scan.l revision 75730
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[] = "$FreeBSD: head/usr.bin/xlint/lint1/scan.l 75730 2001-04-20 09:33:57Z asmodai $"; 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/* Current position (its also updated when an included file is parsed) */ 54pos_t curr_pos = { 1, "" }; 55 56/* 57 * Current position in C source (not updated when an included file is 58 * parsed). 59 */ 60pos_t csrc_pos = { 1, "" }; 61 62static void incline __P((void)); 63static void badchar __P((int)); 64static sbuf_t *allocsb __P((void)); 65static void freesb __P((sbuf_t *)); 66static int inpc __P((void)); 67static int hash __P((const char *)); 68static sym_t *search __P((sbuf_t *)); 69static int name __P((void)); 70static int keyw __P((sym_t *)); 71static int icon __P((int)); 72static int fcon __P((void)); 73static int operator __P((int, op_t)); 74static int ccon __P((void)); 75static int wccon __P((void)); 76static int getescc __P((int)); 77static void directive __P((void)); 78static void comment __P((void)); 79static int string __P((void)); 80static int wcstrg __P((void)); 81 82%} 83 84L [_A-Za-z] 85D [0-9] 86NZD [1-9] 87OD [0-7] 88HD [0-9A-Fa-f] 89EX ([eE][+-]?[0-9]+) 90 91%% 92 93{L}({L}|{D})* return (name()); 940{OD}*[lLuU]* return (icon(8)); 95{NZD}{D}*[lLuU]* return (icon(10)); 960[xX]{HD}+[lLuU]* return (icon(16)); 97{D}+\.{D}*{EX}?[fFlL]? | 98{D}+{EX}[fFlL]? | 99\.{D}+{EX}?[fFlL]? return (fcon()); 100"=" return (operator(T_ASSIGN, ASSIGN)); 101"*=" return (operator(T_OPASS, MULASS)); 102"/=" return (operator(T_OPASS, DIVASS)); 103"%=" return (operator(T_OPASS, MODASS)); 104"+=" return (operator(T_OPASS, ADDASS)); 105"-=" return (operator(T_OPASS, SUBASS)); 106"<<=" return (operator(T_OPASS, SHLASS)); 107">>=" return (operator(T_OPASS, SHRASS)); 108"&=" return (operator(T_OPASS, ANDASS)); 109"^=" return (operator(T_OPASS, XORASS)); 110"|=" return (operator(T_OPASS, ORASS)); 111"||" return (operator(T_LOGOR, LOGOR)); 112"&&" return (operator(T_LOGAND, LOGAND)); 113"|" return (operator(T_OR, OR)); 114"&" return (operator(T_AND, AND)); 115"^" return (operator(T_XOR, XOR)); 116"==" return (operator(T_EQOP, EQ)); 117"!=" return (operator(T_EQOP, NE)); 118"<" return (operator(T_RELOP, LT)); 119">" return (operator(T_RELOP, GT)); 120"<=" return (operator(T_RELOP, LE)); 121">=" return (operator(T_RELOP, GE)); 122"<<" return (operator(T_SHFTOP, SHL)); 123">>" return (operator(T_SHFTOP, SHR)); 124"++" return (operator(T_INCDEC, INC)); 125"--" return (operator(T_INCDEC, DEC)); 126"->" return (operator(T_STROP, ARROW)); 127"." return (operator(T_STROP, POINT)); 128"+" return (operator(T_ADDOP, PLUS)); 129"-" return (operator(T_ADDOP, MINUS)); 130"*" return (operator(T_MULT, MULT)); 131"/" return (operator(T_DIVOP, DIV)); 132"%" return (operator(T_DIVOP, MOD)); 133"!" return (operator(T_UNOP, NOT)); 134"~" return (operator(T_UNOP, COMPL)); 135"\"" return (string()); 136"L\"" return (wcstrg()); 137";" return (T_SEMI); 138"{" return (T_LBRACE); 139"}" return (T_RBRACE); 140"," return (T_COMMA); 141":" return (T_COLON); 142"?" return (T_QUEST); 143"[" return (T_LBRACK); 144"]" return (T_RBRACK); 145"(" return (T_LPARN); 146")" return (T_RPARN); 147"..." return (T_ELLIPSE); 148"'" return (ccon()); 149"L'" return (wccon()); 150^#.*$ directive(); 151\n incline(); 152\t|" "|\f|\v ; 153"/*" comment(); 154. badchar(yytext[0]); 155 156%% 157 158static void 159incline() 160{ 161 curr_pos.p_line++; 162 if (curr_pos.p_file == csrc_pos.p_file) 163 csrc_pos.p_line++; 164} 165 166static void 167badchar(c) 168 int c; 169{ 170 /* unknown character \%o */ 171 error(250, c); 172} 173 174/* 175 * Keywords. 176 * During initialisation they are written to the symbol table. 177 */ 178static struct kwtab { 179 const char *kw_name; /* keyword */ 180 int kw_token; /* token returned by yylex() */ 181 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 182 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 183 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 184 u_int kw_stdc : 1; /* STDC keyword */ 185 u_int kw_gcc : 1; /* GCC keyword */ 186} kwtab[] = { 187 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 188 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 189 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 190 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 191 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 192 { "case", T_CASE, 0, 0, 0, 0, 0 }, 193 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 194 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 195 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 196 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 197 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 198 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 199 { "do", T_DO, 0, 0, 0, 0, 0 }, 200 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 201 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 202 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 203 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 204 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 205 { "for", T_FOR, 0, 0, 0, 0, 0 }, 206 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 207 { "if", T_IF, 0, 0, 0, 0, 0 }, 208 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 209 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 210 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 211 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 212 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 213 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 214 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 215 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 216 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 217 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 218 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 219 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 220 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 221 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 222 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 223 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 224 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 225 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 226 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 227 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 228 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 229 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 230 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 231 { NULL, 0, 0, 0, 0, 0, 0 } 232}; 233 234/* Symbol table */ 235static sym_t *symtab[HSHSIZ1]; 236 237/* bit i of the entry with index i is set */ 238u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 239 240/* least significant i bits are set in the entry with index i */ 241u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 242 243/* least significant i bits are not set in the entry with index i */ 244u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 245 246/* free list for sbuf structures */ 247static sbuf_t *sbfrlst; 248 249/* Typ of next expected symbol */ 250symt_t symtyp; 251 252 253/* 254 * All keywords are written to the symbol table. This saves us looking 255 * in a extra table for each name we found. 256 */ 257void 258initscan() 259{ 260 struct kwtab *kw; 261 sym_t *sym; 262 int h, i; 263 u_quad_t uq; 264 265 for (kw = kwtab; kw->kw_name != NULL; kw++) { 266 if (kw->kw_stdc && tflag) 267 continue; 268 if (kw->kw_gcc && !gflag) 269 continue; 270 sym = getblk(sizeof (sym_t)); 271 sym->s_name = kw->kw_name; 272 sym->s_keyw = 1; 273 sym->s_value.v_quad = kw->kw_token; 274 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 275 sym->s_tspec = kw->kw_tspec; 276 } else if (kw->kw_token == T_SCLASS) { 277 sym->s_scl = kw->kw_scl; 278 } else if (kw->kw_token == T_QUAL) { 279 sym->s_tqual = kw->kw_tqual; 280 } 281 h = hash(sym->s_name); 282 if ((sym->s_link = symtab[h]) != NULL) 283 symtab[h]->s_rlink = &sym->s_link; 284 (symtab[h] = sym)->s_rlink = &symtab[h]; 285 } 286 287 /* initialize bit-masks for quads */ 288 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 289 qbmasks[i] = (u_quad_t)1 << i; 290 uq = ~(u_quad_t)0 << i; 291 qumasks[i] = uq; 292 qlmasks[i] = ~uq; 293 } 294 qumasks[i] = 0; 295 qlmasks[i] = ~(u_quad_t)0; 296} 297 298/* 299 * Get a free sbuf structure, if possible from the free list 300 */ 301static sbuf_t * 302allocsb() 303{ 304 sbuf_t *sb; 305 306 if ((sb = sbfrlst) != NULL) { 307 sbfrlst = sb->sb_nxt; 308 } else { 309 sb = xmalloc(sizeof (sbuf_t)); 310 } 311 (void)memset(sb, 0, sizeof (sb)); 312 return (sb); 313} 314 315/* 316 * Put a sbuf structure to the free list 317 */ 318static void 319freesb(sb) 320 sbuf_t *sb; 321{ 322 sb->sb_nxt = sbfrlst; 323 sbfrlst = sb; 324} 325 326/* 327 * Read a character and ensure that it is positive (except EOF). 328 * Increment line count(s) if necessary. 329 */ 330static int 331inpc() 332{ 333 int c; 334 335 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 336 incline(); 337 return (c); 338} 339 340static int 341hash(s) 342 const char *s; 343{ 344 u_int v; 345 const u_char *us; 346 347 v = 0; 348 for (us = (const u_char *)s; *us != '\0'; us++) { 349 v = (v << sizeof (v)) + *us; 350 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 351 } 352 return (v % HSHSIZ1); 353} 354 355/* 356 * Lex has found a letter followed by zero or more letters or digits. 357 * It looks for a symbol in the symbol table with the same name. This 358 * symbol must either be a keyword or a symbol of the type required by 359 * symtyp (label, member, tag, ...). 360 * 361 * If it is a keyword, the token is returned. In some cases it is described 362 * more deeply by data written to yylval. 363 * 364 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 365 * is stored in yylval. This struct contains the name of the symbol, it's 366 * length and hash value. If there is already a symbol of the same name 367 * and type in the symbol table, the sbuf struct also contains a pointer 368 * to the symbol table entry. 369 */ 370static int 371name() 372{ 373 char *s; 374 sbuf_t *sb; 375 sym_t *sym; 376 int tok; 377 378 sb = allocsb(); 379 sb->sb_name = yytext; 380 sb->sb_len = yyleng; 381 sb->sb_hash = hash(yytext); 382 383 if ((sym = search(sb)) != NULL && sym->s_keyw) { 384 freesb(sb); 385 return (keyw(sym)); 386 } 387 388 sb->sb_sym = sym; 389 390 if (sym != NULL) { 391 if (blklev < sym->s_blklev) 392 lerror("name() 1"); 393 sb->sb_name = sym->s_name; 394 sb->sb_len = strlen(sym->s_name); 395 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 396 } else { 397 s = getblk(yyleng + 1); 398 (void)memcpy(s, yytext, yyleng + 1); 399 sb->sb_name = s; 400 sb->sb_len = yyleng; 401 tok = T_NAME; 402 } 403 404 yylval.y_sb = sb; 405 return (tok); 406} 407 408static sym_t * 409search(sb) 410 sbuf_t *sb; 411{ 412 sym_t *sym; 413 414 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 415 if (strcmp(sym->s_name, sb->sb_name) == 0) { 416 if (sym->s_keyw || sym->s_kind == symtyp) 417 return (sym); 418 } 419 } 420 421 return (NULL); 422} 423 424static int 425keyw(sym) 426 sym_t *sym; 427{ 428 int t; 429 430 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 431 yylval.y_scl = sym->s_scl; 432 } else if (t == T_TYPE || t == T_SOU) { 433 yylval.y_tspec = sym->s_tspec; 434 } else if (t == T_QUAL) { 435 yylval.y_tqual = sym->s_tqual; 436 } 437 return (t); 438} 439 440/* 441 * Convert a string representing an integer into internal representation. 442 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 443 */ 444static int 445icon(base) 446 int base; 447{ 448 int l_suffix, u_suffix; 449 int len; 450 const char *cp; 451 char c, *eptr; 452 tspec_t typ; 453 u_long ul; 454 u_quad_t uq; 455 int ansiu; 456 static tspec_t contypes[2][3] = { 457 { INT, LONG, QUAD }, 458 { UINT, ULONG, UQUAD } 459 }; 460 461 cp = yytext; 462 len = yyleng; 463 464 /* skip 0x */ 465 if (base == 16) { 466 cp += 2; 467 len -= 2; 468 } 469 470 /* read suffixes */ 471 l_suffix = u_suffix = 0; 472 for ( ; ; ) { 473 if ((c = cp[len - 1]) == 'l' || c == 'L') { 474 l_suffix++; 475 } else if (c == 'u' || c == 'U') { 476 u_suffix++; 477 } else { 478 break; 479 } 480 len--; 481 } 482 if (l_suffix > 2 || u_suffix > 1) { 483 /* malformed integer constant */ 484 warning(251); 485 if (l_suffix > 2) 486 l_suffix = 2; 487 if (u_suffix > 1) 488 u_suffix = 1; 489 } 490 if (tflag && u_suffix != 0) { 491 /* suffix U is illegal in traditional C */ 492 warning(97); 493 } 494 typ = contypes[u_suffix][l_suffix]; 495 496 errno = 0; 497 if (l_suffix < 2) { 498 ul = strtoul(cp, &eptr, base); 499 } else { 500 uq = strtouq(cp, &eptr, base); 501 } 502 if (eptr != cp + len) 503 lerror("icon() 1"); 504 if (errno != 0) 505 /* integer constant out of range */ 506 warning(252); 507 508 /* 509 * If the value is to big for the current type, we must choose 510 * another type. 511 */ 512 ansiu = 0; 513 switch (typ) { 514 case INT: 515 if (ul <= INT_MAX) { 516 /* ok */ 517 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 518 typ = UINT; 519 } else if (ul <= LONG_MAX) { 520 typ = LONG; 521 } else { 522 typ = ULONG; 523 } 524 if (typ == UINT || typ == ULONG) { 525 if (tflag) { 526 typ = LONG; 527 } else if (!sflag) { 528 /* 529 * Remember that the constant is unsigned 530 * only in ANSI C 531 */ 532 ansiu = 1; 533 } 534 } 535 break; 536 case UINT: 537 if (ul > (u_int)UINT_MAX) 538 typ = ULONG; 539 break; 540 case LONG: 541 if (ul > LONG_MAX && !tflag) { 542 typ = ULONG; 543 if (!sflag) 544 ansiu = 1; 545 } 546 break; 547 case QUAD: 548 if (uq > QUAD_MAX && !tflag) { 549 typ = UQUAD; 550 if (!sflag) 551 ansiu = 1; 552 } 553 break; 554 /* LINTED (enumeration values not handled in switch) */ 555 default: 556 } 557 558 if (typ != QUAD && typ != UQUAD) { 559 if (isutyp(typ)) { 560 uq = ul; 561 } else { 562 uq = (quad_t)(long)ul; 563 } 564 } 565 566 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 567 568 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 569 yylval.y_val->v_ansiu = ansiu; 570 yylval.y_val->v_quad = (quad_t)uq; 571 572 return (T_CON); 573} 574 575/* 576 * Returns 1 if t is a signed type and the value is negative. 577 * 578 * len is the number of significant bits. If len is -1, len is set 579 * to the width of type t. 580 */ 581int 582sign(q, t, len) 583 quad_t q; 584 tspec_t t; 585 int len; 586{ 587 if (t == PTR || isutyp(t)) 588 return (0); 589 return (msb(q, t, len)); 590} 591 592int 593msb(q, t, len) 594 quad_t q; 595 tspec_t t; 596 int len; 597{ 598 if (len <= 0) 599 len = size(t); 600 return ((q & qbmasks[len - 1]) != 0); 601} 602 603/* 604 * Extends the sign of q. 605 */ 606quad_t 607xsign(q, t, len) 608 quad_t q; 609 tspec_t t; 610 int len; 611{ 612 if (len <= 0) 613 len = size(t); 614 615 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 616 q &= qlmasks[len]; 617 } else { 618 q |= qumasks[len]; 619 } 620 return (q); 621} 622 623/* 624 * Convert a string representing a floating point value into its interal 625 * representation. Type and value are returned in yylval. fcon() 626 * (and yylex()) returns T_CON. 627 * XXX Currently it is not possible to convert constants of type 628 * long double which are greater then DBL_MAX. 629 */ 630static int 631fcon() 632{ 633 const char *cp; 634 int len; 635 tspec_t typ; 636 char c, *eptr; 637 double d; 638 float f; 639 640 cp = yytext; 641 len = yyleng; 642 643 if ((c = cp[len - 1]) == 'f' || c == 'F') { 644 typ = FLOAT; 645 len--; 646 } else if (c == 'l' || c == 'L') { 647 typ = LDOUBLE; 648 len--; 649 } else { 650 typ = DOUBLE; 651 } 652 653 if (tflag && typ != DOUBLE) { 654 /* suffixes F and L are illegal in traditional C */ 655 warning(98); 656 } 657 658 errno = 0; 659 d = strtod(cp, &eptr); 660 if (eptr != cp + len) 661 lerror("fcon() 1"); 662 if (errno != 0) 663 /* floating-point constant out of range */ 664 warning(248); 665 666 if (typ == FLOAT) { 667 f = (float)d; 668 if (isinf(f)) { 669 /* floating-point constant out of range */ 670 warning(248); 671 f = f > 0 ? FLT_MAX : -FLT_MAX; 672 } 673 } 674 675 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 676 if (typ == FLOAT) { 677 yylval.y_val->v_ldbl = f; 678 } else { 679 yylval.y_val->v_ldbl = d; 680 } 681 682 return (T_CON); 683} 684 685static int 686operator(t, o) 687 int t; 688 op_t o; 689{ 690 yylval.y_op = o; 691 return (t); 692} 693 694/* 695 * Called if lex found a leading \'. 696 */ 697static int 698ccon() 699{ 700 int n, val, c; 701 char cv; 702 703 n = 0; 704 val = 0; 705 while ((c = getescc('\'')) >= 0) { 706 val = (val << CHAR_BIT) + c; 707 n++; 708 } 709 if (c == -2) { 710 /* unterminated character constant */ 711 error(253); 712 } else { 713 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 714 /* too many characters in character constant */ 715 error(71); 716 } else if (n > 1) { 717 /* multi-character character constant */ 718 warning(294); 719 } else if (n == 0) { 720 /* empty character constant */ 721 error(73); 722 } 723 } 724 if (n == 1) { 725 cv = (char)val; 726 val = cv; 727 } 728 729 yylval.y_val = xcalloc(1, sizeof (val_t)); 730 yylval.y_val->v_tspec = INT; 731 yylval.y_val->v_quad = val; 732 733 return (T_CON); 734} 735 736/* 737 * Called if lex found a leading L\' 738 */ 739static int 740wccon() 741{ 742 static char buf[MB_LEN_MAX + 1]; 743 int i, c; 744 wchar_t wc; 745 746 i = 0; 747 while ((c = getescc('\'')) >= 0) { 748 if (i < MB_CUR_MAX) 749 buf[i] = (char)c; 750 i++; 751 } 752 753 wc = 0; 754 755 if (c == -2) { 756 /* unterminated character constant */ 757 error(253); 758 } else if (c == 0) { 759 /* empty character constant */ 760 error(73); 761 } else { 762 if (i > MB_CUR_MAX) { 763 i = MB_CUR_MAX; 764 /* too many characters in character constant */ 765 error(71); 766 } else { 767 buf[i] = '\0'; 768 (void)mbtowc(NULL, NULL, 0); 769 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 770 /* invalid multibyte character */ 771 error(291); 772 } 773 } 774 775 yylval.y_val = xcalloc(1, sizeof (val_t)); 776 yylval.y_val->v_tspec = WCHAR; 777 yylval.y_val->v_quad = wc; 778 779 return (T_CON); 780} 781 782/* 783 * Read a character which is part of a character constant or of a string 784 * and handle escapes. 785 * 786 * The Argument is the character which delimits the character constant or 787 * string. 788 * 789 * Returns -1 if the end of the character constant or string is reached, 790 * -2 if the EOF is reached, and the charachter otherwise. 791 */ 792static int 793getescc(d) 794 int d; 795{ 796 static int pbc = -1; 797 int n, c, v; 798 799 if (pbc == -1) { 800 c = inpc(); 801 } else { 802 c = pbc; 803 pbc = -1; 804 } 805 if (c == d) 806 return (-1); 807 switch (c) { 808 case '\n': 809 /* newline in string or char constant */ 810 error(254); 811 return (-2); 812 case EOF: 813 return (-2); 814 case '\\': 815 switch (c = inpc()) { 816 case '"': 817 if (tflag && d == '\'') 818 /* \" inside character constant undef. ... */ 819 warning(262); 820 return ('"'); 821 case '\'': 822 return ('\''); 823 case '?': 824 if (tflag) 825 /* \? undefined in traditional C */ 826 warning(263); 827 return ('?'); 828 case '\\': 829 return ('\\'); 830 case 'a': 831 if (tflag) 832 /* \a undefined in traditional C */ 833 warning(81); 834#ifdef __STDC__ 835 return ('\a'); 836#else 837 return ('\007'); 838#endif 839 case 'b': 840 return ('\b'); 841 case 'f': 842 return ('\f'); 843 case 'n': 844 return ('\n'); 845 case 'r': 846 return ('\r'); 847 case 't': 848 return ('\t'); 849 case 'v': 850 if (tflag) 851 /* \v undefined in traditional C */ 852 warning(264); 853#ifdef __STDC__ 854 return ('\v'); 855#else 856 return ('\013'); 857#endif 858 case '8': case '9': 859 /* bad octal digit %c */ 860 warning(77, c); 861 /* FALLTHROUGH */ 862 case '0': case '1': case '2': case '3': 863 case '4': case '5': case '6': case '7': 864 n = 3; 865 v = 0; 866 do { 867 v = (v << 3) + (c - '0'); 868 c = inpc(); 869 } while (--n && isdigit(c) && (tflag || c <= '7')); 870 if (tflag && n > 0 && isdigit(c)) 871 /* bad octal digit %c */ 872 warning(77, c); 873 pbc = c; 874 if (v > UCHAR_MAX) { 875 /* character escape does not fit in char. */ 876 warning(76); 877 v &= CHAR_MASK; 878 } 879 return (v); 880 case 'x': 881 if (tflag) 882 /* \x undefined in traditional C */ 883 warning(82); 884 v = 0; 885 n = 0; 886 while ((c = inpc()) >= 0 && isxdigit(c)) { 887 c = isdigit(c) ? 888 c - '0' : toupper(c) - 'A' + 10; 889 v = (v << 4) + c; 890 if (n >= 0) { 891 if ((v & ~CHAR_MASK) != 0) { 892 /* overflow in hex escape */ 893 warning(75); 894 n = -1; 895 } else { 896 n++; 897 } 898 } 899 } 900 pbc = c; 901 if (n == 0) { 902 /* no hex digits follow \x */ 903 error(74); 904 } if (n == -1) { 905 v &= CHAR_MASK; 906 } 907 return (v); 908 case '\n': 909 return (getescc(d)); 910 case EOF: 911 return (-2); 912 default: 913 if (isprint(c)) { 914 /* dubious escape \%c */ 915 warning(79, c); 916 } else { 917 /* dubious escape \%o */ 918 warning(80, c); 919 } 920 } 921 } 922 return (c); 923} 924 925/* 926 * Called for preprocessor directives. Currently implemented are: 927 * # lineno 928 * # lineno "filename" 929 */ 930static void 931directive() 932{ 933 const char *cp, *fn; 934 char c, *eptr; 935 size_t fnl; 936 long ln; 937 static int first = 1; 938 939 /* Go to first non-whitespace after # */ 940 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 941 942 if (!isdigit(c)) { 943 error: 944 /* undefined or invalid # directive */ 945 warning(255); 946 return; 947 } 948 ln = strtol(--cp, &eptr, 10); 949 if (cp == eptr) 950 goto error; 951 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 952 goto error; 953 while ((c = *cp++) == ' ' || c == '\t') ; 954 if (c != '\0') { 955 if (c != '"') 956 goto error; 957 fn = cp; 958 while ((c = *cp) != '"' && c != '\0') 959 cp++; 960 if (c != '"') 961 goto error; 962 if ((fnl = cp++ - fn) > PATH_MAX) 963 goto error; 964 while ((c = *cp++) == ' ' || c == '\t') ; 965#if 0 966 if (c != '\0') 967 warning("extra character(s) after directive"); 968#endif 969 curr_pos.p_file = fnnalloc(fn, fnl); 970 /* 971 * If this is the first directive, the name is the name 972 * of the C source file as specified at the command line. 973 * It is written to the output file. 974 */ 975 if (first) { 976 csrc_pos.p_file = curr_pos.p_file; 977 outsrc(curr_pos.p_file); 978 first = 0; 979 } 980 } 981 curr_pos.p_line = (int)ln - 1; 982 if (curr_pos.p_file == csrc_pos.p_file) 983 csrc_pos.p_line = (int)ln - 1; 984} 985 986/* 987 * Handle lint comments. Following comments are currently understood: 988 * ARGSUSEDn 989 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 990 * FALLTHRU FALLTHROUGH 991 * LINTLIBRARY 992 * LINTED NOSTRICT 993 * LONGLONG 994 * NOTREACHED 995 * PRINTFLIKEn 996 * PROTOLIB 997 * SCANFLIKEn 998 * VARARGSn 999 * If one of this comments is recognized, the arguments, if any, are 1000 * parsed and a function which handles this comment is called. 1001 */ 1002static void 1003comment() 1004{ 1005 int c, lc; 1006 static struct { 1007 const char *keywd; 1008 int arg; 1009 void (*func) __P((int)); 1010 } keywtab[] = { 1011 { "ARGSUSED", 1, argsused }, 1012 { "CONSTCOND", 0, constcond }, 1013 { "CONSTANTCOND", 0, constcond }, 1014 { "CONSTANTCONDITION", 0, constcond }, 1015 { "FALLTHRU", 0, fallthru }, 1016 { "FALLTHROUGH", 0, fallthru }, 1017 { "LINTLIBRARY", 0, lintlib }, 1018 { "LINTED", 0, linted }, 1019 { "LONGLONG", 0, longlong }, 1020 { "NOSTRICT", 0, linted }, 1021 { "NOTREACHED", 0, notreach }, 1022 { "PRINTFLIKE", 1, printflike }, 1023 { "PROTOLIB", 1, protolib }, 1024 { "SCANFLIKE", 1, scanflike }, 1025 { "VARARGS", 1, varargs }, 1026 }; 1027 char keywd[32]; 1028 char arg[32]; 1029 int l, i, a; 1030 int eoc; 1031 1032 eoc = 0; 1033 1034 /* Skip white spaces after the start of the comment */ 1035 while ((c = inpc()) != EOF && isspace(c)) ; 1036 1037 /* Read the potential keyword to keywd */ 1038 l = 0; 1039 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1040 keywd[l++] = (char)c; 1041 c = inpc(); 1042 } 1043 keywd[l] = '\0'; 1044 1045 /* look for the keyword */ 1046 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1047 if (strcmp(keywtab[i].keywd, keywd) == 0) 1048 break; 1049 } 1050 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1051 goto skip_rest; 1052 1053 /* skip white spaces after the keyword */ 1054 while (c != EOF && isspace(c)) 1055 c = inpc(); 1056 1057 /* read the argument, if the keyword accepts one and there is one */ 1058 l = 0; 1059 if (keywtab[i].arg) { 1060 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1061 arg[l++] = (char)c; 1062 c = inpc(); 1063 } 1064 } 1065 arg[l] = '\0'; 1066 a = l != 0 ? atoi(arg) : -1; 1067 1068 /* skip white spaces after the argument */ 1069 while (c != EOF && isspace(c)) 1070 c = inpc(); 1071 1072 if (c != '*' || (c = inpc()) != '/') { 1073 if (keywtab[i].func != linted) 1074 /* extra characters in lint comment */ 1075 warning(257); 1076 } else { 1077 /* 1078 * remember that we have already found the end of the 1079 * comment 1080 */ 1081 eoc = 1; 1082 } 1083 1084 if (keywtab[i].func != NULL) 1085 (*keywtab[i].func)(a); 1086 1087 skip_rest: 1088 while (!eoc) { 1089 lc = c; 1090 if ((c = inpc()) == EOF) { 1091 /* unterminated comment */ 1092 error(256); 1093 break; 1094 } 1095 if (lc == '*' && c == '/') 1096 eoc = 1; 1097 } 1098} 1099 1100/* 1101 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1102 * clrwflgs() is called after function definitions and global and 1103 * local declarations and definitions. It is also called between 1104 * the controlling expression and the body of control statements 1105 * (if, switch, for, while). 1106 */ 1107void 1108clrwflgs() 1109{ 1110 nowarn = 0; 1111 quadflg = 0; 1112 ccflg = 0; 1113} 1114 1115/* 1116 * Strings are stored in a dynamically alloceted buffer and passed 1117 * in yylval.y_xstrg to the parser. The parser or the routines called 1118 * by the parser are responsible for freeing this buffer. 1119 */ 1120static int 1121string() 1122{ 1123 u_char *s; 1124 int c; 1125 size_t len, max; 1126 strg_t *strg; 1127 1128 s = xmalloc(max = 64); 1129 1130 len = 0; 1131 while ((c = getescc('"')) >= 0) { 1132 /* +1 to reserve space for a trailing NUL character */ 1133 if (len + 1 == max) 1134 s = xrealloc(s, max *= 2); 1135 s[len++] = (char)c; 1136 } 1137 s[len] = '\0'; 1138 if (c == -2) 1139 /* unterminated string constant */ 1140 error(258); 1141 1142 strg = xcalloc(1, sizeof (strg_t)); 1143 strg->st_tspec = CHAR; 1144 strg->st_len = len; 1145 strg->st_cp = s; 1146 1147 yylval.y_strg = strg; 1148 return (T_STRING); 1149} 1150 1151static int 1152wcstrg() 1153{ 1154 char *s; 1155 int c, i, n, wi; 1156 size_t len, max, wlen; 1157 wchar_t *ws; 1158 strg_t *strg; 1159 1160 s = xmalloc(max = 64); 1161 len = 0; 1162 while ((c = getescc('"')) >= 0) { 1163 /* +1 to save space for a trailing NUL character */ 1164 if (len + 1 >= max) 1165 s = xrealloc(s, max *= 2); 1166 s[len++] = (char)c; 1167 } 1168 s[len] = '\0'; 1169 if (c == -2) 1170 /* unterminated string constant */ 1171 error(258); 1172 1173 /* get length of wide character string */ 1174 (void)mblen(NULL, 0); 1175 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1176 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1177 /* invalid multibyte character */ 1178 error(291); 1179 break; 1180 } 1181 if (n == 0) 1182 n = 1; 1183 } 1184 1185 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1186 1187 /* convert from multibyte to wide char */ 1188 (void)mbtowc(NULL, NULL, 0); 1189 for (i = 0, wi = 0; i < len; i += n, wi++) { 1190 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1191 break; 1192 if (n == 0) 1193 n = 1; 1194 } 1195 ws[wi] = 0; 1196 free(s); 1197 1198 strg = xcalloc(1, sizeof (strg_t)); 1199 strg->st_tspec = WCHAR; 1200 strg->st_len = wlen; 1201 strg->st_wcp = ws; 1202 1203 yylval.y_strg = strg; 1204 return (T_STRING); 1205} 1206 1207/* 1208 * As noted above the scanner does not create new symbol table entries 1209 * for symbols it cannot find in the symbol table. This is to avoid 1210 * putting undeclared symbols into the symbol table if a syntax error 1211 * occurs. 1212 * 1213 * getsym() is called as soon as it is probably ok to put the symbol to 1214 * the symbol table. This does not mean that it is not possible that 1215 * symbols are put to the symbol table which are than not completely 1216 * declared due to syntax errors. To avoid too many problems in this 1217 * case symbols get type int in getsym(). 1218 * 1219 * XXX calls to getsym() should be delayed until decl1*() is called 1220 */ 1221sym_t * 1222getsym(sb) 1223 sbuf_t *sb; 1224{ 1225 dinfo_t *di; 1226 char *s; 1227 sym_t *sym; 1228 1229 sym = sb->sb_sym; 1230 1231 /* 1232 * During member declaration it is possible that name() looked 1233 * for symbols of type FVFT, although it should have looked for 1234 * symbols of type FTAG. Same can happen for labels. Both cases 1235 * are compensated here. 1236 */ 1237 if (symtyp == FMOS || symtyp == FLAB) { 1238 if (sym == NULL || sym->s_kind == FVFT) 1239 sym = search(sb); 1240 } 1241 1242 if (sym != NULL) { 1243 if (sym->s_kind != symtyp) 1244 lerror("storesym() 1"); 1245 symtyp = FVFT; 1246 freesb(sb); 1247 return (sym); 1248 } 1249 1250 /* create a new symbol table entry */ 1251 1252 /* labels must always be allocated at level 1 (outhermost block) */ 1253 if (symtyp == FLAB) { 1254 sym = getlblk(1, sizeof (sym_t)); 1255 s = getlblk(1, sb->sb_len + 1); 1256 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1257 sym->s_name = s; 1258 sym->s_blklev = 1; 1259 di = dcs; 1260 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1261 di = di->d_nxt; 1262 if (di->d_ctx != AUTO) 1263 lerror("storesym() 2"); 1264 } else { 1265 sym = getblk(sizeof (sym_t)); 1266 sym->s_name = sb->sb_name; 1267 sym->s_blklev = blklev; 1268 di = dcs; 1269 } 1270 1271 STRUCT_ASSIGN(sym->s_dpos, curr_pos); 1272 if ((sym->s_kind = symtyp) != FLAB) 1273 sym->s_type = gettyp(INT); 1274 1275 symtyp = FVFT; 1276 1277 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1278 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1279 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1280 1281 *di->d_ldlsym = sym; 1282 di->d_ldlsym = &sym->s_dlnxt; 1283 1284 freesb(sb); 1285 return (sym); 1286} 1287 1288/* 1289 * Remove a symbol forever from the symbol table. s_blklev 1290 * is set to -1 to avoid that the symbol will later be put 1291 * back to the symbol table. 1292 */ 1293void 1294rmsym(sym) 1295 sym_t *sym; 1296{ 1297 if ((*sym->s_rlink = sym->s_link) != NULL) 1298 sym->s_link->s_rlink = sym->s_rlink; 1299 sym->s_blklev = -1; 1300 sym->s_link = NULL; 1301} 1302 1303/* 1304 * Remove a list of symbols declared at one level from the symbol 1305 * table. 1306 */ 1307void 1308rmsyms(syms) 1309 sym_t *syms; 1310{ 1311 sym_t *sym; 1312 1313 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1314 if (sym->s_blklev != -1) { 1315 if ((*sym->s_rlink = sym->s_link) != NULL) 1316 sym->s_link->s_rlink = sym->s_rlink; 1317 sym->s_link = NULL; 1318 sym->s_rlink = NULL; 1319 } 1320 } 1321} 1322 1323/* 1324 * Put a symbol into the symbol table 1325 */ 1326void 1327inssym(bl, sym) 1328 int bl; 1329 sym_t *sym; 1330{ 1331 int h; 1332 1333 h = hash(sym->s_name); 1334 if ((sym->s_link = symtab[h]) != NULL) 1335 symtab[h]->s_rlink = &sym->s_link; 1336 (symtab[h] = sym)->s_rlink = &symtab[h]; 1337 sym->s_blklev = bl; 1338 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1339 lerror("inssym()"); 1340} 1341 1342/* 1343 * Called at level 0 after syntax errors 1344 * Removes all symbols which are not declared at level 0 from the 1345 * symbol table. Also frees all memory which is not associated with 1346 * level 0. 1347 */ 1348void 1349cleanup() 1350{ 1351 sym_t *sym, *nsym; 1352 int i; 1353 1354 for (i = 0; i < HSHSIZ1; i++) { 1355 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1356 nsym = sym->s_link; 1357 if (sym->s_blklev >= 1) { 1358 if ((*sym->s_rlink = nsym) != NULL) 1359 nsym->s_rlink = sym->s_rlink; 1360 } 1361 } 1362 } 1363 1364 for (i = mblklev; i > 0; i--) 1365 freelblk(i); 1366} 1367 1368/* 1369 * Create a new symbol with the name of an existing symbol. 1370 */ 1371sym_t * 1372pushdown(sym) 1373 sym_t *sym; 1374{ 1375 int h; 1376 sym_t *nsym; 1377 1378 h = hash(sym->s_name); 1379 nsym = getblk(sizeof (sym_t)); 1380 if (sym->s_blklev > blklev) 1381 lerror("pushdown()"); 1382 nsym->s_name = sym->s_name; 1383 STRUCT_ASSIGN(nsym->s_dpos, curr_pos); 1384 nsym->s_kind = sym->s_kind; 1385 nsym->s_blklev = blklev; 1386 1387 if ((nsym->s_link = symtab[h]) != NULL) 1388 symtab[h]->s_rlink = &nsym->s_link; 1389 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1390 1391 *dcs->d_ldlsym = nsym; 1392 dcs->d_ldlsym = &nsym->s_dlnxt; 1393 1394 return (nsym); 1395} 1396 1397/* 1398 * Free any dynamically allocated memory referenced by 1399 * the value stack or yylval. 1400 * The type of information in yylval is described by tok. 1401 */ 1402void 1403freeyyv(sp, tok) 1404 void *sp; 1405 int tok; 1406{ 1407 if (tok == T_NAME || tok == T_TYPENAME) { 1408 sbuf_t *sb = *(sbuf_t **)sp; 1409 freesb(sb); 1410 } else if (tok == T_CON) { 1411 val_t *val = *(val_t **)sp; 1412 free(val); 1413 } else if (tok == T_STRING) { 1414 strg_t *strg = *(strg_t **)sp; 1415 if (strg->st_tspec == CHAR) { 1416 free(strg->st_cp); 1417 } else if (strg->st_tspec == WCHAR) { 1418 free(strg->st_wcp); 1419 } else { 1420 lerror("fryylv() 1"); 1421 } 1422 free(strg); 1423 } 1424} 1425