opcodes/ip2k-asm.c

   1 /* Assembler interface for targets using CGEN. -*- C -*-
   2    CGEN: Cpu tools GENerator
   3
   4 THIS FILE IS MACHINE GENERATED WITH CGEN.
   5 - the resultant file is machine generated, cgen-asm.in isn't
   6
   7 Copyright 1996, 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
   8
   9 This file is part of the GNU Binutils and GDB, the GNU debugger.
  10
  11 This program is free software; you can redistribute it and/or modify
  12 it under the terms of the GNU General Public License as published by
  13 the Free Software Foundation; either version 2, or (at your option)
  14 any later version.
  15
  16 This program is distributed in the hope that it will be useful,
  17 but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 GNU General Public License for more details.
  20
  21 You should have received a copy of the GNU General Public License
  22 along with this program; if not, write to the Free Software Foundation, Inc.,
  23 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
  24
  25 /* ??? Eventually more and more of this stuff can go to cpu-independent files.
  26    Keep that in mind.  */
  27
  28 #include "sysdep.h"
  29 #include <stdio.h>
  30 #include "ansidecl.h"
  31 #include "bfd.h"
  32 #include "symcat.h"
  33 #include "ip2k-desc.h"
  34 #include "ip2k-opc.h"
  35 #include "opintl.h"
  36 #include "xregex.h"
  37 #include "libiberty.h"
  38 #include "safe-ctype.h"
  39
  40 #undef  min
  41 #define min(a,b) ((a) < (b) ? (a) : (b))
  42 #undef  max
  43 #define max(a,b) ((a) > (b) ? (a) : (b))
  44
  45 static const char * parse_insn_normal
  46   (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *);
  47 \f
  48 /* -- assembler routines inserted here.  */
  49
  50 /* -- asm.c */
  51
  52 #define PARSE_FUNC_DECL(name) \
  53 static const char *name PARAMS ((CGEN_CPU_DESC, const char **, int, long *))
  54
  55 PARSE_FUNC_DECL (parse_fr);
  56 PARSE_FUNC_DECL (parse_addr16);
  57 PARSE_FUNC_DECL (parse_addr16_p);
  58 PARSE_FUNC_DECL (parse_addr16_cjp);
  59 PARSE_FUNC_DECL (parse_lit8);
  60 PARSE_FUNC_DECL (parse_bit3);
  61
  62
  63 static const char *
  64 parse_fr (cd, strp, opindex, valuep)
  65      CGEN_CPU_DESC cd;
  66      const char **strp;
  67      int opindex;
  68      long *valuep;
  69 {
  70   const char *errmsg;
  71   const char *old_strp;
  72   char *afteroffset;
  73   enum cgen_parse_operand_result result_type;
  74   bfd_vma value;
  75   extern CGEN_KEYWORD ip2k_cgen_opval_register_names;
  76   bfd_vma tempvalue;
  77
  78   old_strp = *strp;
  79   afteroffset = NULL;
  80
  81
  82   /* Check here to see if you're about to try parsing a w as the first arg */
  83   /* and return an error if you are.                                       */
  84   if ( (strncmp(*strp,"w",1)==0) || (strncmp(*strp,"W",1)==0) )
  85     {
  86       (*strp)++;
  87
  88       if ( (strncmp(*strp,",",1)==0) || ISSPACE(**strp) )
  89         {
  90           /* We've been passed a w.  Return with an error message so that  */
  91           /* cgen will try the next parsing option.                        */
  92           errmsg = _("W keyword invalid in FR operand slot.");
  93           return errmsg;
  94         }
  95       *strp = old_strp;
  96     }
  97
  98
  99   /* Attempt parse as register keyword. */
 100   /* old_strp = *strp; */
 101
 102   errmsg = cgen_parse_keyword (cd, strp, & ip2k_cgen_opval_register_names,
 103                                valuep);
 104   if ( *strp != NULL )
 105     if (errmsg == NULL)
 106       return errmsg;
 107
 108   /* Attempt to parse for "(IP)" */
 109   afteroffset = strstr(*strp,"(IP)");
 110
 111   if ( afteroffset == NULL)
 112     {
 113       /* Make sure it's not in lower case */
 114       afteroffset = strstr(*strp,"(ip)");
 115     }
 116
 117   if ( afteroffset != NULL )
 118     {
 119       if ( afteroffset != *strp )
 120         {
 121           /* Invalid offset present.*/
 122           errmsg = _("offset(IP) is not a valid form");
 123           return errmsg;
 124         }
 125       else
 126         {
 127           *strp += 4;
 128           *valuep = 0;
 129           errmsg = NULL;
 130           return errmsg;
 131         }
 132     }
 133
 134   /* Attempt to parse for DP. ex: mov w, offset(DP)  */
 135   /*                              mov offset(DP),w   */
 136
 137   /* Try parsing it as an address and see what comes back */
 138
 139   afteroffset = strstr(*strp,"(DP)");
 140
 141   if ( afteroffset == NULL)
 142     {
 143       /* Maybe it's in lower case */
 144       afteroffset = strstr(*strp,"(dp)");
 145     }
 146
 147   if ( afteroffset != NULL )
 148     {
 149       if ( afteroffset == *strp )
 150         {
 151           /* No offset present. Use 0 by default. */
 152           tempvalue = 0;
 153           errmsg = NULL;
 154         }
 155       else
 156         {
 157           errmsg = cgen_parse_address (cd, strp, opindex,
 158                                        BFD_RELOC_IP2K_FR_OFFSET,
 159                                        & result_type, & tempvalue);
 160         }
 161
 162       if (errmsg == NULL)
 163         {
 164           if (tempvalue <= 127)
 165             {
 166               /* Value is ok.  Fix up the first 2 bits and return */
 167               *valuep = 0x0100 | tempvalue;
 168               *strp += 4; /* skip over the (DP) in *strp */
 169               return errmsg;
 170             }
 171           else
 172             {
 173               /* Found something there in front of (DP) but it's out
 174                  of range. */
 175               errmsg = _("(DP) offset out of range.");
 176               return errmsg;
 177             }
 178         }
 179     }
 180
 181
 182   /* Attempt to parse for SP. ex: mov w, offset(SP)  */
 183   /*                              mov offset(SP), w  */
 184
 185
 186   afteroffset = strstr(*strp,"(SP)");
 187
 188   if (afteroffset == NULL)
 189     {
 190       /* Maybe it's in lower case. */
 191       afteroffset = strstr(*strp, "(sp)");
 192     }
 193
 194   if ( afteroffset != NULL )
 195     {
 196       if ( afteroffset ==  *strp )
 197         {
 198           /* No offset present. Use 0 by default. */
 199           tempvalue = 0;
 200           errmsg = NULL;
 201         }
 202       else
 203         {
 204           errmsg = cgen_parse_address (cd, strp, opindex,
 205                                        BFD_RELOC_IP2K_FR_OFFSET,
 206                                        & result_type, & tempvalue);
 207         }
 208       if (errmsg == NULL)
 209         {
 210           if (tempvalue <= 127)
 211             {
 212               /* Value is ok.  Fix up the first 2 bits and return */
 213               *valuep = 0x0180 | tempvalue;
 214               *strp += 4; /* skip over the (SP) in *strp */
 215               return errmsg;
 216             }
 217           else
 218             {
 219               /* Found something there in front of (SP) but it's out
 220                  of range. */
 221               errmsg = _("(SP) offset out of range.");
 222               return errmsg;
 223             }
 224
 225         }
 226     }
 227
 228
 229   /* Attempt to parse as an address. */
 230   *strp = old_strp;
 231   errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_IP2K_FR9,
 232                                & result_type, & value);
 233   if (errmsg == NULL)
 234     {
 235       *valuep = value;
 236
 237       /* if a parenthesis is found, warn about invalid form */
 238
 239       if (**strp == '(')
 240         {
 241           errmsg = _("illegal use of parentheses");
 242         }
 243       /* if a numeric value is specified, ensure that it is between
 244          1 and 255 */
 245       else if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 246         {
 247           if (value < 0x1 || value > 0xff)
 248             errmsg = _("operand out of range (not between 1 and 255)");
 249         }
 250     }
 251   return errmsg;
 252 }
 253
 254 static const char *
 255 parse_addr16 (cd, strp, opindex, valuep)
 256      CGEN_CPU_DESC cd;
 257      const char **strp;
 258      int opindex;
 259      long *valuep;
 260 {
 261   const char *errmsg;
 262   enum cgen_parse_operand_result result_type;
 263   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 264   bfd_vma value;
 265
 266   if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16H )
 267     code = BFD_RELOC_IP2K_HI8DATA;
 268   else if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16L )
 269     code = BFD_RELOC_IP2K_LO8DATA;
 270   else
 271     {
 272       /* Something is very wrong. opindex has to be one of the above. */
 273       errmsg = _("parse_addr16: invalid opindex.");
 274       return errmsg;
 275     }
 276
 277   errmsg = cgen_parse_address (cd, strp, opindex, code,
 278                                & result_type, & value);
 279   if (errmsg == NULL)
 280     {
 281       /* We either have a relocation or a number now. */
 282       if ( result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER )
 283         {
 284           /* We got a number back. */
 285           if ( code == BFD_RELOC_IP2K_HI8DATA )
 286             value >>= 8;
 287           else    /* code = BFD_RELOC_IP2K_LOW8DATA */
 288             value &= 0x00FF;
 289         }
 290       *valuep = value;
 291     }
 292
 293   return errmsg;
 294 }
 295
 296
 297 static const char *
 298 parse_addr16_p (cd, strp, opindex, valuep)
 299      CGEN_CPU_DESC cd;
 300      const char **strp;
 301      int opindex;
 302      long *valuep;
 303 {
 304   const char *errmsg;
 305   enum cgen_parse_operand_result result_type;
 306   bfd_reloc_code_real_type code = BFD_RELOC_IP2K_PAGE3;
 307   bfd_vma value;
 308
 309   errmsg = cgen_parse_address (cd, strp, opindex, code,
 310                                & result_type, & value);
 311   if (errmsg == NULL)
 312     {
 313       if ( result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER )
 314         *valuep = (value >> 13) & 0x7;
 315       else if ( result_type == CGEN_PARSE_OPERAND_RESULT_QUEUED )
 316         *valuep = value;
 317     }
 318   return errmsg;
 319 }
 320
 321
 322 static const char *
 323 parse_addr16_cjp (cd, strp, opindex, valuep)
 324      CGEN_CPU_DESC cd;
 325      const char **strp;
 326      int opindex;
 327      long *valuep;
 328 {
 329   const char *errmsg;
 330   enum cgen_parse_operand_result result_type;
 331   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 332   bfd_vma value;
 333
 334   if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16CJP )
 335     code = BFD_RELOC_IP2K_ADDR16CJP;
 336   else if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16P )
 337     code = BFD_RELOC_IP2K_PAGE3;
 338
 339   errmsg = cgen_parse_address (cd, strp, opindex, code,
 340                                & result_type, & value);
 341   if (errmsg == NULL)
 342     {
 343       if ( result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER )
 344         {
 345           if ( (value & 0x1) == 0)  /* If the address is even .... */
 346             {
 347               if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16CJP )
 348                 *valuep = (value >> 1) & 0x1FFF;  /* Should mask be 1FFF? */
 349               else if ( opindex == (CGEN_OPERAND_TYPE)IP2K_OPERAND_ADDR16P )
 350                 *valuep = (value >> 14) & 0x7;
 351             }
 352           else
 353             errmsg = _("Byte address required. - must be even.");
 354         }
 355       else if ( result_type == CGEN_PARSE_OPERAND_RESULT_QUEUED )
 356         {
 357           /* This will happen for things like (s2-s1) where s2 and s1
 358              are labels.  */
 359           *valuep = value;
 360         }
 361       else
 362         errmsg = _("cgen_parse_address returned a symbol. Literal required.");
 363     }
 364   return errmsg;
 365 }
 366
 367
 368 static const char *
 369 parse_lit8 (cd, strp, opindex, valuep)
 370      CGEN_CPU_DESC cd;
 371      const char **strp;
 372      int opindex;
 373      long *valuep;
 374 {
 375   const char *errmsg;
 376   enum cgen_parse_operand_result result_type;
 377   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 378   bfd_vma value;
 379
 380   /* Parse %OP relocating operators. */
 381   if (strncmp (*strp, "%bank", 5) == 0)
 382     {
 383       *strp += 5;
 384       code = BFD_RELOC_IP2K_BANK;
 385     }
 386   else if (strncmp (*strp, "%lo8data", 8) == 0)
 387     {
 388       *strp += 8;
 389       code = BFD_RELOC_IP2K_LO8DATA;
 390     }
 391   else if (strncmp (*strp, "%hi8data", 8) == 0)
 392     {
 393       *strp += 8;
 394       code = BFD_RELOC_IP2K_HI8DATA;
 395     }
 396   else if (strncmp (*strp, "%ex8data", 8) == 0)
 397     {
 398       *strp += 8;
 399       code = BFD_RELOC_IP2K_EX8DATA;
 400     }
 401   else if (strncmp (*strp, "%lo8insn", 8) == 0)
 402     {
 403       *strp += 8;
 404       code = BFD_RELOC_IP2K_LO8INSN;
 405     }
 406   else if (strncmp (*strp, "%hi8insn", 8) == 0)
 407     {
 408       *strp += 8;
 409       code = BFD_RELOC_IP2K_HI8INSN;
 410     }
 411
 412
 413   /* Parse %op operand.  */
 414   if (code != BFD_RELOC_NONE)
 415     {
 416       errmsg = cgen_parse_address (cd, strp, opindex, code,
 417                                    & result_type, & value);
 418       if ((errmsg == NULL) &&
 419           (result_type != CGEN_PARSE_OPERAND_RESULT_QUEUED))
 420         errmsg = _("%operator operand is not a symbol");
 421
 422       *valuep = value;
 423     }
 424   /* Parse as a number.  */
 425   else
 426     {
 427       errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep);
 428
 429       /* Truncate to eight bits to accept both signed and unsigned input. */
 430       if (errmsg == NULL)
 431         *valuep &= 0xFF;
 432     }
 433
 434   return errmsg;
 435 }
 436
 437 static const char *
 438 parse_bit3 (cd, strp, opindex, valuep)
 439      CGEN_CPU_DESC cd;
 440      const char **strp;
 441      int opindex;
 442      long *valuep;
 443 {
 444   const char *errmsg;
 445   char mode = 0;
 446   long count = 0;
 447   unsigned long value;
 448
 449   if (strncmp (*strp, "%bit", 4) == 0)
 450     {
 451       *strp += 4;
 452       mode = 1;
 453     }
 454   else if (strncmp (*strp, "%msbbit", 7) == 0)
 455     {
 456       *strp += 7;
 457       mode = 1;
 458     }
 459   else if (strncmp (*strp, "%lsbbit", 7) == 0)
 460     {
 461       *strp += 7;
 462       mode = 2;
 463     }
 464
 465   errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep);
 466   if (errmsg)
 467     return errmsg;
 468
 469   if (mode)
 470     {
 471       value = (unsigned long) *valuep;
 472       if (value == 0)
 473         {
 474           errmsg = _("Attempt to find bit index of 0");
 475           return errmsg;
 476         }
 477
 478       if (mode == 1)
 479         {
 480           count = 31;
 481           while ((value & 0x80000000) == 0)
 482             {
 483               count--;
 484               value <<= 1;
 485             }
 486         }
 487       else if (mode == 2)
 488         {
 489           count = 0;
 490           while ((value & 0x00000001) == 0)
 491             {
 492               count++;
 493               value >>= 1;
 494             }
 495         }
 496
 497       *valuep = count;
 498     }
 499
 500   return errmsg;
 501 }
 502
 503
 504 /* -- dis.c */
 505
 506 const char * ip2k_cgen_parse_operand
 507   PARAMS ((CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *));
 508
 509 /* Main entry point for operand parsing.
 510
 511    This function is basically just a big switch statement.  Earlier versions
 512    used tables to look up the function to use, but
 513    - if the table contains both assembler and disassembler functions then
 514      the disassembler contains much of the assembler and vice-versa,
 515    - there's a lot of inlining possibilities as things grow,
 516    - using a switch statement avoids the function call overhead.
 517
 518    This function could be moved into `parse_insn_normal', but keeping it
 519    separate makes clear the interface between `parse_insn_normal' and each of
 520    the handlers.  */
 521
 522 const char *
 523 ip2k_cgen_parse_operand (cd, opindex, strp, fields)
 524      CGEN_CPU_DESC cd;
 525      int opindex;
 526      const char ** strp;
 527      CGEN_FIELDS * fields;
 528 {
 529   const char * errmsg = NULL;
 530   /* Used by scalar operands that still need to be parsed.  */
 531   long junk ATTRIBUTE_UNUSED;
 532
 533   switch (opindex)
 534     {
 535     case IP2K_OPERAND_ADDR16CJP :
 536       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16CJP, &fields->f_addr16cjp);
 537       break;
 538     case IP2K_OPERAND_ADDR16H :
 539       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16H, &fields->f_imm8);
 540       break;
 541     case IP2K_OPERAND_ADDR16L :
 542       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16L, &fields->f_imm8);
 543       break;
 544     case IP2K_OPERAND_ADDR16P :
 545       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16P, &fields->f_page3);
 546       break;
 547     case IP2K_OPERAND_BITNO :
 548       errmsg = parse_bit3 (cd, strp, IP2K_OPERAND_BITNO, &fields->f_bitno);
 549       break;
 550     case IP2K_OPERAND_CBIT :
 551       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_CBIT, &junk);
 552       break;
 553     case IP2K_OPERAND_DCBIT :
 554       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_DCBIT, &junk);
 555       break;
 556     case IP2K_OPERAND_FR :
 557       errmsg = parse_fr (cd, strp, IP2K_OPERAND_FR, &fields->f_reg);
 558       break;
 559     case IP2K_OPERAND_LIT8 :
 560       errmsg = parse_lit8 (cd, strp, IP2K_OPERAND_LIT8, &fields->f_imm8);
 561       break;
 562     case IP2K_OPERAND_PABITS :
 563       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_PABITS, &junk);
 564       break;
 565     case IP2K_OPERAND_RETI3 :
 566       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_RETI3, &fields->f_reti3);
 567       break;
 568     case IP2K_OPERAND_ZBIT :
 569       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_ZBIT, &junk);
 570       break;
 571
 572     default :
 573       /* xgettext:c-format */
 574       fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex);
 575       abort ();
 576   }
 577
 578   return errmsg;
 579 }
 580
 581 cgen_parse_fn * const ip2k_cgen_parse_handlers[] =
 582 {
 583   parse_insn_normal,
 584 };
 585
 586 void
 587 ip2k_cgen_init_asm (cd)
 588      CGEN_CPU_DESC cd;
 589 {
 590   ip2k_cgen_init_opcode_table (cd);
 591   ip2k_cgen_init_ibld_table (cd);
 592   cd->parse_handlers = & ip2k_cgen_parse_handlers[0];
 593   cd->parse_operand = ip2k_cgen_parse_operand;
 594 }
 595
 596 \f
 597
 598 /* Regex construction routine.
 599
 600    This translates an opcode syntax string into a regex string,
 601    by replacing any non-character syntax element (such as an
 602    opcode) with the pattern '.*'
 603
 604    It then compiles the regex and stores it in the opcode, for
 605    later use by ip2k_cgen_assemble_insn
 606
 607    Returns NULL for success, an error message for failure.  */
 608
 609 char *
 610 ip2k_cgen_build_insn_regex (CGEN_INSN *insn)
 611 {
 612   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
 613   const char *mnem = CGEN_INSN_MNEMONIC (insn);
 614   char rxbuf[CGEN_MAX_RX_ELEMENTS];
 615   char *rx = rxbuf;
 616   const CGEN_SYNTAX_CHAR_TYPE *syn;
 617   int reg_err;
 618
 619   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
 620
 621   /* Mnemonics come first in the syntax string.  */
 622   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 623     return _("missing mnemonic in syntax string");
 624   ++syn;
 625
 626   /* Generate a case sensitive regular expression that emulates case
 627      insensitive matching in the "C" locale.  We cannot generate a case
 628      insensitive regular expression because in Turkish locales, 'i' and 'I'
 629      are not equal modulo case conversion.  */
 630
 631   /* Copy the literal mnemonic out of the insn.  */
 632   for (; *mnem; mnem++)
 633     {
 634       char c = *mnem;
 635
 636       if (ISALPHA (c))
 637         {
 638           *rx++ = '[';
 639           *rx++ = TOLOWER (c);
 640           *rx++ = TOUPPER (c);
 641           *rx++ = ']';
 642         }
 643       else
 644         *rx++ = c;
 645     }
 646
 647   /* Copy any remaining literals from the syntax string into the rx.  */
 648   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
 649     {
 650       if (CGEN_SYNTAX_CHAR_P (* syn))
 651         {
 652           char c = CGEN_SYNTAX_CHAR (* syn);
 653
 654           switch (c)
 655             {
 656               /* Escape any regex metacharacters in the syntax.  */
 657             case '.': case '[': case '\\':
 658             case '*': case '^': case '$':
 659
 660 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
 661             case '?': case '{': case '}':
 662             case '(': case ')': case '*':
 663             case '|': case '+': case ']':
 664 #endif
 665               *rx++ = '\\';
 666               *rx++ = c;
 667               break;
 668
 669             default:
 670               if (ISALPHA (c))
 671                 {
 672                   *rx++ = '[';
 673                   *rx++ = TOLOWER (c);
 674                   *rx++ = TOUPPER (c);
 675                   *rx++ = ']';
 676                 }
 677               else
 678                 *rx++ = c;
 679               break;
 680             }
 681         }
 682       else
 683         {
 684           /* Replace non-syntax fields with globs.  */
 685           *rx++ = '.';
 686           *rx++ = '*';
 687         }
 688     }
 689
 690   /* Trailing whitespace ok.  */
 691   * rx++ = '[';
 692   * rx++ = ' ';
 693   * rx++ = '\t';
 694   * rx++ = ']';
 695   * rx++ = '*';
 696
 697   /* But anchor it after that.  */
 698   * rx++ = '$';
 699   * rx = '\0';
 700
 701   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
 702   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
 703
 704   if (reg_err == 0)
 705     return NULL;
 706   else
 707     {
 708       static char msg[80];
 709
 710       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
 711       regfree ((regex_t *) CGEN_INSN_RX (insn));
 712       free (CGEN_INSN_RX (insn));
 713       (CGEN_INSN_RX (insn)) = NULL;
 714       return msg;
 715     }
 716 }
 717
 718 \f
 719 /* Default insn parser.
 720
 721    The syntax string is scanned and operands are parsed and stored in FIELDS.
 722    Relocs are queued as we go via other callbacks.
 723
 724    ??? Note that this is currently an all-or-nothing parser.  If we fail to
 725    parse the instruction, we return 0 and the caller will start over from
 726    the beginning.  Backtracking will be necessary in parsing subexpressions,
 727    but that can be handled there.  Not handling backtracking here may get
 728    expensive in the case of the m68k.  Deal with later.
 729
 730    Returns NULL for success, an error message for failure.  */
 731
 732 static const char *
 733 parse_insn_normal (CGEN_CPU_DESC cd,
 734                    const CGEN_INSN *insn,
 735                    const char **strp,
 736                    CGEN_FIELDS *fields)
 737 {
 738   /* ??? Runtime added insns not handled yet.  */
 739   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
 740   const char *str = *strp;
 741   const char *errmsg;
 742   const char *p;
 743   const CGEN_SYNTAX_CHAR_TYPE * syn;
 744 #ifdef CGEN_MNEMONIC_OPERANDS
 745   /* FIXME: wip */
 746   int past_opcode_p;
 747 #endif
 748
 749   /* For now we assume the mnemonic is first (there are no leading operands).
 750      We can parse it without needing to set up operand parsing.
 751      GAS's input scrubber will ensure mnemonics are lowercase, but we may
 752      not be called from GAS.  */
 753   p = CGEN_INSN_MNEMONIC (insn);
 754   while (*p && TOLOWER (*p) == TOLOWER (*str))
 755     ++p, ++str;
 756
 757   if (* p)
 758     return _("unrecognized instruction");
 759
 760 #ifndef CGEN_MNEMONIC_OPERANDS
 761   if (* str && ! ISSPACE (* str))
 762     return _("unrecognized instruction");
 763 #endif
 764
 765   CGEN_INIT_PARSE (cd);
 766   cgen_init_parse_operand (cd);
 767 #ifdef CGEN_MNEMONIC_OPERANDS
 768   past_opcode_p = 0;
 769 #endif
 770
 771   /* We don't check for (*str != '\0') here because we want to parse
 772      any trailing fake arguments in the syntax string.  */
 773   syn = CGEN_SYNTAX_STRING (syntax);
 774
 775   /* Mnemonics come first for now, ensure valid string.  */
 776   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 777     abort ();
 778
 779   ++syn;
 780
 781   while (* syn != 0)
 782     {
 783       /* Non operand chars must match exactly.  */
 784       if (CGEN_SYNTAX_CHAR_P (* syn))
 785         {
 786           /* FIXME: While we allow for non-GAS callers above, we assume the
 787              first char after the mnemonic part is a space.  */
 788           /* FIXME: We also take inappropriate advantage of the fact that
 789              GAS's input scrubber will remove extraneous blanks.  */
 790           if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
 791             {
 792 #ifdef CGEN_MNEMONIC_OPERANDS
 793               if (CGEN_SYNTAX_CHAR(* syn) == ' ')
 794                 past_opcode_p = 1;
 795 #endif
 796               ++ syn;
 797               ++ str;
 798             }
 799           else if (*str)
 800             {
 801               /* Syntax char didn't match.  Can't be this insn.  */
 802               static char msg [80];
 803
 804               /* xgettext:c-format */
 805               sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
 806                        CGEN_SYNTAX_CHAR(*syn), *str);
 807               return msg;
 808             }
 809           else
 810             {
 811               /* Ran out of input.  */
 812               static char msg [80];
 813
 814               /* xgettext:c-format */
 815               sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
 816                        CGEN_SYNTAX_CHAR(*syn));
 817               return msg;
 818             }
 819           continue;
 820         }
 821
 822       /* We have an operand of some sort.  */
 823       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn),
 824                                           &str, fields);
 825       if (errmsg)
 826         return errmsg;
 827
 828       /* Done with this operand, continue with next one.  */
 829       ++ syn;
 830     }
 831
 832   /* If we're at the end of the syntax string, we're done.  */
 833   if (* syn == 0)
 834     {
 835       /* FIXME: For the moment we assume a valid `str' can only contain
 836          blanks now.  IE: We needn't try again with a longer version of
 837          the insn and it is assumed that longer versions of insns appear
 838          before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
 839       while (ISSPACE (* str))
 840         ++ str;
 841
 842       if (* str != '\0')
 843         return _("junk at end of line"); /* FIXME: would like to include `str' */
 844
 845       return NULL;
 846     }
 847
 848   /* We couldn't parse it.  */
 849   return _("unrecognized instruction");
 850 }
 851 \f
 852 /* Main entry point.
 853    This routine is called for each instruction to be assembled.
 854    STR points to the insn to be assembled.
 855    We assume all necessary tables have been initialized.
 856    The assembled instruction, less any fixups, is stored in BUF.
 857    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
 858    still needs to be converted to target byte order, otherwise BUF is an array
 859    of bytes in target byte order.
 860    The result is a pointer to the insn's entry in the opcode table,
 861    or NULL if an error occured (an error message will have already been
 862    printed).
 863
 864    Note that when processing (non-alias) macro-insns,
 865    this function recurses.
 866
 867    ??? It's possible to make this cpu-independent.
 868    One would have to deal with a few minor things.
 869    At this point in time doing so would be more of a curiosity than useful
 870    [for example this file isn't _that_ big], but keeping the possibility in
 871    mind helps keep the design clean.  */
 872
 873 const CGEN_INSN *
 874 ip2k_cgen_assemble_insn (CGEN_CPU_DESC cd,
 875                            const char *str,
 876                            CGEN_FIELDS *fields,
 877                            CGEN_INSN_BYTES_PTR buf,
 878                            char **errmsg)
 879 {
 880   const char *start;
 881   CGEN_INSN_LIST *ilist;
 882   const char *parse_errmsg = NULL;
 883   const char *insert_errmsg = NULL;
 884   int recognized_mnemonic = 0;
 885
 886   /* Skip leading white space.  */
 887   while (ISSPACE (* str))
 888     ++ str;
 889
 890   /* The instructions are stored in hashed lists.
 891      Get the first in the list.  */
 892   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
 893
 894   /* Keep looking until we find a match.  */
 895   start = str;
 896   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
 897     {
 898       const CGEN_INSN *insn = ilist->insn;
 899       recognized_mnemonic = 1;
 900
 901 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
 902       /* Not usually needed as unsupported opcodes
 903          shouldn't be in the hash lists.  */
 904       /* Is this insn supported by the selected cpu?  */
 905       if (! ip2k_cgen_insn_supported (cd, insn))
 906         continue;
 907 #endif
 908       /* If the RELAXED attribute is set, this is an insn that shouldn't be
 909          chosen immediately.  Instead, it is used during assembler/linker
 910          relaxation if possible.  */
 911       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
 912         continue;
 913
 914       str = start;
 915
 916       /* Skip this insn if str doesn't look right lexically.  */
 917       if (CGEN_INSN_RX (insn) != NULL &&
 918           regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
 919         continue;
 920
 921       /* Allow parse/insert handlers to obtain length of insn.  */
 922       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
 923
 924       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
 925       if (parse_errmsg != NULL)
 926         continue;
 927
 928       /* ??? 0 is passed for `pc'.  */
 929       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
 930                                                  (bfd_vma) 0);
 931       if (insert_errmsg != NULL)
 932         continue;
 933
 934       /* It is up to the caller to actually output the insn and any
 935          queued relocs.  */
 936       return insn;
 937     }
 938
 939   {
 940     static char errbuf[150];
 941 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
 942     const char *tmp_errmsg;
 943
 944     /* If requesting verbose error messages, use insert_errmsg.
 945        Failing that, use parse_errmsg.  */
 946     tmp_errmsg = (insert_errmsg ? insert_errmsg :
 947                   parse_errmsg ? parse_errmsg :
 948                   recognized_mnemonic ?
 949                   _("unrecognized form of instruction") :
 950                   _("unrecognized instruction"));
 951
 952     if (strlen (start) > 50)
 953       /* xgettext:c-format */
 954       sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
 955     else
 956       /* xgettext:c-format */
 957       sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
 958 #else
 959     if (strlen (start) > 50)
 960       /* xgettext:c-format */
 961       sprintf (errbuf, _("bad instruction `%.50s...'"), start);
 962     else
 963       /* xgettext:c-format */
 964       sprintf (errbuf, _("bad instruction `%.50s'"), start);
 965 #endif
 966
 967     *errmsg = errbuf;
 968     return NULL;
 969   }
 970 }
 971 \f
 972 #if 0 /* This calls back to GAS which we can't do without care.  */
 973
 974 /* Record each member of OPVALS in the assembler's symbol table.
 975    This lets GAS parse registers for us.
 976    ??? Interesting idea but not currently used.  */
 977
 978 /* Record each member of OPVALS in the assembler's symbol table.
 979    FIXME: Not currently used.  */
 980
 981 void
 982 ip2k_cgen_asm_hash_keywords (CGEN_CPU_DESC cd, CGEN_KEYWORD *opvals)
 983 {
 984   CGEN_KEYWORD_SEARCH search = cgen_keyword_search_init (opvals, NULL);
 985   const CGEN_KEYWORD_ENTRY * ke;
 986
 987   while ((ke = cgen_keyword_search_next (& search)) != NULL)
 988     {
 989 #if 0 /* Unnecessary, should be done in the search routine.  */
 990       if (! ip2k_cgen_opval_supported (ke))
 991         continue;
 992 #endif
 993       cgen_asm_record_register (cd, ke->name, ke->value);
 994     }
 995 }
 996
 997 #endif /* 0 */