opcodes/openrisc-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, 2005
   8    Free Software Foundation, Inc.
   9
  10    This file is part of the GNU Binutils and GDB, the GNU debugger.
  11
  12    This program 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 2, or (at your option)
  15    any later version.
  16
  17    This program is distributed in the hope that it will be useful,
  18    but WITHOUT ANY WARRANTY; without even the implied warranty of
  19    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20    GNU General Public 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 /* ??? 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 "openrisc-desc.h"
  35 #include "openrisc-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 * MISSING_CLOSING_PARENTHESIS = N_("missing `)'");
  54
  55 #define CGEN_VERBOSE_ASSEMBLER_ERRORS
  56
  57 long
  58 openrisc_sign_extend_16bit (long value)
  59 {
  60   return ((value & 0xffff) ^ 0x8000) - 0x8000;
  61 }
  62
  63 /* Handle hi().  */
  64
  65 static const char *
  66 parse_hi16 (CGEN_CPU_DESC cd, const char ** strp, int opindex, long * valuep)
  67 {
  68   const char *errmsg;
  69   enum cgen_parse_operand_result result_type;
  70   unsigned long ret;
  71
  72   if (**strp == '#')
  73     ++*strp;
  74
  75   if (strncasecmp (*strp, "hi(", 3) == 0)
  76     {
  77       bfd_vma value;
  78
  79       *strp += 3;
  80       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_HI16,
  81                                    & result_type, & value);
  82       if (**strp != ')')
  83         return MISSING_CLOSING_PARENTHESIS;
  84
  85       ++*strp;
  86       if (errmsg == NULL
  87           && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
  88         value >>= 16;
  89       ret = value;
  90     }
  91   else
  92     {
  93       if (**strp == '-')
  94         {
  95           long value;
  96
  97           errmsg = cgen_parse_signed_integer (cd, strp, opindex, &value);
  98           ret = value;
  99         }
 100       else
 101         {
 102           unsigned long value;
 103
 104           errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, &value);
 105           ret = value;
 106         }
 107     }
 108
 109   *valuep = ((ret & 0xffff) ^ 0x8000) - 0x8000;
 110   return errmsg;
 111 }
 112
 113 /* Handle lo().  */
 114
 115 static const char *
 116 parse_lo16 (CGEN_CPU_DESC cd, const char ** strp, int opindex, long * valuep)
 117 {
 118   const char *errmsg;
 119   enum cgen_parse_operand_result result_type;
 120   unsigned long ret;
 121
 122   if (**strp == '#')
 123     ++*strp;
 124
 125   if (strncasecmp (*strp, "lo(", 3) == 0)
 126     {
 127       bfd_vma value;
 128
 129       *strp += 3;
 130       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_LO16,
 131                                    & result_type, & value);
 132       if (**strp != ')')
 133         return MISSING_CLOSING_PARENTHESIS;
 134
 135       ++*strp;
 136       ret = value;
 137     }
 138   else
 139     {
 140       if (**strp == '-')
 141         {
 142           long value;
 143
 144           errmsg = cgen_parse_signed_integer (cd, strp, opindex, &value);
 145           ret = value;
 146         }
 147       else
 148         {
 149           unsigned long value;
 150
 151           errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, &value);
 152           ret = value;
 153         }
 154     }
 155
 156   *valuep = ((ret & 0xffff) ^ 0x8000) - 0x8000;
 157   return errmsg;
 158 }
 159
 160 /* -- */
 161
 162 const char * openrisc_cgen_parse_operand
 163   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
 164
 165 /* Main entry point for operand parsing.
 166
 167    This function is basically just a big switch statement.  Earlier versions
 168    used tables to look up the function to use, but
 169    - if the table contains both assembler and disassembler functions then
 170      the disassembler contains much of the assembler and vice-versa,
 171    - there's a lot of inlining possibilities as things grow,
 172    - using a switch statement avoids the function call overhead.
 173
 174    This function could be moved into `parse_insn_normal', but keeping it
 175    separate makes clear the interface between `parse_insn_normal' and each of
 176    the handlers.  */
 177
 178 const char *
 179 openrisc_cgen_parse_operand (CGEN_CPU_DESC cd,
 180                            int opindex,
 181                            const char ** strp,
 182                            CGEN_FIELDS * fields)
 183 {
 184   const char * errmsg = NULL;
 185   /* Used by scalar operands that still need to be parsed.  */
 186   long junk ATTRIBUTE_UNUSED;
 187
 188   switch (opindex)
 189     {
 190     case OPENRISC_OPERAND_ABS_26 :
 191       {
 192         bfd_vma value = 0;
 193         errmsg = cgen_parse_address (cd, strp, OPENRISC_OPERAND_ABS_26, 0, NULL,  & value);
 194         fields->f_abs26 = value;
 195       }
 196       break;
 197     case OPENRISC_OPERAND_DISP_26 :
 198       {
 199         bfd_vma value = 0;
 200         errmsg = cgen_parse_address (cd, strp, OPENRISC_OPERAND_DISP_26, 0, NULL,  & value);
 201         fields->f_disp26 = value;
 202       }
 203       break;
 204     case OPENRISC_OPERAND_HI16 :
 205       errmsg = parse_hi16 (cd, strp, OPENRISC_OPERAND_HI16, (long *) (& fields->f_simm16));
 206       break;
 207     case OPENRISC_OPERAND_LO16 :
 208       errmsg = parse_lo16 (cd, strp, OPENRISC_OPERAND_LO16, (long *) (& fields->f_lo16));
 209       break;
 210     case OPENRISC_OPERAND_OP_F_23 :
 211       errmsg = cgen_parse_unsigned_integer (cd, strp, OPENRISC_OPERAND_OP_F_23, (unsigned long *) (& fields->f_op4));
 212       break;
 213     case OPENRISC_OPERAND_OP_F_3 :
 214       errmsg = cgen_parse_unsigned_integer (cd, strp, OPENRISC_OPERAND_OP_F_3, (unsigned long *) (& fields->f_op5));
 215       break;
 216     case OPENRISC_OPERAND_RA :
 217       errmsg = cgen_parse_keyword (cd, strp, & openrisc_cgen_opval_h_gr, & fields->f_r2);
 218       break;
 219     case OPENRISC_OPERAND_RB :
 220       errmsg = cgen_parse_keyword (cd, strp, & openrisc_cgen_opval_h_gr, & fields->f_r3);
 221       break;
 222     case OPENRISC_OPERAND_RD :
 223       errmsg = cgen_parse_keyword (cd, strp, & openrisc_cgen_opval_h_gr, & fields->f_r1);
 224       break;
 225     case OPENRISC_OPERAND_SIMM_16 :
 226       errmsg = cgen_parse_signed_integer (cd, strp, OPENRISC_OPERAND_SIMM_16, (long *) (& fields->f_simm16));
 227       break;
 228     case OPENRISC_OPERAND_UI16NC :
 229       errmsg = parse_lo16 (cd, strp, OPENRISC_OPERAND_UI16NC, (long *) (& fields->f_i16nc));
 230       break;
 231     case OPENRISC_OPERAND_UIMM_16 :
 232       errmsg = cgen_parse_unsigned_integer (cd, strp, OPENRISC_OPERAND_UIMM_16, (unsigned long *) (& fields->f_uimm16));
 233       break;
 234     case OPENRISC_OPERAND_UIMM_5 :
 235       errmsg = cgen_parse_unsigned_integer (cd, strp, OPENRISC_OPERAND_UIMM_5, (unsigned long *) (& fields->f_uimm5));
 236       break;
 237
 238     default :
 239       /* xgettext:c-format */
 240       fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex);
 241       abort ();
 242   }
 243
 244   return errmsg;
 245 }
 246
 247 cgen_parse_fn * const openrisc_cgen_parse_handlers[] =
 248 {
 249   parse_insn_normal,
 250 };
 251
 252 void
 253 openrisc_cgen_init_asm (CGEN_CPU_DESC cd)
 254 {
 255   openrisc_cgen_init_opcode_table (cd);
 256   openrisc_cgen_init_ibld_table (cd);
 257   cd->parse_handlers = & openrisc_cgen_parse_handlers[0];
 258   cd->parse_operand = openrisc_cgen_parse_operand;
 259 }
 260
 261 \f
 262
 263 /* Regex construction routine.
 264
 265    This translates an opcode syntax string into a regex string,
 266    by replacing any non-character syntax element (such as an
 267    opcode) with the pattern '.*'
 268
 269    It then compiles the regex and stores it in the opcode, for
 270    later use by openrisc_cgen_assemble_insn
 271
 272    Returns NULL for success, an error message for failure.  */
 273
 274 char *
 275 openrisc_cgen_build_insn_regex (CGEN_INSN *insn)
 276 {
 277   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
 278   const char *mnem = CGEN_INSN_MNEMONIC (insn);
 279   char rxbuf[CGEN_MAX_RX_ELEMENTS];
 280   char *rx = rxbuf;
 281   const CGEN_SYNTAX_CHAR_TYPE *syn;
 282   int reg_err;
 283
 284   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
 285
 286   /* Mnemonics come first in the syntax string.  */
 287   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 288     return _("missing mnemonic in syntax string");
 289   ++syn;
 290
 291   /* Generate a case sensitive regular expression that emulates case
 292      insensitive matching in the "C" locale.  We cannot generate a case
 293      insensitive regular expression because in Turkish locales, 'i' and 'I'
 294      are not equal modulo case conversion.  */
 295
 296   /* Copy the literal mnemonic out of the insn.  */
 297   for (; *mnem; mnem++)
 298     {
 299       char c = *mnem;
 300
 301       if (ISALPHA (c))
 302         {
 303           *rx++ = '[';
 304           *rx++ = TOLOWER (c);
 305           *rx++ = TOUPPER (c);
 306           *rx++ = ']';
 307         }
 308       else
 309         *rx++ = c;
 310     }
 311
 312   /* Copy any remaining literals from the syntax string into the rx.  */
 313   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
 314     {
 315       if (CGEN_SYNTAX_CHAR_P (* syn))
 316         {
 317           char c = CGEN_SYNTAX_CHAR (* syn);
 318
 319           switch (c)
 320             {
 321               /* Escape any regex metacharacters in the syntax.  */
 322             case '.': case '[': case '\\':
 323             case '*': case '^': case '$':
 324
 325 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
 326             case '?': case '{': case '}':
 327             case '(': case ')': case '*':
 328             case '|': case '+': case ']':
 329 #endif
 330               *rx++ = '\\';
 331               *rx++ = c;
 332               break;
 333
 334             default:
 335               if (ISALPHA (c))
 336                 {
 337                   *rx++ = '[';
 338                   *rx++ = TOLOWER (c);
 339                   *rx++ = TOUPPER (c);
 340                   *rx++ = ']';
 341                 }
 342               else
 343                 *rx++ = c;
 344               break;
 345             }
 346         }
 347       else
 348         {
 349           /* Replace non-syntax fields with globs.  */
 350           *rx++ = '.';
 351           *rx++ = '*';
 352         }
 353     }
 354
 355   /* Trailing whitespace ok.  */
 356   * rx++ = '[';
 357   * rx++ = ' ';
 358   * rx++ = '\t';
 359   * rx++ = ']';
 360   * rx++ = '*';
 361
 362   /* But anchor it after that.  */
 363   * rx++ = '$';
 364   * rx = '\0';
 365
 366   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
 367   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
 368
 369   if (reg_err == 0)
 370     return NULL;
 371   else
 372     {
 373       static char msg[80];
 374
 375       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
 376       regfree ((regex_t *) CGEN_INSN_RX (insn));
 377       free (CGEN_INSN_RX (insn));
 378       (CGEN_INSN_RX (insn)) = NULL;
 379       return msg;
 380     }
 381 }
 382
 383 \f
 384 /* Default insn parser.
 385
 386    The syntax string is scanned and operands are parsed and stored in FIELDS.
 387    Relocs are queued as we go via other callbacks.
 388
 389    ??? Note that this is currently an all-or-nothing parser.  If we fail to
 390    parse the instruction, we return 0 and the caller will start over from
 391    the beginning.  Backtracking will be necessary in parsing subexpressions,
 392    but that can be handled there.  Not handling backtracking here may get
 393    expensive in the case of the m68k.  Deal with later.
 394
 395    Returns NULL for success, an error message for failure.  */
 396
 397 static const char *
 398 parse_insn_normal (CGEN_CPU_DESC cd,
 399                    const CGEN_INSN *insn,
 400                    const char **strp,
 401                    CGEN_FIELDS *fields)
 402 {
 403   /* ??? Runtime added insns not handled yet.  */
 404   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
 405   const char *str = *strp;
 406   const char *errmsg;
 407   const char *p;
 408   const CGEN_SYNTAX_CHAR_TYPE * syn;
 409 #ifdef CGEN_MNEMONIC_OPERANDS
 410   /* FIXME: wip */
 411   int past_opcode_p;
 412 #endif
 413
 414   /* For now we assume the mnemonic is first (there are no leading operands).
 415      We can parse it without needing to set up operand parsing.
 416      GAS's input scrubber will ensure mnemonics are lowercase, but we may
 417      not be called from GAS.  */
 418   p = CGEN_INSN_MNEMONIC (insn);
 419   while (*p && TOLOWER (*p) == TOLOWER (*str))
 420     ++p, ++str;
 421
 422   if (* p)
 423     return _("unrecognized instruction");
 424
 425 #ifndef CGEN_MNEMONIC_OPERANDS
 426   if (* str && ! ISSPACE (* str))
 427     return _("unrecognized instruction");
 428 #endif
 429
 430   CGEN_INIT_PARSE (cd);
 431   cgen_init_parse_operand (cd);
 432 #ifdef CGEN_MNEMONIC_OPERANDS
 433   past_opcode_p = 0;
 434 #endif
 435
 436   /* We don't check for (*str != '\0') here because we want to parse
 437      any trailing fake arguments in the syntax string.  */
 438   syn = CGEN_SYNTAX_STRING (syntax);
 439
 440   /* Mnemonics come first for now, ensure valid string.  */
 441   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
 442     abort ();
 443
 444   ++syn;
 445
 446   while (* syn != 0)
 447     {
 448       /* Non operand chars must match exactly.  */
 449       if (CGEN_SYNTAX_CHAR_P (* syn))
 450         {
 451           /* FIXME: While we allow for non-GAS callers above, we assume the
 452              first char after the mnemonic part is a space.  */
 453           /* FIXME: We also take inappropriate advantage of the fact that
 454              GAS's input scrubber will remove extraneous blanks.  */
 455           if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
 456             {
 457 #ifdef CGEN_MNEMONIC_OPERANDS
 458               if (CGEN_SYNTAX_CHAR(* syn) == ' ')
 459                 past_opcode_p = 1;
 460 #endif
 461               ++ syn;
 462               ++ str;
 463             }
 464           else if (*str)
 465             {
 466               /* Syntax char didn't match.  Can't be this insn.  */
 467               static char msg [80];
 468
 469               /* xgettext:c-format */
 470               sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
 471                        CGEN_SYNTAX_CHAR(*syn), *str);
 472               return msg;
 473             }
 474           else
 475             {
 476               /* Ran out of input.  */
 477               static char msg [80];
 478
 479               /* xgettext:c-format */
 480               sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
 481                        CGEN_SYNTAX_CHAR(*syn));
 482               return msg;
 483             }
 484           continue;
 485         }
 486
 487       /* We have an operand of some sort.  */
 488       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn),
 489                                           &str, fields);
 490       if (errmsg)
 491         return errmsg;
 492
 493       /* Done with this operand, continue with next one.  */
 494       ++ syn;
 495     }
 496
 497   /* If we're at the end of the syntax string, we're done.  */
 498   if (* syn == 0)
 499     {
 500       /* FIXME: For the moment we assume a valid `str' can only contain
 501          blanks now.  IE: We needn't try again with a longer version of
 502          the insn and it is assumed that longer versions of insns appear
 503          before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
 504       while (ISSPACE (* str))
 505         ++ str;
 506
 507       if (* str != '\0')
 508         return _("junk at end of line"); /* FIXME: would like to include `str' */
 509
 510       return NULL;
 511     }
 512
 513   /* We couldn't parse it.  */
 514   return _("unrecognized instruction");
 515 }
 516 \f
 517 /* Main entry point.
 518    This routine is called for each instruction to be assembled.
 519    STR points to the insn to be assembled.
 520    We assume all necessary tables have been initialized.
 521    The assembled instruction, less any fixups, is stored in BUF.
 522    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
 523    still needs to be converted to target byte order, otherwise BUF is an array
 524    of bytes in target byte order.
 525    The result is a pointer to the insn's entry in the opcode table,
 526    or NULL if an error occured (an error message will have already been
 527    printed).
 528
 529    Note that when processing (non-alias) macro-insns,
 530    this function recurses.
 531
 532    ??? It's possible to make this cpu-independent.
 533    One would have to deal with a few minor things.
 534    At this point in time doing so would be more of a curiosity than useful
 535    [for example this file isn't _that_ big], but keeping the possibility in
 536    mind helps keep the design clean.  */
 537
 538 const CGEN_INSN *
 539 openrisc_cgen_assemble_insn (CGEN_CPU_DESC cd,
 540                            const char *str,
 541                            CGEN_FIELDS *fields,
 542                            CGEN_INSN_BYTES_PTR buf,
 543                            char **errmsg)
 544 {
 545   const char *start;
 546   CGEN_INSN_LIST *ilist;
 547   const char *parse_errmsg = NULL;
 548   const char *insert_errmsg = NULL;
 549   int recognized_mnemonic = 0;
 550
 551   /* Skip leading white space.  */
 552   while (ISSPACE (* str))
 553     ++ str;
 554
 555   /* The instructions are stored in hashed lists.
 556      Get the first in the list.  */
 557   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
 558
 559   /* Keep looking until we find a match.  */
 560   start = str;
 561   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
 562     {
 563       const CGEN_INSN *insn = ilist->insn;
 564       recognized_mnemonic = 1;
 565
 566 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
 567       /* Not usually needed as unsupported opcodes
 568          shouldn't be in the hash lists.  */
 569       /* Is this insn supported by the selected cpu?  */
 570       if (! openrisc_cgen_insn_supported (cd, insn))
 571         continue;
 572 #endif
 573       /* If the RELAXED attribute is set, this is an insn that shouldn't be
 574          chosen immediately.  Instead, it is used during assembler/linker
 575          relaxation if possible.  */
 576       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
 577         continue;
 578
 579       str = start;
 580
 581       /* Skip this insn if str doesn't look right lexically.  */
 582       if (CGEN_INSN_RX (insn) != NULL &&
 583           regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
 584         continue;
 585
 586       /* Allow parse/insert handlers to obtain length of insn.  */
 587       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
 588
 589       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
 590       if (parse_errmsg != NULL)
 591         continue;
 592
 593       /* ??? 0 is passed for `pc'.  */
 594       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
 595                                                  (bfd_vma) 0);
 596       if (insert_errmsg != NULL)
 597         continue;
 598
 599       /* It is up to the caller to actually output the insn and any
 600          queued relocs.  */
 601       return insn;
 602     }
 603
 604   {
 605     static char errbuf[150];
 606 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
 607     const char *tmp_errmsg;
 608
 609     /* If requesting verbose error messages, use insert_errmsg.
 610        Failing that, use parse_errmsg.  */
 611     tmp_errmsg = (insert_errmsg ? insert_errmsg :
 612                   parse_errmsg ? parse_errmsg :
 613                   recognized_mnemonic ?
 614                   _("unrecognized form of instruction") :
 615                   _("unrecognized instruction"));
 616
 617     if (strlen (start) > 50)
 618       /* xgettext:c-format */
 619       sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
 620     else
 621       /* xgettext:c-format */
 622       sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
 623 #else
 624     if (strlen (start) > 50)
 625       /* xgettext:c-format */
 626       sprintf (errbuf, _("bad instruction `%.50s...'"), start);
 627     else
 628       /* xgettext:c-format */
 629       sprintf (errbuf, _("bad instruction `%.50s'"), start);
 630 #endif
 631
 632     *errmsg = errbuf;
 633     return NULL;
 634   }
 635 }