 {
 596   or1k_cgen_init_opcode_table (cd);
 597   or1k_cgen_init_ibld_table (cd);
 598   cd->parse_handlers = & or1k_cgen_parse_handlers[0];
 599   cd->parse_operand = or1k_cgen_parse_operand;
 600 #ifdef CGEN_ASM_INIT_HOOK
 601 CGEN_ASM_INIT_HOOK
 602 #endif
 603 }
 604
 605 \f
 606
 607 /* Regex construction routine.
 608
 609    This translates an opcode syntax string into a regex string,
 610    by replacing any non-character syntax element (such as an
 611    opcode) with the pattern '.*'
 612
 613    It then compiles the regex and stores it in the opcode, for
 614    later use by or1k_cgen_assemble_insn
 615
 616    Returns NULL for success, an error message for failure.  */
 617
 618 char *
 619 or1k_cgen_build_insn_regex (CGEN_INSN *insn)
 620 {
 621   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
 622   const char *mnem = CGEN_INSN_MNEMONIC (insn);
 623   char rxbuf[CGEN_MAX_RX_ELEMENTS];
 624   char *rx = rxbuf;
 625   const CGEN_SYNTAX_CHAR_TYPE *syn;
 626   int reg_err;
 627
 628   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
 629
 630   /* Mnemonics come first in the syntax string.  */
 631   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 632     return _("missing mnemonic in syntax string");
 633   ++syn;
 634
 635   /* Generate a case sensitive regular expression that emulates case
 636      insensitive matching in the "C" locale.  We cannot generate a case
 637      insensitive regular expression because in Turkish locales, 'i' and 'I'
 638      are not equal modulo case conversion.  */
 639
 640   /* Copy the literal mnemonic out of the insn.  */
 641   for (; *mnem; mnem++)
 642     {
 643       char c = *mnem;
 644
 645       if (ISALPHA (c))
 646         {
 647           *rx++ = '[';
 648           *rx++ = TOLOWER (c);
 649           *rx++ = TOUPPER (c);
 650           *rx++ = ']';
 651         }
 652       else
 653         *rx++ = c;
 654     }
 655
 656   /* Copy any remaining literals from the syntax string into the rx.  */
 657   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
 658     {
 659       if (CGEN_SYNTAX_CHAR_P (* syn))
 660         {
 661           char c = CGEN_SYNTAX_CHAR (* syn);
 662
 663           switch (c)
 664             {
 665               /* Escape any regex metacharacters in the syntax.  */
 666             case '.': case '[': case '\\':
 667             case '*': case '^': case '$':
 668
 669 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
 670             case '?': case '{': case '}':
 671             case '(': case ')': case '*':
 672             case '|': case '+': case ']':
 673 #endif
 674               *rx++ = '\\';
 675               *rx++ = c;
 676               break;
 677
 678             default:
 679               if (ISALPHA (c))
 680                 {
 681                   *rx++ = '[';
 682                   *rx++ = TOLOWER (c);
 683                   *rx++ = TOUPPER (c);
 684                   *rx++ = ']';
 685                 }
 686               else
 687                 *rx++ = c;
 688               break;
 689             }
 690         }
 691       else
 692         {
 693           /* Replace non-syntax fields with globs.  */
 694           *rx++ = '.';
 695           *rx++ = '*';
 696         }
 697     }
 698
 699   /* Trailing whitespace ok.  */
 700   * rx++ = '[';
 701   * rx++ = ' ';
 702   * rx++ = '\t';
 703   * rx++ = ']';
 704   * rx++ = '*';
 705
 706   /* But anchor it after that.  */
 707   * rx++ = '$';
 708   * rx = '\0';
 709
 710   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
 711   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
 712
 713   if (reg_err == 0)
 714     return NULL;
 715   else
 716     {
 717       static char msg[80];
 718
 719       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
 720       regfree ((regex_t *) CGEN_INSN_RX (insn));
 721       free (CGEN_INSN_RX (insn));
 722       (CGEN_INSN_RX (insn)) = NULL;
 723       return msg;
 724     }
 725 }
 726
 727 \f
 728 /* Default insn parser.
 729
 730    The syntax string is scanned and operands are parsed and stored in FIELDS.
 731    Relocs are queued as we go via other callbacks.
 732
 733    ??? Note that this is currently an all-or-nothing parser.  If we fail to
 734    parse the instruction, we return 0 and the caller will start over from
 735    the beginning.  Backtracking will be necessary in parsing subexpressions,
 736    but that can be handled there.  Not handling backtracking here may get
 737    expensive in the case of the m68k.  Deal with later.
 738
 739    Returns NULL for success, an error message for failure.  */
 740
 741 static const char *
 742 parse_insn_normal (CGEN_CPU_DESC cd,
 743                    const CGEN_INSN *insn,
 744                    const char **strp,
 745                    CGEN_FIELDS *fields)
 746 {
 747   /* ??? Runtime added insns not handled yet.  */
 748   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
 749   const char *str = *strp;
 750   const char *errmsg;
 751   const char *p;
 752   const CGEN_SYNTAX_CHAR_TYPE * syn;
 753 #ifdef CGEN_MNEMONIC_OPERANDS
 754   /* FIXME: wip */
 755   int past_opcode_p;
 756 #endif
 757
 758   /* For now we assume the mnemonic is first (there are no leading operands).
 759      We can parse it without needing to set up operand parsing.
 760      GAS's input scrubber will ensure mnemonics are lowercase, but we may
 761      not be called from GAS.  */
 762   p = CGEN_INSN_MNEMONIC (insn);
 763   while (*p && TOLOWER (*p) == TOLOWER (*str))
 764     ++p, ++str;
 765
 766   if (* p)
 767     return _("unrecognized instruction");
 768
 769 #ifndef CGEN_MNEMONIC_OPERANDS
 770   if (* str && ! ISSPACE (* str))
 771     return _("unrecognized instruction");
 772 #endif
 773
 774   CGEN_INIT_PARSE (cd);
 775   cgen_init_parse_operand (cd);
 776 #ifdef CGEN_MNEMONIC_OPERANDS
 777   past_opcode_p = 0;
 778 #endif
 779
 780   /* We don't check for (*str != '\0') here because we want to parse
 781      any trailing fake arguments in the syntax string.  */
 782   syn = CGEN_SYNTAX_STRING (syntax);
 783
 784   /* Mnemonics come first for now, ensure valid string.  */
 785   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 786     abort ();
 787
 788   ++syn;
 789
 790   while (* syn != 0)
 791     {
 792       /* Non operand chars must match exactly.  */
 793       if (CGEN_SYNTAX_CHAR_P (* syn))
 794         {
 795           /* FIXME: While we allow for non-GAS callers above, we assume the
 796              first char after the mnemonic part is a space.  */
 797           /* FIXME: We also take inappropriate advantage of the fact that
 798              GAS's input scrubber will remove extraneous blanks.  */
 799           if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
 800             {
 801 #ifdef CGEN_MNEMONIC_OPERANDS
 802               if (CGEN_SYNTAX_CHAR(* syn) == ' ')
 803                 past_opcode_p = 1;
 804 #endif
 805               ++ syn;
 806               ++ str;
 807             }
 808           else if (*str)
 809             {
 810               /* Syntax char didn't match.  Can't be this insn.  */
 811               static char msg [80];
 812
 813               /* xgettext:c-format */
 814               sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
 815                        CGEN_SYNTAX_CHAR(*syn), *str);
 816               return msg;
 817             }
 818           else
 819             {
 820               /* Ran out of input.  */
 821               static char msg [80];
 822
 823               /* xgettext:c-format */
 824               sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
 825                        CGEN_SYNTAX_CHAR(*syn));
 826               return msg;
 827             }
 828           continue;
 829         }
 830
 831 #ifdef CGEN_MNEMONIC_OPERANDS
 832       (void) past_opcode_p;
 833 #endif
 834       /* We have an operand of some sort.  */
 835       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn), &str, fields);
 836       if (errmsg)
 837         return errmsg;
 838
 839       /* Done with this operand, continue with next one.  */
 840       ++ syn;
 841     }
 842
 843   /* If we're at the end of the syntax string, we're done.  */
 844   if (* syn == 0)
 845     {
 846       /* FIXME: For the moment we assume a valid `str' can only contain
 847          blanks now.  IE: We needn't try again with a longer version of
 848          the insn and it is assumed that longer versions of insns appear
 849          before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
 850       while (ISSPACE (* str))
 851         ++ str;
 852
 853       if (* str != '\0')
 854         return _("junk at end of line"); /* FIXME: would like to include `str' */
 855
 856       return NULL;
 857     }
 858
 859   /* We couldn't parse it.  */
 860   return _("unrecognized instruction");
 861 }
 862 \f
 863 /* Main entry point.
 864    This routine is called for each instruction to be assembled.
 865    STR points to the insn to be assembled.
 866    We assume all necessary tables have been initialized.
 867    The assembled instruction, less any fixups, is stored in BUF.
 868    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
 869    still needs to be converted to target byte order, otherwise BUF is an array
 870    of bytes in target byte order.
 871    The result is a pointer to the insn's entry in the opcode table,
 872    or NULL if an error occured (an error message will have already been
 873    printed).
 874
 875    Note that when processing (non-alias) macro-insns,
 876    this function recurses.
 877
 878    ??? It's possible to make this cpu-independent.
 879    One would have to deal with a few minor things.
 880    At this point in time doing so would be more of a curiosity than useful
 881    [for example this file isn't _that_ big], but keeping the possibility in
 882    mind helps keep the design clean.  */
 883
 884 const CGEN_INSN *
 885 or1k_cgen_assemble_insn (CGEN_CPU_DESC cd,
 886                            const char *str,
 887                            CGEN_FIELDS *fields,
 888                            CGEN_INSN_BYTES_PTR buf,
 889                            char **errmsg)
 890 {
 891   const char *start;
 892   CGEN_INSN_LIST *ilist;
 893   const char *parse_errmsg = NULL;
 894   const char *insert_errmsg = NULL;
 895   int recognized_mnemonic = 0;
 896
 897   /* Skip leading white space.  */
 898   while (ISSPACE (* str))
 899     ++ str;
 900
 901   /* The instructions are stored in hashed lists.
 902      Get the first in the list.  */
 903   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
 904
 905   /* Keep looking until we find a match.  */
 906   start = str;
 907   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
 908     {
 909       const CGEN_INSN *insn = ilist->insn;
 910       recognized_mnemonic = 1;
 911
 912 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
 913       /* Not usually needed as unsupported opcodes
 914          shouldn't be in the hash lists.  */
 915       /* Is this insn supported by the selected cpu?  */
 916       if (! or1k_cgen_insn_supported (cd, insn))
 917         continue;
 918 #endif
 919       /* If the RELAXED attribute is set, this is an insn that shouldn't be
 920          chosen immediately.  Instead, it is used during assembler/linker
 921          relaxation if possible.  */
 922       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
 923         continue;
 924
 925       str = start;
 926
 927       /* Skip this insn if str doesn't look right lexically.  */
 928       if (CGEN_INSN_RX (insn) != NULL &&
 929           regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
 930         continue;
 931
 932       /* Allow parse/insert handlers to obtain length of insn.  */
 933       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
 934
 935       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
 936       if (parse_errmsg != NULL)
 937         continue;
 938
 939       /* ??? 0 is passed for `pc'.  */
 940       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
 941                                                  (bfd_vma) 0);
 942       if (insert_errmsg != NULL)
 943         continue;
 944
 945       /* It is up to the caller to actually output the insn and any
 946          queued relocs.  */
 947       return insn;
 948     }
 949
 950   {
 951     static char errbuf[150];
 952     const char *tmp_errmsg;
 953 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
 954 #define be_verbose 1
 955 #else
 956 #define be_verbose 0
 957 #endif
 958
 959     if (be_verbose)
 960       {
 961         /* If requesting verbose error messages, use insert_errmsg.
 962            Failing that, use parse_errmsg.  */
 963         tmp_errmsg = (insert_errmsg ? insert_errmsg :
 964                       parse_errmsg ? parse_errmsg :
 965                       recognized_mnemonic ?
 966                       _("unrecognized form of instruction") :
 967                       _("unrecognized instruction"));
 968
 969         if (strlen (start) > 50)
 970           /* xgettext:c-format */
 971           sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
 972         else
 973           /* xgettext:c-format */
 974           sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
 975       }
 976     else
 977       {
 978         if (strlen (start) > 50)
 979           /* xgettext:c-format */
 980           sprintf (errbuf, _("bad instruction `%.50s...'"), start);
 981         else
 982           /* xgettext:c-format */
 983           sprintf (errbuf, _("bad instruction `%.50s'"), start);
 984       }
 985
 986     *errmsg = errbuf;
 987     return NULL;
 988   }
 989 }