scan.l revision 1.54
1 %{ 2 /* $NetBSD: scan.l,v 1.54 2014/02/18 22:01:36 christos Exp $ */ 3 4 /* 5 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved. 6 * Copyright (c) 1994, 1995 Jochen Pohl 7 * All Rights Reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed by Jochen Pohl for 20 * The NetBSD Project. 21 * 4. The name of the author may not be used to endorse or promote products 22 * derived from this software without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 27 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 28 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 29 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 30 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 31 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 32 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 33 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 #include <sys/cdefs.h> 37 #if defined(__RCSID) && !defined(lint) 38 __RCSID("$NetBSD: scan.l,v 1.54 2014/02/18 22:01:36 christos Exp $"); 39 #endif 40 41 #include <stdlib.h> 42 #include <string.h> 43 #include <limits.h> 44 #include <float.h> 45 #include <ctype.h> 46 #include <errno.h> 47 #include <math.h> 48 49 #include "lint1.h" 50 #include "cgram.h" 51 52 #define CHAR_MASK (~(~0 << CHAR_BIT)) 53 54 /* Current position (its also updated when an included file is parsed) */ 55 pos_t curr_pos = { 1, "", 0 }; 56 57 /* 58 * Current position in C source (not updated when an included file is 59 * parsed). 60 */ 61 pos_t csrc_pos = { 1, "", 0 }; 62 63 static void incline(void); 64 static void badchar(int); 65 static sbuf_t *allocsb(void); 66 static void freesb(sbuf_t *); 67 static int inpc(void); 68 static int hash(const char *); 69 static sym_t *search(sbuf_t *); 70 static int name(void); 71 static int keyw(sym_t *); 72 static int icon(int); 73 static int fcon(void); 74 static int operator(int, op_t); 75 static int ccon(void); 76 static int wccon(void); 77 static int getescc(int); 78 static void directive(void); 79 static void comment(void); 80 static void slashslashcomment(void); 81 static int string(void); 82 static int wcstrg(void); 83 84 %} 85 86 L [_A-Za-z] 87 D [0-9] 88 NZD [1-9] 89 OD [0-7] 90 HD [0-9A-Fa-f] 91 EX ([eE][+-]?[0-9]+) 92 HX (p[+-]?[0-9A-Fa-f]+) 93 TL ([fFlL]?[i]?) 94 95 %option nounput 96 97 %% 98 99 {L}({L}|{D})* return (name()); 100 0{OD}*[lLuU]* return (icon(8)); 101 {NZD}{D}*[lLuU]* return (icon(10)); 102 0[xX]{HD}+[lLuU]* return (icon(16)); 103 {D}+\.{D}*{EX}?{TL} | 104 {D}+{EX}{TL} | 105 0[xX]{HD}+\.{HD}*{HX}{TL} | 106 0[xX]{HD}+{HX}{TL} | 107 \.{D}+{EX}?{TL} return (fcon()); 108 "=" return (operator(T_ASSIGN, ASSIGN)); 109 "*=" return (operator(T_OPASS, MULASS)); 110 "/=" return (operator(T_OPASS, DIVASS)); 111 "%=" return (operator(T_OPASS, MODASS)); 112 "+=" return (operator(T_OPASS, ADDASS)); 113 "-=" return (operator(T_OPASS, SUBASS)); 114 "<<=" return (operator(T_OPASS, SHLASS)); 115 ">>=" return (operator(T_OPASS, SHRASS)); 116 "&=" return (operator(T_OPASS, ANDASS)); 117 "^=" return (operator(T_OPASS, XORASS)); 118 "|=" return (operator(T_OPASS, ORASS)); 119 "||" return (operator(T_LOGOR, LOGOR)); 120 "&&" return (operator(T_LOGAND, LOGAND)); 121 "|" return (operator(T_OR, OR)); 122 "&" return (operator(T_AND, AND)); 123 "^" return (operator(T_XOR, XOR)); 124 "==" return (operator(T_EQOP, EQ)); 125 "!=" return (operator(T_EQOP, NE)); 126 "<" return (operator(T_RELOP, LT)); 127 ">" return (operator(T_RELOP, GT)); 128 "<=" return (operator(T_RELOP, LE)); 129 ">=" return (operator(T_RELOP, GE)); 130 "<<" return (operator(T_SHFTOP, SHL)); 131 ">>" return (operator(T_SHFTOP, SHR)); 132 "++" return (operator(T_INCDEC, INC)); 133 "--" return (operator(T_INCDEC, DEC)); 134 "->" return (operator(T_STROP, ARROW)); 135 "." return (operator(T_STROP, POINT)); 136 "+" return (operator(T_ADDOP, PLUS)); 137 "-" return (operator(T_ADDOP, MINUS)); 138 "*" return (operator(T_MULT, MULT)); 139 "/" return (operator(T_DIVOP, DIV)); 140 "%" return (operator(T_DIVOP, MOD)); 141 "!" return (operator(T_UNOP, NOT)); 142 "~" return (operator(T_UNOP, COMPL)); 143 "\"" return (string()); 144 "L\"" return (wcstrg()); 145 ";" return (T_SEMI); 146 "{" return (T_LBRACE); 147 "}" return (T_RBRACE); 148 "," return (T_COMMA); 149 ":" return (T_COLON); 150 "?" return (T_QUEST); 151 "[" return (T_LBRACK); 152 "]" return (T_RBRACK); 153 "(" return (T_LPARN); 154 ")" return (T_RPARN); 155 "..." return (T_ELLIPSE); 156 "'" return (ccon()); 157 "L'" return (wccon()); 158 ^#.*$ directive(); 159 \n incline(); 160 \t|" "|\f|\v ; 161 "/*" comment(); 162 "//" slashslashcomment(); 163 . badchar(yytext[0]); 164 165 %% 166 167 static void 168 incline(void) 169 { 170 curr_pos.p_line++; 171 curr_pos.p_uniq = 0; 172 if (curr_pos.p_file == csrc_pos.p_file) { 173 csrc_pos.p_line++; 174 csrc_pos.p_uniq = 0; 175 } 176 } 177 178 static void 179 badchar(int c) 180 { 181 182 /* unknown character \%o */ 183 error(250, c); 184 } 185 186 /* 187 * Keywords. 188 * During initialisation they are written to the symbol table. 189 */ 190 static struct kwtab { 191 const char *kw_name; /* keyword */ 192 int kw_token; /* token returned by yylex() */ 193 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 194 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 195 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 196 u_int kw_c89; /* c89 keyword */ 197 u_int kw_c99; /* c99 keyword */ 198 u_int kw_gcc; /* GCC keyword */ 199 } kwtab[] = { 200 { "__alignof__", T_ALIGNOF, 0, 0, 0, 0, 0, 0 }, 201 { "__attribute__",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1 }, 202 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0, 0, 1 }, 203 { "__packed__", T_AT_PACKED, 0, 0, 0, 0, 0, 1 }, 204 { "packed", T_AT_PACKED, 0, 0, 0, 0, 0, 1 }, 205 { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1 }, 206 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1 }, 207 { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1 }, 208 { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1 }, 209 { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1 }, 210 { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1 }, 211 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1 }, 212 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1 }, 213 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1 }, 214 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1 }, 215 { "asm", T_ASM, 0, 0, 0, 0, 0, 1 }, 216 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0 }, 217 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0 }, 218 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0 }, 219 { "break", T_BREAK, 0, 0, 0, 0, 0, 0 }, 220 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0 }, 221 { "case", T_CASE, 0, 0, 0, 0, 0, 0 }, 222 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0 }, 223 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0 }, 224 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0 }, 225 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0 }, 226 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0 }, 227 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0 }, 228 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0 }, 229 { "do", T_DO, 0, 0, 0, 0, 0, 0 }, 230 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0 }, 231 { "else", T_ELSE, 0, 0, 0, 0, 0, 0 }, 232 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0 }, 233 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1 }, 234 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0 }, 235 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0 }, 236 { "for", T_FOR, 0, 0, 0, 0, 0, 0 }, 237 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0 }, 238 { "if", T_IF, 0, 0, 0, 0, 0, 0 }, 239 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0 }, 240 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0 }, 241 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0 }, 242 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0 }, 243 { "int", T_TYPE, 0, INT, 0, 0, 0, 0 }, 244 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0 }, 245 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0 }, 246 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0 }, 247 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0 }, 248 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0 }, 249 { "return", T_RETURN, 0, 0, 0, 0, 0, 0 }, 250 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0 }, 251 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0 }, 252 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0 }, 253 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0 }, 254 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0 }, 255 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0 }, 256 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0 }, 257 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0 }, 258 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0 }, 259 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0 }, 260 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1 }, 261 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1 }, 262 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1 }, 263 { "union", T_SOU, 0, UNION, 0, 0, 0, 0 }, 264 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0 }, 265 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0 }, 266 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0 }, 267 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0 }, 268 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0 }, 269 { "while", T_WHILE, 0, 0, 0, 0, 0, 0 }, 270 { NULL, 0, 0, 0, 0, 0, 0, 0 } 271 }; 272 273 /* Symbol table */ 274 static sym_t *symtab[HSHSIZ1]; 275 276 /* bit i of the entry with index i is set */ 277 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 278 279 /* least significant i bits are set in the entry with index i */ 280 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 281 282 /* least significant i bits are not set in the entry with index i */ 283 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 284 285 /* free list for sbuf structures */ 286 static sbuf_t *sbfrlst; 287 288 /* Typ of next expected symbol */ 289 symt_t symtyp; 290 291 292 /* 293 * All keywords are written to the symbol table. This saves us looking 294 * in a extra table for each name we found. 295 */ 296 void 297 initscan(void) 298 { 299 struct kwtab *kw; 300 sym_t *sym; 301 size_t h, i; 302 uint64_t uq; 303 304 for (kw = kwtab; kw->kw_name != NULL; kw++) { 305 if ((kw->kw_c89 || kw->kw_c99) && tflag) 306 continue; 307 if (kw->kw_c99 && !(Sflag || gflag)) 308 continue; 309 if (kw->kw_gcc && !gflag) 310 continue; 311 sym = getblk(sizeof (sym_t)); 312 sym->s_name = kw->kw_name; 313 sym->s_keyw = 1; 314 sym->s_value.v_quad = kw->kw_token; 315 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 316 sym->s_tspec = kw->kw_tspec; 317 } else if (kw->kw_token == T_SCLASS) { 318 sym->s_scl = kw->kw_scl; 319 } else if (kw->kw_token == T_QUAL) { 320 sym->s_tqual = kw->kw_tqual; 321 } 322 h = hash(sym->s_name); 323 if ((sym->s_link = symtab[h]) != NULL) 324 symtab[h]->s_rlink = &sym->s_link; 325 sym->s_rlink = &symtab[h]; 326 symtab[h] = sym; 327 } 328 329 /* initialize bit-masks for quads */ 330 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 331 qbmasks[i] = (uint64_t)1 << i; 332 uq = ~(uint64_t)0 << i; 333 qumasks[i] = uq; 334 qlmasks[i] = ~uq; 335 } 336 qumasks[i] = 0; 337 qlmasks[i] = ~(uint64_t)0; 338 } 339 340 /* 341 * Get a free sbuf structure, if possible from the free list 342 */ 343 static sbuf_t * 344 allocsb(void) 345 { 346 sbuf_t *sb; 347 348 if ((sb = sbfrlst) != NULL) { 349 sbfrlst = sb->sb_nxt; 350 } else { 351 sb = xmalloc(sizeof (sbuf_t)); 352 } 353 (void)memset(sb, 0, sizeof (*sb)); 354 return (sb); 355 } 356 357 /* 358 * Put a sbuf structure to the free list 359 */ 360 static void 361 freesb(sbuf_t *sb) 362 { 363 364 sb->sb_nxt = sbfrlst; 365 sbfrlst = sb; 366 } 367 368 /* 369 * Read a character and ensure that it is positive (except EOF). 370 * Increment line count(s) if necessary. 371 */ 372 static int 373 inpc(void) 374 { 375 int c; 376 377 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 378 incline(); 379 return (c); 380 } 381 382 static int 383 hash(const char *s) 384 { 385 u_int v; 386 const u_char *us; 387 388 v = 0; 389 for (us = (const u_char *)s; *us != '\0'; us++) { 390 v = (v << sizeof (v)) + *us; 391 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 392 } 393 return (v % HSHSIZ1); 394 } 395 396 /* 397 * Lex has found a letter followed by zero or more letters or digits. 398 * It looks for a symbol in the symbol table with the same name. This 399 * symbol must either be a keyword or a symbol of the type required by 400 * symtyp (label, member, tag, ...). 401 * 402 * If it is a keyword, the token is returned. In some cases it is described 403 * more deeply by data written to yylval. 404 * 405 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 406 * is stored in yylval. This struct contains the name of the symbol, it's 407 * length and hash value. If there is already a symbol of the same name 408 * and type in the symbol table, the sbuf struct also contains a pointer 409 * to the symbol table entry. 410 */ 411 static int 412 name(void) 413 { 414 char *s; 415 sbuf_t *sb; 416 sym_t *sym; 417 int tok; 418 419 sb = allocsb(); 420 sb->sb_name = yytext; 421 sb->sb_len = yyleng; 422 sb->sb_hash = hash(yytext); 423 if ((sym = search(sb)) != NULL && sym->s_keyw) { 424 freesb(sb); 425 return (keyw(sym)); 426 } 427 428 sb->sb_sym = sym; 429 430 if (sym != NULL) { 431 if (blklev < sym->s_blklev) 432 LERROR("name()"); 433 sb->sb_name = sym->s_name; 434 sb->sb_len = strlen(sym->s_name); 435 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 436 } else { 437 s = getblk(yyleng + 1); 438 (void)memcpy(s, yytext, yyleng + 1); 439 sb->sb_name = s; 440 sb->sb_len = yyleng; 441 tok = T_NAME; 442 } 443 444 yylval.y_sb = sb; 445 return (tok); 446 } 447 448 static sym_t * 449 search(sbuf_t *sb) 450 { 451 sym_t *sym; 452 453 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 454 if (strcmp(sym->s_name, sb->sb_name) == 0) { 455 if (sym->s_keyw || sym->s_kind == symtyp) 456 return (sym); 457 } 458 } 459 460 return (NULL); 461 } 462 463 static int 464 keyw(sym_t *sym) 465 { 466 int t; 467 468 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 469 yylval.y_scl = sym->s_scl; 470 } else if (t == T_TYPE || t == T_SOU) { 471 yylval.y_tspec = sym->s_tspec; 472 } else if (t == T_QUAL) { 473 yylval.y_tqual = sym->s_tqual; 474 } 475 return (t); 476 } 477 478 /* 479 * Convert a string representing an integer into internal representation. 480 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 481 */ 482 static int 483 icon(int base) 484 { 485 int l_suffix, u_suffix; 486 int len; 487 const char *cp; 488 char c, *eptr; 489 tspec_t typ; 490 uint64_t uq = 0; 491 int ansiu; 492 static tspec_t contypes[2][3] = { 493 { INT, LONG, QUAD }, 494 { UINT, ULONG, UQUAD } 495 }; 496 497 cp = yytext; 498 len = yyleng; 499 500 /* skip 0x */ 501 if (base == 16) { 502 cp += 2; 503 len -= 2; 504 } 505 506 /* read suffixes */ 507 l_suffix = u_suffix = 0; 508 for ( ; ; ) { 509 if ((c = cp[len - 1]) == 'l' || c == 'L') { 510 l_suffix++; 511 } else if (c == 'u' || c == 'U') { 512 u_suffix++; 513 } else { 514 break; 515 } 516 len--; 517 } 518 if (l_suffix > 2 || u_suffix > 1) { 519 /* malformed integer constant */ 520 warning(251); 521 if (l_suffix > 2) 522 l_suffix = 2; 523 if (u_suffix > 1) 524 u_suffix = 1; 525 } 526 if (tflag && u_suffix != 0) { 527 /* suffix U is illegal in traditional C */ 528 warning(97); 529 } 530 typ = contypes[u_suffix][l_suffix]; 531 532 errno = 0; 533 534 uq = strtouq(cp, &eptr, base); 535 if (eptr != cp + len) 536 LERROR("icon()"); 537 if (errno != 0) 538 /* integer constant out of range */ 539 warning(252); 540 541 /* 542 * If the value is too big for the current type, we must choose 543 * another type. 544 */ 545 ansiu = 0; 546 switch (typ) { 547 case INT: 548 if (uq <= TARG_INT_MAX) { 549 /* ok */ 550 } else if (uq <= TARG_UINT_MAX && base != 10) { 551 typ = UINT; 552 } else if (uq <= TARG_LONG_MAX) { 553 typ = LONG; 554 } else { 555 typ = ULONG; 556 if (uq > TARG_ULONG_MAX) { 557 /* integer constant out of range */ 558 warning(252); 559 } 560 } 561 if (typ == UINT || typ == ULONG) { 562 if (tflag) { 563 typ = LONG; 564 } else if (!sflag) { 565 /* 566 * Remember that the constant is unsigned 567 * only in ANSI C 568 */ 569 ansiu = 1; 570 } 571 } 572 break; 573 case UINT: 574 if (uq > TARG_UINT_MAX) { 575 typ = ULONG; 576 if (uq > TARG_ULONG_MAX) { 577 /* integer constant out of range */ 578 warning(252); 579 } 580 } 581 break; 582 case LONG: 583 if (uq > TARG_LONG_MAX && !tflag) { 584 typ = ULONG; 585 if (!sflag) 586 ansiu = 1; 587 if (uq > TARG_ULONG_MAX) { 588 /* integer constant out of range */ 589 warning(252); 590 } 591 } 592 break; 593 case ULONG: 594 if (uq > TARG_ULONG_MAX) { 595 /* integer constant out of range */ 596 warning(252); 597 } 598 break; 599 case QUAD: 600 if (uq > TARG_QUAD_MAX && !tflag) { 601 typ = UQUAD; 602 if (!sflag) 603 ansiu = 1; 604 } 605 break; 606 case UQUAD: 607 if (uq > TARG_UQUAD_MAX) { 608 /* integer constant out of range */ 609 warning(252); 610 } 611 break; 612 /* LINTED206: (enumeration values not handled in switch) */ 613 case STRUCT: 614 case VOID: 615 case LDOUBLE: 616 case FUNC: 617 case ARRAY: 618 case PTR: 619 case ENUM: 620 case UNION: 621 case SIGNED: 622 case NOTSPEC: 623 case DOUBLE: 624 case FLOAT: 625 case USHORT: 626 case SHORT: 627 case UCHAR: 628 case SCHAR: 629 case CHAR: 630 case BOOL: 631 case UNSIGN: 632 case FCOMPLEX: 633 case DCOMPLEX: 634 case LCOMPLEX: 635 case COMPLEX: 636 break; 637 638 case NTSPEC: /* this value unused */ 639 break; 640 } 641 642 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 643 644 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 645 yylval.y_val->v_ansiu = ansiu; 646 yylval.y_val->v_quad = (int64_t)uq; 647 648 return (T_CON); 649 } 650 651 /* 652 * Returns 1 if t is a signed type and the value is negative. 653 * 654 * len is the number of significant bits. If len is -1, len is set 655 * to the width of type t. 656 */ 657 int 658 sign(int64_t q, tspec_t t, int len) 659 { 660 661 if (t == PTR || isutyp(t)) 662 return (0); 663 return (msb(q, t, len)); 664 } 665 666 int 667 msb(int64_t q, tspec_t t, int len) 668 { 669 670 if (len <= 0) 671 len = size(t); 672 return ((q & qbmasks[len - 1]) != 0); 673 } 674 675 /* 676 * Extends the sign of q. 677 */ 678 int64_t 679 xsign(int64_t q, tspec_t t, int len) 680 { 681 682 if (len <= 0) 683 len = size(t); 684 685 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 686 q &= qlmasks[len]; 687 } else { 688 q |= qumasks[len]; 689 } 690 return (q); 691 } 692 693 /* 694 * Convert a string representing a floating point value into its interal 695 * representation. Type and value are returned in yylval. fcon() 696 * (and yylex()) returns T_CON. 697 * XXX Currently it is not possible to convert constants of type 698 * long double which are greater than DBL_MAX. 699 */ 700 static int 701 fcon(void) 702 { 703 const char *cp; 704 int len; 705 tspec_t typ; 706 char c, *eptr; 707 double d; 708 float f = 0; 709 710 cp = yytext; 711 len = yyleng; 712 713 if (cp[len - 1] == 'i') { 714 /* imaginary, do nothing for now */ 715 len--; 716 } 717 if ((c = cp[len - 1]) == 'f' || c == 'F') { 718 typ = FLOAT; 719 len--; 720 } else if (c == 'l' || c == 'L') { 721 typ = LDOUBLE; 722 len--; 723 } else { 724 if (c == 'd' || c == 'D') 725 len--; 726 typ = DOUBLE; 727 } 728 729 if (tflag && typ != DOUBLE) { 730 /* suffixes F and L are illegal in traditional C */ 731 warning(98); 732 } 733 734 errno = 0; 735 d = strtod(cp, &eptr); 736 if (eptr != cp + len) { 737 switch (*eptr) { 738 /* 739 * XXX: non-native non-current strtod() may not handle hex 740 * floats, ignore the rest if we find traces of hex float 741 * syntax... 742 */ 743 case 'p': 744 case 'P': 745 case 'x': 746 case 'X': 747 d = 0; 748 errno = 0; 749 break; 750 default: 751 LERROR("fcon(%s->%s)", cp, eptr); 752 } 753 } 754 if (errno != 0) 755 /* floating-point constant out of range */ 756 warning(248); 757 758 if (typ == FLOAT) { 759 f = (float)d; 760 if (!finite(f)) { 761 /* floating-point constant out of range */ 762 warning(248); 763 f = f > 0 ? FLT_MAX : -FLT_MAX; 764 } 765 } 766 767 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 768 if (typ == FLOAT) { 769 yylval.y_val->v_ldbl = f; 770 } else { 771 yylval.y_val->v_ldbl = d; 772 } 773 774 return (T_CON); 775 } 776 777 static int 778 operator(int t, op_t o) 779 { 780 781 yylval.y_op = o; 782 return (t); 783 } 784 785 /* 786 * Called if lex found a leading \'. 787 */ 788 static int 789 ccon(void) 790 { 791 size_t n; 792 int val, c; 793 char cv; 794 795 n = 0; 796 val = 0; 797 while ((c = getescc('\'')) >= 0) { 798 val = (val << CHAR_BIT) + c; 799 n++; 800 } 801 if (c == -2) { 802 /* unterminated character constant */ 803 error(253); 804 } else { 805 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 806 /* too many characters in character constant */ 807 error(71); 808 } else if (n > 1) { 809 /* multi-character character constant */ 810 warning(294); 811 } else if (n == 0) { 812 /* empty character constant */ 813 error(73); 814 } 815 } 816 if (n == 1) { 817 cv = (char)val; 818 val = cv; 819 } 820 821 yylval.y_val = xcalloc(1, sizeof (val_t)); 822 yylval.y_val->v_tspec = INT; 823 yylval.y_val->v_quad = val; 824 825 return (T_CON); 826 } 827 828 /* 829 * Called if lex found a leading L\' 830 */ 831 static int 832 wccon(void) 833 { 834 static char buf[MB_LEN_MAX + 1]; 835 size_t i; 836 int c; 837 wchar_t wc; 838 839 i = 0; 840 while ((c = getescc('\'')) >= 0) { 841 if (i < MB_CUR_MAX) 842 buf[i] = (char)c; 843 i++; 844 } 845 846 wc = 0; 847 848 if (c == -2) { 849 /* unterminated character constant */ 850 error(253); 851 } else if (c == 0) { 852 /* empty character constant */ 853 error(73); 854 } else { 855 if (i > MB_CUR_MAX) { 856 i = MB_CUR_MAX; 857 /* too many characters in character constant */ 858 error(71); 859 } else { 860 buf[i] = '\0'; 861 (void)mbtowc(NULL, NULL, 0); 862 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 863 /* invalid multibyte character */ 864 error(291); 865 } 866 } 867 868 yylval.y_val = xcalloc(1, sizeof (val_t)); 869 yylval.y_val->v_tspec = WCHAR; 870 yylval.y_val->v_quad = wc; 871 872 return (T_CON); 873 } 874 875 /* 876 * Read a character which is part of a character constant or of a string 877 * and handle escapes. 878 * 879 * The Argument is the character which delimits the character constant or 880 * string. 881 * 882 * Returns -1 if the end of the character constant or string is reached, 883 * -2 if the EOF is reached, and the character otherwise. 884 */ 885 static int 886 getescc(int d) 887 { 888 static int pbc = -1; 889 int n, c, v; 890 891 if (pbc == -1) { 892 c = inpc(); 893 } else { 894 c = pbc; 895 pbc = -1; 896 } 897 if (c == d) 898 return (-1); 899 switch (c) { 900 case '\n': 901 if (tflag) { 902 /* newline in string or char constant */ 903 error(254); 904 return (-2); 905 } 906 return (c); 907 case EOF: 908 return (-2); 909 case '\\': 910 switch (c = inpc()) { 911 case '"': 912 if (tflag && d == '\'') 913 /* \" inside character constant undef. ... */ 914 warning(262); 915 return ('"'); 916 case '\'': 917 return ('\''); 918 case '?': 919 if (tflag) 920 /* \? undefined in traditional C */ 921 warning(263); 922 return ('?'); 923 case '\\': 924 return ('\\'); 925 case 'a': 926 if (tflag) 927 /* \a undefined in traditional C */ 928 warning(81); 929 return ('\a'); 930 case 'b': 931 return ('\b'); 932 case 'f': 933 return ('\f'); 934 case 'n': 935 return ('\n'); 936 case 'r': 937 return ('\r'); 938 case 't': 939 return ('\t'); 940 case 'v': 941 if (tflag) 942 /* \v undefined in traditional C */ 943 warning(264); 944 return ('\v'); 945 case '8': case '9': 946 /* bad octal digit %c */ 947 warning(77, c); 948 /* FALLTHROUGH */ 949 case '0': case '1': case '2': case '3': 950 case '4': case '5': case '6': case '7': 951 n = 3; 952 v = 0; 953 do { 954 v = (v << 3) + (c - '0'); 955 c = inpc(); 956 } while (--n && isdigit(c) && (tflag || c <= '7')); 957 if (tflag && n > 0 && isdigit(c)) 958 /* bad octal digit %c */ 959 warning(77, c); 960 pbc = c; 961 if (v > TARG_UCHAR_MAX) { 962 /* character escape does not fit in char. */ 963 warning(76); 964 v &= CHAR_MASK; 965 } 966 return (v); 967 case 'x': 968 if (tflag) 969 /* \x undefined in traditional C */ 970 warning(82); 971 v = 0; 972 n = 0; 973 while ((c = inpc()) >= 0 && isxdigit(c)) { 974 c = isdigit(c) ? 975 c - '0' : toupper(c) - 'A' + 10; 976 v = (v << 4) + c; 977 if (n >= 0) { 978 if ((v & ~CHAR_MASK) != 0) { 979 /* overflow in hex escape */ 980 warning(75); 981 n = -1; 982 } else { 983 n++; 984 } 985 } 986 } 987 pbc = c; 988 if (n == 0) { 989 /* no hex digits follow \x */ 990 error(74); 991 } if (n == -1) { 992 v &= CHAR_MASK; 993 } 994 return (v); 995 case '\n': 996 return (getescc(d)); 997 case EOF: 998 return (-2); 999 default: 1000 if (isprint(c)) { 1001 /* dubious escape \%c */ 1002 warning(79, c); 1003 } else { 1004 /* dubious escape \%o */ 1005 warning(80, c); 1006 } 1007 } 1008 } 1009 return (c); 1010 } 1011 1012 /* 1013 * Called for preprocessor directives. Currently implemented are: 1014 * # lineno 1015 * # lineno "filename" 1016 */ 1017 static void 1018 directive(void) 1019 { 1020 const char *cp, *fn; 1021 char c, *eptr; 1022 size_t fnl; 1023 long ln; 1024 static int first = 1; 1025 1026 /* Go to first non-whitespace after # */ 1027 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1028 continue; 1029 1030 if (!isdigit((unsigned char)c)) { 1031 error: 1032 /* undefined or invalid # directive */ 1033 warning(255); 1034 return; 1035 } 1036 ln = strtol(--cp, &eptr, 10); 1037 if (cp == eptr) 1038 goto error; 1039 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1040 goto error; 1041 while ((c = *cp++) == ' ' || c == '\t') 1042 continue; 1043 if (c != '\0') { 1044 if (c != '"') 1045 goto error; 1046 fn = cp; 1047 while ((c = *cp) != '"' && c != '\0') 1048 cp++; 1049 if (c != '"') 1050 goto error; 1051 if ((fnl = cp++ - fn) > PATH_MAX) 1052 goto error; 1053 while ((c = *cp++) == ' ' || c == '\t') 1054 continue; 1055 #if 0 1056 if (c != '\0') 1057 warning("extra character(s) after directive"); 1058 #endif 1059 1060 /* empty string means stdin */ 1061 if (fnl == 0) { 1062 fn = "{standard input}"; 1063 fnl = 16; /* strlen (fn) */ 1064 } 1065 curr_pos.p_file = fnnalloc(fn, fnl); 1066 /* 1067 * If this is the first directive, the name is the name 1068 * of the C source file as specified at the command line. 1069 * It is written to the output file. 1070 */ 1071 if (first) { 1072 csrc_pos.p_file = curr_pos.p_file; 1073 outsrc(curr_pos.p_file); 1074 first = 0; 1075 } 1076 } 1077 curr_pos.p_line = (int)ln - 1; 1078 curr_pos.p_uniq = 0; 1079 if (curr_pos.p_file == csrc_pos.p_file) { 1080 csrc_pos.p_line = (int)ln - 1; 1081 csrc_pos.p_uniq = 0; 1082 } 1083 } 1084 1085 /* 1086 * Handle lint comments. Following comments are currently understood: 1087 * ARGSUSEDn 1088 * BITFIELDTYPE 1089 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1090 * FALLTHRU FALLTHROUGH 1091 * LINTLIBRARY 1092 * LINTEDn NOSTRICTn 1093 * LONGLONG 1094 * NOTREACHED 1095 * PRINTFLIKEn 1096 * PROTOLIB 1097 * SCANFLIKEn 1098 * VARARGSn 1099 * If one of this comments is recognized, the arguments, if any, are 1100 * parsed and a function which handles this comment is called. 1101 */ 1102 static void 1103 comment(void) 1104 { 1105 int c, lc; 1106 static struct { 1107 const char *keywd; 1108 int arg; 1109 void (*func)(int); 1110 } keywtab[] = { 1111 { "ARGSUSED", 1, argsused }, 1112 { "BITFIELDTYPE", 0, bitfieldtype }, 1113 { "CONSTCOND", 0, constcond }, 1114 { "CONSTANTCOND", 0, constcond }, 1115 { "CONSTANTCONDITION", 0, constcond }, 1116 { "FALLTHRU", 0, fallthru }, 1117 { "FALLTHROUGH", 0, fallthru }, 1118 { "LINTLIBRARY", 0, lintlib }, 1119 { "LINTED", 1, linted }, 1120 { "LONGLONG", 0, longlong }, 1121 { "NOSTRICT", 1, linted }, 1122 { "NOTREACHED", 0, notreach }, 1123 { "PRINTFLIKE", 1, printflike }, 1124 { "PROTOLIB", 1, protolib }, 1125 { "SCANFLIKE", 1, scanflike }, 1126 { "VARARGS", 1, varargs }, 1127 }; 1128 char keywd[32]; 1129 char arg[32]; 1130 size_t l, i; 1131 int a; 1132 int eoc; 1133 1134 eoc = 0; 1135 1136 /* Skip white spaces after the start of the comment */ 1137 while ((c = inpc()) != EOF && isspace(c)) 1138 continue; 1139 1140 /* Read the potential keyword to keywd */ 1141 l = 0; 1142 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1143 keywd[l++] = (char)c; 1144 c = inpc(); 1145 } 1146 keywd[l] = '\0'; 1147 1148 /* look for the keyword */ 1149 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1150 if (strcmp(keywtab[i].keywd, keywd) == 0) 1151 break; 1152 } 1153 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1154 goto skip_rest; 1155 1156 /* skip white spaces after the keyword */ 1157 while (c != EOF && isspace(c)) 1158 c = inpc(); 1159 1160 /* read the argument, if the keyword accepts one and there is one */ 1161 l = 0; 1162 if (keywtab[i].arg) { 1163 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1164 arg[l++] = (char)c; 1165 c = inpc(); 1166 } 1167 } 1168 arg[l] = '\0'; 1169 a = l != 0 ? atoi(arg) : -1; 1170 1171 /* skip white spaces after the argument */ 1172 while (c != EOF && isspace(c)) 1173 c = inpc(); 1174 1175 if (c != '*' || (c = inpc()) != '/') { 1176 if (keywtab[i].func != linted) 1177 /* extra characters in lint comment */ 1178 warning(257); 1179 } else { 1180 /* 1181 * remember that we have already found the end of the 1182 * comment 1183 */ 1184 eoc = 1; 1185 } 1186 1187 if (keywtab[i].func != NULL) 1188 (*keywtab[i].func)(a); 1189 1190 skip_rest: 1191 while (!eoc) { 1192 lc = c; 1193 if ((c = inpc()) == EOF) { 1194 /* unterminated comment */ 1195 error(256); 1196 break; 1197 } 1198 if (lc == '*' && c == '/') 1199 eoc = 1; 1200 } 1201 } 1202 1203 /* 1204 * Handle // style comments 1205 */ 1206 static void 1207 slashslashcomment(void) 1208 { 1209 int c; 1210 1211 if (!Sflag && !gflag) 1212 /* // comments only supported in C99 */ 1213 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1214 1215 while ((c = inpc()) != EOF && c != '\n') 1216 continue; 1217 } 1218 1219 /* 1220 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1221 * clrwflgs() is called after function definitions and global and 1222 * local declarations and definitions. It is also called between 1223 * the controlling expression and the body of control statements 1224 * (if, switch, for, while). 1225 */ 1226 void 1227 clrwflgs(void) 1228 { 1229 1230 lwarn = LWARN_ALL; 1231 quadflg = 0; 1232 ccflg = 0; 1233 } 1234 1235 /* 1236 * Strings are stored in a dynamically alloceted buffer and passed 1237 * in yylval.y_xstrg to the parser. The parser or the routines called 1238 * by the parser are responsible for freeing this buffer. 1239 */ 1240 static int 1241 string(void) 1242 { 1243 u_char *s; 1244 int c; 1245 size_t len, max; 1246 strg_t *strg; 1247 1248 s = xmalloc(max = 64); 1249 1250 len = 0; 1251 while ((c = getescc('"')) >= 0) { 1252 /* +1 to reserve space for a trailing NUL character */ 1253 if (len + 1 == max) 1254 s = xrealloc(s, max *= 2); 1255 s[len++] = (char)c; 1256 } 1257 s[len] = '\0'; 1258 if (c == -2) 1259 /* unterminated string constant */ 1260 error(258); 1261 1262 strg = xcalloc(1, sizeof (strg_t)); 1263 strg->st_tspec = CHAR; 1264 strg->st_len = len; 1265 strg->st_cp = s; 1266 1267 yylval.y_strg = strg; 1268 return (T_STRING); 1269 } 1270 1271 static int 1272 wcstrg(void) 1273 { 1274 char *s; 1275 int c, n; 1276 size_t i, wi; 1277 size_t len, max, wlen; 1278 wchar_t *ws; 1279 strg_t *strg; 1280 1281 s = xmalloc(max = 64); 1282 len = 0; 1283 while ((c = getescc('"')) >= 0) { 1284 /* +1 to save space for a trailing NUL character */ 1285 if (len + 1 >= max) 1286 s = xrealloc(s, max *= 2); 1287 s[len++] = (char)c; 1288 } 1289 s[len] = '\0'; 1290 if (c == -2) 1291 /* unterminated string constant */ 1292 error(258); 1293 1294 /* get length of wide-character string */ 1295 (void)mblen(NULL, 0); 1296 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1297 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1298 /* invalid multibyte character */ 1299 error(291); 1300 break; 1301 } 1302 if (n == 0) 1303 n = 1; 1304 } 1305 1306 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1307 1308 /* convert from multibyte to wide char */ 1309 (void)mbtowc(NULL, NULL, 0); 1310 for (i = 0, wi = 0; i < len; i += n, wi++) { 1311 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1312 break; 1313 if (n == 0) 1314 n = 1; 1315 } 1316 ws[wi] = 0; 1317 free(s); 1318 1319 strg = xcalloc(1, sizeof (strg_t)); 1320 strg->st_tspec = WCHAR; 1321 strg->st_len = wlen; 1322 strg->st_wcp = ws; 1323 1324 yylval.y_strg = strg; 1325 return (T_STRING); 1326 } 1327 1328 /* 1329 * As noted above the scanner does not create new symbol table entries 1330 * for symbols it cannot find in the symbol table. This is to avoid 1331 * putting undeclared symbols into the symbol table if a syntax error 1332 * occurs. 1333 * 1334 * getsym() is called as soon as it is probably ok to put the symbol to 1335 * the symbol table. This does not mean that it is not possible that 1336 * symbols are put to the symbol table which are than not completely 1337 * declared due to syntax errors. To avoid too many problems in this 1338 * case symbols get type int in getsym(). 1339 * 1340 * XXX calls to getsym() should be delayed until decl1*() is called 1341 */ 1342 sym_t * 1343 getsym(sbuf_t *sb) 1344 { 1345 dinfo_t *di; 1346 char *s; 1347 sym_t *sym; 1348 1349 sym = sb->sb_sym; 1350 1351 /* 1352 * During member declaration it is possible that name() looked 1353 * for symbols of type FVFT, although it should have looked for 1354 * symbols of type FTAG. Same can happen for labels. Both cases 1355 * are compensated here. 1356 */ 1357 if (symtyp == FMOS || symtyp == FLAB) { 1358 if (sym == NULL || sym->s_kind == FVFT) 1359 sym = search(sb); 1360 } 1361 1362 if (sym != NULL) { 1363 if (sym->s_kind != symtyp) 1364 LERROR("storesym()"); 1365 symtyp = FVFT; 1366 freesb(sb); 1367 return (sym); 1368 } 1369 1370 /* create a new symbol table entry */ 1371 1372 /* labels must always be allocated at level 1 (outhermost block) */ 1373 if (symtyp == FLAB) { 1374 sym = getlblk(1, sizeof (sym_t)); 1375 s = getlblk(1, sb->sb_len + 1); 1376 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1377 sym->s_name = s; 1378 sym->s_blklev = 1; 1379 di = dcs; 1380 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1381 di = di->d_nxt; 1382 if (di->d_ctx != AUTO) 1383 LERROR("storesym()"); 1384 } else { 1385 sym = getblk(sizeof (sym_t)); 1386 sym->s_name = sb->sb_name; 1387 sym->s_blklev = blklev; 1388 di = dcs; 1389 } 1390 1391 UNIQUE_CURR_POS(sym->s_dpos); 1392 if ((sym->s_kind = symtyp) != FLAB) 1393 sym->s_type = gettyp(INT); 1394 1395 symtyp = FVFT; 1396 1397 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1398 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1399 sym->s_rlink = &symtab[sb->sb_hash]; 1400 symtab[sb->sb_hash] = sym; 1401 1402 *di->d_ldlsym = sym; 1403 di->d_ldlsym = &sym->s_dlnxt; 1404 1405 freesb(sb); 1406 return (sym); 1407 } 1408 1409 /* 1410 * Construct a temporary symbol. The symbol starts with a digit, so that 1411 * it is illegal. 1412 */ 1413 sym_t * 1414 mktempsym(type_t *t) 1415 { 1416 static int n = 0; 1417 int h; 1418 char *s = getlblk(blklev, 64); 1419 sym_t *sym = getblk(sizeof (sym_t)); 1420 1421 (void)snprintf(s, 64, "%.8d_tmp", n++); 1422 h = hash(s); 1423 1424 sym->s_name = s; 1425 sym->s_type = t; 1426 sym->s_blklev = blklev; 1427 sym->s_scl = AUTO; 1428 sym->s_kind = FVFT; 1429 sym->s_used = 1; 1430 sym->s_set = 1; 1431 1432 if ((sym->s_link = symtab[h]) != NULL) 1433 symtab[h]->s_rlink = &sym->s_link; 1434 sym->s_rlink = &symtab[h]; 1435 symtab[h] = sym; 1436 1437 *dcs->d_ldlsym = sym; 1438 dcs->d_ldlsym = &sym->s_dlnxt; 1439 1440 return sym; 1441 } 1442 1443 /* 1444 * Remove a symbol forever from the symbol table. s_blklev 1445 * is set to -1 to avoid that the symbol will later be put 1446 * back to the symbol table. 1447 */ 1448 void 1449 rmsym(sym_t *sym) 1450 { 1451 1452 if ((*sym->s_rlink = sym->s_link) != NULL) 1453 sym->s_link->s_rlink = sym->s_rlink; 1454 sym->s_blklev = -1; 1455 sym->s_link = NULL; 1456 } 1457 1458 /* 1459 * Remove a list of symbols declared at one level from the symbol 1460 * table. 1461 */ 1462 void 1463 rmsyms(sym_t *syms) 1464 { 1465 sym_t *sym; 1466 1467 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1468 if (sym->s_blklev != -1) { 1469 if ((*sym->s_rlink = sym->s_link) != NULL) 1470 sym->s_link->s_rlink = sym->s_rlink; 1471 sym->s_link = NULL; 1472 sym->s_rlink = NULL; 1473 } 1474 } 1475 } 1476 1477 /* 1478 * Put a symbol into the symbol table 1479 */ 1480 void 1481 inssym(int bl, sym_t *sym) 1482 { 1483 int h; 1484 1485 h = hash(sym->s_name); 1486 if ((sym->s_link = symtab[h]) != NULL) 1487 symtab[h]->s_rlink = &sym->s_link; 1488 sym->s_rlink = &symtab[h]; 1489 symtab[h] = sym; 1490 sym->s_blklev = bl; 1491 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1492 LERROR("inssym()"); 1493 } 1494 1495 /* 1496 * Called at level 0 after syntax errors 1497 * Removes all symbols which are not declared at level 0 from the 1498 * symbol table. Also frees all memory which is not associated with 1499 * level 0. 1500 */ 1501 void 1502 cleanup(void) 1503 { 1504 sym_t *sym, *nsym; 1505 int i; 1506 1507 for (i = 0; i < HSHSIZ1; i++) { 1508 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1509 nsym = sym->s_link; 1510 if (sym->s_blklev >= 1) { 1511 if ((*sym->s_rlink = nsym) != NULL) 1512 nsym->s_rlink = sym->s_rlink; 1513 } 1514 } 1515 } 1516 1517 for (i = mblklev; i > 0; i--) 1518 freelblk(i); 1519 } 1520 1521 /* 1522 * Create a new symbol with the name of an existing symbol. 1523 */ 1524 sym_t * 1525 pushdown(sym_t *sym) 1526 { 1527 int h; 1528 sym_t *nsym; 1529 1530 h = hash(sym->s_name); 1531 nsym = getblk(sizeof (sym_t)); 1532 if (sym->s_blklev > blklev) 1533 LERROR("pushdown()"); 1534 nsym->s_name = sym->s_name; 1535 UNIQUE_CURR_POS(nsym->s_dpos); 1536 nsym->s_kind = sym->s_kind; 1537 nsym->s_blklev = blklev; 1538 1539 if ((nsym->s_link = symtab[h]) != NULL) 1540 symtab[h]->s_rlink = &nsym->s_link; 1541 nsym->s_rlink = &symtab[h]; 1542 symtab[h] = nsym; 1543 1544 *dcs->d_ldlsym = nsym; 1545 dcs->d_ldlsym = &nsym->s_dlnxt; 1546 1547 return (nsym); 1548 } 1549 1550 /* 1551 * Free any dynamically allocated memory referenced by 1552 * the value stack or yylval. 1553 * The type of information in yylval is described by tok. 1554 */ 1555 void 1556 freeyyv(void *sp, int tok) 1557 { 1558 if (tok == T_NAME || tok == T_TYPENAME) { 1559 sbuf_t *sb = *(sbuf_t **)sp; 1560 freesb(sb); 1561 } else if (tok == T_CON) { 1562 val_t *val = *(val_t **)sp; 1563 free(val); 1564 } else if (tok == T_STRING) { 1565 strg_t *strg = *(strg_t **)sp; 1566 if (strg->st_tspec == CHAR) { 1567 free(strg->st_cp); 1568 } else if (strg->st_tspec == WCHAR) { 1569 free(strg->st_wcp); 1570 } else { 1571 LERROR("fryylv()"); 1572 } 1573 free(strg); 1574 } 1575 } 1576
Indexes created Tue Oct 21 16:10:11 GMT 2025