scan.l revision 12099
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 <math.h> 46#include <err.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 } 559 560 if (typ != QUAD && typ != UQUAD) { 561 if (isutyp(typ)) { 562 uq = ul; 563 } else { 564 uq = (quad_t)(long)ul; 565 } 566 } 567 568 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 569 570 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 571 yylval.y_val->v_ansiu = ansiu; 572 yylval.y_val->v_quad = (quad_t)uq; 573 574 return (T_CON); 575} 576 577/* 578 * Returns 1 if t is a signed type and the value is negative. 579 * 580 * len is the number of significant bits. If len is -1, len is set 581 * to the width of type t. 582 */ 583int 584sign(q, t, len) 585 quad_t q; 586 tspec_t t; 587 int len; 588{ 589 if (t == PTR || isutyp(t)) 590 return (0); 591 return (msb(q, t, len)); 592} 593 594int 595msb(q, t, len) 596 quad_t q; 597 tspec_t t; 598 int len; 599{ 600 if (len <= 0) 601 len = size(t); 602 return ((q & qbmasks[len - 1]) != 0); 603} 604 605/* 606 * Extends the sign of q. 607 */ 608quad_t 609xsign(q, t, len) 610 quad_t q; 611 tspec_t t; 612 int len; 613{ 614 if (len <= 0) 615 len = size(t); 616 617 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 618 q &= qlmasks[len]; 619 } else { 620 q |= qumasks[len]; 621 } 622 return (q); 623} 624 625/* 626 * Convert a string representing a floating point value into its interal 627 * representation. Type and value are returned in yylval. fcon() 628 * (and yylex()) returns T_CON. 629 * XXX Currently it is not possible to convert constants of type 630 * long double which are greater then DBL_MAX. 631 */ 632static int 633fcon() 634{ 635 const char *cp; 636 int len; 637 tspec_t typ; 638 char c, *eptr; 639 double d; 640 float f; 641 642 cp = yytext; 643 len = yyleng; 644 645 if ((c = cp[len - 1]) == 'f' || c == 'F') { 646 typ = FLOAT; 647 len--; 648 } else if (c == 'l' || c == 'L') { 649 typ = LDOUBLE; 650 len--; 651 } else { 652 typ = DOUBLE; 653 } 654 655 if (tflag && typ != DOUBLE) { 656 /* suffixes F and L are illegal in traditional C */ 657 warning(98); 658 } 659 660 errno = 0; 661 d = strtod(cp, &eptr); 662 if (eptr != cp + len) 663 lerror("fcon() 1"); 664 if (errno != 0) 665 /* floating-point constant out of range */ 666 warning(248); 667 668 if (typ == FLOAT) { 669 f = (float)d; 670 if (isinf(f)) { 671 /* floating-point constant out of range */ 672 warning(248); 673 f = f > 0 ? FLT_MAX : -FLT_MAX; 674 } 675 } 676 677 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 678 if (typ == FLOAT) { 679 yylval.y_val->v_ldbl = f; 680 } else { 681 yylval.y_val->v_ldbl = d; 682 } 683 684 return (T_CON); 685} 686 687static int 688operator(t, o) 689 int t; 690 op_t o; 691{ 692 yylval.y_op = o; 693 return (t); 694} 695 696/* 697 * Called if lex found a leading \'. 698 */ 699static int 700ccon() 701{ 702 int n, val, c; 703 char cv; 704 705 n = 0; 706 val = 0; 707 while ((c = getescc('\'')) >= 0) { 708 val = (val << CHAR_BIT) + c; 709 n++; 710 } 711 if (c == -2) { 712 /* unterminated character constant */ 713 error(253); 714 } else { 715 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 716 /* too many characters in character constant */ 717 error(71); 718 } else if (n > 1) { 719 /* multi-character character constant */ 720 warning(294); 721 } else if (n == 0) { 722 /* empty character constant */ 723 error(73); 724 } 725 } 726 if (n == 1) { 727 cv = (char)val; 728 val = cv; 729 } 730 731 yylval.y_val = xcalloc(1, sizeof (val_t)); 732 yylval.y_val->v_tspec = INT; 733 yylval.y_val->v_quad = val; 734 735 return (T_CON); 736} 737 738/* 739 * Called if lex found a leading L\' 740 */ 741static int 742wccon() 743{ 744 static char buf[MB_LEN_MAX + 1]; 745 int i, c; 746 wchar_t wc; 747 748 i = 0; 749 while ((c = getescc('\'')) >= 0) { 750 if (i < MB_CUR_MAX) 751 buf[i] = (char)c; 752 i++; 753 } 754 755 wc = 0; 756 757 if (c == -2) { 758 /* unterminated character constant */ 759 error(253); 760 } else if (c == 0) { 761 /* empty character constant */ 762 error(73); 763 } else { 764 if (i > MB_CUR_MAX) { 765 i = MB_CUR_MAX; 766 /* too many characters in character constant */ 767 error(71); 768 } else { 769 buf[i] = '\0'; 770 (void)mbtowc(NULL, NULL, 0); 771 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 772 /* invalid multibyte character */ 773 error(291); 774 } 775 } 776 777 yylval.y_val = xcalloc(1, sizeof (val_t)); 778 yylval.y_val->v_tspec = WCHAR; 779 yylval.y_val->v_quad = wc; 780 781 return (T_CON); 782} 783 784/* 785 * Read a character which is part of a character constant or of a string 786 * and handle escapes. 787 * 788 * The Argument is the character which delimits the character constant or 789 * string. 790 * 791 * Returns -1 if the end of the character constant or string is reached, 792 * -2 if the EOF is reached, and the charachter otherwise. 793 */ 794static int 795getescc(d) 796 int d; 797{ 798 static int pbc = -1; 799 int n, c, v; 800 801 if (pbc == -1) { 802 c = inpc(); 803 } else { 804 c = pbc; 805 pbc = -1; 806 } 807 if (c == d) 808 return (-1); 809 switch (c) { 810 case '\n': 811 /* newline in string or char constant */ 812 error(254); 813 return (-2); 814 case EOF: 815 return (-2); 816 case '\\': 817 switch (c = inpc()) { 818 case '"': 819 if (tflag && d == '\'') 820 /* \" inside character constant undef. ... */ 821 warning(262); 822 return ('"'); 823 case '\'': 824 return ('\''); 825 case '?': 826 if (tflag) 827 /* \? undefined in traditional C */ 828 warning(263); 829 return ('?'); 830 case '\\': 831 return ('\\'); 832 case 'a': 833 if (tflag) 834 /* \a undefined in traditional C */ 835 warning(81); 836#ifdef __STDC__ 837 return ('\a'); 838#else 839 return ('\007'); 840#endif 841 case 'b': 842 return ('\b'); 843 case 'f': 844 return ('\f'); 845 case 'n': 846 return ('\n'); 847 case 'r': 848 return ('\r'); 849 case 't': 850 return ('\t'); 851 case 'v': 852 if (tflag) 853 /* \v undefined in traditional C */ 854 warning(264); 855#ifdef __STDC__ 856 return ('\v'); 857#else 858 return ('\013'); 859#endif 860 case '8': case '9': 861 /* bad octal digit %c */ 862 warning(77, c); 863 /* FALLTHROUGH */ 864 case '0': case '1': case '2': case '3': 865 case '4': case '5': case '6': case '7': 866 n = 3; 867 v = 0; 868 do { 869 v = (v << 3) + (c - '0'); 870 c = inpc(); 871 } while (--n && isdigit(c) && (tflag || c <= '7')); 872 if (tflag && n > 0 && isdigit(c)) 873 /* bad octal digit %c */ 874 warning(77, c); 875 pbc = c; 876 if (v > UCHAR_MAX) { 877 /* character escape does not fit in char. */ 878 warning(76); 879 v &= CHAR_MASK; 880 } 881 return (v); 882 case 'x': 883 if (tflag) 884 /* \x undefined in traditional C */ 885 warning(82); 886 v = 0; 887 n = 0; 888 while ((c = inpc()) >= 0 && isxdigit(c)) { 889 c = isdigit(c) ? 890 c - '0' : toupper(c) - 'A' + 10; 891 v = (v << 4) + c; 892 if (n >= 0) { 893 if ((v & ~CHAR_MASK) != 0) { 894 /* overflow in hex escape */ 895 warning(75); 896 n = -1; 897 } else { 898 n++; 899 } 900 } 901 } 902 pbc = c; 903 if (n == 0) { 904 /* no hex digits follow \x */ 905 error(74); 906 } if (n == -1) { 907 v &= CHAR_MASK; 908 } 909 return (v); 910 case '\n': 911 return (getescc(d)); 912 case EOF: 913 return (-2); 914 default: 915 if (isprint(c)) { 916 /* dubious escape \%c */ 917 warning(79, c); 918 } else { 919 /* dubious escape \%o */ 920 warning(80, c); 921 } 922 } 923 } 924 return (c); 925} 926 927/* 928 * Called for preprocessor directives. Currently implemented are: 929 * # lineno 930 * # lineno "filename" 931 */ 932static void 933directive() 934{ 935 const char *cp, *fn; 936 char c, *eptr; 937 size_t fnl; 938 long ln; 939 static int first = 1; 940 941 /* Go to first non-whitespace after # */ 942 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 943 944 if (!isdigit(c)) { 945 error: 946 /* undefined or invalid # directive */ 947 warning(255); 948 return; 949 } 950 ln = strtol(--cp, &eptr, 10); 951 if (cp == eptr) 952 goto error; 953 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 954 goto error; 955 while ((c = *cp++) == ' ' || c == '\t') ; 956 if (c != '\0') { 957 if (c != '"') 958 goto error; 959 fn = cp; 960 while ((c = *cp) != '"' && c != '\0') 961 cp++; 962 if (c != '"') 963 goto error; 964 if ((fnl = cp++ - fn) > PATH_MAX) 965 goto error; 966 while ((c = *cp++) == ' ' || c == '\t') ; 967#if 0 968 if (c != '\0') 969 warning("extra character(s) after directive"); 970#endif 971 curr_pos.p_file = fnnalloc(fn, fnl); 972 /* 973 * If this is the first directive, the name is the name 974 * of the C source file as specified at the command line. 975 * It is written to the output file. 976 */ 977 if (first) { 978 csrc_pos.p_file = curr_pos.p_file; 979 outsrc(curr_pos.p_file); 980 first = 0; 981 } 982 } 983 curr_pos.p_line = (int)ln - 1; 984 if (curr_pos.p_file == csrc_pos.p_file) 985 csrc_pos.p_line = (int)ln - 1; 986} 987 988/* 989 * Handle lint comments. Following comments are currently understood: 990 * ARGSUSEDn 991 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 992 * FALLTHRU FALLTHROUGH 993 * LINTLIBRARY 994 * LINTED NOSTRICT 995 * LONGLONG 996 * NOTREACHED 997 * PRINTFLIKEn 998 * PROTOLIB 999 * SCANFLIKEn 1000 * VARARGSn 1001 * If one of this comments is recognized, the arguments, if any, are 1002 * parsed and a function which handles this comment is called. 1003 */ 1004static void 1005comment() 1006{ 1007 int c, lc; 1008 static struct { 1009 const char *keywd; 1010 int arg; 1011 void (*func) __P((int)); 1012 } keywtab[] = { 1013 { "ARGSUSED", 1, argsused }, 1014 { "CONSTCOND", 0, constcond }, 1015 { "CONSTANTCOND", 0, constcond }, 1016 { "CONSTANTCONDITION", 0, constcond }, 1017 { "FALLTHRU", 0, fallthru }, 1018 { "FALLTHROUGH", 0, fallthru }, 1019 { "LINTLIBRARY", 0, lintlib }, 1020 { "LINTED", 0, linted }, 1021 { "LONGLONG", 0, longlong }, 1022 { "NOSTRICT", 0, linted }, 1023 { "NOTREACHED", 0, notreach }, 1024 { "PRINTFLIKE", 1, printflike }, 1025 { "PROTOLIB", 1, protolib }, 1026 { "SCANFLIKE", 1, scanflike }, 1027 { "VARARGS", 1, varargs }, 1028 }; 1029 char keywd[32]; 1030 char arg[32]; 1031 int l, i, a; 1032 int eoc; 1033 1034 eoc = 0; 1035 1036 /* Skip white spaces after the start of the comment */ 1037 while ((c = inpc()) != EOF && isspace(c)) ; 1038 1039 /* Read the potential keyword to keywd */ 1040 l = 0; 1041 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1042 keywd[l++] = (char)c; 1043 c = inpc(); 1044 } 1045 keywd[l] = '\0'; 1046 1047 /* look for the keyword */ 1048 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1049 if (strcmp(keywtab[i].keywd, keywd) == 0) 1050 break; 1051 } 1052 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1053 goto skip_rest; 1054 1055 /* skip white spaces after the keyword */ 1056 while (c != EOF && isspace(c)) 1057 c = inpc(); 1058 1059 /* read the argument, if the keyword accepts one and there is one */ 1060 l = 0; 1061 if (keywtab[i].arg) { 1062 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1063 arg[l++] = (char)c; 1064 c = inpc(); 1065 } 1066 } 1067 arg[l] = '\0'; 1068 a = l != 0 ? atoi(arg) : -1; 1069 1070 /* skip white spaces after the argument */ 1071 while (c != EOF && isspace(c)) 1072 c = inpc(); 1073 1074 if (c != '*' || (c = inpc()) != '/') { 1075 if (keywtab[i].func != linted) 1076 /* extra characters in lint comment */ 1077 warning(257); 1078 } else { 1079 /* 1080 * remember that we have already found the end of the 1081 * comment 1082 */ 1083 eoc = 1; 1084 } 1085 1086 if (keywtab[i].func != NULL) 1087 (*keywtab[i].func)(a); 1088 1089 skip_rest: 1090 while (!eoc) { 1091 lc = c; 1092 if ((c = inpc()) == EOF) { 1093 /* unterminated comment */ 1094 error(256); 1095 break; 1096 } 1097 if (lc == '*' && c == '/') 1098 eoc = 1; 1099 } 1100} 1101 1102/* 1103 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1104 * clrwflgs() is called after function definitions and global and 1105 * local declarations and definitions. It is also called between 1106 * the controlling expression and the body of control statements 1107 * (if, switch, for, while). 1108 */ 1109void 1110clrwflgs() 1111{ 1112 nowarn = 0; 1113 quadflg = 0; 1114 ccflg = 0; 1115} 1116 1117/* 1118 * Strings are stored in a dynamically alloceted buffer and passed 1119 * in yylval.y_xstrg to the parser. The parser or the routines called 1120 * by the parser are responsible for freeing this buffer. 1121 */ 1122static int 1123string() 1124{ 1125 u_char *s; 1126 int c; 1127 size_t len, max; 1128 strg_t *strg; 1129 1130 s = xmalloc(max = 64); 1131 1132 len = 0; 1133 while ((c = getescc('"')) >= 0) { 1134 /* +1 to reserve space for a trailing NUL character */ 1135 if (len + 1 == max) 1136 s = xrealloc(s, max *= 2); 1137 s[len++] = (char)c; 1138 } 1139 s[len] = '\0'; 1140 if (c == -2) 1141 /* unterminated string constant */ 1142 error(258); 1143 1144 strg = xcalloc(1, sizeof (strg_t)); 1145 strg->st_tspec = CHAR; 1146 strg->st_len = len; 1147 strg->st_cp = s; 1148 1149 yylval.y_strg = strg; 1150 return (T_STRING); 1151} 1152 1153static int 1154wcstrg() 1155{ 1156 char *s; 1157 int c, i, n, wi; 1158 size_t len, max, wlen; 1159 wchar_t *ws; 1160 strg_t *strg; 1161 1162 s = xmalloc(max = 64); 1163 len = 0; 1164 while ((c = getescc('"')) >= 0) { 1165 /* +1 to save space for a trailing NUL character */ 1166 if (len + 1 >= max) 1167 s = xrealloc(s, max *= 2); 1168 s[len++] = (char)c; 1169 } 1170 s[len] = '\0'; 1171 if (c == -2) 1172 /* unterminated string constant */ 1173 error(258); 1174 1175 /* get length of wide character string */ 1176 (void)mblen(NULL, 0); 1177 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1178 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1179 /* invalid multibyte character */ 1180 error(291); 1181 break; 1182 } 1183 if (n == 0) 1184 n = 1; 1185 } 1186 1187 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1188 1189 /* convert from multibyte to wide char */ 1190 (void)mbtowc(NULL, NULL, 0); 1191 for (i = 0, wi = 0; i < len; i += n, wi++) { 1192 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1193 break; 1194 if (n == 0) 1195 n = 1; 1196 } 1197 ws[wi] = 0; 1198 free(s); 1199 1200 strg = xcalloc(1, sizeof (strg_t)); 1201 strg->st_tspec = WCHAR; 1202 strg->st_len = wlen; 1203 strg->st_wcp = ws; 1204 1205 yylval.y_strg = strg; 1206 return (T_STRING); 1207} 1208 1209/* 1210 * As noted above the scanner does not create new symbol table entries 1211 * for symbols it cannot find in the symbol table. This is to avoid 1212 * putting undeclared symbols into the symbol table if a syntax error 1213 * occurs. 1214 * 1215 * getsym() is called as soon as it is probably ok to put the symbol to 1216 * the symbol table. This does not mean that it is not possible that 1217 * symbols are put to the symbol table which are than not completely 1218 * declared due to syntax errors. To avoid too many problems in this 1219 * case symbols get type int in getsym(). 1220 * 1221 * XXX calls to getsym() should be delayed until decl1*() is called 1222 */ 1223sym_t * 1224getsym(sb) 1225 sbuf_t *sb; 1226{ 1227 dinfo_t *di; 1228 char *s; 1229 sym_t *sym; 1230 1231 sym = sb->sb_sym; 1232 1233 /* 1234 * During member declaration it is possible that name() looked 1235 * for symbols of type FVFT, although it should have looked for 1236 * symbols of type FTAG. Same can happen for labels. Both cases 1237 * are compensated here. 1238 */ 1239 if (symtyp == FMOS || symtyp == FLAB) { 1240 if (sym == NULL || sym->s_kind == FVFT) 1241 sym = search(sb); 1242 } 1243 1244 if (sym != NULL) { 1245 if (sym->s_kind != symtyp) 1246 lerror("storesym() 1"); 1247 symtyp = FVFT; 1248 freesb(sb); 1249 return (sym); 1250 } 1251 1252 /* create a new symbol table entry */ 1253 1254 /* labels must always be allocated at level 1 (outhermost block) */ 1255 if (symtyp == FLAB) { 1256 sym = getlblk(1, sizeof (sym_t)); 1257 s = getlblk(1, sb->sb_len + 1); 1258 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1259 sym->s_name = s; 1260 sym->s_blklev = 1; 1261 di = dcs; 1262 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1263 di = di->d_nxt; 1264 if (di->d_ctx != AUTO) 1265 lerror("storesym() 2"); 1266 } else { 1267 sym = getblk(sizeof (sym_t)); 1268 sym->s_name = sb->sb_name; 1269 sym->s_blklev = blklev; 1270 di = dcs; 1271 } 1272 1273 STRUCT_ASSIGN(sym->s_dpos, curr_pos); 1274 if ((sym->s_kind = symtyp) != FLAB) 1275 sym->s_type = gettyp(INT); 1276 1277 symtyp = FVFT; 1278 1279 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1280 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1281 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1282 1283 *di->d_ldlsym = sym; 1284 di->d_ldlsym = &sym->s_dlnxt; 1285 1286 freesb(sb); 1287 return (sym); 1288} 1289 1290/* 1291 * Remove a symbol forever from the symbol table. s_blklev 1292 * is set to -1 to avoid that the symbol will later be put 1293 * back to the symbol table. 1294 */ 1295void 1296rmsym(sym) 1297 sym_t *sym; 1298{ 1299 if ((*sym->s_rlink = sym->s_link) != NULL) 1300 sym->s_link->s_rlink = sym->s_rlink; 1301 sym->s_blklev = -1; 1302 sym->s_link = NULL; 1303} 1304 1305/* 1306 * Remove a list of symbols declared at one level from the symbol 1307 * table. 1308 */ 1309void 1310rmsyms(syms) 1311 sym_t *syms; 1312{ 1313 sym_t *sym; 1314 1315 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1316 if (sym->s_blklev != -1) { 1317 if ((*sym->s_rlink = sym->s_link) != NULL) 1318 sym->s_link->s_rlink = sym->s_rlink; 1319 sym->s_link = NULL; 1320 sym->s_rlink = NULL; 1321 } 1322 } 1323} 1324 1325/* 1326 * Put a symbol into the symbol table 1327 */ 1328void 1329inssym(bl, sym) 1330 int bl; 1331 sym_t *sym; 1332{ 1333 int h; 1334 1335 h = hash(sym->s_name); 1336 if ((sym->s_link = symtab[h]) != NULL) 1337 symtab[h]->s_rlink = &sym->s_link; 1338 (symtab[h] = sym)->s_rlink = &symtab[h]; 1339 sym->s_blklev = bl; 1340 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1341 lerror("inssym()"); 1342} 1343 1344/* 1345 * Called at level 0 after syntax errors 1346 * Removes all symbols which are not declared at level 0 from the 1347 * symbol table. Also frees all memory which is not associated with 1348 * level 0. 1349 */ 1350void 1351cleanup() 1352{ 1353 sym_t *sym, *nsym; 1354 int i; 1355 1356 for (i = 0; i < HSHSIZ1; i++) { 1357 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1358 nsym = sym->s_link; 1359 if (sym->s_blklev >= 1) { 1360 if ((*sym->s_rlink = nsym) != NULL) 1361 nsym->s_rlink = sym->s_rlink; 1362 } 1363 } 1364 } 1365 1366 for (i = mblklev; i > 0; i--) 1367 freelblk(i); 1368} 1369 1370/* 1371 * Create a new symbol with the name of an existing symbol. 1372 */ 1373sym_t * 1374pushdown(sym) 1375 sym_t *sym; 1376{ 1377 int h; 1378 sym_t *nsym; 1379 1380 h = hash(sym->s_name); 1381 nsym = getblk(sizeof (sym_t)); 1382 if (sym->s_blklev > blklev) 1383 lerror("pushdown()"); 1384 nsym->s_name = sym->s_name; 1385 STRUCT_ASSIGN(nsym->s_dpos, curr_pos); 1386 nsym->s_kind = sym->s_kind; 1387 nsym->s_blklev = blklev; 1388 1389 if ((nsym->s_link = symtab[h]) != NULL) 1390 symtab[h]->s_rlink = &nsym->s_link; 1391 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1392 1393 *dcs->d_ldlsym = nsym; 1394 dcs->d_ldlsym = &nsym->s_dlnxt; 1395 1396 return (nsym); 1397} 1398 1399/* 1400 * Free any dynamically allocated memory referenced by 1401 * the value stack or yylval. 1402 * The type of information in yylval is described by tok. 1403 */ 1404void 1405freeyyv(sp, tok) 1406 void *sp; 1407 int tok; 1408{ 1409 if (tok == T_NAME || tok == T_TYPENAME) { 1410 sbuf_t *sb = *(sbuf_t **)sp; 1411 freesb(sb); 1412 } else if (tok == T_CON) { 1413 val_t *val = *(val_t **)sp; 1414 free(val); 1415 } else if (tok == T_STRING) { 1416 strg_t *strg = *(strg_t **)sp; 1417 if (strg->st_tspec == CHAR) { 1418 free(strg->st_cp); 1419 } else if (strg->st_tspec == WCHAR) { 1420 free(strg->st_wcp); 1421 } else { 1422 lerror("fryylv() 1"); 1423 } 1424 free(strg); 1425 } 1426} 1427