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 /* -- */
 423
 424 const char * or1k_cgen_parse_operand
 425   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
 426
 427 /* Main entry point for operand parsing.
 428
 429    This function is basically just a big switch statement.  Earlier versions
 430    used tables to look up the function to use, but
 431    - if the table contains both assembler and disassembler functions then
 432      the disassembler contains much of the assembler and vice-versa,
 433    - there's a lot of inlining possibilities as things grow,
 434    - using a switch statement avoids the function call overhead.
 435
 436    This function could be moved into `parse_insn_normal', but keeping it
 437    separate makes clear the interface between `parse_insn_normal' and each of
 438    the handlers.  */
 439
 440 const char *
 441 or1k_cgen_parse_operand (CGEN_CPU_DESC cd,
 442                            int opindex,
 443                            const char ** strp,
 444                            CGEN_FIELDS * fields)
 445 {
 446   const char * errmsg = NULL;
 447   /* Used by scalar operands that still need to be parsed.  */
 448   long junk ATTRIBUTE_UNUSED;
 449
 450   switch (opindex)
 451     {
 452     case OR1K_OPERAND_DISP21 :
 453       {
 454         bfd_vma value = 0;
 455         errmsg = parse_disp21 (cd, strp, OR1K_OPERAND_DISP21, 0, NULL,  & value);
 456         fields->f_disp21 = value;
 457       }
 458       break;
 459     case OR1K_OPERAND_DISP26 :
 460       {
 461         bfd_vma value = 0;
 462         errmsg = parse_disp26 (cd, strp, OR1K_OPERAND_DISP26, 0, NULL,  & value);
 463         fields->f_disp26 = value;
 464       }
 465       break;
 466     case OR1K_OPERAND_RA :
 467       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r2);
 468       break;
 469     case OR1K_OPERAND_RADF :
 470       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r1);
 471       break;
 472     case OR1K_OPERAND_RASF :
 473       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r2);
 474       break;
 475     case OR1K_OPERAND_RB :
 476       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r3);
 477       break;
 478     case OR1K_OPERAND_RBDF :
 479       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r1);
 480       break;
 481     case OR1K_OPERAND_RBSF :
 482       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r3);
 483       break;
 484     case OR1K_OPERAND_RD :
 485       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_gpr, & fields->f_r1);
 486       break;
 487     case OR1K_OPERAND_RDDF :
 488       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fdr, & fields->f_r1);
 489       break;
 490     case OR1K_OPERAND_RDSF :
 491       errmsg = cgen_parse_keyword (cd, strp, & or1k_cgen_opval_h_fsr, & fields->f_r1);
 492       break;
 493     case OR1K_OPERAND_SIMM16 :
 494       errmsg = parse_simm16 (cd, strp, OR1K_OPERAND_SIMM16, (long *) (& fields->f_simm16));
 495       break;
 496     case OR1K_OPERAND_SIMM16_SPLIT :
 497       errmsg = parse_simm16_split (cd, strp, OR1K_OPERAND_SIMM16_SPLIT, (long *) (& fields->f_simm16_split));
 498       break;
 499     case OR1K_OPERAND_UIMM16 :
 500       errmsg = parse_uimm16 (cd, strp, OR1K_OPERAND_UIMM16, (unsigned long *) (& fields->f_uimm16));
 501       break;
 502     case OR1K_OPERAND_UIMM16_SPLIT :
 503       errmsg = parse_uimm16_split (cd, strp, OR1K_OPERAND_UIMM16_SPLIT, (unsigned long *) (& fields->f_uimm16_split));
 504       break;
 505     case OR1K_OPERAND_UIMM6 :
 506       errmsg = cgen_parse_unsigned_integer (cd, strp, OR1K_OPERAND_UIMM6, (unsigned long *) (& fields->f_uimm6));
 507       break;
 508
 509     default :
 510       /* xgettext:c-format */
 511       opcodes_error_handler
 512         (_("internal error: unrecognized field %d while parsing"),
 513          opindex);
 514       abort ();
 515   }
 516
 517   return errmsg;
 518 }
 519
 520 cgen_parse_fn * const or1k_cgen_parse_handlers[] =
 521 {
 522   parse_insn_normal,
 523 };
 524
 525 void
 526 or1k_cgen_init_asm (CGEN_CPU_DESC cd)
 527 {
 528   or1k_cgen_init_opcode_table (cd);
 529   or1k_cgen_init_ibld_table (cd);
 530   cd->parse_handlers = & or1k_cgen_parse_handlers[0];
 531   cd->parse_operand = or1k_cgen_parse_operand;
 532 #ifdef CGEN_ASM_INIT_HOOK
 533 CGEN_ASM_INIT_HOOK
 534 #endif
 535 }
 536
 537 \f
 538
 539 /* Regex construction routine.
 540
 541    This translates an opcode syntax string into a regex string,
 542    by replacing any non-character syntax element (such as an
 543    opcode) with the pattern '.*'
 544
 545    It then compiles the regex and stores it in the opcode, for
 546    later use by or1k_cgen_assemble_insn
 547
 548    Returns NULL for success, an error message for failure.  */
 549
 550 char *
 551 or1k_cgen_build_insn_regex (CGEN_INSN *insn)
 552 {
 553   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
 554   const char *mnem = CGEN_INSN_MNEMONIC (insn);
 555   char rxbuf[CGEN_MAX_RX_ELEMENTS];
 556   char *rx = rxbuf;
 557   const CGEN_SYNTAX_CHAR_TYPE *syn;
 558   int reg_err;
 559
 560   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
 561
 562   /* Mnemonics come first in the syntax string.  */
 563   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 564     return _("missing mnemonic in syntax string");
 565   ++syn;
 566
 567   /* Generate a case sensitive regular expression that emulates case
 568      insensitive matching in the "C" locale.  We cannot generate a case
 569      insensitive regular expression because in Turkish locales, 'i' and 'I'
 570      are not equal modulo case conversion.  */
 571
 572   /* Copy the literal mnemonic out of the insn.  */
 573   for (; *mnem; mnem++)
 574     {
 575       char c = *mnem;
 576
 577       if (ISALPHA (c))
 578         {
 579           *rx++ = '[';
 580           *rx++ = TOLOWER (c);
 581           *rx++ = TOUPPER (c);
 582           *rx++ = ']';
 583         }
 584       else
 585         *rx++ = c;
 586     }
 587
 588   /* Copy any remaining literals from the syntax string into the rx.  */
 589   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
 590     {
 591       if (CGEN_SYNTAX_CHAR_P (* syn))
 592         {
 593           char c = CGEN_SYNTAX_CHAR (* syn);
 594
 595           switch (c)
 596             {
 597               /* Escape any regex metacharacters in the syntax.  */
 598             case '.': case '[': case '\\':
 599             case '*': case '^': case '$':
 600
 601 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
 602             case '?': case '{': case '}':
 603             case '(': case ')': case '*':
 604             case '|': case '+': case ']':
 605 #endif
 606               *rx++ = '\\';
 607               *rx++ = c;
 608               break;
 609
 610             default:
 611               if (ISALPHA (c))
 612                 {
 613                   *rx++ = '[';
 614                   *rx++ = TOLOWER (c);
 615                   *rx++ = TOUPPER (c);
 616                   *rx++ = ']';
 617                 }
 618               else
 619                 *rx++ = c;
 620               break;
 621             }
 622         }
 623       else
 624         {
 625           /* Replace non-syntax fields with globs.  */
 626           *rx++ = '.';
 627           *rx++ = '*';
 628         }
 629     }
 630
 631   /* Trailing whitespace ok.  */
 632   * rx++ = '[';
 633   * rx++ = ' ';
 634   * rx++ = '\t';
 635   * rx++ = ']';
 636   * rx++ = '*';
 637
 638   /* But anchor it after that.  */
 639   * rx++ = '$';
 640   * rx = '\0';
 641
 642   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
 643   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
 644
 645   if (reg_err == 0)
 646     return NULL;
 647   else
 648     {
 649       static char msg[80];
 650
 651       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
 652       regfree ((regex_t *) CGEN_INSN_RX (insn));
 653       free (CGEN_INSN_RX (insn));
 654       (CGEN_INSN_RX (insn)) = NULL;
 655       return msg;
 656     }
 657 }
 658
 659 \f
 660 /* Default insn parser.
 661
 662    The syntax string is scanned and operands are parsed and stored in FIELDS.
 663    Relocs are queued as we go via other callbacks.
 664
 665    ??? Note that this is currently an all-or-nothing parser.  If we fail to
 666    parse the instruction, we return 0 and the caller will start over from
 667    the beginning.  Backtracking will be necessary in parsing subexpressions,
 668    but that can be handled there.  Not handling backtracking here may get
 669    expensive in the case of the m68k.  Deal with later.
 670
 671    Returns NULL for success, an error message for failure.  */
 672
 673 static const char *
 674 parse_insn_normal (CGEN_CPU_DESC cd,
 675                    const CGEN_INSN *insn,
 676                    const char **strp,
 677                    CGEN_FIELDS *fields)
 678 {
 679   /* ??? Runtime added insns not handled yet.  */
 680   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
 681   const char *str = *strp;
 682   const char *errmsg;
 683   const char *p;
 684   const CGEN_SYNTAX_CHAR_TYPE * syn;
 685 #ifdef CGEN_MNEMONIC_OPERANDS
 686   /* FIXME: wip */
 687   int past_opcode_p;
 688 #endif
 689
 690   /* For now we assume the mnemonic is first (there are no leading operands).
 691      We can parse it without needing to set up operand parsing.
 692      GAS's input scrubber will ensure mnemonics are lowercase, but we may
 693      not be called from GAS.  */
 694   p = CGEN_INSN_MNEMONIC (insn);
 695   while (*p && TOLOWER (*p) == TOLOWER (*str))
 696     ++p, ++str;
 697
 698   if (* p)
 699     return _("unrecognized instruction");
 700
 701 #ifndef CGEN_MNEMONIC_OPERANDS
 702   if (* str && ! ISSPACE (* str))
 703     return _("unrecognized instruction");
 704 #endif
 705
 706   CGEN_INIT_PARSE (cd);
 707   cgen_init_parse_operand (cd);
 708 #ifdef CGEN_MNEMONIC_OPERANDS
 709   past_opcode_p = 0;
 710 #endif
 711
 712   /* We don't check for (*str != '\0') here because we want to parse
 713      any trailing fake arguments in the syntax string.  */
 714   syn = CGEN_SYNTAX_STRING (syntax);
 715
 716   /* Mnemonics come first for now, ensure valid string.  */
 717   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 718     abort ();
 719
 720   ++syn;
 721
 722   while (* syn != 0)
 723     {
 724       /* Non operand chars must match exactly.  */
 725       if (CGEN_SYNTAX_CHAR_P (* syn))
 726         {
 727           /* FIXME: While we allow for non-GAS callers above, we assume the
 728              first char after the mnemonic part is a space.  */
 729           /* FIXME: We also take inappropriate advantage of the fact that
 730              GAS's input scrubber will remove extraneous blanks.  */
 731           if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
 732             {
 733 #ifdef CGEN_MNEMONIC_OPERANDS
 734               if (CGEN_SYNTAX_CHAR(* syn) == ' ')
 735                 past_opcode_p = 1;
 736 #endif
 737               ++ syn;
 738               ++ str;
 739             }
 740           else if (*str)
 741             {
 742               /* Syntax char didn't match.  Can't be this insn.  */
 743               static char msg [80];
 744
 745               /* xgettext:c-format */
 746               sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
 747                        CGEN_SYNTAX_CHAR(*syn), *str);
 748               return msg;
 749             }
 750           else
 751             {
 752               /* Ran out of input.  */
 753               static char msg [80];
 754
 755               /* xgettext:c-format */
 756               sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
 757                        CGEN_SYNTAX_CHAR(*syn));
 758               return msg;
 759             }
 760           continue;
 761         }
 762
 763 #ifdef CGEN_MNEMONIC_OPERANDS
 764       (void) past_opcode_p;
 765 #endif
 766       /* We have an operand of some sort.  */
 767       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn), &str, fields);
 768       if (errmsg)
 769         return errmsg;
 770
 771       /* Done with this operand, continue with next one.  */
 772       ++ syn;
 773     }
 774
 775   /* If we're at the end of the syntax string, we're done.  */
 776   if (* syn == 0)
 777     {
 778       /* FIXME: For the moment we assume a valid `str' can only contain
 779          blanks now.  IE: We needn't try again with a longer version of
 780          the insn and it is assumed that longer versions of insns appear
 781          before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
 782       while (ISSPACE (* str))
 783         ++ str;
 784
 785       if (* str != '\0')
 786         return _("junk at end of line"); /* FIXME: would like to include `str' */
 787
 788       return NULL;
 789     }
 790
 791   /* We couldn't parse it.  */
 792   return _("unrecognized instruction");
 793 }
 794 \f
 795 /* Main entry point.
 796    This routine is called for each instruction to be assembled.
 797    STR points to the insn to be assembled.
 798    We assume all necessary tables have been initialized.
 799    The assembled instruction, less any fixups, is stored in BUF.
 800    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
 801    still needs to be converted to target byte order, otherwise BUF is an array
 802    of bytes in target byte order.
 803    The result is a pointer to the insn's entry in the opcode table,
 804    or NULL if an error occured (an error message will have already been
 805    printed).
 806
 807    Note that when processing (non-alias) macro-insns,
 808    this function recurses.
 809
 810    ??? It's possible to make this cpu-independent.
 811    One would have to deal with a few minor things.
 812    At this point in time doing so would be more of a curiosity than useful
 813    [for example this file isn't _that_ big], but keeping the possibility in
 814    mind helps keep the design clean.  */
 815
 816 const CGEN_INSN *
 817 or1k_cgen_assemble_insn (CGEN_CPU_DESC cd,
 818                            const char *str,
 819                            CGEN_FIELDS *fields,
 820                            CGEN_INSN_BYTES_PTR buf,
 821                            char **errmsg)
 822 {
 823   const char *start;
 824   CGEN_INSN_LIST *ilist;
 825   const char *parse_errmsg = NULL;
 826   const char *insert_errmsg = NULL;
 827   int recognized_mnemonic = 0;
 828
 829   /* Skip leading white space.  */
 830   while (ISSPACE (* str))
 831     ++ str;
 832
 833   /* The instructions are stored in hashed lists.
 834      Get the first in the list.  */
 835   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
 836
 837   /* Keep looking until we find a match.  */
 838   start = str;
 839   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
 840     {
 841       const CGEN_INSN *insn = ilist->insn;
 842       recognized_mnemonic = 1;
 843
 844 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
 845       /* Not usually needed as unsupported opcodes
 846          shouldn't be in the hash lists.  */
 847       /* Is this insn supported by the selected cpu?  */
 848       if (! or1k_cgen_insn_supported (cd, insn))
 849         continue;
 850 #endif
 851       /* If the RELAXED attribute is set, this is an insn that shouldn't be
 852          chosen immediately.  Instead, it is used during assembler/linker
 853          relaxation if possible.  */
 854       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
 855         continue;
 856
 857       str = start;
 858
 859       /* Skip this insn if str doesn't look right lexically.  */
 860       if (CGEN_INSN_RX (insn) != NULL &&
 861           regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
 862         continue;
 863
 864       /* Allow parse/insert handlers to obtain length of insn.  */
 865       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
 866
 867       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
 868       if (parse_errmsg != NULL)
 869         continue;
 870
 871       /* ??? 0 is passed for `pc'.  */
 872       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
 873                                                  (bfd_vma) 0);
 874       if (insert_errmsg != NULL)
 875         continue;
 876
 877       /* It is up to the caller to actually output the insn and any
 878          queued relocs.  */
 879       return insn;
 880     }
 881
 882   {
 883     static char errbuf[150];
 884     const char *tmp_errmsg;
 885 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
 886 #define be_verbose 1
 887 #else
 888 #define be_verbose 0
 889 #endif
 890
 891     if (be_verbose)
 892       {
 893         /* If requesting verbose error messages, use insert_errmsg.
 894            Failing that, use parse_errmsg.  */
 895         tmp_errmsg = (insert_errmsg ? insert_errmsg :
 896                       parse_errmsg ? parse_errmsg :
 897                       recognized_mnemonic ?
 898                       _("unrecognized form of instruction") :
 899                       _("unrecognized instruction"));
 900
 901         if (strlen (start) > 50)
 902           /* xgettext:c-format */
 903           sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
 904         else
 905           /* xgettext:c-format */
 906           sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
 907       }
 908     else
 909       {
 910         if (strlen (start) > 50)
 911           /* xgettext:c-format */
 912           sprintf (errbuf, _("bad instruction `%.50s...'"), start);
 913         else
 914           /* xgettext:c-format */
 915           sprintf (errbuf, _("bad instruction `%.50s'"), start);
 916       }
 917
 918     *errmsg = errbuf;
 919     return NULL;
 920   }
 921 }