1%{ 2/* $NetBSD: scan.l,v 1.26 2002/01/31 22:30:21 tv 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#if defined(__RCSID) && !defined(lint) 38__RCSID("$NetBSD: scan.l,v 1.26 2002/01/31 22:30:21 tv Exp $"); 39#endif 40__FBSDID("$FreeBSD$"); 41 42#include <stdlib.h> 43#include <string.h> 44#include <limits.h> 45#include <float.h> 46#include <ctype.h> 47#include <errno.h> 48#include <err.h> 49#include <math.h> 50 51#include "lint1.h" 52#include "cgram.h" 53 54#define CHAR_MASK (~(~0 << CHAR_BIT)) 55 56/* Current position (its also updated when an included file is parsed) */ 57pos_t curr_pos = { 1, "", 0 }; 58 59/* 60 * Current position in C source (not updated when an included file is 61 * parsed). 62 */ 63pos_t csrc_pos = { 1, "", 0 }; 64 65static void incline(void); 66static void badchar(int); 67static sbuf_t *allocsb(void); 68static void freesb(sbuf_t *); 69static int inpc(void); 70static int hash(const char *); 71static sym_t *search(sbuf_t *); 72static int name(void); 73static int keyw(sym_t *); 74static int icon(int); 75static int fcon(void); 76static int operator(int, op_t); 77static int ccon(void); 78static int wccon(void); 79static int getescc(int); 80static void directive(void); 81static void comment(void); 82static void slashslashcomment(void); 83static int string(void); 84static int wcstrg(void); 85 86%} 87 88%option nounput 89 90L [_A-Za-z] 91D [0-9] 92NZD [1-9] 93OD [0-7] 94HD [0-9A-Fa-f] 95EX ([eE][+-]?[0-9]+) 96 97%% 98 99{L}({L}|{D})* return (name()); 1000{OD}*[lLuU]* return (icon(8)); 101{NZD}{D}*[lLuU]* return (icon(10)); 1020[xX]{HD}+[lLuU]* return (icon(16)); 103{D}+\.{D}*{EX}?[fFlL]? | 104{D}+{EX}[fFlL]? | 105\.{D}+{EX}?[fFlL]? return (fcon()); 106"=" return (operator(T_ASSIGN, ASSIGN)); 107"*=" return (operator(T_OPASS, MULASS)); 108"/=" return (operator(T_OPASS, DIVASS)); 109"%=" return (operator(T_OPASS, MODASS)); 110"+=" return (operator(T_OPASS, ADDASS)); 111"-=" return (operator(T_OPASS, SUBASS)); 112"<<=" return (operator(T_OPASS, SHLASS)); 113">>=" return (operator(T_OPASS, SHRASS)); 114"&=" return (operator(T_OPASS, ANDASS)); 115"^=" return (operator(T_OPASS, XORASS)); 116"|=" return (operator(T_OPASS, ORASS)); 117"||" return (operator(T_LOGOR, LOGOR)); 118"&&" return (operator(T_LOGAND, LOGAND)); 119"|" return (operator(T_OR, OR)); 120"&" return (operator(T_AND, AND)); 121"^" return (operator(T_XOR, XOR)); 122"==" return (operator(T_EQOP, EQ)); 123"!=" return (operator(T_EQOP, NE)); 124"<" return (operator(T_RELOP, LT)); 125">" return (operator(T_RELOP, GT)); 126"<=" return (operator(T_RELOP, LE)); 127">=" return (operator(T_RELOP, GE)); 128"<<" return (operator(T_SHFTOP, SHL)); 129">>" return (operator(T_SHFTOP, SHR)); 130"++" return (operator(T_INCDEC, INC)); 131"--" return (operator(T_INCDEC, DEC)); 132"->" return (operator(T_STROP, ARROW)); 133"." return (operator(T_STROP, POINT)); 134"+" return (operator(T_ADDOP, PLUS)); 135"-" return (operator(T_ADDOP, MINUS)); 136"*" return (operator(T_MULT, MULT)); 137"/" return (operator(T_DIVOP, DIV)); 138"%" return (operator(T_DIVOP, MOD)); 139"!" return (operator(T_UNOP, NOT)); 140"~" return (operator(T_UNOP, COMPL)); 141"\"" return (string()); 142"L\"" return (wcstrg()); 143";" return (T_SEMI); 144"{" return (T_LBRACE); 145"}" return (T_RBRACE); 146"," return (T_COMMA); 147":" return (T_COLON); 148"?" return (T_QUEST); 149"[" return (T_LBRACK); 150"]" return (T_RBRACK); 151"(" return (T_LPARN); 152")" return (T_RPARN); 153"..." return (T_ELLIPSE); 154"'" return (ccon()); 155"L'" return (wccon()); 156^#.*$ directive(); 157\n incline(); 158\t|" "|\f|\v ; 159"/*" comment(); 160"//" slashslashcomment(); 161. badchar(yytext[0]); 162 163%% 164 165static void 166incline(void) 167{ 168 curr_pos.p_line++; 169 curr_pos.p_uniq = 0; 170 if (curr_pos.p_file == csrc_pos.p_file) { 171 csrc_pos.p_line++; 172 csrc_pos.p_uniq = 0; 173 } 174} 175 176static void 177badchar(int c) 178{ 179 180 /* unknown character \%o */ 181 error(250, c); 182} 183 184/* 185 * Keywords. 186 * During initialisation they are written to the symbol table. 187 */ 188static struct kwtab { 189 const char *kw_name; /* keyword */ 190 int kw_token; /* token returned by yylex() */ 191 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 192 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 193 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 194 u_int kw_stdc : 1; /* STDC keyword */ 195 u_int kw_gcc : 1; /* GCC keyword */ 196} kwtab[] = { 197 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 198 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 199 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 200 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 201 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 202 { "case", T_CASE, 0, 0, 0, 0, 0 }, 203 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 204 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 205 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 206 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 207 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 208 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 209 { "do", T_DO, 0, 0, 0, 0, 0 }, 210 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 211 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 212 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 213 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 214 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 215 { "for", T_FOR, 0, 0, 0, 0, 0 }, 216 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 217 { "if", T_IF, 0, 0, 0, 0, 0 }, 218 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 219 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 220 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 221 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 222 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 }, 223 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 224 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 225 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 226 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 227 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 228 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 229 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 230 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 231 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 232 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 233 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 234 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 235 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 236 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 237 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 238 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 239 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 240 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 241 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 242 { NULL, 0, 0, 0, 0, 0, 0 } 243}; 244 245/* Symbol table */ 246static sym_t *symtab[HSHSIZ1]; 247 248/* bit i of the entry with index i is set */ 249uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 250 251/* least significant i bits are set in the entry with index i */ 252uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 253 254/* least significant i bits are not set in the entry with index i */ 255uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 256 257/* free list for sbuf structures */ 258static sbuf_t *sbfrlst; 259 260/* Typ of next expected symbol */ 261symt_t symtyp; 262 263 264/* 265 * All keywords are written to the symbol table. This saves us looking 266 * in an extra table for each name we found. 267 */ 268void 269initscan(void) 270{ 271 struct kwtab *kw; 272 sym_t *sym; 273 int h, i; 274 uint64_t uq; 275 276 for (kw = kwtab; kw->kw_name != NULL; kw++) { 277 if (kw->kw_stdc && tflag) 278 continue; 279 if (kw->kw_gcc && !gflag) 280 continue; 281 sym = getblk(sizeof (sym_t)); 282 sym->s_name = kw->kw_name; 283 sym->s_keyw = 1; 284 sym->s_value.v_quad = kw->kw_token; 285 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 286 sym->s_tspec = kw->kw_tspec; 287 } else if (kw->kw_token == T_SCLASS) { 288 sym->s_scl = kw->kw_scl; 289 } else if (kw->kw_token == T_QUAL) { 290 sym->s_tqual = kw->kw_tqual; 291 } 292 h = hash(sym->s_name); 293 if ((sym->s_link = symtab[h]) != NULL) 294 symtab[h]->s_rlink = &sym->s_link; 295 (symtab[h] = sym)->s_rlink = &symtab[h]; 296 } 297 298 /* initialize bit-masks for quads */ 299 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 300 qbmasks[i] = (uint64_t)1 << i; 301 uq = ~(uint64_t)0 << i; 302 qumasks[i] = uq; 303 qlmasks[i] = ~uq; 304 } 305 qumasks[i] = 0; 306 qlmasks[i] = ~(uint64_t)0; 307} 308 309/* 310 * Get a free sbuf structure, if possible from the free list 311 */ 312static sbuf_t * 313allocsb(void) 314{ 315 sbuf_t *sb; 316 317 if ((sb = sbfrlst) != NULL) { 318 sbfrlst = sb->sb_nxt; 319 } else { 320 if ((sb = malloc(sizeof (sbuf_t))) == NULL) 321 nomem(); 322 } 323 (void)memset(sb, 0, sizeof (*sb)); 324 return (sb); 325} 326 327/* 328 * Put a sbuf structure to the free list 329 */ 330static void 331freesb(sbuf_t *sb) 332{ 333 334 sb->sb_nxt = sbfrlst; 335 sbfrlst = sb; 336} 337 338/* 339 * Read a character and ensure that it is positive (except EOF). 340 * Increment line count(s) if necessary. 341 */ 342static int 343inpc(void) 344{ 345 int c; 346 347 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 348 incline(); 349 return (c); 350} 351 352static int 353hash(const char *s) 354{ 355 u_int v; 356 const u_char *us; 357 358 v = 0; 359 for (us = (const u_char *)s; *us != '\0'; us++) { 360 v = (v << sizeof (v)) + *us; 361 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 362 } 363 return (v % HSHSIZ1); 364} 365 366/* 367 * Lex has found a letter followed by zero or more letters or digits. 368 * It looks for a symbol in the symbol table with the same name. This 369 * symbol must either be a keyword or a symbol of the type required by 370 * symtyp (label, member, tag, ...). 371 * 372 * If it is a keyword, the token is returned. In some cases it is described 373 * more deeply by data written to yylval. 374 * 375 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 376 * is stored in yylval. This struct contains the name of the symbol, it's 377 * length and hash value. If there is already a symbol of the same name 378 * and type in the symbol table, the sbuf struct also contains a pointer 379 * to the symbol table entry. 380 */ 381static int 382name(void) 383{ 384 char *s; 385 sbuf_t *sb; 386 sym_t *sym; 387 int tok; 388 389 sb = allocsb(); 390 sb->sb_name = yytext; 391 sb->sb_len = yyleng; 392 sb->sb_hash = hash(yytext); 393 394 if ((sym = search(sb)) != NULL && sym->s_keyw) { 395 freesb(sb); 396 return (keyw(sym)); 397 } 398 399 sb->sb_sym = sym; 400 401 if (sym != NULL) { 402 if (blklev < sym->s_blklev) 403 lerror("name() 1"); 404 sb->sb_name = sym->s_name; 405 sb->sb_len = strlen(sym->s_name); 406 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 407 } else { 408 s = getblk(yyleng + 1); 409 (void)memcpy(s, yytext, yyleng + 1); 410 sb->sb_name = s; 411 sb->sb_len = yyleng; 412 tok = T_NAME; 413 } 414 415 yylval.y_sb = sb; 416 return (tok); 417} 418 419static sym_t * 420search(sbuf_t *sb) 421{ 422 sym_t *sym; 423 424 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 425 if (strcmp(sym->s_name, sb->sb_name) == 0) { 426 if (sym->s_keyw || sym->s_kind == symtyp) 427 return (sym); 428 } 429 } 430 431 return (NULL); 432} 433 434static int 435keyw(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 */ 453static int 454icon(int base) 455{ 456 int l_suffix, u_suffix; 457 int len; 458 const char *cp; 459 char c, *eptr; 460 tspec_t typ; 461 u_long ul = 0; 462 uint64_t uq = 0; 463 int ansiu; 464 static tspec_t contypes[2][3] = { 465 { INT, LONG, QUAD }, 466 { UINT, ULONG, UQUAD } 467 }; 468 469 cp = yytext; 470 len = yyleng; 471 472 /* skip 0x */ 473 if (base == 16) { 474 cp += 2; 475 len -= 2; 476 } 477 478 /* read suffixes */ 479 l_suffix = u_suffix = 0; 480 for ( ; ; ) { 481 if ((c = cp[len - 1]) == 'l' || c == 'L') { 482 l_suffix++; 483 } else if (c == 'u' || c == 'U') { 484 u_suffix++; 485 } else { 486 break; 487 } 488 len--; 489 } 490 if (l_suffix > 2 || u_suffix > 1) { 491 /* malformed integer constant */ 492 warning(251); 493 if (l_suffix > 2) 494 l_suffix = 2; 495 if (u_suffix > 1) 496 u_suffix = 1; 497 } 498 if (tflag && u_suffix != 0) { 499 /* suffix U is illegal in traditional C */ 500 warning(97); 501 } 502 typ = contypes[u_suffix][l_suffix]; 503 504 errno = 0; 505 if (l_suffix < 2) { 506 ul = strtoul(cp, &eptr, base); 507 } else { 508 uq = strtouq(cp, &eptr, base); 509 } 510 if (eptr != cp + len) 511 lerror("icon() 1"); 512 if (errno != 0) 513 /* integer constant out of range */ 514 warning(252); 515 516 /* 517 * If the value is to big for the current type, we must choose 518 * another type. 519 */ 520 ansiu = 0; 521 switch (typ) { 522 case INT: 523 if (ul <= INT_MAX) { 524 /* ok */ 525 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 526 typ = UINT; 527 } else if (ul <= LONG_MAX) { 528 typ = LONG; 529 } else { 530 typ = ULONG; 531 } 532 if (typ == UINT || typ == ULONG) { 533 if (tflag) { 534 typ = LONG; 535 } else if (!sflag) { 536 /* 537 * Remember that the constant is unsigned 538 * only in ANSI C 539 */ 540 ansiu = 1; 541 } 542 } 543 break; 544 case UINT: 545 if (ul > (u_int)UINT_MAX) 546 typ = ULONG; 547 break; 548 case LONG: 549 if (ul > LONG_MAX && !tflag) { 550 typ = ULONG; 551 if (!sflag) 552 ansiu = 1; 553 } 554 break; 555 case QUAD: 556 if (uq > QUAD_MAX && !tflag) { 557 typ = UQUAD; 558 if (!sflag) 559 ansiu = 1; 560 } 561 break; 562 /* LINTED (enumeration values not handled in switch) */ 563 case STRUCT: 564 case VOID: 565 case LDOUBLE: 566 case FUNC: 567 case ARRAY: 568 case PTR: 569 case ENUM: 570 case UNION: 571 case SIGNED: 572 case NOTSPEC: 573 case DOUBLE: 574 case FLOAT: 575 case UQUAD: 576 case ULONG: 577 case USHORT: 578 case SHORT: 579 case UCHAR: 580 case SCHAR: 581 case CHAR: 582 case UNSIGN: 583 break; 584 } 585 586 if (typ != QUAD && typ != UQUAD) { 587 if (isutyp(typ)) { 588 uq = ul; 589 } else { 590 uq = (int64_t)(long)ul; 591 } 592 } 593 594 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 595 596 if ((yylval.y_val = calloc(1, sizeof(val_t))) == NULL) 597 nomem(); 598 yylval.y_val->v_tspec = typ; 599 yylval.y_val->v_ansiu = ansiu; 600 yylval.y_val->v_quad = (int64_t)uq; 601 602 return (T_CON); 603} 604 605/* 606 * Returns 1 if t is a signed type and the value is negative. 607 * 608 * len is the number of significant bits. If len is -1, len is set 609 * to the width of type t. 610 */ 611int 612sign(int64_t q, tspec_t t, int len) 613{ 614 615 if (t == PTR || isutyp(t)) 616 return (0); 617 return (msb(q, t, len)); 618} 619 620int 621msb(int64_t q, tspec_t t, int len) 622{ 623 624 if (len <= 0) 625 len = size(t); 626 return ((q & qbmasks[len - 1]) != 0); 627} 628 629/* 630 * Extends the sign of q. 631 */ 632int64_t 633xsign(int64_t q, tspec_t t, int len) 634{ 635 636 if (len <= 0) 637 len = size(t); 638 639 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 640 q &= qlmasks[len]; 641 } else { 642 q |= qumasks[len]; 643 } 644 return (q); 645} 646 647/* 648 * Convert a string representing a floating point value into its interal 649 * representation. Type and value are returned in yylval. fcon() 650 * (and yylex()) returns T_CON. 651 * XXX Currently it is not possible to convert constants of type 652 * long double which are greater than DBL_MAX. 653 */ 654static int 655fcon(void) 656{ 657 const char *cp; 658 int len; 659 tspec_t typ; 660 char c, *eptr; 661 double d; 662 float f = 0; 663 664 cp = yytext; 665 len = yyleng; 666 667 if ((c = cp[len - 1]) == 'f' || c == 'F') { 668 typ = FLOAT; 669 len--; 670 } else if (c == 'l' || c == 'L') { 671 typ = LDOUBLE; 672 len--; 673 } else { 674 typ = DOUBLE; 675 } 676 677 if (tflag && typ != DOUBLE) { 678 /* suffixes F and L are illegal in traditional C */ 679 warning(98); 680 } 681 682 errno = 0; 683 d = strtod(cp, &eptr); 684 if (eptr != cp + len) 685 lerror("fcon() 1"); 686 if (errno != 0) 687 /* floating-point constant out of range */ 688 warning(248); 689 690 if (typ == FLOAT) { 691 f = (float)d; 692 if (!finite(f)) { 693 /* floating-point constant out of range */ 694 warning(248); 695 f = f > 0 ? FLT_MAX : -FLT_MAX; 696 } 697 } 698 699 if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL) 700 nomem(); 701 yylval.y_val->v_tspec = typ; 702 if (typ == FLOAT) { 703 yylval.y_val->v_ldbl = f; 704 } else { 705 yylval.y_val->v_ldbl = d; 706 } 707 708 return (T_CON); 709} 710 711static int 712operator(int t, op_t o) 713{ 714 715 yylval.y_op = o; 716 return (t); 717} 718 719/* 720 * Called if lex found a leading \'. 721 */ 722static int 723ccon(void) 724{ 725 int n, val, c; 726 char cv; 727 728 n = 0; 729 val = 0; 730 while ((c = getescc('\'')) >= 0) { 731 val = (val << CHAR_BIT) + c; 732 n++; 733 } 734 if (c == -2) { 735 /* unterminated character constant */ 736 error(253); 737 } else { 738 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 739 /* too many characters in character constant */ 740 error(71); 741 } else if (n > 1) { 742 /* multi-character character constant */ 743 warning(294); 744 } else if (n == 0) { 745 /* empty character constant */ 746 error(73); 747 } 748 } 749 if (n == 1) { 750 cv = (char)val; 751 val = cv; 752 } 753 754 yylval.y_val = xcalloc(1, sizeof (val_t)); 755 yylval.y_val->v_tspec = INT; 756 yylval.y_val->v_quad = val; 757 758 return (T_CON); 759} 760 761/* 762 * Called if lex found a leading L\' 763 */ 764static int 765wccon(void) 766{ 767 static char buf[MB_LEN_MAX + 1]; 768 int i, c; 769 wchar_t wc; 770 771 i = 0; 772 while ((c = getescc('\'')) >= 0) { 773 if (i < MB_CUR_MAX) 774 buf[i] = (char)c; 775 i++; 776 } 777 778 wc = 0; 779 780 if (c == -2) { 781 /* unterminated character constant */ 782 error(253); 783 } else if (c == 0) { 784 /* empty character constant */ 785 error(73); 786 } else { 787 if (i > MB_CUR_MAX) { 788 i = MB_CUR_MAX; 789 /* too many characters in character constant */ 790 error(71); 791 } else { 792 buf[i] = '\0'; 793 (void)mbtowc(NULL, NULL, 0); 794 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 795 /* invalid multibyte character */ 796 error(291); 797 } 798 } 799 800 if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL) 801 nomem(); 802 yylval.y_val->v_tspec = WCHAR; 803 yylval.y_val->v_quad = wc; 804 805 return (T_CON); 806} 807 808/* 809 * Read a character which is part of a character constant or of a string 810 * and handle escapes. 811 * 812 * The Argument is the character which delimits the character constant or 813 * string. 814 * 815 * Returns -1 if the end of the character constant or string is reached, 816 * -2 if the EOF is reached, and the character otherwise. 817 */ 818static int 819getescc(int d) 820{ 821 static int pbc = -1; 822 int n, c, v; 823 824 if (pbc == -1) { 825 c = inpc(); 826 } else { 827 c = pbc; 828 pbc = -1; 829 } 830 if (c == d) 831 return (-1); 832 switch (c) { 833 case '\n': 834 if (tflag) { 835 /* newline in string or char constant */ 836 error(254); 837 return (-2); 838 } 839 return (c); 840 case EOF: 841 return (-2); 842 case '\\': 843 switch (c = inpc()) { 844 case '"': 845 if (tflag && d == '\'') 846 /* \" inside character constant undef. ... */ 847 warning(262); 848 return ('"'); 849 case '\'': 850 return ('\''); 851 case '?': 852 if (tflag) 853 /* \? undefined in traditional C */ 854 warning(263); 855 return ('?'); 856 case '\\': 857 return ('\\'); 858 case 'a': 859 if (tflag) 860 /* \a undefined in traditional C */ 861 warning(81); 862 return ('\a'); 863 case 'b': 864 return ('\b'); 865 case 'f': 866 return ('\f'); 867 case 'n': 868 return ('\n'); 869 case 'r': 870 return ('\r'); 871 case 't': 872 return ('\t'); 873 case 'v': 874 if (tflag) 875 /* \v undefined in traditional C */ 876 warning(264); 877 return ('\v'); 878 case '8': case '9': 879 /* bad octal digit %c */ 880 warning(77, c); 881 /* FALLTHROUGH */ 882 case '0': case '1': case '2': case '3': 883 case '4': case '5': case '6': case '7': 884 n = 3; 885 v = 0; 886 do { 887 v = (v << 3) + (c - '0'); 888 c = inpc(); 889 } while (--n && isdigit(c) && (tflag || c <= '7')); 890 if (tflag && n > 0 && isdigit(c)) 891 /* bad octal digit %c */ 892 warning(77, c); 893 pbc = c; 894 if (v > UCHAR_MAX) { 895 /* character escape does not fit in char. */ 896 warning(76); 897 v &= CHAR_MASK; 898 } 899 return (v); 900 case 'x': 901 if (tflag) 902 /* \x undefined in traditional C */ 903 warning(82); 904 v = 0; 905 n = 0; 906 while ((c = inpc()) >= 0 && isxdigit(c)) { 907 c = isdigit(c) ? 908 c - '0' : toupper(c) - 'A' + 10; 909 v = (v << 4) + c; 910 if (n >= 0) { 911 if ((v & ~CHAR_MASK) != 0) { 912 /* overflow in hex escape */ 913 warning(75); 914 n = -1; 915 } else { 916 n++; 917 } 918 } 919 } 920 pbc = c; 921 if (n == 0) { 922 /* no hex digits follow \x */ 923 error(74); 924 } if (n == -1) { 925 v &= CHAR_MASK; 926 } 927 return (v); 928 case '\n': 929 return (getescc(d)); 930 case EOF: 931 return (-2); 932 default: 933 if (isprint(c)) { 934 /* dubious escape \%c */ 935 warning(79, c); 936 } else { 937 /* dubious escape \%o */ 938 warning(80, c); 939 } 940 } 941 } 942 return (c); 943} 944 945/* 946 * Called for preprocessor directives. Currently implemented are: 947 * # lineno 948 * # lineno "filename" 949 */ 950static void 951directive(void) 952{ 953 const char *cp, *fn; 954 char c, *eptr; 955 size_t fnl; 956 long ln; 957 static int first = 1; 958 959 /* Go to first non-whitespace after # */ 960 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 961 continue; 962 963 if (!isdigit((unsigned char)c)) { 964 error: 965 /* undefined or invalid # directive */ 966 warning(255); 967 return; 968 } 969 ln = strtol(--cp, &eptr, 10); 970 if (cp == eptr) 971 goto error; 972 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 973 goto error; 974 while ((c = *cp++) == ' ' || c == '\t') 975 continue; 976 if (c != '\0') { 977 if (c != '"') 978 goto error; 979 fn = cp; 980 while ((c = *cp) != '"' && c != '\0') 981 cp++; 982 if (c != '"') 983 goto error; 984 if ((fnl = cp++ - fn) > PATH_MAX) 985 goto error; 986 while ((c = *cp++) == ' ' || c == '\t') 987 continue; 988#if 0 989 if (c != '\0') 990 warning("extra character(s) after directive"); 991#endif 992 993 /* empty string means stdin */ 994 if (fnl == 0) { 995 fn = "{standard input}"; 996 fnl = 16; /* strlen (fn) */ 997 } 998 curr_pos.p_file = fnnalloc(fn, fnl); 999 /* 1000 * If this is the first directive, the name is the name 1001 * of the C source file as specified at the command line. 1002 * It is written to the output file. 1003 */ 1004 if (first) { 1005 csrc_pos.p_file = curr_pos.p_file; 1006 outsrc(curr_pos.p_file); 1007 first = 0; 1008 } 1009 } 1010 curr_pos.p_line = (int)ln - 1; 1011 curr_pos.p_uniq = 0; 1012 if (curr_pos.p_file == csrc_pos.p_file) { 1013 csrc_pos.p_line = (int)ln - 1; 1014 csrc_pos.p_uniq = 0; 1015 } 1016} 1017 1018/* 1019 * Handle lint comments. Following comments are currently understood: 1020 * ARGSUSEDn 1021 * BITFIELDTYPE 1022 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1023 * FALLTHRU FALLTHROUGH 1024 * LINTLIBRARY 1025 * LINTED NOSTRICT 1026 * LONGLONG 1027 * NOTREACHED 1028 * PRINTFLIKEn 1029 * PROTOLIB 1030 * SCANFLIKEn 1031 * VARARGSn 1032 * If one of this comments is recognized, the arguments, if any, are 1033 * parsed and a function which handles this comment is called. 1034 */ 1035static void 1036comment(void) 1037{ 1038 int c, lc; 1039 static struct { 1040 const char *keywd; 1041 int arg; 1042 void (*func)(int); 1043 } keywtab[] = { 1044 { "ARGSUSED", 1, argsused }, 1045 { "BITFIELDTYPE", 0, bitfieldtype }, 1046 { "CONSTCOND", 0, constcond }, 1047 { "CONSTANTCOND", 0, constcond }, 1048 { "CONSTANTCONDITION", 0, constcond }, 1049 { "FALLTHRU", 0, fallthru }, 1050 { "FALLTHROUGH", 0, fallthru }, 1051 { "LINTLIBRARY", 0, lintlib }, 1052 { "LINTED", 0, linted }, 1053 { "LONGLONG", 0, longlong }, 1054 { "NOSTRICT", 0, linted }, 1055 { "NOTREACHED", 0, notreach }, 1056 { "PRINTFLIKE", 1, printflike }, 1057 { "PROTOLIB", 1, protolib }, 1058 { "SCANFLIKE", 1, scanflike }, 1059 { "VARARGS", 1, varargs }, 1060 }; 1061 char keywd[32]; 1062 char arg[32]; 1063 int l, i, a; 1064 int eoc; 1065 1066 eoc = 0; 1067 1068 /* Skip white spaces after the start of the comment */ 1069 while ((c = inpc()) != EOF && isspace(c)) 1070 continue; 1071 1072 /* Read the potential keyword to keywd */ 1073 l = 0; 1074 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1075 keywd[l++] = (char)c; 1076 c = inpc(); 1077 } 1078 keywd[l] = '\0'; 1079 1080 /* look for the keyword */ 1081 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1082 if (strcmp(keywtab[i].keywd, keywd) == 0) 1083 break; 1084 } 1085 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1086 goto skip_rest; 1087 1088 /* skip white spaces after the keyword */ 1089 while (c != EOF && isspace(c)) 1090 c = inpc(); 1091 1092 /* read the argument, if the keyword accepts one and there is one */ 1093 l = 0; 1094 if (keywtab[i].arg) { 1095 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1096 arg[l++] = (char)c; 1097 c = inpc(); 1098 } 1099 } 1100 arg[l] = '\0'; 1101 a = l != 0 ? atoi(arg) : -1; 1102 1103 /* skip white spaces after the argument */ 1104 while (c != EOF && isspace(c)) 1105 c = inpc(); 1106 1107 if (c != '*' || (c = inpc()) != '/') { 1108 if (keywtab[i].func != linted) 1109 /* extra characters in lint comment */ 1110 warning(257); 1111 } else { 1112 /* 1113 * remember that we have already found the end of the 1114 * comment 1115 */ 1116 eoc = 1; 1117 } 1118 1119 if (keywtab[i].func != NULL) 1120 (*keywtab[i].func)(a); 1121 1122 skip_rest: 1123 while (!eoc) { 1124 lc = c; 1125 if ((c = inpc()) == EOF) { 1126 /* unterminated comment */ 1127 error(256); 1128 break; 1129 } 1130 if (lc == '*' && c == '/') 1131 eoc = 1; 1132 } 1133} 1134 1135/* 1136 * Handle // style comments 1137 */ 1138static void 1139slashslashcomment(void) 1140{ 1141 int c; 1142 1143 if (sflag < 2 && !gflag) 1144 /* // comments only supported in C99 */ 1145 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1146 1147 while ((c = inpc()) != EOF && c != '\n') 1148 continue; 1149} 1150 1151/* 1152 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1153 * clrwflgs() is called after function definitions and global and 1154 * local declarations and definitions. It is also called between 1155 * the controlling expression and the body of control statements 1156 * (if, switch, for, while). 1157 */ 1158void 1159clrwflgs(void) 1160{ 1161 1162 nowarn = 0; 1163 quadflg = 0; 1164 ccflg = 0; 1165} 1166 1167/* 1168 * Strings are stored in a dynamically alloceted buffer and passed 1169 * in yylval.y_xstrg to the parser. The parser or the routines called 1170 * by the parser are responsible for freeing this buffer. 1171 */ 1172static int 1173string(void) 1174{ 1175 u_char *s; 1176 int c; 1177 size_t len, max; 1178 strg_t *strg; 1179 1180 if ((s = malloc(max = 64)) == NULL) 1181 nomem(); 1182 1183 len = 0; 1184 while ((c = getescc('"')) >= 0) { 1185 /* +1 to reserve space for a trailing NUL character */ 1186 if (len + 1 == max) 1187 if ((s = realloc(s, max *= 2)) == NULL) 1188 nomem(); 1189 s[len++] = (char)c; 1190 } 1191 s[len] = '\0'; 1192 if (c == -2) 1193 /* unterminated string constant */ 1194 error(258); 1195 1196 if ((strg = calloc(1, sizeof (strg_t))) == NULL) 1197 nomem(); 1198 strg->st_tspec = CHAR; 1199 strg->st_len = len; 1200 strg->st_cp = s; 1201 1202 yylval.y_strg = strg; 1203 return (T_STRING); 1204} 1205 1206static int 1207wcstrg(void) 1208{ 1209 char *s; 1210 int c, i, n, wi; 1211 size_t len, max, wlen; 1212 wchar_t *ws; 1213 strg_t *strg; 1214 1215 if ((s = malloc(max = 64)) == NULL) 1216 nomem(); 1217 len = 0; 1218 while ((c = getescc('"')) >= 0) { 1219 /* +1 to save space for a trailing NUL character */ 1220 if (len + 1 >= max) 1221 if ((s = realloc(s, max *= 2)) == NULL) 1222 nomem(); 1223 s[len++] = (char)c; 1224 } 1225 s[len] = '\0'; 1226 if (c == -2) 1227 /* unterminated string constant */ 1228 error(258); 1229 1230 /* get length of wide character string */ 1231 (void)mblen(NULL, 0); 1232 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1233 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1234 /* invalid multibyte character */ 1235 error(291); 1236 break; 1237 } 1238 if (n == 0) 1239 n = 1; 1240 } 1241 1242 if ((ws = malloc((wlen + 1) * sizeof (wchar_t))) == NULL) 1243 nomem(); 1244 1245 /* convert from multibyte to wide char */ 1246 (void)mbtowc(NULL, NULL, 0); 1247 for (i = 0, wi = 0; i < len; i += n, wi++) { 1248 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1249 break; 1250 if (n == 0) 1251 n = 1; 1252 } 1253 ws[wi] = 0; 1254 free(s); 1255 1256 if ((strg = calloc(1, sizeof (strg_t))) == NULL) 1257 nomem(); 1258 strg->st_tspec = WCHAR; 1259 strg->st_len = wlen; 1260 strg->st_wcp = ws; 1261 1262 yylval.y_strg = strg; 1263 return (T_STRING); 1264} 1265 1266/* 1267 * As noted above the scanner does not create new symbol table entries 1268 * for symbols it cannot find in the symbol table. This is to avoid 1269 * putting undeclared symbols into the symbol table if a syntax error 1270 * occurs. 1271 * 1272 * getsym() is called as soon as it is probably ok to put the symbol to 1273 * the symbol table. This does not mean that it is not possible that 1274 * symbols are put to the symbol table which are than not completely 1275 * declared due to syntax errors. To avoid too many problems in this 1276 * case symbols get type int in getsym(). 1277 * 1278 * XXX calls to getsym() should be delayed until decl1*() is called 1279 */ 1280sym_t * 1281getsym(sbuf_t *sb) 1282{ 1283 dinfo_t *di; 1284 char *s; 1285 sym_t *sym; 1286 1287 sym = sb->sb_sym; 1288 1289 /* 1290 * During member declaration it is possible that name() looked 1291 * for symbols of type FVFT, although it should have looked for 1292 * symbols of type FTAG. Same can happen for labels. Both cases 1293 * are compensated here. 1294 */ 1295 if (symtyp == FMOS || symtyp == FLAB) { 1296 if (sym == NULL || sym->s_kind == FVFT) 1297 sym = search(sb); 1298 } 1299 1300 if (sym != NULL) { 1301 if (sym->s_kind != symtyp) 1302 lerror("storesym() 1"); 1303 symtyp = FVFT; 1304 freesb(sb); 1305 return (sym); 1306 } 1307 1308 /* create a new symbol table entry */ 1309 1310 /* labels must always be allocated at level 1 (outhermost block) */ 1311 if (symtyp == FLAB) { 1312 sym = getlblk(1, sizeof (sym_t)); 1313 s = getlblk(1, sb->sb_len + 1); 1314 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1315 sym->s_name = s; 1316 sym->s_blklev = 1; 1317 di = dcs; 1318 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1319 di = di->d_nxt; 1320 if (di->d_ctx != AUTO) 1321 lerror("storesym() 2"); 1322 } else { 1323 sym = getblk(sizeof (sym_t)); 1324 sym->s_name = sb->sb_name; 1325 sym->s_blklev = blklev; 1326 di = dcs; 1327 } 1328 1329 UNIQUE_CURR_POS(sym->s_dpos); 1330 if ((sym->s_kind = symtyp) != FLAB) 1331 sym->s_type = gettyp(INT); 1332 1333 symtyp = FVFT; 1334 1335 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1336 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1337 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1338 1339 *di->d_ldlsym = sym; 1340 di->d_ldlsym = &sym->s_dlnxt; 1341 1342 freesb(sb); 1343 return (sym); 1344} 1345 1346/* 1347 * Remove a symbol forever from the symbol table. s_blklev 1348 * is set to -1 to avoid that the symbol will later be put 1349 * back to the symbol table. 1350 */ 1351void 1352rmsym(sym_t *sym) 1353{ 1354 1355 if ((*sym->s_rlink = sym->s_link) != NULL) 1356 sym->s_link->s_rlink = sym->s_rlink; 1357 sym->s_blklev = -1; 1358 sym->s_link = NULL; 1359} 1360 1361/* 1362 * Remove a list of symbols declared at one level from the symbol 1363 * table. 1364 */ 1365void 1366rmsyms(sym_t *syms) 1367{ 1368 sym_t *sym; 1369 1370 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1371 if (sym->s_blklev != -1) { 1372 if ((*sym->s_rlink = sym->s_link) != NULL) 1373 sym->s_link->s_rlink = sym->s_rlink; 1374 sym->s_link = NULL; 1375 sym->s_rlink = NULL; 1376 } 1377 } 1378} 1379 1380/* 1381 * Put a symbol into the symbol table 1382 */ 1383void 1384inssym(int bl, sym_t *sym) 1385{ 1386 int h; 1387 1388 h = hash(sym->s_name); 1389 if ((sym->s_link = symtab[h]) != NULL) 1390 symtab[h]->s_rlink = &sym->s_link; 1391 (symtab[h] = sym)->s_rlink = &symtab[h]; 1392 sym->s_blklev = bl; 1393 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1394 lerror("inssym()"); 1395} 1396 1397/* 1398 * Called at level 0 after syntax errors 1399 * Removes all symbols which are not declared at level 0 from the 1400 * symbol table. Also frees all memory which is not associated with 1401 * level 0. 1402 */ 1403void 1404cleanup(void) 1405{ 1406 sym_t *sym, *nsym; 1407 int i; 1408 1409 for (i = 0; i < HSHSIZ1; i++) { 1410 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1411 nsym = sym->s_link; 1412 if (sym->s_blklev >= 1) { 1413 if ((*sym->s_rlink = nsym) != NULL) 1414 nsym->s_rlink = sym->s_rlink; 1415 } 1416 } 1417 } 1418 1419 for (i = mblklev; i > 0; i--) 1420 freelblk(i); 1421} 1422 1423/* 1424 * Create a new symbol with the name of an existing symbol. 1425 */ 1426sym_t * 1427pushdown(sym_t *sym) 1428{ 1429 int h; 1430 sym_t *nsym; 1431 1432 h = hash(sym->s_name); 1433 nsym = getblk(sizeof (sym_t)); 1434 if (sym->s_blklev > blklev) 1435 lerror("pushdown()"); 1436 nsym->s_name = sym->s_name; 1437 UNIQUE_CURR_POS(nsym->s_dpos); 1438 nsym->s_kind = sym->s_kind; 1439 nsym->s_blklev = blklev; 1440 1441 if ((nsym->s_link = symtab[h]) != NULL) 1442 symtab[h]->s_rlink = &nsym->s_link; 1443 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1444 1445 *dcs->d_ldlsym = nsym; 1446 dcs->d_ldlsym = &nsym->s_dlnxt; 1447 1448 return (nsym); 1449} 1450 1451/* 1452 * Free any dynamically allocated memory referenced by 1453 * the value stack or yylval. 1454 * The type of information in yylval is described by tok. 1455 */ 1456void 1457freeyyv(void *sp, int tok) 1458{ 1459 if (tok == T_NAME || tok == T_TYPENAME) { 1460 sbuf_t *sb = *(sbuf_t **)sp; 1461 freesb(sb); 1462 } else if (tok == T_CON) { 1463 val_t *val = *(val_t **)sp; 1464 free(val); 1465 } else if (tok == T_STRING) { 1466 strg_t *strg = *(strg_t **)sp; 1467 if (strg->st_tspec == CHAR) { 1468 free(strg->st_cp); 1469 } else if (strg->st_tspec == WCHAR) { 1470 free(strg->st_wcp); 1471 } else { 1472 lerror("fryylv() 1"); 1473 } 1474 free(strg); 1475 } 1476} 1477