parser.c

   1 /* parser.c   source line parser for the Netwide Assembler
   2  *
   3  * The Netwide Assembler is copyright (C) 1996 Simon Tatham and
   4  * Julian Hall. All rights reserved. The software is
   5  * redistributable under the license given in the file "LICENSE"
   6  * distributed in the NASM archive.
   7  *
   8  * initial version 27/iii/95 by Simon Tatham
   9  */
  10
  11 #include "compiler.h"
  12
  13 #include <stdio.h>
  14 #include <stdlib.h>
  15 #include <stddef.h>
  16 #include <string.h>
  17 #include <ctype.h>
  18 #include <inttypes.h>
  19
  20 #include "nasm.h"
  21 #include "insns.h"
  22 #include "nasmlib.h"
  23 #include "stdscan.h"
  24 #include "parser.h"
  25 #include "float.h"
  26
  27 extern int in_abs_seg;          /* ABSOLUTE segment flag */
  28 extern int32_t abs_seg;            /* ABSOLUTE segment */
  29 extern int32_t abs_offset;         /* ABSOLUTE segment offset */
  30
  31 #include "regflags.c"           /* List of register flags */
  32
  33 static int is_comma_next(void);
  34
  35 static int i;
  36 static struct tokenval tokval;
  37 static efunc error;
  38 static struct ofmt *outfmt;     /* Structure of addresses of output routines */
  39 static struct location *location;         /* Pointer to current line's segment,offset */
  40
  41 void parser_global_info(struct ofmt *output, struct location * locp)
  42 {
  43     outfmt = output;
  44     location = locp;
  45 }
  46
  47 static int prefix_slot(enum prefixes prefix)
  48 {
  49     switch (prefix) {
  50     case R_CS:
  51     case R_DS:
  52     case R_SS:
  53     case R_ES:
  54     case R_FS:
  55     case R_GS:
  56         return PPS_SEG;
  57     case P_LOCK:
  58     case P_REP:
  59     case P_REPE:
  60     case P_REPZ:
  61     case P_REPNE:
  62     case P_REPNZ:
  63         return PPS_LREP;
  64     case P_O16:
  65     case P_O32:
  66     case P_O64:
  67     case P_OSP:
  68         return PPS_OSIZE;
  69     case P_A16:
  70     case P_A32:
  71     case P_A64:
  72     case P_ASP:
  73         return PPS_ASIZE;
  74     default:
  75         error(ERR_PANIC, "Invalid value %d passed to prefix_slot()", prefix);
  76         return -1;
  77     }
  78 }
  79
  80 static void process_size_override(insn * result, int operand)
  81 {
  82     if (tasm_compatible_mode) {
  83         switch ((int)tokval.t_integer) {
  84             /* For TASM compatibility a size override inside the
  85              * brackets changes the size of the operand, not the
  86              * address type of the operand as it does in standard
  87              * NASM syntax. Hence:
  88              *
  89              *  mov     eax,[DWORD val]
  90              *
  91              * is valid syntax in TASM compatibility mode. Note that
  92              * you lose the ability to override the default address
  93              * type for the instruction, but we never use anything
  94              * but 32-bit flat model addressing in our code.
  95              */
  96         case S_BYTE:
  97             result->oprs[operand].type |= BITS8;
  98             break;
  99         case S_WORD:
 100             result->oprs[operand].type |= BITS16;
 101             break;
 102         case S_DWORD:
 103         case S_LONG:
 104             result->oprs[operand].type |= BITS32;
 105             break;
 106         case S_QWORD:
 107             result->oprs[operand].type |= BITS64;
 108             break;
 109         case S_TWORD:
 110             result->oprs[operand].type |= BITS80;
 111             break;
 112         case S_OWORD:
 113             result->oprs[operand].type |= BITS128;
 114             break;
 115         default:
 116             error(ERR_NONFATAL,
 117                   "invalid operand size specification");
 118             break;
 119         }
 120     } else {
 121         /* Standard NASM compatible syntax */
 122         switch ((int)tokval.t_integer) {
 123         case S_NOSPLIT:
 124             result->oprs[operand].eaflags |= EAF_TIMESTWO;
 125             break;
 126         case S_REL:
 127             result->oprs[operand].eaflags |= EAF_REL;
 128             break;
 129         case S_ABS:
 130             result->oprs[operand].eaflags |= EAF_ABS;
 131             break;
 132         case S_BYTE:
 133             result->oprs[operand].disp_size = 8;
 134             result->oprs[operand].eaflags |= EAF_BYTEOFFS;
 135             break;
 136         case P_A16:
 137         case P_A32:
 138         case P_A64:
 139             if (result->prefixes[PPS_ASIZE] &&
 140                 result->prefixes[PPS_ASIZE] != tokval.t_integer)
 141                 error(ERR_NONFATAL,
 142                       "conflicting address size specifications");
 143             else
 144                 result->prefixes[PPS_ASIZE] = tokval.t_integer;
 145             break;
 146         case S_WORD:
 147             result->oprs[operand].disp_size = 16;
 148             result->oprs[operand].eaflags |= EAF_WORDOFFS;
 149             break;
 150         case S_DWORD:
 151         case S_LONG:
 152             result->oprs[operand].disp_size = 32;
 153             result->oprs[operand].eaflags |= EAF_WORDOFFS;
 154             break;
 155         case S_QWORD:
 156             result->oprs[operand].disp_size = 64;
 157             result->oprs[operand].eaflags |= EAF_WORDOFFS;
 158             break;
 159         default:
 160             error(ERR_NONFATAL, "invalid size specification in"
 161                   " effective address");
 162             break;
 163         }
 164     }
 165 }
 166
 167 insn *parse_line(int pass, char *buffer, insn * result,
 168                  efunc errfunc, evalfunc evaluate, ldfunc ldef)
 169 {
 170     int operand;
 171     int critical;
 172     struct eval_hints hints;
 173     int j;
 174     bool first;
 175     bool insn_is_label = false;
 176
 177 restart_parse:
 178     first = true;
 179     result->forw_ref = false;
 180     error = errfunc;
 181
 182     stdscan_reset();
 183     stdscan_bufptr = buffer;
 184     i = stdscan(NULL, &tokval);
 185
 186     result->label = NULL;       /* Assume no label */
 187     result->eops = NULL;        /* must do this, whatever happens */
 188     result->operands = 0;       /* must initialize this */
 189
 190     if (i == 0) {               /* blank line - ignore */
 191         result->opcode = -1;    /* and no instruction either */
 192         return result;
 193     }
 194     if (i != TOKEN_ID && i != TOKEN_INSN && i != TOKEN_PREFIX &&
 195         (i != TOKEN_REG || (REG_SREG & ~reg_flags[tokval.t_integer]))) {
 196         error(ERR_NONFATAL, "label or instruction expected"
 197               " at start of line");
 198         result->opcode = -1;
 199         return result;
 200     }
 201
 202     if (i == TOKEN_ID || (insn_is_label && i == TOKEN_INSN)) {
 203         /* there's a label here */
 204         first = false;
 205         result->label = tokval.t_charptr;
 206         i = stdscan(NULL, &tokval);
 207         if (i == ':') {         /* skip over the optional colon */
 208             i = stdscan(NULL, &tokval);
 209         } else if (i == 0) {
 210             error(ERR_WARNING | ERR_WARN_OL | ERR_PASS1,
 211                   "label alone on a line without a colon might be in error");
 212         }
 213         if (i != TOKEN_INSN || tokval.t_integer != I_EQU) {
 214             /*
 215              * FIXME: location->segment could be NO_SEG, in which case
 216              * it is possible we should be passing 'abs_seg'. Look into this.
 217              * Work out whether that is *really* what we should be doing.
 218              * Generally fix things. I think this is right as it is, but
 219              * am still not certain.
 220              */
 221             ldef(result->label, in_abs_seg ? abs_seg : location->segment,
 222                  location->offset, NULL, true, false, outfmt, errfunc);
 223         }
 224     }
 225
 226     if (i == 0) {
 227         result->opcode = -1;    /* this line contains just a label */
 228         return result;
 229     }
 230
 231     for (j = 0; j < MAXPREFIX; j++)
 232         result->prefixes[j] = P_none;
 233     result->times = 1L;
 234
 235     while (i == TOKEN_PREFIX ||
 236            (i == TOKEN_REG && !(REG_SREG & ~reg_flags[tokval.t_integer])))
 237     {
 238         first = false;
 239
 240         /*
 241          * Handle special case: the TIMES prefix.
 242          */
 243         if (i == TOKEN_PREFIX && tokval.t_integer == P_TIMES) {
 244             expr *value;
 245
 246             i = stdscan(NULL, &tokval);
 247             value =
 248                 evaluate(stdscan, NULL, &tokval, NULL, pass0, error, NULL);
 249             i = tokval.t_type;
 250             if (!value) {       /* but, error in evaluator */
 251                 result->opcode = -1;    /* unrecoverable parse error: */
 252                 return result;  /* ignore this instruction */
 253             }
 254             if (!is_simple(value)) {
 255                 error(ERR_NONFATAL,
 256                       "non-constant argument supplied to TIMES");
 257                 result->times = 1L;
 258             } else {
 259                 result->times = value->value;
 260                 if (value->value < 0) {
 261                     error(ERR_NONFATAL, "TIMES value %d is negative",
 262                           value->value);
 263                     result->times = 0;
 264                 }
 265             }
 266         } else {
 267             int slot = prefix_slot(tokval.t_integer);
 268             if (result->prefixes[slot]) {
 269                if (result->prefixes[slot] == tokval.t_integer)
 270                     error(ERR_WARNING,
 271                       "instruction has redundant prefixes");
 272                else
 273                     error(ERR_NONFATAL,
 274                       "instruction has conflicting prefixes");
 275             }
 276             result->prefixes[slot] = tokval.t_integer;
 277             i = stdscan(NULL, &tokval);
 278         }
 279     }
 280
 281     if (i != TOKEN_INSN) {
 282         int j;
 283         enum prefixes pfx;
 284
 285         for (j = 0; j < MAXPREFIX; j++)
 286             if ((pfx = result->prefixes[j]) != P_none)
 287                 break;
 288
 289         if (i == 0 && pfx != P_none) {
 290             /*
 291              * Instruction prefixes are present, but no actual
 292              * instruction. This is allowed: at this point we
 293              * invent a notional instruction of RESB 0.
 294              */
 295             result->opcode = I_RESB;
 296             result->operands = 1;
 297             result->oprs[0].type = IMMEDIATE;
 298             result->oprs[0].offset = 0L;
 299             result->oprs[0].segment = result->oprs[0].wrt = NO_SEG;
 300             return result;
 301         } else {
 302             error(ERR_NONFATAL, "parser: instruction expected");
 303             result->opcode = -1;
 304             return result;
 305         }
 306     }
 307
 308     result->opcode = tokval.t_integer;
 309     result->condition = tokval.t_inttwo;
 310
 311     /*
 312      * RESB, RESW and RESD cannot be satisfied with incorrectly
 313      * evaluated operands, since the correct values _must_ be known
 314      * on the first pass. Hence, even in pass one, we set the
 315      * `critical' flag on calling evaluate(), so that it will bomb
 316      * out on undefined symbols. Nasty, but there's nothing we can
 317      * do about it.
 318      *
 319      * For the moment, EQU has the same difficulty, so we'll
 320      * include that.
 321      */
 322     if (result->opcode == I_RESB || result->opcode == I_RESW ||
 323         result->opcode == I_RESD || result->opcode == I_RESQ ||
 324         result->opcode == I_REST || result->opcode == I_RESO ||
 325         result->opcode == I_EQU || result->opcode == I_INCBIN) {
 326         critical = (pass0 < 2 ? 1 : 2);
 327
 328     } else
 329         critical = (pass == 2 ? 2 : 0);
 330
 331     if (result->opcode == I_DB || result->opcode == I_DW ||
 332         result->opcode == I_DD || result->opcode == I_DQ ||
 333         result->opcode == I_DT || result->opcode == I_DO ||
 334         result->opcode == I_INCBIN) {
 335         extop *eop, **tail = &result->eops, **fixptr;
 336         int oper_num = 0;
 337
 338         result->eops_float = false;
 339
 340         /*
 341          * Begin to read the DB/DW/DD/DQ/DT/DO/INCBIN operands.
 342          */
 343         while (1) {
 344             i = stdscan(NULL, &tokval);
 345             if (i == 0)
 346                 break;
 347             else if (first && i == ':') {
 348                 insn_is_label = true;
 349                 goto restart_parse;
 350             }
 351             first = false;
 352             fixptr = tail;
 353             eop = *tail = nasm_malloc(sizeof(extop));
 354             tail = &eop->next;
 355             eop->next = NULL;
 356             eop->type = EOT_NOTHING;
 357             oper_num++;
 358
 359             if (i == TOKEN_NUM && tokval.t_charptr && is_comma_next()) {
 360                 eop->type = EOT_DB_STRING;
 361                 eop->stringval = tokval.t_charptr;
 362                 eop->stringlen = tokval.t_inttwo;
 363                 i = stdscan(NULL, &tokval);     /* eat the comma */
 364                 continue;
 365             }
 366
 367             if ((i == TOKEN_FLOAT && is_comma_next())
 368                 || i == '-' || i == '+') {
 369                 int32_t sign = +1;
 370
 371                 if (i == '+' || i == '-') {
 372                     char *save = stdscan_bufptr;
 373                     int token = i;
 374                     sign = (i == '-') ? -1 : 1;
 375                     i = stdscan(NULL, &tokval);
 376                     if (i != TOKEN_FLOAT || !is_comma_next()) {
 377                         stdscan_bufptr = save;
 378                         i = tokval.t_type = token;
 379                     }
 380                 }
 381
 382                 if (i == TOKEN_FLOAT) {
 383                     eop->type = EOT_DB_STRING;
 384                     result->eops_float = true;
 385                     switch (result->opcode) {
 386                     case I_DB:
 387                         eop->stringlen = 1;
 388                         break;
 389                     case I_DW:
 390                         eop->stringlen = 2;
 391                         break;
 392                     case I_DD:
 393                         eop->stringlen = 4;
 394                         break;
 395                     case I_DQ:
 396                         eop->stringlen = 8;
 397                         break;
 398                     case I_DT:
 399                         eop->stringlen = 10;
 400                         break;
 401                     case I_DO:
 402                         eop->stringlen = 16;
 403                         break;
 404                     default:
 405                         error(ERR_NONFATAL, "floating-point constant"
 406                               " encountered in unknown instruction");
 407                         /*
 408                          * fix suggested by Pedro Gimeno... original line
 409                          * was:
 410                          * eop->type = EOT_NOTHING;
 411                          */
 412                         eop->stringlen = 0;
 413                         break;
 414                     }
 415                     eop = nasm_realloc(eop, sizeof(extop) + eop->stringlen);
 416                     tail = &eop->next;
 417                     *fixptr = eop;
 418                     eop->stringval = (char *)eop + sizeof(extop);
 419                     if (!eop->stringlen ||
 420                         !float_const(tokval.t_charptr, sign,
 421                                      (uint8_t *)eop->stringval,
 422                                      eop->stringlen, error))
 423                         eop->type = EOT_NOTHING;
 424                     i = stdscan(NULL, &tokval); /* eat the comma */
 425                     continue;
 426                 }
 427             }
 428
 429             /* anything else */
 430             {
 431                 expr *value;
 432                 value = evaluate(stdscan, NULL, &tokval, NULL,
 433                                  critical, error, NULL);
 434                 i = tokval.t_type;
 435                 if (!value) {   /* error in evaluator */
 436                     result->opcode = -1;        /* unrecoverable parse error: */
 437                     return result;      /* ignore this instruction */
 438                 }
 439                 if (is_unknown(value)) {
 440                     eop->type = EOT_DB_NUMBER;
 441                     eop->offset = 0;    /* doesn't matter what we put */
 442                     eop->segment = eop->wrt = NO_SEG;   /* likewise */
 443                 } else if (is_reloc(value)) {
 444                     eop->type = EOT_DB_NUMBER;
 445                     eop->offset = reloc_value(value);
 446                     eop->segment = reloc_seg(value);
 447                     eop->wrt = reloc_wrt(value);
 448                 } else {
 449                     error(ERR_NONFATAL,
 450                           "operand %d: expression is not simple"
 451                           " or relocatable", oper_num);
 452                 }
 453             }
 454
 455             /*
 456              * We're about to call stdscan(), which will eat the
 457              * comma that we're currently sitting on between
 458              * arguments. However, we'd better check first that it
 459              * _is_ a comma.
 460              */
 461             if (i == 0)         /* also could be EOL */
 462                 break;
 463             if (i != ',') {
 464                 error(ERR_NONFATAL, "comma expected after operand %d",
 465                       oper_num);
 466                 result->opcode = -1;    /* unrecoverable parse error: */
 467                 return result;  /* ignore this instruction */
 468             }
 469         }
 470
 471         if (result->opcode == I_INCBIN) {
 472             /*
 473              * Correct syntax for INCBIN is that there should be
 474              * one string operand, followed by one or two numeric
 475              * operands.
 476              */
 477             if (!result->eops || result->eops->type != EOT_DB_STRING)
 478                 error(ERR_NONFATAL, "`incbin' expects a file name");
 479             else if (result->eops->next &&
 480                      result->eops->next->type != EOT_DB_NUMBER)
 481                 error(ERR_NONFATAL, "`incbin': second parameter is",
 482                       " non-numeric");
 483             else if (result->eops->next && result->eops->next->next &&
 484                      result->eops->next->next->type != EOT_DB_NUMBER)
 485                 error(ERR_NONFATAL, "`incbin': third parameter is",
 486                       " non-numeric");
 487             else if (result->eops->next && result->eops->next->next &&
 488                      result->eops->next->next->next)
 489                 error(ERR_NONFATAL,
 490                       "`incbin': more than three parameters");
 491             else
 492                 return result;
 493             /*
 494              * If we reach here, one of the above errors happened.
 495              * Throw the instruction away.
 496              */
 497             result->opcode = -1;
 498             return result;
 499         } else /* DB ... */ if (oper_num == 0)
 500             error(ERR_WARNING | ERR_PASS1,
 501                   "no operand for data declaration");
 502         else
 503             result->operands = oper_num;
 504
 505         return result;
 506     }
 507
 508     /* right. Now we begin to parse the operands. There may be up to four
 509      * of these, separated by commas, and terminated by a zero token. */
 510
 511     for (operand = 0; operand < MAX_OPERANDS; operand++) {
 512         expr *value;            /* used most of the time */
 513         int mref;               /* is this going to be a memory ref? */
 514         int bracket;            /* is it a [] mref, or a & mref? */
 515         int setsize = 0;
 516
 517         result->oprs[operand].disp_size = 0;    /* have to zero this whatever */
 518         result->oprs[operand].eaflags = 0;      /* and this */
 519         result->oprs[operand].opflags = 0;
 520
 521         i = stdscan(NULL, &tokval);
 522         if (i == 0)
 523             break;              /* end of operands: get out of here */
 524         else if (first && i == ':') {
 525             insn_is_label = true;
 526             goto restart_parse;
 527         }
 528         first = false;
 529         result->oprs[operand].type = 0; /* so far, no override */
 530         while (i == TOKEN_SPECIAL) {    /* size specifiers */
 531             switch ((int)tokval.t_integer) {
 532             case S_BYTE:
 533                 if (!setsize)   /* we want to use only the first */
 534                     result->oprs[operand].type |= BITS8;
 535                 setsize = 1;
 536                 break;
 537             case S_WORD:
 538                 if (!setsize)
 539                     result->oprs[operand].type |= BITS16;
 540                 setsize = 1;
 541                 break;
 542             case S_DWORD:
 543             case S_LONG:
 544                 if (!setsize)
 545                     result->oprs[operand].type |= BITS32;
 546                 setsize = 1;
 547                 break;
 548             case S_QWORD:
 549                 if (!setsize)
 550                     result->oprs[operand].type |= BITS64;
 551                 setsize = 1;
 552                 break;
 553             case S_TWORD:
 554                 if (!setsize)
 555                     result->oprs[operand].type |= BITS80;
 556                 setsize = 1;
 557                 break;
 558             case S_OWORD:
 559                 if (!setsize)
 560                     result->oprs[operand].type |= BITS128;
 561                 setsize = 1;
 562                 break;
 563             case S_TO:
 564                 result->oprs[operand].type |= TO;
 565                 break;
 566             case S_STRICT:
 567                 result->oprs[operand].type |= STRICT;
 568                 break;
 569             case S_FAR:
 570                 result->oprs[operand].type |= FAR;
 571                 break;
 572             case S_NEAR:
 573                 result->oprs[operand].type |= NEAR;
 574                 break;
 575             case S_SHORT:
 576                 result->oprs[operand].type |= SHORT;
 577                 break;
 578             default:
 579                 error(ERR_NONFATAL, "invalid operand size specification");
 580             }
 581             i = stdscan(NULL, &tokval);
 582         }
 583
 584         if (i == '[' || i == '&') {     /* memory reference */
 585             mref = true;
 586             bracket = (i == '[');
 587             i = stdscan(NULL, &tokval); /* then skip the colon */
 588             while (i == TOKEN_SPECIAL || i == TOKEN_PREFIX) {
 589                 process_size_override(result, operand);
 590                 i = stdscan(NULL, &tokval);
 591             }
 592         } else {                /* immediate operand, or register */
 593             mref = false;
 594             bracket = false;    /* placate optimisers */
 595         }
 596
 597         if ((result->oprs[operand].type & FAR) && !mref &&
 598             result->opcode != I_JMP && result->opcode != I_CALL) {
 599             error(ERR_NONFATAL, "invalid use of FAR operand specifier");
 600         }
 601
 602         value = evaluate(stdscan, NULL, &tokval,
 603                          &result->oprs[operand].opflags,
 604                          critical, error, &hints);
 605         i = tokval.t_type;
 606         if (result->oprs[operand].opflags & OPFLAG_FORWARD) {
 607             result->forw_ref = true;
 608         }
 609         if (!value) {           /* error in evaluator */
 610             result->opcode = -1;        /* unrecoverable parse error: */
 611             return result;      /* ignore this instruction */
 612         }
 613         if (i == ':' && mref) { /* it was seg:offset */
 614             /*
 615              * Process the segment override.
 616              */
 617             if (value[1].type != 0 || value->value != 1 ||
 618                 REG_SREG & ~reg_flags[value->type])
 619                 error(ERR_NONFATAL, "invalid segment override");
 620             else if (result->prefixes[PPS_SEG])
 621                 error(ERR_NONFATAL,
 622                       "instruction has conflicting segment overrides");
 623             else {
 624                 result->prefixes[PPS_SEG] = value->type;
 625                 if (!(REG_FSGS & ~reg_flags[value->type]))
 626                     result->oprs[operand].eaflags |= EAF_FSGS;
 627             }
 628
 629             i = stdscan(NULL, &tokval); /* then skip the colon */
 630             while (i == TOKEN_SPECIAL || i == TOKEN_PREFIX) {
 631                 process_size_override(result, operand);
 632                 i = stdscan(NULL, &tokval);
 633             }
 634             value = evaluate(stdscan, NULL, &tokval,
 635                              &result->oprs[operand].opflags,
 636                              critical, error, &hints);
 637             i = tokval.t_type;
 638             if (result->oprs[operand].opflags & OPFLAG_FORWARD) {
 639                 result->forw_ref = true;
 640             }
 641             /* and get the offset */
 642             if (!value) {       /* but, error in evaluator */
 643                 result->opcode = -1;    /* unrecoverable parse error: */
 644                 return result;  /* ignore this instruction */
 645             }
 646         }
 647         if (mref && bracket) {  /* find ] at the end */
 648             if (i != ']') {
 649                 error(ERR_NONFATAL, "parser: expecting ]");
 650                 do {            /* error recovery again */
 651                     i = stdscan(NULL, &tokval);
 652                 } while (i != 0 && i != ',');
 653             } else              /* we got the required ] */
 654                 i = stdscan(NULL, &tokval);
 655         } else {                /* immediate operand */
 656             if (i != 0 && i != ',' && i != ':') {
 657                 error(ERR_NONFATAL, "comma or end of line expected");
 658                 do {            /* error recovery */
 659                     i = stdscan(NULL, &tokval);
 660                 } while (i != 0 && i != ',');
 661             } else if (i == ':') {
 662                 result->oprs[operand].type |= COLON;
 663             }
 664         }
 665
 666         /* now convert the exprs returned from evaluate() into operand
 667          * descriptions... */
 668
 669         if (mref) {             /* it's a memory reference */
 670             expr *e = value;
 671             int b, i, s;        /* basereg, indexreg, scale */
 672             int64_t o;             /* offset */
 673
 674             b = i = -1, o = s = 0;
 675             result->oprs[operand].hintbase = hints.base;
 676             result->oprs[operand].hinttype = hints.type;
 677
 678             if (e->type && e->type <= EXPR_REG_END) {   /* this bit's a register */
 679                 if (e->value == 1)      /* in fact it can be basereg */
 680                     b = e->type;
 681                 else            /* no, it has to be indexreg */
 682                     i = e->type, s = e->value;
 683                 e++;
 684             }
 685             if (e->type && e->type <= EXPR_REG_END) {   /* it's a 2nd register */
 686                 if (b != -1)    /* If the first was the base, ... */
 687                     i = e->type, s = e->value;  /* second has to be indexreg */
 688
 689                 else if (e->value != 1) {       /* If both want to be index */
 690                     error(ERR_NONFATAL,
 691                           "beroset-p-592-invalid effective address");
 692                     result->opcode = -1;
 693                     return result;
 694                 } else
 695                     b = e->type;
 696                 e++;
 697             }
 698             if (e->type != 0) { /* is there an offset? */
 699                 if (e->type <= EXPR_REG_END) {  /* in fact, is there an error? */
 700                     error(ERR_NONFATAL,
 701                           "beroset-p-603-invalid effective address");
 702                     result->opcode = -1;
 703                     return result;
 704                 } else {
 705                     if (e->type == EXPR_UNKNOWN) {
 706                         o = 0;  /* doesn't matter what */
 707                         result->oprs[operand].wrt = NO_SEG;     /* nor this */
 708                         result->oprs[operand].segment = NO_SEG; /* or this */
 709                         while (e->type)
 710                             e++;        /* go to the end of the line */
 711                     } else {
 712                         if (e->type == EXPR_SIMPLE) {
 713                             o = e->value;
 714                             e++;
 715                         }
 716                         if (e->type == EXPR_WRT) {
 717                             result->oprs[operand].wrt = e->value;
 718                             e++;
 719                         } else
 720                             result->oprs[operand].wrt = NO_SEG;
 721                         /*
 722                          * Look for a segment base type.
 723                          */
 724                         if (e->type && e->type < EXPR_SEGBASE) {
 725                             error(ERR_NONFATAL,
 726                                   "beroset-p-630-invalid effective address");
 727                             result->opcode = -1;
 728                             return result;
 729                         }
 730                         while (e->type && e->value == 0)
 731                             e++;
 732                         if (e->type && e->value != 1) {
 733                             error(ERR_NONFATAL,
 734                                   "beroset-p-637-invalid effective address");
 735                             result->opcode = -1;
 736                             return result;
 737                         }
 738                         if (e->type) {
 739                             result->oprs[operand].segment =
 740                                 e->type - EXPR_SEGBASE;
 741                             e++;
 742                         } else
 743                             result->oprs[operand].segment = NO_SEG;
 744                         while (e->type && e->value == 0)
 745                             e++;
 746                         if (e->type) {
 747                             error(ERR_NONFATAL,
 748                                   "beroset-p-650-invalid effective address");
 749                             result->opcode = -1;
 750                             return result;
 751                         }
 752                     }
 753                 }
 754             } else {
 755                 o = 0;
 756                 result->oprs[operand].wrt = NO_SEG;
 757                 result->oprs[operand].segment = NO_SEG;
 758             }
 759
 760             if (e->type != 0) { /* there'd better be nothing left! */
 761                 error(ERR_NONFATAL,
 762                       "beroset-p-663-invalid effective address");
 763                 result->opcode = -1;
 764                 return result;
 765             }
 766
 767             /* It is memory, but it can match any r/m operand */
 768             result->oprs[operand].type |= MEMORY_ANY;
 769
 770             if (b == -1 && (i == -1 || s == 0)) {
 771                 int is_rel = globalbits == 64 &&
 772                     !(result->oprs[operand].eaflags & EAF_ABS) &&
 773                     ((globalrel &&
 774                       !(result->oprs[operand].eaflags & EAF_FSGS)) ||
 775                      (result->oprs[operand].eaflags & EAF_REL));
 776
 777                 result->oprs[operand].type |= is_rel ? IP_REL : MEM_OFFS;
 778             }
 779             result->oprs[operand].basereg = b;
 780             result->oprs[operand].indexreg = i;
 781             result->oprs[operand].scale = s;
 782             result->oprs[operand].offset = o;
 783         } else {                /* it's not a memory reference */
 784
 785             if (is_just_unknown(value)) {       /* it's immediate but unknown */
 786                 result->oprs[operand].type |= IMMEDIATE;
 787                 result->oprs[operand].offset = 0;       /* don't care */
 788                 result->oprs[operand].segment = NO_SEG; /* don't care again */
 789                 result->oprs[operand].wrt = NO_SEG;     /* still don't care */
 790             } else if (is_reloc(value)) {       /* it's immediate */
 791                 result->oprs[operand].type |= IMMEDIATE;
 792                 result->oprs[operand].offset = reloc_value(value);
 793                 result->oprs[operand].segment = reloc_seg(value);
 794                 result->oprs[operand].wrt = reloc_wrt(value);
 795                 if (is_simple(value)) {
 796                     if (reloc_value(value) == 1)
 797                         result->oprs[operand].type |= UNITY;
 798                     if (optimizing >= 0 &&
 799                         !(result->oprs[operand].type & STRICT)) {
 800                         int64_t v64 = reloc_value(value);
 801                         int32_t v32 = (int32_t)v64;
 802                         int16_t v16 = (int16_t)v32;
 803
 804                         if (v64 >= -128 && v64 <= 127)
 805                             result->oprs[operand].type |= SBYTE64;
 806                         if (v32 >= -128 && v32 <= 127)
 807                             result->oprs[operand].type |= SBYTE32;
 808                         if (v16 >= -128 && v16 <= 127)
 809                             result->oprs[operand].type |= SBYTE16;
 810                     }
 811                 }
 812             } else {            /* it's a register */
 813                 unsigned int rs;
 814
 815                 if (value->type >= EXPR_SIMPLE || value->value != 1) {
 816                     error(ERR_NONFATAL, "invalid operand type");
 817                     result->opcode = -1;
 818                     return result;
 819                 }
 820
 821                 /*
 822                  * check that its only 1 register, not an expression...
 823                  */
 824                 for (i = 1; value[i].type; i++)
 825                     if (value[i].value) {
 826                         error(ERR_NONFATAL, "invalid operand type");
 827                         result->opcode = -1;
 828                         return result;
 829                     }
 830
 831                 /* clear overrides, except TO which applies to FPU regs */
 832                 if (result->oprs[operand].type & ~TO) {
 833                     /*
 834                      * we want to produce a warning iff the specified size
 835                      * is different from the register size
 836                      */
 837                     rs = result->oprs[operand].type & SIZE_MASK;
 838                 } else
 839                     rs = 0;
 840
 841                 result->oprs[operand].type &= TO;
 842                 result->oprs[operand].type |= REGISTER;
 843                 result->oprs[operand].type |= reg_flags[value->type];
 844                 result->oprs[operand].basereg = value->type;
 845
 846                 if (rs && (result->oprs[operand].type & SIZE_MASK) != rs)
 847                     error(ERR_WARNING | ERR_PASS1,
 848                           "register size specification ignored");
 849             }
 850         }
 851     }
 852
 853     result->operands = operand; /* set operand count */
 854
 855 /* clear remaining operands */
 856 while (operand < MAX_OPERANDS)
 857     result->oprs[operand++].type = 0;
 858
 859     /*
 860      * Transform RESW, RESD, RESQ, REST, RESO into RESB.
 861      */
 862     switch (result->opcode) {
 863     case I_RESW:
 864         result->opcode = I_RESB;
 865         result->oprs[0].offset *= 2;
 866         break;
 867     case I_RESD:
 868         result->opcode = I_RESB;
 869         result->oprs[0].offset *= 4;
 870         break;
 871     case I_RESQ:
 872         result->opcode = I_RESB;
 873         result->oprs[0].offset *= 8;
 874         break;
 875     case I_REST:
 876         result->opcode = I_RESB;
 877         result->oprs[0].offset *= 10;
 878         break;
 879     case I_RESO:
 880         result->opcode = I_RESB;
 881         result->oprs[0].offset *= 16;
 882         break;
 883     default:
 884         break;
 885     }
 886
 887     return result;
 888 }
 889
 890 static int is_comma_next(void)
 891 {
 892     char *p;
 893     int i;
 894     struct tokenval tv;
 895
 896     p = stdscan_bufptr;
 897     i = stdscan(NULL, &tv);
 898     stdscan_bufptr = p;
 899     return (i == ',' || i == ';' || !i);
 900 }
 901
 902 void cleanup_insn(insn * i)
 903 {
 904     extop *e;
 905
 906     while (i->eops) {
 907         e = i->eops;
 908         i->eops = i->eops->next;
 909         nasm_free(e);
 910     }
 911 }