scan.l revision 1.23
1 %{ 2 /* $NetBSD: scan.l,v 1.23 2002/01/03 05:37:39 thorpej 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 1.23 2002/01/03 05:37:39 thorpej 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 case NTSPEC: /* this value unused */ 583 break; 584 } 585 586 if (typ != QUAD && typ != UQUAD) { 587 if (isutyp(typ)) { 588 uq = ul; 589 } else { 590 uq = (quad_t)(long)ul; 591 } 592 } 593 594 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 595 596 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 597 yylval.y_val->v_ansiu = ansiu; 598 yylval.y_val->v_quad = (quad_t)uq; 599 600 return (T_CON); 601 } 602 603 /* 604 * Returns 1 if t is a signed type and the value is negative. 605 * 606 * len is the number of significant bits. If len is -1, len is set 607 * to the width of type t. 608 */ 609 int 610 sign(quad_t q, tspec_t t, int len) 611 { 612 613 if (t == PTR || isutyp(t)) 614 return (0); 615 return (msb(q, t, len)); 616 } 617 618 int 619 msb(quad_t q, tspec_t t, int len) 620 { 621 622 if (len <= 0) 623 len = size(t); 624 return ((q & qbmasks[len - 1]) != 0); 625 } 626 627 /* 628 * Extends the sign of q. 629 */ 630 quad_t 631 xsign(quad_t q, tspec_t t, int len) 632 { 633 634 if (len <= 0) 635 len = size(t); 636 637 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 638 q &= qlmasks[len]; 639 } else { 640 q |= qumasks[len]; 641 } 642 return (q); 643 } 644 645 /* 646 * Convert a string representing a floating point value into its interal 647 * representation. Type and value are returned in yylval. fcon() 648 * (and yylex()) returns T_CON. 649 * XXX Currently it is not possible to convert constants of type 650 * long double which are greater than DBL_MAX. 651 */ 652 static int 653 fcon(void) 654 { 655 const char *cp; 656 int len; 657 tspec_t typ; 658 char c, *eptr; 659 double d; 660 float f = 0; 661 662 cp = yytext; 663 len = yyleng; 664 665 if ((c = cp[len - 1]) == 'f' || c == 'F') { 666 typ = FLOAT; 667 len--; 668 } else if (c == 'l' || c == 'L') { 669 typ = LDOUBLE; 670 len--; 671 } else { 672 typ = DOUBLE; 673 } 674 675 if (tflag && typ != DOUBLE) { 676 /* suffixes F and L are illegal in traditional C */ 677 warning(98); 678 } 679 680 errno = 0; 681 d = strtod(cp, &eptr); 682 if (eptr != cp + len) 683 lerror("fcon() 1"); 684 if (errno != 0) 685 /* floating-point constant out of range */ 686 warning(248); 687 688 if (typ == FLOAT) { 689 f = (float)d; 690 if (isinf(f)) { 691 /* floating-point constant out of range */ 692 warning(248); 693 f = f > 0 ? FLT_MAX : -FLT_MAX; 694 } 695 } 696 697 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 698 if (typ == FLOAT) { 699 yylval.y_val->v_ldbl = f; 700 } else { 701 yylval.y_val->v_ldbl = d; 702 } 703 704 return (T_CON); 705 } 706 707 static int 708 operator(int t, op_t o) 709 { 710 711 yylval.y_op = o; 712 return (t); 713 } 714 715 /* 716 * Called if lex found a leading \'. 717 */ 718 static int 719 ccon(void) 720 { 721 int n, val, c; 722 char cv; 723 724 n = 0; 725 val = 0; 726 while ((c = getescc('\'')) >= 0) { 727 val = (val << CHAR_BIT) + c; 728 n++; 729 } 730 if (c == -2) { 731 /* unterminated character constant */ 732 error(253); 733 } else { 734 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 735 /* too many characters in character constant */ 736 error(71); 737 } else if (n > 1) { 738 /* multi-character character constant */ 739 warning(294); 740 } else if (n == 0) { 741 /* empty character constant */ 742 error(73); 743 } 744 } 745 if (n == 1) { 746 cv = (char)val; 747 val = cv; 748 } 749 750 yylval.y_val = xcalloc(1, sizeof (val_t)); 751 yylval.y_val->v_tspec = INT; 752 yylval.y_val->v_quad = val; 753 754 return (T_CON); 755 } 756 757 /* 758 * Called if lex found a leading L\' 759 */ 760 static int 761 wccon(void) 762 { 763 static char buf[MB_LEN_MAX + 1]; 764 int i, c; 765 wchar_t wc; 766 767 i = 0; 768 while ((c = getescc('\'')) >= 0) { 769 if (i < MB_CUR_MAX) 770 buf[i] = (char)c; 771 i++; 772 } 773 774 wc = 0; 775 776 if (c == -2) { 777 /* unterminated character constant */ 778 error(253); 779 } else if (c == 0) { 780 /* empty character constant */ 781 error(73); 782 } else { 783 if (i > MB_CUR_MAX) { 784 i = MB_CUR_MAX; 785 /* too many characters in character constant */ 786 error(71); 787 } else { 788 buf[i] = '\0'; 789 (void)mbtowc(NULL, NULL, 0); 790 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 791 /* invalid multibyte character */ 792 error(291); 793 } 794 } 795 796 yylval.y_val = xcalloc(1, sizeof (val_t)); 797 yylval.y_val->v_tspec = WCHAR; 798 yylval.y_val->v_quad = wc; 799 800 return (T_CON); 801 } 802 803 /* 804 * Read a character which is part of a character constant or of a string 805 * and handle escapes. 806 * 807 * The Argument is the character which delimits the character constant or 808 * string. 809 * 810 * Returns -1 if the end of the character constant or string is reached, 811 * -2 if the EOF is reached, and the character otherwise. 812 */ 813 static int 814 getescc(int d) 815 { 816 static int pbc = -1; 817 int n, c, v; 818 819 if (pbc == -1) { 820 c = inpc(); 821 } else { 822 c = pbc; 823 pbc = -1; 824 } 825 if (c == d) 826 return (-1); 827 switch (c) { 828 case '\n': 829 if (tflag) { 830 /* newline in string or char constant */ 831 error(254); 832 return (-2); 833 } 834 return (c); 835 case EOF: 836 return (-2); 837 case '\\': 838 switch (c = inpc()) { 839 case '"': 840 if (tflag && d == '\'') 841 /* \" inside character constant undef. ... */ 842 warning(262); 843 return ('"'); 844 case '\'': 845 return ('\''); 846 case '?': 847 if (tflag) 848 /* \? undefined in traditional C */ 849 warning(263); 850 return ('?'); 851 case '\\': 852 return ('\\'); 853 case 'a': 854 if (tflag) 855 /* \a undefined in traditional C */ 856 warning(81); 857 return ('\a'); 858 case 'b': 859 return ('\b'); 860 case 'f': 861 return ('\f'); 862 case 'n': 863 return ('\n'); 864 case 'r': 865 return ('\r'); 866 case 't': 867 return ('\t'); 868 case 'v': 869 if (tflag) 870 /* \v undefined in traditional C */ 871 warning(264); 872 return ('\v'); 873 case '8': case '9': 874 /* bad octal digit %c */ 875 warning(77, c); 876 /* FALLTHROUGH */ 877 case '0': case '1': case '2': case '3': 878 case '4': case '5': case '6': case '7': 879 n = 3; 880 v = 0; 881 do { 882 v = (v << 3) + (c - '0'); 883 c = inpc(); 884 } while (--n && isdigit(c) && (tflag || c <= '7')); 885 if (tflag && n > 0 && isdigit(c)) 886 /* bad octal digit %c */ 887 warning(77, c); 888 pbc = c; 889 if (v > UCHAR_MAX) { 890 /* character escape does not fit in char. */ 891 warning(76); 892 v &= CHAR_MASK; 893 } 894 return (v); 895 case 'x': 896 if (tflag) 897 /* \x undefined in traditional C */ 898 warning(82); 899 v = 0; 900 n = 0; 901 while ((c = inpc()) >= 0 && isxdigit(c)) { 902 c = isdigit(c) ? 903 c - '0' : toupper(c) - 'A' + 10; 904 v = (v << 4) + c; 905 if (n >= 0) { 906 if ((v & ~CHAR_MASK) != 0) { 907 /* overflow in hex escape */ 908 warning(75); 909 n = -1; 910 } else { 911 n++; 912 } 913 } 914 } 915 pbc = c; 916 if (n == 0) { 917 /* no hex digits follow \x */ 918 error(74); 919 } if (n == -1) { 920 v &= CHAR_MASK; 921 } 922 return (v); 923 case '\n': 924 return (getescc(d)); 925 case EOF: 926 return (-2); 927 default: 928 if (isprint(c)) { 929 /* dubious escape \%c */ 930 warning(79, c); 931 } else { 932 /* dubious escape \%o */ 933 warning(80, c); 934 } 935 } 936 } 937 return (c); 938 } 939 940 /* 941 * Called for preprocessor directives. Currently implemented are: 942 * # lineno 943 * # lineno "filename" 944 */ 945 static void 946 directive(void) 947 { 948 const char *cp, *fn; 949 char c, *eptr; 950 size_t fnl; 951 long ln; 952 static int first = 1; 953 954 /* Go to first non-whitespace after # */ 955 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 956 continue; 957 958 if (!isdigit((unsigned char)c)) { 959 error: 960 /* undefined or invalid # directive */ 961 warning(255); 962 return; 963 } 964 ln = strtol(--cp, &eptr, 10); 965 if (cp == eptr) 966 goto error; 967 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 968 goto error; 969 while ((c = *cp++) == ' ' || c == '\t') 970 continue; 971 if (c != '\0') { 972 if (c != '"') 973 goto error; 974 fn = cp; 975 while ((c = *cp) != '"' && c != '\0') 976 cp++; 977 if (c != '"') 978 goto error; 979 if ((fnl = cp++ - fn) > PATH_MAX) 980 goto error; 981 while ((c = *cp++) == ' ' || c == '\t') 982 continue; 983 #if 0 984 if (c != '\0') 985 warning("extra character(s) after directive"); 986 #endif 987 988 /* empty string means stdin */ 989 if (fnl == 0) { 990 fn = "{standard input}"; 991 fnl = 16; /* strlen (fn) */ 992 } 993 curr_pos.p_file = fnnalloc(fn, fnl); 994 /* 995 * If this is the first directive, the name is the name 996 * of the C source file as specified at the command line. 997 * It is written to the output file. 998 */ 999 if (first) { 1000 csrc_pos.p_file = curr_pos.p_file; 1001 outsrc(curr_pos.p_file); 1002 first = 0; 1003 } 1004 } 1005 curr_pos.p_line = (int)ln - 1; 1006 curr_pos.p_uniq = 0; 1007 if (curr_pos.p_file == csrc_pos.p_file) { 1008 csrc_pos.p_line = (int)ln - 1; 1009 csrc_pos.p_uniq = 0; 1010 } 1011 } 1012 1013 /* 1014 * Handle lint comments. Following comments are currently understood: 1015 * ARGSUSEDn 1016 * BITFIELDTYPE 1017 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1018 * FALLTHRU FALLTHROUGH 1019 * LINTLIBRARY 1020 * LINTED NOSTRICT 1021 * LONGLONG 1022 * NOTREACHED 1023 * PRINTFLIKEn 1024 * PROTOLIB 1025 * SCANFLIKEn 1026 * VARARGSn 1027 * If one of this comments is recognized, the arguments, if any, are 1028 * parsed and a function which handles this comment is called. 1029 */ 1030 static void 1031 comment(void) 1032 { 1033 int c, lc; 1034 static struct { 1035 const char *keywd; 1036 int arg; 1037 void (*func)(int); 1038 } keywtab[] = { 1039 { "ARGSUSED", 1, argsused }, 1040 { "BITFIELDTYPE", 0, bitfieldtype }, 1041 { "CONSTCOND", 0, constcond }, 1042 { "CONSTANTCOND", 0, constcond }, 1043 { "CONSTANTCONDITION", 0, constcond }, 1044 { "FALLTHRU", 0, fallthru }, 1045 { "FALLTHROUGH", 0, fallthru }, 1046 { "LINTLIBRARY", 0, lintlib }, 1047 { "LINTED", 0, linted }, 1048 { "LONGLONG", 0, longlong }, 1049 { "NOSTRICT", 0, linted }, 1050 { "NOTREACHED", 0, notreach }, 1051 { "PRINTFLIKE", 1, printflike }, 1052 { "PROTOLIB", 1, protolib }, 1053 { "SCANFLIKE", 1, scanflike }, 1054 { "VARARGS", 1, varargs }, 1055 }; 1056 char keywd[32]; 1057 char arg[32]; 1058 int l, i, a; 1059 int eoc; 1060 1061 eoc = 0; 1062 1063 /* Skip white spaces after the start of the comment */ 1064 while ((c = inpc()) != EOF && isspace(c)) 1065 continue; 1066 1067 /* Read the potential keyword to keywd */ 1068 l = 0; 1069 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1070 keywd[l++] = (char)c; 1071 c = inpc(); 1072 } 1073 keywd[l] = '\0'; 1074 1075 /* look for the keyword */ 1076 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1077 if (strcmp(keywtab[i].keywd, keywd) == 0) 1078 break; 1079 } 1080 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1081 goto skip_rest; 1082 1083 /* skip white spaces after the keyword */ 1084 while (c != EOF && isspace(c)) 1085 c = inpc(); 1086 1087 /* read the argument, if the keyword accepts one and there is one */ 1088 l = 0; 1089 if (keywtab[i].arg) { 1090 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1091 arg[l++] = (char)c; 1092 c = inpc(); 1093 } 1094 } 1095 arg[l] = '\0'; 1096 a = l != 0 ? atoi(arg) : -1; 1097 1098 /* skip white spaces after the argument */ 1099 while (c != EOF && isspace(c)) 1100 c = inpc(); 1101 1102 if (c != '*' || (c = inpc()) != '/') { 1103 if (keywtab[i].func != linted) 1104 /* extra characters in lint comment */ 1105 warning(257); 1106 } else { 1107 /* 1108 * remember that we have already found the end of the 1109 * comment 1110 */ 1111 eoc = 1; 1112 } 1113 1114 if (keywtab[i].func != NULL) 1115 (*keywtab[i].func)(a); 1116 1117 skip_rest: 1118 while (!eoc) { 1119 lc = c; 1120 if ((c = inpc()) == EOF) { 1121 /* unterminated comment */ 1122 error(256); 1123 break; 1124 } 1125 if (lc == '*' && c == '/') 1126 eoc = 1; 1127 } 1128 } 1129 1130 /* 1131 * Handle // style comments 1132 */ 1133 static void 1134 slashslashcomment(void) 1135 { 1136 int c; 1137 1138 if (sflag < 2 && !gflag) 1139 /* // comments only supported in C99 */ 1140 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1141 1142 while ((c = inpc()) != EOF && c != '\n') 1143 continue; 1144 } 1145 1146 /* 1147 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1148 * clrwflgs() is called after function definitions and global and 1149 * local declarations and definitions. It is also called between 1150 * the controlling expression and the body of control statements 1151 * (if, switch, for, while). 1152 */ 1153 void 1154 clrwflgs(void) 1155 { 1156 1157 nowarn = 0; 1158 quadflg = 0; 1159 ccflg = 0; 1160 } 1161 1162 /* 1163 * Strings are stored in a dynamically alloceted buffer and passed 1164 * in yylval.y_xstrg to the parser. The parser or the routines called 1165 * by the parser are responsible for freeing this buffer. 1166 */ 1167 static int 1168 string(void) 1169 { 1170 u_char *s; 1171 int c; 1172 size_t len, max; 1173 strg_t *strg; 1174 1175 s = xmalloc(max = 64); 1176 1177 len = 0; 1178 while ((c = getescc('"')) >= 0) { 1179 /* +1 to reserve 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 strg = xcalloc(1, sizeof (strg_t)); 1190 strg->st_tspec = CHAR; 1191 strg->st_len = len; 1192 strg->st_cp = s; 1193 1194 yylval.y_strg = strg; 1195 return (T_STRING); 1196 } 1197 1198 static int 1199 wcstrg(void) 1200 { 1201 char *s; 1202 int c, i, n, wi; 1203 size_t len, max, wlen; 1204 wchar_t *ws; 1205 strg_t *strg; 1206 1207 s = xmalloc(max = 64); 1208 len = 0; 1209 while ((c = getescc('"')) >= 0) { 1210 /* +1 to save space for a trailing NUL character */ 1211 if (len + 1 >= max) 1212 s = xrealloc(s, max *= 2); 1213 s[len++] = (char)c; 1214 } 1215 s[len] = '\0'; 1216 if (c == -2) 1217 /* unterminated string constant */ 1218 error(258); 1219 1220 /* get length of wide character string */ 1221 (void)mblen(NULL, 0); 1222 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1223 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1224 /* invalid multibyte character */ 1225 error(291); 1226 break; 1227 } 1228 if (n == 0) 1229 n = 1; 1230 } 1231 1232 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1233 1234 /* convert from multibyte to wide char */ 1235 (void)mbtowc(NULL, NULL, 0); 1236 for (i = 0, wi = 0; i < len; i += n, wi++) { 1237 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1238 break; 1239 if (n == 0) 1240 n = 1; 1241 } 1242 ws[wi] = 0; 1243 free(s); 1244 1245 strg = xcalloc(1, sizeof (strg_t)); 1246 strg->st_tspec = WCHAR; 1247 strg->st_len = wlen; 1248 strg->st_wcp = ws; 1249 1250 yylval.y_strg = strg; 1251 return (T_STRING); 1252 } 1253 1254 /* 1255 * As noted above the scanner does not create new symbol table entries 1256 * for symbols it cannot find in the symbol table. This is to avoid 1257 * putting undeclared symbols into the symbol table if a syntax error 1258 * occurs. 1259 * 1260 * getsym() is called as soon as it is probably ok to put the symbol to 1261 * the symbol table. This does not mean that it is not possible that 1262 * symbols are put to the symbol table which are than not completely 1263 * declared due to syntax errors. To avoid too many problems in this 1264 * case symbols get type int in getsym(). 1265 * 1266 * XXX calls to getsym() should be delayed until decl1*() is called 1267 */ 1268 sym_t * 1269 getsym(sbuf_t *sb) 1270 { 1271 dinfo_t *di; 1272 char *s; 1273 sym_t *sym; 1274 1275 sym = sb->sb_sym; 1276 1277 /* 1278 * During member declaration it is possible that name() looked 1279 * for symbols of type FVFT, although it should have looked for 1280 * symbols of type FTAG. Same can happen for labels. Both cases 1281 * are compensated here. 1282 */ 1283 if (symtyp == FMOS || symtyp == FLAB) { 1284 if (sym == NULL || sym->s_kind == FVFT) 1285 sym = search(sb); 1286 } 1287 1288 if (sym != NULL) { 1289 if (sym->s_kind != symtyp) 1290 lerror("storesym() 1"); 1291 symtyp = FVFT; 1292 freesb(sb); 1293 return (sym); 1294 } 1295 1296 /* create a new symbol table entry */ 1297 1298 /* labels must always be allocated at level 1 (outhermost block) */ 1299 if (symtyp == FLAB) { 1300 sym = getlblk(1, sizeof (sym_t)); 1301 s = getlblk(1, sb->sb_len + 1); 1302 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1303 sym->s_name = s; 1304 sym->s_blklev = 1; 1305 di = dcs; 1306 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1307 di = di->d_nxt; 1308 if (di->d_ctx != AUTO) 1309 lerror("storesym() 2"); 1310 } else { 1311 sym = getblk(sizeof (sym_t)); 1312 sym->s_name = sb->sb_name; 1313 sym->s_blklev = blklev; 1314 di = dcs; 1315 } 1316 1317 UNIQUE_CURR_POS(sym->s_dpos); 1318 if ((sym->s_kind = symtyp) != FLAB) 1319 sym->s_type = gettyp(INT); 1320 1321 symtyp = FVFT; 1322 1323 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1324 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1325 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1326 1327 *di->d_ldlsym = sym; 1328 di->d_ldlsym = &sym->s_dlnxt; 1329 1330 freesb(sb); 1331 return (sym); 1332 } 1333 1334 /* 1335 * Remove a symbol forever from the symbol table. s_blklev 1336 * is set to -1 to avoid that the symbol will later be put 1337 * back to the symbol table. 1338 */ 1339 void 1340 rmsym(sym_t *sym) 1341 { 1342 1343 if ((*sym->s_rlink = sym->s_link) != NULL) 1344 sym->s_link->s_rlink = sym->s_rlink; 1345 sym->s_blklev = -1; 1346 sym->s_link = NULL; 1347 } 1348 1349 /* 1350 * Remove a list of symbols declared at one level from the symbol 1351 * table. 1352 */ 1353 void 1354 rmsyms(sym_t *syms) 1355 { 1356 sym_t *sym; 1357 1358 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1359 if (sym->s_blklev != -1) { 1360 if ((*sym->s_rlink = sym->s_link) != NULL) 1361 sym->s_link->s_rlink = sym->s_rlink; 1362 sym->s_link = NULL; 1363 sym->s_rlink = NULL; 1364 } 1365 } 1366 } 1367 1368 /* 1369 * Put a symbol into the symbol table 1370 */ 1371 void 1372 inssym(int bl, sym_t *sym) 1373 { 1374 int h; 1375 1376 h = hash(sym->s_name); 1377 if ((sym->s_link = symtab[h]) != NULL) 1378 symtab[h]->s_rlink = &sym->s_link; 1379 (symtab[h] = sym)->s_rlink = &symtab[h]; 1380 sym->s_blklev = bl; 1381 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1382 lerror("inssym()"); 1383 } 1384 1385 /* 1386 * Called at level 0 after syntax errors 1387 * Removes all symbols which are not declared at level 0 from the 1388 * symbol table. Also frees all memory which is not associated with 1389 * level 0. 1390 */ 1391 void 1392 cleanup(void) 1393 { 1394 sym_t *sym, *nsym; 1395 int i; 1396 1397 for (i = 0; i < HSHSIZ1; i++) { 1398 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1399 nsym = sym->s_link; 1400 if (sym->s_blklev >= 1) { 1401 if ((*sym->s_rlink = nsym) != NULL) 1402 nsym->s_rlink = sym->s_rlink; 1403 } 1404 } 1405 } 1406 1407 for (i = mblklev; i > 0; i--) 1408 freelblk(i); 1409 } 1410 1411 /* 1412 * Create a new symbol with the name of an existing symbol. 1413 */ 1414 sym_t * 1415 pushdown(sym_t *sym) 1416 { 1417 int h; 1418 sym_t *nsym; 1419 1420 h = hash(sym->s_name); 1421 nsym = getblk(sizeof (sym_t)); 1422 if (sym->s_blklev > blklev) 1423 lerror("pushdown()"); 1424 nsym->s_name = sym->s_name; 1425 UNIQUE_CURR_POS(nsym->s_dpos); 1426 nsym->s_kind = sym->s_kind; 1427 nsym->s_blklev = blklev; 1428 1429 if ((nsym->s_link = symtab[h]) != NULL) 1430 symtab[h]->s_rlink = &nsym->s_link; 1431 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1432 1433 *dcs->d_ldlsym = nsym; 1434 dcs->d_ldlsym = &nsym->s_dlnxt; 1435 1436 return (nsym); 1437 } 1438 1439 /* 1440 * Free any dynamically allocated memory referenced by 1441 * the value stack or yylval. 1442 * The type of information in yylval is described by tok. 1443 */ 1444 void 1445 freeyyv(void *sp, int tok) 1446 { 1447 if (tok == T_NAME || tok == T_TYPENAME) { 1448 sbuf_t *sb = *(sbuf_t **)sp; 1449 freesb(sb); 1450 } else if (tok == T_CON) { 1451 val_t *val = *(val_t **)sp; 1452 free(val); 1453 } else if (tok == T_STRING) { 1454 strg_t *strg = *(strg_t **)sp; 1455 if (strg->st_tspec == CHAR) { 1456 free(strg->st_cp); 1457 } else if (strg->st_tspec == WCHAR) { 1458 free(strg->st_wcp); 1459 } else { 1460 lerror("fryylv() 1"); 1461 } 1462 free(strg); 1463 } 1464 } 1465
Indexes created Wed Oct 01 19:09:53 GMT 2025