scan.l revision 1.18
1 %{ 2 /* $NetBSD: scan.l,v 1.18 2001/05/24 12:10:39 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 1.18 2001/05/24 12:10:39 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 __P((void)); 66 static void badchar __P((int)); 67 static sbuf_t *allocsb __P((void)); 68 static void freesb __P((sbuf_t *)); 69 static int inpc __P((void)); 70 static int hash __P((const char *)); 71 static sym_t *search __P((sbuf_t *)); 72 static int name __P((void)); 73 static int keyw __P((sym_t *)); 74 static int icon __P((int)); 75 static int fcon __P((void)); 76 static int operator __P((int, op_t)); 77 static int ccon __P((void)); 78 static int wccon __P((void)); 79 static int getescc __P((int)); 80 static void directive __P((void)); 81 static void comment __P((void)); 82 static void slashslashcomment __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 "//" slashslashcomment(); 159 . badchar(yytext[0]); 160 161 %% 162 163 static void 164 incline() 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(c) 176 int c; 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() 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() 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(sb) 329 sbuf_t *sb; 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() 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(s) 351 const char *s; 352 { 353 u_int v; 354 const u_char *us; 355 356 v = 0; 357 for (us = (const u_char *)s; *us != '\0'; us++) { 358 v = (v << sizeof (v)) + *us; 359 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 360 } 361 return (v % HSHSIZ1); 362 } 363 364 /* 365 * Lex has found a letter followed by zero or more letters or digits. 366 * It looks for a symbol in the symbol table with the same name. This 367 * symbol must either be a keyword or a symbol of the type required by 368 * symtyp (label, member, tag, ...). 369 * 370 * If it is a keyword, the token is returned. In some cases it is described 371 * more deeply by data written to yylval. 372 * 373 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 374 * is stored in yylval. This struct contains the name of the symbol, it's 375 * length and hash value. If there is already a symbol of the same name 376 * and type in the symbol table, the sbuf struct also contains a pointer 377 * to the symbol table entry. 378 */ 379 static int 380 name() 381 { 382 char *s; 383 sbuf_t *sb; 384 sym_t *sym; 385 int tok; 386 387 sb = allocsb(); 388 sb->sb_name = yytext; 389 sb->sb_len = yyleng; 390 sb->sb_hash = hash(yytext); 391 392 if ((sym = search(sb)) != NULL && sym->s_keyw) { 393 freesb(sb); 394 return (keyw(sym)); 395 } 396 397 sb->sb_sym = sym; 398 399 if (sym != NULL) { 400 if (blklev < sym->s_blklev) 401 lerror("name() 1"); 402 sb->sb_name = sym->s_name; 403 sb->sb_len = strlen(sym->s_name); 404 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 405 } else { 406 s = getblk(yyleng + 1); 407 (void)memcpy(s, yytext, yyleng + 1); 408 sb->sb_name = s; 409 sb->sb_len = yyleng; 410 tok = T_NAME; 411 } 412 413 yylval.y_sb = sb; 414 return (tok); 415 } 416 417 static sym_t * 418 search(sb) 419 sbuf_t *sb; 420 { 421 sym_t *sym; 422 423 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 424 if (strcmp(sym->s_name, sb->sb_name) == 0) { 425 if (sym->s_keyw || sym->s_kind == symtyp) 426 return (sym); 427 } 428 } 429 430 return (NULL); 431 } 432 433 static int 434 keyw(sym) 435 sym_t *sym; 436 { 437 int t; 438 439 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 440 yylval.y_scl = sym->s_scl; 441 } else if (t == T_TYPE || t == T_SOU) { 442 yylval.y_tspec = sym->s_tspec; 443 } else if (t == T_QUAL) { 444 yylval.y_tqual = sym->s_tqual; 445 } 446 return (t); 447 } 448 449 /* 450 * Convert a string representing an integer into internal representation. 451 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 452 */ 453 static int 454 icon(base) 455 int base; 456 { 457 int l_suffix, u_suffix; 458 int len; 459 const char *cp; 460 char c, *eptr; 461 tspec_t typ; 462 u_long ul = 0; 463 u_quad_t uq = 0; 464 int ansiu; 465 static tspec_t contypes[2][3] = { 466 { INT, LONG, QUAD }, 467 { UINT, ULONG, UQUAD } 468 }; 469 470 cp = yytext; 471 len = yyleng; 472 473 /* skip 0x */ 474 if (base == 16) { 475 cp += 2; 476 len -= 2; 477 } 478 479 /* read suffixes */ 480 l_suffix = u_suffix = 0; 481 for ( ; ; ) { 482 if ((c = cp[len - 1]) == 'l' || c == 'L') { 483 l_suffix++; 484 } else if (c == 'u' || c == 'U') { 485 u_suffix++; 486 } else { 487 break; 488 } 489 len--; 490 } 491 if (l_suffix > 2 || u_suffix > 1) { 492 /* malformed integer constant */ 493 warning(251); 494 if (l_suffix > 2) 495 l_suffix = 2; 496 if (u_suffix > 1) 497 u_suffix = 1; 498 } 499 if (tflag && u_suffix != 0) { 500 /* suffix U is illegal in traditional C */ 501 warning(97); 502 } 503 typ = contypes[u_suffix][l_suffix]; 504 505 errno = 0; 506 if (l_suffix < 2) { 507 ul = strtoul(cp, &eptr, base); 508 } else { 509 uq = strtouq(cp, &eptr, base); 510 } 511 if (eptr != cp + len) 512 lerror("icon() 1"); 513 if (errno != 0) 514 /* integer constant out of range */ 515 warning(252); 516 517 /* 518 * If the value is to big for the current type, we must choose 519 * another type. 520 */ 521 ansiu = 0; 522 switch (typ) { 523 case INT: 524 if (ul <= INT_MAX) { 525 /* ok */ 526 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 527 typ = UINT; 528 } else if (ul <= LONG_MAX) { 529 typ = LONG; 530 } else { 531 typ = ULONG; 532 } 533 if (typ == UINT || typ == ULONG) { 534 if (tflag) { 535 typ = LONG; 536 } else if (!sflag) { 537 /* 538 * Remember that the constant is unsigned 539 * only in ANSI C 540 */ 541 ansiu = 1; 542 } 543 } 544 break; 545 case UINT: 546 if (ul > (u_int)UINT_MAX) 547 typ = ULONG; 548 break; 549 case LONG: 550 if (ul > LONG_MAX && !tflag) { 551 typ = ULONG; 552 if (!sflag) 553 ansiu = 1; 554 } 555 break; 556 case QUAD: 557 if (uq > QUAD_MAX && !tflag) { 558 typ = UQUAD; 559 if (!sflag) 560 ansiu = 1; 561 } 562 break; 563 /* LINTED (enumeration values not handled in switch) */ 564 case STRUCT: 565 case VOID: 566 case LDOUBLE: 567 case FUNC: 568 case ARRAY: 569 case PTR: 570 case ENUM: 571 case UNION: 572 case SIGNED: 573 case NOTSPEC: 574 case DOUBLE: 575 case FLOAT: 576 case UQUAD: 577 case ULONG: 578 case USHORT: 579 case SHORT: 580 case UCHAR: 581 case SCHAR: 582 case CHAR: 583 case UNSIGN: 584 break; 585 } 586 587 if (typ != QUAD && typ != UQUAD) { 588 if (isutyp(typ)) { 589 uq = ul; 590 } else { 591 uq = (quad_t)(long)ul; 592 } 593 } 594 595 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 596 597 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 598 yylval.y_val->v_ansiu = ansiu; 599 yylval.y_val->v_quad = (quad_t)uq; 600 601 return (T_CON); 602 } 603 604 /* 605 * Returns 1 if t is a signed type and the value is negative. 606 * 607 * len is the number of significant bits. If len is -1, len is set 608 * to the width of type t. 609 */ 610 int 611 sign(q, t, len) 612 quad_t q; 613 tspec_t t; 614 int len; 615 { 616 if (t == PTR || isutyp(t)) 617 return (0); 618 return (msb(q, t, len)); 619 } 620 621 int 622 msb(q, t, len) 623 quad_t q; 624 tspec_t t; 625 int len; 626 { 627 if (len <= 0) 628 len = size(t); 629 return ((q & qbmasks[len - 1]) != 0); 630 } 631 632 /* 633 * Extends the sign of q. 634 */ 635 quad_t 636 xsign(q, t, len) 637 quad_t q; 638 tspec_t t; 639 int len; 640 { 641 if (len <= 0) 642 len = size(t); 643 644 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 645 q &= qlmasks[len]; 646 } else { 647 q |= qumasks[len]; 648 } 649 return (q); 650 } 651 652 /* 653 * Convert a string representing a floating point value into its interal 654 * representation. Type and value are returned in yylval. fcon() 655 * (and yylex()) returns T_CON. 656 * XXX Currently it is not possible to convert constants of type 657 * long double which are greater then DBL_MAX. 658 */ 659 static int 660 fcon() 661 { 662 const char *cp; 663 int len; 664 tspec_t typ; 665 char c, *eptr; 666 double d; 667 float f = 0; 668 669 cp = yytext; 670 len = yyleng; 671 672 if ((c = cp[len - 1]) == 'f' || c == 'F') { 673 typ = FLOAT; 674 len--; 675 } else if (c == 'l' || c == 'L') { 676 typ = LDOUBLE; 677 len--; 678 } else { 679 typ = DOUBLE; 680 } 681 682 if (tflag && typ != DOUBLE) { 683 /* suffixes F and L are illegal in traditional C */ 684 warning(98); 685 } 686 687 errno = 0; 688 d = strtod(cp, &eptr); 689 if (eptr != cp + len) 690 lerror("fcon() 1"); 691 if (errno != 0) 692 /* floating-point constant out of range */ 693 warning(248); 694 695 if (typ == FLOAT) { 696 f = (float)d; 697 if (isinf(f)) { 698 /* floating-point constant out of range */ 699 warning(248); 700 f = f > 0 ? FLT_MAX : -FLT_MAX; 701 } 702 } 703 704 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 705 if (typ == FLOAT) { 706 yylval.y_val->v_ldbl = f; 707 } else { 708 yylval.y_val->v_ldbl = d; 709 } 710 711 return (T_CON); 712 } 713 714 static int 715 operator(t, o) 716 int t; 717 op_t o; 718 { 719 yylval.y_op = o; 720 return (t); 721 } 722 723 /* 724 * Called if lex found a leading \'. 725 */ 726 static int 727 ccon() 728 { 729 int n, val, c; 730 char cv; 731 732 n = 0; 733 val = 0; 734 while ((c = getescc('\'')) >= 0) { 735 val = (val << CHAR_BIT) + c; 736 n++; 737 } 738 if (c == -2) { 739 /* unterminated character constant */ 740 error(253); 741 } else { 742 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 743 /* too many characters in character constant */ 744 error(71); 745 } else if (n > 1) { 746 /* multi-character character constant */ 747 warning(294); 748 } else if (n == 0) { 749 /* empty character constant */ 750 error(73); 751 } 752 } 753 if (n == 1) { 754 cv = (char)val; 755 val = cv; 756 } 757 758 yylval.y_val = xcalloc(1, sizeof (val_t)); 759 yylval.y_val->v_tspec = INT; 760 yylval.y_val->v_quad = val; 761 762 return (T_CON); 763 } 764 765 /* 766 * Called if lex found a leading L\' 767 */ 768 static int 769 wccon() 770 { 771 static char buf[MB_LEN_MAX + 1]; 772 int i, c; 773 wchar_t wc; 774 775 i = 0; 776 while ((c = getescc('\'')) >= 0) { 777 if (i < MB_CUR_MAX) 778 buf[i] = (char)c; 779 i++; 780 } 781 782 wc = 0; 783 784 if (c == -2) { 785 /* unterminated character constant */ 786 error(253); 787 } else if (c == 0) { 788 /* empty character constant */ 789 error(73); 790 } else { 791 if (i > MB_CUR_MAX) { 792 i = MB_CUR_MAX; 793 /* too many characters in character constant */ 794 error(71); 795 } else { 796 buf[i] = '\0'; 797 (void)mbtowc(NULL, NULL, 0); 798 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 799 /* invalid multibyte character */ 800 error(291); 801 } 802 } 803 804 yylval.y_val = xcalloc(1, sizeof (val_t)); 805 yylval.y_val->v_tspec = WCHAR; 806 yylval.y_val->v_quad = wc; 807 808 return (T_CON); 809 } 810 811 /* 812 * Read a character which is part of a character constant or of a string 813 * and handle escapes. 814 * 815 * The Argument is the character which delimits the character constant or 816 * string. 817 * 818 * Returns -1 if the end of the character constant or string is reached, 819 * -2 if the EOF is reached, and the character otherwise. 820 */ 821 static int 822 getescc(d) 823 int d; 824 { 825 static int pbc = -1; 826 int n, c, v; 827 828 if (pbc == -1) { 829 c = inpc(); 830 } else { 831 c = pbc; 832 pbc = -1; 833 } 834 if (c == d) 835 return (-1); 836 switch (c) { 837 case '\n': 838 if (tflag) { 839 /* newline in string or char constant */ 840 error(254); 841 return (-2); 842 } 843 return (c); 844 case EOF: 845 return (-2); 846 case '\\': 847 switch (c = inpc()) { 848 case '"': 849 if (tflag && d == '\'') 850 /* \" inside character constant undef. ... */ 851 warning(262); 852 return ('"'); 853 case '\'': 854 return ('\''); 855 case '?': 856 if (tflag) 857 /* \? undefined in traditional C */ 858 warning(263); 859 return ('?'); 860 case '\\': 861 return ('\\'); 862 case 'a': 863 if (tflag) 864 /* \a undefined in traditional C */ 865 warning(81); 866 #ifdef __STDC__ 867 return ('\a'); 868 #else 869 return ('\007'); 870 #endif 871 case 'b': 872 return ('\b'); 873 case 'f': 874 return ('\f'); 875 case 'n': 876 return ('\n'); 877 case 'r': 878 return ('\r'); 879 case 't': 880 return ('\t'); 881 case 'v': 882 if (tflag) 883 /* \v undefined in traditional C */ 884 warning(264); 885 #ifdef __STDC__ 886 return ('\v'); 887 #else 888 return ('\013'); 889 #endif 890 case '8': case '9': 891 /* bad octal digit %c */ 892 warning(77, c); 893 /* FALLTHROUGH */ 894 case '0': case '1': case '2': case '3': 895 case '4': case '5': case '6': case '7': 896 n = 3; 897 v = 0; 898 do { 899 v = (v << 3) + (c - '0'); 900 c = inpc(); 901 } while (--n && isdigit(c) && (tflag || c <= '7')); 902 if (tflag && n > 0 && isdigit(c)) 903 /* bad octal digit %c */ 904 warning(77, c); 905 pbc = c; 906 if (v > UCHAR_MAX) { 907 /* character escape does not fit in char. */ 908 warning(76); 909 v &= CHAR_MASK; 910 } 911 return (v); 912 case 'x': 913 if (tflag) 914 /* \x undefined in traditional C */ 915 warning(82); 916 v = 0; 917 n = 0; 918 while ((c = inpc()) >= 0 && isxdigit(c)) { 919 c = isdigit(c) ? 920 c - '0' : toupper(c) - 'A' + 10; 921 v = (v << 4) + c; 922 if (n >= 0) { 923 if ((v & ~CHAR_MASK) != 0) { 924 /* overflow in hex escape */ 925 warning(75); 926 n = -1; 927 } else { 928 n++; 929 } 930 } 931 } 932 pbc = c; 933 if (n == 0) { 934 /* no hex digits follow \x */ 935 error(74); 936 } if (n == -1) { 937 v &= CHAR_MASK; 938 } 939 return (v); 940 case '\n': 941 return (getescc(d)); 942 case EOF: 943 return (-2); 944 default: 945 if (isprint(c)) { 946 /* dubious escape \%c */ 947 warning(79, c); 948 } else { 949 /* dubious escape \%o */ 950 warning(80, c); 951 } 952 } 953 } 954 return (c); 955 } 956 957 /* 958 * Called for preprocessor directives. Currently implemented are: 959 * # lineno 960 * # lineno "filename" 961 */ 962 static void 963 directive() 964 { 965 const char *cp, *fn; 966 char c, *eptr; 967 size_t fnl; 968 long ln; 969 static int first = 1; 970 971 /* Go to first non-whitespace after # */ 972 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 973 974 if (!isdigit((unsigned char)c)) { 975 error: 976 /* undefined or invalid # directive */ 977 warning(255); 978 return; 979 } 980 ln = strtol(--cp, &eptr, 10); 981 if (cp == eptr) 982 goto error; 983 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 984 goto error; 985 while ((c = *cp++) == ' ' || c == '\t') 986 continue; 987 if (c != '\0') { 988 if (c != '"') 989 goto error; 990 fn = cp; 991 while ((c = *cp) != '"' && c != '\0') 992 cp++; 993 if (c != '"') 994 goto error; 995 if ((fnl = cp++ - fn) > PATH_MAX) 996 goto error; 997 while ((c = *cp++) == ' ' || c == '\t') 998 continue; 999 #if 0 1000 if (c != '\0') 1001 warning("extra character(s) after directive"); 1002 #endif 1003 1004 /* empty string means stdin */ 1005 if (fnl == 0) { 1006 fn = "{standard input}"; 1007 fnl = 16; /* strlen (fn) */ 1008 } 1009 curr_pos.p_file = fnnalloc(fn, fnl); 1010 /* 1011 * If this is the first directive, the name is the name 1012 * of the C source file as specified at the command line. 1013 * It is written to the output file. 1014 */ 1015 if (first) { 1016 csrc_pos.p_file = curr_pos.p_file; 1017 outsrc(curr_pos.p_file); 1018 first = 0; 1019 } 1020 } 1021 curr_pos.p_line = (int)ln - 1; 1022 curr_pos.p_uniq = 0; 1023 if (curr_pos.p_file == csrc_pos.p_file) { 1024 csrc_pos.p_line = (int)ln - 1; 1025 csrc_pos.p_uniq = 0; 1026 } 1027 } 1028 1029 /* 1030 * Handle lint comments. Following comments are currently understood: 1031 * ARGSUSEDn 1032 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1033 * FALLTHRU FALLTHROUGH 1034 * LINTLIBRARY 1035 * LINTED NOSTRICT 1036 * LONGLONG 1037 * NOTREACHED 1038 * PRINTFLIKEn 1039 * PROTOLIB 1040 * SCANFLIKEn 1041 * VARARGSn 1042 * If one of this comments is recognized, the arguments, if any, are 1043 * parsed and a function which handles this comment is called. 1044 */ 1045 static void 1046 comment() 1047 { 1048 int c, lc; 1049 static struct { 1050 const char *keywd; 1051 int arg; 1052 void (*func) __P((int)); 1053 } keywtab[] = { 1054 { "ARGSUSED", 1, argsused }, 1055 { "CONSTCOND", 0, constcond }, 1056 { "CONSTANTCOND", 0, constcond }, 1057 { "CONSTANTCONDITION", 0, constcond }, 1058 { "FALLTHRU", 0, fallthru }, 1059 { "FALLTHROUGH", 0, fallthru }, 1060 { "LINTLIBRARY", 0, lintlib }, 1061 { "LINTED", 0, linted }, 1062 { "LONGLONG", 0, longlong }, 1063 { "NOSTRICT", 0, linted }, 1064 { "NOTREACHED", 0, notreach }, 1065 { "PRINTFLIKE", 1, printflike }, 1066 { "PROTOLIB", 1, protolib }, 1067 { "SCANFLIKE", 1, scanflike }, 1068 { "VARARGS", 1, varargs }, 1069 }; 1070 char keywd[32]; 1071 char arg[32]; 1072 int l, i, a; 1073 int eoc; 1074 1075 eoc = 0; 1076 1077 /* Skip white spaces after the start of the comment */ 1078 while ((c = inpc()) != EOF && isspace(c)) 1079 continue; 1080 1081 /* Read the potential keyword to keywd */ 1082 l = 0; 1083 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1084 keywd[l++] = (char)c; 1085 c = inpc(); 1086 } 1087 keywd[l] = '\0'; 1088 1089 /* look for the keyword */ 1090 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1091 if (strcmp(keywtab[i].keywd, keywd) == 0) 1092 break; 1093 } 1094 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1095 goto skip_rest; 1096 1097 /* skip white spaces after the keyword */ 1098 while (c != EOF && isspace(c)) 1099 c = inpc(); 1100 1101 /* read the argument, if the keyword accepts one and there is one */ 1102 l = 0; 1103 if (keywtab[i].arg) { 1104 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1105 arg[l++] = (char)c; 1106 c = inpc(); 1107 } 1108 } 1109 arg[l] = '\0'; 1110 a = l != 0 ? atoi(arg) : -1; 1111 1112 /* skip white spaces after the argument */ 1113 while (c != EOF && isspace(c)) 1114 c = inpc(); 1115 1116 if (c != '*' || (c = inpc()) != '/') { 1117 if (keywtab[i].func != linted) 1118 /* extra characters in lint comment */ 1119 warning(257); 1120 } else { 1121 /* 1122 * remember that we have already found the end of the 1123 * comment 1124 */ 1125 eoc = 1; 1126 } 1127 1128 if (keywtab[i].func != NULL) 1129 (*keywtab[i].func)(a); 1130 1131 skip_rest: 1132 while (!eoc) { 1133 lc = c; 1134 if ((c = inpc()) == EOF) { 1135 /* unterminated comment */ 1136 error(256); 1137 break; 1138 } 1139 if (lc == '*' && c == '/') 1140 eoc = 1; 1141 } 1142 } 1143 1144 /* 1145 * Handle // style comments 1146 */ 1147 static void 1148 slashslashcomment() 1149 { 1150 int c; 1151 1152 if (sflag < 2 && !gflag) 1153 /* // comments only support in C99 */ 1154 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1155 1156 while ((c = inpc()) != EOF && c != '\n') 1157 continue; 1158 } 1159 1160 /* 1161 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1162 * clrwflgs() is called after function definitions and global and 1163 * local declarations and definitions. It is also called between 1164 * the controlling expression and the body of control statements 1165 * (if, switch, for, while). 1166 */ 1167 void 1168 clrwflgs() 1169 { 1170 nowarn = 0; 1171 quadflg = 0; 1172 ccflg = 0; 1173 } 1174 1175 /* 1176 * Strings are stored in a dynamically alloceted buffer and passed 1177 * in yylval.y_xstrg to the parser. The parser or the routines called 1178 * by the parser are responsible for freeing this buffer. 1179 */ 1180 static int 1181 string() 1182 { 1183 u_char *s; 1184 int c; 1185 size_t len, max; 1186 strg_t *strg; 1187 1188 s = xmalloc(max = 64); 1189 1190 len = 0; 1191 while ((c = getescc('"')) >= 0) { 1192 /* +1 to reserve space for a trailing NUL character */ 1193 if (len + 1 == max) 1194 s = xrealloc(s, max *= 2); 1195 s[len++] = (char)c; 1196 } 1197 s[len] = '\0'; 1198 if (c == -2) 1199 /* unterminated string constant */ 1200 error(258); 1201 1202 strg = xcalloc(1, sizeof (strg_t)); 1203 strg->st_tspec = CHAR; 1204 strg->st_len = len; 1205 strg->st_cp = s; 1206 1207 yylval.y_strg = strg; 1208 return (T_STRING); 1209 } 1210 1211 static int 1212 wcstrg() 1213 { 1214 char *s; 1215 int c, i, n, wi; 1216 size_t len, max, wlen; 1217 wchar_t *ws; 1218 strg_t *strg; 1219 1220 s = xmalloc(max = 64); 1221 len = 0; 1222 while ((c = getescc('"')) >= 0) { 1223 /* +1 to save space for a trailing NUL character */ 1224 if (len + 1 >= max) 1225 s = xrealloc(s, max *= 2); 1226 s[len++] = (char)c; 1227 } 1228 s[len] = '\0'; 1229 if (c == -2) 1230 /* unterminated string constant */ 1231 error(258); 1232 1233 /* get length of wide character string */ 1234 (void)mblen(NULL, 0); 1235 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1236 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1237 /* invalid multibyte character */ 1238 error(291); 1239 break; 1240 } 1241 if (n == 0) 1242 n = 1; 1243 } 1244 1245 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1246 1247 /* convert from multibyte to wide char */ 1248 (void)mbtowc(NULL, NULL, 0); 1249 for (i = 0, wi = 0; i < len; i += n, wi++) { 1250 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1251 break; 1252 if (n == 0) 1253 n = 1; 1254 } 1255 ws[wi] = 0; 1256 free(s); 1257 1258 strg = xcalloc(1, sizeof (strg_t)); 1259 strg->st_tspec = WCHAR; 1260 strg->st_len = wlen; 1261 strg->st_wcp = ws; 1262 1263 yylval.y_strg = strg; 1264 return (T_STRING); 1265 } 1266 1267 /* 1268 * As noted above the scanner does not create new symbol table entries 1269 * for symbols it cannot find in the symbol table. This is to avoid 1270 * putting undeclared symbols into the symbol table if a syntax error 1271 * occurs. 1272 * 1273 * getsym() is called as soon as it is probably ok to put the symbol to 1274 * the symbol table. This does not mean that it is not possible that 1275 * symbols are put to the symbol table which are than not completely 1276 * declared due to syntax errors. To avoid too many problems in this 1277 * case symbols get type int in getsym(). 1278 * 1279 * XXX calls to getsym() should be delayed until decl1*() is called 1280 */ 1281 sym_t * 1282 getsym(sb) 1283 sbuf_t *sb; 1284 { 1285 dinfo_t *di; 1286 char *s; 1287 sym_t *sym; 1288 1289 sym = sb->sb_sym; 1290 1291 /* 1292 * During member declaration it is possible that name() looked 1293 * for symbols of type FVFT, although it should have looked for 1294 * symbols of type FTAG. Same can happen for labels. Both cases 1295 * are compensated here. 1296 */ 1297 if (symtyp == FMOS || symtyp == FLAB) { 1298 if (sym == NULL || sym->s_kind == FVFT) 1299 sym = search(sb); 1300 } 1301 1302 if (sym != NULL) { 1303 if (sym->s_kind != symtyp) 1304 lerror("storesym() 1"); 1305 symtyp = FVFT; 1306 freesb(sb); 1307 return (sym); 1308 } 1309 1310 /* create a new symbol table entry */ 1311 1312 /* labels must always be allocated at level 1 (outhermost block) */ 1313 if (symtyp == FLAB) { 1314 sym = getlblk(1, sizeof (sym_t)); 1315 s = getlblk(1, sb->sb_len + 1); 1316 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1317 sym->s_name = s; 1318 sym->s_blklev = 1; 1319 di = dcs; 1320 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1321 di = di->d_nxt; 1322 if (di->d_ctx != AUTO) 1323 lerror("storesym() 2"); 1324 } else { 1325 sym = getblk(sizeof (sym_t)); 1326 sym->s_name = sb->sb_name; 1327 sym->s_blklev = blklev; 1328 di = dcs; 1329 } 1330 1331 UNIQUE_CURR_POS(sym->s_dpos); 1332 if ((sym->s_kind = symtyp) != FLAB) 1333 sym->s_type = gettyp(INT); 1334 1335 symtyp = FVFT; 1336 1337 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1338 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1339 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1340 1341 *di->d_ldlsym = sym; 1342 di->d_ldlsym = &sym->s_dlnxt; 1343 1344 freesb(sb); 1345 return (sym); 1346 } 1347 1348 /* 1349 * Remove a symbol forever from the symbol table. s_blklev 1350 * is set to -1 to avoid that the symbol will later be put 1351 * back to the symbol table. 1352 */ 1353 void 1354 rmsym(sym) 1355 sym_t *sym; 1356 { 1357 if ((*sym->s_rlink = sym->s_link) != NULL) 1358 sym->s_link->s_rlink = sym->s_rlink; 1359 sym->s_blklev = -1; 1360 sym->s_link = NULL; 1361 } 1362 1363 /* 1364 * Remove a list of symbols declared at one level from the symbol 1365 * table. 1366 */ 1367 void 1368 rmsyms(syms) 1369 sym_t *syms; 1370 { 1371 sym_t *sym; 1372 1373 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1374 if (sym->s_blklev != -1) { 1375 if ((*sym->s_rlink = sym->s_link) != NULL) 1376 sym->s_link->s_rlink = sym->s_rlink; 1377 sym->s_link = NULL; 1378 sym->s_rlink = NULL; 1379 } 1380 } 1381 } 1382 1383 /* 1384 * Put a symbol into the symbol table 1385 */ 1386 void 1387 inssym(bl, sym) 1388 int bl; 1389 sym_t *sym; 1390 { 1391 int h; 1392 1393 h = hash(sym->s_name); 1394 if ((sym->s_link = symtab[h]) != NULL) 1395 symtab[h]->s_rlink = &sym->s_link; 1396 (symtab[h] = sym)->s_rlink = &symtab[h]; 1397 sym->s_blklev = bl; 1398 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1399 lerror("inssym()"); 1400 } 1401 1402 /* 1403 * Called at level 0 after syntax errors 1404 * Removes all symbols which are not declared at level 0 from the 1405 * symbol table. Also frees all memory which is not associated with 1406 * level 0. 1407 */ 1408 void 1409 cleanup() 1410 { 1411 sym_t *sym, *nsym; 1412 int i; 1413 1414 for (i = 0; i < HSHSIZ1; i++) { 1415 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1416 nsym = sym->s_link; 1417 if (sym->s_blklev >= 1) { 1418 if ((*sym->s_rlink = nsym) != NULL) 1419 nsym->s_rlink = sym->s_rlink; 1420 } 1421 } 1422 } 1423 1424 for (i = mblklev; i > 0; i--) 1425 freelblk(i); 1426 } 1427 1428 /* 1429 * Create a new symbol with the name of an existing symbol. 1430 */ 1431 sym_t * 1432 pushdown(sym) 1433 sym_t *sym; 1434 { 1435 int h; 1436 sym_t *nsym; 1437 1438 h = hash(sym->s_name); 1439 nsym = getblk(sizeof (sym_t)); 1440 if (sym->s_blklev > blklev) 1441 lerror("pushdown()"); 1442 nsym->s_name = sym->s_name; 1443 UNIQUE_CURR_POS(nsym->s_dpos); 1444 nsym->s_kind = sym->s_kind; 1445 nsym->s_blklev = blklev; 1446 1447 if ((nsym->s_link = symtab[h]) != NULL) 1448 symtab[h]->s_rlink = &nsym->s_link; 1449 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1450 1451 *dcs->d_ldlsym = nsym; 1452 dcs->d_ldlsym = &nsym->s_dlnxt; 1453 1454 return (nsym); 1455 } 1456 1457 /* 1458 * Free any dynamically allocated memory referenced by 1459 * the value stack or yylval. 1460 * The type of information in yylval is described by tok. 1461 */ 1462 void 1463 freeyyv(sp, tok) 1464 void *sp; 1465 int tok; 1466 { 1467 if (tok == T_NAME || tok == T_TYPENAME) { 1468 sbuf_t *sb = *(sbuf_t **)sp; 1469 freesb(sb); 1470 } else if (tok == T_CON) { 1471 val_t *val = *(val_t **)sp; 1472 free(val); 1473 } else if (tok == T_STRING) { 1474 strg_t *strg = *(strg_t **)sp; 1475 if (strg->st_tspec == CHAR) { 1476 free(strg->st_cp); 1477 } else if (strg->st_tspec == WCHAR) { 1478 free(strg->st_wcp); 1479 } else { 1480 lerror("fryylv() 1"); 1481 } 1482 free(strg); 1483 } 1484 } 1485
Indexes created Wed Oct 01 15:09:59 GMT 2025