opcodes/ip2k-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) 1996-2024 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 "ip2k-desc.h"
  36 #include "ip2k-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 *
  55 parse_fr (CGEN_CPU_DESC cd,
  56           const char **strp,
  57           int opindex,
  58           unsigned long *valuep)
  59 {
  60   const char *errmsg;
  61   const char *old_strp;
  62   char *afteroffset;
  63   enum cgen_parse_operand_result result_type;
  64   bfd_vma value;
  65   extern CGEN_KEYWORD ip2k_cgen_opval_register_names;
  66   bfd_vma tempvalue;
  67
  68   old_strp = *strp;
  69   afteroffset = NULL;
  70
  71   /* Check here to see if you're about to try parsing a w as the first arg
  72      and return an error if you are.  */
  73   if ((strncmp (*strp, "w", 1) == 0) || (strncmp (*strp, "W", 1) == 0))
  74     {
  75       (*strp)++;
  76
  77       if ((strncmp (*strp, ",", 1) == 0) || ISSPACE (**strp))
  78         {
  79           /* We've been passed a w.  Return with an error message so that
  80              cgen will try the next parsing option.  */
  81           errmsg = _("W keyword invalid in FR operand slot.");
  82           return errmsg;
  83         }
  84       *strp = old_strp;
  85     }
  86
  87   /* Attempt parse as register keyword. */
  88   errmsg = cgen_parse_keyword (cd, strp, & ip2k_cgen_opval_register_names,
  89                                (long *) valuep);
  90   if (*strp != NULL
  91       && errmsg == NULL)
  92     return errmsg;
  93
  94   /* Attempt to parse for "(IP)".  */
  95   afteroffset = strstr (*strp, "(IP)");
  96
  97   if (afteroffset == NULL)
  98     /* Make sure it's not in lower case.  */
  99     afteroffset = strstr (*strp, "(ip)");
 100
 101   if (afteroffset != NULL)
 102     {
 103       if (afteroffset != *strp)
 104         {
 105           /* Invalid offset present.  */
 106           errmsg = _("offset(IP) is not a valid form");
 107           return errmsg;
 108         }
 109       else
 110         {
 111           *strp += 4;
 112           *valuep = 0;
 113           errmsg = NULL;
 114           return errmsg;
 115         }
 116     }
 117
 118   /* Attempt to parse for DP. ex: mov w, offset(DP)
 119                                   mov offset(DP),w   */
 120
 121   /* Try parsing it as an address and see what comes back.  */
 122   afteroffset = strstr (*strp, "(DP)");
 123
 124   if (afteroffset == NULL)
 125     /* Maybe it's in lower case.  */
 126     afteroffset = strstr (*strp, "(dp)");
 127
 128   if (afteroffset != NULL)
 129     {
 130       if (afteroffset == *strp)
 131         {
 132           /* No offset present. Use 0 by default.  */
 133           tempvalue = 0;
 134           errmsg = NULL;
 135         }
 136       else
 137         errmsg = cgen_parse_address (cd, strp, opindex,
 138                                      BFD_RELOC_IP2K_FR_OFFSET,
 139                                      & result_type, & tempvalue);
 140
 141       if (errmsg == NULL)
 142         {
 143           if (tempvalue <= 127)
 144             {
 145               /* Value is ok.  Fix up the first 2 bits and return.  */
 146               *valuep = 0x0100 | tempvalue;
 147               *strp += 4; /* Skip over the (DP) in *strp.  */
 148               return errmsg;
 149             }
 150           else
 151             {
 152               /* Found something there in front of (DP) but it's out
 153                  of range.  */
 154               errmsg = _("(DP) offset out of range.");
 155               return errmsg;
 156             }
 157         }
 158     }
 159
 160
 161   /* Attempt to parse for SP. ex: mov w, offset(SP)
 162                                   mov offset(SP), w.  */
 163   afteroffset = strstr (*strp, "(SP)");
 164
 165   if (afteroffset == NULL)
 166     /* Maybe it's in lower case.  */
 167     afteroffset = strstr (*strp, "(sp)");
 168
 169   if (afteroffset != NULL)
 170     {
 171       if (afteroffset == *strp)
 172         {
 173           /* No offset present. Use 0 by default.  */
 174           tempvalue = 0;
 175           errmsg = NULL;
 176         }
 177       else
 178         errmsg = cgen_parse_address (cd, strp, opindex,
 179                                      BFD_RELOC_IP2K_FR_OFFSET,
 180                                      & result_type, & tempvalue);
 181
 182       if (errmsg == NULL)
 183         {
 184           if (tempvalue <= 127)
 185             {
 186               /* Value is ok.  Fix up the first 2 bits and return.  */
 187               *valuep = 0x0180 | tempvalue;
 188               *strp += 4; /* Skip over the (SP) in *strp.  */
 189               return errmsg;
 190             }
 191           else
 192             {
 193               /* Found something there in front of (SP) but it's out
 194                  of range.  */
 195               errmsg = _("(SP) offset out of range.");
 196               return errmsg;
 197             }
 198         }
 199     }
 200
 201   /* Attempt to parse as an address.  */
 202   *strp = old_strp;
 203   errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_IP2K_FR9,
 204                                & result_type, & value);
 205   if (errmsg == NULL)
 206     {
 207       *valuep = value;
 208
 209       /* If a parenthesis is found, warn about invalid form.  */
 210       if (**strp == '(')
 211         errmsg = _("illegal use of parentheses");
 212
 213       /* If a numeric value is specified, ensure that it is between
 214          1 and 255.  */
 215       else if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 216         {
 217           if (value < 0x1 || value > 0xff)
 218             errmsg = _("operand out of range (not between 1 and 255)");
 219         }
 220     }
 221   return errmsg;
 222 }
 223
 224 static const char *
 225 parse_addr16 (CGEN_CPU_DESC cd,
 226               const char **strp,
 227               int opindex,
 228               unsigned long *valuep)
 229 {
 230   const char *errmsg;
 231   enum cgen_parse_operand_result result_type;
 232   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 233   bfd_vma value;
 234
 235   if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16H)
 236     code = BFD_RELOC_IP2K_HI8DATA;
 237   else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16L)
 238     code = BFD_RELOC_IP2K_LO8DATA;
 239   else
 240     {
 241       /* Something is very wrong. opindex has to be one of the above.  */
 242       errmsg = _("parse_addr16: invalid opindex.");
 243       return errmsg;
 244     }
 245
 246   errmsg = cgen_parse_address (cd, strp, opindex, code,
 247                                & result_type, & value);
 248   if (errmsg == NULL)
 249     {
 250       /* We either have a relocation or a number now.  */
 251       if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 252         {
 253           /* We got a number back.  */
 254           if (code == BFD_RELOC_IP2K_HI8DATA)
 255             value >>= 8;
 256           else
 257             /* code = BFD_RELOC_IP2K_LOW8DATA.  */
 258             value &= 0x00FF;
 259         }
 260       *valuep = value;
 261     }
 262
 263   return errmsg;
 264 }
 265
 266 static const char *
 267 parse_addr16_cjp (CGEN_CPU_DESC cd,
 268                   const char **strp,
 269                   int opindex,
 270                   unsigned long *valuep)
 271 {
 272   const char *errmsg;
 273   enum cgen_parse_operand_result result_type;
 274   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 275   bfd_vma value;
 276
 277   if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16CJP)
 278     code = BFD_RELOC_IP2K_ADDR16CJP;
 279   else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16P)
 280     code = BFD_RELOC_IP2K_PAGE3;
 281
 282   errmsg = cgen_parse_address (cd, strp, opindex, code,
 283                                & result_type, & value);
 284   if (errmsg == NULL)
 285     {
 286       if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 287         {
 288           if ((value & 0x1) == 0)  /* If the address is even .... */
 289             {
 290               if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16CJP)
 291                 *valuep = (value >> 1) & 0x1FFF;  /* Should mask be 1FFF?  */
 292               else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16P)
 293                 *valuep = (value >> 14) & 0x7;
 294             }
 295           else
 296             errmsg = _("Byte address required. - must be even.");
 297         }
 298       else if (result_type == CGEN_PARSE_OPERAND_RESULT_QUEUED)
 299         {
 300           /* This will happen for things like (s2-s1) where s2 and s1
 301              are labels.  */
 302           *valuep = value;
 303         }
 304       else
 305         errmsg = _("cgen_parse_address returned a symbol. Literal required.");
 306     }
 307   return errmsg;
 308 }
 309
 310 static const char *
 311 parse_lit8 (CGEN_CPU_DESC cd,
 312             const char **strp,
 313             int opindex,
 314             long *valuep)
 315 {
 316   const char *errmsg;
 317   enum cgen_parse_operand_result result_type;
 318   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 319   bfd_vma value;
 320
 321   /* Parse %OP relocating operators.  */
 322   if (strncmp (*strp, "%bank", 5) == 0)
 323     {
 324       *strp += 5;
 325       code = BFD_RELOC_IP2K_BANK;
 326     }
 327   else if (strncmp (*strp, "%lo8data", 8) == 0)
 328     {
 329       *strp += 8;
 330       code = BFD_RELOC_IP2K_LO8DATA;
 331     }
 332   else if (strncmp (*strp, "%hi8data", 8) == 0)
 333     {
 334       *strp += 8;
 335       code = BFD_RELOC_IP2K_HI8DATA;
 336     }
 337   else if (strncmp (*strp, "%ex8data", 8) == 0)
 338     {
 339       *strp += 8;
 340       code = BFD_RELOC_IP2K_EX8DATA;
 341     }
 342   else if (strncmp (*strp, "%lo8insn", 8) == 0)
 343     {
 344       *strp += 8;
 345       code = BFD_RELOC_IP2K_LO8INSN;
 346     }
 347   else if (strncmp (*strp, "%hi8insn", 8) == 0)
 348     {
 349       *strp += 8;
 350       code = BFD_RELOC_IP2K_HI8INSN;
 351     }
 352
 353   /* Parse %op operand.  */
 354   if (code != BFD_RELOC_NONE)
 355     {
 356       errmsg = cgen_parse_address (cd, strp, opindex, code,
 357                                    & result_type, & value);
 358       if ((errmsg == NULL) &&
 359           (result_type != CGEN_PARSE_OPERAND_RESULT_QUEUED))
 360         errmsg = _("percent-operator operand is not a symbol");
 361
 362       *valuep = value;
 363     }
 364   /* Parse as a number.  */
 365   else
 366     {
 367       errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep);
 368
 369       /* Truncate to eight bits to accept both signed and unsigned input.  */
 370       if (errmsg == NULL)
 371         *valuep &= 0xFF;
 372     }
 373
 374   return errmsg;
 375 }
 376
 377 static const char *
 378 parse_bit3 (CGEN_CPU_DESC cd,
 379             const char **strp,
 380             int opindex,
 381             unsigned long *valuep)
 382 {
 383   const char *errmsg;
 384   char mode = 0;
 385   long count = 0;
 386   unsigned long value;
 387
 388   if (strncmp (*strp, "%bit", 4) == 0)
 389     {
 390       *strp += 4;
 391       mode = 1;
 392     }
 393   else if (strncmp (*strp, "%msbbit", 7) == 0)
 394     {
 395       *strp += 7;
 396       mode = 1;
 397     }
 398   else if (strncmp (*strp, "%lsbbit", 7) == 0)
 399     {
 400       *strp += 7;
 401       mode = 2;
 402     }
 403
 404   errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep);
 405   if (errmsg)
 406     return errmsg;
 407
 408   if (mode)
 409     {
 410       value = * valuep;
 411       if (value == 0)
 412         {
 413           errmsg = _("Attempt to find bit index of 0");
 414           return errmsg;
 415         }
 416
 417       if (mode == 1)
 418         {
 419           count = 31;
 420           while ((value & 0x80000000) == 0)
 421             {
 422               count--;
 423               value <<= 1;
 424             }
 425         }
 426       else if (mode == 2)
 427         {
 428           count = 0;
 429           while ((value & 0x00000001) == 0)
 430             {
 431               count++;
 432               value >>= 1;
 433             }
 434         }
 435
 436       *valuep = count;
 437     }
 438
 439   return errmsg;
 440 }
 441
 442 /* -- dis.c */
 443
 444 const char * ip2k_cgen_parse_operand
 445   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
 446
 447 /* Main entry point for operand parsing.
 448
 449    This function is basically just a big switch statement.  Earlier versions
 450    used tables to look up the function to use, but
 451    - if the table contains both assembler and disassembler functions then
 452      the disassembler contains much of the assembler and vice-versa,
 453    - there's a lot of inlining possibilities as things grow,
 454    - using a switch statement avoids the function call overhead.
 455
 456    This function could be moved into `parse_insn_normal', but keeping it
 457    separate makes clear the interface between `parse_insn_normal' and each of
 458    the handlers.  */
 459
 460 const char *
 461 ip2k_cgen_parse_operand (CGEN_CPU_DESC cd,
 462                            int opindex,
 463                            const char ** strp,
 464                            CGEN_FIELDS * fields)
 465 {
 466   const char * errmsg = NULL;
 467   /* Used by scalar operands that still need to be parsed.  */
 468   long junk ATTRIBUTE_UNUSED;
 469
 470   switch (opindex)
 471     {
 472     case IP2K_OPERAND_ADDR16CJP :
 473       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16CJP, (unsigned long *) (& fields->f_addr16cjp));
 474       break;
 475     case IP2K_OPERAND_ADDR16H :
 476       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16H, (unsigned long *) (& fields->f_imm8));
 477       break;
 478     case IP2K_OPERAND_ADDR16L :
 479       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16L, (unsigned long *) (& fields->f_imm8));
 480       break;
 481     case IP2K_OPERAND_ADDR16P :
 482       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16P, (unsigned long *) (& fields->f_page3));
 483       break;
 484     case IP2K_OPERAND_BITNO :
 485       errmsg = parse_bit3 (cd, strp, IP2K_OPERAND_BITNO, (unsigned long *) (& fields->f_bitno));
 486       break;
 487     case IP2K_OPERAND_CBIT :
 488       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_CBIT, (unsigned long *) (& junk));
 489       break;
 490     case IP2K_OPERAND_DCBIT :
 491       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_DCBIT, (unsigned long *) (& junk));
 492       break;
 493     case IP2K_OPERAND_FR :
 494       errmsg = parse_fr (cd, strp, IP2K_OPERAND_FR, (unsigned long *) (& fields->f_reg));
 495       break;
 496     case IP2K_OPERAND_LIT8 :
 497       errmsg = parse_lit8 (cd, strp, IP2K_OPERAND_LIT8, (long *) (& fields->f_imm8));
 498       break;
 499     case IP2K_OPERAND_PABITS :
 500       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_PABITS, (unsigned long *) (& junk));
 501       break;
 502     case IP2K_OPERAND_RETI3 :
 503       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_RETI3, (unsigned long *) (& fields->f_reti3));
 504       break;
 505     case IP2K_OPERAND_ZBIT :
 506       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_ZBIT, (unsigned long *) (& junk));
 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 ip2k_cgen_parse_handlers[] =
 521 {
 522   parse_insn_normal,
 523 };
 524
 525 void
 526 ip2k_cgen_init_asm (CGEN_CPU_DESC cd)
 527 {
 528   ip2k_cgen_init_opcode_table (cd);
 529   ip2k_cgen_init_ibld_table (cd);
 530   cd->parse_handlers = & ip2k_cgen_parse_handlers[0];
 531   cd->parse_operand = ip2k_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 ip2k_cgen_assemble_insn
 547
 548    Returns NULL for success, an error message for failure.  */
 549
 550 char *
 551 ip2k_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 ip2k_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 (! ip2k_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 }