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 (C) 1996-2017 Free Software Foundation, Inc.
   8
   9    This file is part of libopcodes.
  10
  11    This library 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 3, or (at your option)
  14    any later version.
  15
  16    It is distributed in the hope that it will be useful, but WITHOUT
  17    ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
  18    or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
  19    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    51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
  24
  25
  26 /* ??? Eventually more and more of this stuff can go to cpu-independent files.
  27    Keep that in mind.  */
  28
  29 #include "sysdep.h"
  30 #include <stdio.h>
  31 #include "ansidecl.h"
  32 #include "bfd.h"
  33 #include "symcat.h"
  34 #include "ip2k-desc.h"
  35 #include "ip2k-opc.h"
  36 #include "opintl.h"
  37 #include "xregex.h"
  38 #include "libiberty.h"
  39 #include "safe-ctype.h"
  40
  41 #undef  min
  42 #define min(a,b) ((a) < (b) ? (a) : (b))
  43 #undef  max
  44 #define max(a,b) ((a) > (b) ? (a) : (b))
  45
  46 static const char * parse_insn_normal
  47   (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *);
  48 \f
  49 /* -- assembler routines inserted here.  */
  50
  51 /* -- asm.c */
  52
  53 static const char *
  54 parse_fr (CGEN_CPU_DESC cd,
  55           const char **strp,
  56           int opindex,
  57           unsigned long *valuep)
  58 {
  59   const char *errmsg;
  60   const char *old_strp;
  61   char *afteroffset;
  62   enum cgen_parse_operand_result result_type;
  63   bfd_vma value;
  64   extern CGEN_KEYWORD ip2k_cgen_opval_register_names;
  65   bfd_vma tempvalue;
  66
  67   old_strp = *strp;
  68   afteroffset = NULL;
  69
  70   /* Check here to see if you're about to try parsing a w as the first arg
  71      and return an error if you are.  */
  72   if ((strncmp (*strp, "w", 1) == 0) || (strncmp (*strp, "W", 1) == 0))
  73     {
  74       (*strp)++;
  75
  76       if ((strncmp (*strp, ",", 1) == 0) || ISSPACE (**strp))
  77         {
  78           /* We've been passed a w.  Return with an error message so that
  79              cgen will try the next parsing option.  */
  80           errmsg = _("W keyword invalid in FR operand slot.");
  81           return errmsg;
  82         }
  83       *strp = old_strp;
  84     }
  85
  86   /* Attempt parse as register keyword. */
  87   errmsg = cgen_parse_keyword (cd, strp, & ip2k_cgen_opval_register_names,
  88                                (long *) valuep);
  89   if (*strp != NULL
  90       && errmsg == NULL)
  91     return errmsg;
  92
  93   /* Attempt to parse for "(IP)".  */
  94   afteroffset = strstr (*strp, "(IP)");
  95
  96   if (afteroffset == NULL)
  97     /* Make sure it's not in lower case.  */
  98     afteroffset = strstr (*strp, "(ip)");
  99
 100   if (afteroffset != NULL)
 101     {
 102       if (afteroffset != *strp)
 103         {
 104           /* Invalid offset present.  */
 105           errmsg = _("offset(IP) is not a valid form");
 106           return errmsg;
 107         }
 108       else
 109         {
 110           *strp += 4;
 111           *valuep = 0;
 112           errmsg = NULL;
 113           return errmsg;
 114         }
 115     }
 116
 117   /* Attempt to parse for DP. ex: mov w, offset(DP)
 118                                   mov offset(DP),w   */
 119
 120   /* Try parsing it as an address and see what comes back.  */
 121   afteroffset = strstr (*strp, "(DP)");
 122
 123   if (afteroffset == NULL)
 124     /* Maybe it's in lower case.  */
 125     afteroffset = strstr (*strp, "(dp)");
 126
 127   if (afteroffset != NULL)
 128     {
 129       if (afteroffset == *strp)
 130         {
 131           /* No offset present. Use 0 by default.  */
 132           tempvalue = 0;
 133           errmsg = NULL;
 134         }
 135       else
 136         errmsg = cgen_parse_address (cd, strp, opindex,
 137                                      BFD_RELOC_IP2K_FR_OFFSET,
 138                                      & result_type, & tempvalue);
 139
 140       if (errmsg == NULL)
 141         {
 142           if (tempvalue <= 127)
 143             {
 144               /* Value is ok.  Fix up the first 2 bits and return.  */
 145               *valuep = 0x0100 | tempvalue;
 146               *strp += 4; /* Skip over the (DP) in *strp.  */
 147               return errmsg;
 148             }
 149           else
 150             {
 151               /* Found something there in front of (DP) but it's out
 152                  of range.  */
 153               errmsg = _("(DP) offset out of range.");
 154               return errmsg;
 155             }
 156         }
 157     }
 158
 159
 160   /* Attempt to parse for SP. ex: mov w, offset(SP)
 161                                   mov offset(SP), w.  */
 162   afteroffset = strstr (*strp, "(SP)");
 163
 164   if (afteroffset == NULL)
 165     /* Maybe it's in lower case.  */
 166     afteroffset = strstr (*strp, "(sp)");
 167
 168   if (afteroffset != NULL)
 169     {
 170       if (afteroffset == *strp)
 171         {
 172           /* No offset present. Use 0 by default.  */
 173           tempvalue = 0;
 174           errmsg = NULL;
 175         }
 176       else
 177         errmsg = cgen_parse_address (cd, strp, opindex,
 178                                      BFD_RELOC_IP2K_FR_OFFSET,
 179                                      & result_type, & tempvalue);
 180
 181       if (errmsg == NULL)
 182         {
 183           if (tempvalue <= 127)
 184             {
 185               /* Value is ok.  Fix up the first 2 bits and return.  */
 186               *valuep = 0x0180 | tempvalue;
 187               *strp += 4; /* Skip over the (SP) in *strp.  */
 188               return errmsg;
 189             }
 190           else
 191             {
 192               /* Found something there in front of (SP) but it's out
 193                  of range.  */
 194               errmsg = _("(SP) offset out of range.");
 195               return errmsg;
 196             }
 197         }
 198     }
 199
 200   /* Attempt to parse as an address.  */
 201   *strp = old_strp;
 202   errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_IP2K_FR9,
 203                                & result_type, & value);
 204   if (errmsg == NULL)
 205     {
 206       *valuep = value;
 207
 208       /* If a parenthesis is found, warn about invalid form.  */
 209       if (**strp == '(')
 210         errmsg = _("illegal use of parentheses");
 211
 212       /* If a numeric value is specified, ensure that it is between
 213          1 and 255.  */
 214       else if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 215         {
 216           if (value < 0x1 || value > 0xff)
 217             errmsg = _("operand out of range (not between 1 and 255)");
 218         }
 219     }
 220   return errmsg;
 221 }
 222
 223 static const char *
 224 parse_addr16 (CGEN_CPU_DESC cd,
 225               const char **strp,
 226               int opindex,
 227               unsigned long *valuep)
 228 {
 229   const char *errmsg;
 230   enum cgen_parse_operand_result result_type;
 231   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 232   bfd_vma value;
 233
 234   if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16H)
 235     code = BFD_RELOC_IP2K_HI8DATA;
 236   else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16L)
 237     code = BFD_RELOC_IP2K_LO8DATA;
 238   else
 239     {
 240       /* Something is very wrong. opindex has to be one of the above.  */
 241       errmsg = _("parse_addr16: invalid opindex.");
 242       return errmsg;
 243     }
 244
 245   errmsg = cgen_parse_address (cd, strp, opindex, code,
 246                                & result_type, & value);
 247   if (errmsg == NULL)
 248     {
 249       /* We either have a relocation or a number now.  */
 250       if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 251         {
 252           /* We got a number back.  */
 253           if (code == BFD_RELOC_IP2K_HI8DATA)
 254             value >>= 8;
 255           else
 256             /* code = BFD_RELOC_IP2K_LOW8DATA.  */
 257             value &= 0x00FF;
 258         }
 259       *valuep = value;
 260     }
 261
 262   return errmsg;
 263 }
 264
 265 static const char *
 266 parse_addr16_cjp (CGEN_CPU_DESC cd,
 267                   const char **strp,
 268                   int opindex,
 269                   unsigned long *valuep)
 270 {
 271   const char *errmsg;
 272   enum cgen_parse_operand_result result_type;
 273   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 274   bfd_vma value;
 275
 276   if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16CJP)
 277     code = BFD_RELOC_IP2K_ADDR16CJP;
 278   else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16P)
 279     code = BFD_RELOC_IP2K_PAGE3;
 280
 281   errmsg = cgen_parse_address (cd, strp, opindex, code,
 282                                & result_type, & value);
 283   if (errmsg == NULL)
 284     {
 285       if (result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
 286         {
 287           if ((value & 0x1) == 0)  /* If the address is even .... */
 288             {
 289               if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16CJP)
 290                 *valuep = (value >> 1) & 0x1FFF;  /* Should mask be 1FFF?  */
 291               else if (opindex == (CGEN_OPERAND_TYPE) IP2K_OPERAND_ADDR16P)
 292                 *valuep = (value >> 14) & 0x7;
 293             }
 294           else
 295             errmsg = _("Byte address required. - must be even.");
 296         }
 297       else if (result_type == CGEN_PARSE_OPERAND_RESULT_QUEUED)
 298         {
 299           /* This will happen for things like (s2-s1) where s2 and s1
 300              are labels.  */
 301           *valuep = value;
 302         }
 303       else
 304         errmsg = _("cgen_parse_address returned a symbol. Literal required.");
 305     }
 306   return errmsg;
 307 }
 308
 309 static const char *
 310 parse_lit8 (CGEN_CPU_DESC cd,
 311             const char **strp,
 312             int opindex,
 313             long *valuep)
 314 {
 315   const char *errmsg;
 316   enum cgen_parse_operand_result result_type;
 317   bfd_reloc_code_real_type code = BFD_RELOC_NONE;
 318   bfd_vma value;
 319
 320   /* Parse %OP relocating operators.  */
 321   if (strncmp (*strp, "%bank", 5) == 0)
 322     {
 323       *strp += 5;
 324       code = BFD_RELOC_IP2K_BANK;
 325     }
 326   else if (strncmp (*strp, "%lo8data", 8) == 0)
 327     {
 328       *strp += 8;
 329       code = BFD_RELOC_IP2K_LO8DATA;
 330     }
 331   else if (strncmp (*strp, "%hi8data", 8) == 0)
 332     {
 333       *strp += 8;
 334       code = BFD_RELOC_IP2K_HI8DATA;
 335     }
 336   else if (strncmp (*strp, "%ex8data", 8) == 0)
 337     {
 338       *strp += 8;
 339       code = BFD_RELOC_IP2K_EX8DATA;
 340     }
 341   else if (strncmp (*strp, "%lo8insn", 8) == 0)
 342     {
 343       *strp += 8;
 344       code = BFD_RELOC_IP2K_LO8INSN;
 345     }
 346   else if (strncmp (*strp, "%hi8insn", 8) == 0)
 347     {
 348       *strp += 8;
 349       code = BFD_RELOC_IP2K_HI8INSN;
 350     }
 351
 352   /* Parse %op operand.  */
 353   if (code != BFD_RELOC_NONE)
 354     {
 355       errmsg = cgen_parse_address (cd, strp, opindex, code,
 356                                    & result_type, & value);
 357       if ((errmsg == NULL) &&
 358           (result_type != CGEN_PARSE_OPERAND_RESULT_QUEUED))
 359         errmsg = _("percent-operator operand is not a symbol");
 360
 361       *valuep = value;
 362     }
 363   /* Parse as a number.  */
 364   else
 365     {
 366       errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep);
 367
 368       /* Truncate to eight bits to accept both signed and unsigned input.  */
 369       if (errmsg == NULL)
 370         *valuep &= 0xFF;
 371     }
 372
 373   return errmsg;
 374 }
 375
 376 static const char *
 377 parse_bit3 (CGEN_CPU_DESC cd,
 378             const char **strp,
 379             int opindex,
 380             unsigned long *valuep)
 381 {
 382   const char *errmsg;
 383   char mode = 0;
 384   long count = 0;
 385   unsigned long value;
 386
 387   if (strncmp (*strp, "%bit", 4) == 0)
 388     {
 389       *strp += 4;
 390       mode = 1;
 391     }
 392   else if (strncmp (*strp, "%msbbit", 7) == 0)
 393     {
 394       *strp += 7;
 395       mode = 1;
 396     }
 397   else if (strncmp (*strp, "%lsbbit", 7) == 0)
 398     {
 399       *strp += 7;
 400       mode = 2;
 401     }
 402
 403   errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep);
 404   if (errmsg)
 405     return errmsg;
 406
 407   if (mode)
 408     {
 409       value = * valuep;
 410       if (value == 0)
 411         {
 412           errmsg = _("Attempt to find bit index of 0");
 413           return errmsg;
 414         }
 415
 416       if (mode == 1)
 417         {
 418           count = 31;
 419           while ((value & 0x80000000) == 0)
 420             {
 421               count--;
 422               value <<= 1;
 423             }
 424         }
 425       else if (mode == 2)
 426         {
 427           count = 0;
 428           while ((value & 0x00000001) == 0)
 429             {
 430               count++;
 431               value >>= 1;
 432             }
 433         }
 434
 435       *valuep = count;
 436     }
 437
 438   return errmsg;
 439 }
 440
 441 /* -- dis.c */
 442
 443 const char * ip2k_cgen_parse_operand
 444   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
 445
 446 /* Main entry point for operand parsing.
 447
 448    This function is basically just a big switch statement.  Earlier versions
 449    used tables to look up the function to use, but
 450    - if the table contains both assembler and disassembler functions then
 451      the disassembler contains much of the assembler and vice-versa,
 452    - there's a lot of inlining possibilities as things grow,
 453    - using a switch statement avoids the function call overhead.
 454
 455    This function could be moved into `parse_insn_normal', but keeping it
 456    separate makes clear the interface between `parse_insn_normal' and each of
 457    the handlers.  */
 458
 459 const char *
 460 ip2k_cgen_parse_operand (CGEN_CPU_DESC cd,
 461                            int opindex,
 462                            const char ** strp,
 463                            CGEN_FIELDS * fields)
 464 {
 465   const char * errmsg = NULL;
 466   /* Used by scalar operands that still need to be parsed.  */
 467   long junk ATTRIBUTE_UNUSED;
 468
 469   switch (opindex)
 470     {
 471     case IP2K_OPERAND_ADDR16CJP :
 472       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16CJP, (unsigned long *) (& fields->f_addr16cjp));
 473       break;
 474     case IP2K_OPERAND_ADDR16H :
 475       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16H, (unsigned long *) (& fields->f_imm8));
 476       break;
 477     case IP2K_OPERAND_ADDR16L :
 478       errmsg = parse_addr16 (cd, strp, IP2K_OPERAND_ADDR16L, (unsigned long *) (& fields->f_imm8));
 479       break;
 480     case IP2K_OPERAND_ADDR16P :
 481       errmsg = parse_addr16_cjp (cd, strp, IP2K_OPERAND_ADDR16P, (unsigned long *) (& fields->f_page3));
 482       break;
 483     case IP2K_OPERAND_BITNO :
 484       errmsg = parse_bit3 (cd, strp, IP2K_OPERAND_BITNO, (unsigned long *) (& fields->f_bitno));
 485       break;
 486     case IP2K_OPERAND_CBIT :
 487       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_CBIT, (unsigned long *) (& junk));
 488       break;
 489     case IP2K_OPERAND_DCBIT :
 490       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_DCBIT, (unsigned long *) (& junk));
 491       break;
 492     case IP2K_OPERAND_FR :
 493       errmsg = parse_fr (cd, strp, IP2K_OPERAND_FR, (unsigned long *) (& fields->f_reg));
 494       break;
 495     case IP2K_OPERAND_LIT8 :
 496       errmsg = parse_lit8 (cd, strp, IP2K_OPERAND_LIT8, (long *) (& fields->f_imm8));
 497       break;
 498     case IP2K_OPERAND_PABITS :
 499       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_PABITS, (unsigned long *) (& junk));
 500       break;
 501     case IP2K_OPERAND_RETI3 :
 502       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_RETI3, (unsigned long *) (& fields->f_reti3));
 503       break;
 504     case IP2K_OPERAND_ZBIT :
 505       errmsg = cgen_parse_unsigned_integer (cd, strp, IP2K_OPERAND_ZBIT, (unsigned long *) (& junk));
 506       break;
 507
 508     default :
 509       /* xgettext:c-format */
 510       fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex);
 511       abort ();
 512   }
 513
 514   return errmsg;
 515 }
 516
 517 cgen_parse_fn * const ip2k_cgen_parse_handlers[] =
 518 {
 519   parse_insn_normal,
 520 };
 521
 522 void
 523 ip2k_cgen_init_asm (CGEN_CPU_DESC cd)
 524 {
 525   ip2k_cgen_init_opcode_table (cd);
 526   ip2k_cgen_init_ibld_table (cd);
 527   cd->parse_handlers = & ip2k_cgen_parse_handlers[0];
 528   cd->parse_operand = ip2k_cgen_parse_operand;
 529 #ifdef CGEN_ASM_INIT_HOOK
 530 CGEN_ASM_INIT_HOOK
 531 #endif
 532 }
 533
 534 \f
 535
 536 /* Regex construction routine.
 537
 538    This translates an opcode syntax string into a regex string,
 539    by replacing any non-character syntax element (such as an
 540    opcode) with the pattern '.*'
 541
 542    It then compiles the regex and stores it in the opcode, for
 543    later use by ip2k_cgen_assemble_insn
 544
 545    Returns NULL for success, an error message for failure.  */
 546
 547 char *
 548 ip2k_cgen_build_insn_regex (CGEN_INSN *insn)
 549 {
 550   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
 551   const char *mnem = CGEN_INSN_MNEMONIC (insn);
 552   char rxbuf[CGEN_MAX_RX_ELEMENTS];
 553   char *rx = rxbuf;
 554   const CGEN_SYNTAX_CHAR_TYPE *syn;
 555   int reg_err;
 556
 557   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
 558
 559   /* Mnemonics come first in the syntax string.  */
 560   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 561     return _("missing mnemonic in syntax string");
 562   ++syn;
 563
 564   /* Generate a case sensitive regular expression that emulates case
 565      insensitive matching in the "C" locale.  We cannot generate a case
 566      insensitive regular expression because in Turkish locales, 'i' and 'I'
 567      are not equal modulo case conversion.  */
 568
 569   /* Copy the literal mnemonic out of the insn.  */
 570   for (; *mnem; mnem++)
 571     {
 572       char c = *mnem;
 573
 574       if (ISALPHA (c))
 575         {
 576           *rx++ = '[';
 577           *rx++ = TOLOWER (c);
 578           *rx++ = TOUPPER (c);
 579           *rx++ = ']';
 580         }
 581       else
 582         *rx++ = c;
 583     }
 584
 585   /* Copy any remaining literals from the syntax string into the rx.  */
 586   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
 587     {
 588       if (CGEN_SYNTAX_CHAR_P (* syn))
 589         {
 590           char c = CGEN_SYNTAX_CHAR (* syn);
 591
 592           switch (c)
 593             {
 594               /* Escape any regex metacharacters in the syntax.  */
 595             case '.': case '[': case '\\':
 596             case '*': case '^': case '$':
 597
 598 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
 599             case '?': case '{': case '}':
 600             case '(': case ')': case '*':
 601             case '|': case '+': case ']':
 602 #endif
 603               *rx++ = '\\';
 604               *rx++ = c;
 605               break;
 606
 607             default:
 608               if (ISALPHA (c))
 609                 {
 610                   *rx++ = '[';
 611                   *rx++ = TOLOWER (c);
 612                   *rx++ = TOUPPER (c);
 613                   *rx++ = ']';
 614                 }
 615               else
 616                 *rx++ = c;
 617               break;
 618             }
 619         }
 620       else
 621         {
 622           /* Replace non-syntax fields with globs.  */
 623           *rx++ = '.';
 624           *rx++ = '*';
 625         }
 626     }
 627
 628   /* Trailing whitespace ok.  */
 629   * rx++ = '[';
 630   * rx++ = ' ';
 631   * rx++ = '\t';
 632   * rx++ = ']';
 633   * rx++ = '*';
 634
 635   /* But anchor it after that.  */
 636   * rx++ = '$';
 637   * rx = '\0';
 638
 639   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
 640   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
 641
 642   if (reg_err == 0)
 643     return NULL;
 644   else
 645     {
 646       static char msg[80];
 647
 648       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
 649       regfree ((regex_t *) CGEN_INSN_RX (insn));
 650       free (CGEN_INSN_RX (insn));
 651       (CGEN_INSN_RX (insn)) = NULL;
 652       return msg;
 653     }
 654 }
 655
 656 \f
 657 /* Default insn parser.
 658
 659    The syntax string is scanned and operands are parsed and stored in FIELDS.
 660    Relocs are queued as we go via other callbacks.
 661
 662    ??? Note that this is currently an all-or-nothing parser.  If we fail to
 663    parse the instruction, we return 0 and the caller will start over from
 664    the beginning.  Backtracking will be necessary in parsing subexpressions,
 665    but that can be handled there.  Not handling backtracking here may get
 666    expensive in the case of the m68k.  Deal with later.
 667
 668    Returns NULL for success, an error message for failure.  */
 669
 670 static const char *
 671 parse_insn_normal (CGEN_CPU_DESC cd,
 672                    const CGEN_INSN *insn,
 673                    const char **strp,
 674                    CGEN_FIELDS *fields)
 675 {
 676   /* ??? Runtime added insns not handled yet.  */
 677   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
 678   const char *str = *strp;
 679   const char *errmsg;
 680   const char *p;
 681   const CGEN_SYNTAX_CHAR_TYPE * syn;
 682 #ifdef CGEN_MNEMONIC_OPERANDS
 683   /* FIXME: wip */
 684   int past_opcode_p;
 685 #endif
 686
 687   /* For now we assume the mnemonic is first (there are no leading operands).
 688      We can parse it without needing to set up operand parsing.
 689      GAS's input scrubber will ensure mnemonics are lowercase, but we may
 690      not be called from GAS.  */
 691   p = CGEN_INSN_MNEMONIC (insn);
 692   while (*p && TOLOWER (*p) == TOLOWER (*str))
 693     ++p, ++str;
 694
 695   if (* p)
 696     return _("unrecognized instruction");
 697
 698 #ifndef CGEN_MNEMONIC_OPERANDS
 699   if (* str && ! ISSPACE (* str))
 700     return _("unrecognized instruction");
 701 #endif
 702
 703   CGEN_INIT_PARSE (cd);
 704   cgen_init_parse_operand (cd);
 705 #ifdef CGEN_MNEMONIC_OPERANDS
 706   past_opcode_p = 0;
 707 #endif
 708
 709   /* We don't check for (*str != '\0') here because we want to parse
 710      any trailing fake arguments in the syntax string.  */
 711   syn = CGEN_SYNTAX_STRING (syntax);
 712
 713   /* Mnemonics come first for now, ensure valid string.  */
 714   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 715     abort ();
 716
 717   ++syn;
 718
 719   while (* syn != 0)
 720     {
 721       /* Non operand chars must match exactly.  */
 722       if (CGEN_SYNTAX_CHAR_P (* syn))
 723         {
 724           /* FIXME: While we allow for non-GAS callers above, we assume the
 725              first char after the mnemonic part is a space.  */
 726           /* FIXME: We also take inappropriate advantage of the fact that
 727              GAS's input scrubber will remove extraneous blanks.  */
 728           if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
 729             {
 730 #ifdef CGEN_MNEMONIC_OPERANDS
 731               if (CGEN_SYNTAX_CHAR(* syn) == ' ')
 732                 past_opcode_p = 1;
 733 #endif
 734               ++ syn;
 735               ++ str;
 736             }
 737           else if (*str)
 738             {
 739               /* Syntax char didn't match.  Can't be this insn.  */
 740               static char msg [80];
 741
 742               /* xgettext:c-format */
 743               sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
 744                        CGEN_SYNTAX_CHAR(*syn), *str);
 745               return msg;
 746             }
 747           else
 748             {
 749               /* Ran out of input.  */
 750               static char msg [80];
 751
 752               /* xgettext:c-format */
 753               sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
 754                        CGEN_SYNTAX_CHAR(*syn));
 755               return msg;
 756             }
 757           continue;
 758         }
 759
 760 #ifdef CGEN_MNEMONIC_OPERANDS
 761       (void) past_opcode_p;
 762 #endif
 763       /* We have an operand of some sort.  */
 764       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn), &str, fields);
 765       if (errmsg)
 766         return errmsg;
 767
 768       /* Done with this operand, continue with next one.  */
 769       ++ syn;
 770     }
 771
 772   /* If we're at the end of the syntax string, we're done.  */
 773   if (* syn == 0)
 774     {
 775       /* FIXME: For the moment we assume a valid `str' can only contain
 776          blanks now.  IE: We needn't try again with a longer version of
 777          the insn and it is assumed that longer versions of insns appear
 778          before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
 779       while (ISSPACE (* str))
 780         ++ str;
 781
 782       if (* str != '\0')
 783         return _("junk at end of line"); /* FIXME: would like to include `str' */
 784
 785       return NULL;
 786     }
 787
 788   /* We couldn't parse it.  */
 789   return _("unrecognized instruction");
 790 }
 791 \f
 792 /* Main entry point.
 793    This routine is called for each instruction to be assembled.
 794    STR points to the insn to be assembled.
 795    We assume all necessary tables have been initialized.
 796    The assembled instruction, less any fixups, is stored in BUF.
 797    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
 798    still needs to be converted to target byte order, otherwise BUF is an array
 799    of bytes in target byte order.
 800    The result is a pointer to the insn's entry in the opcode table,
 801    or NULL if an error occured (an error message will have already been
 802    printed).
 803
 804    Note that when processing (non-alias) macro-insns,
 805    this function recurses.
 806
 807    ??? It's possible to make this cpu-independent.
 808    One would have to deal with a few minor things.
 809    At this point in time doing so would be more of a curiosity than useful
 810    [for example this file isn't _that_ big], but keeping the possibility in
 811    mind helps keep the design clean.  */
 812
 813 const CGEN_INSN *
 814 ip2k_cgen_assemble_insn (CGEN_CPU_DESC cd,
 815                            const char *str,
 816                            CGEN_FIELDS *fields,
 817                            CGEN_INSN_BYTES_PTR buf,
 818                            char **errmsg)
 819 {
 820   const char *start;
 821   CGEN_INSN_LIST *ilist;
 822   const char *parse_errmsg = NULL;
 823   const char *insert_errmsg = NULL;
 824   int recognized_mnemonic = 0;
 825
 826   /* Skip leading white space.  */
 827   while (ISSPACE (* str))
 828     ++ str;
 829
 830   /* The instructions are stored in hashed lists.
 831      Get the first in the list.  */
 832   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
 833
 834   /* Keep looking until we find a match.  */
 835   start = str;
 836   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
 837     {
 838       const CGEN_INSN *insn = ilist->insn;
 839       recognized_mnemonic = 1;
 840
 841 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
 842       /* Not usually needed as unsupported opcodes
 843          shouldn't be in the hash lists.  */
 844       /* Is this insn supported by the selected cpu?  */
 845       if (! ip2k_cgen_insn_supported (cd, insn))
 846         continue;
 847 #endif
 848       /* If the RELAXED attribute is set, this is an insn that shouldn't be
 849          chosen immediately.  Instead, it is used during assembler/linker
 850          relaxation if possible.  */
 851       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
 852         continue;
 853
 854       str = start;
 855
 856       /* Skip this insn if str doesn't look right lexically.  */
 857       if (CGEN_INSN_RX (insn) != NULL &&
 858           regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
 859         continue;
 860
 861       /* Allow parse/insert handlers to obtain length of insn.  */
 862       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
 863
 864       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
 865       if (parse_errmsg != NULL)
 866         continue;
 867
 868       /* ??? 0 is passed for `pc'.  */
 869       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
 870                                                  (bfd_vma) 0);
 871       if (insert_errmsg != NULL)
 872         continue;
 873
 874       /* It is up to the caller to actually output the insn and any
 875          queued relocs.  */
 876       return insn;
 877     }
 878
 879   {
 880     static char errbuf[150];
 881     const char *tmp_errmsg;
 882 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
 883 #define be_verbose 1
 884 #else
 885 #define be_verbose 0
 886 #endif
 887
 888     if (be_verbose)
 889       {
 890         /* If requesting verbose error messages, use insert_errmsg.
 891            Failing that, use parse_errmsg.  */
 892         tmp_errmsg = (insert_errmsg ? insert_errmsg :
 893                       parse_errmsg ? parse_errmsg :
 894                       recognized_mnemonic ?
 895                       _("unrecognized form of instruction") :
 896                       _("unrecognized instruction"));
 897
 898         if (strlen (start) > 50)
 899           /* xgettext:c-format */
 900           sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
 901         else
 902           /* xgettext:c-format */
 903           sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
 904       }
 905     else
 906       {
 907         if (strlen (start) > 50)
 908           /* xgettext:c-format */
 909           sprintf (errbuf, _("bad instruction `%.50s...'"), start);
 910         else
 911           /* xgettext:c-format */
 912           sprintf (errbuf, _("bad instruction `%.50s'"), start);
 913       }
 914
 915     *errmsg = errbuf;
 916     return NULL;
 917   }
 918 }