lex.c revision 1.27
1 /* $NetBSD: lex.c,v 1.27 2021/04/12 15:55:26 christos 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 1.27 2021/04/12 15:55:26 christos 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 ((int)(~(~0U << CHAR_SIZE))) 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 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 #ifdef DEBUG 86 printf("parsing %s:%d\n", curr_pos.p_file, curr_pos.p_line); 87 #endif 88 if (curr_pos.p_file == csrc_pos.p_file) { 89 csrc_pos.p_line++; 90 csrc_pos.p_uniq = 0; 91 } 92 } 93 94 void 95 lex_unknown_character(int c) 96 { 97 98 /* unknown character \%o */ 99 error(250, c); 100 } 101 102 #define kwdef(name, token, scl, tspec, tqual, c89, c99, gcc, attr, deco) \ 103 { \ 104 name, token, scl, tspec, tqual, \ 105 (c89) > 0, (c99) > 0, (gcc) > 0, (attr) > 0, deco, \ 106 } 107 #define kwdef_token(name, token, c89, c99, gcc, attr, deco) \ 108 kwdef(name, token, 0, 0, 0, c89, c99, gcc, attr, deco) 109 #define kwdef_sclass(name, sclass, c89, c99, gcc, attr, deco) \ 110 kwdef(name, T_SCLASS, sclass, 0, 0, c89, c99, gcc, attr, deco) 111 #define kwdef_type(name, tspec, c89, c99, gcc, attr, deco) \ 112 kwdef(name, T_TYPE, 0, tspec, 0, c89, c99, gcc, attr, deco) 113 #define kwdef_tqual(name, tqual, c89, c99, gcc, attr, deco) \ 114 kwdef(name, T_QUAL, 0, 0, tqual, c89, c99, gcc, attr, deco) 115 116 /* 117 * Keywords. 118 * During initialization they are written to the symbol table. 119 */ 120 static struct kwtab { 121 const char *kw_name; /* keyword */ 122 int kw_token; /* token returned by yylex() */ 123 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 124 tspec_t kw_tspec; /* type spec. if kw_token 125 * T_TYPE or T_STRUCT_OR_UNION */ 126 tqual_t kw_tqual; /* type qual. if kw_token T_QUAL */ 127 bool kw_c89 : 1; /* C89 keyword */ 128 bool kw_c99 : 1; /* C99 keyword */ 129 bool kw_gcc : 1; /* GCC keyword */ 130 bool kw_attr : 1; /* GCC attribute, keyword */ 131 u_int kw_deco : 3; /* 1 = name, 2 = __name, 4 = __name__ */ 132 } kwtab[] = { 133 #ifdef INT128_SIZE 134 kwdef_type( "__int128_t", INT128, 0,1,0,0,1), 135 kwdef_type( "__uint128_t", UINT128, 0,1,0,0,1), 136 #endif 137 kwdef_tqual( "__thread", THREAD, 0,0,1,0,1), 138 kwdef_token( "_Alignas", T_ALIGNAS, 0,0,0,0,1), 139 kwdef_token( "_Alignof", T_ALIGNOF, 0,0,0,0,1), 140 kwdef_type( "_Bool", BOOL, 0,1,0,0,1), 141 kwdef_type( "_Complex", COMPLEX, 0,1,0,0,1), 142 kwdef_token( "_Generic", T_GENERIC, 0,1,0,0,1), 143 kwdef_token( "_Noreturn", T_NORETURN, 0,1,0,0,1), 144 kwdef_tqual( "_Thread_local", THREAD, 0,1,0,0,1), 145 kwdef_token( "alias", T_AT_ALIAS, 0,0,1,1,5), 146 kwdef_token( "aligned", T_AT_ALIGNED, 0,0,1,1,5), 147 kwdef_token( "alignof", T_ALIGNOF, 0,0,0,0,4), 148 kwdef_token( "alloc_size", T_AT_ALLOC_SIZE, 0,0,1,1,5), 149 kwdef_token( "always_inline", T_AT_ALWAYS_INLINE, 0,0,1,1,5), 150 kwdef_token( "asm", T_ASM, 0,0,1,0,7), 151 kwdef_token( "attribute", T_ATTRIBUTE, 0,0,1,0,6), 152 kwdef_sclass( "auto", AUTO, 0,0,0,0,1), 153 kwdef_token( "bounded", T_AT_BOUNDED, 0,0,1,1,5), 154 kwdef_token( "break", T_BREAK, 0,0,0,0,1), 155 kwdef_token( "buffer", T_AT_BUFFER, 0,0,1,1,5), 156 kwdef_token( "builtin_offsetof", T_BUILTIN_OFFSETOF, 0,0,1,0,2), 157 kwdef_token( "case", T_CASE, 0,0,0,0,1), 158 kwdef_type( "char", CHAR, 0,0,0,0,1), 159 kwdef_token( "cold", T_AT_COLD, 0,0,1,1,5), 160 kwdef_tqual( "const", CONST, 1,0,0,0,7), 161 kwdef_token( "constructor", T_AT_CONSTRUCTOR, 0,0,1,1,5), 162 kwdef_token( "continue", T_CONTINUE, 0,0,0,0,1), 163 kwdef_token( "default", T_DEFAULT, 0,0,0,0,1), 164 kwdef_token( "deprecated", T_AT_DEPRECATED, 0,0,1,1,5), 165 kwdef_token( "destructor", T_AT_DESTRUCTOR, 0,0,1,1,5), 166 kwdef_token( "do", T_DO, 0,0,0,0,1), 167 kwdef_type( "double", DOUBLE, 0,0,0,0,1), 168 kwdef_token( "else", T_ELSE, 0,0,0,0,1), 169 kwdef_token( "enum", T_ENUM, 0,0,0,0,1), 170 kwdef_token( "extension", T_EXTENSION, 0,0,1,0,4), 171 kwdef_sclass( "extern", EXTERN, 0,0,0,0,1), 172 kwdef_token( "fallthrough", T_AT_FALLTHROUGH, 0,0,1,1,5), 173 kwdef_type( "float", FLOAT, 0,0,0,0,1), 174 kwdef_token( "for", T_FOR, 0,0,0,0,1), 175 kwdef_token( "format", T_AT_FORMAT, 0,0,1,1,5), 176 kwdef_token( "format_arg", T_AT_FORMAT_ARG, 0,0,1,1,5), 177 kwdef_token( "gnu_inline", T_AT_GNU_INLINE, 0,0,1,1,5), 178 kwdef_token( "gnu_printf", T_AT_FORMAT_GNU_PRINTF, 0,0,1,1,5), 179 kwdef_token( "goto", T_GOTO, 0,0,0,0,1), 180 kwdef_token( "if", T_IF, 0,0,0,0,1), 181 kwdef_token( "imag", T_IMAG, 0,1,0,0,4), 182 kwdef_sclass( "inline", INLINE, 0,1,0,0,7), 183 kwdef_type( "int", INT, 0,0,0,0,1), 184 kwdef_type( "long", LONG, 0,0,0,0,1), 185 kwdef_token( "malloc", T_AT_MALLOC, 0,0,1,1,5), 186 kwdef_token( "may_alias", T_AT_MAY_ALIAS, 0,0,1,1,5), 187 kwdef_token( "minbytes", T_AT_MINBYTES, 0,0,1,1,5), 188 kwdef_token( "mode", T_AT_MODE, 0,0,1,1,5), 189 kwdef_token( "no_instrument_function", 190 T_AT_NO_INSTRUMENT_FUNCTION, 0,0,1,1,5), 191 kwdef_token( "nonnull", T_AT_NONNULL, 0,0,1,1,5), 192 kwdef_token( "noinline", T_AT_NOINLINE, 0,0,1,1,5), 193 kwdef_token( "noreturn", T_AT_NORETURN, 0,0,1,1,5), 194 kwdef_token( "nothrow", T_AT_NOTHROW, 0,0,1,1,5), 195 kwdef_token( "optimize", T_AT_OPTIMIZE, 0,0,1,1,5), 196 kwdef_token( "packed", T_AT_PACKED, 0,0,1,1,5), 197 kwdef_token( "packed", T_PACKED, 0,0,0,0,2), 198 kwdef_token( "pcs", T_AT_PCS, 0,0,0,0,5), 199 kwdef_token( "printf", T_AT_FORMAT_PRINTF, 0,0,1,1,5), 200 kwdef_token( "pure", T_AT_PURE, 0,0,1,1,5), 201 kwdef_token( "real", T_REAL, 0,1,0,0,4), 202 kwdef_sclass( "register", REG, 0,0,0,0,1), 203 kwdef_tqual( "restrict", RESTRICT, 0,1,0,0,5), 204 kwdef_token( "return", T_RETURN, 0,0,0,0,1), 205 kwdef_token( "returns_twice", T_AT_RETURNS_TWICE, 0,0,1,1,5), 206 kwdef_token( "scanf", T_AT_FORMAT_SCANF, 0,0,1,1,5), 207 kwdef_token( "section", T_AT_SECTION, 0,0,1,1,7), 208 kwdef_token( "sentinel", T_AT_SENTINEL, 0,0,1,1,5), 209 kwdef_type( "short", SHORT, 0,0,0,0,1), 210 kwdef_type( "signed", SIGNED, 1,0,0,0,3), 211 kwdef_token( "sizeof", T_SIZEOF, 0,0,0,0,1), 212 kwdef_sclass( "static", STATIC, 0,0,0,0,1), 213 kwdef_token( "strfmon", T_AT_FORMAT_STRFMON, 0,0,1,1,5), 214 kwdef_token( "strftime", T_AT_FORMAT_STRFTIME, 0,0,1,1,5), 215 kwdef_token( "string", T_AT_STRING, 0,0,1,1,5), 216 kwdef("struct", T_STRUCT_OR_UNION, 0, STRUCT, 0, 0,0,0,0,1), 217 kwdef_token( "switch", T_SWITCH, 0,0,0,0,1), 218 kwdef_token( "symbolrename", T_SYMBOLRENAME, 0,0,0,0,2), 219 kwdef_token( "syslog", T_AT_FORMAT_SYSLOG, 0,0,1,1,5), 220 kwdef_token( "transparent_union", T_AT_TUNION, 0,0,1,1,5), 221 kwdef_token( "tls_model", T_AT_TLS_MODEL, 0,0,1,1,5), 222 kwdef_sclass( "typedef", TYPEDEF, 0,0,0,0,1), 223 kwdef_token( "typeof", T_TYPEOF, 0,0,1,0,7), 224 kwdef("union", T_STRUCT_OR_UNION, 0, UNION, 0, 0,0,0,0,1), 225 kwdef_type( "unsigned", UNSIGN, 0,0,0,0,1), 226 kwdef_token( "unused", T_AT_UNUSED, 0,0,1,1,5), 227 kwdef_token( "used", T_AT_USED, 0,0,1,1,5), 228 kwdef_token( "visibility", T_AT_VISIBILITY, 0,0,1,1,5), 229 kwdef_type( "void", VOID, 0,0,0,0,1), 230 kwdef_tqual( "volatile", VOLATILE, 1,0,0,0,7), 231 kwdef_token("warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0,0,1,1,5), 232 kwdef_token( "weak", T_AT_WEAK, 0,0,1,1,5), 233 kwdef_token( "while", T_WHILE, 0,0,0,0,1), 234 kwdef(NULL, 0, 0, 0, 0, 0, 0, 0, 0, 0), 235 #undef kwdef 236 #undef kwdef_token 237 #undef kwdef_sclass 238 #undef kwdef_type 239 #undef kwdef_tqual 240 }; 241 242 /* Symbol table */ 243 static sym_t *symtab[HSHSIZ1]; 244 245 /* bit i of the entry with index i is set */ 246 uint64_t qbmasks[64]; 247 248 /* least significant i bits are set in the entry with index i */ 249 uint64_t qlmasks[64 + 1]; 250 251 /* least significant i bits are not set in the entry with index i */ 252 uint64_t qumasks[64 + 1]; 253 254 /* free list for sbuf structures */ 255 static sbuf_t *sbfrlst; 256 257 /* type of next expected symbol */ 258 symt_t symtyp; 259 260 261 static void 262 add_keyword(struct kwtab *kw, int deco) 263 { 264 sym_t *sym; 265 size_t h; 266 char buf[256]; 267 const char *name; 268 269 if ((kw->kw_deco & deco) == 0) 270 return; 271 272 switch (deco) { 273 case 1: 274 name = kw->kw_name; 275 break; 276 case 2: 277 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 278 name = strdup(buf); 279 break; 280 default: 281 lint_assert(deco == 4); 282 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 283 name = strdup(buf); 284 break; 285 } 286 287 if (name == NULL) 288 err(1, "Can't init symbol table"); 289 290 sym = getblk(sizeof(*sym)); 291 sym->s_name = name; 292 sym->s_keyword = kw; 293 sym->s_value.v_quad = kw->kw_token; 294 if (kw->kw_token == T_TYPE || kw->kw_token == T_STRUCT_OR_UNION) { 295 sym->s_tspec = kw->kw_tspec; 296 } else if (kw->kw_token == T_SCLASS) { 297 sym->s_scl = kw->kw_scl; 298 } else if (kw->kw_token == T_QUAL) { 299 sym->s_tqual = kw->kw_tqual; 300 } 301 h = hash(sym->s_name); 302 if ((sym->s_link = symtab[h]) != NULL) 303 symtab[h]->s_rlink = &sym->s_link; 304 sym->s_rlink = &symtab[h]; 305 symtab[h] = sym; 306 } 307 308 /* 309 * All keywords are written to the symbol table. This saves us looking 310 * in a extra table for each name we found. 311 */ 312 void 313 initscan(void) 314 { 315 struct kwtab *kw; 316 size_t i; 317 uint64_t uq; 318 319 for (kw = kwtab; kw->kw_name != NULL; kw++) { 320 if ((kw->kw_c89 || kw->kw_c99) && tflag) 321 continue; 322 if (kw->kw_c99 && !(Sflag || gflag)) 323 continue; 324 if (kw->kw_gcc && !gflag) 325 continue; 326 add_keyword(kw, 1); 327 add_keyword(kw, 2); 328 add_keyword(kw, 4); 329 } 330 331 /* initialize bit-masks for quads */ 332 for (i = 0; i < 64; i++) { 333 qbmasks[i] = (uint64_t)1 << i; 334 uq = ~(uint64_t)0 << i; 335 qumasks[i] = uq; 336 qlmasks[i] = ~uq; 337 } 338 qumasks[i] = 0; 339 qlmasks[i] = ~(uint64_t)0; 340 } 341 342 /* 343 * Get a free sbuf structure, if possible from the free list 344 */ 345 static sbuf_t * 346 allocsb(void) 347 { 348 sbuf_t *sb; 349 350 if ((sb = sbfrlst) != NULL) { 351 sbfrlst = sb->sb_next; 352 #ifdef BLKDEBUG 353 (void)memset(sb, 0, sizeof(*sb)); 354 #else 355 sb->sb_next = NULL; 356 #endif 357 } else { 358 sb = xmalloc(sizeof(*sb)); 359 (void)memset(sb, 0, sizeof(*sb)); 360 } 361 return sb; 362 } 363 364 /* 365 * Put a sbuf structure to the free list 366 */ 367 static void 368 freesb(sbuf_t *sb) 369 { 370 371 (void)memset(sb, ZERO, sizeof(*sb)); 372 sb->sb_next = sbfrlst; 373 sbfrlst = sb; 374 } 375 376 /* 377 * Read a character and ensure that it is positive (except EOF). 378 * Increment line count(s) if necessary. 379 */ 380 static int 381 inpc(void) 382 { 383 int c; 384 385 if ((c = lex_input()) != EOF && (c &= CHAR_MASK) == '\n') 386 lex_next_line(); 387 return c; 388 } 389 390 static int 391 hash(const char *s) 392 { 393 u_int v; 394 const u_char *us; 395 396 v = 0; 397 for (us = (const u_char *)s; *us != '\0'; us++) { 398 v = (v << sizeof(v)) + *us; 399 v ^= v >> (sizeof(v) * CHAR_BIT - sizeof(v)); 400 } 401 return v % HSHSIZ1; 402 } 403 404 /* 405 * Lex has found a letter followed by zero or more letters or digits. 406 * It looks for a symbol in the symbol table with the same name. This 407 * symbol must either be a keyword or a symbol of the type required by 408 * symtyp (label, member, tag, ...). 409 * 410 * If it is a keyword, the token is returned. In some cases it is described 411 * more deeply by data written to yylval. 412 * 413 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 414 * is stored in yylval. This struct contains the name of the symbol, its 415 * length and hash value. If there is already a symbol of the same name 416 * and type in the symbol table, the sbuf struct also contains a pointer 417 * to the symbol table entry. 418 */ 419 extern int 420 lex_name(const char *yytext, size_t yyleng) 421 { 422 char *s; 423 sbuf_t *sb; 424 sym_t *sym; 425 int tok; 426 427 sb = allocsb(); 428 sb->sb_name = yytext; 429 sb->sb_len = yyleng; 430 sb->sb_hash = hash(yytext); 431 if ((sym = search(sb)) != NULL && sym->s_keyword != NULL) { 432 freesb(sb); 433 return keyw(sym); 434 } 435 436 sb->sb_sym = sym; 437 438 if (sym != NULL) { 439 lint_assert(block_level >= sym->s_block_level); 440 sb->sb_name = sym->s_name; 441 sb->sb_len = strlen(sym->s_name); 442 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 443 } else { 444 s = getblk(yyleng + 1); 445 (void)memcpy(s, yytext, yyleng + 1); 446 sb->sb_name = s; 447 sb->sb_len = yyleng; 448 tok = T_NAME; 449 } 450 451 yylval.y_sb = sb; 452 return tok; 453 } 454 455 static sym_t * 456 search(sbuf_t *sb) 457 { 458 sym_t *sym; 459 460 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 461 if (strcmp(sym->s_name, sb->sb_name) == 0) { 462 if (sym->s_keyword != NULL) { 463 struct kwtab *kw = sym->s_keyword; 464 if (!kw->kw_attr || attron) 465 return sym; 466 } else if (!attron && sym->s_kind == symtyp) 467 return sym; 468 } 469 } 470 471 return NULL; 472 } 473 474 static int 475 keyw(sym_t *sym) 476 { 477 int t; 478 479 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 480 yylval.y_scl = sym->s_scl; 481 } else if (t == T_TYPE || t == T_STRUCT_OR_UNION) { 482 yylval.y_tspec = sym->s_tspec; 483 } else if (t == T_QUAL) { 484 yylval.y_tqual = sym->s_tqual; 485 } 486 return t; 487 } 488 489 /* 490 * Convert a string representing an integer into internal representation. 491 * Return T_CON, storing the numeric value in yylval, for yylex. 492 */ 493 int 494 lex_integer_constant(const char *yytext, size_t yyleng, int base) 495 { 496 int l_suffix, u_suffix; 497 int len; 498 const char *cp; 499 char c, *eptr; 500 tspec_t typ; 501 bool ansiu; 502 bool warned = false; 503 #ifdef TARG_INT128_MAX 504 __uint128_t uq = 0; 505 static tspec_t contypes[2][4] = { 506 { INT, LONG, QUAD, INT128, }, 507 { UINT, ULONG, UQUAD, UINT128, } 508 }; 509 #else 510 uint64_t uq = 0; 511 static tspec_t contypes[2][3] = { 512 { INT, LONG, QUAD, }, 513 { UINT, ULONG, UQUAD, } 514 }; 515 #endif 516 517 cp = yytext; 518 len = yyleng; 519 520 /* skip 0[xX] or 0[bB] */ 521 if (base == 16 || base == 2) { 522 cp += 2; 523 len -= 2; 524 } 525 526 /* read suffixes */ 527 l_suffix = u_suffix = 0; 528 for (;;) { 529 if ((c = cp[len - 1]) == 'l' || c == 'L') { 530 l_suffix++; 531 } else if (c == 'u' || c == 'U') { 532 u_suffix++; 533 } else { 534 break; 535 } 536 len--; 537 } 538 if (l_suffix > 2 || u_suffix > 1) { 539 /* malformed integer constant */ 540 warning(251); 541 if (l_suffix > 2) 542 l_suffix = 2; 543 if (u_suffix > 1) 544 u_suffix = 1; 545 } 546 if (tflag && u_suffix != 0) { 547 /* suffix U is illegal in traditional C */ 548 warning(97); 549 } 550 typ = contypes[u_suffix][l_suffix]; 551 552 errno = 0; 553 554 uq = strtouq(cp, &eptr, base); 555 lint_assert(eptr == cp + len); 556 if (errno != 0) { 557 /* integer constant out of range */ 558 warning(252); 559 warned = true; 560 } 561 562 /* 563 * If the value is too big for the current type, we must choose 564 * another type. 565 */ 566 ansiu = false; 567 switch (typ) { 568 case INT: 569 if (uq <= TARG_INT_MAX) { 570 /* ok */ 571 } else if (uq <= TARG_UINT_MAX && base != 10) { 572 typ = UINT; 573 } else if (uq <= TARG_LONG_MAX) { 574 typ = LONG; 575 } else { 576 typ = ULONG; 577 if (uq > TARG_ULONG_MAX && !warned) { 578 /* integer constant out of range */ 579 warning(252); 580 } 581 } 582 if (typ == UINT || typ == ULONG) { 583 if (tflag) { 584 typ = LONG; 585 } else if (!sflag) { 586 /* 587 * Remember that the constant is unsigned 588 * only in ANSI C 589 */ 590 ansiu = true; 591 } 592 } 593 break; 594 case UINT: 595 if (uq > TARG_UINT_MAX) { 596 typ = ULONG; 597 if (uq > TARG_ULONG_MAX && !warned) { 598 /* integer constant out of range */ 599 warning(252); 600 } 601 } 602 break; 603 case LONG: 604 if (uq > TARG_LONG_MAX && !tflag) { 605 typ = ULONG; 606 if (!sflag) 607 ansiu = true; 608 if (uq > TARG_ULONG_MAX && !warned) { 609 /* integer constant out of range */ 610 warning(252); 611 } 612 } 613 break; 614 case ULONG: 615 if (uq > TARG_ULONG_MAX && !warned) { 616 /* integer constant out of range */ 617 warning(252); 618 } 619 break; 620 case QUAD: 621 if (uq > TARG_QUAD_MAX && !tflag) { 622 typ = UQUAD; 623 if (!sflag) 624 ansiu = true; 625 } 626 break; 627 case UQUAD: 628 if (uq > TARG_UQUAD_MAX && !warned) { 629 /* integer constant out of range */ 630 warning(252); 631 } 632 break; 633 #ifdef INT128_SIZE 634 case INT128: 635 #ifdef TARG_INT128_MAX 636 if (uq > TARG_INT128_MAX && !tflag) { 637 typ = UINT128; 638 if (!sflag) 639 ansiu = true; 640 } 641 #endif 642 break; 643 case UINT128: 644 #ifdef TARG_INT128_MAX 645 if (uq > TARG_UINT128_MAX && !warned) { 646 /* integer constant out of range */ 647 warning(252); 648 } 649 #endif 650 break; 651 #endif 652 /* LINTED206: (enumeration values not handled in switch) */ 653 case STRUCT: 654 case VOID: 655 case LDOUBLE: 656 case FUNC: 657 case ARRAY: 658 case PTR: 659 case ENUM: 660 case UNION: 661 case SIGNED: 662 case NOTSPEC: 663 case DOUBLE: 664 case FLOAT: 665 case USHORT: 666 case SHORT: 667 case UCHAR: 668 case SCHAR: 669 case CHAR: 670 case BOOL: 671 case UNSIGN: 672 case FCOMPLEX: 673 case DCOMPLEX: 674 case LCOMPLEX: 675 case COMPLEX: 676 break; 677 } 678 679 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 680 681 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 682 yylval.y_val->v_tspec = typ; 683 yylval.y_val->v_ansiu = ansiu; 684 yylval.y_val->v_quad = (int64_t)uq; 685 686 return T_CON; 687 } 688 689 /* 690 * Returns whether t is a signed type and the value is negative. 691 * 692 * len is the number of significant bits. If len is -1, len is set 693 * to the width of type t. 694 */ 695 static bool 696 sign(int64_t q, tspec_t t, int len) 697 { 698 699 if (t == PTR || is_uinteger(t)) 700 return false; 701 return msb(q, t, len) != 0; 702 } 703 704 int 705 msb(int64_t q, tspec_t t, int len) 706 { 707 708 if (len <= 0) 709 len = size_in_bits(t); 710 return (q & qbmasks[len - 1]) != 0 ? 1 : 0; 711 } 712 713 /* 714 * Extends the sign of q. 715 */ 716 int64_t 717 xsign(int64_t q, tspec_t t, int len) 718 { 719 720 if (len <= 0) 721 len = size_in_bits(t); 722 723 if (t == PTR || is_uinteger(t) || !sign(q, t, len)) { 724 q &= qlmasks[len]; 725 } else { 726 q |= qumasks[len]; 727 } 728 return q; 729 } 730 731 /* 732 * Convert a string representing a floating point value into its integral 733 * representation. Type and value are returned in yylval. fcon() 734 * (and yylex()) returns T_CON. 735 * XXX Currently it is not possible to convert constants of type 736 * long double which are greater than DBL_MAX. 737 */ 738 int 739 lex_floating_constant(const char *yytext, size_t yyleng) 740 { 741 const char *cp; 742 int len; 743 tspec_t typ; 744 char c, *eptr; 745 double d; 746 float f = 0; 747 748 cp = yytext; 749 len = yyleng; 750 751 if (cp[len - 1] == 'i') { 752 /* imaginary, do nothing for now */ 753 len--; 754 } 755 if ((c = cp[len - 1]) == 'f' || c == 'F') { 756 typ = FLOAT; 757 len--; 758 } else if (c == 'l' || c == 'L') { 759 typ = LDOUBLE; 760 len--; 761 } else { 762 if (c == 'd' || c == 'D') 763 len--; 764 typ = DOUBLE; 765 } 766 767 if (tflag && typ != DOUBLE) { 768 /* suffixes F and L are illegal in traditional C */ 769 warning(98); 770 } 771 772 errno = 0; 773 d = strtod(cp, &eptr); 774 if (eptr != cp + len) { 775 switch (*eptr) { 776 /* 777 * XXX: non-native non-current strtod() may not handle hex 778 * floats, ignore the rest if we find traces of hex float 779 * syntax... 780 */ 781 case 'p': 782 case 'P': 783 case 'x': 784 case 'X': 785 d = 0; 786 errno = 0; 787 break; 788 default: 789 INTERNAL_ERROR("fcon(%s->%s)", cp, eptr); 790 } 791 } 792 if (errno != 0) 793 /* floating-point constant out of range */ 794 warning(248); 795 796 if (typ == FLOAT) { 797 f = (float)d; 798 if (finite(f) == 0) { 799 /* floating-point constant out of range */ 800 warning(248); 801 f = f > 0 ? FLT_MAX : -FLT_MAX; 802 } 803 } 804 805 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 806 yylval.y_val->v_tspec = typ; 807 if (typ == FLOAT) { 808 yylval.y_val->v_ldbl = f; 809 } else { 810 yylval.y_val->v_ldbl = d; 811 } 812 813 return T_CON; 814 } 815 816 int 817 lex_operator(int t, op_t o) 818 { 819 820 yylval.y_op = o; 821 return t; 822 } 823 824 /* 825 * Called if lex found a leading \'. 826 */ 827 int 828 lex_character_constant(void) 829 { 830 size_t n; 831 int val, c; 832 char cv; 833 834 n = 0; 835 val = 0; 836 while ((c = get_escaped_char('\'')) >= 0) { 837 val = (val << CHAR_SIZE) + c; 838 n++; 839 } 840 if (c == -2) { 841 /* unterminated character constant */ 842 error(253); 843 } else { 844 /* XXX: should rather be sizeof(TARG_INT) */ 845 if (n > sizeof(int) || (n > 1 && (pflag || hflag))) { 846 /* too many characters in character constant */ 847 error(71); 848 } else if (n > 1) { 849 /* multi-character character constant */ 850 warning(294); 851 } else if (n == 0) { 852 /* empty character constant */ 853 error(73); 854 } 855 } 856 if (n == 1) { 857 cv = (char)val; 858 val = cv; 859 } 860 861 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 862 yylval.y_val->v_tspec = INT; 863 yylval.y_val->v_quad = val; 864 865 return T_CON; 866 } 867 868 /* 869 * Called if lex found a leading L\' 870 */ 871 int 872 lex_wide_character_constant(void) 873 { 874 static char buf[MB_LEN_MAX + 1]; 875 size_t i, imax; 876 int c; 877 wchar_t wc; 878 879 imax = MB_CUR_MAX; 880 881 i = 0; 882 while ((c = get_escaped_char('\'')) >= 0) { 883 if (i < imax) 884 buf[i] = (char)c; 885 i++; 886 } 887 888 wc = 0; 889 890 if (c == -2) { 891 /* unterminated character constant */ 892 error(253); 893 } else if (c == 0) { 894 /* empty character constant */ 895 error(73); 896 } else { 897 if (i > imax) { 898 i = imax; 899 /* too many characters in character constant */ 900 error(71); 901 } else { 902 buf[i] = '\0'; 903 (void)mbtowc(NULL, NULL, 0); 904 if (mbtowc(&wc, buf, imax) < 0) 905 /* invalid multibyte character */ 906 error(291); 907 } 908 } 909 910 yylval.y_val = xcalloc(1, sizeof(*yylval.y_val)); 911 yylval.y_val->v_tspec = WCHAR; 912 yylval.y_val->v_quad = wc; 913 914 return T_CON; 915 } 916 917 /* 918 * Read a character which is part of a character constant or of a string 919 * and handle escapes. 920 * 921 * The argument is the character which delimits the character constant or 922 * string. 923 * 924 * Returns -1 if the end of the character constant or string is reached, 925 * -2 if the EOF is reached, and the character otherwise. 926 */ 927 static int 928 get_escaped_char(int delim) 929 { 930 static int pbc = -1; 931 int n, c, v; 932 933 if (pbc == -1) { 934 c = inpc(); 935 } else { 936 c = pbc; 937 pbc = -1; 938 } 939 if (c == delim) 940 return -1; 941 switch (c) { 942 case '\n': 943 if (tflag) { 944 /* newline in string or char constant */ 945 error(254); 946 return -2; 947 } 948 return c; 949 case EOF: 950 return -2; 951 case '\\': 952 switch (c = inpc()) { 953 case '"': 954 if (tflag && delim == '\'') 955 /* \" inside character constants undef... */ 956 warning(262); 957 return '"'; 958 case '\'': 959 return '\''; 960 case '?': 961 if (tflag) 962 /* \? undefined in traditional C */ 963 warning(263); 964 return '?'; 965 case '\\': 966 return '\\'; 967 case 'a': 968 if (tflag) 969 /* \a undefined in traditional C */ 970 warning(81); 971 return '\a'; 972 case 'b': 973 return '\b'; 974 case 'f': 975 return '\f'; 976 case 'n': 977 return '\n'; 978 case 'r': 979 return '\r'; 980 case 't': 981 return '\t'; 982 case 'v': 983 if (tflag) 984 /* \v undefined in traditional C */ 985 warning(264); 986 return '\v'; 987 case '8': case '9': 988 /* bad octal digit %c */ 989 warning(77, c); 990 /* FALLTHROUGH */ 991 case '0': case '1': case '2': case '3': 992 case '4': case '5': case '6': case '7': 993 n = 3; 994 v = 0; 995 do { 996 v = (v << 3) + (c - '0'); 997 c = inpc(); 998 } while (--n > 0 && isdigit(c) && (tflag || c <= '7')); 999 if (tflag && n > 0 && isdigit(c)) 1000 /* bad octal digit %c */ 1001 warning(77, c); 1002 pbc = c; 1003 if (v > TARG_UCHAR_MAX) { 1004 /* character escape does not fit in character */ 1005 warning(76); 1006 v &= CHAR_MASK; 1007 } 1008 return v; 1009 case 'x': 1010 if (tflag) 1011 /* \x undefined in traditional C */ 1012 warning(82); 1013 v = 0; 1014 n = 0; 1015 while ((c = inpc()) >= 0 && isxdigit(c)) { 1016 c = isdigit(c) ? 1017 c - '0' : toupper(c) - 'A' + 10; 1018 v = (v << 4) + c; 1019 if (n >= 0) { 1020 if ((v & ~CHAR_MASK) != 0) { 1021 /* overflow in hex escape */ 1022 warning(75); 1023 n = -1; 1024 } else { 1025 n++; 1026 } 1027 } 1028 } 1029 pbc = c; 1030 if (n == 0) { 1031 /* no hex digits follow \x */ 1032 error(74); 1033 } if (n == -1) { 1034 v &= CHAR_MASK; 1035 } 1036 return v; 1037 case '\n': 1038 return get_escaped_char(delim); 1039 case EOF: 1040 return -2; 1041 default: 1042 if (isprint(c)) { 1043 /* dubious escape \%c */ 1044 warning(79, c); 1045 } else { 1046 /* dubious escape \%o */ 1047 warning(80, c); 1048 } 1049 } 1050 } 1051 return c; 1052 } 1053 1054 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */ 1055 static void 1056 parse_line_directive_flags(const char *p, 1057 bool *is_begin, bool *is_end, bool *is_system) 1058 { 1059 1060 *is_begin = false; 1061 *is_end = false; 1062 *is_system = false; 1063 1064 while (*p != '\0') { 1065 while (ch_isspace(*p)) 1066 p++; 1067 1068 const char *word_start = p; 1069 while (*p != '\0' && !ch_isspace(*p)) 1070 p++; 1071 const char *word_end = p; 1072 1073 if (word_end - word_start == 1 && word_start[0] == '1') 1074 *is_begin = true; 1075 if (word_end - word_start == 1 && word_start[0] == '2') 1076 *is_end = true; 1077 if (word_end - word_start == 1 && word_start[0] == '3') 1078 *is_system = true; 1079 /* Flag '4' would only be interesting if lint handled C++. */ 1080 } 1081 1082 #if 0 1083 if (*p != '\0') { 1084 /* syntax error '%s' */ 1085 warning(249, "extra character(s) after directive"); 1086 } 1087 #endif 1088 } 1089 1090 /* 1091 * Called for preprocessor directives. Currently implemented are: 1092 * # lineno 1093 * # lineno "filename" 1094 * # lineno "filename" GCC-flag... 1095 */ 1096 void 1097 lex_directive(const char *yytext) 1098 { 1099 const char *cp, *fn; 1100 char c, *eptr; 1101 size_t fnl; 1102 long ln; 1103 bool is_begin, is_end, is_system; 1104 1105 static bool first = true; 1106 1107 /* Go to first non-whitespace after # */ 1108 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1109 continue; 1110 1111 if (!ch_isdigit(c)) { 1112 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6])) 1113 return; 1114 error: 1115 /* undefined or invalid # directive */ 1116 warning(255); 1117 return; 1118 } 1119 ln = strtol(--cp, &eptr, 10); 1120 if (cp == eptr) 1121 goto error; 1122 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1123 goto error; 1124 while ((c = *cp++) == ' ' || c == '\t') 1125 continue; 1126 if (c != '\0') { 1127 if (c != '"') 1128 goto error; 1129 fn = cp; 1130 while ((c = *cp) != '"' && c != '\0') 1131 cp++; 1132 if (c != '"') 1133 goto error; 1134 if ((fnl = cp++ - fn) > PATH_MAX) 1135 goto error; 1136 /* empty string means stdin */ 1137 if (fnl == 0) { 1138 fn = "{standard input}"; 1139 fnl = 16; /* strlen (fn) */ 1140 } 1141 curr_pos.p_file = record_filename(fn, fnl); 1142 /* 1143 * If this is the first directive, the name is the name 1144 * of the C source file as specified at the command line. 1145 * It is written to the output file. 1146 */ 1147 if (first) { 1148 csrc_pos.p_file = curr_pos.p_file; 1149 outsrc(transform_filename(curr_pos.p_file, 1150 strlen(curr_pos.p_file))); 1151 first = false; 1152 } 1153 1154 parse_line_directive_flags(cp, &is_begin, &is_end, &is_system); 1155 update_location(curr_pos.p_file, (int)ln, is_begin, is_end); 1156 in_system_header = is_system; 1157 } 1158 curr_pos.p_line = (int)ln - 1; 1159 curr_pos.p_uniq = 0; 1160 if (curr_pos.p_file == csrc_pos.p_file) { 1161 csrc_pos.p_line = (int)ln - 1; 1162 csrc_pos.p_uniq = 0; 1163 } 1164 } 1165 1166 /* 1167 * Handle lint comments such as ARGSUSED. 1168 * 1169 * If one of these comments is recognized, the argument, if any, is 1170 * parsed and a function which handles this comment is called. 1171 */ 1172 void 1173 lex_comment(void) 1174 { 1175 int c, lc; 1176 static const struct { 1177 const char *keywd; 1178 bool arg; 1179 void (*func)(int); 1180 } keywtab[] = { 1181 { "ARGSUSED", true, argsused }, 1182 { "BITFIELDTYPE", false, bitfieldtype }, 1183 { "CONSTCOND", false, constcond }, 1184 { "CONSTANTCOND", false, constcond }, 1185 { "CONSTANTCONDITION", false, constcond }, 1186 { "FALLTHRU", false, fallthru }, 1187 { "FALLTHROUGH", false, fallthru }, 1188 { "LINTLIBRARY", false, lintlib }, 1189 { "LINTED", true, linted }, 1190 { "LONGLONG", false, longlong }, 1191 { "NOSTRICT", true, linted }, 1192 { "NOTREACHED", false, not_reached }, 1193 { "PRINTFLIKE", true, printflike }, 1194 { "PROTOLIB", true, protolib }, 1195 { "SCANFLIKE", true, scanflike }, 1196 { "VARARGS", true, varargs }, 1197 }; 1198 char keywd[32]; 1199 char arg[32]; 1200 size_t l, i; 1201 int a; 1202 bool eoc; 1203 1204 eoc = false; 1205 1206 /* Skip whitespace after the start of the comment */ 1207 while ((c = inpc()) != EOF && isspace(c)) 1208 continue; 1209 1210 /* Read the potential keyword to keywd */ 1211 l = 0; 1212 while (c != EOF && isupper(c) && l < sizeof(keywd) - 1) { 1213 keywd[l++] = (char)c; 1214 c = inpc(); 1215 } 1216 keywd[l] = '\0'; 1217 1218 /* look for the keyword */ 1219 for (i = 0; i < sizeof(keywtab) / sizeof(keywtab[0]); i++) { 1220 if (strcmp(keywtab[i].keywd, keywd) == 0) 1221 break; 1222 } 1223 if (i == sizeof(keywtab) / sizeof(keywtab[0])) 1224 goto skip_rest; 1225 1226 /* skip whitespace after the keyword */ 1227 while (c != EOF && isspace(c)) 1228 c = inpc(); 1229 1230 /* read the argument, if the keyword accepts one and there is one */ 1231 l = 0; 1232 if (keywtab[i].arg) { 1233 while (c != EOF && isdigit(c) && l < sizeof(arg) - 1) { 1234 arg[l++] = (char)c; 1235 c = inpc(); 1236 } 1237 } 1238 arg[l] = '\0'; 1239 a = l != 0 ? atoi(arg) : -1; 1240 1241 /* skip whitespace after the argument */ 1242 while (c != EOF && isspace(c)) 1243 c = inpc(); 1244 1245 if (c != '*' || (c = inpc()) != '/') { 1246 if (keywtab[i].func != linted) 1247 /* extra characters in lint comment */ 1248 warning(257); 1249 } else { 1250 /* 1251 * remember that we have already found the end of the 1252 * comment 1253 */ 1254 eoc = true; 1255 } 1256 1257 if (keywtab[i].func != NULL) 1258 (*keywtab[i].func)(a); 1259 1260 skip_rest: 1261 while (!eoc) { 1262 lc = c; 1263 if ((c = inpc()) == EOF) { 1264 /* unterminated comment */ 1265 error(256); 1266 break; 1267 } 1268 if (lc == '*' && c == '/') 1269 eoc = true; 1270 } 1271 } 1272 1273 /* 1274 * Handle // style comments 1275 */ 1276 void 1277 lex_slash_slash_comment(void) 1278 { 1279 int c; 1280 1281 if (!Sflag && !gflag) 1282 /* %s C does not support // comments */ 1283 gnuism(312, tflag ? "traditional" : "ANSI"); 1284 1285 while ((c = inpc()) != EOF && c != '\n') 1286 continue; 1287 } 1288 1289 /* 1290 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1291 * clear_warn_flags() is called after function definitions and global and 1292 * local declarations and definitions. It is also called between 1293 * the controlling expression and the body of control statements 1294 * (if, switch, for, while). 1295 */ 1296 void 1297 clear_warn_flags(void) 1298 { 1299 1300 lwarn = LWARN_ALL; 1301 quadflg = false; 1302 constcond_flag = false; 1303 } 1304 1305 /* 1306 * Strings are stored in a dynamically allocated buffer and passed 1307 * in yylval.y_xstrg to the parser. The parser or the routines called 1308 * by the parser are responsible for freeing this buffer. 1309 */ 1310 int 1311 lex_string(void) 1312 { 1313 u_char *s; 1314 int c; 1315 size_t len, max; 1316 strg_t *strg; 1317 1318 s = xmalloc(max = 64); 1319 1320 len = 0; 1321 while ((c = get_escaped_char('"')) >= 0) { 1322 /* +1 to reserve space for a trailing NUL character */ 1323 if (len + 1 == max) 1324 s = xrealloc(s, max *= 2); 1325 s[len++] = (char)c; 1326 } 1327 s[len] = '\0'; 1328 if (c == -2) 1329 /* unterminated string constant */ 1330 error(258); 1331 1332 strg = xcalloc(1, sizeof(*strg)); 1333 strg->st_tspec = CHAR; 1334 strg->st_len = len; 1335 strg->st_cp = s; 1336 1337 yylval.y_string = strg; 1338 return T_STRING; 1339 } 1340 1341 int 1342 lex_wide_string(void) 1343 { 1344 char *s; 1345 int c, n; 1346 size_t i, wi; 1347 size_t len, max, wlen; 1348 wchar_t *ws; 1349 strg_t *strg; 1350 1351 s = xmalloc(max = 64); 1352 len = 0; 1353 while ((c = get_escaped_char('"')) >= 0) { 1354 /* +1 to save space for a trailing NUL character */ 1355 if (len + 1 >= max) 1356 s = xrealloc(s, max *= 2); 1357 s[len++] = (char)c; 1358 } 1359 s[len] = '\0'; 1360 if (c == -2) 1361 /* unterminated string constant */ 1362 error(258); 1363 1364 /* get length of wide-character string */ 1365 (void)mblen(NULL, 0); 1366 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1367 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1368 /* invalid multibyte character */ 1369 error(291); 1370 break; 1371 } 1372 if (n == 0) 1373 n = 1; 1374 } 1375 1376 ws = xmalloc((wlen + 1) * sizeof(*ws)); 1377 1378 /* convert from multibyte to wide char */ 1379 (void)mbtowc(NULL, NULL, 0); 1380 for (i = 0, wi = 0; i < len; i += n, wi++) { 1381 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1382 break; 1383 if (n == 0) 1384 n = 1; 1385 } 1386 ws[wi] = 0; 1387 free(s); 1388 1389 strg = xcalloc(1, sizeof(*strg)); 1390 strg->st_tspec = WCHAR; 1391 strg->st_len = wlen; 1392 strg->st_wcp = ws; 1393 1394 yylval.y_string = strg; 1395 return T_STRING; 1396 } 1397 1398 /* 1399 * As noted above the scanner does not create new symbol table entries 1400 * for symbols it cannot find in the symbol table. This is to avoid 1401 * putting undeclared symbols into the symbol table if a syntax error 1402 * occurs. 1403 * 1404 * getsym() is called as soon as it is probably ok to put the symbol to 1405 * the symbol table. This does not mean that it is not possible that 1406 * symbols are put to the symbol table which are not completely 1407 * declared due to syntax errors. To avoid too many problems in this 1408 * case, symbols get type int in getsym(). 1409 * 1410 * XXX calls to getsym() should be delayed until decl1*() is called. 1411 */ 1412 sym_t * 1413 getsym(sbuf_t *sb) 1414 { 1415 dinfo_t *di; 1416 char *s; 1417 sym_t *sym; 1418 1419 sym = sb->sb_sym; 1420 1421 /* 1422 * During member declaration it is possible that name() looked 1423 * for symbols of type FVFT, although it should have looked for 1424 * symbols of type FTAG. Same can happen for labels. Both cases 1425 * are compensated here. 1426 */ 1427 if (symtyp == FMEMBER || symtyp == FLABEL) { 1428 if (sym == NULL || sym->s_kind == FVFT) 1429 sym = search(sb); 1430 } 1431 1432 if (sym != NULL) { 1433 if (sym->s_kind != symtyp) 1434 INTERNAL_ERROR("getsym(%d, %d)", sym->s_kind, symtyp); 1435 symtyp = FVFT; 1436 freesb(sb); 1437 return sym; 1438 } 1439 1440 /* create a new symbol table entry */ 1441 1442 /* labels must always be allocated at level 1 (outermost block) */ 1443 if (symtyp == FLABEL) { 1444 sym = getlblk(1, sizeof(*sym)); 1445 s = getlblk(1, sb->sb_len + 1); 1446 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1447 sym->s_name = s; 1448 sym->s_block_level = 1; 1449 di = dcs; 1450 while (di->d_next != NULL && di->d_next->d_next != NULL) 1451 di = di->d_next; 1452 lint_assert(di->d_ctx == AUTO); 1453 } else { 1454 sym = getblk(sizeof(*sym)); 1455 sym->s_name = sb->sb_name; 1456 sym->s_block_level = block_level; 1457 di = dcs; 1458 } 1459 1460 UNIQUE_CURR_POS(sym->s_def_pos); 1461 if ((sym->s_kind = symtyp) != FLABEL) 1462 sym->s_type = gettyp(INT); 1463 1464 symtyp = FVFT; 1465 1466 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1467 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1468 sym->s_rlink = &symtab[sb->sb_hash]; 1469 symtab[sb->sb_hash] = sym; 1470 1471 *di->d_ldlsym = sym; 1472 di->d_ldlsym = &sym->s_dlnxt; 1473 1474 freesb(sb); 1475 return sym; 1476 } 1477 1478 /* 1479 * Construct a temporary symbol. The symbol starts with a digit, so that 1480 * it is illegal. 1481 */ 1482 sym_t * 1483 mktempsym(type_t *t) 1484 { 1485 static int n = 0; 1486 int h; 1487 char *s = getlblk(block_level, 64); 1488 sym_t *sym = getblk(sizeof(*sym)); 1489 1490 (void)snprintf(s, 64, "%.8d_tmp", n++); 1491 h = hash(s); 1492 1493 sym->s_name = s; 1494 sym->s_type = t; 1495 sym->s_block_level = block_level; 1496 sym->s_scl = AUTO; 1497 sym->s_kind = FVFT; 1498 sym->s_used = true; 1499 sym->s_set = true; 1500 1501 if ((sym->s_link = symtab[h]) != NULL) 1502 symtab[h]->s_rlink = &sym->s_link; 1503 sym->s_rlink = &symtab[h]; 1504 symtab[h] = sym; 1505 1506 *dcs->d_ldlsym = sym; 1507 dcs->d_ldlsym = &sym->s_dlnxt; 1508 1509 return sym; 1510 } 1511 1512 /* 1513 * Remove a symbol forever from the symbol table. s_block_level 1514 * is set to -1 to avoid that the symbol will later be put 1515 * back to the symbol table. 1516 */ 1517 void 1518 rmsym(sym_t *sym) 1519 { 1520 1521 if ((*sym->s_rlink = sym->s_link) != NULL) 1522 sym->s_link->s_rlink = sym->s_rlink; 1523 sym->s_block_level = -1; 1524 sym->s_link = NULL; 1525 } 1526 1527 /* 1528 * Remove a list of symbols declared at one level from the symbol 1529 * table. 1530 */ 1531 void 1532 rmsyms(sym_t *syms) 1533 { 1534 sym_t *sym; 1535 1536 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1537 if (sym->s_block_level != -1) { 1538 if ((*sym->s_rlink = sym->s_link) != NULL) 1539 sym->s_link->s_rlink = sym->s_rlink; 1540 sym->s_link = NULL; 1541 sym->s_rlink = NULL; 1542 } 1543 } 1544 } 1545 1546 /* 1547 * Put a symbol into the symbol table. 1548 */ 1549 void 1550 inssym(int bl, sym_t *sym) 1551 { 1552 int h; 1553 1554 h = hash(sym->s_name); 1555 if ((sym->s_link = symtab[h]) != NULL) 1556 symtab[h]->s_rlink = &sym->s_link; 1557 sym->s_rlink = &symtab[h]; 1558 symtab[h] = sym; 1559 sym->s_block_level = bl; 1560 lint_assert(sym->s_link == NULL || 1561 sym->s_block_level >= sym->s_link->s_block_level); 1562 } 1563 1564 /* 1565 * Called at level 0 after syntax errors. 1566 * 1567 * Removes all symbols which are not declared at level 0 from the 1568 * symbol table. Also frees all memory which is not associated with 1569 * level 0. 1570 */ 1571 void 1572 cleanup(void) 1573 { 1574 sym_t *sym, *nsym; 1575 int i; 1576 1577 for (i = 0; i < HSHSIZ1; i++) { 1578 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1579 nsym = sym->s_link; 1580 if (sym->s_block_level >= 1) { 1581 if ((*sym->s_rlink = nsym) != NULL) 1582 nsym->s_rlink = sym->s_rlink; 1583 } 1584 } 1585 } 1586 1587 for (i = mem_block_level; i > 0; i--) 1588 freelblk(i); 1589 } 1590 1591 /* 1592 * Create a new symbol with the name of an existing symbol. 1593 */ 1594 sym_t * 1595 pushdown(const sym_t *sym) 1596 { 1597 int h; 1598 sym_t *nsym; 1599 1600 h = hash(sym->s_name); 1601 nsym = getblk(sizeof(*nsym)); 1602 lint_assert(sym->s_block_level <= block_level); 1603 nsym->s_name = sym->s_name; 1604 UNIQUE_CURR_POS(nsym->s_def_pos); 1605 nsym->s_kind = sym->s_kind; 1606 nsym->s_block_level = block_level; 1607 1608 if ((nsym->s_link = symtab[h]) != NULL) 1609 symtab[h]->s_rlink = &nsym->s_link; 1610 nsym->s_rlink = &symtab[h]; 1611 symtab[h] = nsym; 1612 1613 *dcs->d_ldlsym = nsym; 1614 dcs->d_ldlsym = &nsym->s_dlnxt; 1615 1616 return nsym; 1617 } 1618 1619 /* 1620 * Free any dynamically allocated memory referenced by 1621 * the value stack or yylval. 1622 * The type of information in yylval is described by tok. 1623 */ 1624 void 1625 freeyyv(void *sp, int tok) 1626 { 1627 if (tok == T_NAME || tok == T_TYPENAME) { 1628 sbuf_t *sb = *(sbuf_t **)sp; 1629 freesb(sb); 1630 } else if (tok == T_CON) { 1631 val_t *val = *(val_t **)sp; 1632 free(val); 1633 } else if (tok == T_STRING) { 1634 strg_t *strg = *(strg_t **)sp; 1635 if (strg->st_tspec == CHAR) { 1636 free(strg->st_cp); 1637 } else { 1638 lint_assert(strg->st_tspec == WCHAR); 1639 free(strg->st_wcp); 1640 } 1641 free(strg); 1642 } 1643 } 1644
Indexes created Mon Sep 29 21:09:56 GMT 2025