1 %{ 2 /* $NetBSD: scan.l,v 1.10 1997/11/03 22:36:50 cgd Exp $ */ 3 4 /* 5 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved. 6 * Copyright (c) 1994, 1995 Jochen Pohl 7 * All Rights Reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed by Jochen Pohl for 20 * The NetBSD Project. 21 * 4. The name of the author may not be used to endorse or promote products 22 * derived from this software without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 27 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 28 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 29 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 30 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 31 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 32 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 33 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 #ifndef lint 37 static char rcsid[] = "$NetBSD: scan.l,v 1.10 1997/11/03 22:36:50 cgd Exp $"; 38 #endif 39 40 #include <stdlib.h> 41 #include <string.h> 42 #include <limits.h> 43 #include <float.h> 44 #include <ctype.h> 45 #include <errno.h> 46 #include <math.h> 47 #include <err.h> 48 49 #include "lint1.h" 50 #include "y.tab.h" 51 52 #define CHAR_MASK (~(~0 << CHAR_BIT)) 53 54 /* XXX declaration of strtouq() is missing in stdlib.h ? */ 55 extern u_quad_t strtouq __P((const char *, char **, int)); 56 57 /* Current position (its also updated when an included file is parsed) */ 58 pos_t curr_pos = { 1, "", 0 }; 59 60 /* 61 * Current position in C source (not updated when an included file is 62 * parsed). 63 */ 64 pos_t csrc_pos = { 1, "", 0 }; 65 66 static void incline __P((void)); 67 static void badchar __P((int)); 68 static sbuf_t *allocsb __P((void)); 69 static void freesb __P((sbuf_t *)); 70 static int inpc __P((void)); 71 static int hash __P((const char *)); 72 static sym_t *search __P((sbuf_t *)); 73 static int name __P((void)); 74 static int keyw __P((sym_t *)); 75 static int icon __P((int)); 76 static int fcon __P((void)); 77 static int operator __P((int, op_t)); 78 static int ccon __P((void)); 79 static int wccon __P((void)); 80 static int getescc __P((int)); 81 static void directive __P((void)); 82 static void comment __P((void)); 83 static int string __P((void)); 84 static int wcstrg __P((void)); 85 86 %} 87 88 L [_A-Za-z] 89 D [0-9] 90 NZD [1-9] 91 OD [0-7] 92 HD [0-9A-Fa-f] 93 EX ([eE][+-]?[0-9]+) 94 95 %% 96 97 {L}({L}|{D})* return (name()); 98 0{OD}*[lLuU]* return (icon(8)); 99 {NZD}{D}*[lLuU]* return (icon(10)); 100 0[xX]{HD}+[lLuU]* return (icon(16)); 101 {D}+\.{D}*{EX}?[fFlL]? | 102 {D}+{EX}[fFlL]? | 103 \.{D}+{EX}?[fFlL]? return (fcon()); 104 "=" return (operator(T_ASSIGN, ASSIGN)); 105 "*=" return (operator(T_OPASS, MULASS)); 106 "/=" return (operator(T_OPASS, DIVASS)); 107 "%=" return (operator(T_OPASS, MODASS)); 108 "+=" return (operator(T_OPASS, ADDASS)); 109 "-=" return (operator(T_OPASS, SUBASS)); 110 "<<=" return (operator(T_OPASS, SHLASS)); 111 ">>=" return (operator(T_OPASS, SHRASS)); 112 "&=" return (operator(T_OPASS, ANDASS)); 113 "^=" return (operator(T_OPASS, XORASS)); 114 "|=" return (operator(T_OPASS, ORASS)); 115 "||" return (operator(T_LOGOR, LOGOR)); 116 "&&" return (operator(T_LOGAND, LOGAND)); 117 "|" return (operator(T_OR, OR)); 118 "&" return (operator(T_AND, AND)); 119 "^" return (operator(T_XOR, XOR)); 120 "==" return (operator(T_EQOP, EQ)); 121 "!=" return (operator(T_EQOP, NE)); 122 "<" return (operator(T_RELOP, LT)); 123 ">" return (operator(T_RELOP, GT)); 124 "<=" return (operator(T_RELOP, LE)); 125 ">=" return (operator(T_RELOP, GE)); 126 "<<" return (operator(T_SHFTOP, SHL)); 127 ">>" return (operator(T_SHFTOP, SHR)); 128 "++" return (operator(T_INCDEC, INC)); 129 "--" return (operator(T_INCDEC, DEC)); 130 "->" return (operator(T_STROP, ARROW)); 131 "." return (operator(T_STROP, POINT)); 132 "+" return (operator(T_ADDOP, PLUS)); 133 "-" return (operator(T_ADDOP, MINUS)); 134 "*" return (operator(T_MULT, MULT)); 135 "/" return (operator(T_DIVOP, DIV)); 136 "%" return (operator(T_DIVOP, MOD)); 137 "!" return (operator(T_UNOP, NOT)); 138 "~" return (operator(T_UNOP, COMPL)); 139 "\"" return (string()); 140 "L\"" return (wcstrg()); 141 ";" return (T_SEMI); 142 "{" return (T_LBRACE); 143 "}" return (T_RBRACE); 144 "," return (T_COMMA); 145 ":" return (T_COLON); 146 "?" return (T_QUEST); 147 "[" return (T_LBRACK); 148 "]" return (T_RBRACK); 149 "(" return (T_LPARN); 150 ")" return (T_RPARN); 151 "..." return (T_ELLIPSE); 152 "'" return (ccon()); 153 "L'" return (wccon()); 154 ^#.*$ directive(); 155 \n incline(); 156 \t|" "|\f|\v ; 157 "/*" comment(); 158 . badchar(yytext[0]); 159 160 %% 161 162 static void 163 incline() 164 { 165 curr_pos.p_line++; 166 curr_pos.p_uniq = 0; 167 if (curr_pos.p_file == csrc_pos.p_file) { 168 csrc_pos.p_line++; 169 csrc_pos.p_uniq = 0; 170 } 171 } 172 173 static void 174 badchar(c) 175 int c; 176 { 177 /* unknown character \%o */ 178 error(250, c); 179 } 180 181 /* 182 * Keywords. 183 * During initialisation they are written to the symbol table. 184 */ 185 static struct kwtab { 186 const char *kw_name; /* keyword */ 187 int kw_token; /* token returned by yylex() */ 188 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 189 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 190 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 191 u_int kw_stdc : 1; /* STDC keyword */ 192 u_int kw_gcc : 1; /* GCC keyword */ 193 } kwtab[] = { 194 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 195 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 196 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 197 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 198 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 199 { "case", T_CASE, 0, 0, 0, 0, 0 }, 200 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 201 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 202 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 203 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 204 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 205 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 206 { "do", T_DO, 0, 0, 0, 0, 0 }, 207 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 208 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 209 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 210 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 211 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 212 { "for", T_FOR, 0, 0, 0, 0, 0 }, 213 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 214 { "if", T_IF, 0, 0, 0, 0, 0 }, 215 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 216 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 217 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 218 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 219 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 }, 220 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 221 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 222 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 223 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 224 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 225 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 226 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 227 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 228 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 229 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 230 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 231 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 232 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 233 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 234 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 235 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 236 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 237 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 238 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 239 { NULL, 0, 0, 0, 0, 0, 0 } 240 }; 241 242 /* Symbol table */ 243 static sym_t *symtab[HSHSIZ1]; 244 245 /* bit i of the entry with index i is set */ 246 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 247 248 /* least significant i bits are set in the entry with index i */ 249 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 250 251 /* least significant i bits are not set in the entry with index i */ 252 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 253 254 /* free list for sbuf structures */ 255 static sbuf_t *sbfrlst; 256 257 /* Typ of next expected symbol */ 258 symt_t symtyp; 259 260 261 /* 262 * All keywords are written to the symbol table. This saves us looking 263 * in a extra table for each name we found. 264 */ 265 void 266 initscan() 267 { 268 struct kwtab *kw; 269 sym_t *sym; 270 int h, i; 271 u_quad_t uq; 272 273 for (kw = kwtab; kw->kw_name != NULL; kw++) { 274 if (kw->kw_stdc && tflag) 275 continue; 276 if (kw->kw_gcc && !gflag) 277 continue; 278 sym = getblk(sizeof (sym_t)); 279 sym->s_name = kw->kw_name; 280 sym->s_keyw = 1; 281 sym->s_value.v_quad = kw->kw_token; 282 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 283 sym->s_tspec = kw->kw_tspec; 284 } else if (kw->kw_token == T_SCLASS) { 285 sym->s_scl = kw->kw_scl; 286 } else if (kw->kw_token == T_QUAL) { 287 sym->s_tqual = kw->kw_tqual; 288 } 289 h = hash(sym->s_name); 290 if ((sym->s_link = symtab[h]) != NULL) 291 symtab[h]->s_rlink = &sym->s_link; 292 (symtab[h] = sym)->s_rlink = &symtab[h]; 293 } 294 295 /* initialize bit-masks for quads */ 296 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 297 qbmasks[i] = (u_quad_t)1 << i; 298 uq = ~(u_quad_t)0 << i; 299 qumasks[i] = uq; 300 qlmasks[i] = ~uq; 301 } 302 qumasks[i] = 0; 303 qlmasks[i] = ~(u_quad_t)0; 304 } 305 306 /* 307 * Get a free sbuf structure, if possible from the free list 308 */ 309 static sbuf_t * 310 allocsb() 311 { 312 sbuf_t *sb; 313 314 if ((sb = sbfrlst) != NULL) { 315 sbfrlst = sb->sb_nxt; 316 } else { 317 sb = xmalloc(sizeof (sbuf_t)); 318 } 319 (void)memset(sb, 0, sizeof (sb)); 320 return (sb); 321 } 322 323 /* 324 * Put a sbuf structure to the free list 325 */ 326 static void 327 freesb(sb) 328 sbuf_t *sb; 329 { 330 sb->sb_nxt = sbfrlst; 331 sbfrlst = sb; 332 } 333 334 /* 335 * Read a character and ensure that it is positive (except EOF). 336 * Increment line count(s) if necessary. 337 */ 338 static int 339 inpc() 340 { 341 int c; 342 343 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 344 incline(); 345 return (c); 346 } 347 348 static int 349 hash(s) 350 const char *s; 351 { 352 u_int v; 353 const u_char *us; 354 355 v = 0; 356 for (us = (const u_char *)s; *us != '\0'; us++) { 357 v = (v << sizeof (v)) + *us; 358 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 359 } 360 return (v % HSHSIZ1); 361 } 362 363 /* 364 * Lex has found a letter followed by zero or more letters or digits. 365 * It looks for a symbol in the symbol table with the same name. This 366 * symbol must either be a keyword or a symbol of the type required by 367 * symtyp (label, member, tag, ...). 368 * 369 * If it is a keyword, the token is returned. In some cases it is described 370 * more deeply by data written to yylval. 371 * 372 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 373 * is stored in yylval. This struct contains the name of the symbol, it's 374 * length and hash value. If there is already a symbol of the same name 375 * and type in the symbol table, the sbuf struct also contains a pointer 376 * to the symbol table entry. 377 */ 378 static int 379 name() 380 { 381 char *s; 382 sbuf_t *sb; 383 sym_t *sym; 384 int tok; 385 386 sb = allocsb(); 387 sb->sb_name = yytext; 388 sb->sb_len = yyleng; 389 sb->sb_hash = hash(yytext); 390 391 if ((sym = search(sb)) != NULL && sym->s_keyw) { 392 freesb(sb); 393 return (keyw(sym)); 394 } 395 396 sb->sb_sym = sym; 397 398 if (sym != NULL) { 399 if (blklev < sym->s_blklev) 400 lerror("name() 1"); 401 sb->sb_name = sym->s_name; 402 sb->sb_len = strlen(sym->s_name); 403 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 404 } else { 405 s = getblk(yyleng + 1); 406 (void)memcpy(s, yytext, yyleng + 1); 407 sb->sb_name = s; 408 sb->sb_len = yyleng; 409 tok = T_NAME; 410 } 411 412 yylval.y_sb = sb; 413 return (tok); 414 } 415 416 static sym_t * 417 search(sb) 418 sbuf_t *sb; 419 { 420 sym_t *sym; 421 422 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 423 if (strcmp(sym->s_name, sb->sb_name) == 0) { 424 if (sym->s_keyw || sym->s_kind == symtyp) 425 return (sym); 426 } 427 } 428 429 return (NULL); 430 } 431 432 static int 433 keyw(sym) 434 sym_t *sym; 435 { 436 int t; 437 438 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 439 yylval.y_scl = sym->s_scl; 440 } else if (t == T_TYPE || t == T_SOU) { 441 yylval.y_tspec = sym->s_tspec; 442 } else if (t == T_QUAL) { 443 yylval.y_tqual = sym->s_tqual; 444 } 445 return (t); 446 } 447 448 /* 449 * Convert a string representing an integer into internal representation. 450 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 451 */ 452 static int 453 icon(base) 454 int base; 455 { 456 int l_suffix, u_suffix; 457 int len; 458 const char *cp; 459 char c, *eptr; 460 tspec_t typ; 461 u_long ul; 462 u_quad_t uq; 463 int ansiu; 464 static tspec_t contypes[2][3] = { 465 { INT, LONG, QUAD }, 466 { UINT, ULONG, UQUAD } 467 }; 468 469 cp = yytext; 470 len = yyleng; 471 472 /* skip 0x */ 473 if (base == 16) { 474 cp += 2; 475 len -= 2; 476 } 477 478 /* read suffixes */ 479 l_suffix = u_suffix = 0; 480 for ( ; ; ) { 481 if ((c = cp[len - 1]) == 'l' || c == 'L') { 482 l_suffix++; 483 } else if (c == 'u' || c == 'U') { 484 u_suffix++; 485 } else { 486 break; 487 } 488 len--; 489 } 490 if (l_suffix > 2 || u_suffix > 1) { 491 /* malformed integer constant */ 492 warning(251); 493 if (l_suffix > 2) 494 l_suffix = 2; 495 if (u_suffix > 1) 496 u_suffix = 1; 497 } 498 if (tflag && u_suffix != 0) { 499 /* suffix U is illegal in traditional C */ 500 warning(97); 501 } 502 typ = contypes[u_suffix][l_suffix]; 503 504 errno = 0; 505 if (l_suffix < 2) { 506 ul = strtoul(cp, &eptr, base); 507 } else { 508 uq = strtouq(cp, &eptr, base); 509 } 510 if (eptr != cp + len) 511 lerror("icon() 1"); 512 if (errno != 0) 513 /* integer constant out of range */ 514 warning(252); 515 516 /* 517 * If the value is to big for the current type, we must choose 518 * another type. 519 */ 520 ansiu = 0; 521 switch (typ) { 522 case INT: 523 if (ul <= INT_MAX) { 524 /* ok */ 525 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 526 typ = UINT; 527 } else if (ul <= LONG_MAX) { 528 typ = LONG; 529 } else { 530 typ = ULONG; 531 } 532 if (typ == UINT || typ == ULONG) { 533 if (tflag) { 534 typ = LONG; 535 } else if (!sflag) { 536 /* 537 * Remember that the constant is unsigned 538 * only in ANSI C 539 */ 540 ansiu = 1; 541 } 542 } 543 break; 544 case UINT: 545 if (ul > (u_int)UINT_MAX) 546 typ = ULONG; 547 break; 548 case LONG: 549 if (ul > LONG_MAX && !tflag) { 550 typ = ULONG; 551 if (!sflag) 552 ansiu = 1; 553 } 554 break; 555 case QUAD: 556 if (uq > QUAD_MAX && !tflag) { 557 typ = UQUAD; 558 if (!sflag) 559 ansiu = 1; 560 } 561 break; 562 /* LINTED (enumeration values not handled in switch) */ 563 } 564 565 if (typ != QUAD && typ != UQUAD) { 566 if (isutyp(typ)) { 567 uq = ul; 568 } else { 569 uq = (quad_t)(long)ul; 570 } 571 } 572 573 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 574 575 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 576 yylval.y_val->v_ansiu = ansiu; 577 yylval.y_val->v_quad = (quad_t)uq; 578 579 return (T_CON); 580 } 581 582 /* 583 * Returns 1 if t is a signed type and the value is negative. 584 * 585 * len is the number of significant bits. If len is -1, len is set 586 * to the width of type t. 587 */ 588 int 589 sign(q, t, len) 590 quad_t q; 591 tspec_t t; 592 int len; 593 { 594 if (t == PTR || isutyp(t)) 595 return (0); 596 return (msb(q, t, len)); 597 } 598 599 int 600 msb(q, t, len) 601 quad_t q; 602 tspec_t t; 603 int len; 604 { 605 if (len <= 0) 606 len = size(t); 607 return ((q & qbmasks[len - 1]) != 0); 608 } 609 610 /* 611 * Extends the sign of q. 612 */ 613 quad_t 614 xsign(q, t, len) 615 quad_t q; 616 tspec_t t; 617 int len; 618 { 619 if (len <= 0) 620 len = size(t); 621 622 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 623 q &= qlmasks[len]; 624 } else { 625 q |= qumasks[len]; 626 } 627 return (q); 628 } 629 630 /* 631 * Convert a string representing a floating point value into its interal 632 * representation. Type and value are returned in yylval. fcon() 633 * (and yylex()) returns T_CON. 634 * XXX Currently it is not possible to convert constants of type 635 * long double which are greater then DBL_MAX. 636 */ 637 static int 638 fcon() 639 { 640 const char *cp; 641 int len; 642 tspec_t typ; 643 char c, *eptr; 644 double d; 645 float f; 646 647 cp = yytext; 648 len = yyleng; 649 650 if ((c = cp[len - 1]) == 'f' || c == 'F') { 651 typ = FLOAT; 652 len--; 653 } else if (c == 'l' || c == 'L') { 654 typ = LDOUBLE; 655 len--; 656 } else { 657 typ = DOUBLE; 658 } 659 660 if (tflag && typ != DOUBLE) { 661 /* suffixes F and L are illegal in traditional C */ 662 warning(98); 663 } 664 665 errno = 0; 666 d = strtod(cp, &eptr); 667 if (eptr != cp + len) 668 lerror("fcon() 1"); 669 if (errno != 0) 670 /* floating-point constant out of range */ 671 warning(248); 672 673 if (typ == FLOAT) { 674 f = (float)d; 675 if (isinf(f)) { 676 /* floating-point constant out of range */ 677 warning(248); 678 f = f > 0 ? FLT_MAX : -FLT_MAX; 679 } 680 } 681 682 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 683 if (typ == FLOAT) { 684 yylval.y_val->v_ldbl = f; 685 } else { 686 yylval.y_val->v_ldbl = d; 687 } 688 689 return (T_CON); 690 } 691 692 static int 693 operator(t, o) 694 int t; 695 op_t o; 696 { 697 yylval.y_op = o; 698 return (t); 699 } 700 701 /* 702 * Called if lex found a leading \'. 703 */ 704 static int 705 ccon() 706 { 707 int n, val, c; 708 char cv; 709 710 n = 0; 711 val = 0; 712 while ((c = getescc('\'')) >= 0) { 713 val = (val << CHAR_BIT) + c; 714 n++; 715 } 716 if (c == -2) { 717 /* unterminated character constant */ 718 error(253); 719 } else { 720 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 721 /* too many characters in character constant */ 722 error(71); 723 } else if (n > 1) { 724 /* multi-character character constant */ 725 warning(294); 726 } else if (n == 0) { 727 /* empty character constant */ 728 error(73); 729 } 730 } 731 if (n == 1) { 732 cv = (char)val; 733 val = cv; 734 } 735 736 yylval.y_val = xcalloc(1, sizeof (val_t)); 737 yylval.y_val->v_tspec = INT; 738 yylval.y_val->v_quad = val; 739 740 return (T_CON); 741 } 742 743 /* 744 * Called if lex found a leading L\' 745 */ 746 static int 747 wccon() 748 { 749 static char buf[MB_LEN_MAX + 1]; 750 int i, c; 751 wchar_t wc; 752 753 i = 0; 754 while ((c = getescc('\'')) >= 0) { 755 if (i < MB_CUR_MAX) 756 buf[i] = (char)c; 757 i++; 758 } 759 760 wc = 0; 761 762 if (c == -2) { 763 /* unterminated character constant */ 764 error(253); 765 } else if (c == 0) { 766 /* empty character constant */ 767 error(73); 768 } else { 769 if (i > MB_CUR_MAX) { 770 i = MB_CUR_MAX; 771 /* too many characters in character constant */ 772 error(71); 773 } else { 774 buf[i] = '\0'; 775 (void)mbtowc(NULL, NULL, 0); 776 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 777 /* invalid multibyte character */ 778 error(291); 779 } 780 } 781 782 yylval.y_val = xcalloc(1, sizeof (val_t)); 783 yylval.y_val->v_tspec = WCHAR; 784 yylval.y_val->v_quad = wc; 785 786 return (T_CON); 787 } 788 789 /* 790 * Read a character which is part of a character constant or of a string 791 * and handle escapes. 792 * 793 * The Argument is the character which delimits the character constant or 794 * string. 795 * 796 * Returns -1 if the end of the character constant or string is reached, 797 * -2 if the EOF is reached, and the charachter otherwise. 798 */ 799 static int 800 getescc(d) 801 int d; 802 { 803 static int pbc = -1; 804 int n, c, v; 805 806 if (pbc == -1) { 807 c = inpc(); 808 } else { 809 c = pbc; 810 pbc = -1; 811 } 812 if (c == d) 813 return (-1); 814 switch (c) { 815 case '\n': 816 /* newline in string or char constant */ 817 error(254); 818 return (-2); 819 case EOF: 820 return (-2); 821 case '\\': 822 switch (c = inpc()) { 823 case '"': 824 if (tflag && d == '\'') 825 /* \" inside character constant undef. ... */ 826 warning(262); 827 return ('"'); 828 case '\'': 829 return ('\''); 830 case '?': 831 if (tflag) 832 /* \? undefined in traditional C */ 833 warning(263); 834 return ('?'); 835 case '\\': 836 return ('\\'); 837 case 'a': 838 if (tflag) 839 /* \a undefined in traditional C */ 840 warning(81); 841 #ifdef __STDC__ 842 return ('\a'); 843 #else 844 return ('\007'); 845 #endif 846 case 'b': 847 return ('\b'); 848 case 'f': 849 return ('\f'); 850 case 'n': 851 return ('\n'); 852 case 'r': 853 return ('\r'); 854 case 't': 855 return ('\t'); 856 case 'v': 857 if (tflag) 858 /* \v undefined in traditional C */ 859 warning(264); 860 #ifdef __STDC__ 861 return ('\v'); 862 #else 863 return ('\013'); 864 #endif 865 case '8': case '9': 866 /* bad octal digit %c */ 867 warning(77, c); 868 /* FALLTHROUGH */ 869 case '0': case '1': case '2': case '3': 870 case '4': case '5': case '6': case '7': 871 n = 3; 872 v = 0; 873 do { 874 v = (v << 3) + (c - '0'); 875 c = inpc(); 876 } while (--n && isdigit(c) && (tflag || c <= '7')); 877 if (tflag && n > 0 && isdigit(c)) 878 /* bad octal digit %c */ 879 warning(77, c); 880 pbc = c; 881 if (v > UCHAR_MAX) { 882 /* character escape does not fit in char. */ 883 warning(76); 884 v &= CHAR_MASK; 885 } 886 return (v); 887 case 'x': 888 if (tflag) 889 /* \x undefined in traditional C */ 890 warning(82); 891 v = 0; 892 n = 0; 893 while ((c = inpc()) >= 0 && isxdigit(c)) { 894 c = isdigit(c) ? 895 c - '0' : toupper(c) - 'A' + 10; 896 v = (v << 4) + c; 897 if (n >= 0) { 898 if ((v & ~CHAR_MASK) != 0) { 899 /* overflow in hex escape */ 900 warning(75); 901 n = -1; 902 } else { 903 n++; 904 } 905 } 906 } 907 pbc = c; 908 if (n == 0) { 909 /* no hex digits follow \x */ 910 error(74); 911 } if (n == -1) { 912 v &= CHAR_MASK; 913 } 914 return (v); 915 case '\n': 916 return (getescc(d)); 917 case EOF: 918 return (-2); 919 default: 920 if (isprint(c)) { 921 /* dubious escape \%c */ 922 warning(79, c); 923 } else { 924 /* dubious escape \%o */ 925 warning(80, c); 926 } 927 } 928 } 929 return (c); 930 } 931 932 /* 933 * Called for preprocessor directives. Currently implemented are: 934 * # lineno 935 * # lineno "filename" 936 */ 937 static void 938 directive() 939 { 940 const char *cp, *fn; 941 char c, *eptr; 942 size_t fnl; 943 long ln; 944 static int first = 1; 945 946 /* Go to first non-whitespace after # */ 947 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 948 949 if (!isdigit(c)) { 950 error: 951 /* undefined or invalid # directive */ 952 warning(255); 953 return; 954 } 955 ln = strtol(--cp, &eptr, 10); 956 if (cp == eptr) 957 goto error; 958 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 959 goto error; 960 while ((c = *cp++) == ' ' || c == '\t') ; 961 if (c != '\0') { 962 if (c != '"') 963 goto error; 964 fn = cp; 965 while ((c = *cp) != '"' && c != '\0') 966 cp++; 967 if (c != '"') 968 goto error; 969 if ((fnl = cp++ - fn) > PATH_MAX) 970 goto error; 971 while ((c = *cp++) == ' ' || c == '\t') ; 972 #if 0 973 if (c != '\0') 974 warning("extra character(s) after directive"); 975 #endif 976 977 /* empty string means stdin */ 978 if (fnl == 0) { 979 fn = "{standard input}"; 980 fnl = 16; /* strlen (fn) */ 981 } 982 curr_pos.p_file = fnnalloc(fn, fnl); 983 /* 984 * If this is the first directive, the name is the name 985 * of the C source file as specified at the command line. 986 * It is written to the output file. 987 */ 988 if (first) { 989 csrc_pos.p_file = curr_pos.p_file; 990 outsrc(curr_pos.p_file); 991 first = 0; 992 } 993 } 994 curr_pos.p_line = (int)ln - 1; 995 curr_pos.p_uniq = 0; 996 if (curr_pos.p_file == csrc_pos.p_file) { 997 csrc_pos.p_line = (int)ln - 1; 998 csrc_pos.p_uniq = 0; 999 } 1000 } 1001 1002 /* 1003 * Handle lint comments. Following comments are currently understood: 1004 * ARGSUSEDn 1005 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1006 * FALLTHRU FALLTHROUGH 1007 * LINTLIBRARY 1008 * LINTED NOSTRICT 1009 * LONGLONG 1010 * NOTREACHED 1011 * PRINTFLIKEn 1012 * PROTOLIB 1013 * SCANFLIKEn 1014 * VARARGSn 1015 * If one of this comments is recognized, the arguments, if any, are 1016 * parsed and a function which handles this comment is called. 1017 */ 1018 static void 1019 comment() 1020 { 1021 int c, lc; 1022 static struct { 1023 const char *keywd; 1024 int arg; 1025 void (*func) __P((int)); 1026 } keywtab[] = { 1027 { "ARGSUSED", 1, argsused }, 1028 { "CONSTCOND", 0, constcond }, 1029 { "CONSTANTCOND", 0, constcond }, 1030 { "CONSTANTCONDITION", 0, constcond }, 1031 { "FALLTHRU", 0, fallthru }, 1032 { "FALLTHROUGH", 0, fallthru }, 1033 { "LINTLIBRARY", 0, lintlib }, 1034 { "LINTED", 0, linted }, 1035 { "LONGLONG", 0, longlong }, 1036 { "NOSTRICT", 0, linted }, 1037 { "NOTREACHED", 0, notreach }, 1038 { "PRINTFLIKE", 1, printflike }, 1039 { "PROTOLIB", 1, protolib }, 1040 { "SCANFLIKE", 1, scanflike }, 1041 { "VARARGS", 1, varargs }, 1042 }; 1043 char keywd[32]; 1044 char arg[32]; 1045 int l, i, a; 1046 int eoc; 1047 1048 eoc = 0; 1049 1050 /* Skip white spaces after the start of the comment */ 1051 while ((c = inpc()) != EOF && isspace(c)) ; 1052 1053 /* Read the potential keyword to keywd */ 1054 l = 0; 1055 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1056 keywd[l++] = (char)c; 1057 c = inpc(); 1058 } 1059 keywd[l] = '\0'; 1060 1061 /* look for the keyword */ 1062 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1063 if (strcmp(keywtab[i].keywd, keywd) == 0) 1064 break; 1065 } 1066 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1067 goto skip_rest; 1068 1069 /* skip white spaces after the keyword */ 1070 while (c != EOF && isspace(c)) 1071 c = inpc(); 1072 1073 /* read the argument, if the keyword accepts one and there is one */ 1074 l = 0; 1075 if (keywtab[i].arg) { 1076 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1077 arg[l++] = (char)c; 1078 c = inpc(); 1079 } 1080 } 1081 arg[l] = '\0'; 1082 a = l != 0 ? atoi(arg) : -1; 1083 1084 /* skip white spaces after the argument */ 1085 while (c != EOF && isspace(c)) 1086 c = inpc(); 1087 1088 if (c != '*' || (c = inpc()) != '/') { 1089 if (keywtab[i].func != linted) 1090 /* extra characters in lint comment */ 1091 warning(257); 1092 } else { 1093 /* 1094 * remember that we have already found the end of the 1095 * comment 1096 */ 1097 eoc = 1; 1098 } 1099 1100 if (keywtab[i].func != NULL) 1101 (*keywtab[i].func)(a); 1102 1103 skip_rest: 1104 while (!eoc) { 1105 lc = c; 1106 if ((c = inpc()) == EOF) { 1107 /* unterminated comment */ 1108 error(256); 1109 break; 1110 } 1111 if (lc == '*' && c == '/') 1112 eoc = 1; 1113 } 1114 } 1115 1116 /* 1117 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1118 * clrwflgs() is called after function definitions and global and 1119 * local declarations and definitions. It is also called between 1120 * the controlling expression and the body of control statements 1121 * (if, switch, for, while). 1122 */ 1123 void 1124 clrwflgs() 1125 { 1126 nowarn = 0; 1127 quadflg = 0; 1128 ccflg = 0; 1129 } 1130 1131 /* 1132 * Strings are stored in a dynamically alloceted buffer and passed 1133 * in yylval.y_xstrg to the parser. The parser or the routines called 1134 * by the parser are responsible for freeing this buffer. 1135 */ 1136 static int 1137 string() 1138 { 1139 u_char *s; 1140 int c; 1141 size_t len, max; 1142 strg_t *strg; 1143 1144 s = xmalloc(max = 64); 1145 1146 len = 0; 1147 while ((c = getescc('"')) >= 0) { 1148 /* +1 to reserve space for a trailing NUL character */ 1149 if (len + 1 == max) 1150 s = xrealloc(s, max *= 2); 1151 s[len++] = (char)c; 1152 } 1153 s[len] = '\0'; 1154 if (c == -2) 1155 /* unterminated string constant */ 1156 error(258); 1157 1158 strg = xcalloc(1, sizeof (strg_t)); 1159 strg->st_tspec = CHAR; 1160 strg->st_len = len; 1161 strg->st_cp = s; 1162 1163 yylval.y_strg = strg; 1164 return (T_STRING); 1165 } 1166 1167 static int 1168 wcstrg() 1169 { 1170 char *s; 1171 int c, i, n, wi; 1172 size_t len, max, wlen; 1173 wchar_t *ws; 1174 strg_t *strg; 1175 1176 s = xmalloc(max = 64); 1177 len = 0; 1178 while ((c = getescc('"')) >= 0) { 1179 /* +1 to save space for a trailing NUL character */ 1180 if (len + 1 >= max) 1181 s = xrealloc(s, max *= 2); 1182 s[len++] = (char)c; 1183 } 1184 s[len] = '\0'; 1185 if (c == -2) 1186 /* unterminated string constant */ 1187 error(258); 1188 1189 /* get length of wide character string */ 1190 (void)mblen(NULL, 0); 1191 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1192 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1193 /* invalid multibyte character */ 1194 error(291); 1195 break; 1196 } 1197 if (n == 0) 1198 n = 1; 1199 } 1200 1201 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1202 1203 /* convert from multibyte to wide char */ 1204 (void)mbtowc(NULL, NULL, 0); 1205 for (i = 0, wi = 0; i < len; i += n, wi++) { 1206 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1207 break; 1208 if (n == 0) 1209 n = 1; 1210 } 1211 ws[wi] = 0; 1212 free(s); 1213 1214 strg = xcalloc(1, sizeof (strg_t)); 1215 strg->st_tspec = WCHAR; 1216 strg->st_len = wlen; 1217 strg->st_wcp = ws; 1218 1219 yylval.y_strg = strg; 1220 return (T_STRING); 1221 } 1222 1223 /* 1224 * As noted above the scanner does not create new symbol table entries 1225 * for symbols it cannot find in the symbol table. This is to avoid 1226 * putting undeclared symbols into the symbol table if a syntax error 1227 * occurs. 1228 * 1229 * getsym() is called as soon as it is probably ok to put the symbol to 1230 * the symbol table. This does not mean that it is not possible that 1231 * symbols are put to the symbol table which are than not completely 1232 * declared due to syntax errors. To avoid too many problems in this 1233 * case symbols get type int in getsym(). 1234 * 1235 * XXX calls to getsym() should be delayed until decl1*() is called 1236 */ 1237 sym_t * 1238 getsym(sb) 1239 sbuf_t *sb; 1240 { 1241 dinfo_t *di; 1242 char *s; 1243 sym_t *sym; 1244 1245 sym = sb->sb_sym; 1246 1247 /* 1248 * During member declaration it is possible that name() looked 1249 * for symbols of type FVFT, although it should have looked for 1250 * symbols of type FTAG. Same can happen for labels. Both cases 1251 * are compensated here. 1252 */ 1253 if (symtyp == FMOS || symtyp == FLAB) { 1254 if (sym == NULL || sym->s_kind == FVFT) 1255 sym = search(sb); 1256 } 1257 1258 if (sym != NULL) { 1259 if (sym->s_kind != symtyp) 1260 lerror("storesym() 1"); 1261 symtyp = FVFT; 1262 freesb(sb); 1263 return (sym); 1264 } 1265 1266 /* create a new symbol table entry */ 1267 1268 /* labels must always be allocated at level 1 (outhermost block) */ 1269 if (symtyp == FLAB) { 1270 sym = getlblk(1, sizeof (sym_t)); 1271 s = getlblk(1, sb->sb_len + 1); 1272 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1273 sym->s_name = s; 1274 sym->s_blklev = 1; 1275 di = dcs; 1276 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1277 di = di->d_nxt; 1278 if (di->d_ctx != AUTO) 1279 lerror("storesym() 2"); 1280 } else { 1281 sym = getblk(sizeof (sym_t)); 1282 sym->s_name = sb->sb_name; 1283 sym->s_blklev = blklev; 1284 di = dcs; 1285 } 1286 1287 UNIQUE_CURR_POS(sym->s_dpos); 1288 if ((sym->s_kind = symtyp) != FLAB) 1289 sym->s_type = gettyp(INT); 1290 1291 symtyp = FVFT; 1292 1293 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1294 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1295 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1296 1297 *di->d_ldlsym = sym; 1298 di->d_ldlsym = &sym->s_dlnxt; 1299 1300 freesb(sb); 1301 return (sym); 1302 } 1303 1304 /* 1305 * Remove a symbol forever from the symbol table. s_blklev 1306 * is set to -1 to avoid that the symbol will later be put 1307 * back to the symbol table. 1308 */ 1309 void 1310 rmsym(sym) 1311 sym_t *sym; 1312 { 1313 if ((*sym->s_rlink = sym->s_link) != NULL) 1314 sym->s_link->s_rlink = sym->s_rlink; 1315 sym->s_blklev = -1; 1316 sym->s_link = NULL; 1317 } 1318 1319 /* 1320 * Remove a list of symbols declared at one level from the symbol 1321 * table. 1322 */ 1323 void 1324 rmsyms(syms) 1325 sym_t *syms; 1326 { 1327 sym_t *sym; 1328 1329 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1330 if (sym->s_blklev != -1) { 1331 if ((*sym->s_rlink = sym->s_link) != NULL) 1332 sym->s_link->s_rlink = sym->s_rlink; 1333 sym->s_link = NULL; 1334 sym->s_rlink = NULL; 1335 } 1336 } 1337 } 1338 1339 /* 1340 * Put a symbol into the symbol table 1341 */ 1342 void 1343 inssym(bl, sym) 1344 int bl; 1345 sym_t *sym; 1346 { 1347 int h; 1348 1349 h = hash(sym->s_name); 1350 if ((sym->s_link = symtab[h]) != NULL) 1351 symtab[h]->s_rlink = &sym->s_link; 1352 (symtab[h] = sym)->s_rlink = &symtab[h]; 1353 sym->s_blklev = bl; 1354 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1355 lerror("inssym()"); 1356 } 1357 1358 /* 1359 * Called at level 0 after syntax errors 1360 * Removes all symbols which are not declared at level 0 from the 1361 * symbol table. Also frees all memory which is not associated with 1362 * level 0. 1363 */ 1364 void 1365 cleanup() 1366 { 1367 sym_t *sym, *nsym; 1368 int i; 1369 1370 for (i = 0; i < HSHSIZ1; i++) { 1371 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1372 nsym = sym->s_link; 1373 if (sym->s_blklev >= 1) { 1374 if ((*sym->s_rlink = nsym) != NULL) 1375 nsym->s_rlink = sym->s_rlink; 1376 } 1377 } 1378 } 1379 1380 for (i = mblklev; i > 0; i--) 1381 freelblk(i); 1382 } 1383 1384 /* 1385 * Create a new symbol with the name of an existing symbol. 1386 */ 1387 sym_t * 1388 pushdown(sym) 1389 sym_t *sym; 1390 { 1391 int h; 1392 sym_t *nsym; 1393 1394 h = hash(sym->s_name); 1395 nsym = getblk(sizeof (sym_t)); 1396 if (sym->s_blklev > blklev) 1397 lerror("pushdown()"); 1398 nsym->s_name = sym->s_name; 1399 UNIQUE_CURR_POS(nsym->s_dpos); 1400 nsym->s_kind = sym->s_kind; 1401 nsym->s_blklev = blklev; 1402 1403 if ((nsym->s_link = symtab[h]) != NULL) 1404 symtab[h]->s_rlink = &nsym->s_link; 1405 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1406 1407 *dcs->d_ldlsym = nsym; 1408 dcs->d_ldlsym = &nsym->s_dlnxt; 1409 1410 return (nsym); 1411 } 1412 1413 /* 1414 * Free any dynamically allocated memory referenced by 1415 * the value stack or yylval. 1416 * The type of information in yylval is described by tok. 1417 */ 1418 void 1419 freeyyv(sp, tok) 1420 void *sp; 1421 int tok; 1422 { 1423 if (tok == T_NAME || tok == T_TYPENAME) { 1424 sbuf_t *sb = *(sbuf_t **)sp; 1425 freesb(sb); 1426 } else if (tok == T_CON) { 1427 val_t *val = *(val_t **)sp; 1428 free(val); 1429 } else if (tok == T_STRING) { 1430 strg_t *strg = *(strg_t **)sp; 1431 if (strg->st_tspec == CHAR) { 1432 free(strg->st_cp); 1433 } else if (strg->st_tspec == WCHAR) { 1434 free(strg->st_wcp); 1435 } else { 1436 lerror("fryylv() 1"); 1437 } 1438 free(strg); 1439 } 1440 } 1441