scan.l revision 1.73
1 %{ 2 /* $NetBSD: scan.l,v 1.73 2016/12/29 23:54:29 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 1.73 2016/12/29 23:54:29 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 ((int)(~(~0U << 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 /* Are we parsing a gcc attribute? */ 64 int attron; 65 66 static void incline(void); 67 static void badchar(int); 68 static sbuf_t *allocsb(void); 69 static void freesb(sbuf_t *); 70 static int inpc(void); 71 static int hash(const char *); 72 static sym_t *search(sbuf_t *); 73 static int name(void); 74 static int keyw(sym_t *); 75 static int icon(int); 76 static int fcon(void); 77 static int operator(int, op_t); 78 static int ccon(void); 79 static int wccon(void); 80 static int getescc(int); 81 static void directive(void); 82 static void comment(void); 83 static void slashslashcomment(void); 84 static int string(void); 85 static int wcstrg(void); 86 87 %} 88 89 90 L [_A-Za-z] 91 D [0-9] 92 NZD [1-9] 93 BD [0-1] 94 OD [0-7] 95 HD [0-9A-Fa-f] 96 EX ([eE][+-]?[0-9]+) 97 HX (p[+-]?[0-9A-Fa-f]+) 98 TL ([fFlL]?[i]?) 99 100 %option nounput 101 102 %% 103 104 {L}({L}|{D})* return (name()); 105 0[bB]{BD}+[lLuU]* return (icon(2)); 106 0{OD}*[lLuU]* return (icon(8)); 107 {NZD}{D}*[lLuU]* return (icon(10)); 108 0[xX]{HD}+[lLuU]* return (icon(16)); 109 {D}+\.{D}*{EX}?{TL} | 110 {D}+{EX}{TL} | 111 0[xX]{HD}+\.{HD}*{HX}{TL} | 112 0[xX]{HD}+{HX}{TL} | 113 \.{D}+{EX}?{TL} return (fcon()); 114 "=" return (operator(T_ASSIGN, ASSIGN)); 115 "*=" return (operator(T_OPASS, MULASS)); 116 "/=" return (operator(T_OPASS, DIVASS)); 117 "%=" return (operator(T_OPASS, MODASS)); 118 "+=" return (operator(T_OPASS, ADDASS)); 119 "-=" return (operator(T_OPASS, SUBASS)); 120 "<<=" return (operator(T_OPASS, SHLASS)); 121 ">>=" return (operator(T_OPASS, SHRASS)); 122 "&=" return (operator(T_OPASS, ANDASS)); 123 "^=" return (operator(T_OPASS, XORASS)); 124 "|=" return (operator(T_OPASS, ORASS)); 125 "||" return (operator(T_LOGOR, LOGOR)); 126 "&&" return (operator(T_LOGAND, LOGAND)); 127 "|" return (operator(T_OR, OR)); 128 "&" return (operator(T_AND, AND)); 129 "^" return (operator(T_XOR, XOR)); 130 "==" return (operator(T_EQOP, EQ)); 131 "!=" return (operator(T_EQOP, NE)); 132 "<" return (operator(T_RELOP, LT)); 133 ">" return (operator(T_RELOP, GT)); 134 "<=" return (operator(T_RELOP, LE)); 135 ">=" return (operator(T_RELOP, GE)); 136 "<<" return (operator(T_SHFTOP, SHL)); 137 ">>" return (operator(T_SHFTOP, SHR)); 138 "++" return (operator(T_INCDEC, INC)); 139 "--" return (operator(T_INCDEC, DEC)); 140 "->" return (operator(T_STROP, ARROW)); 141 "." return (operator(T_STROP, POINT)); 142 "+" return (operator(T_ADDOP, PLUS)); 143 "-" return (operator(T_ADDOP, MINUS)); 144 "*" return (operator(T_MULT, MULT)); 145 "/" return (operator(T_DIVOP, DIV)); 146 "%" return (operator(T_DIVOP, MOD)); 147 "!" return (operator(T_UNOP, NOT)); 148 "~" return (operator(T_UNOP, COMPL)); 149 "\"" return (string()); 150 "L\"" return (wcstrg()); 151 ";" return (T_SEMI); 152 "{" return (T_LBRACE); 153 "}" return (T_RBRACE); 154 "," return (T_COMMA); 155 ":" return (T_COLON); 156 "?" return (T_QUEST); 157 "[" return (T_LBRACK); 158 "]" return (T_RBRACK); 159 "(" return (T_LPARN); 160 ")" return (T_RPARN); 161 "..." return (T_ELLIPSE); 162 "'" return (ccon()); 163 "L'" return (wccon()); 164 ^#.*$ directive(); 165 \n incline(); 166 \t|" "|\f|\v ; 167 "/*" comment(); 168 "//" slashslashcomment(); 169 . badchar(yytext[0]); 170 171 %% 172 173 static void 174 incline(void) 175 { 176 curr_pos.p_line++; 177 curr_pos.p_uniq = 0; 178 if (curr_pos.p_file == csrc_pos.p_file) { 179 csrc_pos.p_line++; 180 csrc_pos.p_uniq = 0; 181 } 182 } 183 184 static void 185 badchar(int c) 186 { 187 188 /* unknown character \%o */ 189 error(250, c); 190 } 191 192 /* 193 * Keywords. 194 * During initialisation they are written to the symbol table. 195 */ 196 static struct kwtab { 197 const char *kw_name; /* keyword */ 198 int kw_token; /* token returned by yylex() */ 199 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 200 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 201 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 202 u_int kw_c89 : 1; /* c89 keyword */ 203 u_int kw_c99 : 1; /* c99 keyword */ 204 u_int kw_gcc : 1; /* GCC keyword */ 205 u_int kw_attr : 1; /* GCC attribute, keyword */ 206 u_int kw_deco : 3; /* name[1] __name[2] __name__[4] */ 207 } kwtab[] = { 208 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 }, 209 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 }, 210 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 }, 211 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 212 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 }, 213 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 }, 214 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 }, 215 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 }, 216 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 }, 217 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 }, 218 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 }, 219 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 }, 220 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 }, 221 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 }, 222 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 }, 223 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 }, 224 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 }, 225 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 }, 226 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 }, 227 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 }, 228 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 }, 229 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 }, 230 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 }, 231 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 }, 232 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 }, 233 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 }, 234 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 }, 235 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 }, 236 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 }, 237 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 }, 238 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 }, 239 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 }, 240 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 }, 241 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 }, 242 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 }, 243 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 }, 244 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 245 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 }, 246 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 247 0, 0, 0, 0,0,1,1,5 }, 248 { "non_null", T_AT_NON_NULL, 0, 0, 0, 0,0,1,1,5 }, 249 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 }, 250 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 }, 251 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 }, 252 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 253 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 254 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 255 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 256 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 257 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 258 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 259 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 260 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 261 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 262 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 263 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 264 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 265 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 266 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 267 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 268 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 269 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 270 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 271 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 272 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 273 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,5 }, 274 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 275 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 276 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 277 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 278 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 279 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 280 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 281 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 282 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 283 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 284 }; 285 286 /* Symbol table */ 287 static sym_t *symtab[HSHSIZ1]; 288 289 /* bit i of the entry with index i is set */ 290 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 291 292 /* least significant i bits are set in the entry with index i */ 293 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 294 295 /* least significant i bits are not set in the entry with index i */ 296 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 297 298 /* free list for sbuf structures */ 299 static sbuf_t *sbfrlst; 300 301 /* Typ of next expected symbol */ 302 symt_t symtyp; 303 304 305 static void 306 addkw(struct kwtab *kw, int deco) 307 { 308 sym_t *sym; 309 size_t h; 310 char buf[256]; 311 const char *name; 312 313 if (!(kw->kw_deco & deco)) 314 return; 315 316 switch (deco) { 317 case 1: 318 name = kw->kw_name; 319 break; 320 case 2: 321 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 322 name = strdup(buf); 323 break; 324 case 4: 325 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 326 name = strdup(buf); 327 break; 328 default: 329 abort(); 330 } 331 332 if (name == NULL) 333 err(1, "Can't init symbol table"); 334 335 sym = getblk(sizeof (sym_t)); 336 sym->s_name = name; 337 sym->s_keyw = kw; 338 sym->s_value.v_quad = kw->kw_token; 339 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 340 sym->s_tspec = kw->kw_tspec; 341 } else if (kw->kw_token == T_SCLASS) { 342 sym->s_scl = kw->kw_scl; 343 } else if (kw->kw_token == T_QUAL) { 344 sym->s_tqual = kw->kw_tqual; 345 } 346 h = hash(sym->s_name); 347 if ((sym->s_link = symtab[h]) != NULL) 348 symtab[h]->s_rlink = &sym->s_link; 349 sym->s_rlink = &symtab[h]; 350 symtab[h] = sym; 351 } 352 353 /* 354 * All keywords are written to the symbol table. This saves us looking 355 * in a extra table for each name we found. 356 */ 357 void 358 initscan(void) 359 { 360 struct kwtab *kw; 361 size_t i; 362 uint64_t uq; 363 364 for (kw = kwtab; kw->kw_name != NULL; kw++) { 365 if ((kw->kw_c89 || kw->kw_c99) && tflag) 366 continue; 367 if (kw->kw_c99 && !(Sflag || gflag)) 368 continue; 369 if (kw->kw_gcc && !gflag) 370 continue; 371 addkw(kw, 1); 372 addkw(kw, 2); 373 addkw(kw, 4); 374 } 375 376 /* initialize bit-masks for quads */ 377 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 378 qbmasks[i] = (uint64_t)1 << i; 379 uq = ~(uint64_t)0 << i; 380 qumasks[i] = uq; 381 qlmasks[i] = ~uq; 382 } 383 qumasks[i] = 0; 384 qlmasks[i] = ~(uint64_t)0; 385 } 386 387 /* 388 * Get a free sbuf structure, if possible from the free list 389 */ 390 static sbuf_t * 391 allocsb(void) 392 { 393 sbuf_t *sb; 394 395 if ((sb = sbfrlst) != NULL) { 396 sbfrlst = sb->sb_nxt; 397 #ifdef BLKDEBUG 398 (void)memset(sb, 0, sizeof (*sb)); 399 #else 400 sb->sb_nxt = NULL; 401 #endif 402 } else { 403 sb = xmalloc(sizeof (sbuf_t)); 404 (void)memset(sb, 0, sizeof (*sb)); 405 } 406 return (sb); 407 } 408 409 /* 410 * Put a sbuf structure to the free list 411 */ 412 static void 413 freesb(sbuf_t *sb) 414 { 415 416 (void)memset(sb, ZERO, sizeof (*sb)); 417 sb->sb_nxt = sbfrlst; 418 sbfrlst = sb; 419 } 420 421 /* 422 * Read a character and ensure that it is positive (except EOF). 423 * Increment line count(s) if necessary. 424 */ 425 static int 426 inpc(void) 427 { 428 int c; 429 430 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 431 incline(); 432 return (c); 433 } 434 435 static int 436 hash(const char *s) 437 { 438 u_int v; 439 const u_char *us; 440 441 v = 0; 442 for (us = (const u_char *)s; *us != '\0'; us++) { 443 v = (v << sizeof (v)) + *us; 444 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 445 } 446 return (v % HSHSIZ1); 447 } 448 449 /* 450 * Lex has found a letter followed by zero or more letters or digits. 451 * It looks for a symbol in the symbol table with the same name. This 452 * symbol must either be a keyword or a symbol of the type required by 453 * symtyp (label, member, tag, ...). 454 * 455 * If it is a keyword, the token is returned. In some cases it is described 456 * more deeply by data written to yylval. 457 * 458 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 459 * is stored in yylval. This struct contains the name of the symbol, its 460 * length and hash value. If there is already a symbol of the same name 461 * and type in the symbol table, the sbuf struct also contains a pointer 462 * to the symbol table entry. 463 */ 464 static int 465 name(void) 466 { 467 char *s; 468 sbuf_t *sb; 469 sym_t *sym; 470 int tok; 471 472 sb = allocsb(); 473 sb->sb_name = yytext; 474 sb->sb_len = yyleng; 475 sb->sb_hash = hash(yytext); 476 if ((sym = search(sb)) != NULL && sym->s_keyw) { 477 freesb(sb); 478 return (keyw(sym)); 479 } 480 481 sb->sb_sym = sym; 482 483 if (sym != NULL) { 484 if (blklev < sym->s_blklev) 485 LERROR("name()"); 486 sb->sb_name = sym->s_name; 487 sb->sb_len = strlen(sym->s_name); 488 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 489 } else { 490 s = getblk(yyleng + 1); 491 (void)memcpy(s, yytext, yyleng + 1); 492 sb->sb_name = s; 493 sb->sb_len = yyleng; 494 tok = T_NAME; 495 } 496 497 yylval.y_sb = sb; 498 return (tok); 499 } 500 501 static sym_t * 502 search(sbuf_t *sb) 503 { 504 sym_t *sym; 505 506 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 507 if (strcmp(sym->s_name, sb->sb_name) == 0) { 508 if (sym->s_keyw) { 509 struct kwtab *kw = sym->s_keyw; 510 if (!kw->kw_attr || attron) 511 return (sym); 512 } else if (!attron && sym->s_kind == symtyp) 513 return sym; 514 } 515 } 516 517 return (NULL); 518 } 519 520 static int 521 keyw(sym_t *sym) 522 { 523 int t; 524 525 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 526 yylval.y_scl = sym->s_scl; 527 } else if (t == T_TYPE || t == T_SOU) { 528 yylval.y_tspec = sym->s_tspec; 529 } else if (t == T_QUAL) { 530 yylval.y_tqual = sym->s_tqual; 531 } 532 return (t); 533 } 534 535 /* 536 * Convert a string representing an integer into internal representation. 537 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 538 */ 539 static int 540 icon(int base) 541 { 542 int l_suffix, u_suffix; 543 int len; 544 const char *cp; 545 char c, *eptr; 546 tspec_t typ; 547 uint64_t uq = 0; 548 int ansiu; 549 static tspec_t contypes[2][3] = { 550 { INT, LONG, QUAD }, 551 { UINT, ULONG, UQUAD } 552 }; 553 554 cp = yytext; 555 len = yyleng; 556 557 /* skip 0[xX] or 0[bB] */ 558 if (base == 16 || base == 2) { 559 cp += 2; 560 len -= 2; 561 } 562 563 /* read suffixes */ 564 l_suffix = u_suffix = 0; 565 for ( ; ; ) { 566 if ((c = cp[len - 1]) == 'l' || c == 'L') { 567 l_suffix++; 568 } else if (c == 'u' || c == 'U') { 569 u_suffix++; 570 } else { 571 break; 572 } 573 len--; 574 } 575 if (l_suffix > 2 || u_suffix > 1) { 576 /* malformed integer constant */ 577 warning(251); 578 if (l_suffix > 2) 579 l_suffix = 2; 580 if (u_suffix > 1) 581 u_suffix = 1; 582 } 583 if (tflag && u_suffix != 0) { 584 /* suffix U is illegal in traditional C */ 585 warning(97); 586 } 587 typ = contypes[u_suffix][l_suffix]; 588 589 errno = 0; 590 591 uq = strtouq(cp, &eptr, base); 592 if (eptr != cp + len) 593 LERROR("icon()"); 594 if (errno != 0) 595 /* integer constant out of range */ 596 warning(252); 597 598 /* 599 * If the value is too big for the current type, we must choose 600 * another type. 601 */ 602 ansiu = 0; 603 switch (typ) { 604 case INT: 605 if (uq <= TARG_INT_MAX) { 606 /* ok */ 607 } else if (uq <= TARG_UINT_MAX && base != 10) { 608 typ = UINT; 609 } else if (uq <= TARG_LONG_MAX) { 610 typ = LONG; 611 } else { 612 typ = ULONG; 613 if (uq > TARG_ULONG_MAX) { 614 /* integer constant out of range */ 615 warning(252); 616 } 617 } 618 if (typ == UINT || typ == ULONG) { 619 if (tflag) { 620 typ = LONG; 621 } else if (!sflag) { 622 /* 623 * Remember that the constant is unsigned 624 * only in ANSI C 625 */ 626 ansiu = 1; 627 } 628 } 629 break; 630 case UINT: 631 if (uq > TARG_UINT_MAX) { 632 typ = ULONG; 633 if (uq > TARG_ULONG_MAX) { 634 /* integer constant out of range */ 635 warning(252); 636 } 637 } 638 break; 639 case LONG: 640 if (uq > TARG_LONG_MAX && !tflag) { 641 typ = ULONG; 642 if (!sflag) 643 ansiu = 1; 644 if (uq > TARG_ULONG_MAX) { 645 /* integer constant out of range */ 646 warning(252); 647 } 648 } 649 break; 650 case ULONG: 651 if (uq > TARG_ULONG_MAX) { 652 /* integer constant out of range */ 653 warning(252); 654 } 655 break; 656 case QUAD: 657 if (uq > TARG_QUAD_MAX && !tflag) { 658 typ = UQUAD; 659 if (!sflag) 660 ansiu = 1; 661 } 662 break; 663 case UQUAD: 664 if (uq > TARG_UQUAD_MAX) { 665 /* integer constant out of range */ 666 warning(252); 667 } 668 break; 669 /* LINTED206: (enumeration values not handled in switch) */ 670 case STRUCT: 671 case VOID: 672 case LDOUBLE: 673 case FUNC: 674 case ARRAY: 675 case PTR: 676 case ENUM: 677 case UNION: 678 case SIGNED: 679 case NOTSPEC: 680 case DOUBLE: 681 case FLOAT: 682 case USHORT: 683 case SHORT: 684 case UCHAR: 685 case SCHAR: 686 case CHAR: 687 case BOOL: 688 case UNSIGN: 689 case FCOMPLEX: 690 case DCOMPLEX: 691 case LCOMPLEX: 692 case COMPLEX: 693 break; 694 695 case NTSPEC: /* this value unused */ 696 break; 697 } 698 699 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 700 701 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 702 yylval.y_val->v_ansiu = ansiu; 703 yylval.y_val->v_quad = (int64_t)uq; 704 705 return (T_CON); 706 } 707 708 /* 709 * Returns 1 if t is a signed type and the value is negative. 710 * 711 * len is the number of significant bits. If len is -1, len is set 712 * to the width of type t. 713 */ 714 int 715 sign(int64_t q, tspec_t t, int len) 716 { 717 718 if (t == PTR || isutyp(t)) 719 return (0); 720 return (msb(q, t, len)); 721 } 722 723 int 724 msb(int64_t q, tspec_t t, int len) 725 { 726 727 if (len <= 0) 728 len = size(t); 729 return ((q & qbmasks[len - 1]) != 0); 730 } 731 732 /* 733 * Extends the sign of q. 734 */ 735 int64_t 736 xsign(int64_t q, tspec_t t, int len) 737 { 738 739 if (len <= 0) 740 len = size(t); 741 742 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 743 q &= qlmasks[len]; 744 } else { 745 q |= qumasks[len]; 746 } 747 return (q); 748 } 749 750 /* 751 * Convert a string representing a floating point value into its interal 752 * representation. Type and value are returned in yylval. fcon() 753 * (and yylex()) returns T_CON. 754 * XXX Currently it is not possible to convert constants of type 755 * long double which are greater than DBL_MAX. 756 */ 757 static int 758 fcon(void) 759 { 760 const char *cp; 761 int len; 762 tspec_t typ; 763 char c, *eptr; 764 double d; 765 float f = 0; 766 767 cp = yytext; 768 len = yyleng; 769 770 if (cp[len - 1] == 'i') { 771 /* imaginary, do nothing for now */ 772 len--; 773 } 774 if ((c = cp[len - 1]) == 'f' || c == 'F') { 775 typ = FLOAT; 776 len--; 777 } else if (c == 'l' || c == 'L') { 778 typ = LDOUBLE; 779 len--; 780 } else { 781 if (c == 'd' || c == 'D') 782 len--; 783 typ = DOUBLE; 784 } 785 786 if (tflag && typ != DOUBLE) { 787 /* suffixes F and L are illegal in traditional C */ 788 warning(98); 789 } 790 791 errno = 0; 792 d = strtod(cp, &eptr); 793 if (eptr != cp + len) { 794 switch (*eptr) { 795 /* 796 * XXX: non-native non-current strtod() may not handle hex 797 * floats, ignore the rest if we find traces of hex float 798 * syntax... 799 */ 800 case 'p': 801 case 'P': 802 case 'x': 803 case 'X': 804 d = 0; 805 errno = 0; 806 break; 807 default: 808 LERROR("fcon(%s->%s)", cp, eptr); 809 } 810 } 811 if (errno != 0) 812 /* floating-point constant out of range */ 813 warning(248); 814 815 if (typ == FLOAT) { 816 f = (float)d; 817 if (!finite(f)) { 818 /* floating-point constant out of range */ 819 warning(248); 820 f = f > 0 ? FLT_MAX : -FLT_MAX; 821 } 822 } 823 824 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 825 if (typ == FLOAT) { 826 yylval.y_val->v_ldbl = f; 827 } else { 828 yylval.y_val->v_ldbl = d; 829 } 830 831 return (T_CON); 832 } 833 834 static int 835 operator(int t, op_t o) 836 { 837 838 yylval.y_op = o; 839 return (t); 840 } 841 842 /* 843 * Called if lex found a leading \'. 844 */ 845 static int 846 ccon(void) 847 { 848 size_t n; 849 int val, c; 850 char cv; 851 852 n = 0; 853 val = 0; 854 while ((c = getescc('\'')) >= 0) { 855 val = (val << CHAR_BIT) + c; 856 n++; 857 } 858 if (c == -2) { 859 /* unterminated character constant */ 860 error(253); 861 } else { 862 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 863 /* too many characters in character constant */ 864 error(71); 865 } else if (n > 1) { 866 /* multi-character character constant */ 867 warning(294); 868 } else if (n == 0) { 869 /* empty character constant */ 870 error(73); 871 } 872 } 873 if (n == 1) { 874 cv = (char)val; 875 val = cv; 876 } 877 878 yylval.y_val = xcalloc(1, sizeof (val_t)); 879 yylval.y_val->v_tspec = INT; 880 yylval.y_val->v_quad = val; 881 882 return (T_CON); 883 } 884 885 /* 886 * Called if lex found a leading L\' 887 */ 888 static int 889 wccon(void) 890 { 891 static char buf[MB_LEN_MAX + 1]; 892 size_t i; 893 int c; 894 wchar_t wc; 895 896 i = 0; 897 while ((c = getescc('\'')) >= 0) { 898 if (i < MB_CUR_MAX) 899 buf[i] = (char)c; 900 i++; 901 } 902 903 wc = 0; 904 905 if (c == -2) { 906 /* unterminated character constant */ 907 error(253); 908 } else if (c == 0) { 909 /* empty character constant */ 910 error(73); 911 } else { 912 if (i > MB_CUR_MAX) { 913 i = MB_CUR_MAX; 914 /* too many characters in character constant */ 915 error(71); 916 } else { 917 buf[i] = '\0'; 918 (void)mbtowc(NULL, NULL, 0); 919 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 920 /* invalid multibyte character */ 921 error(291); 922 } 923 } 924 925 yylval.y_val = xcalloc(1, sizeof (val_t)); 926 yylval.y_val->v_tspec = WCHAR; 927 yylval.y_val->v_quad = wc; 928 929 return (T_CON); 930 } 931 932 /* 933 * Read a character which is part of a character constant or of a string 934 * and handle escapes. 935 * 936 * The Argument is the character which delimits the character constant or 937 * string. 938 * 939 * Returns -1 if the end of the character constant or string is reached, 940 * -2 if the EOF is reached, and the character otherwise. 941 */ 942 static int 943 getescc(int d) 944 { 945 static int pbc = -1; 946 int n, c, v; 947 948 if (pbc == -1) { 949 c = inpc(); 950 } else { 951 c = pbc; 952 pbc = -1; 953 } 954 if (c == d) 955 return (-1); 956 switch (c) { 957 case '\n': 958 if (tflag) { 959 /* newline in string or char constant */ 960 error(254); 961 return (-2); 962 } 963 return (c); 964 case EOF: 965 return (-2); 966 case '\\': 967 switch (c = inpc()) { 968 case '"': 969 if (tflag && d == '\'') 970 /* \" inside character constant undef. ... */ 971 warning(262); 972 return ('"'); 973 case '\'': 974 return ('\''); 975 case '?': 976 if (tflag) 977 /* \? undefined in traditional C */ 978 warning(263); 979 return ('?'); 980 case '\\': 981 return ('\\'); 982 case 'a': 983 if (tflag) 984 /* \a undefined in traditional C */ 985 warning(81); 986 return ('\a'); 987 case 'b': 988 return ('\b'); 989 case 'f': 990 return ('\f'); 991 case 'n': 992 return ('\n'); 993 case 'r': 994 return ('\r'); 995 case 't': 996 return ('\t'); 997 case 'v': 998 if (tflag) 999 /* \v undefined in traditional C */ 1000 warning(264); 1001 return ('\v'); 1002 case '8': case '9': 1003 /* bad octal digit %c */ 1004 warning(77, c); 1005 /* FALLTHROUGH */ 1006 case '0': case '1': case '2': case '3': 1007 case '4': case '5': case '6': case '7': 1008 n = 3; 1009 v = 0; 1010 do { 1011 v = (v << 3) + (c - '0'); 1012 c = inpc(); 1013 } while (--n && isdigit(c) && (tflag || c <= '7')); 1014 if (tflag && n > 0 && isdigit(c)) 1015 /* bad octal digit %c */ 1016 warning(77, c); 1017 pbc = c; 1018 if (v > TARG_UCHAR_MAX) { 1019 /* character escape does not fit in char. */ 1020 warning(76); 1021 v &= CHAR_MASK; 1022 } 1023 return (v); 1024 case 'x': 1025 if (tflag) 1026 /* \x undefined in traditional C */ 1027 warning(82); 1028 v = 0; 1029 n = 0; 1030 while ((c = inpc()) >= 0 && isxdigit(c)) { 1031 c = isdigit(c) ? 1032 c - '0' : toupper(c) - 'A' + 10; 1033 v = (v << 4) + c; 1034 if (n >= 0) { 1035 if ((v & ~CHAR_MASK) != 0) { 1036 /* overflow in hex escape */ 1037 warning(75); 1038 n = -1; 1039 } else { 1040 n++; 1041 } 1042 } 1043 } 1044 pbc = c; 1045 if (n == 0) { 1046 /* no hex digits follow \x */ 1047 error(74); 1048 } if (n == -1) { 1049 v &= CHAR_MASK; 1050 } 1051 return (v); 1052 case '\n': 1053 return (getescc(d)); 1054 case EOF: 1055 return (-2); 1056 default: 1057 if (isprint(c)) { 1058 /* dubious escape \%c */ 1059 warning(79, c); 1060 } else { 1061 /* dubious escape \%o */ 1062 warning(80, c); 1063 } 1064 } 1065 } 1066 return (c); 1067 } 1068 1069 /* 1070 * Called for preprocessor directives. Currently implemented are: 1071 * # lineno 1072 * # lineno "filename" 1073 */ 1074 static void 1075 directive(void) 1076 { 1077 const char *cp, *fn; 1078 char c, *eptr; 1079 size_t fnl; 1080 long ln; 1081 static int first = 1; 1082 1083 /* Go to first non-whitespace after # */ 1084 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1085 continue; 1086 1087 if (!isdigit((unsigned char)c)) { 1088 error: 1089 /* undefined or invalid # directive */ 1090 warning(255); 1091 return; 1092 } 1093 ln = strtol(--cp, &eptr, 10); 1094 if (cp == eptr) 1095 goto error; 1096 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1097 goto error; 1098 while ((c = *cp++) == ' ' || c == '\t') 1099 continue; 1100 if (c != '\0') { 1101 if (c != '"') 1102 goto error; 1103 fn = cp; 1104 while ((c = *cp) != '"' && c != '\0') 1105 cp++; 1106 if (c != '"') 1107 goto error; 1108 if ((fnl = cp++ - fn) > PATH_MAX) 1109 goto error; 1110 while ((c = *cp++) == ' ' || c == '\t') 1111 continue; 1112 #if 0 1113 if (c != '\0') 1114 warning("extra character(s) after directive"); 1115 #endif 1116 1117 /* empty string means stdin */ 1118 if (fnl == 0) { 1119 fn = "{standard input}"; 1120 fnl = 16; /* strlen (fn) */ 1121 } 1122 curr_pos.p_file = fnnalloc(fn, fnl); 1123 /* 1124 * If this is the first directive, the name is the name 1125 * of the C source file as specified at the command line. 1126 * It is written to the output file. 1127 */ 1128 if (first) { 1129 csrc_pos.p_file = curr_pos.p_file; 1130 outsrc(fnxform(curr_pos.p_file, 1131 strlen(curr_pos.p_file))); 1132 first = 0; 1133 } 1134 } 1135 curr_pos.p_line = (int)ln - 1; 1136 curr_pos.p_uniq = 0; 1137 if (curr_pos.p_file == csrc_pos.p_file) { 1138 csrc_pos.p_line = (int)ln - 1; 1139 csrc_pos.p_uniq = 0; 1140 } 1141 } 1142 1143 /* 1144 * Handle lint comments. Following comments are currently understood: 1145 * ARGSUSEDn 1146 * BITFIELDTYPE 1147 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1148 * FALLTHRU FALLTHROUGH 1149 * LINTLIBRARY 1150 * LINTEDn NOSTRICTn 1151 * LONGLONG 1152 * NOTREACHED 1153 * PRINTFLIKEn 1154 * PROTOLIB 1155 * SCANFLIKEn 1156 * VARARGSn 1157 * If one of this comments is recognized, the arguments, if any, are 1158 * parsed and a function which handles this comment is called. 1159 */ 1160 static void 1161 comment(void) 1162 { 1163 int c, lc; 1164 static struct { 1165 const char *keywd; 1166 int arg; 1167 void (*func)(int); 1168 } keywtab[] = { 1169 { "ARGSUSED", 1, argsused }, 1170 { "BITFIELDTYPE", 0, bitfieldtype }, 1171 { "CONSTCOND", 0, constcond }, 1172 { "CONSTANTCOND", 0, constcond }, 1173 { "CONSTANTCONDITION", 0, constcond }, 1174 { "FALLTHRU", 0, fallthru }, 1175 { "FALLTHROUGH", 0, fallthru }, 1176 { "LINTLIBRARY", 0, lintlib }, 1177 { "LINTED", 1, linted }, 1178 { "LONGLONG", 0, longlong }, 1179 { "NOSTRICT", 1, linted }, 1180 { "NOTREACHED", 0, notreach }, 1181 { "PRINTFLIKE", 1, printflike }, 1182 { "PROTOLIB", 1, protolib }, 1183 { "SCANFLIKE", 1, scanflike }, 1184 { "VARARGS", 1, varargs }, 1185 }; 1186 char keywd[32]; 1187 char arg[32]; 1188 size_t l, i; 1189 int a; 1190 int eoc; 1191 1192 eoc = 0; 1193 1194 /* Skip white spaces after the start of the comment */ 1195 while ((c = inpc()) != EOF && isspace(c)) 1196 continue; 1197 1198 /* Read the potential keyword to keywd */ 1199 l = 0; 1200 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1201 keywd[l++] = (char)c; 1202 c = inpc(); 1203 } 1204 keywd[l] = '\0'; 1205 1206 /* look for the keyword */ 1207 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1208 if (strcmp(keywtab[i].keywd, keywd) == 0) 1209 break; 1210 } 1211 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1212 goto skip_rest; 1213 1214 /* skip white spaces after the keyword */ 1215 while (c != EOF && isspace(c)) 1216 c = inpc(); 1217 1218 /* read the argument, if the keyword accepts one and there is one */ 1219 l = 0; 1220 if (keywtab[i].arg) { 1221 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1222 arg[l++] = (char)c; 1223 c = inpc(); 1224 } 1225 } 1226 arg[l] = '\0'; 1227 a = l != 0 ? atoi(arg) : -1; 1228 1229 /* skip white spaces after the argument */ 1230 while (c != EOF && isspace(c)) 1231 c = inpc(); 1232 1233 if (c != '*' || (c = inpc()) != '/') { 1234 if (keywtab[i].func != linted) 1235 /* extra characters in lint comment */ 1236 warning(257); 1237 } else { 1238 /* 1239 * remember that we have already found the end of the 1240 * comment 1241 */ 1242 eoc = 1; 1243 } 1244 1245 if (keywtab[i].func != NULL) 1246 (*keywtab[i].func)(a); 1247 1248 skip_rest: 1249 while (!eoc) { 1250 lc = c; 1251 if ((c = inpc()) == EOF) { 1252 /* unterminated comment */ 1253 error(256); 1254 break; 1255 } 1256 if (lc == '*' && c == '/') 1257 eoc = 1; 1258 } 1259 } 1260 1261 /* 1262 * Handle // style comments 1263 */ 1264 static void 1265 slashslashcomment(void) 1266 { 1267 int c; 1268 1269 if (!Sflag && !gflag) 1270 /* // comments only supported in C99 */ 1271 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1272 1273 while ((c = inpc()) != EOF && c != '\n') 1274 continue; 1275 } 1276 1277 /* 1278 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1279 * clrwflgs() is called after function definitions and global and 1280 * local declarations and definitions. It is also called between 1281 * the controlling expression and the body of control statements 1282 * (if, switch, for, while). 1283 */ 1284 void 1285 clrwflgs(void) 1286 { 1287 1288 lwarn = LWARN_ALL; 1289 quadflg = 0; 1290 ccflg = 0; 1291 } 1292 1293 /* 1294 * Strings are stored in a dynamically alloceted buffer and passed 1295 * in yylval.y_xstrg to the parser. The parser or the routines called 1296 * by the parser are responsible for freeing this buffer. 1297 */ 1298 static int 1299 string(void) 1300 { 1301 u_char *s; 1302 int c; 1303 size_t len, max; 1304 strg_t *strg; 1305 1306 s = xmalloc(max = 64); 1307 1308 len = 0; 1309 while ((c = getescc('"')) >= 0) { 1310 /* +1 to reserve space for a trailing NUL character */ 1311 if (len + 1 == max) 1312 s = xrealloc(s, max *= 2); 1313 s[len++] = (char)c; 1314 } 1315 s[len] = '\0'; 1316 if (c == -2) 1317 /* unterminated string constant */ 1318 error(258); 1319 1320 strg = xcalloc(1, sizeof (strg_t)); 1321 strg->st_tspec = CHAR; 1322 strg->st_len = len; 1323 strg->st_cp = s; 1324 1325 yylval.y_strg = strg; 1326 return (T_STRING); 1327 } 1328 1329 static int 1330 wcstrg(void) 1331 { 1332 char *s; 1333 int c, n; 1334 size_t i, wi; 1335 size_t len, max, wlen; 1336 wchar_t *ws; 1337 strg_t *strg; 1338 1339 s = xmalloc(max = 64); 1340 len = 0; 1341 while ((c = getescc('"')) >= 0) { 1342 /* +1 to save space for a trailing NUL character */ 1343 if (len + 1 >= max) 1344 s = xrealloc(s, max *= 2); 1345 s[len++] = (char)c; 1346 } 1347 s[len] = '\0'; 1348 if (c == -2) 1349 /* unterminated string constant */ 1350 error(258); 1351 1352 /* get length of wide-character string */ 1353 (void)mblen(NULL, 0); 1354 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1355 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1356 /* invalid multibyte character */ 1357 error(291); 1358 break; 1359 } 1360 if (n == 0) 1361 n = 1; 1362 } 1363 1364 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1365 1366 /* convert from multibyte to wide char */ 1367 (void)mbtowc(NULL, NULL, 0); 1368 for (i = 0, wi = 0; i < len; i += n, wi++) { 1369 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1370 break; 1371 if (n == 0) 1372 n = 1; 1373 } 1374 ws[wi] = 0; 1375 free(s); 1376 1377 strg = xcalloc(1, sizeof (strg_t)); 1378 strg->st_tspec = WCHAR; 1379 strg->st_len = wlen; 1380 strg->st_wcp = ws; 1381 1382 yylval.y_strg = strg; 1383 return (T_STRING); 1384 } 1385 1386 /* 1387 * As noted above the scanner does not create new symbol table entries 1388 * for symbols it cannot find in the symbol table. This is to avoid 1389 * putting undeclared symbols into the symbol table if a syntax error 1390 * occurs. 1391 * 1392 * getsym() is called as soon as it is probably ok to put the symbol to 1393 * the symbol table. This does not mean that it is not possible that 1394 * symbols are put to the symbol table which are than not completely 1395 * declared due to syntax errors. To avoid too many problems in this 1396 * case symbols get type int in getsym(). 1397 * 1398 * XXX calls to getsym() should be delayed until decl1*() is called 1399 */ 1400 sym_t * 1401 getsym(sbuf_t *sb) 1402 { 1403 dinfo_t *di; 1404 char *s; 1405 sym_t *sym; 1406 1407 sym = sb->sb_sym; 1408 1409 /* 1410 * During member declaration it is possible that name() looked 1411 * for symbols of type FVFT, although it should have looked for 1412 * symbols of type FTAG. Same can happen for labels. Both cases 1413 * are compensated here. 1414 */ 1415 if (symtyp == FMOS || symtyp == FLAB) { 1416 if (sym == NULL || sym->s_kind == FVFT) 1417 sym = search(sb); 1418 } 1419 1420 if (sym != NULL) { 1421 if (sym->s_kind != symtyp) 1422 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1423 symtyp = FVFT; 1424 freesb(sb); 1425 return (sym); 1426 } 1427 1428 /* create a new symbol table entry */ 1429 1430 /* labels must always be allocated at level 1 (outhermost block) */ 1431 if (symtyp == FLAB) { 1432 sym = getlblk(1, sizeof (sym_t)); 1433 s = getlblk(1, sb->sb_len + 1); 1434 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1435 sym->s_name = s; 1436 sym->s_blklev = 1; 1437 di = dcs; 1438 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1439 di = di->d_nxt; 1440 if (di->d_ctx != AUTO) 1441 LERROR("storesym()"); 1442 } else { 1443 sym = getblk(sizeof (sym_t)); 1444 sym->s_name = sb->sb_name; 1445 sym->s_blklev = blklev; 1446 di = dcs; 1447 } 1448 1449 UNIQUE_CURR_POS(sym->s_dpos); 1450 if ((sym->s_kind = symtyp) != FLAB) 1451 sym->s_type = gettyp(INT); 1452 1453 symtyp = FVFT; 1454 1455 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1456 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1457 sym->s_rlink = &symtab[sb->sb_hash]; 1458 symtab[sb->sb_hash] = sym; 1459 1460 *di->d_ldlsym = sym; 1461 di->d_ldlsym = &sym->s_dlnxt; 1462 1463 freesb(sb); 1464 return (sym); 1465 } 1466 1467 /* 1468 * Construct a temporary symbol. The symbol starts with a digit, so that 1469 * it is illegal. 1470 */ 1471 sym_t * 1472 mktempsym(type_t *t) 1473 { 1474 static int n = 0; 1475 int h; 1476 char *s = getlblk(blklev, 64); 1477 sym_t *sym = getblk(sizeof (sym_t)); 1478 1479 (void)snprintf(s, 64, "%.8d_tmp", n++); 1480 h = hash(s); 1481 1482 sym->s_name = s; 1483 sym->s_type = t; 1484 sym->s_blklev = blklev; 1485 sym->s_scl = AUTO; 1486 sym->s_kind = FVFT; 1487 sym->s_used = 1; 1488 sym->s_set = 1; 1489 1490 if ((sym->s_link = symtab[h]) != NULL) 1491 symtab[h]->s_rlink = &sym->s_link; 1492 sym->s_rlink = &symtab[h]; 1493 symtab[h] = sym; 1494 1495 *dcs->d_ldlsym = sym; 1496 dcs->d_ldlsym = &sym->s_dlnxt; 1497 1498 return sym; 1499 } 1500 1501 /* 1502 * Remove a symbol forever from the symbol table. s_blklev 1503 * is set to -1 to avoid that the symbol will later be put 1504 * back to the symbol table. 1505 */ 1506 void 1507 rmsym(sym_t *sym) 1508 { 1509 1510 if ((*sym->s_rlink = sym->s_link) != NULL) 1511 sym->s_link->s_rlink = sym->s_rlink; 1512 sym->s_blklev = -1; 1513 sym->s_link = NULL; 1514 } 1515 1516 /* 1517 * Remove a list of symbols declared at one level from the symbol 1518 * table. 1519 */ 1520 void 1521 rmsyms(sym_t *syms) 1522 { 1523 sym_t *sym; 1524 1525 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1526 if (sym->s_blklev != -1) { 1527 if ((*sym->s_rlink = sym->s_link) != NULL) 1528 sym->s_link->s_rlink = sym->s_rlink; 1529 sym->s_link = NULL; 1530 sym->s_rlink = NULL; 1531 } 1532 } 1533 } 1534 1535 /* 1536 * Put a symbol into the symbol table 1537 */ 1538 void 1539 inssym(int bl, sym_t *sym) 1540 { 1541 int h; 1542 1543 h = hash(sym->s_name); 1544 if ((sym->s_link = symtab[h]) != NULL) 1545 symtab[h]->s_rlink = &sym->s_link; 1546 sym->s_rlink = &symtab[h]; 1547 symtab[h] = sym; 1548 sym->s_blklev = bl; 1549 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1550 LERROR("inssym()"); 1551 } 1552 1553 /* 1554 * Called at level 0 after syntax errors 1555 * Removes all symbols which are not declared at level 0 from the 1556 * symbol table. Also frees all memory which is not associated with 1557 * level 0. 1558 */ 1559 void 1560 cleanup(void) 1561 { 1562 sym_t *sym, *nsym; 1563 int i; 1564 1565 for (i = 0; i < HSHSIZ1; i++) { 1566 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1567 nsym = sym->s_link; 1568 if (sym->s_blklev >= 1) { 1569 if ((*sym->s_rlink = nsym) != NULL) 1570 nsym->s_rlink = sym->s_rlink; 1571 } 1572 } 1573 } 1574 1575 for (i = mblklev; i > 0; i--) 1576 freelblk(i); 1577 } 1578 1579 /* 1580 * Create a new symbol with the name of an existing symbol. 1581 */ 1582 sym_t * 1583 pushdown(sym_t *sym) 1584 { 1585 int h; 1586 sym_t *nsym; 1587 1588 h = hash(sym->s_name); 1589 nsym = getblk(sizeof (sym_t)); 1590 if (sym->s_blklev > blklev) 1591 LERROR("pushdown()"); 1592 nsym->s_name = sym->s_name; 1593 UNIQUE_CURR_POS(nsym->s_dpos); 1594 nsym->s_kind = sym->s_kind; 1595 nsym->s_blklev = blklev; 1596 1597 if ((nsym->s_link = symtab[h]) != NULL) 1598 symtab[h]->s_rlink = &nsym->s_link; 1599 nsym->s_rlink = &symtab[h]; 1600 symtab[h] = nsym; 1601 1602 *dcs->d_ldlsym = nsym; 1603 dcs->d_ldlsym = &nsym->s_dlnxt; 1604 1605 return (nsym); 1606 } 1607 1608 /* 1609 * Free any dynamically allocated memory referenced by 1610 * the value stack or yylval. 1611 * The type of information in yylval is described by tok. 1612 */ 1613 void 1614 freeyyv(void *sp, int tok) 1615 { 1616 if (tok == T_NAME || tok == T_TYPENAME) { 1617 sbuf_t *sb = *(sbuf_t **)sp; 1618 freesb(sb); 1619 } else if (tok == T_CON) { 1620 val_t *val = *(val_t **)sp; 1621 free(val); 1622 } else if (tok == T_STRING) { 1623 strg_t *strg = *(strg_t **)sp; 1624 if (strg->st_tspec == CHAR) { 1625 free(strg->st_cp); 1626 } else if (strg->st_tspec == WCHAR) { 1627 free(strg->st_wcp); 1628 } else { 1629 LERROR("fryylv()"); 1630 } 1631 free(strg); 1632 } 1633 } 1634
Indexes created Wed Oct 01 07:09:59 GMT 2025