scan.l revision 17142
1%{ 2/* $NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $ */ 3 4/* 5 * Copyright (c) 1994, 1995 Jochen Pohl 6 * All Rights Reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by Jochen Pohl for 19 * The NetBSD Project. 20 * 4. The name of the author may not be used to endorse or promote products 21 * derived from this software without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 24 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 25 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 26 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 27 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 28 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 32 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35#ifndef lint 36static char rcsid[] = "$NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $"; 37#endif 38 39#include <stdlib.h> 40#include <string.h> 41#include <limits.h> 42#include <float.h> 43#include <ctype.h> 44#include <errno.h> 45#include <err.h> 46#include <math.h> 47 48#include "lint1.h" 49#include "y.tab.h" 50 51#define CHAR_MASK (~(~0 << CHAR_BIT)) 52 53/* XXX declaration of strtouq() is missing in stdlib.h ? */ 54extern u_quad_t strtouq __P((const char *, char **, int)); 55 56/* Current position (its also updated when an included file is parsed) */ 57pos_t curr_pos = { 1, "" }; 58 59/* 60 * Current position in C source (not updated when an included file is 61 * parsed). 62 */ 63pos_t csrc_pos = { 1, "" }; 64 65static void incline __P((void)); 66static void badchar __P((int)); 67static sbuf_t *allocsb __P((void)); 68static void freesb __P((sbuf_t *)); 69static int inpc __P((void)); 70static int hash __P((const char *)); 71static sym_t *search __P((sbuf_t *)); 72static int name __P((void)); 73static int keyw __P((sym_t *)); 74static int icon __P((int)); 75static int fcon __P((void)); 76static int operator __P((int, op_t)); 77static int ccon __P((void)); 78static int wccon __P((void)); 79static int getescc __P((int)); 80static void directive __P((void)); 81static void comment __P((void)); 82static int string __P((void)); 83static int wcstrg __P((void)); 84 85%} 86 87L [_A-Za-z] 88D [0-9] 89NZD [1-9] 90OD [0-7] 91HD [0-9A-Fa-f] 92EX ([eE][+-]?[0-9]+) 93 94%% 95 96{L}({L}|{D})* return (name()); 970{OD}*[lLuU]* return (icon(8)); 98{NZD}{D}*[lLuU]* return (icon(10)); 990[xX]{HD}+[lLuU]* return (icon(16)); 100{D}+\.{D}*{EX}?[fFlL]? | 101{D}+{EX}[fFlL]? | 102\.{D}+{EX}?[fFlL]? return (fcon()); 103"=" return (operator(T_ASSIGN, ASSIGN)); 104"*=" return (operator(T_OPASS, MULASS)); 105"/=" return (operator(T_OPASS, DIVASS)); 106"%=" return (operator(T_OPASS, MODASS)); 107"+=" return (operator(T_OPASS, ADDASS)); 108"-=" return (operator(T_OPASS, SUBASS)); 109"<<=" return (operator(T_OPASS, SHLASS)); 110">>=" return (operator(T_OPASS, SHRASS)); 111"&=" return (operator(T_OPASS, ANDASS)); 112"^=" return (operator(T_OPASS, XORASS)); 113"|=" return (operator(T_OPASS, ORASS)); 114"||" return (operator(T_LOGOR, LOGOR)); 115"&&" return (operator(T_LOGAND, LOGAND)); 116"|" return (operator(T_OR, OR)); 117"&" return (operator(T_AND, AND)); 118"^" return (operator(T_XOR, XOR)); 119"==" return (operator(T_EQOP, EQ)); 120"!=" return (operator(T_EQOP, NE)); 121"<" return (operator(T_RELOP, LT)); 122">" return (operator(T_RELOP, GT)); 123"<=" return (operator(T_RELOP, LE)); 124">=" return (operator(T_RELOP, GE)); 125"<<" return (operator(T_SHFTOP, SHL)); 126">>" return (operator(T_SHFTOP, SHR)); 127"++" return (operator(T_INCDEC, INC)); 128"--" return (operator(T_INCDEC, DEC)); 129"->" return (operator(T_STROP, ARROW)); 130"." return (operator(T_STROP, POINT)); 131"+" return (operator(T_ADDOP, PLUS)); 132"-" return (operator(T_ADDOP, MINUS)); 133"*" return (operator(T_MULT, MULT)); 134"/" return (operator(T_DIVOP, DIV)); 135"%" return (operator(T_DIVOP, MOD)); 136"!" return (operator(T_UNOP, NOT)); 137"~" return (operator(T_UNOP, COMPL)); 138"\"" return (string()); 139"L\"" return (wcstrg()); 140";" return (T_SEMI); 141"{" return (T_LBRACE); 142"}" return (T_RBRACE); 143"," return (T_COMMA); 144":" return (T_COLON); 145"?" return (T_QUEST); 146"[" return (T_LBRACK); 147"]" return (T_RBRACK); 148"(" return (T_LPARN); 149")" return (T_RPARN); 150"..." return (T_ELLIPSE); 151"'" return (ccon()); 152"L'" return (wccon()); 153^#.*$ directive(); 154\n incline(); 155\t|" "|\f|\v ; 156"/*" comment(); 157. badchar(yytext[0]); 158 159%% 160 161static void 162incline() 163{ 164 curr_pos.p_line++; 165 if (curr_pos.p_file == csrc_pos.p_file) 166 csrc_pos.p_line++; 167} 168 169static void 170badchar(c) 171 int c; 172{ 173 /* unknown character \%o */ 174 error(250, c); 175} 176 177/* 178 * Keywords. 179 * During initialisation they are written to the symbol table. 180 */ 181static struct kwtab { 182 const char *kw_name; /* keyword */ 183 int kw_token; /* token returned by yylex() */ 184 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 185 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 186 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 187 u_int kw_stdc : 1; /* STDC keyword */ 188 u_int kw_gcc : 1; /* GCC keyword */ 189} kwtab[] = { 190 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 191 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 192 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 193 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 194 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 195 { "case", T_CASE, 0, 0, 0, 0, 0 }, 196 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 197 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 198 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 199 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 200 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 201 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 202 { "do", T_DO, 0, 0, 0, 0, 0 }, 203 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 204 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 205 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 206 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 207 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 208 { "for", T_FOR, 0, 0, 0, 0, 0 }, 209 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 210 { "if", T_IF, 0, 0, 0, 0, 0 }, 211 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 212 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 213 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 214 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 215 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 216 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 217 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 218 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 219 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 220 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 221 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 222 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 223 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 224 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 225 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 226 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 227 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 228 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 229 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 230 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 231 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 232 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 233 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 234 { NULL, 0, 0, 0, 0, 0, 0 } 235}; 236 237/* Symbol table */ 238static sym_t *symtab[HSHSIZ1]; 239 240/* bit i of the entry with index i is set */ 241u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 242 243/* least significant i bits are set in the entry with index i */ 244u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 245 246/* least significant i bits are not set in the entry with index i */ 247u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 248 249/* free list for sbuf structures */ 250static sbuf_t *sbfrlst; 251 252/* Typ of next expected symbol */ 253symt_t symtyp; 254 255 256/* 257 * All keywords are written to the symbol table. This saves us looking 258 * in a extra table for each name we found. 259 */ 260void 261initscan() 262{ 263 struct kwtab *kw; 264 sym_t *sym; 265 int h, i; 266 u_quad_t uq; 267 268 for (kw = kwtab; kw->kw_name != NULL; kw++) { 269 if (kw->kw_stdc && tflag) 270 continue; 271 if (kw->kw_gcc && !gflag) 272 continue; 273 sym = getblk(sizeof (sym_t)); 274 sym->s_name = kw->kw_name; 275 sym->s_keyw = 1; 276 sym->s_value.v_quad = kw->kw_token; 277 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 278 sym->s_tspec = kw->kw_tspec; 279 } else if (kw->kw_token == T_SCLASS) { 280 sym->s_scl = kw->kw_scl; 281 } else if (kw->kw_token == T_QUAL) { 282 sym->s_tqual = kw->kw_tqual; 283 } 284 h = hash(sym->s_name); 285 if ((sym->s_link = symtab[h]) != NULL) 286 symtab[h]->s_rlink = &sym->s_link; 287 (symtab[h] = sym)->s_rlink = &symtab[h]; 288 } 289 290 /* initialize bit-masks for quads */ 291 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 292 qbmasks[i] = (u_quad_t)1 << i; 293 uq = ~(u_quad_t)0 << i; 294 qumasks[i] = uq; 295 qlmasks[i] = ~uq; 296 } 297 qumasks[i] = 0; 298 qlmasks[i] = ~(u_quad_t)0; 299} 300 301/* 302 * Get a free sbuf structure, if possible from the free list 303 */ 304static sbuf_t * 305allocsb() 306{ 307 sbuf_t *sb; 308 309 if ((sb = sbfrlst) != NULL) { 310 sbfrlst = sb->sb_nxt; 311 } else { 312 sb = xmalloc(sizeof (sbuf_t)); 313 } 314 (void)memset(sb, 0, sizeof (sb)); 315 return (sb); 316} 317 318/* 319 * Put a sbuf structure to the free list 320 */ 321static void 322freesb(sb) 323 sbuf_t *sb; 324{ 325 sb->sb_nxt = sbfrlst; 326 sbfrlst = sb; 327} 328 329/* 330 * Read a character and ensure that it is positive (except EOF). 331 * Increment line count(s) if necessary. 332 */ 333static int 334inpc() 335{ 336 int c; 337 338 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 339 incline(); 340 return (c); 341} 342 343static int 344hash(s) 345 const char *s; 346{ 347 u_int v; 348 const u_char *us; 349 350 v = 0; 351 for (us = (const u_char *)s; *us != '\0'; us++) { 352 v = (v << sizeof (v)) + *us; 353 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 354 } 355 return (v % HSHSIZ1); 356} 357 358/* 359 * Lex has found a letter followed by zero or more letters or digits. 360 * It looks for a symbol in the symbol table with the same name. This 361 * symbol must either be a keyword or a symbol of the type required by 362 * symtyp (label, member, tag, ...). 363 * 364 * If it is a keyword, the token is returned. In some cases it is described 365 * more deeply by data written to yylval. 366 * 367 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 368 * is stored in yylval. This struct contains the name of the symbol, it's 369 * length and hash value. If there is already a symbol of the same name 370 * and type in the symbol table, the sbuf struct also contains a pointer 371 * to the symbol table entry. 372 */ 373static int 374name() 375{ 376 char *s; 377 sbuf_t *sb; 378 sym_t *sym; 379 int tok; 380 381 sb = allocsb(); 382 sb->sb_name = yytext; 383 sb->sb_len = yyleng; 384 sb->sb_hash = hash(yytext); 385 386 if ((sym = search(sb)) != NULL && sym->s_keyw) { 387 freesb(sb); 388 return (keyw(sym)); 389 } 390 391 sb->sb_sym = sym; 392 393 if (sym != NULL) { 394 if (blklev < sym->s_blklev) 395 lerror("name() 1"); 396 sb->sb_name = sym->s_name; 397 sb->sb_len = strlen(sym->s_name); 398 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 399 } else { 400 s = getblk(yyleng + 1); 401 (void)memcpy(s, yytext, yyleng + 1); 402 sb->sb_name = s; 403 sb->sb_len = yyleng; 404 tok = T_NAME; 405 } 406 407 yylval.y_sb = sb; 408 return (tok); 409} 410 411static sym_t * 412search(sb) 413 sbuf_t *sb; 414{ 415 sym_t *sym; 416 417 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 418 if (strcmp(sym->s_name, sb->sb_name) == 0) { 419 if (sym->s_keyw || sym->s_kind == symtyp) 420 return (sym); 421 } 422 } 423 424 return (NULL); 425} 426 427static int 428keyw(sym) 429 sym_t *sym; 430{ 431 int t; 432 433 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 434 yylval.y_scl = sym->s_scl; 435 } else if (t == T_TYPE || t == T_SOU) { 436 yylval.y_tspec = sym->s_tspec; 437 } else if (t == T_QUAL) { 438 yylval.y_tqual = sym->s_tqual; 439 } 440 return (t); 441} 442 443/* 444 * Convert a string representing an integer into internal representation. 445 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 446 */ 447static int 448icon(base) 449 int base; 450{ 451 int l_suffix, u_suffix; 452 int len; 453 const char *cp; 454 char c, *eptr; 455 tspec_t typ; 456 u_long ul; 457 u_quad_t uq; 458 int ansiu; 459 static tspec_t contypes[2][3] = { 460 { INT, LONG, QUAD }, 461 { UINT, ULONG, UQUAD } 462 }; 463 464 cp = yytext; 465 len = yyleng; 466 467 /* skip 0x */ 468 if (base == 16) { 469 cp += 2; 470 len -= 2; 471 } 472 473 /* read suffixes */ 474 l_suffix = u_suffix = 0; 475 for ( ; ; ) { 476 if ((c = cp[len - 1]) == 'l' || c == 'L') { 477 l_suffix++; 478 } else if (c == 'u' || c == 'U') { 479 u_suffix++; 480 } else { 481 break; 482 } 483 len--; 484 } 485 if (l_suffix > 2 || u_suffix > 1) { 486 /* malformed integer constant */ 487 warning(251); 488 if (l_suffix > 2) 489 l_suffix = 2; 490 if (u_suffix > 1) 491 u_suffix = 1; 492 } 493 if (tflag && u_suffix != 0) { 494 /* suffix U is illegal in traditional C */ 495 warning(97); 496 } 497 typ = contypes[u_suffix][l_suffix]; 498 499 errno = 0; 500 if (l_suffix < 2) { 501 ul = strtoul(cp, &eptr, base); 502 } else { 503 uq = strtouq(cp, &eptr, base); 504 } 505 if (eptr != cp + len) 506 lerror("icon() 1"); 507 if (errno != 0) 508 /* integer constant out of range */ 509 warning(252); 510 511 /* 512 * If the value is to big for the current type, we must choose 513 * another type. 514 */ 515 ansiu = 0; 516 switch (typ) { 517 case INT: 518 if (ul <= INT_MAX) { 519 /* ok */ 520 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 521 typ = UINT; 522 } else if (ul <= LONG_MAX) { 523 typ = LONG; 524 } else { 525 typ = ULONG; 526 } 527 if (typ == UINT || typ == ULONG) { 528 if (tflag) { 529 typ = LONG; 530 } else if (!sflag) { 531 /* 532 * Remember that the constant is unsigned 533 * only in ANSI C 534 */ 535 ansiu = 1; 536 } 537 } 538 break; 539 case UINT: 540 if (ul > (u_int)UINT_MAX) 541 typ = ULONG; 542 break; 543 case LONG: 544 if (ul > LONG_MAX && !tflag) { 545 typ = ULONG; 546 if (!sflag) 547 ansiu = 1; 548 } 549 break; 550 case QUAD: 551 if (uq > QUAD_MAX && !tflag) { 552 typ = UQUAD; 553 if (!sflag) 554 ansiu = 1; 555 } 556 break; 557 /* LINTED (enumeration values not handled in switch) */ 558 default: 559 } 560 561 if (typ != QUAD && typ != UQUAD) { 562 if (isutyp(typ)) { 563 uq = ul; 564 } else { 565 uq = (quad_t)(long)ul; 566 } 567 } 568 569 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 570 571 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 572 yylval.y_val->v_ansiu = ansiu; 573 yylval.y_val->v_quad = (quad_t)uq; 574 575 return (T_CON); 576} 577 578/* 579 * Returns 1 if t is a signed type and the value is negative. 580 * 581 * len is the number of significant bits. If len is -1, len is set 582 * to the width of type t. 583 */ 584int 585sign(q, t, len) 586 quad_t q; 587 tspec_t t; 588 int len; 589{ 590 if (t == PTR || isutyp(t)) 591 return (0); 592 return (msb(q, t, len)); 593} 594 595int 596msb(q, t, len) 597 quad_t q; 598 tspec_t t; 599 int len; 600{ 601 if (len <= 0) 602 len = size(t); 603 return ((q & qbmasks[len - 1]) != 0); 604} 605 606/* 607 * Extends the sign of q. 608 */ 609quad_t 610xsign(q, t, len) 611 quad_t q; 612 tspec_t t; 613 int len; 614{ 615 if (len <= 0) 616 len = size(t); 617 618 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 619 q &= qlmasks[len]; 620 } else { 621 q |= qumasks[len]; 622 } 623 return (q); 624} 625 626/* 627 * Convert a string representing a floating point value into its interal 628 * representation. Type and value are returned in yylval. fcon() 629 * (and yylex()) returns T_CON. 630 * XXX Currently it is not possible to convert constants of type 631 * long double which are greater then DBL_MAX. 632 */ 633static int 634fcon() 635{ 636 const char *cp; 637 int len; 638 tspec_t typ; 639 char c, *eptr; 640 double d; 641 float f; 642 643 cp = yytext; 644 len = yyleng; 645 646 if ((c = cp[len - 1]) == 'f' || c == 'F') { 647 typ = FLOAT; 648 len--; 649 } else if (c == 'l' || c == 'L') { 650 typ = LDOUBLE; 651 len--; 652 } else { 653 typ = DOUBLE; 654 } 655 656 if (tflag && typ != DOUBLE) { 657 /* suffixes F and L are illegal in traditional C */ 658 warning(98); 659 } 660 661 errno = 0; 662 d = strtod(cp, &eptr); 663 if (eptr != cp + len) 664 lerror("fcon() 1"); 665 if (errno != 0) 666 /* floating-point constant out of range */ 667 warning(248); 668 669 if (typ == FLOAT) { 670 f = (float)d; 671 if (isinf(f)) { 672 /* floating-point constant out of range */ 673 warning(248); 674 f = f > 0 ? FLT_MAX : -FLT_MAX; 675 } 676 } 677 678 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 679 if (typ == FLOAT) { 680 yylval.y_val->v_ldbl = f; 681 } else { 682 yylval.y_val->v_ldbl = d; 683 } 684 685 return (T_CON); 686} 687 688static int 689operator(t, o) 690 int t; 691 op_t o; 692{ 693 yylval.y_op = o; 694 return (t); 695} 696 697/* 698 * Called if lex found a leading \'. 699 */ 700static int 701ccon() 702{ 703 int n, val, c; 704 char cv; 705 706 n = 0; 707 val = 0; 708 while ((c = getescc('\'')) >= 0) { 709 val = (val << CHAR_BIT) + c; 710 n++; 711 } 712 if (c == -2) { 713 /* unterminated character constant */ 714 error(253); 715 } else { 716 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 717 /* too many characters in character constant */ 718 error(71); 719 } else if (n > 1) { 720 /* multi-character character constant */ 721 warning(294); 722 } else if (n == 0) { 723 /* empty character constant */ 724 error(73); 725 } 726 } 727 if (n == 1) { 728 cv = (char)val; 729 val = cv; 730 } 731 732 yylval.y_val = xcalloc(1, sizeof (val_t)); 733 yylval.y_val->v_tspec = INT; 734 yylval.y_val->v_quad = val; 735 736 return (T_CON); 737} 738 739/* 740 * Called if lex found a leading L\' 741 */ 742static int 743wccon() 744{ 745 static char buf[MB_LEN_MAX + 1]; 746 int i, c; 747 wchar_t wc; 748 749 i = 0; 750 while ((c = getescc('\'')) >= 0) { 751 if (i < MB_CUR_MAX) 752 buf[i] = (char)c; 753 i++; 754 } 755 756 wc = 0; 757 758 if (c == -2) { 759 /* unterminated character constant */ 760 error(253); 761 } else if (c == 0) { 762 /* empty character constant */ 763 error(73); 764 } else { 765 if (i > MB_CUR_MAX) { 766 i = MB_CUR_MAX; 767 /* too many characters in character constant */ 768 error(71); 769 } else { 770 buf[i] = '\0'; 771 (void)mbtowc(NULL, NULL, 0); 772 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 773 /* invalid multibyte character */ 774 error(291); 775 } 776 } 777 778 yylval.y_val = xcalloc(1, sizeof (val_t)); 779 yylval.y_val->v_tspec = WCHAR; 780 yylval.y_val->v_quad = wc; 781 782 return (T_CON); 783} 784 785/* 786 * Read a character which is part of a character constant or of a string 787 * and handle escapes. 788 * 789 * The Argument is the character which delimits the character constant or 790 * string. 791 * 792 * Returns -1 if the end of the character constant or string is reached, 793 * -2 if the EOF is reached, and the charachter otherwise. 794 */ 795static int 796getescc(d) 797 int d; 798{ 799 static int pbc = -1; 800 int n, c, v; 801 802 if (pbc == -1) { 803 c = inpc(); 804 } else { 805 c = pbc; 806 pbc = -1; 807 } 808 if (c == d) 809 return (-1); 810 switch (c) { 811 case '\n': 812 /* newline in string or char constant */ 813 error(254); 814 return (-2); 815 case EOF: 816 return (-2); 817 case '\\': 818 switch (c = inpc()) { 819 case '"': 820 if (tflag && d == '\'') 821 /* \" inside character constant undef. ... */ 822 warning(262); 823 return ('"'); 824 case '\'': 825 return ('\''); 826 case '?': 827 if (tflag) 828 /* \? undefined in traditional C */ 829 warning(263); 830 return ('?'); 831 case '\\': 832 return ('\\'); 833 case 'a': 834 if (tflag) 835 /* \a undefined in traditional C */ 836 warning(81); 837#ifdef __STDC__ 838 return ('\a'); 839#else 840 return ('\007'); 841#endif 842 case 'b': 843 return ('\b'); 844 case 'f': 845 return ('\f'); 846 case 'n': 847 return ('\n'); 848 case 'r': 849 return ('\r'); 850 case 't': 851 return ('\t'); 852 case 'v': 853 if (tflag) 854 /* \v undefined in traditional C */ 855 warning(264); 856#ifdef __STDC__ 857 return ('\v'); 858#else 859 return ('\013'); 860#endif 861 case '8': case '9': 862 /* bad octal digit %c */ 863 warning(77, c); 864 /* FALLTHROUGH */ 865 case '0': case '1': case '2': case '3': 866 case '4': case '5': case '6': case '7': 867 n = 3; 868 v = 0; 869 do { 870 v = (v << 3) + (c - '0'); 871 c = inpc(); 872 } while (--n && isdigit(c) && (tflag || c <= '7')); 873 if (tflag && n > 0 && isdigit(c)) 874 /* bad octal digit %c */ 875 warning(77, c); 876 pbc = c; 877 if (v > UCHAR_MAX) { 878 /* character escape does not fit in char. */ 879 warning(76); 880 v &= CHAR_MASK; 881 } 882 return (v); 883 case 'x': 884 if (tflag) 885 /* \x undefined in traditional C */ 886 warning(82); 887 v = 0; 888 n = 0; 889 while ((c = inpc()) >= 0 && isxdigit(c)) { 890 c = isdigit(c) ? 891 c - '0' : toupper(c) - 'A' + 10; 892 v = (v << 4) + c; 893 if (n >= 0) { 894 if ((v & ~CHAR_MASK) != 0) { 895 /* overflow in hex escape */ 896 warning(75); 897 n = -1; 898 } else { 899 n++; 900 } 901 } 902 } 903 pbc = c; 904 if (n == 0) { 905 /* no hex digits follow \x */ 906 error(74); 907 } if (n == -1) { 908 v &= CHAR_MASK; 909 } 910 return (v); 911 case '\n': 912 return (getescc(d)); 913 case EOF: 914 return (-2); 915 default: 916 if (isprint(c)) { 917 /* dubious escape \%c */ 918 warning(79, c); 919 } else { 920 /* dubious escape \%o */ 921 warning(80, c); 922 } 923 } 924 } 925 return (c); 926} 927 928/* 929 * Called for preprocessor directives. Currently implemented are: 930 * # lineno 931 * # lineno "filename" 932 */ 933static void 934directive() 935{ 936 const char *cp, *fn; 937 char c, *eptr; 938 size_t fnl; 939 long ln; 940 static int first = 1; 941 942 /* Go to first non-whitespace after # */ 943 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 944 945 if (!isdigit(c)) { 946 error: 947 /* undefined or invalid # directive */ 948 warning(255); 949 return; 950 } 951 ln = strtol(--cp, &eptr, 10); 952 if (cp == eptr) 953 goto error; 954 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 955 goto error; 956 while ((c = *cp++) == ' ' || c == '\t') ; 957 if (c != '\0') { 958 if (c != '"') 959 goto error; 960 fn = cp; 961 while ((c = *cp) != '"' && c != '\0') 962 cp++; 963 if (c != '"') 964 goto error; 965 if ((fnl = cp++ - fn) > PATH_MAX) 966 goto error; 967 while ((c = *cp++) == ' ' || c == '\t') ; 968#if 0 969 if (c != '\0') 970 warning("extra character(s) after directive"); 971#endif 972 curr_pos.p_file = fnnalloc(fn, fnl); 973 /* 974 * If this is the first directive, the name is the name 975 * of the C source file as specified at the command line. 976 * It is written to the output file. 977 */ 978 if (first) { 979 csrc_pos.p_file = curr_pos.p_file; 980 outsrc(curr_pos.p_file); 981 first = 0; 982 } 983 } 984 curr_pos.p_line = (int)ln - 1; 985 if (curr_pos.p_file == csrc_pos.p_file) 986 csrc_pos.p_line = (int)ln - 1; 987} 988 989/* 990 * Handle lint comments. Following comments are currently understood: 991 * ARGSUSEDn 992 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 993 * FALLTHRU FALLTHROUGH 994 * LINTLIBRARY 995 * LINTED NOSTRICT 996 * LONGLONG 997 * NOTREACHED 998 * PRINTFLIKEn 999 * PROTOLIB 1000 * SCANFLIKEn 1001 * VARARGSn 1002 * If one of this comments is recognized, the arguments, if any, are 1003 * parsed and a function which handles this comment is called. 1004 */ 1005static void 1006comment() 1007{ 1008 int c, lc; 1009 static struct { 1010 const char *keywd; 1011 int arg; 1012 void (*func) __P((int)); 1013 } keywtab[] = { 1014 { "ARGSUSED", 1, argsused }, 1015 { "CONSTCOND", 0, constcond }, 1016 { "CONSTANTCOND", 0, constcond }, 1017 { "CONSTANTCONDITION", 0, constcond }, 1018 { "FALLTHRU", 0, fallthru }, 1019 { "FALLTHROUGH", 0, fallthru }, 1020 { "LINTLIBRARY", 0, lintlib }, 1021 { "LINTED", 0, linted }, 1022 { "LONGLONG", 0, longlong }, 1023 { "NOSTRICT", 0, linted }, 1024 { "NOTREACHED", 0, notreach }, 1025 { "PRINTFLIKE", 1, printflike }, 1026 { "PROTOLIB", 1, protolib }, 1027 { "SCANFLIKE", 1, scanflike }, 1028 { "VARARGS", 1, varargs }, 1029 }; 1030 char keywd[32]; 1031 char arg[32]; 1032 int l, i, a; 1033 int eoc; 1034 1035 eoc = 0; 1036 1037 /* Skip white spaces after the start of the comment */ 1038 while ((c = inpc()) != EOF && isspace(c)) ; 1039 1040 /* Read the potential keyword to keywd */ 1041 l = 0; 1042 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1043 keywd[l++] = (char)c; 1044 c = inpc(); 1045 } 1046 keywd[l] = '\0'; 1047 1048 /* look for the keyword */ 1049 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1050 if (strcmp(keywtab[i].keywd, keywd) == 0) 1051 break; 1052 } 1053 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1054 goto skip_rest; 1055 1056 /* skip white spaces after the keyword */ 1057 while (c != EOF && isspace(c)) 1058 c = inpc(); 1059 1060 /* read the argument, if the keyword accepts one and there is one */ 1061 l = 0; 1062 if (keywtab[i].arg) { 1063 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1064 arg[l++] = (char)c; 1065 c = inpc(); 1066 } 1067 } 1068 arg[l] = '\0'; 1069 a = l != 0 ? atoi(arg) : -1; 1070 1071 /* skip white spaces after the argument */ 1072 while (c != EOF && isspace(c)) 1073 c = inpc(); 1074 1075 if (c != '*' || (c = inpc()) != '/') { 1076 if (keywtab[i].func != linted) 1077 /* extra characters in lint comment */ 1078 warning(257); 1079 } else { 1080 /* 1081 * remember that we have already found the end of the 1082 * comment 1083 */ 1084 eoc = 1; 1085 } 1086 1087 if (keywtab[i].func != NULL) 1088 (*keywtab[i].func)(a); 1089 1090 skip_rest: 1091 while (!eoc) { 1092 lc = c; 1093 if ((c = inpc()) == EOF) { 1094 /* unterminated comment */ 1095 error(256); 1096 break; 1097 } 1098 if (lc == '*' && c == '/') 1099 eoc = 1; 1100 } 1101} 1102 1103/* 1104 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1105 * clrwflgs() is called after function definitions and global and 1106 * local declarations and definitions. It is also called between 1107 * the controlling expression and the body of control statements 1108 * (if, switch, for, while). 1109 */ 1110void 1111clrwflgs() 1112{ 1113 nowarn = 0; 1114 quadflg = 0; 1115 ccflg = 0; 1116} 1117 1118/* 1119 * Strings are stored in a dynamically alloceted buffer and passed 1120 * in yylval.y_xstrg to the parser. The parser or the routines called 1121 * by the parser are responsible for freeing this buffer. 1122 */ 1123static int 1124string() 1125{ 1126 u_char *s; 1127 int c; 1128 size_t len, max; 1129 strg_t *strg; 1130 1131 s = xmalloc(max = 64); 1132 1133 len = 0; 1134 while ((c = getescc('"')) >= 0) { 1135 /* +1 to reserve space for a trailing NUL character */ 1136 if (len + 1 == max) 1137 s = xrealloc(s, max *= 2); 1138 s[len++] = (char)c; 1139 } 1140 s[len] = '\0'; 1141 if (c == -2) 1142 /* unterminated string constant */ 1143 error(258); 1144 1145 strg = xcalloc(1, sizeof (strg_t)); 1146 strg->st_tspec = CHAR; 1147 strg->st_len = len; 1148 strg->st_cp = s; 1149 1150 yylval.y_strg = strg; 1151 return (T_STRING); 1152} 1153 1154static int 1155wcstrg() 1156{ 1157 char *s; 1158 int c, i, n, wi; 1159 size_t len, max, wlen; 1160 wchar_t *ws; 1161 strg_t *strg; 1162 1163 s = xmalloc(max = 64); 1164 len = 0; 1165 while ((c = getescc('"')) >= 0) { 1166 /* +1 to save space for a trailing NUL character */ 1167 if (len + 1 >= max) 1168 s = xrealloc(s, max *= 2); 1169 s[len++] = (char)c; 1170 } 1171 s[len] = '\0'; 1172 if (c == -2) 1173 /* unterminated string constant */ 1174 error(258); 1175 1176 /* get length of wide character string */ 1177 (void)mblen(NULL, 0); 1178 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1179 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1180 /* invalid multibyte character */ 1181 error(291); 1182 break; 1183 } 1184 if (n == 0) 1185 n = 1; 1186 } 1187 1188 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1189 1190 /* convert from multibyte to wide char */ 1191 (void)mbtowc(NULL, NULL, 0); 1192 for (i = 0, wi = 0; i < len; i += n, wi++) { 1193 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1194 break; 1195 if (n == 0) 1196 n = 1; 1197 } 1198 ws[wi] = 0; 1199 free(s); 1200 1201 strg = xcalloc(1, sizeof (strg_t)); 1202 strg->st_tspec = WCHAR; 1203 strg->st_len = wlen; 1204 strg->st_wcp = ws; 1205 1206 yylval.y_strg = strg; 1207 return (T_STRING); 1208} 1209 1210/* 1211 * As noted above the scanner does not create new symbol table entries 1212 * for symbols it cannot find in the symbol table. This is to avoid 1213 * putting undeclared symbols into the symbol table if a syntax error 1214 * occurs. 1215 * 1216 * getsym() is called as soon as it is probably ok to put the symbol to 1217 * the symbol table. This does not mean that it is not possible that 1218 * symbols are put to the symbol table which are than not completely 1219 * declared due to syntax errors. To avoid too many problems in this 1220 * case symbols get type int in getsym(). 1221 * 1222 * XXX calls to getsym() should be delayed until decl1*() is called 1223 */ 1224sym_t * 1225getsym(sb) 1226 sbuf_t *sb; 1227{ 1228 dinfo_t *di; 1229 char *s; 1230 sym_t *sym; 1231 1232 sym = sb->sb_sym; 1233 1234 /* 1235 * During member declaration it is possible that name() looked 1236 * for symbols of type FVFT, although it should have looked for 1237 * symbols of type FTAG. Same can happen for labels. Both cases 1238 * are compensated here. 1239 */ 1240 if (symtyp == FMOS || symtyp == FLAB) { 1241 if (sym == NULL || sym->s_kind == FVFT) 1242 sym = search(sb); 1243 } 1244 1245 if (sym != NULL) { 1246 if (sym->s_kind != symtyp) 1247 lerror("storesym() 1"); 1248 symtyp = FVFT; 1249 freesb(sb); 1250 return (sym); 1251 } 1252 1253 /* create a new symbol table entry */ 1254 1255 /* labels must always be allocated at level 1 (outhermost block) */ 1256 if (symtyp == FLAB) { 1257 sym = getlblk(1, sizeof (sym_t)); 1258 s = getlblk(1, sb->sb_len + 1); 1259 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1260 sym->s_name = s; 1261 sym->s_blklev = 1; 1262 di = dcs; 1263 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1264 di = di->d_nxt; 1265 if (di->d_ctx != AUTO) 1266 lerror("storesym() 2"); 1267 } else { 1268 sym = getblk(sizeof (sym_t)); 1269 sym->s_name = sb->sb_name; 1270 sym->s_blklev = blklev; 1271 di = dcs; 1272 } 1273 1274 STRUCT_ASSIGN(sym->s_dpos, curr_pos); 1275 if ((sym->s_kind = symtyp) != FLAB) 1276 sym->s_type = gettyp(INT); 1277 1278 symtyp = FVFT; 1279 1280 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1281 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1282 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1283 1284 *di->d_ldlsym = sym; 1285 di->d_ldlsym = &sym->s_dlnxt; 1286 1287 freesb(sb); 1288 return (sym); 1289} 1290 1291/* 1292 * Remove a symbol forever from the symbol table. s_blklev 1293 * is set to -1 to avoid that the symbol will later be put 1294 * back to the symbol table. 1295 */ 1296void 1297rmsym(sym) 1298 sym_t *sym; 1299{ 1300 if ((*sym->s_rlink = sym->s_link) != NULL) 1301 sym->s_link->s_rlink = sym->s_rlink; 1302 sym->s_blklev = -1; 1303 sym->s_link = NULL; 1304} 1305 1306/* 1307 * Remove a list of symbols declared at one level from the symbol 1308 * table. 1309 */ 1310void 1311rmsyms(syms) 1312 sym_t *syms; 1313{ 1314 sym_t *sym; 1315 1316 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1317 if (sym->s_blklev != -1) { 1318 if ((*sym->s_rlink = sym->s_link) != NULL) 1319 sym->s_link->s_rlink = sym->s_rlink; 1320 sym->s_link = NULL; 1321 sym->s_rlink = NULL; 1322 } 1323 } 1324} 1325 1326/* 1327 * Put a symbol into the symbol table 1328 */ 1329void 1330inssym(bl, sym) 1331 int bl; 1332 sym_t *sym; 1333{ 1334 int h; 1335 1336 h = hash(sym->s_name); 1337 if ((sym->s_link = symtab[h]) != NULL) 1338 symtab[h]->s_rlink = &sym->s_link; 1339 (symtab[h] = sym)->s_rlink = &symtab[h]; 1340 sym->s_blklev = bl; 1341 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1342 lerror("inssym()"); 1343} 1344 1345/* 1346 * Called at level 0 after syntax errors 1347 * Removes all symbols which are not declared at level 0 from the 1348 * symbol table. Also frees all memory which is not associated with 1349 * level 0. 1350 */ 1351void 1352cleanup() 1353{ 1354 sym_t *sym, *nsym; 1355 int i; 1356 1357 for (i = 0; i < HSHSIZ1; i++) { 1358 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1359 nsym = sym->s_link; 1360 if (sym->s_blklev >= 1) { 1361 if ((*sym->s_rlink = nsym) != NULL) 1362 nsym->s_rlink = sym->s_rlink; 1363 } 1364 } 1365 } 1366 1367 for (i = mblklev; i > 0; i--) 1368 freelblk(i); 1369} 1370 1371/* 1372 * Create a new symbol with the name of an existing symbol. 1373 */ 1374sym_t * 1375pushdown(sym) 1376 sym_t *sym; 1377{ 1378 int h; 1379 sym_t *nsym; 1380 1381 h = hash(sym->s_name); 1382 nsym = getblk(sizeof (sym_t)); 1383 if (sym->s_blklev > blklev) 1384 lerror("pushdown()"); 1385 nsym->s_name = sym->s_name; 1386 STRUCT_ASSIGN(nsym->s_dpos, curr_pos); 1387 nsym->s_kind = sym->s_kind; 1388 nsym->s_blklev = blklev; 1389 1390 if ((nsym->s_link = symtab[h]) != NULL) 1391 symtab[h]->s_rlink = &nsym->s_link; 1392 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1393 1394 *dcs->d_ldlsym = nsym; 1395 dcs->d_ldlsym = &nsym->s_dlnxt; 1396 1397 return (nsym); 1398} 1399 1400/* 1401 * Free any dynamically allocated memory referenced by 1402 * the value stack or yylval. 1403 * The type of information in yylval is described by tok. 1404 */ 1405void 1406freeyyv(sp, tok) 1407 void *sp; 1408 int tok; 1409{ 1410 if (tok == T_NAME || tok == T_TYPENAME) { 1411 sbuf_t *sb = *(sbuf_t **)sp; 1412 freesb(sb); 1413 } else if (tok == T_CON) { 1414 val_t *val = *(val_t **)sp; 1415 free(val); 1416 } else if (tok == T_STRING) { 1417 strg_t *strg = *(strg_t **)sp; 1418 if (strg->st_tspec == CHAR) { 1419 free(strg->st_cp); 1420 } else if (strg->st_tspec == WCHAR) { 1421 free(strg->st_wcp); 1422 } else { 1423 lerror("fryylv() 1"); 1424 } 1425 free(strg); 1426 } 1427} 1428