scan.l revision 1.14.2.1
1 %{ 2 /* $NetBSD: scan.l,v 1.14.2.1 2000/06/23 16:40:19 minoura 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 1.14.2.1 2000/06/23 16:40:19 minoura 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 int string __P((void)); 83 static int wcstrg __P((void)); 84 85 %} 86 87 L [_A-Za-z] 88 D [0-9] 89 NZD [1-9] 90 OD [0-7] 91 HD [0-9A-Fa-f] 92 EX ([eE][+-]?[0-9]+) 93 94 %% 95 96 {L}({L}|{D})* return (name()); 97 0{OD}*[lLuU]* return (icon(8)); 98 {NZD}{D}*[lLuU]* return (icon(10)); 99 0[xX]{HD}+[lLuU]* return (icon(16)); 100 {D}+\.{D}*{EX}?[fFlL]? | 101 {D}+{EX}[fFlL]? | 102 \.{D}+{EX}?[fFlL]? return (fcon()); 103 "=" return (operator(T_ASSIGN, ASSIGN)); 104 "*=" return (operator(T_OPASS, MULASS)); 105 "/=" return (operator(T_OPASS, DIVASS)); 106 "%=" return (operator(T_OPASS, MODASS)); 107 "+=" return (operator(T_OPASS, ADDASS)); 108 "-=" return (operator(T_OPASS, SUBASS)); 109 "<<=" return (operator(T_OPASS, SHLASS)); 110 ">>=" return (operator(T_OPASS, SHRASS)); 111 "&=" return (operator(T_OPASS, ANDASS)); 112 "^=" return (operator(T_OPASS, XORASS)); 113 "|=" return (operator(T_OPASS, ORASS)); 114 "||" return (operator(T_LOGOR, LOGOR)); 115 "&&" return (operator(T_LOGAND, LOGAND)); 116 "|" return (operator(T_OR, OR)); 117 "&" return (operator(T_AND, AND)); 118 "^" return (operator(T_XOR, XOR)); 119 "==" return (operator(T_EQOP, EQ)); 120 "!=" return (operator(T_EQOP, NE)); 121 "<" return (operator(T_RELOP, LT)); 122 ">" return (operator(T_RELOP, GT)); 123 "<=" return (operator(T_RELOP, LE)); 124 ">=" return (operator(T_RELOP, GE)); 125 "<<" return (operator(T_SHFTOP, SHL)); 126 ">>" return (operator(T_SHFTOP, SHR)); 127 "++" return (operator(T_INCDEC, INC)); 128 "--" return (operator(T_INCDEC, DEC)); 129 "->" return (operator(T_STROP, ARROW)); 130 "." return (operator(T_STROP, POINT)); 131 "+" return (operator(T_ADDOP, PLUS)); 132 "-" return (operator(T_ADDOP, MINUS)); 133 "*" return (operator(T_MULT, MULT)); 134 "/" return (operator(T_DIVOP, DIV)); 135 "%" return (operator(T_DIVOP, MOD)); 136 "!" return (operator(T_UNOP, NOT)); 137 "~" return (operator(T_UNOP, COMPL)); 138 "\"" return (string()); 139 "L\"" return (wcstrg()); 140 ";" return (T_SEMI); 141 "{" return (T_LBRACE); 142 "}" return (T_RBRACE); 143 "," return (T_COMMA); 144 ":" return (T_COLON); 145 "?" return (T_QUEST); 146 "[" return (T_LBRACK); 147 "]" return (T_RBRACK); 148 "(" return (T_LPARN); 149 ")" return (T_RPARN); 150 "..." return (T_ELLIPSE); 151 "'" return (ccon()); 152 "L'" return (wccon()); 153 ^#.*$ directive(); 154 \n incline(); 155 \t|" "|\f|\v ; 156 "/*" comment(); 157 . badchar(yytext[0]); 158 159 %% 160 161 static void 162 incline() 163 { 164 curr_pos.p_line++; 165 curr_pos.p_uniq = 0; 166 if (curr_pos.p_file == csrc_pos.p_file) { 167 csrc_pos.p_line++; 168 csrc_pos.p_uniq = 0; 169 } 170 } 171 172 static void 173 badchar(c) 174 int c; 175 { 176 /* unknown character \%o */ 177 error(250, c); 178 } 179 180 /* 181 * Keywords. 182 * During initialisation they are written to the symbol table. 183 */ 184 static struct kwtab { 185 const char *kw_name; /* keyword */ 186 int kw_token; /* token returned by yylex() */ 187 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 188 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 189 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 190 u_int kw_stdc : 1; /* STDC keyword */ 191 u_int kw_gcc : 1; /* GCC keyword */ 192 } kwtab[] = { 193 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 194 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 195 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 196 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 197 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 198 { "case", T_CASE, 0, 0, 0, 0, 0 }, 199 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 200 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 201 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 202 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 203 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 204 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 205 { "do", T_DO, 0, 0, 0, 0, 0 }, 206 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 207 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 208 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 209 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 210 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 211 { "for", T_FOR, 0, 0, 0, 0, 0 }, 212 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 213 { "if", T_IF, 0, 0, 0, 0, 0 }, 214 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 215 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 216 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 217 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 218 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 }, 219 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 220 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 221 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 222 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 223 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 224 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 225 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 226 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 227 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 228 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 229 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 230 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 231 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 232 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 233 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 234 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 235 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 236 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 237 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 238 { NULL, 0, 0, 0, 0, 0, 0 } 239 }; 240 241 /* Symbol table */ 242 static sym_t *symtab[HSHSIZ1]; 243 244 /* bit i of the entry with index i is set */ 245 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 246 247 /* least significant i bits are set in the entry with index i */ 248 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 249 250 /* least significant i bits are not set in the entry with index i */ 251 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 252 253 /* free list for sbuf structures */ 254 static sbuf_t *sbfrlst; 255 256 /* Typ of next expected symbol */ 257 symt_t symtyp; 258 259 260 /* 261 * All keywords are written to the symbol table. This saves us looking 262 * in a extra table for each name we found. 263 */ 264 void 265 initscan() 266 { 267 struct kwtab *kw; 268 sym_t *sym; 269 int h, i; 270 u_quad_t uq; 271 272 for (kw = kwtab; kw->kw_name != NULL; kw++) { 273 if (kw->kw_stdc && tflag) 274 continue; 275 if (kw->kw_gcc && !gflag) 276 continue; 277 sym = getblk(sizeof (sym_t)); 278 sym->s_name = kw->kw_name; 279 sym->s_keyw = 1; 280 sym->s_value.v_quad = kw->kw_token; 281 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 282 sym->s_tspec = kw->kw_tspec; 283 } else if (kw->kw_token == T_SCLASS) { 284 sym->s_scl = kw->kw_scl; 285 } else if (kw->kw_token == T_QUAL) { 286 sym->s_tqual = kw->kw_tqual; 287 } 288 h = hash(sym->s_name); 289 if ((sym->s_link = symtab[h]) != NULL) 290 symtab[h]->s_rlink = &sym->s_link; 291 (symtab[h] = sym)->s_rlink = &symtab[h]; 292 } 293 294 /* initialize bit-masks for quads */ 295 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 296 qbmasks[i] = (u_quad_t)1 << i; 297 uq = ~(u_quad_t)0 << i; 298 qumasks[i] = uq; 299 qlmasks[i] = ~uq; 300 } 301 qumasks[i] = 0; 302 qlmasks[i] = ~(u_quad_t)0; 303 } 304 305 /* 306 * Get a free sbuf structure, if possible from the free list 307 */ 308 static sbuf_t * 309 allocsb() 310 { 311 sbuf_t *sb; 312 313 if ((sb = sbfrlst) != NULL) { 314 sbfrlst = sb->sb_nxt; 315 } else { 316 sb = xmalloc(sizeof (sbuf_t)); 317 } 318 (void)memset(sb, 0, sizeof (sb)); 319 return (sb); 320 } 321 322 /* 323 * Put a sbuf structure to the free list 324 */ 325 static void 326 freesb(sb) 327 sbuf_t *sb; 328 { 329 sb->sb_nxt = sbfrlst; 330 sbfrlst = sb; 331 } 332 333 /* 334 * Read a character and ensure that it is positive (except EOF). 335 * Increment line count(s) if necessary. 336 */ 337 static int 338 inpc() 339 { 340 int c; 341 342 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 343 incline(); 344 return (c); 345 } 346 347 static int 348 hash(s) 349 const char *s; 350 { 351 u_int v; 352 const u_char *us; 353 354 v = 0; 355 for (us = (const u_char *)s; *us != '\0'; us++) { 356 v = (v << sizeof (v)) + *us; 357 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 358 } 359 return (v % HSHSIZ1); 360 } 361 362 /* 363 * Lex has found a letter followed by zero or more letters or digits. 364 * It looks for a symbol in the symbol table with the same name. This 365 * symbol must either be a keyword or a symbol of the type required by 366 * symtyp (label, member, tag, ...). 367 * 368 * If it is a keyword, the token is returned. In some cases it is described 369 * more deeply by data written to yylval. 370 * 371 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 372 * is stored in yylval. This struct contains the name of the symbol, it's 373 * length and hash value. If there is already a symbol of the same name 374 * and type in the symbol table, the sbuf struct also contains a pointer 375 * to the symbol table entry. 376 */ 377 static int 378 name() 379 { 380 char *s; 381 sbuf_t *sb; 382 sym_t *sym; 383 int tok; 384 385 sb = allocsb(); 386 sb->sb_name = yytext; 387 sb->sb_len = yyleng; 388 sb->sb_hash = hash(yytext); 389 390 if ((sym = search(sb)) != NULL && sym->s_keyw) { 391 freesb(sb); 392 return (keyw(sym)); 393 } 394 395 sb->sb_sym = sym; 396 397 if (sym != NULL) { 398 if (blklev < sym->s_blklev) 399 lerror("name() 1"); 400 sb->sb_name = sym->s_name; 401 sb->sb_len = strlen(sym->s_name); 402 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 403 } else { 404 s = getblk(yyleng + 1); 405 (void)memcpy(s, yytext, yyleng + 1); 406 sb->sb_name = s; 407 sb->sb_len = yyleng; 408 tok = T_NAME; 409 } 410 411 yylval.y_sb = sb; 412 return (tok); 413 } 414 415 static sym_t * 416 search(sb) 417 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) 433 sym_t *sym; 434 { 435 int t; 436 437 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 438 yylval.y_scl = sym->s_scl; 439 } else if (t == T_TYPE || t == T_SOU) { 440 yylval.y_tspec = sym->s_tspec; 441 } else if (t == T_QUAL) { 442 yylval.y_tqual = sym->s_tqual; 443 } 444 return (t); 445 } 446 447 /* 448 * Convert a string representing an integer into internal representation. 449 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 450 */ 451 static int 452 icon(base) 453 int base; 454 { 455 int l_suffix, u_suffix; 456 int len; 457 const char *cp; 458 char c, *eptr; 459 tspec_t typ; 460 u_long ul = 0; 461 u_quad_t uq = 0; 462 int ansiu; 463 static tspec_t contypes[2][3] = { 464 { INT, LONG, QUAD }, 465 { UINT, ULONG, UQUAD } 466 }; 467 468 cp = yytext; 469 len = yyleng; 470 471 /* skip 0x */ 472 if (base == 16) { 473 cp += 2; 474 len -= 2; 475 } 476 477 /* read suffixes */ 478 l_suffix = u_suffix = 0; 479 for ( ; ; ) { 480 if ((c = cp[len - 1]) == 'l' || c == 'L') { 481 l_suffix++; 482 } else if (c == 'u' || c == 'U') { 483 u_suffix++; 484 } else { 485 break; 486 } 487 len--; 488 } 489 if (l_suffix > 2 || u_suffix > 1) { 490 /* malformed integer constant */ 491 warning(251); 492 if (l_suffix > 2) 493 l_suffix = 2; 494 if (u_suffix > 1) 495 u_suffix = 1; 496 } 497 if (tflag && u_suffix != 0) { 498 /* suffix U is illegal in traditional C */ 499 warning(97); 500 } 501 typ = contypes[u_suffix][l_suffix]; 502 503 errno = 0; 504 if (l_suffix < 2) { 505 ul = strtoul(cp, &eptr, base); 506 } else { 507 uq = strtouq(cp, &eptr, base); 508 } 509 if (eptr != cp + len) 510 lerror("icon() 1"); 511 if (errno != 0) 512 /* integer constant out of range */ 513 warning(252); 514 515 /* 516 * If the value is to big for the current type, we must choose 517 * another type. 518 */ 519 ansiu = 0; 520 switch (typ) { 521 case INT: 522 if (ul <= INT_MAX) { 523 /* ok */ 524 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 525 typ = UINT; 526 } else if (ul <= LONG_MAX) { 527 typ = LONG; 528 } else { 529 typ = ULONG; 530 } 531 if (typ == UINT || typ == ULONG) { 532 if (tflag) { 533 typ = LONG; 534 } else if (!sflag) { 535 /* 536 * Remember that the constant is unsigned 537 * only in ANSI C 538 */ 539 ansiu = 1; 540 } 541 } 542 break; 543 case UINT: 544 if (ul > (u_int)UINT_MAX) 545 typ = ULONG; 546 break; 547 case LONG: 548 if (ul > LONG_MAX && !tflag) { 549 typ = ULONG; 550 if (!sflag) 551 ansiu = 1; 552 } 553 break; 554 case QUAD: 555 if (uq > QUAD_MAX && !tflag) { 556 typ = UQUAD; 557 if (!sflag) 558 ansiu = 1; 559 } 560 break; 561 /* LINTED (enumeration values not handled in switch) */ 562 case STRUCT: 563 case VOID: 564 case LDOUBLE: 565 case FUNC: 566 case ARRAY: 567 case PTR: 568 case ENUM: 569 case UNION: 570 case SIGNED: 571 case NOTSPEC: 572 case DOUBLE: 573 case FLOAT: 574 case UQUAD: 575 case ULONG: 576 case USHORT: 577 case SHORT: 578 case UCHAR: 579 case SCHAR: 580 case CHAR: 581 case UNSIGN: 582 break; 583 } 584 585 if (typ != QUAD && typ != UQUAD) { 586 if (isutyp(typ)) { 587 uq = ul; 588 } else { 589 uq = (quad_t)(long)ul; 590 } 591 } 592 593 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 594 595 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 596 yylval.y_val->v_ansiu = ansiu; 597 yylval.y_val->v_quad = (quad_t)uq; 598 599 return (T_CON); 600 } 601 602 /* 603 * Returns 1 if t is a signed type and the value is negative. 604 * 605 * len is the number of significant bits. If len is -1, len is set 606 * to the width of type t. 607 */ 608 int 609 sign(q, t, len) 610 quad_t q; 611 tspec_t t; 612 int len; 613 { 614 if (t == PTR || isutyp(t)) 615 return (0); 616 return (msb(q, t, len)); 617 } 618 619 int 620 msb(q, t, len) 621 quad_t q; 622 tspec_t t; 623 int len; 624 { 625 if (len <= 0) 626 len = size(t); 627 return ((q & qbmasks[len - 1]) != 0); 628 } 629 630 /* 631 * Extends the sign of q. 632 */ 633 quad_t 634 xsign(q, t, len) 635 quad_t q; 636 tspec_t t; 637 int len; 638 { 639 if (len <= 0) 640 len = size(t); 641 642 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 643 q &= qlmasks[len]; 644 } else { 645 q |= qumasks[len]; 646 } 647 return (q); 648 } 649 650 /* 651 * Convert a string representing a floating point value into its interal 652 * representation. Type and value are returned in yylval. fcon() 653 * (and yylex()) returns T_CON. 654 * XXX Currently it is not possible to convert constants of type 655 * long double which are greater then DBL_MAX. 656 */ 657 static int 658 fcon() 659 { 660 const char *cp; 661 int len; 662 tspec_t typ; 663 char c, *eptr; 664 double d; 665 float f = 0; 666 667 cp = yytext; 668 len = yyleng; 669 670 if ((c = cp[len - 1]) == 'f' || c == 'F') { 671 typ = FLOAT; 672 len--; 673 } else if (c == 'l' || c == 'L') { 674 typ = LDOUBLE; 675 len--; 676 } else { 677 typ = DOUBLE; 678 } 679 680 if (tflag && typ != DOUBLE) { 681 /* suffixes F and L are illegal in traditional C */ 682 warning(98); 683 } 684 685 errno = 0; 686 d = strtod(cp, &eptr); 687 if (eptr != cp + len) 688 lerror("fcon() 1"); 689 if (errno != 0) 690 /* floating-point constant out of range */ 691 warning(248); 692 693 if (typ == FLOAT) { 694 f = (float)d; 695 if (isinf(f)) { 696 /* floating-point constant out of range */ 697 warning(248); 698 f = f > 0 ? FLT_MAX : -FLT_MAX; 699 } 700 } 701 702 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 703 if (typ == FLOAT) { 704 yylval.y_val->v_ldbl = f; 705 } else { 706 yylval.y_val->v_ldbl = d; 707 } 708 709 return (T_CON); 710 } 711 712 static int 713 operator(t, o) 714 int t; 715 op_t o; 716 { 717 yylval.y_op = o; 718 return (t); 719 } 720 721 /* 722 * Called if lex found a leading \'. 723 */ 724 static int 725 ccon() 726 { 727 int n, val, c; 728 char cv; 729 730 n = 0; 731 val = 0; 732 while ((c = getescc('\'')) >= 0) { 733 val = (val << CHAR_BIT) + c; 734 n++; 735 } 736 if (c == -2) { 737 /* unterminated character constant */ 738 error(253); 739 } else { 740 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 741 /* too many characters in character constant */ 742 error(71); 743 } else if (n > 1) { 744 /* multi-character character constant */ 745 warning(294); 746 } else if (n == 0) { 747 /* empty character constant */ 748 error(73); 749 } 750 } 751 if (n == 1) { 752 cv = (char)val; 753 val = cv; 754 } 755 756 yylval.y_val = xcalloc(1, sizeof (val_t)); 757 yylval.y_val->v_tspec = INT; 758 yylval.y_val->v_quad = val; 759 760 return (T_CON); 761 } 762 763 /* 764 * Called if lex found a leading L\' 765 */ 766 static int 767 wccon() 768 { 769 static char buf[MB_LEN_MAX + 1]; 770 int i, c; 771 wchar_t wc; 772 773 i = 0; 774 while ((c = getescc('\'')) >= 0) { 775 if (i < MB_CUR_MAX) 776 buf[i] = (char)c; 777 i++; 778 } 779 780 wc = 0; 781 782 if (c == -2) { 783 /* unterminated character constant */ 784 error(253); 785 } else if (c == 0) { 786 /* empty character constant */ 787 error(73); 788 } else { 789 if (i > MB_CUR_MAX) { 790 i = MB_CUR_MAX; 791 /* too many characters in character constant */ 792 error(71); 793 } else { 794 buf[i] = '\0'; 795 (void)mbtowc(NULL, NULL, 0); 796 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 797 /* invalid multibyte character */ 798 error(291); 799 } 800 } 801 802 yylval.y_val = xcalloc(1, sizeof (val_t)); 803 yylval.y_val->v_tspec = WCHAR; 804 yylval.y_val->v_quad = wc; 805 806 return (T_CON); 807 } 808 809 /* 810 * Read a character which is part of a character constant or of a string 811 * and handle escapes. 812 * 813 * The Argument is the character which delimits the character constant or 814 * string. 815 * 816 * Returns -1 if the end of the character constant or string is reached, 817 * -2 if the EOF is reached, and the charachter otherwise. 818 */ 819 static int 820 getescc(d) 821 int d; 822 { 823 static int pbc = -1; 824 int n, c, v; 825 826 if (pbc == -1) { 827 c = inpc(); 828 } else { 829 c = pbc; 830 pbc = -1; 831 } 832 if (c == d) 833 return (-1); 834 switch (c) { 835 case '\n': 836 /* newline in string or char constant */ 837 error(254); 838 return (-2); 839 case EOF: 840 return (-2); 841 case '\\': 842 switch (c = inpc()) { 843 case '"': 844 if (tflag && d == '\'') 845 /* \" inside character constant undef. ... */ 846 warning(262); 847 return ('"'); 848 case '\'': 849 return ('\''); 850 case '?': 851 if (tflag) 852 /* \? undefined in traditional C */ 853 warning(263); 854 return ('?'); 855 case '\\': 856 return ('\\'); 857 case 'a': 858 if (tflag) 859 /* \a undefined in traditional C */ 860 warning(81); 861 #ifdef __STDC__ 862 return ('\a'); 863 #else 864 return ('\007'); 865 #endif 866 case 'b': 867 return ('\b'); 868 case 'f': 869 return ('\f'); 870 case 'n': 871 return ('\n'); 872 case 'r': 873 return ('\r'); 874 case 't': 875 return ('\t'); 876 case 'v': 877 if (tflag) 878 /* \v undefined in traditional C */ 879 warning(264); 880 #ifdef __STDC__ 881 return ('\v'); 882 #else 883 return ('\013'); 884 #endif 885 case '8': case '9': 886 /* bad octal digit %c */ 887 warning(77, c); 888 /* FALLTHROUGH */ 889 case '0': case '1': case '2': case '3': 890 case '4': case '5': case '6': case '7': 891 n = 3; 892 v = 0; 893 do { 894 v = (v << 3) + (c - '0'); 895 c = inpc(); 896 } while (--n && isdigit(c) && (tflag || c <= '7')); 897 if (tflag && n > 0 && isdigit(c)) 898 /* bad octal digit %c */ 899 warning(77, c); 900 pbc = c; 901 if (v > UCHAR_MAX) { 902 /* character escape does not fit in char. */ 903 warning(76); 904 v &= CHAR_MASK; 905 } 906 return (v); 907 case 'x': 908 if (tflag) 909 /* \x undefined in traditional C */ 910 warning(82); 911 v = 0; 912 n = 0; 913 while ((c = inpc()) >= 0 && isxdigit(c)) { 914 c = isdigit(c) ? 915 c - '0' : toupper(c) - 'A' + 10; 916 v = (v << 4) + c; 917 if (n >= 0) { 918 if ((v & ~CHAR_MASK) != 0) { 919 /* overflow in hex escape */ 920 warning(75); 921 n = -1; 922 } else { 923 n++; 924 } 925 } 926 } 927 pbc = c; 928 if (n == 0) { 929 /* no hex digits follow \x */ 930 error(74); 931 } if (n == -1) { 932 v &= CHAR_MASK; 933 } 934 return (v); 935 case '\n': 936 return (getescc(d)); 937 case EOF: 938 return (-2); 939 default: 940 if (isprint(c)) { 941 /* dubious escape \%c */ 942 warning(79, c); 943 } else { 944 /* dubious escape \%o */ 945 warning(80, c); 946 } 947 } 948 } 949 return (c); 950 } 951 952 /* 953 * Called for preprocessor directives. Currently implemented are: 954 * # lineno 955 * # lineno "filename" 956 */ 957 static void 958 directive() 959 { 960 const char *cp, *fn; 961 char c, *eptr; 962 size_t fnl; 963 long ln; 964 static int first = 1; 965 966 /* Go to first non-whitespace after # */ 967 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 968 969 if (!isdigit((unsigned char)c)) { 970 error: 971 /* undefined or invalid # directive */ 972 warning(255); 973 return; 974 } 975 ln = strtol(--cp, &eptr, 10); 976 if (cp == eptr) 977 goto error; 978 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 979 goto error; 980 while ((c = *cp++) == ' ' || c == '\t') ; 981 if (c != '\0') { 982 if (c != '"') 983 goto error; 984 fn = cp; 985 while ((c = *cp) != '"' && c != '\0') 986 cp++; 987 if (c != '"') 988 goto error; 989 if ((fnl = cp++ - fn) > PATH_MAX) 990 goto error; 991 while ((c = *cp++) == ' ' || c == '\t') ; 992 #if 0 993 if (c != '\0') 994 warning("extra character(s) after directive"); 995 #endif 996 997 /* empty string means stdin */ 998 if (fnl == 0) { 999 fn = "{standard input}"; 1000 fnl = 16; /* strlen (fn) */ 1001 } 1002 curr_pos.p_file = fnnalloc(fn, fnl); 1003 /* 1004 * If this is the first directive, the name is the name 1005 * of the C source file as specified at the command line. 1006 * It is written to the output file. 1007 */ 1008 if (first) { 1009 csrc_pos.p_file = curr_pos.p_file; 1010 outsrc(curr_pos.p_file); 1011 first = 0; 1012 } 1013 } 1014 curr_pos.p_line = (int)ln - 1; 1015 curr_pos.p_uniq = 0; 1016 if (curr_pos.p_file == csrc_pos.p_file) { 1017 csrc_pos.p_line = (int)ln - 1; 1018 csrc_pos.p_uniq = 0; 1019 } 1020 } 1021 1022 /* 1023 * Handle lint comments. Following comments are currently understood: 1024 * ARGSUSEDn 1025 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1026 * FALLTHRU FALLTHROUGH 1027 * LINTLIBRARY 1028 * LINTED NOSTRICT 1029 * LONGLONG 1030 * NOTREACHED 1031 * PRINTFLIKEn 1032 * PROTOLIB 1033 * SCANFLIKEn 1034 * VARARGSn 1035 * If one of this comments is recognized, the arguments, if any, are 1036 * parsed and a function which handles this comment is called. 1037 */ 1038 static void 1039 comment() 1040 { 1041 int c, lc; 1042 static struct { 1043 const char *keywd; 1044 int arg; 1045 void (*func) __P((int)); 1046 } keywtab[] = { 1047 { "ARGSUSED", 1, argsused }, 1048 { "CONSTCOND", 0, constcond }, 1049 { "CONSTANTCOND", 0, constcond }, 1050 { "CONSTANTCONDITION", 0, constcond }, 1051 { "FALLTHRU", 0, fallthru }, 1052 { "FALLTHROUGH", 0, fallthru }, 1053 { "LINTLIBRARY", 0, lintlib }, 1054 { "LINTED", 0, linted }, 1055 { "LONGLONG", 0, longlong }, 1056 { "NOSTRICT", 0, linted }, 1057 { "NOTREACHED", 0, notreach }, 1058 { "PRINTFLIKE", 1, printflike }, 1059 { "PROTOLIB", 1, protolib }, 1060 { "SCANFLIKE", 1, scanflike }, 1061 { "VARARGS", 1, varargs }, 1062 }; 1063 char keywd[32]; 1064 char arg[32]; 1065 int l, i, a; 1066 int eoc; 1067 1068 eoc = 0; 1069 1070 /* Skip white spaces after the start of the comment */ 1071 while ((c = inpc()) != EOF && isspace(c)) ; 1072 1073 /* Read the potential keyword to keywd */ 1074 l = 0; 1075 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1076 keywd[l++] = (char)c; 1077 c = inpc(); 1078 } 1079 keywd[l] = '\0'; 1080 1081 /* look for the keyword */ 1082 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1083 if (strcmp(keywtab[i].keywd, keywd) == 0) 1084 break; 1085 } 1086 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1087 goto skip_rest; 1088 1089 /* skip white spaces after the keyword */ 1090 while (c != EOF && isspace(c)) 1091 c = inpc(); 1092 1093 /* read the argument, if the keyword accepts one and there is one */ 1094 l = 0; 1095 if (keywtab[i].arg) { 1096 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1097 arg[l++] = (char)c; 1098 c = inpc(); 1099 } 1100 } 1101 arg[l] = '\0'; 1102 a = l != 0 ? atoi(arg) : -1; 1103 1104 /* skip white spaces after the argument */ 1105 while (c != EOF && isspace(c)) 1106 c = inpc(); 1107 1108 if (c != '*' || (c = inpc()) != '/') { 1109 if (keywtab[i].func != linted) 1110 /* extra characters in lint comment */ 1111 warning(257); 1112 } else { 1113 /* 1114 * remember that we have already found the end of the 1115 * comment 1116 */ 1117 eoc = 1; 1118 } 1119 1120 if (keywtab[i].func != NULL) 1121 (*keywtab[i].func)(a); 1122 1123 skip_rest: 1124 while (!eoc) { 1125 lc = c; 1126 if ((c = inpc()) == EOF) { 1127 /* unterminated comment */ 1128 error(256); 1129 break; 1130 } 1131 if (lc == '*' && c == '/') 1132 eoc = 1; 1133 } 1134 } 1135 1136 /* 1137 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1138 * clrwflgs() is called after function definitions and global and 1139 * local declarations and definitions. It is also called between 1140 * the controlling expression and the body of control statements 1141 * (if, switch, for, while). 1142 */ 1143 void 1144 clrwflgs() 1145 { 1146 nowarn = 0; 1147 quadflg = 0; 1148 ccflg = 0; 1149 } 1150 1151 /* 1152 * Strings are stored in a dynamically alloceted buffer and passed 1153 * in yylval.y_xstrg to the parser. The parser or the routines called 1154 * by the parser are responsible for freeing this buffer. 1155 */ 1156 static int 1157 string() 1158 { 1159 u_char *s; 1160 int c; 1161 size_t len, max; 1162 strg_t *strg; 1163 1164 s = xmalloc(max = 64); 1165 1166 len = 0; 1167 while ((c = getescc('"')) >= 0) { 1168 /* +1 to reserve space for a trailing NUL character */ 1169 if (len + 1 == max) 1170 s = xrealloc(s, max *= 2); 1171 s[len++] = (char)c; 1172 } 1173 s[len] = '\0'; 1174 if (c == -2) 1175 /* unterminated string constant */ 1176 error(258); 1177 1178 strg = xcalloc(1, sizeof (strg_t)); 1179 strg->st_tspec = CHAR; 1180 strg->st_len = len; 1181 strg->st_cp = s; 1182 1183 yylval.y_strg = strg; 1184 return (T_STRING); 1185 } 1186 1187 static int 1188 wcstrg() 1189 { 1190 char *s; 1191 int c, i, n, wi; 1192 size_t len, max, wlen; 1193 wchar_t *ws; 1194 strg_t *strg; 1195 1196 s = xmalloc(max = 64); 1197 len = 0; 1198 while ((c = getescc('"')) >= 0) { 1199 /* +1 to save space for a trailing NUL character */ 1200 if (len + 1 >= max) 1201 s = xrealloc(s, max *= 2); 1202 s[len++] = (char)c; 1203 } 1204 s[len] = '\0'; 1205 if (c == -2) 1206 /* unterminated string constant */ 1207 error(258); 1208 1209 /* get length of wide character string */ 1210 (void)mblen(NULL, 0); 1211 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1212 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1213 /* invalid multibyte character */ 1214 error(291); 1215 break; 1216 } 1217 if (n == 0) 1218 n = 1; 1219 } 1220 1221 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1222 1223 /* convert from multibyte to wide char */ 1224 (void)mbtowc(NULL, NULL, 0); 1225 for (i = 0, wi = 0; i < len; i += n, wi++) { 1226 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1227 break; 1228 if (n == 0) 1229 n = 1; 1230 } 1231 ws[wi] = 0; 1232 free(s); 1233 1234 strg = xcalloc(1, sizeof (strg_t)); 1235 strg->st_tspec = WCHAR; 1236 strg->st_len = wlen; 1237 strg->st_wcp = ws; 1238 1239 yylval.y_strg = strg; 1240 return (T_STRING); 1241 } 1242 1243 /* 1244 * As noted above the scanner does not create new symbol table entries 1245 * for symbols it cannot find in the symbol table. This is to avoid 1246 * putting undeclared symbols into the symbol table if a syntax error 1247 * occurs. 1248 * 1249 * getsym() is called as soon as it is probably ok to put the symbol to 1250 * the symbol table. This does not mean that it is not possible that 1251 * symbols are put to the symbol table which are than not completely 1252 * declared due to syntax errors. To avoid too many problems in this 1253 * case symbols get type int in getsym(). 1254 * 1255 * XXX calls to getsym() should be delayed until decl1*() is called 1256 */ 1257 sym_t * 1258 getsym(sb) 1259 sbuf_t *sb; 1260 { 1261 dinfo_t *di; 1262 char *s; 1263 sym_t *sym; 1264 1265 sym = sb->sb_sym; 1266 1267 /* 1268 * During member declaration it is possible that name() looked 1269 * for symbols of type FVFT, although it should have looked for 1270 * symbols of type FTAG. Same can happen for labels. Both cases 1271 * are compensated here. 1272 */ 1273 if (symtyp == FMOS || symtyp == FLAB) { 1274 if (sym == NULL || sym->s_kind == FVFT) 1275 sym = search(sb); 1276 } 1277 1278 if (sym != NULL) { 1279 if (sym->s_kind != symtyp) 1280 lerror("storesym() 1"); 1281 symtyp = FVFT; 1282 freesb(sb); 1283 return (sym); 1284 } 1285 1286 /* create a new symbol table entry */ 1287 1288 /* labels must always be allocated at level 1 (outhermost block) */ 1289 if (symtyp == FLAB) { 1290 sym = getlblk(1, sizeof (sym_t)); 1291 s = getlblk(1, sb->sb_len + 1); 1292 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1293 sym->s_name = s; 1294 sym->s_blklev = 1; 1295 di = dcs; 1296 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1297 di = di->d_nxt; 1298 if (di->d_ctx != AUTO) 1299 lerror("storesym() 2"); 1300 } else { 1301 sym = getblk(sizeof (sym_t)); 1302 sym->s_name = sb->sb_name; 1303 sym->s_blklev = blklev; 1304 di = dcs; 1305 } 1306 1307 UNIQUE_CURR_POS(sym->s_dpos); 1308 if ((sym->s_kind = symtyp) != FLAB) 1309 sym->s_type = gettyp(INT); 1310 1311 symtyp = FVFT; 1312 1313 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1314 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1315 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1316 1317 *di->d_ldlsym = sym; 1318 di->d_ldlsym = &sym->s_dlnxt; 1319 1320 freesb(sb); 1321 return (sym); 1322 } 1323 1324 /* 1325 * Remove a symbol forever from the symbol table. s_blklev 1326 * is set to -1 to avoid that the symbol will later be put 1327 * back to the symbol table. 1328 */ 1329 void 1330 rmsym(sym) 1331 sym_t *sym; 1332 { 1333 if ((*sym->s_rlink = sym->s_link) != NULL) 1334 sym->s_link->s_rlink = sym->s_rlink; 1335 sym->s_blklev = -1; 1336 sym->s_link = NULL; 1337 } 1338 1339 /* 1340 * Remove a list of symbols declared at one level from the symbol 1341 * table. 1342 */ 1343 void 1344 rmsyms(syms) 1345 sym_t *syms; 1346 { 1347 sym_t *sym; 1348 1349 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1350 if (sym->s_blklev != -1) { 1351 if ((*sym->s_rlink = sym->s_link) != NULL) 1352 sym->s_link->s_rlink = sym->s_rlink; 1353 sym->s_link = NULL; 1354 sym->s_rlink = NULL; 1355 } 1356 } 1357 } 1358 1359 /* 1360 * Put a symbol into the symbol table 1361 */ 1362 void 1363 inssym(bl, sym) 1364 int bl; 1365 sym_t *sym; 1366 { 1367 int h; 1368 1369 h = hash(sym->s_name); 1370 if ((sym->s_link = symtab[h]) != NULL) 1371 symtab[h]->s_rlink = &sym->s_link; 1372 (symtab[h] = sym)->s_rlink = &symtab[h]; 1373 sym->s_blklev = bl; 1374 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1375 lerror("inssym()"); 1376 } 1377 1378 /* 1379 * Called at level 0 after syntax errors 1380 * Removes all symbols which are not declared at level 0 from the 1381 * symbol table. Also frees all memory which is not associated with 1382 * level 0. 1383 */ 1384 void 1385 cleanup() 1386 { 1387 sym_t *sym, *nsym; 1388 int i; 1389 1390 for (i = 0; i < HSHSIZ1; i++) { 1391 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1392 nsym = sym->s_link; 1393 if (sym->s_blklev >= 1) { 1394 if ((*sym->s_rlink = nsym) != NULL) 1395 nsym->s_rlink = sym->s_rlink; 1396 } 1397 } 1398 } 1399 1400 for (i = mblklev; i > 0; i--) 1401 freelblk(i); 1402 } 1403 1404 /* 1405 * Create a new symbol with the name of an existing symbol. 1406 */ 1407 sym_t * 1408 pushdown(sym) 1409 sym_t *sym; 1410 { 1411 int h; 1412 sym_t *nsym; 1413 1414 h = hash(sym->s_name); 1415 nsym = getblk(sizeof (sym_t)); 1416 if (sym->s_blklev > blklev) 1417 lerror("pushdown()"); 1418 nsym->s_name = sym->s_name; 1419 UNIQUE_CURR_POS(nsym->s_dpos); 1420 nsym->s_kind = sym->s_kind; 1421 nsym->s_blklev = blklev; 1422 1423 if ((nsym->s_link = symtab[h]) != NULL) 1424 symtab[h]->s_rlink = &nsym->s_link; 1425 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1426 1427 *dcs->d_ldlsym = nsym; 1428 dcs->d_ldlsym = &nsym->s_dlnxt; 1429 1430 return (nsym); 1431 } 1432 1433 /* 1434 * Free any dynamically allocated memory referenced by 1435 * the value stack or yylval. 1436 * The type of information in yylval is described by tok. 1437 */ 1438 void 1439 freeyyv(sp, tok) 1440 void *sp; 1441 int tok; 1442 { 1443 if (tok == T_NAME || tok == T_TYPENAME) { 1444 sbuf_t *sb = *(sbuf_t **)sp; 1445 freesb(sb); 1446 } else if (tok == T_CON) { 1447 val_t *val = *(val_t **)sp; 1448 free(val); 1449 } else if (tok == T_STRING) { 1450 strg_t *strg = *(strg_t **)sp; 1451 if (strg->st_tspec == CHAR) { 1452 free(strg->st_cp); 1453 } else if (strg->st_tspec == WCHAR) { 1454 free(strg->st_wcp); 1455 } else { 1456 lerror("fryylv() 1"); 1457 } 1458 free(strg); 1459 } 1460 } 1461
Indexes created Mon Oct 20 20:10:13 GMT 2025