lex.c revision 1.73
1 /* $NetBSD: lex.c,v 1.73 2021/08/28 13:29:26 rillig Exp $ */ 2 3 /* 4 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved. 5 * Copyright (c) 1994, 1995 Jochen Pohl 6 * All Rights Reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by Jochen Pohl for 19 * The NetBSD Project. 20 * 4. The name of the author may not be used to endorse or promote products 21 * derived from this software without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 24 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 25 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 26 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 27 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 28 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 32 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35 #if HAVE_NBTOOL_CONFIG_H 36 #include "nbtool_config.h" 37 #endif 38 39 #include <sys/cdefs.h> 40 #if defined(__RCSID) && !defined(lint) 41 __RCSID("$NetBSD: lex.c,v 1.73 2021/08/28 13:29:26 rillig Exp $"); 42 #endif 43 44 #include <ctype.h> 45 #include <errno.h> 46 #include <float.h> 47 #include <limits.h> 48 #include <math.h> 49 #include <stdlib.h> 50 #include <string.h> 51 52 #include "lint1.h" 53 #include "cgram.h" 54 55 #define CHAR_MASK ((1U << CHAR_SIZE) - 1) 56 57 58 /* Current position (it's also updated when an included file is parsed) */ 59 pos_t curr_pos = { "", 1, 0 }; 60 61 /* 62 * Current position in C source (not updated when an included file is 63 * parsed). 64 */ 65 pos_t csrc_pos = { "", 1, 0 }; 66 67 /* Are we parsing a gcc attribute? */ 68 bool attron; 69 70 bool in_system_header = false; 71 72 static sbuf_t *allocsb(void); 73 static void freesb(sbuf_t *); 74 static int inpc(void); 75 static unsigned int hash(const char *); 76 static sym_t * search(sbuf_t *); 77 static int keyw(sym_t *); 78 static int get_escaped_char(int); 79 80 void 81 lex_next_line(void) 82 { 83 curr_pos.p_line++; 84 curr_pos.p_uniq = 0; 85 debug_step("parsing %s:%d", curr_pos.p_file, curr_pos.p_line); 86 if (curr_pos.p_file == csrc_pos.p_file) { 87 csrc_pos.p_line++; 88 csrc_pos.p_uniq = 0; 89 } 90 } 91 92 void 93 lex_unknown_character(int c) 94 { 95 96 /* unknown character \%o */ 97 error(250, c); 98 } 99 100 #define kwdef(name, token, scl, tspec, tqual, c89, c99, gcc, attr, deco) \ 101 { \ 102 name, token, scl, tspec, tqual, \ 103 (c89) > 0, (c99) > 0, (gcc) > 0, (attr) > 0, \ 104 ((deco) & 1) != 0, ((deco) & 2) != 0, ((deco) & 4) != 0, \ 105 } 106 #define kwdef_token(name, token, c89, c99, gcc, attr, deco) \ 107 kwdef(name, token, 0, 0, 0, c89, c99, gcc, attr, deco) 108 #define kwdef_sclass(name, sclass, c89, c99, gcc, attr, deco) \ 109 kwdef(name, T_SCLASS, sclass, 0, 0, c89, c99, gcc, attr, deco) 110 #define kwdef_type(name, tspec, c89, c99, gcc, attr, deco) \ 111 kwdef(name, T_TYPE, 0, tspec, 0, c89, c99, gcc, attr, deco) 112 #define kwdef_tqual(name, tqual, c89, c99, gcc, attr, deco) \ 113 kwdef(name, T_QUAL, 0, 0, tqual, c89, c99, gcc, attr, deco) 114 #define kwdef_keyword(name, token) \ 115 kwdef(name, token, 0, 0, 0, 0, 0, 0, 0, 1) 116 #define kwdef_gcc_attr(name, token) \ 117 kwdef(name, token, 0, 0, 0, 0, 0, 1, 1, 5) 118 119 /* 120 * Keywords. 121 * During initialization they are written to the symbol table. 122 */ 123 static struct kwtab { 124 const char *kw_name; /* keyword */ 125 int kw_token; /* token returned by yylex() */ 126 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 127 tspec_t kw_tspec; /* type spec. if kw_token 128 * T_TYPE or T_STRUCT_OR_UNION */ 129 tqual_t kw_tqual; /* type qual. if kw_token T_QUAL */ 130 bool kw_c89 : 1; /* C89 keyword */ 131 bool kw_c99 : 1; /* C99 keyword */ 132 bool kw_gcc : 1; /* GCC keyword */ 133 bool kw_attr : 1; /* GCC attribute, keyword */ 134 bool kw_plain : 1; /* 'name' */ 135 bool kw_leading : 1; /* '__name' */ 136 bool kw_both : 1; /* '__name__' */ 137 } kwtab[] = { 138 kwdef_gcc_attr( "alias", T_AT_ALIAS), 139 kwdef_keyword( "_Alignas", T_ALIGNAS), 140 kwdef_keyword( "_Alignof", T_ALIGNOF), 141 kwdef_gcc_attr( "aligned", T_AT_ALIGNED), 142 kwdef_token( "__alignof__", T_ALIGNOF, 0,0,0,0,1), 143 kwdef_gcc_attr( "alloc_size", T_AT_ALLOC_SIZE), 144 kwdef_gcc_attr( "always_inline",T_AT_ALWAYS_INLINE), 145 kwdef_token( "asm", T_ASM, 0,0,1,0,7), 146 kwdef_token( "attribute", T_ATTRIBUTE, 0,0,1,0,6), 147 kwdef_sclass( "auto", AUTO, 0,0,0,0,1), 148 kwdef_type( "_Bool", BOOL, 0,1,0,0,1), 149 kwdef_gcc_attr( "bounded", T_AT_BOUNDED), 150 kwdef_keyword( "break", T_BREAK), 151 kwdef_gcc_attr( "buffer", T_AT_BUFFER), 152 kwdef_token( "__builtin_offsetof", T_BUILTIN_OFFSETOF, 0,0,1,0,1), 153 kwdef_keyword( "case", T_CASE), 154 kwdef_type( "char", CHAR, 0,0,0,0,1), 155 kwdef_gcc_attr( "cold", T_AT_COLD), 156 kwdef_gcc_attr( "common", T_AT_COMMON), 157 kwdef_type( "_Complex", COMPLEX, 0,1,0,0,1), 158 kwdef_tqual( "const", CONST, 1,0,0,0,7), 159 kwdef_gcc_attr( "constructor", T_AT_CONSTRUCTOR), 160 kwdef_keyword( "continue", T_CONTINUE), 161 kwdef_keyword( "default", T_DEFAULT), 162 kwdef_gcc_attr( "deprecated", T_AT_DEPRECATED), 163 kwdef_gcc_attr( "destructor", T_AT_DESTRUCTOR), 164 kwdef_keyword( "do", T_DO), 165 kwdef_type( "double", DOUBLE, 0,0,0,0,1), 166 kwdef_keyword( "else", T_ELSE), 167 kwdef_keyword( "enum", T_ENUM), 168 kwdef_token( "__extension__",T_EXTENSION, 0,0,1,0,1), 169 kwdef_sclass( "extern", EXTERN, 0,0,0,0,1), 170 kwdef_gcc_attr( "fallthrough", T_AT_FALLTHROUGH), 171 kwdef_type( "float", FLOAT, 0,0,0,0,1), 172 kwdef_keyword( "for", T_FOR), 173 kwdef_gcc_attr( "format", T_AT_FORMAT), 174 kwdef_gcc_attr( "format_arg", T_AT_FORMAT_ARG), 175 kwdef_token( "_Generic", T_GENERIC, 0,1,0,0,1), 176 kwdef_gcc_attr( "gnu_inline", T_AT_GNU_INLINE), 177 kwdef_gcc_attr( "gnu_printf", T_AT_FORMAT_GNU_PRINTF), 178 kwdef_keyword( "goto", T_GOTO), 179 kwdef_gcc_attr( "hot", T_AT_HOT), 180 kwdef_keyword( "if", T_IF), 181 kwdef_token( "__imag__", T_IMAG, 0,0,1,0,1), 182 kwdef_sclass( "inline", INLINE, 0,1,0,0,7), 183 kwdef_type( "int", INT, 0,0,0,0,1), 184 #ifdef INT128_SIZE 185 kwdef_type( "__int128_t", INT128, 0,1,0,0,1), 186 #endif 187 kwdef_type( "long", LONG, 0,0,0,0,1), 188 kwdef_gcc_attr( "malloc", T_AT_MALLOC), 189 kwdef_gcc_attr( "may_alias", T_AT_MAY_ALIAS), 190 kwdef_gcc_attr( "minbytes", T_AT_MINBYTES), 191 kwdef_gcc_attr( "mode", T_AT_MODE), 192 kwdef_gcc_attr("no_instrument_function", 193 T_AT_NO_INSTRUMENT_FUNCTION), 194 kwdef_gcc_attr( "noinline", T_AT_NOINLINE), 195 kwdef_gcc_attr( "nonnull", T_AT_NONNULL), 196 kwdef_gcc_attr( "nonstring", T_AT_NONSTRING), 197 kwdef_token( "_Noreturn", T_NORETURN, 0,1,0,0,1), 198 kwdef_gcc_attr( "noreturn", T_AT_NORETURN), 199 kwdef_gcc_attr( "nothrow", T_AT_NOTHROW), 200 kwdef_gcc_attr( "optimize", T_AT_OPTIMIZE), 201 kwdef_gcc_attr( "packed", T_AT_PACKED), 202 kwdef_token( "__packed", T_PACKED, 0,0,0,0,1), 203 kwdef_gcc_attr( "pcs", T_AT_PCS), 204 kwdef_gcc_attr( "printf", T_AT_FORMAT_PRINTF), 205 kwdef_gcc_attr( "pure", T_AT_PURE), 206 kwdef_token( "__real__", T_REAL, 0,0,1,0,1), 207 kwdef_sclass( "register", REG, 0,0,0,0,1), 208 kwdef_gcc_attr( "regparm", T_AT_REGPARM), 209 kwdef_tqual( "restrict", RESTRICT, 0,1,0,0,5), 210 kwdef_keyword( "return", T_RETURN), 211 kwdef_gcc_attr( "returns_nonnull",T_AT_RETURNS_NONNULL), 212 kwdef_gcc_attr( "returns_twice",T_AT_RETURNS_TWICE), 213 kwdef_gcc_attr( "scanf", T_AT_FORMAT_SCANF), 214 kwdef_token( "section", T_AT_SECTION, 0,0,1,1,7), 215 kwdef_gcc_attr( "sentinel", T_AT_SENTINEL), 216 kwdef_type( "short", SHORT, 0,0,0,0,1), 217 kwdef_type( "signed", SIGNED, 1,0,0,0,3), 218 kwdef_keyword( "sizeof", T_SIZEOF), 219 kwdef_sclass( "static", STATIC, 0,0,0,0,1), 220 kwdef_gcc_attr( "strfmon", T_AT_FORMAT_STRFMON), 221 kwdef_gcc_attr( "strftime", T_AT_FORMAT_STRFTIME), 222 kwdef_gcc_attr( "string", T_AT_STRING), 223 kwdef("struct", T_STRUCT_OR_UNION, 0, STRUCT, 0, 0,0,0,0,1), 224 kwdef_keyword( "switch", T_SWITCH), 225 kwdef_token( "__symbolrename", T_SYMBOLRENAME, 0,0,0,0,1), 226 kwdef_gcc_attr( "syslog", T_AT_FORMAT_SYSLOG), 227 kwdef_tqual( "__thread", THREAD, 0,0,1,0,1), 228 kwdef_tqual( "_Thread_local", THREAD, 0,1,0,0,1), 229 kwdef_gcc_attr( "tls_model", T_AT_TLS_MODEL), 230 kwdef_gcc_attr( "transparent_union", T_AT_TUNION), 231 kwdef_sclass( "typedef", TYPEDEF, 0,0,0,0,1), 232 kwdef_token( "typeof", T_TYPEOF, 0,0,1,0,7), 233 #ifdef INT128_SIZE 234 kwdef_type( "__uint128_t", UINT128, 0,1,0,0,1), 235 #endif 236 kwdef("union", T_STRUCT_OR_UNION, 0, UNION, 0, 0,0,0,0,1), 237 kwdef_type( "unsigned", UNSIGN, 0,0,0,0,1), 238 kwdef_gcc_attr( "unused", T_AT_UNUSED), 239 kwdef_gcc_attr( "used", T_AT_USED), 240 kwdef_gcc_attr( "visibility", T_AT_VISIBILITY), 241 kwdef_type( "void", VOID, 0,0,0,0,1), 242 kwdef_tqual( "volatile", VOLATILE, 1,0,0,0,7), 243 kwdef_gcc_attr( "warn_unused_result", T_AT_WARN_UNUSED_RESULT), 244 kwdef_gcc_attr( "weak", T_AT_WEAK), 245 kwdef_keyword( "while", T_WHILE), 246 kwdef(NULL, 0, 0, 0, 0, 0, 0, 0, 0, 0), 247 #undef kwdef 248 #undef kwdef_token 249 #undef kwdef_sclass 250 #undef kwdef_type 251 #undef kwdef_tqual 252 #undef kwdef_keyword 253 #undef kwdef_gcc_attr 254 }; 255 256 /* Symbol table */ 257 static sym_t *symtab[HSHSIZ1]; 258 259 /* free list for sbuf structures */ 260 static sbuf_t *sbfrlst; 261 262 /* type of next expected symbol */ 263 symt_t symtyp; 264 265 266 static void 267 symtab_add(sym_t *sym) 268 { 269 size_t h; 270 271 h = hash(sym->s_name); 272 if ((sym->s_link = symtab[h]) != NULL) 273 symtab[h]->s_rlink = &sym->s_link; 274 sym->s_rlink = &symtab[h]; 275 symtab[h] = sym; 276 } 277 278 static void 279 symtab_remove(sym_t *sym) 280 { 281 282 if ((*sym->s_rlink = sym->s_link) != NULL) 283 sym->s_link->s_rlink = sym->s_rlink; 284 sym->s_link = NULL; 285 } 286 287 288 static void 289 add_keyword(const struct kwtab *kw, bool leading, bool trailing) 290 { 291 sym_t *sym; 292 char buf[256]; 293 const char *name; 294 295 if (!leading && !trailing) { 296 name = kw->kw_name; 297 } else { 298 (void)snprintf(buf, sizeof(buf), "%s%s%s", 299 leading ? "__" : "", kw->kw_name, trailing ? "__" : ""); 300 name = xstrdup(buf); 301 } 302 303 sym = getblk(sizeof(*sym)); 304 sym->s_name = name; 305 sym->s_keyword = kw; 306 sym->s_value.v_quad = kw->kw_token; 307 if (kw->kw_token == T_TYPE || kw->kw_token == T_STRUCT_OR_UNION) { 308 sym->s_tspec = kw->kw_tspec; 309 } else if (kw->kw_token == T_SCLASS) { 310 sym->s_scl = kw->kw_scl; 311 } else if (kw->kw_token == T_QUAL) { 312 sym->s_tqual = kw->kw_tqual; 313 } 314 315 symtab_add(sym); 316 } 317 318 /* 319 * All keywords are written to the symbol table. This saves us looking 320 * in an extra table for each name we found. 321 */ 322 void 323 initscan(void) 324 { 325 struct kwtab *kw; 326 327 for (kw = kwtab; kw->kw_name != NULL; kw++) { 328 if ((kw->kw_c89 || kw->kw_c99) && tflag) 329 continue; 330 if (kw->kw_c99 && !(Sflag || gflag)) 331 continue; 332 if (kw->kw_gcc && !gflag) 333 continue; 334 if (kw->kw_plain) 335 add_keyword(kw, false, false); 336 if (kw->kw_leading) 337 add_keyword(kw, true, false); 338 if (kw->kw_both) 339 add_keyword(kw, true, true); 340 } 341 } 342 343 /* 344 * Get a free sbuf structure, if possible from the free list 345 */ 346 static sbuf_t * 347 allocsb(void) 348 { 349 sbuf_t *sb; 350 351 if ((sb = sbfrlst) != NULL) { 352 sbfrlst = sb->sb_next; 353 #ifdef BLKDEBUG 354 (void)memset(sb, 0, sizeof(*sb)); 355 #else 356 sb->sb_next = NULL; 357 #endif 358 } else { 359 sb = xmalloc(sizeof(*sb)); 360 (void)memset(sb, 0, sizeof(*sb)); 361 } 362 return sb; 363 } 364 365 /* 366 * Put a sbuf structure to the free list 367 */ 368 static void 369 freesb(sbuf_t *sb) 370 { 371 372 (void)memset(sb, ZERO, sizeof(*sb)); 373 sb->sb_next = sbfrlst; 374 sbfrlst = sb; 375 } 376 377 /* 378 * Read a character and ensure that it is positive (except EOF). 379 * Increment line count(s) if necessary. 380 */ 381 static int 382 inpc(void) 383 { 384 int c; 385 386 if ((c = lex_input()) == EOF) 387 return c; 388 c &= CHAR_MASK; 389 if (c == '\0') 390 return EOF; /* lex returns 0 on EOF. */ 391 if (c == '\n') 392 lex_next_line(); 393 return c; 394 } 395 396 static unsigned int 397 hash(const char *s) 398 { 399 unsigned int v; 400 const unsigned char *us; 401 402 v = 0; 403 for (us = (const unsigned char *)s; *us != '\0'; us++) { 404 v = (v << sizeof(v)) + *us; 405 v ^= v >> (sizeof(v) * CHAR_BIT - sizeof(v)); 406 } 407 return v % HSHSIZ1; 408 } 409 410 /* 411 * Lex has found a letter followed by zero or more letters or digits. 412 * It looks for a symbol in the symbol table with the same name. This 413 * symbol must either be a keyword or a symbol of the type required by 414 * symtyp (label, member, tag, ...). 415 * 416 * If it is a keyword, the token is returned. In some cases it is described 417 * more deeply by data written to yylval. 418 * 419 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 420 * is stored in yylval. This struct contains the name of the symbol, its 421 * length and hash value. If there is already a symbol of the same name 422 * and type in the symbol table, the sbuf struct also contains a pointer 423 * to the symbol table entry. 424 */ 425 extern int 426 lex_name(const char *yytext, size_t yyleng) 427 { 428 char *s; 429 sbuf_t *sb; 430 sym_t *sym; 431 int tok; 432 433 sb = allocsb(); 434 sb->sb_name = yytext; 435 sb->sb_len = yyleng; 436 if ((sym = search(sb)) != NULL && sym->s_keyword != NULL) { 437 freesb(sb); 438 return keyw(sym); 439 } 440 441 sb->sb_sym = sym; 442 443 if (sym != NULL) { 444 lint_assert(block_level >= sym->s_block_level); 445 sb->sb_name = sym->s_name; 446 sb->sb_len = strlen(sym->s_name); 447 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 448 } else { 449 s = getblk(yyleng + 1); 450 (void)memcpy(s, yytext, yyleng + 1); 451 sb->sb_name = s; 452 sb->sb_len = yyleng; 453 tok = T_NAME; 454 } 455 456 yylval.y_name = sb; 457 return tok; 458 } 459 460 static sym_t * 461 search(sbuf_t *sb) 462 { 463 unsigned int h; 464 sym_t *sym; 465 const struct kwtab *kw; 466 467 h = hash(sb->sb_name); 468 for (sym = symtab[h]; sym != NULL; sym = sym->s_link) { 469 if (strcmp(sym->s_name, sb->sb_name) != 0) 470 continue; 471 kw = sym->s_keyword; 472 473 if (kw != NULL && !kw->kw_attr) 474 return sym; 475 if (kw != NULL && attron) 476 return sym; 477 if (kw == NULL && !attron && sym->s_kind == symtyp) 478 return sym; 479 } 480 481 return NULL; 482 } 483 484 static int 485 keyw(sym_t *sym) 486 { 487 int t; 488 489 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 490 yylval.y_scl = sym->s_scl; 491 } else if (t == T_TYPE || t == T_STRUCT_OR_UNION) { 492 yylval.y_tspec = sym->s_tspec; 493 } else if (t == T_QUAL) { 494 yylval.y_tqual = sym->s_tqual; 495 } 496 return t; 497 } 498 499 /* 500 * Convert a string representing an integer into internal representation. 501 * Return T_CON, storing the numeric value in yylval, for yylex. 502 */ 503 int 504 lex_integer_constant(const char *yytext, size_t yyleng, int base) 505 { 506 int l_suffix, u_suffix; 507 size_t len; 508 const char *cp; 509 char c, *eptr; 510 tspec_t typ; 511 bool ansiu; 512 bool warned = false; 513 #ifdef TARG_INT128_MAX 514 __uint128_t uq = 0; 515 static tspec_t contypes[2][4] = { 516 { INT, LONG, QUAD, INT128, }, 517 { UINT, ULONG, UQUAD, UINT128, } 518 }; 519 #else 520 uint64_t uq = 0; 521 static tspec_t contypes[2][3] = { 522 { INT, LONG, QUAD, }, 523 { UINT, ULONG, UQUAD, } 524 }; 525 #endif 526 527 cp = yytext; 528 len = yyleng; 529 530 /* skip 0[xX] or 0[bB] */ 531 if (base == 16 || base == 2) { 532 cp += 2; 533 len -= 2; 534 } 535 536 /* read suffixes */ 537 l_suffix = u_suffix = 0; 538 for (;;) { 539 if ((c = cp[len - 1]) == 'l' || c == 'L') { 540 l_suffix++; 541 } else if (c == 'u' || c == 'U') { 542 u_suffix++; 543 } else { 544 break; 545 } 546 len--; 547 } 548 if (l_suffix > 2 || u_suffix > 1) { 549 /* malformed integer constant */ 550 warning(251); 551 if (l_suffix > 2) 552 l_suffix = 2; 553 if (u_suffix > 1) 554 u_suffix = 1; 555 } 556 if (tflag && u_suffix != 0) { 557 /* suffix U is illegal in traditional C */ 558 warning(97); 559 } 560 typ = contypes[u_suffix][l_suffix]; 561 562 errno = 0; 563 564 uq = strtouq(cp, &eptr, base); 565 lint_assert(eptr == cp + len); 566 if (errno != 0) { 567 /* integer constant out of range */ 568 warning(252); 569 warned = true; 570 } 571 572 /* 573 * If the value is too big for the current type, we must choose 574 * another type. 575 */ 576 ansiu = false; 577 switch (typ) { 578 case INT: 579 if (uq <= TARG_INT_MAX) { 580 /* ok */ 581 } else if (uq <= TARG_UINT_MAX && base != 10) { 582 typ = UINT; 583 } else if (uq <= TARG_LONG_MAX) { 584 typ = LONG; 585 } else { 586 typ = ULONG; 587 if (uq > TARG_ULONG_MAX && !warned) { 588 /* integer constant out of range */ 589 warning(252); 590 } 591 } 592 if (typ == UINT || typ == ULONG) { 593 if (tflag) { 594 typ = LONG; 595 } else if (!sflag) { 596 /* 597 * Remember that the constant is unsigned 598 * only in ANSI C 599 */ 600 ansiu = true; 601 } 602 } 603 break; 604 case UINT: 605 if (uq > TARG_UINT_MAX) { 606 typ = ULONG; 607 if (uq > TARG_ULONG_MAX && !warned) { 608 /* integer constant out of range */ 609 warning(252); 610 } 611 } 612 break; 613 case LONG: 614 if (uq > TARG_LONG_MAX && !tflag) { 615 typ = ULONG; 616 if (!sflag) 617 ansiu = true; 618 if (uq > TARG_ULONG_MAX && !warned) { 619 /* integer constant out of range */ 620 warning(252); 621 } 622 } 623 break; 624 case ULONG: 625 if (uq > TARG_ULONG_MAX && !warned) { 626 /* integer constant out of range */ 627 warning(252); 628 } 629 break; 630 case QUAD: 631 if (uq > TARG_QUAD_MAX && !tflag) { 632 typ = UQUAD; 633 if (!sflag) 634 ansiu = true; 635 } 636 break; 637 case UQUAD: 638 if (uq > TARG_UQUAD_MAX && !warned) { 639 /* integer constant out of range */ 640 warning(252); 641 } 642 break; 643 #ifdef INT128_SIZE 644 case INT128: 645 #ifdef TARG_INT128_MAX 646 if (uq > TARG_INT128_MAX && !tflag) { 647 typ = UINT128; 648 if (!sflag) 649 ansiu = true; 650 } 651 #endif 652 break; 653 case UINT128: 654 #ifdef TARG_INT128_MAX 655 if (uq > TARG_UINT128_MAX && !warned) { 656 /* integer constant out of range */ 657 warning(252); 658 } 659 #endif 660 break; 661 #endif 662 default: 663 break; 664 } 665 666 uq = (uint64_t)convert_integer((int64_t)uq, typ, 0); 667 668 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 669 yylval.y_val->v_tspec = typ; 670 yylval.y_val->v_unsigned_since_c90 = ansiu; 671 yylval.y_val->v_quad = (int64_t)uq; 672 673 return T_CON; 674 } 675 676 /* 677 * Extend or truncate q to match t. If t is signed, sign-extend. 678 * 679 * len is the number of significant bits. If len is -1, len is set 680 * to the width of type t. 681 */ 682 int64_t 683 convert_integer(int64_t q, tspec_t t, unsigned int len) 684 { 685 uint64_t vbits; 686 687 if (len == 0) 688 len = size_in_bits(t); 689 690 vbits = value_bits(len); 691 return t == PTR || is_uinteger(t) || ((q & bit(len - 1)) == 0) 692 ? (int64_t)(q & vbits) 693 : (int64_t)(q | ~vbits); 694 } 695 696 /* 697 * Convert a string representing a floating point value into its integral 698 * representation. Type and value are returned in yylval. fcon() 699 * (and yylex()) returns T_CON. 700 * XXX Currently it is not possible to convert constants of type 701 * long double which are greater than DBL_MAX. 702 */ 703 int 704 lex_floating_constant(const char *yytext, size_t yyleng) 705 { 706 const char *cp; 707 int len; 708 tspec_t typ; 709 char c, *eptr; 710 double d; 711 float f = 0; 712 713 cp = yytext; 714 len = yyleng; 715 716 if (cp[len - 1] == 'i') { 717 /* imaginary, do nothing for now */ 718 len--; 719 } 720 if ((c = cp[len - 1]) == 'f' || c == 'F') { 721 typ = FLOAT; 722 len--; 723 } else if (c == 'l' || c == 'L') { 724 typ = LDOUBLE; 725 len--; 726 } else { 727 if (c == 'd' || c == 'D') 728 len--; 729 typ = DOUBLE; 730 } 731 732 if (tflag && typ != DOUBLE) { 733 /* suffixes F and L are illegal in traditional C */ 734 warning(98); 735 } 736 737 errno = 0; 738 d = strtod(cp, &eptr); 739 if (eptr != cp + len) { 740 switch (*eptr) { 741 /* 742 * XXX: non-native non-current strtod() may not handle hex 743 * floats, ignore the rest if we find traces of hex float 744 * syntax... 745 */ 746 case 'p': 747 case 'P': 748 case 'x': 749 case 'X': 750 d = 0; 751 errno = 0; 752 break; 753 default: 754 INTERNAL_ERROR("fcon(%s->%s)", cp, eptr); 755 } 756 } 757 if (errno != 0) 758 /* floating-point constant out of range */ 759 warning(248); 760 761 if (typ == FLOAT) { 762 f = (float)d; 763 if (finite(f) == 0) { 764 /* floating-point constant out of range */ 765 warning(248); 766 f = f > 0 ? FLT_MAX : -FLT_MAX; 767 } 768 } 769 770 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 771 yylval.y_val->v_tspec = typ; 772 if (typ == FLOAT) { 773 yylval.y_val->v_ldbl = f; 774 } else { 775 yylval.y_val->v_ldbl = d; 776 } 777 778 return T_CON; 779 } 780 781 int 782 lex_operator(int t, op_t o) 783 { 784 785 yylval.y_op = o; 786 return t; 787 } 788 789 /* 790 * Called if lex found a leading \'. 791 */ 792 int 793 lex_character_constant(void) 794 { 795 size_t n; 796 int val, c; 797 798 n = 0; 799 val = 0; 800 while ((c = get_escaped_char('\'')) >= 0) { 801 val = (val << CHAR_SIZE) + c; 802 n++; 803 } 804 if (c == -2) { 805 /* unterminated character constant */ 806 error(253); 807 } else if (n > sizeof(int) || (n > 1 && (pflag || hflag))) { 808 /* XXX: should rather be sizeof(TARG_INT) */ 809 810 /* too many characters in character constant */ 811 error(71); 812 } else if (n > 1) { 813 /* multi-character character constant */ 814 warning(294); 815 } else if (n == 0) { 816 /* empty character constant */ 817 error(73); 818 } 819 if (n == 1) { 820 /* 821 * XXX: use the target platform's 'char' instead of the 822 * 'char' from the execution environment, to be able to 823 * run lint for powerpc on x86_64. 824 */ 825 val = (char)val; 826 } 827 828 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 829 yylval.y_val->v_tspec = INT; 830 yylval.y_val->v_quad = val; 831 832 return T_CON; 833 } 834 835 /* 836 * Called if lex found a leading L\' 837 */ 838 int 839 lex_wide_character_constant(void) 840 { 841 static char buf[MB_LEN_MAX + 1]; 842 size_t n, nmax; 843 int c; 844 wchar_t wc; 845 846 nmax = MB_CUR_MAX; 847 848 n = 0; 849 while ((c = get_escaped_char('\'')) >= 0) { 850 if (n < nmax) 851 buf[n] = (char)c; 852 n++; 853 } 854 855 wc = 0; 856 857 if (c == -2) { 858 /* unterminated character constant */ 859 error(253); 860 } else if (n == 0) { 861 /* empty character constant */ 862 error(73); 863 } else if (n > nmax) { 864 n = nmax; 865 /* too many characters in character constant */ 866 error(71); 867 } else { 868 buf[n] = '\0'; 869 (void)mbtowc(NULL, NULL, 0); 870 if (mbtowc(&wc, buf, nmax) < 0) 871 /* invalid multibyte character */ 872 error(291); 873 } 874 875 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 876 yylval.y_val->v_tspec = WCHAR; 877 yylval.y_val->v_quad = wc; 878 879 return T_CON; 880 } 881 882 /* 883 * Read a character which is part of a character constant or of a string 884 * and handle escapes. 885 * 886 * The argument is the character which delimits the character constant or 887 * string. 888 * 889 * Returns -1 if the end of the character constant or string is reached, 890 * -2 if the EOF is reached, and the character otherwise. 891 */ 892 static int 893 get_escaped_char(int delim) 894 { 895 static int pbc = -1; 896 int n, c, v; 897 898 if (pbc == -1) { 899 c = inpc(); 900 } else { 901 c = pbc; 902 pbc = -1; 903 } 904 if (c == delim) 905 return -1; 906 switch (c) { 907 case '\n': 908 if (tflag) { 909 /* newline in string or char constant */ 910 error(254); 911 return -2; 912 } 913 return c; 914 case 0: 915 /* syntax error '%s' */ 916 error(249, "EOF or null byte in literal"); 917 return -2; 918 case EOF: 919 return -2; 920 case '\\': 921 switch (c = inpc()) { 922 case '"': 923 if (tflag && delim == '\'') 924 /* \" inside character constants undef... */ 925 warning(262); 926 return '"'; 927 case '\'': 928 return '\''; 929 case '?': 930 if (tflag) 931 /* \? undefined in traditional C */ 932 warning(263); 933 return '?'; 934 case '\\': 935 return '\\'; 936 case 'a': 937 if (tflag) 938 /* \a undefined in traditional C */ 939 warning(81); 940 return '\a'; 941 case 'b': 942 return '\b'; 943 case 'f': 944 return '\f'; 945 case 'n': 946 return '\n'; 947 case 'r': 948 return '\r'; 949 case 't': 950 return '\t'; 951 case 'v': 952 if (tflag) 953 /* \v undefined in traditional C */ 954 warning(264); 955 return '\v'; 956 case '8': case '9': 957 /* bad octal digit %c */ 958 warning(77, c); 959 /* FALLTHROUGH */ 960 case '0': case '1': case '2': case '3': 961 case '4': case '5': case '6': case '7': 962 n = 3; 963 v = 0; 964 do { 965 v = (v << 3) + (c - '0'); 966 c = inpc(); 967 } while (--n > 0 && '0' <= c && c <= '7'); 968 pbc = c; 969 if (v > TARG_UCHAR_MAX) { 970 /* character escape does not fit in character */ 971 warning(76); 972 v &= CHAR_MASK; 973 } 974 return v; 975 case 'x': 976 if (tflag) 977 /* \x undefined in traditional C */ 978 warning(82); 979 v = 0; 980 n = 0; 981 while ((c = inpc()) >= 0 && isxdigit(c)) { 982 c = isdigit(c) ? 983 c - '0' : toupper(c) - 'A' + 10; 984 v = (v << 4) + c; 985 if (n >= 0) { 986 if ((v & ~CHAR_MASK) != 0) { 987 /* overflow in hex escape */ 988 warning(75); 989 n = -1; 990 } else { 991 n++; 992 } 993 } 994 } 995 pbc = c; 996 if (n == 0) { 997 /* no hex digits follow \x */ 998 error(74); 999 } if (n == -1) { 1000 v &= CHAR_MASK; 1001 } 1002 return v; 1003 case '\n': 1004 return get_escaped_char(delim); 1005 case EOF: 1006 return -2; 1007 default: 1008 if (isprint(c)) { 1009 /* dubious escape \%c */ 1010 warning(79, c); 1011 } else { 1012 /* dubious escape \%o */ 1013 warning(80, c); 1014 } 1015 } 1016 } 1017 return c; 1018 } 1019 1020 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */ 1021 static void 1022 parse_line_directive_flags(const char *p, 1023 bool *is_begin, bool *is_end, bool *is_system) 1024 { 1025 1026 *is_begin = false; 1027 *is_end = false; 1028 *is_system = false; 1029 1030 while (*p != '\0') { 1031 while (ch_isspace(*p)) 1032 p++; 1033 1034 const char *word_start = p; 1035 while (*p != '\0' && !ch_isspace(*p)) 1036 p++; 1037 const char *word_end = p; 1038 1039 if (word_end - word_start == 1 && word_start[0] == '1') 1040 *is_begin = true; 1041 if (word_end - word_start == 1 && word_start[0] == '2') 1042 *is_end = true; 1043 if (word_end - word_start == 1 && word_start[0] == '3') 1044 *is_system = true; 1045 /* Flag '4' would only be interesting if lint handled C++. */ 1046 } 1047 1048 #if 0 1049 if (*p != '\0') { 1050 /* syntax error '%s' */ 1051 warning(249, "extra character(s) after directive"); 1052 } 1053 #endif 1054 } 1055 1056 /* 1057 * Called for preprocessor directives. Currently implemented are: 1058 * # lineno 1059 * # lineno "filename" 1060 * # lineno "filename" GCC-flag... 1061 */ 1062 void 1063 lex_directive(const char *yytext) 1064 { 1065 const char *cp, *fn; 1066 char c, *eptr; 1067 size_t fnl; 1068 long ln; 1069 bool is_begin, is_end, is_system; 1070 1071 static bool first = true; 1072 1073 /* Go to first non-whitespace after # */ 1074 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1075 continue; 1076 1077 if (!ch_isdigit(c)) { 1078 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6])) 1079 return; 1080 error: 1081 /* undefined or invalid # directive */ 1082 warning(255); 1083 return; 1084 } 1085 ln = strtol(--cp, &eptr, 10); 1086 if (cp == eptr) 1087 goto error; 1088 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1089 goto error; 1090 while ((c = *cp++) == ' ' || c == '\t') 1091 continue; 1092 if (c != '\0') { 1093 if (c != '"') 1094 goto error; 1095 fn = cp; 1096 while ((c = *cp) != '"' && c != '\0') 1097 cp++; 1098 if (c != '"') 1099 goto error; 1100 if ((fnl = cp++ - fn) > PATH_MAX) 1101 goto error; 1102 /* empty string means stdin */ 1103 if (fnl == 0) { 1104 fn = "{standard input}"; 1105 fnl = 16; /* strlen (fn) */ 1106 } 1107 curr_pos.p_file = record_filename(fn, fnl); 1108 /* 1109 * If this is the first directive, the name is the name 1110 * of the C source file as specified at the command line. 1111 * It is written to the output file. 1112 */ 1113 if (first) { 1114 csrc_pos.p_file = curr_pos.p_file; 1115 outsrc(transform_filename(curr_pos.p_file, 1116 strlen(curr_pos.p_file))); 1117 first = false; 1118 } 1119 1120 parse_line_directive_flags(cp, &is_begin, &is_end, &is_system); 1121 update_location(curr_pos.p_file, (int)ln, is_begin, is_end); 1122 in_system_header = is_system; 1123 } 1124 curr_pos.p_line = (int)ln - 1; 1125 curr_pos.p_uniq = 0; 1126 if (curr_pos.p_file == csrc_pos.p_file) { 1127 csrc_pos.p_line = (int)ln - 1; 1128 csrc_pos.p_uniq = 0; 1129 } 1130 } 1131 1132 /* 1133 * Handle lint comments such as ARGSUSED. 1134 * 1135 * If one of these comments is recognized, the argument, if any, is 1136 * parsed and a function which handles this comment is called. 1137 */ 1138 void 1139 lex_comment(void) 1140 { 1141 int c, lc; 1142 static const struct { 1143 const char *keywd; 1144 bool arg; 1145 void (*func)(int); 1146 } keywtab[] = { 1147 { "ARGSUSED", true, argsused }, 1148 { "BITFIELDTYPE", false, bitfieldtype }, 1149 { "CONSTCOND", false, constcond }, 1150 { "CONSTANTCOND", false, constcond }, 1151 { "CONSTANTCONDITION", false, constcond }, 1152 { "FALLTHRU", false, fallthru }, 1153 { "FALLTHROUGH", false, fallthru }, 1154 { "LINTLIBRARY", false, lintlib }, 1155 { "LINTED", true, linted }, 1156 { "LONGLONG", false, longlong }, 1157 { "NOSTRICT", true, linted }, 1158 { "NOTREACHED", false, not_reached }, 1159 { "PRINTFLIKE", true, printflike }, 1160 { "PROTOLIB", true, protolib }, 1161 { "SCANFLIKE", true, scanflike }, 1162 { "VARARGS", true, varargs }, 1163 }; 1164 char keywd[32]; 1165 char arg[32]; 1166 size_t l, i; 1167 int a; 1168 bool eoc; 1169 1170 eoc = false; 1171 1172 /* Skip whitespace after the start of the comment */ 1173 while ((c = inpc()) != EOF && isspace(c)) 1174 continue; 1175 1176 /* Read the potential keyword to keywd */ 1177 l = 0; 1178 while (c != EOF && isupper(c) && l < sizeof(keywd) - 1) { 1179 keywd[l++] = (char)c; 1180 c = inpc(); 1181 } 1182 keywd[l] = '\0'; 1183 1184 /* look for the keyword */ 1185 for (i = 0; i < sizeof(keywtab) / sizeof(keywtab[0]); i++) { 1186 if (strcmp(keywtab[i].keywd, keywd) == 0) 1187 break; 1188 } 1189 if (i == sizeof(keywtab) / sizeof(keywtab[0])) 1190 goto skip_rest; 1191 1192 /* skip whitespace after the keyword */ 1193 while (c != EOF && isspace(c)) 1194 c = inpc(); 1195 1196 /* read the argument, if the keyword accepts one and there is one */ 1197 l = 0; 1198 if (keywtab[i].arg) { 1199 while (c != EOF && isdigit(c) && l < sizeof(arg) - 1) { 1200 arg[l++] = (char)c; 1201 c = inpc(); 1202 } 1203 } 1204 arg[l] = '\0'; 1205 a = l != 0 ? atoi(arg) : -1; 1206 1207 /* skip whitespace after the argument */ 1208 while (c != EOF && isspace(c)) 1209 c = inpc(); 1210 1211 if (c != '*' || (c = inpc()) != '/') { 1212 if (keywtab[i].func != linted) 1213 /* extra characters in lint comment */ 1214 warning(257); 1215 } else { 1216 /* 1217 * remember that we have already found the end of the 1218 * comment 1219 */ 1220 eoc = true; 1221 } 1222 1223 if (keywtab[i].func != NULL) 1224 (*keywtab[i].func)(a); 1225 1226 skip_rest: 1227 while (!eoc) { 1228 lc = c; 1229 if ((c = inpc()) == EOF) { 1230 /* unterminated comment */ 1231 error(256); 1232 break; 1233 } 1234 if (lc == '*' && c == '/') 1235 eoc = true; 1236 } 1237 } 1238 1239 /* 1240 * Handle // style comments 1241 */ 1242 void 1243 lex_slash_slash_comment(void) 1244 { 1245 int c; 1246 1247 if (!Sflag && !gflag) 1248 /* %s C does not support // comments */ 1249 gnuism(312, tflag ? "traditional" : "ANSI"); 1250 1251 while ((c = inpc()) != EOF && c != '\n') 1252 continue; 1253 } 1254 1255 /* 1256 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1257 * clear_warn_flags() is called after function definitions and global and 1258 * local declarations and definitions. It is also called between 1259 * the controlling expression and the body of control statements 1260 * (if, switch, for, while). 1261 */ 1262 void 1263 clear_warn_flags(void) 1264 { 1265 1266 lwarn = LWARN_ALL; 1267 quadflg = false; 1268 constcond_flag = false; 1269 } 1270 1271 /* 1272 * Strings are stored in a dynamically allocated buffer and passed 1273 * in yylval.y_xstrg to the parser. The parser or the routines called 1274 * by the parser are responsible for freeing this buffer. 1275 */ 1276 int 1277 lex_string(void) 1278 { 1279 unsigned char *s; 1280 int c; 1281 size_t len, max; 1282 strg_t *strg; 1283 1284 s = xmalloc(max = 64); 1285 1286 len = 0; 1287 while ((c = get_escaped_char('"')) >= 0) { 1288 /* +1 to reserve space for a trailing NUL character */ 1289 if (len + 1 == max) 1290 s = xrealloc(s, max *= 2); 1291 s[len++] = (char)c; 1292 } 1293 s[len] = '\0'; 1294 if (c == -2) 1295 /* unterminated string constant */ 1296 error(258); 1297 1298 strg = xcalloc(1, sizeof(*strg)); 1299 strg->st_tspec = CHAR; 1300 strg->st_len = len; 1301 strg->st_cp = s; 1302 1303 yylval.y_string = strg; 1304 return T_STRING; 1305 } 1306 1307 int 1308 lex_wide_string(void) 1309 { 1310 char *s; 1311 int c, n; 1312 size_t i, wi; 1313 size_t len, max, wlen; 1314 wchar_t *ws; 1315 strg_t *strg; 1316 1317 s = xmalloc(max = 64); 1318 len = 0; 1319 while ((c = get_escaped_char('"')) >= 0) { 1320 /* +1 to save space for a trailing NUL character */ 1321 if (len + 1 >= max) 1322 s = xrealloc(s, max *= 2); 1323 s[len++] = (char)c; 1324 } 1325 s[len] = '\0'; 1326 if (c == -2) 1327 /* unterminated string constant */ 1328 error(258); 1329 1330 /* get length of wide-character string */ 1331 (void)mblen(NULL, 0); 1332 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1333 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1334 /* invalid multibyte character */ 1335 error(291); 1336 break; 1337 } 1338 if (n == 0) 1339 n = 1; 1340 } 1341 1342 ws = xmalloc((wlen + 1) * sizeof(*ws)); 1343 1344 /* convert from multibyte to wide char */ 1345 (void)mbtowc(NULL, NULL, 0); 1346 for (i = 0, wi = 0; i < len; i += n, wi++) { 1347 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1348 break; 1349 if (n == 0) 1350 n = 1; 1351 } 1352 ws[wi] = 0; 1353 free(s); 1354 1355 strg = xcalloc(1, sizeof(*strg)); 1356 strg->st_tspec = WCHAR; 1357 strg->st_len = wlen; 1358 strg->st_wcp = ws; 1359 1360 yylval.y_string = strg; 1361 return T_STRING; 1362 } 1363 1364 /* 1365 * As noted above, the scanner does not create new symbol table entries 1366 * for symbols it cannot find in the symbol table. This is to avoid 1367 * putting undeclared symbols into the symbol table if a syntax error 1368 * occurs. 1369 * 1370 * getsym() is called as soon as it is probably ok to put the symbol in the 1371 * symbol table. It is still possible that symbols are put in the symbol 1372 * table that are not completely declared due to syntax errors. To avoid too 1373 * many problems in this case, symbols get type 'int' in getsym(). 1374 * 1375 * XXX calls to getsym() should be delayed until decl1*() is called. 1376 */ 1377 sym_t * 1378 getsym(sbuf_t *sb) 1379 { 1380 dinfo_t *di; 1381 char *s; 1382 sym_t *sym; 1383 1384 sym = sb->sb_sym; 1385 1386 /* 1387 * During member declaration it is possible that name() looked 1388 * for symbols of type FVFT, although it should have looked for 1389 * symbols of type FTAG. Same can happen for labels. Both cases 1390 * are compensated here. 1391 */ 1392 if (symtyp == FMEMBER || symtyp == FLABEL) { 1393 if (sym == NULL || sym->s_kind == FVFT) 1394 sym = search(sb); 1395 } 1396 1397 if (sym != NULL) { 1398 if (sym->s_kind != symtyp) 1399 INTERNAL_ERROR("getsym(%d, %d)", sym->s_kind, symtyp); 1400 symtyp = FVFT; 1401 freesb(sb); 1402 return sym; 1403 } 1404 1405 /* create a new symbol table entry */ 1406 1407 /* labels must always be allocated at level 1 (outermost block) */ 1408 if (symtyp == FLABEL) { 1409 sym = getlblk(1, sizeof(*sym)); 1410 s = getlblk(1, sb->sb_len + 1); 1411 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1412 sym->s_name = s; 1413 sym->s_block_level = 1; 1414 di = dcs; 1415 while (di->d_next != NULL && di->d_next->d_next != NULL) 1416 di = di->d_next; 1417 lint_assert(di->d_ctx == AUTO); 1418 } else { 1419 sym = getblk(sizeof(*sym)); 1420 sym->s_name = sb->sb_name; 1421 sym->s_block_level = block_level; 1422 di = dcs; 1423 } 1424 1425 UNIQUE_CURR_POS(sym->s_def_pos); 1426 if ((sym->s_kind = symtyp) != FLABEL) 1427 sym->s_type = gettyp(INT); 1428 1429 symtyp = FVFT; 1430 1431 symtab_add(sym); 1432 1433 *di->d_ldlsym = sym; 1434 di->d_ldlsym = &sym->s_dlnxt; 1435 1436 freesb(sb); 1437 return sym; 1438 } 1439 1440 /* 1441 * Construct a temporary symbol. The symbol name starts with a digit, making 1442 * the name illegal. 1443 */ 1444 sym_t * 1445 mktempsym(type_t *t) 1446 { 1447 static int n = 0; 1448 char *s = getlblk(block_level, 64); 1449 sym_t *sym = getblk(sizeof(*sym)); 1450 scl_t scl; 1451 1452 (void)snprintf(s, 64, "%.8d_tmp", n++); 1453 1454 scl = dcs->d_scl; 1455 if (scl == NOSCL) 1456 scl = block_level > 0 ? AUTO : EXTERN; 1457 1458 sym->s_name = s; 1459 sym->s_type = t; 1460 sym->s_block_level = block_level; 1461 sym->s_scl = scl; 1462 sym->s_kind = FVFT; 1463 sym->s_used = true; 1464 sym->s_set = true; 1465 1466 symtab_add(sym); 1467 1468 *dcs->d_ldlsym = sym; 1469 dcs->d_ldlsym = &sym->s_dlnxt; 1470 1471 return sym; 1472 } 1473 1474 /* Remove a symbol forever from the symbol table. */ 1475 void 1476 rmsym(sym_t *sym) 1477 { 1478 1479 debug_step("rmsym '%s' %d '%s'", 1480 sym->s_name, (int)sym->s_kind, type_name(sym->s_type)); 1481 symtab_remove(sym); 1482 1483 /* avoid that the symbol will later be put back to the symbol table */ 1484 sym->s_block_level = -1; 1485 } 1486 1487 /* 1488 * Remove a list of symbols declared at one level from the symbol 1489 * table. 1490 */ 1491 void 1492 rmsyms(sym_t *syms) 1493 { 1494 sym_t *sym; 1495 1496 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1497 if (sym->s_block_level != -1) { 1498 debug_step("rmsyms '%s' %d '%s'", 1499 sym->s_name, (int)sym->s_kind, 1500 type_name(sym->s_type)); 1501 symtab_remove(sym); 1502 sym->s_rlink = NULL; 1503 } 1504 } 1505 } 1506 1507 /* 1508 * Put a symbol into the symbol table. 1509 */ 1510 void 1511 inssym(int bl, sym_t *sym) 1512 { 1513 1514 debug_step("inssym '%s' %d '%s'", 1515 sym->s_name, sym->s_kind, type_name(sym->s_type)); 1516 symtab_add(sym); 1517 sym->s_block_level = bl; 1518 lint_assert(sym->s_link == NULL || 1519 sym->s_block_level >= sym->s_link->s_block_level); 1520 } 1521 1522 /* 1523 * Called at level 0 after syntax errors. 1524 * 1525 * Removes all symbols which are not declared at level 0 from the 1526 * symbol table. Also frees all memory which is not associated with 1527 * level 0. 1528 */ 1529 void 1530 cleanup(void) 1531 { 1532 sym_t *sym, *nsym; 1533 int i; 1534 1535 for (i = 0; i < HSHSIZ1; i++) { 1536 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1537 nsym = sym->s_link; 1538 if (sym->s_block_level >= 1) 1539 symtab_remove(sym); 1540 } 1541 } 1542 1543 for (i = mem_block_level; i > 0; i--) 1544 freelblk(i); 1545 } 1546 1547 /* 1548 * Create a new symbol with the name of an existing symbol. 1549 */ 1550 sym_t * 1551 pushdown(const sym_t *sym) 1552 { 1553 sym_t *nsym; 1554 1555 debug_step("pushdown '%s' %d '%s'", 1556 sym->s_name, (int)sym->s_kind, type_name(sym->s_type)); 1557 nsym = getblk(sizeof(*nsym)); 1558 lint_assert(sym->s_block_level <= block_level); 1559 nsym->s_name = sym->s_name; 1560 UNIQUE_CURR_POS(nsym->s_def_pos); 1561 nsym->s_kind = sym->s_kind; 1562 nsym->s_block_level = block_level; 1563 1564 symtab_add(nsym); 1565 1566 *dcs->d_ldlsym = nsym; 1567 dcs->d_ldlsym = &nsym->s_dlnxt; 1568 1569 return nsym; 1570 } 1571 1572 /* 1573 * Free any dynamically allocated memory referenced by 1574 * the value stack or yylval. 1575 * The type of information in yylval is described by tok. 1576 */ 1577 void 1578 freeyyv(void *sp, int tok) 1579 { 1580 if (tok == T_NAME || tok == T_TYPENAME) { 1581 sbuf_t *sb = *(sbuf_t **)sp; 1582 freesb(sb); 1583 } else if (tok == T_CON) { 1584 val_t *val = *(val_t **)sp; 1585 free(val); 1586 } else if (tok == T_STRING) { 1587 strg_t *strg = *(strg_t **)sp; 1588 if (strg->st_tspec == CHAR) { 1589 free(strg->st_cp); 1590 } else { 1591 lint_assert(strg->st_tspec == WCHAR); 1592 free(strg->st_wcp); 1593 } 1594 free(strg); 1595 } 1596 } 1597
Indexes created Fri Sep 26 20:09:58 GMT 2025