lex.c revision 1.103
1 /* $NetBSD: lex.c,v 1.103 2022/02/27 18:29:14 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 1.103 2022/02/27 18:29:14 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 bool in_gcc_attribute; 68 bool in_system_header; 69 70 static int inpc(void); 71 static unsigned int hash(const char *); 72 static sym_t * search(sbuf_t *); 73 static int keyw(sym_t *); 74 static int get_escaped_char(int); 75 76 void 77 lex_next_line(void) 78 { 79 curr_pos.p_line++; 80 curr_pos.p_uniq = 0; 81 debug_step("parsing %s:%d", curr_pos.p_file, curr_pos.p_line); 82 if (curr_pos.p_file == csrc_pos.p_file) { 83 csrc_pos.p_line++; 84 csrc_pos.p_uniq = 0; 85 } 86 } 87 88 void 89 lex_unknown_character(int c) 90 { 91 92 /* unknown character \%o */ 93 error(250, c); 94 } 95 96 #define kwdef(name, token, scl, tspec, tqual, c90, c99, gcc, attr, deco) \ 97 { \ 98 name, token, scl, tspec, tqual, \ 99 (c90) > 0, (c99) > 0, (gcc) > 0, (attr) > 0, \ 100 ((deco) & 1) != 0, ((deco) & 2) != 0, ((deco) & 4) != 0, \ 101 } 102 #define kwdef_token(name, token, c90, c99, gcc, attr, deco) \ 103 kwdef(name, token, 0, 0, 0, c90, c99, gcc, attr, deco) 104 #define kwdef_sclass(name, sclass, c90, c99, gcc, attr, deco) \ 105 kwdef(name, T_SCLASS, sclass, 0, 0, c90, c99, gcc, attr, deco) 106 #define kwdef_type(name, tspec, c90, c99, gcc, attr, deco) \ 107 kwdef(name, T_TYPE, 0, tspec, 0, c90, c99, gcc, attr, deco) 108 #define kwdef_tqual(name, tqual, c90, c99, gcc, attr, deco) \ 109 kwdef(name, T_QUAL, 0, 0, tqual, c90, c99, gcc, attr, deco) 110 #define kwdef_keyword(name, token) \ 111 kwdef(name, token, 0, 0, 0, 0, 0, 0, 0, 1) 112 #define kwdef_gcc_attr(name, token) \ 113 kwdef(name, token, 0, 0, 0, 0, 0, 1, 1, 5) 114 115 /* During initialization, these keywords are written to the symbol table. */ 116 static struct keyword { 117 const char *kw_name; /* keyword */ 118 int kw_token; /* token returned by yylex() */ 119 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 120 tspec_t kw_tspec; /* type spec. if kw_token 121 * T_TYPE or T_STRUCT_OR_UNION */ 122 tqual_t kw_tqual; /* type qual. if kw_token T_QUAL */ 123 bool kw_c90:1; /* C90 keyword */ 124 bool kw_c99:1; /* C99 keyword */ 125 bool kw_gcc:1; /* GCC keyword */ 126 bool kw_attr:1; /* GCC attribute, keyword */ 127 bool kw_plain:1; /* 'name' */ 128 bool kw_leading:1; /* '__name' */ 129 bool kw_both:1; /* '__name__' */ 130 } keywords[] = { 131 kwdef_gcc_attr( "alias", T_AT_ALIAS), 132 kwdef_keyword( "_Alignas", T_ALIGNAS), 133 kwdef_keyword( "_Alignof", T_ALIGNOF), 134 kwdef_gcc_attr( "aligned", T_AT_ALIGNED), 135 kwdef_token( "__alignof__", T_ALIGNOF, 0,0,0,0,1), 136 kwdef_gcc_attr( "alloc_size", T_AT_ALLOC_SIZE), 137 kwdef_gcc_attr( "always_inline",T_AT_ALWAYS_INLINE), 138 kwdef_token( "asm", T_ASM, 0,0,1,0,7), 139 kwdef_token( "attribute", T_ATTRIBUTE, 0,0,1,0,6), 140 kwdef_sclass( "auto", AUTO, 0,0,0,0,1), 141 kwdef_type( "_Bool", BOOL, 0,1,0,0,1), 142 kwdef_gcc_attr( "bounded", T_AT_BOUNDED), 143 kwdef_keyword( "break", T_BREAK), 144 kwdef_gcc_attr( "buffer", T_AT_BUFFER), 145 kwdef_token( "__builtin_offsetof", T_BUILTIN_OFFSETOF, 0,0,1,0,1), 146 kwdef_keyword( "case", T_CASE), 147 kwdef_type( "char", CHAR, 0,0,0,0,1), 148 kwdef_gcc_attr( "cold", T_AT_COLD), 149 kwdef_gcc_attr( "common", T_AT_COMMON), 150 kwdef_type( "_Complex", COMPLEX, 0,1,0,0,1), 151 kwdef_tqual( "const", CONST, 1,0,0,0,7), 152 kwdef_gcc_attr( "constructor", T_AT_CONSTRUCTOR), 153 kwdef_keyword( "continue", T_CONTINUE), 154 kwdef_keyword( "default", T_DEFAULT), 155 kwdef_gcc_attr( "deprecated", T_AT_DEPRECATED), 156 kwdef_gcc_attr( "destructor", T_AT_DESTRUCTOR), 157 kwdef_gcc_attr( "disable_sanitizer_instrumentation", 158 T_AT_DISABLE_SANITIZER_INSTRUMENTATION), 159 kwdef_keyword( "do", T_DO), 160 kwdef_type( "double", DOUBLE, 0,0,0,0,1), 161 kwdef_keyword( "else", T_ELSE), 162 kwdef_keyword( "enum", T_ENUM), 163 kwdef_token( "__extension__",T_EXTENSION, 0,0,1,0,1), 164 kwdef_sclass( "extern", EXTERN, 0,0,0,0,1), 165 kwdef_gcc_attr( "fallthrough", T_AT_FALLTHROUGH), 166 kwdef_type( "float", FLOAT, 0,0,0,0,1), 167 kwdef_keyword( "for", T_FOR), 168 kwdef_gcc_attr( "format", T_AT_FORMAT), 169 kwdef_gcc_attr( "format_arg", T_AT_FORMAT_ARG), 170 kwdef_token( "_Generic", T_GENERIC, 0,1,0,0,1), 171 kwdef_gcc_attr( "gnu_inline", T_AT_GNU_INLINE), 172 kwdef_gcc_attr( "gnu_printf", T_AT_FORMAT_GNU_PRINTF), 173 kwdef_keyword( "goto", T_GOTO), 174 kwdef_gcc_attr( "hot", T_AT_HOT), 175 kwdef_keyword( "if", T_IF), 176 kwdef_token( "__imag__", T_IMAG, 0,0,1,0,1), 177 kwdef_sclass( "inline", INLINE, 0,1,0,0,7), 178 kwdef_type( "int", INT, 0,0,0,0,1), 179 #ifdef INT128_SIZE 180 kwdef_type( "__int128_t", INT128, 0,1,0,0,1), 181 #endif 182 kwdef_type( "long", LONG, 0,0,0,0,1), 183 kwdef_gcc_attr( "malloc", T_AT_MALLOC), 184 kwdef_gcc_attr( "may_alias", T_AT_MAY_ALIAS), 185 kwdef_gcc_attr( "minbytes", T_AT_MINBYTES), 186 kwdef_gcc_attr( "mode", T_AT_MODE), 187 kwdef_gcc_attr("no_instrument_function", 188 T_AT_NO_INSTRUMENT_FUNCTION), 189 kwdef_gcc_attr( "no_sanitize", T_AT_NO_SANITIZE), 190 kwdef_gcc_attr( "no_sanitize_thread", T_AT_NO_SANITIZE_THREAD), 191 kwdef_gcc_attr( "noinline", T_AT_NOINLINE), 192 kwdef_gcc_attr( "nonnull", T_AT_NONNULL), 193 kwdef_gcc_attr( "nonstring", T_AT_NONSTRING), 194 kwdef_token( "_Noreturn", T_NORETURN, 0,1,0,0,1), 195 kwdef_gcc_attr( "noreturn", T_AT_NORETURN), 196 kwdef_gcc_attr( "nothrow", T_AT_NOTHROW), 197 kwdef_gcc_attr( "optimize", T_AT_OPTIMIZE), 198 kwdef_gcc_attr( "optnone", T_AT_OPTNONE), 199 kwdef_gcc_attr( "packed", T_AT_PACKED), 200 kwdef_token( "__packed", T_PACKED, 0,0,0,0,1), 201 kwdef_gcc_attr( "pcs", T_AT_PCS), 202 kwdef_gcc_attr( "printf", T_AT_FORMAT_PRINTF), 203 kwdef_gcc_attr( "pure", T_AT_PURE), 204 kwdef_token( "__real__", T_REAL, 0,0,1,0,1), 205 kwdef_sclass( "register", REG, 0,0,0,0,1), 206 kwdef_gcc_attr( "regparm", T_AT_REGPARM), 207 kwdef_tqual( "restrict", RESTRICT, 0,1,0,0,7), 208 kwdef_keyword( "return", T_RETURN), 209 kwdef_gcc_attr( "returns_nonnull",T_AT_RETURNS_NONNULL), 210 kwdef_gcc_attr( "returns_twice",T_AT_RETURNS_TWICE), 211 kwdef_gcc_attr( "scanf", T_AT_FORMAT_SCANF), 212 kwdef_token( "section", T_AT_SECTION, 0,0,1,1,7), 213 kwdef_gcc_attr( "sentinel", T_AT_SENTINEL), 214 kwdef_type( "short", SHORT, 0,0,0,0,1), 215 kwdef_type( "signed", SIGNED, 1,0,0,0,3), 216 kwdef_keyword( "sizeof", T_SIZEOF), 217 kwdef_sclass( "static", STATIC, 0,0,0,0,1), 218 kwdef_keyword( "_Static_assert", T_STATIC_ASSERT), 219 kwdef_gcc_attr( "strfmon", T_AT_FORMAT_STRFMON), 220 kwdef_gcc_attr( "strftime", T_AT_FORMAT_STRFTIME), 221 kwdef_gcc_attr( "string", T_AT_STRING), 222 kwdef("struct", T_STRUCT_OR_UNION, 0, STRUCT, 0, 0,0,0,0,1), 223 kwdef_keyword( "switch", T_SWITCH), 224 kwdef_token( "__symbolrename", T_SYMBOLRENAME, 0,0,0,0,1), 225 kwdef_gcc_attr( "syslog", T_AT_FORMAT_SYSLOG), 226 kwdef_gcc_attr( "target", T_AT_TARGET), 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 /* type of next expected symbol */ 260 symt_t symtyp; 261 262 263 static void 264 symtab_add(sym_t *sym) 265 { 266 size_t h; 267 268 h = hash(sym->s_name); 269 if ((sym->s_symtab_next = symtab[h]) != NULL) 270 symtab[h]->s_symtab_ref = &sym->s_symtab_next; 271 sym->s_symtab_ref = &symtab[h]; 272 symtab[h] = sym; 273 } 274 275 static void 276 symtab_remove(sym_t *sym) 277 { 278 279 if ((*sym->s_symtab_ref = sym->s_symtab_next) != NULL) 280 sym->s_symtab_next->s_symtab_ref = sym->s_symtab_ref; 281 sym->s_symtab_next = NULL; 282 } 283 284 285 static void 286 add_keyword(const struct keyword *kw, bool leading, bool trailing) 287 { 288 sym_t *sym; 289 char buf[256]; 290 const char *name; 291 292 if (!leading && !trailing) { 293 name = kw->kw_name; 294 } else { 295 (void)snprintf(buf, sizeof(buf), "%s%s%s", 296 leading ? "__" : "", kw->kw_name, trailing ? "__" : ""); 297 name = xstrdup(buf); 298 } 299 300 sym = block_zero_alloc(sizeof(*sym)); 301 sym->s_name = name; 302 sym->s_keyword = kw; 303 sym->s_value.v_quad = kw->kw_token; 304 if (kw->kw_token == T_TYPE || kw->kw_token == T_STRUCT_OR_UNION) { 305 sym->s_tspec = kw->kw_tspec; 306 } else if (kw->kw_token == T_SCLASS) { 307 sym->s_scl = kw->kw_scl; 308 } else if (kw->kw_token == T_QUAL) { 309 sym->s_tqual = kw->kw_tqual; 310 } 311 312 symtab_add(sym); 313 } 314 315 /* 316 * All keywords are written to the symbol table. This saves us looking 317 * in an extra table for each name we found. 318 */ 319 void 320 initscan(void) 321 { 322 struct keyword *kw; 323 324 for (kw = keywords; kw->kw_name != NULL; kw++) { 325 if ((kw->kw_c90 || kw->kw_c99) && tflag) 326 continue; 327 if (kw->kw_c99 && !(Sflag || gflag)) 328 continue; 329 if (kw->kw_gcc && !gflag) 330 continue; 331 if (kw->kw_plain) 332 add_keyword(kw, false, false); 333 if (kw->kw_leading) 334 add_keyword(kw, true, false); 335 if (kw->kw_both) 336 add_keyword(kw, true, true); 337 } 338 } 339 340 /* 341 * Read a character and ensure that it is positive (except EOF). 342 * Increment line count(s) if necessary. 343 */ 344 static int 345 inpc(void) 346 { 347 int c; 348 349 if ((c = lex_input()) == EOF) 350 return c; 351 c &= CHAR_MASK; 352 if (c == '\0') 353 return EOF; /* lex returns 0 on EOF. */ 354 if (c == '\n') 355 lex_next_line(); 356 return c; 357 } 358 359 static unsigned int 360 hash(const char *s) 361 { 362 unsigned int v; 363 const char *p; 364 365 v = 0; 366 for (p = s; *p != '\0'; p++) { 367 v = (v << 4) + (unsigned char)*p; 368 v ^= v >> 28; 369 } 370 return v % HSHSIZ1; 371 } 372 373 /* 374 * Lex has found a letter followed by zero or more letters or digits. 375 * It looks for a symbol in the symbol table with the same name. This 376 * symbol must either be a keyword or a symbol of the type required by 377 * symtyp (label, member, tag, ...). 378 * 379 * If it is a keyword, the token is returned. In some cases it is described 380 * more deeply by data written to yylval. 381 * 382 * If it is a symbol, T_NAME is returned and the name is stored in yylval. 383 * If there is already a symbol of the same name and type in the symbol 384 * table, yylval.y_name->sb_sym points there. 385 */ 386 extern int 387 lex_name(const char *yytext, size_t yyleng) 388 { 389 char *s; 390 sbuf_t *sb; 391 sym_t *sym; 392 int tok; 393 394 sb = xmalloc(sizeof(*sb)); 395 sb->sb_name = yytext; 396 sb->sb_len = yyleng; 397 if ((sym = search(sb)) != NULL && sym->s_keyword != NULL) { 398 free(sb); 399 return keyw(sym); 400 } 401 402 sb->sb_sym = sym; 403 404 if (sym != NULL) { 405 lint_assert(block_level >= sym->s_block_level); 406 sb->sb_name = sym->s_name; 407 sb->sb_len = strlen(sym->s_name); 408 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 409 } else { 410 s = block_zero_alloc(yyleng + 1); 411 (void)memcpy(s, yytext, yyleng + 1); 412 sb->sb_name = s; 413 sb->sb_len = yyleng; 414 tok = T_NAME; 415 } 416 417 yylval.y_name = sb; 418 return tok; 419 } 420 421 static sym_t * 422 search(sbuf_t *sb) 423 { 424 unsigned int h; 425 sym_t *sym; 426 const struct keyword *kw; 427 428 h = hash(sb->sb_name); 429 for (sym = symtab[h]; sym != NULL; sym = sym->s_symtab_next) { 430 if (strcmp(sym->s_name, sb->sb_name) != 0) 431 continue; 432 kw = sym->s_keyword; 433 434 if (kw != NULL && !kw->kw_attr) 435 return sym; 436 if (kw != NULL && in_gcc_attribute) 437 return sym; 438 if (kw == NULL && !in_gcc_attribute && sym->s_kind == symtyp) 439 return sym; 440 } 441 442 return NULL; 443 } 444 445 static int 446 keyw(sym_t *sym) 447 { 448 int t; 449 450 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 451 yylval.y_scl = sym->s_scl; 452 } else if (t == T_TYPE || t == T_STRUCT_OR_UNION) { 453 yylval.y_tspec = sym->s_tspec; 454 } else if (t == T_QUAL) { 455 yylval.y_tqual = sym->s_tqual; 456 } 457 return t; 458 } 459 460 /* 461 * Convert a string representing an integer into internal representation. 462 * Return T_CON, storing the numeric value in yylval, for yylex. 463 */ 464 int 465 lex_integer_constant(const char *yytext, size_t yyleng, int base) 466 { 467 int l_suffix, u_suffix; 468 size_t len; 469 const char *cp; 470 char c, *eptr; 471 tspec_t typ; 472 bool ansiu; 473 bool warned = false; 474 uint64_t uq = 0; 475 476 /* C11 6.4.4.1p5 */ 477 static const tspec_t suffix_type[2][3] = { 478 { INT, LONG, QUAD, }, 479 { UINT, ULONG, UQUAD, } 480 }; 481 482 cp = yytext; 483 len = yyleng; 484 485 /* skip 0[xX] or 0[bB] */ 486 if (base == 16 || base == 2) { 487 cp += 2; 488 len -= 2; 489 } 490 491 /* read suffixes */ 492 l_suffix = u_suffix = 0; 493 for (;;) { 494 if ((c = cp[len - 1]) == 'l' || c == 'L') { 495 l_suffix++; 496 } else if (c == 'u' || c == 'U') { 497 u_suffix++; 498 } else { 499 break; 500 } 501 len--; 502 } 503 if (l_suffix > 2 || u_suffix > 1) { 504 /* malformed integer constant */ 505 warning(251); 506 if (l_suffix > 2) 507 l_suffix = 2; 508 if (u_suffix > 1) 509 u_suffix = 1; 510 } 511 if (tflag && u_suffix != 0) { 512 /* suffix U is illegal in traditional C */ 513 warning(97); 514 } 515 typ = suffix_type[u_suffix][l_suffix]; 516 517 errno = 0; 518 519 uq = (uint64_t)strtoull(cp, &eptr, base); 520 lint_assert(eptr == cp + len); 521 if (errno != 0) { 522 /* integer constant out of range */ 523 warning(252); 524 warned = true; 525 } 526 527 /* 528 * If the value is too big for the current type, we must choose 529 * another type. 530 */ 531 ansiu = false; 532 switch (typ) { 533 case INT: 534 if (uq <= TARG_INT_MAX) { 535 /* ok */ 536 } else if (uq <= TARG_UINT_MAX && base != 10) { 537 typ = UINT; 538 } else if (uq <= TARG_LONG_MAX) { 539 typ = LONG; 540 } else { 541 typ = ULONG; 542 if (uq > TARG_ULONG_MAX && !warned) { 543 /* integer constant out of range */ 544 warning(252); 545 } 546 } 547 if (typ == UINT || typ == ULONG) { 548 if (tflag) { 549 typ = LONG; 550 } else if (!sflag) { 551 /* 552 * Remember that the constant is unsigned 553 * only in ANSI C 554 */ 555 ansiu = true; 556 } 557 } 558 break; 559 case UINT: 560 if (uq > TARG_UINT_MAX) { 561 typ = ULONG; 562 if (uq > TARG_ULONG_MAX && !warned) { 563 /* integer constant out of range */ 564 warning(252); 565 } 566 } 567 break; 568 case LONG: 569 if (uq > TARG_LONG_MAX && !tflag) { 570 typ = ULONG; 571 if (!sflag) 572 ansiu = true; 573 if (uq > TARG_ULONG_MAX && !warned) { 574 /* integer constant out of range */ 575 warning(252); 576 } 577 } 578 break; 579 case ULONG: 580 if (uq > TARG_ULONG_MAX && !warned) { 581 /* integer constant out of range */ 582 warning(252); 583 } 584 break; 585 case QUAD: 586 if (uq > TARG_QUAD_MAX && !tflag) { 587 typ = UQUAD; 588 if (!sflag) 589 ansiu = true; 590 } 591 break; 592 case UQUAD: 593 if (uq > TARG_UQUAD_MAX && !warned) { 594 /* integer constant out of range */ 595 warning(252); 596 } 597 break; 598 default: 599 break; 600 } 601 602 uq = (uint64_t)convert_integer((int64_t)uq, typ, 0); 603 604 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 605 yylval.y_val->v_tspec = typ; 606 yylval.y_val->v_unsigned_since_c90 = ansiu; 607 yylval.y_val->v_quad = (int64_t)uq; 608 609 return T_CON; 610 } 611 612 /* 613 * Extend or truncate q to match t. If t is signed, sign-extend. 614 * 615 * len is the number of significant bits. If len is -1, len is set 616 * to the width of type t. 617 */ 618 int64_t 619 convert_integer(int64_t q, tspec_t t, unsigned int len) 620 { 621 uint64_t vbits; 622 623 if (len == 0) 624 len = size_in_bits(t); 625 626 vbits = value_bits(len); 627 return t == PTR || is_uinteger(t) || ((q & bit(len - 1)) == 0) 628 ? (int64_t)(q & vbits) 629 : (int64_t)(q | ~vbits); 630 } 631 632 /* 633 * Convert a string representing a floating point value into its numerical 634 * representation. Type and value are returned in yylval. 635 * 636 * XXX Currently it is not possible to convert constants of type 637 * long double which are greater than DBL_MAX. 638 */ 639 int 640 lex_floating_constant(const char *yytext, size_t yyleng) 641 { 642 const char *cp; 643 size_t len; 644 tspec_t typ; 645 char c, *eptr; 646 double d; 647 float f = 0; 648 649 cp = yytext; 650 len = yyleng; 651 652 if (cp[len - 1] == 'i') 653 len--; /* imaginary, do nothing for now */ 654 655 if ((c = cp[len - 1]) == 'f' || c == 'F') { 656 typ = FLOAT; 657 len--; 658 } else if (c == 'l' || c == 'L') { 659 typ = LDOUBLE; 660 len--; 661 } else { 662 if (c == 'd' || c == 'D') 663 len--; 664 typ = DOUBLE; 665 } 666 667 if (tflag && typ != DOUBLE) { 668 /* suffixes F and L are illegal in traditional C */ 669 warning(98); 670 } 671 672 errno = 0; 673 d = strtod(cp, &eptr); 674 if (eptr != cp + len) { 675 switch (*eptr) { 676 /* 677 * XXX: non-native non-current strtod() may not handle hex 678 * floats, ignore the rest if we find traces of hex float 679 * syntax... 680 */ 681 case 'p': 682 case 'P': 683 case 'x': 684 case 'X': 685 d = 0; 686 errno = 0; 687 break; 688 default: 689 INTERNAL_ERROR("lex_floating_constant(%s->%s)", 690 cp, eptr); 691 } 692 } 693 if (errno != 0) 694 /* floating-point constant out of range */ 695 warning(248); 696 697 if (typ == FLOAT) { 698 f = (float)d; 699 if (isfinite(f) == 0) { 700 /* floating-point constant out of range */ 701 warning(248); 702 f = f > 0 ? FLT_MAX : -FLT_MAX; 703 } 704 } 705 706 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 707 yylval.y_val->v_tspec = typ; 708 if (typ == FLOAT) { 709 yylval.y_val->v_ldbl = f; 710 } else { 711 yylval.y_val->v_ldbl = d; 712 } 713 714 return T_CON; 715 } 716 717 int 718 lex_operator(int t, op_t o) 719 { 720 721 yylval.y_op = o; 722 return t; 723 } 724 725 /* 726 * Called if lex found a leading \'. 727 */ 728 int 729 lex_character_constant(void) 730 { 731 size_t n; 732 int val, c; 733 734 n = 0; 735 val = 0; 736 while ((c = get_escaped_char('\'')) >= 0) { 737 val = (val << CHAR_SIZE) + c; 738 n++; 739 } 740 if (c == -2) { 741 /* unterminated character constant */ 742 error(253); 743 } else if (n > sizeof(int) || (n > 1 && (pflag || hflag))) { 744 /* XXX: should rather be sizeof(TARG_INT) */ 745 746 /* too many characters in character constant */ 747 error(71); 748 } else if (n > 1) { 749 /* multi-character character constant */ 750 warning(294); 751 } else if (n == 0) { 752 /* empty character constant */ 753 error(73); 754 } 755 if (n == 1) 756 val = (int)convert_integer(val, CHAR, CHAR_SIZE); 757 758 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 759 yylval.y_val->v_tspec = INT; 760 yylval.y_val->v_quad = val; 761 762 return T_CON; 763 } 764 765 /* 766 * Called if lex found a leading L\' 767 */ 768 int 769 lex_wide_character_constant(void) 770 { 771 static char buf[MB_LEN_MAX + 1]; 772 size_t n, nmax; 773 int c; 774 wchar_t wc; 775 776 nmax = MB_CUR_MAX; 777 778 n = 0; 779 while ((c = get_escaped_char('\'')) >= 0) { 780 if (n < nmax) 781 buf[n] = (char)c; 782 n++; 783 } 784 785 wc = 0; 786 787 if (c == -2) { 788 /* unterminated character constant */ 789 error(253); 790 } else if (n == 0) { 791 /* empty character constant */ 792 error(73); 793 } else if (n > nmax) { 794 n = nmax; 795 /* too many characters in character constant */ 796 error(71); 797 } else { 798 buf[n] = '\0'; 799 (void)mbtowc(NULL, NULL, 0); 800 if (mbtowc(&wc, buf, nmax) < 0) 801 /* invalid multibyte character */ 802 error(291); 803 } 804 805 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 806 yylval.y_val->v_tspec = WCHAR; 807 yylval.y_val->v_quad = wc; 808 809 return T_CON; 810 } 811 812 /* 813 * Read a character which is part of a character constant or of a string 814 * and handle escapes. 815 * 816 * The argument is the character which delimits the character constant or 817 * string. 818 * 819 * Returns -1 if the end of the character constant or string is reached, 820 * -2 if the EOF is reached, and the character otherwise. 821 */ 822 static int 823 get_escaped_char(int delim) 824 { 825 static int pbc = -1; 826 int n, c, v; 827 828 if (pbc == -1) { 829 c = inpc(); 830 } else { 831 c = pbc; 832 pbc = -1; 833 } 834 if (c == delim) 835 return -1; 836 switch (c) { 837 case '\n': 838 if (tflag) { 839 /* newline in string or char constant */ 840 error(254); 841 return -2; 842 } 843 return c; 844 case 0: 845 /* syntax error '%s' */ 846 error(249, "EOF or null byte in literal"); 847 return -2; 848 case EOF: 849 return -2; 850 case '\\': 851 switch (c = inpc()) { 852 case '"': 853 if (tflag && delim == '\'') 854 /* \" inside character constants undef... */ 855 warning(262); 856 return '"'; 857 case '\'': 858 return '\''; 859 case '?': 860 if (tflag) 861 /* \? undefined in traditional C */ 862 warning(263); 863 return '?'; 864 case '\\': 865 return '\\'; 866 case 'a': 867 if (tflag) 868 /* \a undefined in traditional C */ 869 warning(81); 870 return '\a'; 871 case 'b': 872 return '\b'; 873 case 'f': 874 return '\f'; 875 case 'n': 876 return '\n'; 877 case 'r': 878 return '\r'; 879 case 't': 880 return '\t'; 881 case 'v': 882 if (tflag) 883 /* \v undefined in traditional C */ 884 warning(264); 885 return '\v'; 886 case '8': case '9': 887 /* bad octal digit %c */ 888 warning(77, c); 889 /* FALLTHROUGH */ 890 case '0': case '1': case '2': case '3': 891 case '4': case '5': case '6': case '7': 892 n = 3; 893 v = 0; 894 do { 895 v = (v << 3) + (c - '0'); 896 c = inpc(); 897 } while (--n > 0 && '0' <= c && c <= '7'); 898 pbc = c; 899 if (v > TARG_UCHAR_MAX) { 900 /* character escape does not fit in character */ 901 warning(76); 902 v &= CHAR_MASK; 903 } 904 return v; 905 case 'x': 906 if (tflag) 907 /* \x undefined in traditional C */ 908 warning(82); 909 v = 0; 910 n = 0; 911 while (c = inpc(), isxdigit(c)) { 912 c = isdigit(c) ? 913 c - '0' : toupper(c) - 'A' + 10; 914 v = (v << 4) + c; 915 if (n >= 0) { 916 if ((v & ~CHAR_MASK) != 0) { 917 /* overflow in hex escape */ 918 warning(75); 919 n = -1; 920 } else { 921 n++; 922 } 923 } 924 } 925 pbc = c; 926 if (n == 0) { 927 /* no hex digits follow \x */ 928 error(74); 929 } if (n == -1) { 930 v &= CHAR_MASK; 931 } 932 return v; 933 case '\n': 934 return get_escaped_char(delim); 935 case EOF: 936 return -2; 937 default: 938 if (isprint(c)) { 939 /* dubious escape \%c */ 940 warning(79, c); 941 } else { 942 /* dubious escape \%o */ 943 warning(80, c); 944 } 945 } 946 } 947 return c; 948 } 949 950 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */ 951 static void 952 parse_line_directive_flags(const char *p, 953 bool *is_begin, bool *is_end, bool *is_system) 954 { 955 956 *is_begin = false; 957 *is_end = false; 958 *is_system = false; 959 960 while (*p != '\0') { 961 const char *word_start, *word_end; 962 963 while (ch_isspace(*p)) 964 p++; 965 966 word_start = p; 967 while (*p != '\0' && !ch_isspace(*p)) 968 p++; 969 word_end = p; 970 971 if (word_end - word_start == 1 && word_start[0] == '1') 972 *is_begin = true; 973 if (word_end - word_start == 1 && word_start[0] == '2') 974 *is_end = true; 975 if (word_end - word_start == 1 && word_start[0] == '3') 976 *is_system = true; 977 /* Flag '4' is only interesting for C++. */ 978 } 979 } 980 981 /* 982 * Called for preprocessor directives. Currently implemented are: 983 * # pragma [argument...] 984 * # lineno 985 * # lineno "filename" 986 * # lineno "filename" GCC-flag... 987 */ 988 void 989 lex_directive(const char *yytext) 990 { 991 const char *cp, *fn; 992 char c, *eptr; 993 size_t fnl; 994 long ln; 995 bool is_begin, is_end, is_system; 996 997 static bool first = true; 998 999 /* Go to first non-whitespace after # */ 1000 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1001 continue; 1002 1003 if (!ch_isdigit(c)) { 1004 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6])) 1005 return; 1006 error: 1007 /* undefined or invalid # directive */ 1008 warning(255); 1009 return; 1010 } 1011 ln = strtol(--cp, &eptr, 10); 1012 if (eptr == cp) 1013 goto error; 1014 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1015 goto error; 1016 while ((c = *cp++) == ' ' || c == '\t') 1017 continue; 1018 if (c != '\0') { 1019 if (c != '"') 1020 goto error; 1021 fn = cp; 1022 while ((c = *cp) != '"' && c != '\0') 1023 cp++; 1024 if (c != '"') 1025 goto error; 1026 if ((fnl = cp++ - fn) > PATH_MAX) 1027 goto error; 1028 /* empty string means stdin */ 1029 if (fnl == 0) { 1030 fn = "{standard input}"; 1031 fnl = 16; /* strlen (fn) */ 1032 } 1033 curr_pos.p_file = record_filename(fn, fnl); 1034 /* 1035 * If this is the first directive, the name is the name 1036 * of the C source file as specified at the command line. 1037 * It is written to the output file. 1038 */ 1039 if (first) { 1040 csrc_pos.p_file = curr_pos.p_file; 1041 outsrc(transform_filename(curr_pos.p_file, 1042 strlen(curr_pos.p_file))); 1043 first = false; 1044 } 1045 1046 parse_line_directive_flags(cp, &is_begin, &is_end, &is_system); 1047 update_location(curr_pos.p_file, (int)ln, is_begin, is_end); 1048 in_system_header = is_system; 1049 } 1050 curr_pos.p_line = (int)ln - 1; 1051 curr_pos.p_uniq = 0; 1052 if (curr_pos.p_file == csrc_pos.p_file) { 1053 csrc_pos.p_line = (int)ln - 1; 1054 csrc_pos.p_uniq = 0; 1055 } 1056 } 1057 1058 /* 1059 * Handle lint comments such as ARGSUSED. 1060 * 1061 * If one of these comments is recognized, the argument, if any, is 1062 * parsed and a function which handles this comment is called. 1063 */ 1064 void 1065 lex_comment(void) 1066 { 1067 int c, lc; 1068 static const struct { 1069 const char *keywd; 1070 bool arg; 1071 void (*func)(int); 1072 } keywtab[] = { 1073 { "ARGSUSED", true, argsused }, 1074 { "BITFIELDTYPE", false, bitfieldtype }, 1075 { "CONSTCOND", false, constcond }, 1076 { "CONSTANTCOND", false, constcond }, 1077 { "CONSTANTCONDITION", false, constcond }, 1078 { "FALLTHRU", false, fallthru }, 1079 { "FALLTHROUGH", false, fallthru }, 1080 { "FALL THROUGH", false, fallthru }, 1081 { "fallthrough", false, fallthru }, 1082 { "LINTLIBRARY", false, lintlib }, 1083 { "LINTED", true, linted }, 1084 { "LONGLONG", false, longlong }, 1085 { "NOSTRICT", true, linted }, 1086 { "NOTREACHED", false, not_reached }, 1087 { "PRINTFLIKE", true, printflike }, 1088 { "PROTOLIB", true, protolib }, 1089 { "SCANFLIKE", true, scanflike }, 1090 { "VARARGS", true, varargs }, 1091 }; 1092 char keywd[32]; 1093 char arg[32]; 1094 size_t l, i; 1095 int a; 1096 bool eoc; 1097 1098 eoc = false; 1099 1100 /* Skip whitespace after the start of the comment */ 1101 while (c = inpc(), isspace(c)) 1102 continue; 1103 1104 /* Read the potential keyword to keywd */ 1105 l = 0; 1106 while (c != EOF && l < sizeof(keywd) - 1 && 1107 (isalpha(c) || isspace(c))) { 1108 if (islower(c) && l > 0 && ch_isupper(keywd[0])) 1109 break; 1110 keywd[l++] = (char)c; 1111 c = inpc(); 1112 } 1113 while (l > 0 && ch_isspace(keywd[l - 1])) 1114 l--; 1115 keywd[l] = '\0'; 1116 1117 /* look for the keyword */ 1118 for (i = 0; i < sizeof(keywtab) / sizeof(keywtab[0]); i++) { 1119 if (strcmp(keywtab[i].keywd, keywd) == 0) 1120 break; 1121 } 1122 if (i == sizeof(keywtab) / sizeof(keywtab[0])) 1123 goto skip_rest; 1124 1125 /* skip whitespace after the keyword */ 1126 while (isspace(c)) 1127 c = inpc(); 1128 1129 /* read the argument, if the keyword accepts one and there is one */ 1130 l = 0; 1131 if (keywtab[i].arg) { 1132 while (isdigit(c) && l < sizeof(arg) - 1) { 1133 arg[l++] = (char)c; 1134 c = inpc(); 1135 } 1136 } 1137 arg[l] = '\0'; 1138 a = l != 0 ? atoi(arg) : -1; 1139 1140 /* skip whitespace after the argument */ 1141 while (isspace(c)) 1142 c = inpc(); 1143 1144 if (c != '*' || (c = inpc()) != '/') { 1145 if (keywtab[i].func != linted) 1146 /* extra characters in lint comment */ 1147 warning(257); 1148 } else { 1149 /* 1150 * remember that we have already found the end of the 1151 * comment 1152 */ 1153 eoc = true; 1154 } 1155 1156 if (keywtab[i].func != NULL) 1157 (*keywtab[i].func)(a); 1158 1159 skip_rest: 1160 while (!eoc) { 1161 lc = c; 1162 if ((c = inpc()) == EOF) { 1163 /* unterminated comment */ 1164 error(256); 1165 break; 1166 } 1167 if (lc == '*' && c == '/') 1168 eoc = true; 1169 } 1170 } 1171 1172 /* 1173 * Handle // style comments 1174 */ 1175 void 1176 lex_slash_slash_comment(void) 1177 { 1178 int c; 1179 1180 if (!Sflag && !gflag) 1181 /* %s does not support // comments */ 1182 gnuism(312, tflag ? "traditional C" : "C90"); 1183 1184 while ((c = inpc()) != EOF && c != '\n') 1185 continue; 1186 } 1187 1188 /* 1189 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1190 * clear_warn_flags is called after function definitions and global and 1191 * local declarations and definitions. It is also called between 1192 * the controlling expression and the body of control statements 1193 * (if, switch, for, while). 1194 */ 1195 void 1196 clear_warn_flags(void) 1197 { 1198 1199 lwarn = LWARN_ALL; 1200 quadflg = false; 1201 constcond_flag = false; 1202 } 1203 1204 /* 1205 * Strings are stored in a dynamically allocated buffer and passed 1206 * in yylval.y_string to the parser. The parser or the routines called 1207 * by the parser are responsible for freeing this buffer. 1208 */ 1209 int 1210 lex_string(void) 1211 { 1212 unsigned char *s; 1213 int c; 1214 size_t len, max; 1215 strg_t *strg; 1216 1217 s = xmalloc(max = 64); 1218 1219 len = 0; 1220 while ((c = get_escaped_char('"')) >= 0) { 1221 /* +1 to reserve space for a trailing NUL character */ 1222 if (len + 1 == max) 1223 s = xrealloc(s, max *= 2); 1224 s[len++] = (char)c; 1225 } 1226 s[len] = '\0'; 1227 if (c == -2) 1228 /* unterminated string constant */ 1229 error(258); 1230 1231 strg = xcalloc(1, sizeof(*strg)); 1232 strg->st_char = true; 1233 strg->st_len = len; 1234 strg->st_mem = s; 1235 1236 yylval.y_string = strg; 1237 return T_STRING; 1238 } 1239 1240 int 1241 lex_wide_string(void) 1242 { 1243 char *s; 1244 int c, n; 1245 size_t i, wi; 1246 size_t len, max, wlen; 1247 wchar_t *ws; 1248 strg_t *strg; 1249 1250 s = xmalloc(max = 64); 1251 len = 0; 1252 while ((c = get_escaped_char('"')) >= 0) { 1253 /* +1 to save space for a trailing NUL character */ 1254 if (len + 1 >= max) 1255 s = xrealloc(s, max *= 2); 1256 s[len++] = (char)c; 1257 } 1258 s[len] = '\0'; 1259 if (c == -2) 1260 /* unterminated string constant */ 1261 error(258); 1262 1263 /* get length of wide-character string */ 1264 (void)mblen(NULL, 0); 1265 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1266 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1267 /* invalid multibyte character */ 1268 error(291); 1269 break; 1270 } 1271 if (n == 0) 1272 n = 1; 1273 } 1274 1275 ws = xmalloc((wlen + 1) * sizeof(*ws)); 1276 1277 /* convert from multibyte to wide char */ 1278 (void)mbtowc(NULL, NULL, 0); 1279 for (i = 0, wi = 0; i < len; i += n, wi++) { 1280 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1281 break; 1282 if (n == 0) 1283 n = 1; 1284 } 1285 ws[wi] = 0; 1286 free(s); 1287 1288 strg = xcalloc(1, sizeof(*strg)); 1289 strg->st_char = false; 1290 strg->st_len = wlen; 1291 strg->st_mem = ws; 1292 1293 yylval.y_string = strg; 1294 return T_STRING; 1295 } 1296 1297 #ifdef DEBUG 1298 static const char * 1299 symt_name(symt_t kind) 1300 { 1301 static const char *name[] = { 1302 "var-func-type", 1303 "member", 1304 "tag", 1305 "label", 1306 }; 1307 return name[kind]; 1308 } 1309 #endif 1310 1311 /* 1312 * As noted above, the scanner does not create new symbol table entries 1313 * for symbols it cannot find in the symbol table. This is to avoid 1314 * putting undeclared symbols into the symbol table if a syntax error 1315 * occurs. 1316 * 1317 * getsym() is called as soon as it is probably ok to put the symbol in the 1318 * symbol table. It is still possible that symbols are put in the symbol 1319 * table that are not completely declared due to syntax errors. To avoid too 1320 * many problems in this case, symbols get type 'int' in getsym(). 1321 * 1322 * XXX calls to getsym() should be delayed until decl1*() is called. 1323 */ 1324 sym_t * 1325 getsym(sbuf_t *sb) 1326 { 1327 dinfo_t *di; 1328 char *s; 1329 sym_t *sym; 1330 1331 sym = sb->sb_sym; 1332 1333 /* 1334 * During member declaration it is possible that name() looked 1335 * for symbols of type FVFT, although it should have looked for 1336 * symbols of type FTAG. Same can happen for labels. Both cases 1337 * are compensated here. 1338 */ 1339 if (symtyp == FMEMBER || symtyp == FLABEL) { 1340 if (sym == NULL || sym->s_kind == FVFT) 1341 sym = search(sb); 1342 } 1343 1344 if (sym != NULL) { 1345 lint_assert(sym->s_kind == symtyp); 1346 symtyp = FVFT; 1347 free(sb); 1348 return sym; 1349 } 1350 1351 /* create a new symbol table entry */ 1352 1353 /* labels must always be allocated at level 1 (outermost block) */ 1354 if (symtyp == FLABEL) { 1355 sym = level_zero_alloc(1, sizeof(*sym)); 1356 s = level_zero_alloc(1, sb->sb_len + 1); 1357 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1358 sym->s_name = s; 1359 sym->s_block_level = 1; 1360 di = dcs; 1361 while (di->d_next != NULL && di->d_next->d_next != NULL) 1362 di = di->d_next; 1363 lint_assert(di->d_ctx == AUTO); 1364 } else { 1365 sym = block_zero_alloc(sizeof(*sym)); 1366 sym->s_name = sb->sb_name; 1367 sym->s_block_level = block_level; 1368 di = dcs; 1369 } 1370 1371 UNIQUE_CURR_POS(sym->s_def_pos); 1372 if ((sym->s_kind = symtyp) != FLABEL) 1373 sym->s_type = gettyp(INT); 1374 1375 symtyp = FVFT; 1376 1377 symtab_add(sym); 1378 1379 *di->d_ldlsym = sym; 1380 di->d_ldlsym = &sym->s_level_next; 1381 1382 free(sb); 1383 return sym; 1384 } 1385 1386 /* 1387 * Construct a temporary symbol. The symbol name starts with a digit, making 1388 * the name illegal. 1389 */ 1390 sym_t * 1391 mktempsym(type_t *t) 1392 { 1393 static int n = 0; 1394 char *s = level_zero_alloc(block_level, 64); 1395 sym_t *sym = block_zero_alloc(sizeof(*sym)); 1396 scl_t scl; 1397 1398 (void)snprintf(s, 64, "%.8d_tmp", n++); 1399 1400 scl = dcs->d_scl; 1401 if (scl == NOSCL) 1402 scl = block_level > 0 ? AUTO : EXTERN; 1403 1404 sym->s_name = s; 1405 sym->s_type = t; 1406 sym->s_block_level = block_level; 1407 sym->s_scl = scl; 1408 sym->s_kind = FVFT; 1409 sym->s_used = true; 1410 sym->s_set = true; 1411 1412 symtab_add(sym); 1413 1414 *dcs->d_ldlsym = sym; 1415 dcs->d_ldlsym = &sym->s_level_next; 1416 1417 return sym; 1418 } 1419 1420 /* Remove a symbol forever from the symbol table. */ 1421 void 1422 rmsym(sym_t *sym) 1423 { 1424 1425 debug_step("rmsym '%s' %s '%s'", 1426 sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type)); 1427 symtab_remove(sym); 1428 1429 /* avoid that the symbol will later be put back to the symbol table */ 1430 sym->s_block_level = -1; 1431 } 1432 1433 /* 1434 * Remove a list of symbols declared at one level from the symbol 1435 * table. 1436 */ 1437 void 1438 rmsyms(sym_t *syms) 1439 { 1440 sym_t *sym; 1441 1442 for (sym = syms; sym != NULL; sym = sym->s_level_next) { 1443 if (sym->s_block_level != -1) { 1444 debug_step("rmsyms '%s' %s '%s'", 1445 sym->s_name, symt_name(sym->s_kind), 1446 type_name(sym->s_type)); 1447 symtab_remove(sym); 1448 sym->s_symtab_ref = NULL; 1449 } 1450 } 1451 } 1452 1453 /* 1454 * Put a symbol into the symbol table. 1455 */ 1456 void 1457 inssym(int bl, sym_t *sym) 1458 { 1459 1460 debug_step("inssym '%s' %s '%s'", 1461 sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type)); 1462 symtab_add(sym); 1463 sym->s_block_level = bl; 1464 lint_assert(sym->s_symtab_next == NULL || 1465 sym->s_block_level >= sym->s_symtab_next->s_block_level); 1466 } 1467 1468 /* 1469 * Called at level 0 after syntax errors. 1470 * 1471 * Removes all symbols which are not declared at level 0 from the 1472 * symbol table. Also frees all memory which is not associated with 1473 * level 0. 1474 */ 1475 void 1476 cleanup(void) 1477 { 1478 sym_t *sym, *nsym; 1479 size_t i; 1480 1481 for (i = 0; i < HSHSIZ1; i++) { 1482 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1483 nsym = sym->s_symtab_next; 1484 if (sym->s_block_level >= 1) 1485 symtab_remove(sym); 1486 } 1487 } 1488 1489 for (i = mem_block_level; i > 0; i--) 1490 level_free_all(i); 1491 } 1492 1493 /* 1494 * Create a new symbol with the name of an existing symbol. 1495 */ 1496 sym_t * 1497 pushdown(const sym_t *sym) 1498 { 1499 sym_t *nsym; 1500 1501 debug_step("pushdown '%s' %s '%s'", 1502 sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type)); 1503 nsym = block_zero_alloc(sizeof(*nsym)); 1504 lint_assert(sym->s_block_level <= block_level); 1505 nsym->s_name = sym->s_name; 1506 UNIQUE_CURR_POS(nsym->s_def_pos); 1507 nsym->s_kind = sym->s_kind; 1508 nsym->s_block_level = block_level; 1509 1510 symtab_add(nsym); 1511 1512 *dcs->d_ldlsym = nsym; 1513 dcs->d_ldlsym = &nsym->s_level_next; 1514 1515 return nsym; 1516 } 1517 1518 /* 1519 * Free any dynamically allocated memory referenced by 1520 * the value stack or yylval. 1521 * The type of information in yylval is described by tok. 1522 */ 1523 void 1524 freeyyv(void *sp, int tok) 1525 { 1526 if (tok == T_NAME || tok == T_TYPENAME) { 1527 sbuf_t *sb = *(sbuf_t **)sp; 1528 free(sb); 1529 } else if (tok == T_CON) { 1530 val_t *val = *(val_t **)sp; 1531 free(val); 1532 } else if (tok == T_STRING) { 1533 strg_t *strg = *(strg_t **)sp; 1534 free(strg->st_mem); 1535 free(strg); 1536 } 1537 } 1538
Indexes created Fri Sep 26 20:09:58 GMT 2025