lex.c revision 1.75
1 /* $NetBSD: lex.c,v 1.75 2021/08/28 18:58:24 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 1.75 2021/08/28 18:58:24 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 /* FIXME: INT128 doesn't belong here. */ 516 /* TODO: const */ 517 /* TODO: remove #ifdef */ 518 static tspec_t contypes[2][4] = { 519 { INT, LONG, QUAD, INT128, }, 520 { UINT, ULONG, UQUAD, UINT128, } 521 }; 522 #else 523 uint64_t uq = 0; 524 static tspec_t contypes[2][3] = { 525 { INT, LONG, QUAD, }, 526 { UINT, ULONG, UQUAD, } 527 }; 528 #endif 529 530 cp = yytext; 531 len = yyleng; 532 533 /* skip 0[xX] or 0[bB] */ 534 if (base == 16 || base == 2) { 535 cp += 2; 536 len -= 2; 537 } 538 539 /* read suffixes */ 540 l_suffix = u_suffix = 0; 541 for (;;) { 542 if ((c = cp[len - 1]) == 'l' || c == 'L') { 543 l_suffix++; 544 } else if (c == 'u' || c == 'U') { 545 u_suffix++; 546 } else { 547 break; 548 } 549 len--; 550 } 551 if (l_suffix > 2 || u_suffix > 1) { 552 /* malformed integer constant */ 553 warning(251); 554 if (l_suffix > 2) 555 l_suffix = 2; 556 if (u_suffix > 1) 557 u_suffix = 1; 558 } 559 if (tflag && u_suffix != 0) { 560 /* suffix U is illegal in traditional C */ 561 warning(97); 562 } 563 typ = contypes[u_suffix][l_suffix]; 564 565 errno = 0; 566 567 uq = strtoull(cp, &eptr, base); 568 lint_assert(eptr == cp + len); 569 if (errno != 0) { 570 /* integer constant out of range */ 571 warning(252); 572 warned = true; 573 } 574 575 /* 576 * If the value is too big for the current type, we must choose 577 * another type. 578 */ 579 ansiu = false; 580 switch (typ) { 581 case INT: 582 if (uq <= TARG_INT_MAX) { 583 /* ok */ 584 } else if (uq <= TARG_UINT_MAX && base != 10) { 585 typ = UINT; 586 } else if (uq <= TARG_LONG_MAX) { 587 typ = LONG; 588 } else { 589 typ = ULONG; 590 if (uq > TARG_ULONG_MAX && !warned) { 591 /* integer constant out of range */ 592 warning(252); 593 } 594 } 595 if (typ == UINT || typ == ULONG) { 596 if (tflag) { 597 typ = LONG; 598 } else if (!sflag) { 599 /* 600 * Remember that the constant is unsigned 601 * only in ANSI C 602 */ 603 ansiu = true; 604 } 605 } 606 break; 607 case UINT: 608 if (uq > TARG_UINT_MAX) { 609 typ = ULONG; 610 if (uq > TARG_ULONG_MAX && !warned) { 611 /* integer constant out of range */ 612 warning(252); 613 } 614 } 615 break; 616 case LONG: 617 if (uq > TARG_LONG_MAX && !tflag) { 618 typ = ULONG; 619 if (!sflag) 620 ansiu = true; 621 if (uq > TARG_ULONG_MAX && !warned) { 622 /* integer constant out of range */ 623 warning(252); 624 } 625 } 626 break; 627 case ULONG: 628 if (uq > TARG_ULONG_MAX && !warned) { 629 /* integer constant out of range */ 630 warning(252); 631 } 632 break; 633 case QUAD: 634 if (uq > TARG_QUAD_MAX && !tflag) { 635 typ = UQUAD; 636 if (!sflag) 637 ansiu = true; 638 } 639 break; 640 case UQUAD: 641 if (uq > TARG_UQUAD_MAX && !warned) { 642 /* integer constant out of range */ 643 warning(252); 644 } 645 break; 646 #ifdef INT128_SIZE 647 case INT128: 648 #ifdef TARG_INT128_MAX 649 if (uq > TARG_INT128_MAX && !tflag) { 650 typ = UINT128; 651 if (!sflag) 652 ansiu = true; 653 } 654 #endif 655 break; 656 case UINT128: 657 #ifdef TARG_INT128_MAX 658 if (uq > TARG_UINT128_MAX && !warned) { 659 /* integer constant out of range */ 660 warning(252); 661 } 662 #endif 663 break; 664 #endif 665 default: 666 break; 667 } 668 669 uq = (uint64_t)convert_integer((int64_t)uq, typ, 0); 670 671 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 672 yylval.y_val->v_tspec = typ; 673 yylval.y_val->v_unsigned_since_c90 = ansiu; 674 yylval.y_val->v_quad = (int64_t)uq; 675 676 return T_CON; 677 } 678 679 /* 680 * Extend or truncate q to match t. If t is signed, sign-extend. 681 * 682 * len is the number of significant bits. If len is -1, len is set 683 * to the width of type t. 684 */ 685 int64_t 686 convert_integer(int64_t q, tspec_t t, unsigned int len) 687 { 688 uint64_t vbits; 689 690 if (len == 0) 691 len = size_in_bits(t); 692 693 vbits = value_bits(len); 694 return t == PTR || is_uinteger(t) || ((q & bit(len - 1)) == 0) 695 ? (int64_t)(q & vbits) 696 : (int64_t)(q | ~vbits); 697 } 698 699 /* 700 * Convert a string representing a floating point value into its integral 701 * representation. Type and value are returned in yylval. fcon() 702 * (and yylex()) returns T_CON. 703 * XXX Currently it is not possible to convert constants of type 704 * long double which are greater than DBL_MAX. 705 */ 706 int 707 lex_floating_constant(const char *yytext, size_t yyleng) 708 { 709 const char *cp; 710 int len; 711 tspec_t typ; 712 char c, *eptr; 713 double d; 714 float f = 0; 715 716 cp = yytext; 717 len = yyleng; 718 719 if (cp[len - 1] == 'i') { 720 /* imaginary, do nothing for now */ 721 len--; 722 } 723 if ((c = cp[len - 1]) == 'f' || c == 'F') { 724 typ = FLOAT; 725 len--; 726 } else if (c == 'l' || c == 'L') { 727 typ = LDOUBLE; 728 len--; 729 } else { 730 if (c == 'd' || c == 'D') 731 len--; 732 typ = DOUBLE; 733 } 734 735 if (tflag && typ != DOUBLE) { 736 /* suffixes F and L are illegal in traditional C */ 737 warning(98); 738 } 739 740 errno = 0; 741 d = strtod(cp, &eptr); 742 if (eptr != cp + len) { 743 switch (*eptr) { 744 /* 745 * XXX: non-native non-current strtod() may not handle hex 746 * floats, ignore the rest if we find traces of hex float 747 * syntax... 748 */ 749 case 'p': 750 case 'P': 751 case 'x': 752 case 'X': 753 d = 0; 754 errno = 0; 755 break; 756 default: 757 INTERNAL_ERROR("fcon(%s->%s)", cp, eptr); 758 } 759 } 760 if (errno != 0) 761 /* floating-point constant out of range */ 762 warning(248); 763 764 if (typ == FLOAT) { 765 f = (float)d; 766 if (finite(f) == 0) { 767 /* floating-point constant out of range */ 768 warning(248); 769 f = f > 0 ? FLT_MAX : -FLT_MAX; 770 } 771 } 772 773 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 774 yylval.y_val->v_tspec = typ; 775 if (typ == FLOAT) { 776 yylval.y_val->v_ldbl = f; 777 } else { 778 yylval.y_val->v_ldbl = d; 779 } 780 781 return T_CON; 782 } 783 784 int 785 lex_operator(int t, op_t o) 786 { 787 788 yylval.y_op = o; 789 return t; 790 } 791 792 /* 793 * Called if lex found a leading \'. 794 */ 795 int 796 lex_character_constant(void) 797 { 798 size_t n; 799 int val, c; 800 801 n = 0; 802 val = 0; 803 while ((c = get_escaped_char('\'')) >= 0) { 804 val = (val << CHAR_SIZE) + c; 805 n++; 806 } 807 if (c == -2) { 808 /* unterminated character constant */ 809 error(253); 810 } else if (n > sizeof(int) || (n > 1 && (pflag || hflag))) { 811 /* XXX: should rather be sizeof(TARG_INT) */ 812 813 /* too many characters in character constant */ 814 error(71); 815 } else if (n > 1) { 816 /* multi-character character constant */ 817 warning(294); 818 } else if (n == 0) { 819 /* empty character constant */ 820 error(73); 821 } 822 if (n == 1) 823 val = (int)convert_integer(val, CHAR, CHAR_SIZE); 824 825 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 826 yylval.y_val->v_tspec = INT; 827 yylval.y_val->v_quad = val; 828 829 return T_CON; 830 } 831 832 /* 833 * Called if lex found a leading L\' 834 */ 835 int 836 lex_wide_character_constant(void) 837 { 838 static char buf[MB_LEN_MAX + 1]; 839 size_t n, nmax; 840 int c; 841 wchar_t wc; 842 843 nmax = MB_CUR_MAX; 844 845 n = 0; 846 while ((c = get_escaped_char('\'')) >= 0) { 847 if (n < nmax) 848 buf[n] = (char)c; 849 n++; 850 } 851 852 wc = 0; 853 854 if (c == -2) { 855 /* unterminated character constant */ 856 error(253); 857 } else if (n == 0) { 858 /* empty character constant */ 859 error(73); 860 } else if (n > nmax) { 861 n = nmax; 862 /* too many characters in character constant */ 863 error(71); 864 } else { 865 buf[n] = '\0'; 866 (void)mbtowc(NULL, NULL, 0); 867 if (mbtowc(&wc, buf, nmax) < 0) 868 /* invalid multibyte character */ 869 error(291); 870 } 871 872 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 873 yylval.y_val->v_tspec = WCHAR; 874 yylval.y_val->v_quad = wc; 875 876 return T_CON; 877 } 878 879 /* 880 * Read a character which is part of a character constant or of a string 881 * and handle escapes. 882 * 883 * The argument is the character which delimits the character constant or 884 * string. 885 * 886 * Returns -1 if the end of the character constant or string is reached, 887 * -2 if the EOF is reached, and the character otherwise. 888 */ 889 static int 890 get_escaped_char(int delim) 891 { 892 static int pbc = -1; 893 int n, c, v; 894 895 if (pbc == -1) { 896 c = inpc(); 897 } else { 898 c = pbc; 899 pbc = -1; 900 } 901 if (c == delim) 902 return -1; 903 switch (c) { 904 case '\n': 905 if (tflag) { 906 /* newline in string or char constant */ 907 error(254); 908 return -2; 909 } 910 return c; 911 case 0: 912 /* syntax error '%s' */ 913 error(249, "EOF or null byte in literal"); 914 return -2; 915 case EOF: 916 return -2; 917 case '\\': 918 switch (c = inpc()) { 919 case '"': 920 if (tflag && delim == '\'') 921 /* \" inside character constants undef... */ 922 warning(262); 923 return '"'; 924 case '\'': 925 return '\''; 926 case '?': 927 if (tflag) 928 /* \? undefined in traditional C */ 929 warning(263); 930 return '?'; 931 case '\\': 932 return '\\'; 933 case 'a': 934 if (tflag) 935 /* \a undefined in traditional C */ 936 warning(81); 937 return '\a'; 938 case 'b': 939 return '\b'; 940 case 'f': 941 return '\f'; 942 case 'n': 943 return '\n'; 944 case 'r': 945 return '\r'; 946 case 't': 947 return '\t'; 948 case 'v': 949 if (tflag) 950 /* \v undefined in traditional C */ 951 warning(264); 952 return '\v'; 953 case '8': case '9': 954 /* bad octal digit %c */ 955 warning(77, c); 956 /* FALLTHROUGH */ 957 case '0': case '1': case '2': case '3': 958 case '4': case '5': case '6': case '7': 959 n = 3; 960 v = 0; 961 do { 962 v = (v << 3) + (c - '0'); 963 c = inpc(); 964 } while (--n > 0 && '0' <= c && c <= '7'); 965 pbc = c; 966 if (v > TARG_UCHAR_MAX) { 967 /* character escape does not fit in character */ 968 warning(76); 969 v &= CHAR_MASK; 970 } 971 return v; 972 case 'x': 973 if (tflag) 974 /* \x undefined in traditional C */ 975 warning(82); 976 v = 0; 977 n = 0; 978 while ((c = inpc()) >= 0 && isxdigit(c)) { 979 c = isdigit(c) ? 980 c - '0' : toupper(c) - 'A' + 10; 981 v = (v << 4) + c; 982 if (n >= 0) { 983 if ((v & ~CHAR_MASK) != 0) { 984 /* overflow in hex escape */ 985 warning(75); 986 n = -1; 987 } else { 988 n++; 989 } 990 } 991 } 992 pbc = c; 993 if (n == 0) { 994 /* no hex digits follow \x */ 995 error(74); 996 } if (n == -1) { 997 v &= CHAR_MASK; 998 } 999 return v; 1000 case '\n': 1001 return get_escaped_char(delim); 1002 case EOF: 1003 return -2; 1004 default: 1005 if (isprint(c)) { 1006 /* dubious escape \%c */ 1007 warning(79, c); 1008 } else { 1009 /* dubious escape \%o */ 1010 warning(80, c); 1011 } 1012 } 1013 } 1014 return c; 1015 } 1016 1017 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */ 1018 static void 1019 parse_line_directive_flags(const char *p, 1020 bool *is_begin, bool *is_end, bool *is_system) 1021 { 1022 1023 *is_begin = false; 1024 *is_end = false; 1025 *is_system = false; 1026 1027 while (*p != '\0') { 1028 while (ch_isspace(*p)) 1029 p++; 1030 1031 const char *word_start = p; 1032 while (*p != '\0' && !ch_isspace(*p)) 1033 p++; 1034 const char *word_end = p; 1035 1036 if (word_end - word_start == 1 && word_start[0] == '1') 1037 *is_begin = true; 1038 if (word_end - word_start == 1 && word_start[0] == '2') 1039 *is_end = true; 1040 if (word_end - word_start == 1 && word_start[0] == '3') 1041 *is_system = true; 1042 /* Flag '4' would only be interesting if lint handled C++. */ 1043 } 1044 1045 #if 0 1046 if (*p != '\0') { 1047 /* syntax error '%s' */ 1048 warning(249, "extra character(s) after directive"); 1049 } 1050 #endif 1051 } 1052 1053 /* 1054 * Called for preprocessor directives. Currently implemented are: 1055 * # lineno 1056 * # lineno "filename" 1057 * # lineno "filename" GCC-flag... 1058 */ 1059 void 1060 lex_directive(const char *yytext) 1061 { 1062 const char *cp, *fn; 1063 char c, *eptr; 1064 size_t fnl; 1065 long ln; 1066 bool is_begin, is_end, is_system; 1067 1068 static bool first = true; 1069 1070 /* Go to first non-whitespace after # */ 1071 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1072 continue; 1073 1074 if (!ch_isdigit(c)) { 1075 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6])) 1076 return; 1077 error: 1078 /* undefined or invalid # directive */ 1079 warning(255); 1080 return; 1081 } 1082 ln = strtol(--cp, &eptr, 10); 1083 if (cp == eptr) 1084 goto error; 1085 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1086 goto error; 1087 while ((c = *cp++) == ' ' || c == '\t') 1088 continue; 1089 if (c != '\0') { 1090 if (c != '"') 1091 goto error; 1092 fn = cp; 1093 while ((c = *cp) != '"' && c != '\0') 1094 cp++; 1095 if (c != '"') 1096 goto error; 1097 if ((fnl = cp++ - fn) > PATH_MAX) 1098 goto error; 1099 /* empty string means stdin */ 1100 if (fnl == 0) { 1101 fn = "{standard input}"; 1102 fnl = 16; /* strlen (fn) */ 1103 } 1104 curr_pos.p_file = record_filename(fn, fnl); 1105 /* 1106 * If this is the first directive, the name is the name 1107 * of the C source file as specified at the command line. 1108 * It is written to the output file. 1109 */ 1110 if (first) { 1111 csrc_pos.p_file = curr_pos.p_file; 1112 outsrc(transform_filename(curr_pos.p_file, 1113 strlen(curr_pos.p_file))); 1114 first = false; 1115 } 1116 1117 parse_line_directive_flags(cp, &is_begin, &is_end, &is_system); 1118 update_location(curr_pos.p_file, (int)ln, is_begin, is_end); 1119 in_system_header = is_system; 1120 } 1121 curr_pos.p_line = (int)ln - 1; 1122 curr_pos.p_uniq = 0; 1123 if (curr_pos.p_file == csrc_pos.p_file) { 1124 csrc_pos.p_line = (int)ln - 1; 1125 csrc_pos.p_uniq = 0; 1126 } 1127 } 1128 1129 /* 1130 * Handle lint comments such as ARGSUSED. 1131 * 1132 * If one of these comments is recognized, the argument, if any, is 1133 * parsed and a function which handles this comment is called. 1134 */ 1135 void 1136 lex_comment(void) 1137 { 1138 int c, lc; 1139 static const struct { 1140 const char *keywd; 1141 bool arg; 1142 void (*func)(int); 1143 } keywtab[] = { 1144 { "ARGSUSED", true, argsused }, 1145 { "BITFIELDTYPE", false, bitfieldtype }, 1146 { "CONSTCOND", false, constcond }, 1147 { "CONSTANTCOND", false, constcond }, 1148 { "CONSTANTCONDITION", false, constcond }, 1149 { "FALLTHRU", false, fallthru }, 1150 { "FALLTHROUGH", false, fallthru }, 1151 { "LINTLIBRARY", false, lintlib }, 1152 { "LINTED", true, linted }, 1153 { "LONGLONG", false, longlong }, 1154 { "NOSTRICT", true, linted }, 1155 { "NOTREACHED", false, not_reached }, 1156 { "PRINTFLIKE", true, printflike }, 1157 { "PROTOLIB", true, protolib }, 1158 { "SCANFLIKE", true, scanflike }, 1159 { "VARARGS", true, varargs }, 1160 }; 1161 char keywd[32]; 1162 char arg[32]; 1163 size_t l, i; 1164 int a; 1165 bool eoc; 1166 1167 eoc = false; 1168 1169 /* Skip whitespace after the start of the comment */ 1170 while ((c = inpc()) != EOF && isspace(c)) 1171 continue; 1172 1173 /* Read the potential keyword to keywd */ 1174 l = 0; 1175 while (c != EOF && isupper(c) && l < sizeof(keywd) - 1) { 1176 keywd[l++] = (char)c; 1177 c = inpc(); 1178 } 1179 keywd[l] = '\0'; 1180 1181 /* look for the keyword */ 1182 for (i = 0; i < sizeof(keywtab) / sizeof(keywtab[0]); i++) { 1183 if (strcmp(keywtab[i].keywd, keywd) == 0) 1184 break; 1185 } 1186 if (i == sizeof(keywtab) / sizeof(keywtab[0])) 1187 goto skip_rest; 1188 1189 /* skip whitespace after the keyword */ 1190 while (c != EOF && isspace(c)) 1191 c = inpc(); 1192 1193 /* read the argument, if the keyword accepts one and there is one */ 1194 l = 0; 1195 if (keywtab[i].arg) { 1196 while (c != EOF && isdigit(c) && l < sizeof(arg) - 1) { 1197 arg[l++] = (char)c; 1198 c = inpc(); 1199 } 1200 } 1201 arg[l] = '\0'; 1202 a = l != 0 ? atoi(arg) : -1; 1203 1204 /* skip whitespace after the argument */ 1205 while (c != EOF && isspace(c)) 1206 c = inpc(); 1207 1208 if (c != '*' || (c = inpc()) != '/') { 1209 if (keywtab[i].func != linted) 1210 /* extra characters in lint comment */ 1211 warning(257); 1212 } else { 1213 /* 1214 * remember that we have already found the end of the 1215 * comment 1216 */ 1217 eoc = true; 1218 } 1219 1220 if (keywtab[i].func != NULL) 1221 (*keywtab[i].func)(a); 1222 1223 skip_rest: 1224 while (!eoc) { 1225 lc = c; 1226 if ((c = inpc()) == EOF) { 1227 /* unterminated comment */ 1228 error(256); 1229 break; 1230 } 1231 if (lc == '*' && c == '/') 1232 eoc = true; 1233 } 1234 } 1235 1236 /* 1237 * Handle // style comments 1238 */ 1239 void 1240 lex_slash_slash_comment(void) 1241 { 1242 int c; 1243 1244 if (!Sflag && !gflag) 1245 /* %s C does not support // comments */ 1246 gnuism(312, tflag ? "traditional" : "ANSI"); 1247 1248 while ((c = inpc()) != EOF && c != '\n') 1249 continue; 1250 } 1251 1252 /* 1253 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1254 * clear_warn_flags() is called after function definitions and global and 1255 * local declarations and definitions. It is also called between 1256 * the controlling expression and the body of control statements 1257 * (if, switch, for, while). 1258 */ 1259 void 1260 clear_warn_flags(void) 1261 { 1262 1263 lwarn = LWARN_ALL; 1264 quadflg = false; 1265 constcond_flag = false; 1266 } 1267 1268 /* 1269 * Strings are stored in a dynamically allocated buffer and passed 1270 * in yylval.y_xstrg to the parser. The parser or the routines called 1271 * by the parser are responsible for freeing this buffer. 1272 */ 1273 int 1274 lex_string(void) 1275 { 1276 unsigned char *s; 1277 int c; 1278 size_t len, max; 1279 strg_t *strg; 1280 1281 s = xmalloc(max = 64); 1282 1283 len = 0; 1284 while ((c = get_escaped_char('"')) >= 0) { 1285 /* +1 to reserve space for a trailing NUL character */ 1286 if (len + 1 == max) 1287 s = xrealloc(s, max *= 2); 1288 s[len++] = (char)c; 1289 } 1290 s[len] = '\0'; 1291 if (c == -2) 1292 /* unterminated string constant */ 1293 error(258); 1294 1295 strg = xcalloc(1, sizeof(*strg)); 1296 strg->st_tspec = CHAR; 1297 strg->st_len = len; 1298 strg->st_cp = s; 1299 1300 yylval.y_string = strg; 1301 return T_STRING; 1302 } 1303 1304 int 1305 lex_wide_string(void) 1306 { 1307 char *s; 1308 int c, n; 1309 size_t i, wi; 1310 size_t len, max, wlen; 1311 wchar_t *ws; 1312 strg_t *strg; 1313 1314 s = xmalloc(max = 64); 1315 len = 0; 1316 while ((c = get_escaped_char('"')) >= 0) { 1317 /* +1 to save space for a trailing NUL character */ 1318 if (len + 1 >= max) 1319 s = xrealloc(s, max *= 2); 1320 s[len++] = (char)c; 1321 } 1322 s[len] = '\0'; 1323 if (c == -2) 1324 /* unterminated string constant */ 1325 error(258); 1326 1327 /* get length of wide-character string */ 1328 (void)mblen(NULL, 0); 1329 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1330 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1331 /* invalid multibyte character */ 1332 error(291); 1333 break; 1334 } 1335 if (n == 0) 1336 n = 1; 1337 } 1338 1339 ws = xmalloc((wlen + 1) * sizeof(*ws)); 1340 1341 /* convert from multibyte to wide char */ 1342 (void)mbtowc(NULL, NULL, 0); 1343 for (i = 0, wi = 0; i < len; i += n, wi++) { 1344 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1345 break; 1346 if (n == 0) 1347 n = 1; 1348 } 1349 ws[wi] = 0; 1350 free(s); 1351 1352 strg = xcalloc(1, sizeof(*strg)); 1353 strg->st_tspec = WCHAR; 1354 strg->st_len = wlen; 1355 strg->st_wcp = ws; 1356 1357 yylval.y_string = strg; 1358 return T_STRING; 1359 } 1360 1361 /* 1362 * As noted above, the scanner does not create new symbol table entries 1363 * for symbols it cannot find in the symbol table. This is to avoid 1364 * putting undeclared symbols into the symbol table if a syntax error 1365 * occurs. 1366 * 1367 * getsym() is called as soon as it is probably ok to put the symbol in the 1368 * symbol table. It is still possible that symbols are put in the symbol 1369 * table that are not completely declared due to syntax errors. To avoid too 1370 * many problems in this case, symbols get type 'int' in getsym(). 1371 * 1372 * XXX calls to getsym() should be delayed until decl1*() is called. 1373 */ 1374 sym_t * 1375 getsym(sbuf_t *sb) 1376 { 1377 dinfo_t *di; 1378 char *s; 1379 sym_t *sym; 1380 1381 sym = sb->sb_sym; 1382 1383 /* 1384 * During member declaration it is possible that name() looked 1385 * for symbols of type FVFT, although it should have looked for 1386 * symbols of type FTAG. Same can happen for labels. Both cases 1387 * are compensated here. 1388 */ 1389 if (symtyp == FMEMBER || symtyp == FLABEL) { 1390 if (sym == NULL || sym->s_kind == FVFT) 1391 sym = search(sb); 1392 } 1393 1394 if (sym != NULL) { 1395 if (sym->s_kind != symtyp) 1396 INTERNAL_ERROR("getsym(%d, %d)", sym->s_kind, symtyp); 1397 symtyp = FVFT; 1398 freesb(sb); 1399 return sym; 1400 } 1401 1402 /* create a new symbol table entry */ 1403 1404 /* labels must always be allocated at level 1 (outermost block) */ 1405 if (symtyp == FLABEL) { 1406 sym = getlblk(1, sizeof(*sym)); 1407 s = getlblk(1, sb->sb_len + 1); 1408 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1409 sym->s_name = s; 1410 sym->s_block_level = 1; 1411 di = dcs; 1412 while (di->d_next != NULL && di->d_next->d_next != NULL) 1413 di = di->d_next; 1414 lint_assert(di->d_ctx == AUTO); 1415 } else { 1416 sym = getblk(sizeof(*sym)); 1417 sym->s_name = sb->sb_name; 1418 sym->s_block_level = block_level; 1419 di = dcs; 1420 } 1421 1422 UNIQUE_CURR_POS(sym->s_def_pos); 1423 if ((sym->s_kind = symtyp) != FLABEL) 1424 sym->s_type = gettyp(INT); 1425 1426 symtyp = FVFT; 1427 1428 symtab_add(sym); 1429 1430 *di->d_ldlsym = sym; 1431 di->d_ldlsym = &sym->s_dlnxt; 1432 1433 freesb(sb); 1434 return sym; 1435 } 1436 1437 /* 1438 * Construct a temporary symbol. The symbol name starts with a digit, making 1439 * the name illegal. 1440 */ 1441 sym_t * 1442 mktempsym(type_t *t) 1443 { 1444 static int n = 0; 1445 char *s = getlblk(block_level, 64); 1446 sym_t *sym = getblk(sizeof(*sym)); 1447 scl_t scl; 1448 1449 (void)snprintf(s, 64, "%.8d_tmp", n++); 1450 1451 scl = dcs->d_scl; 1452 if (scl == NOSCL) 1453 scl = block_level > 0 ? AUTO : EXTERN; 1454 1455 sym->s_name = s; 1456 sym->s_type = t; 1457 sym->s_block_level = block_level; 1458 sym->s_scl = scl; 1459 sym->s_kind = FVFT; 1460 sym->s_used = true; 1461 sym->s_set = true; 1462 1463 symtab_add(sym); 1464 1465 *dcs->d_ldlsym = sym; 1466 dcs->d_ldlsym = &sym->s_dlnxt; 1467 1468 return sym; 1469 } 1470 1471 /* Remove a symbol forever from the symbol table. */ 1472 void 1473 rmsym(sym_t *sym) 1474 { 1475 1476 debug_step("rmsym '%s' %d '%s'", 1477 sym->s_name, (int)sym->s_kind, type_name(sym->s_type)); 1478 symtab_remove(sym); 1479 1480 /* avoid that the symbol will later be put back to the symbol table */ 1481 sym->s_block_level = -1; 1482 } 1483 1484 /* 1485 * Remove a list of symbols declared at one level from the symbol 1486 * table. 1487 */ 1488 void 1489 rmsyms(sym_t *syms) 1490 { 1491 sym_t *sym; 1492 1493 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1494 if (sym->s_block_level != -1) { 1495 debug_step("rmsyms '%s' %d '%s'", 1496 sym->s_name, (int)sym->s_kind, 1497 type_name(sym->s_type)); 1498 symtab_remove(sym); 1499 sym->s_rlink = NULL; 1500 } 1501 } 1502 } 1503 1504 /* 1505 * Put a symbol into the symbol table. 1506 */ 1507 void 1508 inssym(int bl, sym_t *sym) 1509 { 1510 1511 debug_step("inssym '%s' %d '%s'", 1512 sym->s_name, sym->s_kind, type_name(sym->s_type)); 1513 symtab_add(sym); 1514 sym->s_block_level = bl; 1515 lint_assert(sym->s_link == NULL || 1516 sym->s_block_level >= sym->s_link->s_block_level); 1517 } 1518 1519 /* 1520 * Called at level 0 after syntax errors. 1521 * 1522 * Removes all symbols which are not declared at level 0 from the 1523 * symbol table. Also frees all memory which is not associated with 1524 * level 0. 1525 */ 1526 void 1527 cleanup(void) 1528 { 1529 sym_t *sym, *nsym; 1530 int i; 1531 1532 for (i = 0; i < HSHSIZ1; i++) { 1533 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1534 nsym = sym->s_link; 1535 if (sym->s_block_level >= 1) 1536 symtab_remove(sym); 1537 } 1538 } 1539 1540 for (i = mem_block_level; i > 0; i--) 1541 freelblk(i); 1542 } 1543 1544 /* 1545 * Create a new symbol with the name of an existing symbol. 1546 */ 1547 sym_t * 1548 pushdown(const sym_t *sym) 1549 { 1550 sym_t *nsym; 1551 1552 debug_step("pushdown '%s' %d '%s'", 1553 sym->s_name, (int)sym->s_kind, type_name(sym->s_type)); 1554 nsym = getblk(sizeof(*nsym)); 1555 lint_assert(sym->s_block_level <= block_level); 1556 nsym->s_name = sym->s_name; 1557 UNIQUE_CURR_POS(nsym->s_def_pos); 1558 nsym->s_kind = sym->s_kind; 1559 nsym->s_block_level = block_level; 1560 1561 symtab_add(nsym); 1562 1563 *dcs->d_ldlsym = nsym; 1564 dcs->d_ldlsym = &nsym->s_dlnxt; 1565 1566 return nsym; 1567 } 1568 1569 /* 1570 * Free any dynamically allocated memory referenced by 1571 * the value stack or yylval. 1572 * The type of information in yylval is described by tok. 1573 */ 1574 void 1575 freeyyv(void *sp, int tok) 1576 { 1577 if (tok == T_NAME || tok == T_TYPENAME) { 1578 sbuf_t *sb = *(sbuf_t **)sp; 1579 freesb(sb); 1580 } else if (tok == T_CON) { 1581 val_t *val = *(val_t **)sp; 1582 free(val); 1583 } else if (tok == T_STRING) { 1584 strg_t *strg = *(strg_t **)sp; 1585 if (strg->st_tspec == CHAR) { 1586 free(strg->st_cp); 1587 } else { 1588 lint_assert(strg->st_tspec == WCHAR); 1589 free(strg->st_wcp); 1590 } 1591 free(strg); 1592 } 1593 } 1594
Indexes created Sat Sep 27 14:09:57 GMT 2025