scan.l revision 1.19
1 %{ 2 /* $NetBSD: scan.l,v 1.19 2001/05/28 12:40:37 lukem 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 #include <sys/cdefs.h> 37 #ifndef lint 38 __RCSID("$NetBSD: scan.l,v 1.19 2001/05/28 12:40:37 lukem Exp $"); 39 #endif 40 41 #include <stdlib.h> 42 #include <string.h> 43 #include <limits.h> 44 #include <float.h> 45 #include <ctype.h> 46 #include <errno.h> 47 #include <math.h> 48 #include <err.h> 49 50 #include "lint1.h" 51 #include "cgram.h" 52 53 #define CHAR_MASK (~(~0 << CHAR_BIT)) 54 #define YY_NO_UNPUT 55 56 /* Current position (its also updated when an included file is parsed) */ 57 pos_t curr_pos = { 1, "", 0 }; 58 59 /* 60 * Current position in C source (not updated when an included file is 61 * parsed). 62 */ 63 pos_t csrc_pos = { 1, "", 0 }; 64 65 static void incline(void); 66 static void badchar(int); 67 static sbuf_t *allocsb(void); 68 static void freesb(sbuf_t *); 69 static int inpc(void); 70 static int hash(const char *); 71 static sym_t *search(sbuf_t *); 72 static int name(void); 73 static int keyw(sym_t *); 74 static int icon(int); 75 static int fcon(void); 76 static int operator(int, op_t); 77 static int ccon(void); 78 static int wccon(void); 79 static int getescc(int); 80 static void directive(void); 81 static void comment(void); 82 static void slashslashcomment(void); 83 static int string(void); 84 static int wcstrg(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 "//" slashslashcomment(); 159 . badchar(yytext[0]); 160 161 %% 162 163 static void 164 incline(void) 165 { 166 curr_pos.p_line++; 167 curr_pos.p_uniq = 0; 168 if (curr_pos.p_file == csrc_pos.p_file) { 169 csrc_pos.p_line++; 170 csrc_pos.p_uniq = 0; 171 } 172 } 173 174 static void 175 badchar(int c) 176 { 177 178 /* unknown character \%o */ 179 error(250, c); 180 } 181 182 /* 183 * Keywords. 184 * During initialisation they are written to the symbol table. 185 */ 186 static struct kwtab { 187 const char *kw_name; /* keyword */ 188 int kw_token; /* token returned by yylex() */ 189 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 190 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 191 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 192 u_int kw_stdc : 1; /* STDC keyword */ 193 u_int kw_gcc : 1; /* GCC keyword */ 194 } kwtab[] = { 195 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 196 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 197 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 198 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 199 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 200 { "case", T_CASE, 0, 0, 0, 0, 0 }, 201 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 202 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 203 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 204 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 205 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 206 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 207 { "do", T_DO, 0, 0, 0, 0, 0 }, 208 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 209 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 210 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 211 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 212 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 213 { "for", T_FOR, 0, 0, 0, 0, 0 }, 214 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 215 { "if", T_IF, 0, 0, 0, 0, 0 }, 216 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 217 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 218 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 219 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 220 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 }, 221 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 222 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 223 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 224 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 225 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 226 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 227 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 228 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 229 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 230 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 231 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 232 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 233 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 234 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 235 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 236 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 237 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 238 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 239 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 240 { NULL, 0, 0, 0, 0, 0, 0 } 241 }; 242 243 /* Symbol table */ 244 static sym_t *symtab[HSHSIZ1]; 245 246 /* bit i of the entry with index i is set */ 247 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 248 249 /* least significant i bits are set in the entry with index i */ 250 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 251 252 /* least significant i bits are not set in the entry with index i */ 253 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 254 255 /* free list for sbuf structures */ 256 static sbuf_t *sbfrlst; 257 258 /* Typ of next expected symbol */ 259 symt_t symtyp; 260 261 262 /* 263 * All keywords are written to the symbol table. This saves us looking 264 * in a extra table for each name we found. 265 */ 266 void 267 initscan(void) 268 { 269 struct kwtab *kw; 270 sym_t *sym; 271 int h, i; 272 u_quad_t uq; 273 274 for (kw = kwtab; kw->kw_name != NULL; kw++) { 275 if (kw->kw_stdc && tflag) 276 continue; 277 if (kw->kw_gcc && !gflag) 278 continue; 279 sym = getblk(sizeof (sym_t)); 280 sym->s_name = kw->kw_name; 281 sym->s_keyw = 1; 282 sym->s_value.v_quad = kw->kw_token; 283 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 284 sym->s_tspec = kw->kw_tspec; 285 } else if (kw->kw_token == T_SCLASS) { 286 sym->s_scl = kw->kw_scl; 287 } else if (kw->kw_token == T_QUAL) { 288 sym->s_tqual = kw->kw_tqual; 289 } 290 h = hash(sym->s_name); 291 if ((sym->s_link = symtab[h]) != NULL) 292 symtab[h]->s_rlink = &sym->s_link; 293 (symtab[h] = sym)->s_rlink = &symtab[h]; 294 } 295 296 /* initialize bit-masks for quads */ 297 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 298 qbmasks[i] = (u_quad_t)1 << i; 299 uq = ~(u_quad_t)0 << i; 300 qumasks[i] = uq; 301 qlmasks[i] = ~uq; 302 } 303 qumasks[i] = 0; 304 qlmasks[i] = ~(u_quad_t)0; 305 } 306 307 /* 308 * Get a free sbuf structure, if possible from the free list 309 */ 310 static sbuf_t * 311 allocsb(void) 312 { 313 sbuf_t *sb; 314 315 if ((sb = sbfrlst) != NULL) { 316 sbfrlst = sb->sb_nxt; 317 } else { 318 sb = xmalloc(sizeof (sbuf_t)); 319 } 320 (void)memset(sb, 0, sizeof (sb)); 321 return (sb); 322 } 323 324 /* 325 * Put a sbuf structure to the free list 326 */ 327 static void 328 freesb(sbuf_t *sb) 329 { 330 331 sb->sb_nxt = sbfrlst; 332 sbfrlst = sb; 333 } 334 335 /* 336 * Read a character and ensure that it is positive (except EOF). 337 * Increment line count(s) if necessary. 338 */ 339 static int 340 inpc(void) 341 { 342 int c; 343 344 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 345 incline(); 346 return (c); 347 } 348 349 static int 350 hash(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(void) 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(sbuf_t *sb) 418 { 419 sym_t *sym; 420 421 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 422 if (strcmp(sym->s_name, sb->sb_name) == 0) { 423 if (sym->s_keyw || sym->s_kind == symtyp) 424 return (sym); 425 } 426 } 427 428 return (NULL); 429 } 430 431 static int 432 keyw(sym_t *sym) 433 { 434 int t; 435 436 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 437 yylval.y_scl = sym->s_scl; 438 } else if (t == T_TYPE || t == T_SOU) { 439 yylval.y_tspec = sym->s_tspec; 440 } else if (t == T_QUAL) { 441 yylval.y_tqual = sym->s_tqual; 442 } 443 return (t); 444 } 445 446 /* 447 * Convert a string representing an integer into internal representation. 448 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 449 */ 450 static int 451 icon(int base) 452 { 453 int l_suffix, u_suffix; 454 int len; 455 const char *cp; 456 char c, *eptr; 457 tspec_t typ; 458 u_long ul = 0; 459 u_quad_t uq = 0; 460 int ansiu; 461 static tspec_t contypes[2][3] = { 462 { INT, LONG, QUAD }, 463 { UINT, ULONG, UQUAD } 464 }; 465 466 cp = yytext; 467 len = yyleng; 468 469 /* skip 0x */ 470 if (base == 16) { 471 cp += 2; 472 len -= 2; 473 } 474 475 /* read suffixes */ 476 l_suffix = u_suffix = 0; 477 for ( ; ; ) { 478 if ((c = cp[len - 1]) == 'l' || c == 'L') { 479 l_suffix++; 480 } else if (c == 'u' || c == 'U') { 481 u_suffix++; 482 } else { 483 break; 484 } 485 len--; 486 } 487 if (l_suffix > 2 || u_suffix > 1) { 488 /* malformed integer constant */ 489 warning(251); 490 if (l_suffix > 2) 491 l_suffix = 2; 492 if (u_suffix > 1) 493 u_suffix = 1; 494 } 495 if (tflag && u_suffix != 0) { 496 /* suffix U is illegal in traditional C */ 497 warning(97); 498 } 499 typ = contypes[u_suffix][l_suffix]; 500 501 errno = 0; 502 if (l_suffix < 2) { 503 ul = strtoul(cp, &eptr, base); 504 } else { 505 uq = strtouq(cp, &eptr, base); 506 } 507 if (eptr != cp + len) 508 lerror("icon() 1"); 509 if (errno != 0) 510 /* integer constant out of range */ 511 warning(252); 512 513 /* 514 * If the value is to big for the current type, we must choose 515 * another type. 516 */ 517 ansiu = 0; 518 switch (typ) { 519 case INT: 520 if (ul <= INT_MAX) { 521 /* ok */ 522 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 523 typ = UINT; 524 } else if (ul <= LONG_MAX) { 525 typ = LONG; 526 } else { 527 typ = ULONG; 528 } 529 if (typ == UINT || typ == ULONG) { 530 if (tflag) { 531 typ = LONG; 532 } else if (!sflag) { 533 /* 534 * Remember that the constant is unsigned 535 * only in ANSI C 536 */ 537 ansiu = 1; 538 } 539 } 540 break; 541 case UINT: 542 if (ul > (u_int)UINT_MAX) 543 typ = ULONG; 544 break; 545 case LONG: 546 if (ul > LONG_MAX && !tflag) { 547 typ = ULONG; 548 if (!sflag) 549 ansiu = 1; 550 } 551 break; 552 case QUAD: 553 if (uq > QUAD_MAX && !tflag) { 554 typ = UQUAD; 555 if (!sflag) 556 ansiu = 1; 557 } 558 break; 559 /* LINTED (enumeration values not handled in switch) */ 560 case STRUCT: 561 case VOID: 562 case LDOUBLE: 563 case FUNC: 564 case ARRAY: 565 case PTR: 566 case ENUM: 567 case UNION: 568 case SIGNED: 569 case NOTSPEC: 570 case DOUBLE: 571 case FLOAT: 572 case UQUAD: 573 case ULONG: 574 case USHORT: 575 case SHORT: 576 case UCHAR: 577 case SCHAR: 578 case CHAR: 579 case UNSIGN: 580 break; 581 } 582 583 if (typ != QUAD && typ != UQUAD) { 584 if (isutyp(typ)) { 585 uq = ul; 586 } else { 587 uq = (quad_t)(long)ul; 588 } 589 } 590 591 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 592 593 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 594 yylval.y_val->v_ansiu = ansiu; 595 yylval.y_val->v_quad = (quad_t)uq; 596 597 return (T_CON); 598 } 599 600 /* 601 * Returns 1 if t is a signed type and the value is negative. 602 * 603 * len is the number of significant bits. If len is -1, len is set 604 * to the width of type t. 605 */ 606 int 607 sign(quad_t q, tspec_t t, int len) 608 { 609 610 if (t == PTR || isutyp(t)) 611 return (0); 612 return (msb(q, t, len)); 613 } 614 615 int 616 msb(quad_t q, tspec_t t, int len) 617 { 618 619 if (len <= 0) 620 len = size(t); 621 return ((q & qbmasks[len - 1]) != 0); 622 } 623 624 /* 625 * Extends the sign of q. 626 */ 627 quad_t 628 xsign(quad_t q, tspec_t t, int len) 629 { 630 631 if (len <= 0) 632 len = size(t); 633 634 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 635 q &= qlmasks[len]; 636 } else { 637 q |= qumasks[len]; 638 } 639 return (q); 640 } 641 642 /* 643 * Convert a string representing a floating point value into its interal 644 * representation. Type and value are returned in yylval. fcon() 645 * (and yylex()) returns T_CON. 646 * XXX Currently it is not possible to convert constants of type 647 * long double which are greater then DBL_MAX. 648 */ 649 static int 650 fcon(void) 651 { 652 const char *cp; 653 int len; 654 tspec_t typ; 655 char c, *eptr; 656 double d; 657 float f = 0; 658 659 cp = yytext; 660 len = yyleng; 661 662 if ((c = cp[len - 1]) == 'f' || c == 'F') { 663 typ = FLOAT; 664 len--; 665 } else if (c == 'l' || c == 'L') { 666 typ = LDOUBLE; 667 len--; 668 } else { 669 typ = DOUBLE; 670 } 671 672 if (tflag && typ != DOUBLE) { 673 /* suffixes F and L are illegal in traditional C */ 674 warning(98); 675 } 676 677 errno = 0; 678 d = strtod(cp, &eptr); 679 if (eptr != cp + len) 680 lerror("fcon() 1"); 681 if (errno != 0) 682 /* floating-point constant out of range */ 683 warning(248); 684 685 if (typ == FLOAT) { 686 f = (float)d; 687 if (isinf(f)) { 688 /* floating-point constant out of range */ 689 warning(248); 690 f = f > 0 ? FLT_MAX : -FLT_MAX; 691 } 692 } 693 694 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 695 if (typ == FLOAT) { 696 yylval.y_val->v_ldbl = f; 697 } else { 698 yylval.y_val->v_ldbl = d; 699 } 700 701 return (T_CON); 702 } 703 704 static int 705 operator(int t, op_t o) 706 { 707 708 yylval.y_op = o; 709 return (t); 710 } 711 712 /* 713 * Called if lex found a leading \'. 714 */ 715 static int 716 ccon(void) 717 { 718 int n, val, c; 719 char cv; 720 721 n = 0; 722 val = 0; 723 while ((c = getescc('\'')) >= 0) { 724 val = (val << CHAR_BIT) + c; 725 n++; 726 } 727 if (c == -2) { 728 /* unterminated character constant */ 729 error(253); 730 } else { 731 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 732 /* too many characters in character constant */ 733 error(71); 734 } else if (n > 1) { 735 /* multi-character character constant */ 736 warning(294); 737 } else if (n == 0) { 738 /* empty character constant */ 739 error(73); 740 } 741 } 742 if (n == 1) { 743 cv = (char)val; 744 val = cv; 745 } 746 747 yylval.y_val = xcalloc(1, sizeof (val_t)); 748 yylval.y_val->v_tspec = INT; 749 yylval.y_val->v_quad = val; 750 751 return (T_CON); 752 } 753 754 /* 755 * Called if lex found a leading L\' 756 */ 757 static int 758 wccon(void) 759 { 760 static char buf[MB_LEN_MAX + 1]; 761 int i, c; 762 wchar_t wc; 763 764 i = 0; 765 while ((c = getescc('\'')) >= 0) { 766 if (i < MB_CUR_MAX) 767 buf[i] = (char)c; 768 i++; 769 } 770 771 wc = 0; 772 773 if (c == -2) { 774 /* unterminated character constant */ 775 error(253); 776 } else if (c == 0) { 777 /* empty character constant */ 778 error(73); 779 } else { 780 if (i > MB_CUR_MAX) { 781 i = MB_CUR_MAX; 782 /* too many characters in character constant */ 783 error(71); 784 } else { 785 buf[i] = '\0'; 786 (void)mbtowc(NULL, NULL, 0); 787 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 788 /* invalid multibyte character */ 789 error(291); 790 } 791 } 792 793 yylval.y_val = xcalloc(1, sizeof (val_t)); 794 yylval.y_val->v_tspec = WCHAR; 795 yylval.y_val->v_quad = wc; 796 797 return (T_CON); 798 } 799 800 /* 801 * Read a character which is part of a character constant or of a string 802 * and handle escapes. 803 * 804 * The Argument is the character which delimits the character constant or 805 * string. 806 * 807 * Returns -1 if the end of the character constant or string is reached, 808 * -2 if the EOF is reached, and the character otherwise. 809 */ 810 static int 811 getescc(int d) 812 { 813 static int pbc = -1; 814 int n, c, v; 815 816 if (pbc == -1) { 817 c = inpc(); 818 } else { 819 c = pbc; 820 pbc = -1; 821 } 822 if (c == d) 823 return (-1); 824 switch (c) { 825 case '\n': 826 if (tflag) { 827 /* newline in string or char constant */ 828 error(254); 829 return (-2); 830 } 831 return (c); 832 case EOF: 833 return (-2); 834 case '\\': 835 switch (c = inpc()) { 836 case '"': 837 if (tflag && d == '\'') 838 /* \" inside character constant undef. ... */ 839 warning(262); 840 return ('"'); 841 case '\'': 842 return ('\''); 843 case '?': 844 if (tflag) 845 /* \? undefined in traditional C */ 846 warning(263); 847 return ('?'); 848 case '\\': 849 return ('\\'); 850 case 'a': 851 if (tflag) 852 /* \a undefined in traditional C */ 853 warning(81); 854 return ('\a'); 855 case 'b': 856 return ('\b'); 857 case 'f': 858 return ('\f'); 859 case 'n': 860 return ('\n'); 861 case 'r': 862 return ('\r'); 863 case 't': 864 return ('\t'); 865 case 'v': 866 if (tflag) 867 /* \v undefined in traditional C */ 868 warning(264); 869 return ('\v'); 870 case '8': case '9': 871 /* bad octal digit %c */ 872 warning(77, c); 873 /* FALLTHROUGH */ 874 case '0': case '1': case '2': case '3': 875 case '4': case '5': case '6': case '7': 876 n = 3; 877 v = 0; 878 do { 879 v = (v << 3) + (c - '0'); 880 c = inpc(); 881 } while (--n && isdigit(c) && (tflag || c <= '7')); 882 if (tflag && n > 0 && isdigit(c)) 883 /* bad octal digit %c */ 884 warning(77, c); 885 pbc = c; 886 if (v > UCHAR_MAX) { 887 /* character escape does not fit in char. */ 888 warning(76); 889 v &= CHAR_MASK; 890 } 891 return (v); 892 case 'x': 893 if (tflag) 894 /* \x undefined in traditional C */ 895 warning(82); 896 v = 0; 897 n = 0; 898 while ((c = inpc()) >= 0 && isxdigit(c)) { 899 c = isdigit(c) ? 900 c - '0' : toupper(c) - 'A' + 10; 901 v = (v << 4) + c; 902 if (n >= 0) { 903 if ((v & ~CHAR_MASK) != 0) { 904 /* overflow in hex escape */ 905 warning(75); 906 n = -1; 907 } else { 908 n++; 909 } 910 } 911 } 912 pbc = c; 913 if (n == 0) { 914 /* no hex digits follow \x */ 915 error(74); 916 } if (n == -1) { 917 v &= CHAR_MASK; 918 } 919 return (v); 920 case '\n': 921 return (getescc(d)); 922 case EOF: 923 return (-2); 924 default: 925 if (isprint(c)) { 926 /* dubious escape \%c */ 927 warning(79, c); 928 } else { 929 /* dubious escape \%o */ 930 warning(80, c); 931 } 932 } 933 } 934 return (c); 935 } 936 937 /* 938 * Called for preprocessor directives. Currently implemented are: 939 * # lineno 940 * # lineno "filename" 941 */ 942 static void 943 directive(void) 944 { 945 const char *cp, *fn; 946 char c, *eptr; 947 size_t fnl; 948 long ln; 949 static int first = 1; 950 951 /* Go to first non-whitespace after # */ 952 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 953 continue; 954 955 if (!isdigit((unsigned char)c)) { 956 error: 957 /* undefined or invalid # directive */ 958 warning(255); 959 return; 960 } 961 ln = strtol(--cp, &eptr, 10); 962 if (cp == eptr) 963 goto error; 964 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 965 goto error; 966 while ((c = *cp++) == ' ' || c == '\t') 967 continue; 968 if (c != '\0') { 969 if (c != '"') 970 goto error; 971 fn = cp; 972 while ((c = *cp) != '"' && c != '\0') 973 cp++; 974 if (c != '"') 975 goto error; 976 if ((fnl = cp++ - fn) > PATH_MAX) 977 goto error; 978 while ((c = *cp++) == ' ' || c == '\t') 979 continue; 980 #if 0 981 if (c != '\0') 982 warning("extra character(s) after directive"); 983 #endif 984 985 /* empty string means stdin */ 986 if (fnl == 0) { 987 fn = "{standard input}"; 988 fnl = 16; /* strlen (fn) */ 989 } 990 curr_pos.p_file = fnnalloc(fn, fnl); 991 /* 992 * If this is the first directive, the name is the name 993 * of the C source file as specified at the command line. 994 * It is written to the output file. 995 */ 996 if (first) { 997 csrc_pos.p_file = curr_pos.p_file; 998 outsrc(curr_pos.p_file); 999 first = 0; 1000 } 1001 } 1002 curr_pos.p_line = (int)ln - 1; 1003 curr_pos.p_uniq = 0; 1004 if (curr_pos.p_file == csrc_pos.p_file) { 1005 csrc_pos.p_line = (int)ln - 1; 1006 csrc_pos.p_uniq = 0; 1007 } 1008 } 1009 1010 /* 1011 * Handle lint comments. Following comments are currently understood: 1012 * ARGSUSEDn 1013 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1014 * FALLTHRU FALLTHROUGH 1015 * LINTLIBRARY 1016 * LINTED NOSTRICT 1017 * LONGLONG 1018 * NOTREACHED 1019 * PRINTFLIKEn 1020 * PROTOLIB 1021 * SCANFLIKEn 1022 * VARARGSn 1023 * If one of this comments is recognized, the arguments, if any, are 1024 * parsed and a function which handles this comment is called. 1025 */ 1026 static void 1027 comment(void) 1028 { 1029 int c, lc; 1030 static struct { 1031 const char *keywd; 1032 int arg; 1033 void (*func)(int); 1034 } keywtab[] = { 1035 { "ARGSUSED", 1, argsused }, 1036 { "CONSTCOND", 0, constcond }, 1037 { "CONSTANTCOND", 0, constcond }, 1038 { "CONSTANTCONDITION", 0, constcond }, 1039 { "FALLTHRU", 0, fallthru }, 1040 { "FALLTHROUGH", 0, fallthru }, 1041 { "LINTLIBRARY", 0, lintlib }, 1042 { "LINTED", 0, linted }, 1043 { "LONGLONG", 0, longlong }, 1044 { "NOSTRICT", 0, linted }, 1045 { "NOTREACHED", 0, notreach }, 1046 { "PRINTFLIKE", 1, printflike }, 1047 { "PROTOLIB", 1, protolib }, 1048 { "SCANFLIKE", 1, scanflike }, 1049 { "VARARGS", 1, varargs }, 1050 }; 1051 char keywd[32]; 1052 char arg[32]; 1053 int l, i, a; 1054 int eoc; 1055 1056 eoc = 0; 1057 1058 /* Skip white spaces after the start of the comment */ 1059 while ((c = inpc()) != EOF && isspace(c)) 1060 continue; 1061 1062 /* Read the potential keyword to keywd */ 1063 l = 0; 1064 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1065 keywd[l++] = (char)c; 1066 c = inpc(); 1067 } 1068 keywd[l] = '\0'; 1069 1070 /* look for the keyword */ 1071 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1072 if (strcmp(keywtab[i].keywd, keywd) == 0) 1073 break; 1074 } 1075 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1076 goto skip_rest; 1077 1078 /* skip white spaces after the keyword */ 1079 while (c != EOF && isspace(c)) 1080 c = inpc(); 1081 1082 /* read the argument, if the keyword accepts one and there is one */ 1083 l = 0; 1084 if (keywtab[i].arg) { 1085 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1086 arg[l++] = (char)c; 1087 c = inpc(); 1088 } 1089 } 1090 arg[l] = '\0'; 1091 a = l != 0 ? atoi(arg) : -1; 1092 1093 /* skip white spaces after the argument */ 1094 while (c != EOF && isspace(c)) 1095 c = inpc(); 1096 1097 if (c != '*' || (c = inpc()) != '/') { 1098 if (keywtab[i].func != linted) 1099 /* extra characters in lint comment */ 1100 warning(257); 1101 } else { 1102 /* 1103 * remember that we have already found the end of the 1104 * comment 1105 */ 1106 eoc = 1; 1107 } 1108 1109 if (keywtab[i].func != NULL) 1110 (*keywtab[i].func)(a); 1111 1112 skip_rest: 1113 while (!eoc) { 1114 lc = c; 1115 if ((c = inpc()) == EOF) { 1116 /* unterminated comment */ 1117 error(256); 1118 break; 1119 } 1120 if (lc == '*' && c == '/') 1121 eoc = 1; 1122 } 1123 } 1124 1125 /* 1126 * Handle // style comments 1127 */ 1128 static void 1129 slashslashcomment(void) 1130 { 1131 int c; 1132 1133 if (sflag < 2 && !gflag) 1134 /* // comments only support in C99 */ 1135 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1136 1137 while ((c = inpc()) != EOF && c != '\n') 1138 continue; 1139 } 1140 1141 /* 1142 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1143 * clrwflgs() is called after function definitions and global and 1144 * local declarations and definitions. It is also called between 1145 * the controlling expression and the body of control statements 1146 * (if, switch, for, while). 1147 */ 1148 void 1149 clrwflgs(void) 1150 { 1151 1152 nowarn = 0; 1153 quadflg = 0; 1154 ccflg = 0; 1155 } 1156 1157 /* 1158 * Strings are stored in a dynamically alloceted buffer and passed 1159 * in yylval.y_xstrg to the parser. The parser or the routines called 1160 * by the parser are responsible for freeing this buffer. 1161 */ 1162 static int 1163 string(void) 1164 { 1165 u_char *s; 1166 int c; 1167 size_t len, max; 1168 strg_t *strg; 1169 1170 s = xmalloc(max = 64); 1171 1172 len = 0; 1173 while ((c = getescc('"')) >= 0) { 1174 /* +1 to reserve space for a trailing NUL character */ 1175 if (len + 1 == max) 1176 s = xrealloc(s, max *= 2); 1177 s[len++] = (char)c; 1178 } 1179 s[len] = '\0'; 1180 if (c == -2) 1181 /* unterminated string constant */ 1182 error(258); 1183 1184 strg = xcalloc(1, sizeof (strg_t)); 1185 strg->st_tspec = CHAR; 1186 strg->st_len = len; 1187 strg->st_cp = s; 1188 1189 yylval.y_strg = strg; 1190 return (T_STRING); 1191 } 1192 1193 static int 1194 wcstrg(void) 1195 { 1196 char *s; 1197 int c, i, n, wi; 1198 size_t len, max, wlen; 1199 wchar_t *ws; 1200 strg_t *strg; 1201 1202 s = xmalloc(max = 64); 1203 len = 0; 1204 while ((c = getescc('"')) >= 0) { 1205 /* +1 to save space for a trailing NUL character */ 1206 if (len + 1 >= max) 1207 s = xrealloc(s, max *= 2); 1208 s[len++] = (char)c; 1209 } 1210 s[len] = '\0'; 1211 if (c == -2) 1212 /* unterminated string constant */ 1213 error(258); 1214 1215 /* get length of wide character string */ 1216 (void)mblen(NULL, 0); 1217 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1218 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1219 /* invalid multibyte character */ 1220 error(291); 1221 break; 1222 } 1223 if (n == 0) 1224 n = 1; 1225 } 1226 1227 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1228 1229 /* convert from multibyte to wide char */ 1230 (void)mbtowc(NULL, NULL, 0); 1231 for (i = 0, wi = 0; i < len; i += n, wi++) { 1232 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1233 break; 1234 if (n == 0) 1235 n = 1; 1236 } 1237 ws[wi] = 0; 1238 free(s); 1239 1240 strg = xcalloc(1, sizeof (strg_t)); 1241 strg->st_tspec = WCHAR; 1242 strg->st_len = wlen; 1243 strg->st_wcp = ws; 1244 1245 yylval.y_strg = strg; 1246 return (T_STRING); 1247 } 1248 1249 /* 1250 * As noted above the scanner does not create new symbol table entries 1251 * for symbols it cannot find in the symbol table. This is to avoid 1252 * putting undeclared symbols into the symbol table if a syntax error 1253 * occurs. 1254 * 1255 * getsym() is called as soon as it is probably ok to put the symbol to 1256 * the symbol table. This does not mean that it is not possible that 1257 * symbols are put to the symbol table which are than not completely 1258 * declared due to syntax errors. To avoid too many problems in this 1259 * case symbols get type int in getsym(). 1260 * 1261 * XXX calls to getsym() should be delayed until decl1*() is called 1262 */ 1263 sym_t * 1264 getsym(sbuf_t *sb) 1265 { 1266 dinfo_t *di; 1267 char *s; 1268 sym_t *sym; 1269 1270 sym = sb->sb_sym; 1271 1272 /* 1273 * During member declaration it is possible that name() looked 1274 * for symbols of type FVFT, although it should have looked for 1275 * symbols of type FTAG. Same can happen for labels. Both cases 1276 * are compensated here. 1277 */ 1278 if (symtyp == FMOS || symtyp == FLAB) { 1279 if (sym == NULL || sym->s_kind == FVFT) 1280 sym = search(sb); 1281 } 1282 1283 if (sym != NULL) { 1284 if (sym->s_kind != symtyp) 1285 lerror("storesym() 1"); 1286 symtyp = FVFT; 1287 freesb(sb); 1288 return (sym); 1289 } 1290 1291 /* create a new symbol table entry */ 1292 1293 /* labels must always be allocated at level 1 (outhermost block) */ 1294 if (symtyp == FLAB) { 1295 sym = getlblk(1, sizeof (sym_t)); 1296 s = getlblk(1, sb->sb_len + 1); 1297 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1298 sym->s_name = s; 1299 sym->s_blklev = 1; 1300 di = dcs; 1301 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1302 di = di->d_nxt; 1303 if (di->d_ctx != AUTO) 1304 lerror("storesym() 2"); 1305 } else { 1306 sym = getblk(sizeof (sym_t)); 1307 sym->s_name = sb->sb_name; 1308 sym->s_blklev = blklev; 1309 di = dcs; 1310 } 1311 1312 UNIQUE_CURR_POS(sym->s_dpos); 1313 if ((sym->s_kind = symtyp) != FLAB) 1314 sym->s_type = gettyp(INT); 1315 1316 symtyp = FVFT; 1317 1318 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1319 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1320 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1321 1322 *di->d_ldlsym = sym; 1323 di->d_ldlsym = &sym->s_dlnxt; 1324 1325 freesb(sb); 1326 return (sym); 1327 } 1328 1329 /* 1330 * Remove a symbol forever from the symbol table. s_blklev 1331 * is set to -1 to avoid that the symbol will later be put 1332 * back to the symbol table. 1333 */ 1334 void 1335 rmsym(sym_t *sym) 1336 { 1337 1338 if ((*sym->s_rlink = sym->s_link) != NULL) 1339 sym->s_link->s_rlink = sym->s_rlink; 1340 sym->s_blklev = -1; 1341 sym->s_link = NULL; 1342 } 1343 1344 /* 1345 * Remove a list of symbols declared at one level from the symbol 1346 * table. 1347 */ 1348 void 1349 rmsyms(sym_t *syms) 1350 { 1351 sym_t *sym; 1352 1353 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1354 if (sym->s_blklev != -1) { 1355 if ((*sym->s_rlink = sym->s_link) != NULL) 1356 sym->s_link->s_rlink = sym->s_rlink; 1357 sym->s_link = NULL; 1358 sym->s_rlink = NULL; 1359 } 1360 } 1361 } 1362 1363 /* 1364 * Put a symbol into the symbol table 1365 */ 1366 void 1367 inssym(int bl, sym_t *sym) 1368 { 1369 int h; 1370 1371 h = hash(sym->s_name); 1372 if ((sym->s_link = symtab[h]) != NULL) 1373 symtab[h]->s_rlink = &sym->s_link; 1374 (symtab[h] = sym)->s_rlink = &symtab[h]; 1375 sym->s_blklev = bl; 1376 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1377 lerror("inssym()"); 1378 } 1379 1380 /* 1381 * Called at level 0 after syntax errors 1382 * Removes all symbols which are not declared at level 0 from the 1383 * symbol table. Also frees all memory which is not associated with 1384 * level 0. 1385 */ 1386 void 1387 cleanup(void) 1388 { 1389 sym_t *sym, *nsym; 1390 int i; 1391 1392 for (i = 0; i < HSHSIZ1; i++) { 1393 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1394 nsym = sym->s_link; 1395 if (sym->s_blklev >= 1) { 1396 if ((*sym->s_rlink = nsym) != NULL) 1397 nsym->s_rlink = sym->s_rlink; 1398 } 1399 } 1400 } 1401 1402 for (i = mblklev; i > 0; i--) 1403 freelblk(i); 1404 } 1405 1406 /* 1407 * Create a new symbol with the name of an existing symbol. 1408 */ 1409 sym_t * 1410 pushdown(sym_t *sym) 1411 { 1412 int h; 1413 sym_t *nsym; 1414 1415 h = hash(sym->s_name); 1416 nsym = getblk(sizeof (sym_t)); 1417 if (sym->s_blklev > blklev) 1418 lerror("pushdown()"); 1419 nsym->s_name = sym->s_name; 1420 UNIQUE_CURR_POS(nsym->s_dpos); 1421 nsym->s_kind = sym->s_kind; 1422 nsym->s_blklev = blklev; 1423 1424 if ((nsym->s_link = symtab[h]) != NULL) 1425 symtab[h]->s_rlink = &nsym->s_link; 1426 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1427 1428 *dcs->d_ldlsym = nsym; 1429 dcs->d_ldlsym = &nsym->s_dlnxt; 1430 1431 return (nsym); 1432 } 1433 1434 /* 1435 * Free any dynamically allocated memory referenced by 1436 * the value stack or yylval. 1437 * The type of information in yylval is described by tok. 1438 */ 1439 void 1440 freeyyv(void *sp, int tok) 1441 { 1442 if (tok == T_NAME || tok == T_TYPENAME) { 1443 sbuf_t *sb = *(sbuf_t **)sp; 1444 freesb(sb); 1445 } else if (tok == T_CON) { 1446 val_t *val = *(val_t **)sp; 1447 free(val); 1448 } else if (tok == T_STRING) { 1449 strg_t *strg = *(strg_t **)sp; 1450 if (strg->st_tspec == CHAR) { 1451 free(strg->st_cp); 1452 } else if (strg->st_tspec == WCHAR) { 1453 free(strg->st_wcp); 1454 } else { 1455 lerror("fryylv() 1"); 1456 } 1457 free(strg); 1458 } 1459 } 1460
Indexes created Wed Oct 01 12:09:54 GMT 2025