scan.l revision 1.77.2.1
1 %{ 2 /* $NetBSD: scan.l,v 1.77.2.1 2017/04/21 16:54:16 bouyer 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 1.77.2.1 2017/04/21 16:54:16 bouyer 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 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 }, 219 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 }, 220 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 }, 221 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 }, 222 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 }, 223 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 }, 224 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 }, 225 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 }, 226 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 }, 227 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 }, 228 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 }, 229 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 }, 230 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 }, 231 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 }, 232 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 }, 233 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 }, 234 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 }, 235 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 }, 236 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 }, 237 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 }, 238 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 }, 239 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 }, 240 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 }, 241 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 }, 242 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 }, 243 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 }, 244 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 }, 245 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 }, 246 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 }, 247 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 248 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 }, 249 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 }, 250 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 251 0, 0, 0, 0,0,1,1,5 }, 252 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 }, 253 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 }, 254 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 }, 255 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 }, 256 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 }, 257 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 258 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 259 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 260 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 261 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 262 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 263 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 264 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 265 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 266 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 267 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 268 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 269 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 270 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 271 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 272 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 273 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 }, 274 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 275 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 276 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 277 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 278 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 279 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 }, 280 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 281 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 282 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 283 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 284 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 285 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 286 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 287 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 288 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 289 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 290 }; 291 292 /* Symbol table */ 293 static sym_t *symtab[HSHSIZ1]; 294 295 /* bit i of the entry with index i is set */ 296 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 297 298 /* least significant i bits are set in the entry with index i */ 299 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 300 301 /* least significant i bits are not set in the entry with index i */ 302 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 303 304 /* free list for sbuf structures */ 305 static sbuf_t *sbfrlst; 306 307 /* Typ of next expected symbol */ 308 symt_t symtyp; 309 310 311 static void 312 addkw(struct kwtab *kw, int deco) 313 { 314 sym_t *sym; 315 size_t h; 316 char buf[256]; 317 const char *name; 318 319 if (!(kw->kw_deco & deco)) 320 return; 321 322 switch (deco) { 323 case 1: 324 name = kw->kw_name; 325 break; 326 case 2: 327 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 328 name = strdup(buf); 329 break; 330 case 4: 331 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 332 name = strdup(buf); 333 break; 334 default: 335 abort(); 336 } 337 338 if (name == NULL) 339 err(1, "Can't init symbol table"); 340 341 sym = getblk(sizeof (sym_t)); 342 sym->s_name = name; 343 sym->s_keyw = kw; 344 sym->s_value.v_quad = kw->kw_token; 345 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 346 sym->s_tspec = kw->kw_tspec; 347 } else if (kw->kw_token == T_SCLASS) { 348 sym->s_scl = kw->kw_scl; 349 } else if (kw->kw_token == T_QUAL) { 350 sym->s_tqual = kw->kw_tqual; 351 } 352 h = hash(sym->s_name); 353 if ((sym->s_link = symtab[h]) != NULL) 354 symtab[h]->s_rlink = &sym->s_link; 355 sym->s_rlink = &symtab[h]; 356 symtab[h] = sym; 357 } 358 359 /* 360 * All keywords are written to the symbol table. This saves us looking 361 * in a extra table for each name we found. 362 */ 363 void 364 initscan(void) 365 { 366 struct kwtab *kw; 367 size_t i; 368 uint64_t uq; 369 370 for (kw = kwtab; kw->kw_name != NULL; kw++) { 371 if ((kw->kw_c89 || kw->kw_c99) && tflag) 372 continue; 373 if (kw->kw_c99 && !(Sflag || gflag)) 374 continue; 375 if (kw->kw_gcc && !gflag) 376 continue; 377 addkw(kw, 1); 378 addkw(kw, 2); 379 addkw(kw, 4); 380 } 381 382 /* initialize bit-masks for quads */ 383 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 384 qbmasks[i] = (uint64_t)1 << i; 385 uq = ~(uint64_t)0 << i; 386 qumasks[i] = uq; 387 qlmasks[i] = ~uq; 388 } 389 qumasks[i] = 0; 390 qlmasks[i] = ~(uint64_t)0; 391 } 392 393 /* 394 * Get a free sbuf structure, if possible from the free list 395 */ 396 static sbuf_t * 397 allocsb(void) 398 { 399 sbuf_t *sb; 400 401 if ((sb = sbfrlst) != NULL) { 402 sbfrlst = sb->sb_nxt; 403 #ifdef BLKDEBUG 404 (void)memset(sb, 0, sizeof (*sb)); 405 #else 406 sb->sb_nxt = NULL; 407 #endif 408 } else { 409 sb = xmalloc(sizeof (sbuf_t)); 410 (void)memset(sb, 0, sizeof (*sb)); 411 } 412 return (sb); 413 } 414 415 /* 416 * Put a sbuf structure to the free list 417 */ 418 static void 419 freesb(sbuf_t *sb) 420 { 421 422 (void)memset(sb, ZERO, sizeof (*sb)); 423 sb->sb_nxt = sbfrlst; 424 sbfrlst = sb; 425 } 426 427 /* 428 * Read a character and ensure that it is positive (except EOF). 429 * Increment line count(s) if necessary. 430 */ 431 static int 432 inpc(void) 433 { 434 int c; 435 436 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 437 incline(); 438 return (c); 439 } 440 441 static int 442 hash(const char *s) 443 { 444 u_int v; 445 const u_char *us; 446 447 v = 0; 448 for (us = (const u_char *)s; *us != '\0'; us++) { 449 v = (v << sizeof (v)) + *us; 450 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 451 } 452 return (v % HSHSIZ1); 453 } 454 455 /* 456 * Lex has found a letter followed by zero or more letters or digits. 457 * It looks for a symbol in the symbol table with the same name. This 458 * symbol must either be a keyword or a symbol of the type required by 459 * symtyp (label, member, tag, ...). 460 * 461 * If it is a keyword, the token is returned. In some cases it is described 462 * more deeply by data written to yylval. 463 * 464 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 465 * is stored in yylval. This struct contains the name of the symbol, its 466 * length and hash value. If there is already a symbol of the same name 467 * and type in the symbol table, the sbuf struct also contains a pointer 468 * to the symbol table entry. 469 */ 470 static int 471 name(void) 472 { 473 char *s; 474 sbuf_t *sb; 475 sym_t *sym; 476 int tok; 477 478 sb = allocsb(); 479 sb->sb_name = yytext; 480 sb->sb_len = yyleng; 481 sb->sb_hash = hash(yytext); 482 if ((sym = search(sb)) != NULL && sym->s_keyw) { 483 freesb(sb); 484 return (keyw(sym)); 485 } 486 487 sb->sb_sym = sym; 488 489 if (sym != NULL) { 490 if (blklev < sym->s_blklev) 491 LERROR("name()"); 492 sb->sb_name = sym->s_name; 493 sb->sb_len = strlen(sym->s_name); 494 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 495 } else { 496 s = getblk(yyleng + 1); 497 (void)memcpy(s, yytext, yyleng + 1); 498 sb->sb_name = s; 499 sb->sb_len = yyleng; 500 tok = T_NAME; 501 } 502 503 yylval.y_sb = sb; 504 return (tok); 505 } 506 507 static sym_t * 508 search(sbuf_t *sb) 509 { 510 sym_t *sym; 511 512 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 513 if (strcmp(sym->s_name, sb->sb_name) == 0) { 514 if (sym->s_keyw) { 515 struct kwtab *kw = sym->s_keyw; 516 if (!kw->kw_attr || attron) 517 return (sym); 518 } else if (!attron && sym->s_kind == symtyp) 519 return sym; 520 } 521 } 522 523 return (NULL); 524 } 525 526 static int 527 keyw(sym_t *sym) 528 { 529 int t; 530 531 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 532 yylval.y_scl = sym->s_scl; 533 } else if (t == T_TYPE || t == T_SOU) { 534 yylval.y_tspec = sym->s_tspec; 535 } else if (t == T_QUAL) { 536 yylval.y_tqual = sym->s_tqual; 537 } 538 return (t); 539 } 540 541 /* 542 * Convert a string representing an integer into internal representation. 543 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 544 */ 545 static int 546 icon(int base) 547 { 548 int l_suffix, u_suffix; 549 int len; 550 const char *cp; 551 char c, *eptr; 552 tspec_t typ; 553 uint64_t uq = 0; 554 int ansiu; 555 static tspec_t contypes[2][3] = { 556 { INT, LONG, QUAD }, 557 { UINT, ULONG, UQUAD } 558 }; 559 560 cp = yytext; 561 len = yyleng; 562 563 /* skip 0[xX] or 0[bB] */ 564 if (base == 16 || base == 2) { 565 cp += 2; 566 len -= 2; 567 } 568 569 /* read suffixes */ 570 l_suffix = u_suffix = 0; 571 for ( ; ; ) { 572 if ((c = cp[len - 1]) == 'l' || c == 'L') { 573 l_suffix++; 574 } else if (c == 'u' || c == 'U') { 575 u_suffix++; 576 } else { 577 break; 578 } 579 len--; 580 } 581 if (l_suffix > 2 || u_suffix > 1) { 582 /* malformed integer constant */ 583 warning(251); 584 if (l_suffix > 2) 585 l_suffix = 2; 586 if (u_suffix > 1) 587 u_suffix = 1; 588 } 589 if (tflag && u_suffix != 0) { 590 /* suffix U is illegal in traditional C */ 591 warning(97); 592 } 593 typ = contypes[u_suffix][l_suffix]; 594 595 errno = 0; 596 597 uq = strtouq(cp, &eptr, base); 598 if (eptr != cp + len) 599 LERROR("icon()"); 600 if (errno != 0) 601 /* integer constant out of range */ 602 warning(252); 603 604 /* 605 * If the value is too big for the current type, we must choose 606 * another type. 607 */ 608 ansiu = 0; 609 switch (typ) { 610 case INT: 611 if (uq <= TARG_INT_MAX) { 612 /* ok */ 613 } else if (uq <= TARG_UINT_MAX && base != 10) { 614 typ = UINT; 615 } else if (uq <= TARG_LONG_MAX) { 616 typ = LONG; 617 } else { 618 typ = ULONG; 619 if (uq > TARG_ULONG_MAX) { 620 /* integer constant out of range */ 621 warning(252); 622 } 623 } 624 if (typ == UINT || typ == ULONG) { 625 if (tflag) { 626 typ = LONG; 627 } else if (!sflag) { 628 /* 629 * Remember that the constant is unsigned 630 * only in ANSI C 631 */ 632 ansiu = 1; 633 } 634 } 635 break; 636 case UINT: 637 if (uq > TARG_UINT_MAX) { 638 typ = ULONG; 639 if (uq > TARG_ULONG_MAX) { 640 /* integer constant out of range */ 641 warning(252); 642 } 643 } 644 break; 645 case LONG: 646 if (uq > TARG_LONG_MAX && !tflag) { 647 typ = ULONG; 648 if (!sflag) 649 ansiu = 1; 650 if (uq > TARG_ULONG_MAX) { 651 /* integer constant out of range */ 652 warning(252); 653 } 654 } 655 break; 656 case ULONG: 657 if (uq > TARG_ULONG_MAX) { 658 /* integer constant out of range */ 659 warning(252); 660 } 661 break; 662 case QUAD: 663 if (uq > TARG_QUAD_MAX && !tflag) { 664 typ = UQUAD; 665 if (!sflag) 666 ansiu = 1; 667 } 668 break; 669 case UQUAD: 670 if (uq > TARG_UQUAD_MAX) { 671 /* integer constant out of range */ 672 warning(252); 673 } 674 break; 675 /* LINTED206: (enumeration values not handled in switch) */ 676 case STRUCT: 677 case VOID: 678 case LDOUBLE: 679 case FUNC: 680 case ARRAY: 681 case PTR: 682 case ENUM: 683 case UNION: 684 case SIGNED: 685 case NOTSPEC: 686 case DOUBLE: 687 case FLOAT: 688 case USHORT: 689 case SHORT: 690 case UCHAR: 691 case SCHAR: 692 case CHAR: 693 case BOOL: 694 case UNSIGN: 695 case FCOMPLEX: 696 case DCOMPLEX: 697 case LCOMPLEX: 698 case COMPLEX: 699 break; 700 701 case NTSPEC: /* this value unused */ 702 break; 703 } 704 705 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 706 707 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 708 yylval.y_val->v_ansiu = ansiu; 709 yylval.y_val->v_quad = (int64_t)uq; 710 711 return (T_CON); 712 } 713 714 /* 715 * Returns 1 if t is a signed type and the value is negative. 716 * 717 * len is the number of significant bits. If len is -1, len is set 718 * to the width of type t. 719 */ 720 int 721 sign(int64_t q, tspec_t t, int len) 722 { 723 724 if (t == PTR || isutyp(t)) 725 return (0); 726 return (msb(q, t, len)); 727 } 728 729 int 730 msb(int64_t q, tspec_t t, int len) 731 { 732 733 if (len <= 0) 734 len = size(t); 735 return ((q & qbmasks[len - 1]) != 0); 736 } 737 738 /* 739 * Extends the sign of q. 740 */ 741 int64_t 742 xsign(int64_t q, tspec_t t, int len) 743 { 744 745 if (len <= 0) 746 len = size(t); 747 748 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 749 q &= qlmasks[len]; 750 } else { 751 q |= qumasks[len]; 752 } 753 return (q); 754 } 755 756 /* 757 * Convert a string representing a floating point value into its interal 758 * representation. Type and value are returned in yylval. fcon() 759 * (and yylex()) returns T_CON. 760 * XXX Currently it is not possible to convert constants of type 761 * long double which are greater than DBL_MAX. 762 */ 763 static int 764 fcon(void) 765 { 766 const char *cp; 767 int len; 768 tspec_t typ; 769 char c, *eptr; 770 double d; 771 float f = 0; 772 773 cp = yytext; 774 len = yyleng; 775 776 if (cp[len - 1] == 'i') { 777 /* imaginary, do nothing for now */ 778 len--; 779 } 780 if ((c = cp[len - 1]) == 'f' || c == 'F') { 781 typ = FLOAT; 782 len--; 783 } else if (c == 'l' || c == 'L') { 784 typ = LDOUBLE; 785 len--; 786 } else { 787 if (c == 'd' || c == 'D') 788 len--; 789 typ = DOUBLE; 790 } 791 792 if (tflag && typ != DOUBLE) { 793 /* suffixes F and L are illegal in traditional C */ 794 warning(98); 795 } 796 797 errno = 0; 798 d = strtod(cp, &eptr); 799 if (eptr != cp + len) { 800 switch (*eptr) { 801 /* 802 * XXX: non-native non-current strtod() may not handle hex 803 * floats, ignore the rest if we find traces of hex float 804 * syntax... 805 */ 806 case 'p': 807 case 'P': 808 case 'x': 809 case 'X': 810 d = 0; 811 errno = 0; 812 break; 813 default: 814 LERROR("fcon(%s->%s)", cp, eptr); 815 } 816 } 817 if (errno != 0) 818 /* floating-point constant out of range */ 819 warning(248); 820 821 if (typ == FLOAT) { 822 f = (float)d; 823 if (!finite(f)) { 824 /* floating-point constant out of range */ 825 warning(248); 826 f = f > 0 ? FLT_MAX : -FLT_MAX; 827 } 828 } 829 830 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 831 if (typ == FLOAT) { 832 yylval.y_val->v_ldbl = f; 833 } else { 834 yylval.y_val->v_ldbl = d; 835 } 836 837 return (T_CON); 838 } 839 840 static int 841 operator(int t, op_t o) 842 { 843 844 yylval.y_op = o; 845 return (t); 846 } 847 848 /* 849 * Called if lex found a leading \'. 850 */ 851 static int 852 ccon(void) 853 { 854 size_t n; 855 int val, c; 856 char cv; 857 858 n = 0; 859 val = 0; 860 while ((c = getescc('\'')) >= 0) { 861 val = (val << CHAR_BIT) + c; 862 n++; 863 } 864 if (c == -2) { 865 /* unterminated character constant */ 866 error(253); 867 } else { 868 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 869 /* too many characters in character constant */ 870 error(71); 871 } else if (n > 1) { 872 /* multi-character character constant */ 873 warning(294); 874 } else if (n == 0) { 875 /* empty character constant */ 876 error(73); 877 } 878 } 879 if (n == 1) { 880 cv = (char)val; 881 val = cv; 882 } 883 884 yylval.y_val = xcalloc(1, sizeof (val_t)); 885 yylval.y_val->v_tspec = INT; 886 yylval.y_val->v_quad = val; 887 888 return (T_CON); 889 } 890 891 /* 892 * Called if lex found a leading L\' 893 */ 894 static int 895 wccon(void) 896 { 897 static char buf[MB_LEN_MAX + 1]; 898 size_t i; 899 int c; 900 wchar_t wc; 901 902 i = 0; 903 while ((c = getescc('\'')) >= 0) { 904 if (i < MB_CUR_MAX) 905 buf[i] = (char)c; 906 i++; 907 } 908 909 wc = 0; 910 911 if (c == -2) { 912 /* unterminated character constant */ 913 error(253); 914 } else if (c == 0) { 915 /* empty character constant */ 916 error(73); 917 } else { 918 if (i > MB_CUR_MAX) { 919 i = MB_CUR_MAX; 920 /* too many characters in character constant */ 921 error(71); 922 } else { 923 buf[i] = '\0'; 924 (void)mbtowc(NULL, NULL, 0); 925 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 926 /* invalid multibyte character */ 927 error(291); 928 } 929 } 930 931 yylval.y_val = xcalloc(1, sizeof (val_t)); 932 yylval.y_val->v_tspec = WCHAR; 933 yylval.y_val->v_quad = wc; 934 935 return (T_CON); 936 } 937 938 /* 939 * Read a character which is part of a character constant or of a string 940 * and handle escapes. 941 * 942 * The Argument is the character which delimits the character constant or 943 * string. 944 * 945 * Returns -1 if the end of the character constant or string is reached, 946 * -2 if the EOF is reached, and the character otherwise. 947 */ 948 static int 949 getescc(int d) 950 { 951 static int pbc = -1; 952 int n, c, v; 953 954 if (pbc == -1) { 955 c = inpc(); 956 } else { 957 c = pbc; 958 pbc = -1; 959 } 960 if (c == d) 961 return (-1); 962 switch (c) { 963 case '\n': 964 if (tflag) { 965 /* newline in string or char constant */ 966 error(254); 967 return (-2); 968 } 969 return (c); 970 case EOF: 971 return (-2); 972 case '\\': 973 switch (c = inpc()) { 974 case '"': 975 if (tflag && d == '\'') 976 /* \" inside character constant undef. ... */ 977 warning(262); 978 return ('"'); 979 case '\'': 980 return ('\''); 981 case '?': 982 if (tflag) 983 /* \? undefined in traditional C */ 984 warning(263); 985 return ('?'); 986 case '\\': 987 return ('\\'); 988 case 'a': 989 if (tflag) 990 /* \a undefined in traditional C */ 991 warning(81); 992 return ('\a'); 993 case 'b': 994 return ('\b'); 995 case 'f': 996 return ('\f'); 997 case 'n': 998 return ('\n'); 999 case 'r': 1000 return ('\r'); 1001 case 't': 1002 return ('\t'); 1003 case 'v': 1004 if (tflag) 1005 /* \v undefined in traditional C */ 1006 warning(264); 1007 return ('\v'); 1008 case '8': case '9': 1009 /* bad octal digit %c */ 1010 warning(77, c); 1011 /* FALLTHROUGH */ 1012 case '0': case '1': case '2': case '3': 1013 case '4': case '5': case '6': case '7': 1014 n = 3; 1015 v = 0; 1016 do { 1017 v = (v << 3) + (c - '0'); 1018 c = inpc(); 1019 } while (--n && isdigit(c) && (tflag || c <= '7')); 1020 if (tflag && n > 0 && isdigit(c)) 1021 /* bad octal digit %c */ 1022 warning(77, c); 1023 pbc = c; 1024 if (v > TARG_UCHAR_MAX) { 1025 /* character escape does not fit in char. */ 1026 warning(76); 1027 v &= CHAR_MASK; 1028 } 1029 return (v); 1030 case 'x': 1031 if (tflag) 1032 /* \x undefined in traditional C */ 1033 warning(82); 1034 v = 0; 1035 n = 0; 1036 while ((c = inpc()) >= 0 && isxdigit(c)) { 1037 c = isdigit(c) ? 1038 c - '0' : toupper(c) - 'A' + 10; 1039 v = (v << 4) + c; 1040 if (n >= 0) { 1041 if ((v & ~CHAR_MASK) != 0) { 1042 /* overflow in hex escape */ 1043 warning(75); 1044 n = -1; 1045 } else { 1046 n++; 1047 } 1048 } 1049 } 1050 pbc = c; 1051 if (n == 0) { 1052 /* no hex digits follow \x */ 1053 error(74); 1054 } if (n == -1) { 1055 v &= CHAR_MASK; 1056 } 1057 return (v); 1058 case '\n': 1059 return (getescc(d)); 1060 case EOF: 1061 return (-2); 1062 default: 1063 if (isprint(c)) { 1064 /* dubious escape \%c */ 1065 warning(79, c); 1066 } else { 1067 /* dubious escape \%o */ 1068 warning(80, c); 1069 } 1070 } 1071 } 1072 return (c); 1073 } 1074 1075 /* 1076 * Called for preprocessor directives. Currently implemented are: 1077 * # lineno 1078 * # lineno "filename" 1079 */ 1080 static void 1081 directive(void) 1082 { 1083 const char *cp, *fn; 1084 char c, *eptr; 1085 size_t fnl; 1086 long ln; 1087 static int first = 1; 1088 1089 /* Go to first non-whitespace after # */ 1090 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1091 continue; 1092 1093 if (!isdigit((unsigned char)c)) { 1094 if (strncmp(cp, "pragma", 6) == 0 1095 && isspace((unsigned char)cp[6])) 1096 return; 1097 error: 1098 /* undefined or invalid # directive */ 1099 warning(255); 1100 return; 1101 } 1102 ln = strtol(--cp, &eptr, 10); 1103 if (cp == eptr) 1104 goto error; 1105 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1106 goto error; 1107 while ((c = *cp++) == ' ' || c == '\t') 1108 continue; 1109 if (c != '\0') { 1110 if (c != '"') 1111 goto error; 1112 fn = cp; 1113 while ((c = *cp) != '"' && c != '\0') 1114 cp++; 1115 if (c != '"') 1116 goto error; 1117 if ((fnl = cp++ - fn) > PATH_MAX) 1118 goto error; 1119 while ((c = *cp++) == ' ' || c == '\t') 1120 continue; 1121 #if 0 1122 if (c != '\0') 1123 warning("extra character(s) after directive"); 1124 #endif 1125 1126 /* empty string means stdin */ 1127 if (fnl == 0) { 1128 fn = "{standard input}"; 1129 fnl = 16; /* strlen (fn) */ 1130 } 1131 curr_pos.p_file = fnnalloc(fn, fnl); 1132 /* 1133 * If this is the first directive, the name is the name 1134 * of the C source file as specified at the command line. 1135 * It is written to the output file. 1136 */ 1137 if (first) { 1138 csrc_pos.p_file = curr_pos.p_file; 1139 outsrc(fnxform(curr_pos.p_file, 1140 strlen(curr_pos.p_file))); 1141 first = 0; 1142 } 1143 } 1144 curr_pos.p_line = (int)ln - 1; 1145 curr_pos.p_uniq = 0; 1146 if (curr_pos.p_file == csrc_pos.p_file) { 1147 csrc_pos.p_line = (int)ln - 1; 1148 csrc_pos.p_uniq = 0; 1149 } 1150 } 1151 1152 /* 1153 * Handle lint comments. Following comments are currently understood: 1154 * ARGSUSEDn 1155 * BITFIELDTYPE 1156 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1157 * FALLTHRU FALLTHROUGH 1158 * LINTLIBRARY 1159 * LINTEDn NOSTRICTn 1160 * LONGLONG 1161 * NOTREACHED 1162 * PRINTFLIKEn 1163 * PROTOLIB 1164 * SCANFLIKEn 1165 * VARARGSn 1166 * If one of this comments is recognized, the arguments, if any, are 1167 * parsed and a function which handles this comment is called. 1168 */ 1169 static void 1170 comment(void) 1171 { 1172 int c, lc; 1173 static struct { 1174 const char *keywd; 1175 int arg; 1176 void (*func)(int); 1177 } keywtab[] = { 1178 { "ARGSUSED", 1, argsused }, 1179 { "BITFIELDTYPE", 0, bitfieldtype }, 1180 { "CONSTCOND", 0, constcond }, 1181 { "CONSTANTCOND", 0, constcond }, 1182 { "CONSTANTCONDITION", 0, constcond }, 1183 { "FALLTHRU", 0, fallthru }, 1184 { "FALLTHROUGH", 0, fallthru }, 1185 { "LINTLIBRARY", 0, lintlib }, 1186 { "LINTED", 1, linted }, 1187 { "LONGLONG", 0, longlong }, 1188 { "NOSTRICT", 1, linted }, 1189 { "NOTREACHED", 0, notreach }, 1190 { "PRINTFLIKE", 1, printflike }, 1191 { "PROTOLIB", 1, protolib }, 1192 { "SCANFLIKE", 1, scanflike }, 1193 { "VARARGS", 1, varargs }, 1194 }; 1195 char keywd[32]; 1196 char arg[32]; 1197 size_t l, i; 1198 int a; 1199 int eoc; 1200 1201 eoc = 0; 1202 1203 /* Skip white spaces after the start of the comment */ 1204 while ((c = inpc()) != EOF && isspace(c)) 1205 continue; 1206 1207 /* Read the potential keyword to keywd */ 1208 l = 0; 1209 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1210 keywd[l++] = (char)c; 1211 c = inpc(); 1212 } 1213 keywd[l] = '\0'; 1214 1215 /* look for the keyword */ 1216 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1217 if (strcmp(keywtab[i].keywd, keywd) == 0) 1218 break; 1219 } 1220 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1221 goto skip_rest; 1222 1223 /* skip white spaces after the keyword */ 1224 while (c != EOF && isspace(c)) 1225 c = inpc(); 1226 1227 /* read the argument, if the keyword accepts one and there is one */ 1228 l = 0; 1229 if (keywtab[i].arg) { 1230 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1231 arg[l++] = (char)c; 1232 c = inpc(); 1233 } 1234 } 1235 arg[l] = '\0'; 1236 a = l != 0 ? atoi(arg) : -1; 1237 1238 /* skip white spaces after the argument */ 1239 while (c != EOF && isspace(c)) 1240 c = inpc(); 1241 1242 if (c != '*' || (c = inpc()) != '/') { 1243 if (keywtab[i].func != linted) 1244 /* extra characters in lint comment */ 1245 warning(257); 1246 } else { 1247 /* 1248 * remember that we have already found the end of the 1249 * comment 1250 */ 1251 eoc = 1; 1252 } 1253 1254 if (keywtab[i].func != NULL) 1255 (*keywtab[i].func)(a); 1256 1257 skip_rest: 1258 while (!eoc) { 1259 lc = c; 1260 if ((c = inpc()) == EOF) { 1261 /* unterminated comment */ 1262 error(256); 1263 break; 1264 } 1265 if (lc == '*' && c == '/') 1266 eoc = 1; 1267 } 1268 } 1269 1270 /* 1271 * Handle // style comments 1272 */ 1273 static void 1274 slashslashcomment(void) 1275 { 1276 int c; 1277 1278 if (!Sflag && !gflag) 1279 /* // comments only supported in C99 */ 1280 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1281 1282 while ((c = inpc()) != EOF && c != '\n') 1283 continue; 1284 } 1285 1286 /* 1287 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1288 * clrwflgs() is called after function definitions and global and 1289 * local declarations and definitions. It is also called between 1290 * the controlling expression and the body of control statements 1291 * (if, switch, for, while). 1292 */ 1293 void 1294 clrwflgs(void) 1295 { 1296 1297 lwarn = LWARN_ALL; 1298 quadflg = 0; 1299 ccflg = 0; 1300 } 1301 1302 /* 1303 * Strings are stored in a dynamically alloceted buffer and passed 1304 * in yylval.y_xstrg to the parser. The parser or the routines called 1305 * by the parser are responsible for freeing this buffer. 1306 */ 1307 static int 1308 string(void) 1309 { 1310 u_char *s; 1311 int c; 1312 size_t len, max; 1313 strg_t *strg; 1314 1315 s = xmalloc(max = 64); 1316 1317 len = 0; 1318 while ((c = getescc('"')) >= 0) { 1319 /* +1 to reserve space for a trailing NUL character */ 1320 if (len + 1 == max) 1321 s = xrealloc(s, max *= 2); 1322 s[len++] = (char)c; 1323 } 1324 s[len] = '\0'; 1325 if (c == -2) 1326 /* unterminated string constant */ 1327 error(258); 1328 1329 strg = xcalloc(1, sizeof (strg_t)); 1330 strg->st_tspec = CHAR; 1331 strg->st_len = len; 1332 strg->st_cp = s; 1333 1334 yylval.y_strg = strg; 1335 return (T_STRING); 1336 } 1337 1338 static int 1339 wcstrg(void) 1340 { 1341 char *s; 1342 int c, n; 1343 size_t i, wi; 1344 size_t len, max, wlen; 1345 wchar_t *ws; 1346 strg_t *strg; 1347 1348 s = xmalloc(max = 64); 1349 len = 0; 1350 while ((c = getescc('"')) >= 0) { 1351 /* +1 to save space for a trailing NUL character */ 1352 if (len + 1 >= max) 1353 s = xrealloc(s, max *= 2); 1354 s[len++] = (char)c; 1355 } 1356 s[len] = '\0'; 1357 if (c == -2) 1358 /* unterminated string constant */ 1359 error(258); 1360 1361 /* get length of wide-character string */ 1362 (void)mblen(NULL, 0); 1363 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1364 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1365 /* invalid multibyte character */ 1366 error(291); 1367 break; 1368 } 1369 if (n == 0) 1370 n = 1; 1371 } 1372 1373 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1374 1375 /* convert from multibyte to wide char */ 1376 (void)mbtowc(NULL, NULL, 0); 1377 for (i = 0, wi = 0; i < len; i += n, wi++) { 1378 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1379 break; 1380 if (n == 0) 1381 n = 1; 1382 } 1383 ws[wi] = 0; 1384 free(s); 1385 1386 strg = xcalloc(1, sizeof (strg_t)); 1387 strg->st_tspec = WCHAR; 1388 strg->st_len = wlen; 1389 strg->st_wcp = ws; 1390 1391 yylval.y_strg = strg; 1392 return (T_STRING); 1393 } 1394 1395 /* 1396 * As noted above the scanner does not create new symbol table entries 1397 * for symbols it cannot find in the symbol table. This is to avoid 1398 * putting undeclared symbols into the symbol table if a syntax error 1399 * occurs. 1400 * 1401 * getsym() is called as soon as it is probably ok to put the symbol to 1402 * the symbol table. This does not mean that it is not possible that 1403 * symbols are put to the symbol table which are than not completely 1404 * declared due to syntax errors. To avoid too many problems in this 1405 * case symbols get type int in getsym(). 1406 * 1407 * XXX calls to getsym() should be delayed until decl1*() is called 1408 */ 1409 sym_t * 1410 getsym(sbuf_t *sb) 1411 { 1412 dinfo_t *di; 1413 char *s; 1414 sym_t *sym; 1415 1416 sym = sb->sb_sym; 1417 1418 /* 1419 * During member declaration it is possible that name() looked 1420 * for symbols of type FVFT, although it should have looked for 1421 * symbols of type FTAG. Same can happen for labels. Both cases 1422 * are compensated here. 1423 */ 1424 if (symtyp == FMOS || symtyp == FLAB) { 1425 if (sym == NULL || sym->s_kind == FVFT) 1426 sym = search(sb); 1427 } 1428 1429 if (sym != NULL) { 1430 if (sym->s_kind != symtyp) 1431 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1432 symtyp = FVFT; 1433 freesb(sb); 1434 return (sym); 1435 } 1436 1437 /* create a new symbol table entry */ 1438 1439 /* labels must always be allocated at level 1 (outhermost block) */ 1440 if (symtyp == FLAB) { 1441 sym = getlblk(1, sizeof (sym_t)); 1442 s = getlblk(1, sb->sb_len + 1); 1443 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1444 sym->s_name = s; 1445 sym->s_blklev = 1; 1446 di = dcs; 1447 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1448 di = di->d_nxt; 1449 if (di->d_ctx != AUTO) 1450 LERROR("storesym()"); 1451 } else { 1452 sym = getblk(sizeof (sym_t)); 1453 sym->s_name = sb->sb_name; 1454 sym->s_blklev = blklev; 1455 di = dcs; 1456 } 1457 1458 UNIQUE_CURR_POS(sym->s_dpos); 1459 if ((sym->s_kind = symtyp) != FLAB) 1460 sym->s_type = gettyp(INT); 1461 1462 symtyp = FVFT; 1463 1464 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1465 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1466 sym->s_rlink = &symtab[sb->sb_hash]; 1467 symtab[sb->sb_hash] = sym; 1468 1469 *di->d_ldlsym = sym; 1470 di->d_ldlsym = &sym->s_dlnxt; 1471 1472 freesb(sb); 1473 return (sym); 1474 } 1475 1476 /* 1477 * Construct a temporary symbol. The symbol starts with a digit, so that 1478 * it is illegal. 1479 */ 1480 sym_t * 1481 mktempsym(type_t *t) 1482 { 1483 static int n = 0; 1484 int h; 1485 char *s = getlblk(blklev, 64); 1486 sym_t *sym = getblk(sizeof (sym_t)); 1487 1488 (void)snprintf(s, 64, "%.8d_tmp", n++); 1489 h = hash(s); 1490 1491 sym->s_name = s; 1492 sym->s_type = t; 1493 sym->s_blklev = blklev; 1494 sym->s_scl = AUTO; 1495 sym->s_kind = FVFT; 1496 sym->s_used = 1; 1497 sym->s_set = 1; 1498 1499 if ((sym->s_link = symtab[h]) != NULL) 1500 symtab[h]->s_rlink = &sym->s_link; 1501 sym->s_rlink = &symtab[h]; 1502 symtab[h] = sym; 1503 1504 *dcs->d_ldlsym = sym; 1505 dcs->d_ldlsym = &sym->s_dlnxt; 1506 1507 return sym; 1508 } 1509 1510 /* 1511 * Remove a symbol forever from the symbol table. s_blklev 1512 * is set to -1 to avoid that the symbol will later be put 1513 * back to the symbol table. 1514 */ 1515 void 1516 rmsym(sym_t *sym) 1517 { 1518 1519 if ((*sym->s_rlink = sym->s_link) != NULL) 1520 sym->s_link->s_rlink = sym->s_rlink; 1521 sym->s_blklev = -1; 1522 sym->s_link = NULL; 1523 } 1524 1525 /* 1526 * Remove a list of symbols declared at one level from the symbol 1527 * table. 1528 */ 1529 void 1530 rmsyms(sym_t *syms) 1531 { 1532 sym_t *sym; 1533 1534 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1535 if (sym->s_blklev != -1) { 1536 if ((*sym->s_rlink = sym->s_link) != NULL) 1537 sym->s_link->s_rlink = sym->s_rlink; 1538 sym->s_link = NULL; 1539 sym->s_rlink = NULL; 1540 } 1541 } 1542 } 1543 1544 /* 1545 * Put a symbol into the symbol table 1546 */ 1547 void 1548 inssym(int bl, sym_t *sym) 1549 { 1550 int h; 1551 1552 h = hash(sym->s_name); 1553 if ((sym->s_link = symtab[h]) != NULL) 1554 symtab[h]->s_rlink = &sym->s_link; 1555 sym->s_rlink = &symtab[h]; 1556 symtab[h] = sym; 1557 sym->s_blklev = bl; 1558 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1559 LERROR("inssym()"); 1560 } 1561 1562 /* 1563 * Called at level 0 after syntax errors 1564 * Removes all symbols which are not declared at level 0 from the 1565 * symbol table. Also frees all memory which is not associated with 1566 * level 0. 1567 */ 1568 void 1569 cleanup(void) 1570 { 1571 sym_t *sym, *nsym; 1572 int i; 1573 1574 for (i = 0; i < HSHSIZ1; i++) { 1575 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1576 nsym = sym->s_link; 1577 if (sym->s_blklev >= 1) { 1578 if ((*sym->s_rlink = nsym) != NULL) 1579 nsym->s_rlink = sym->s_rlink; 1580 } 1581 } 1582 } 1583 1584 for (i = mblklev; i > 0; i--) 1585 freelblk(i); 1586 } 1587 1588 /* 1589 * Create a new symbol with the name of an existing symbol. 1590 */ 1591 sym_t * 1592 pushdown(sym_t *sym) 1593 { 1594 int h; 1595 sym_t *nsym; 1596 1597 h = hash(sym->s_name); 1598 nsym = getblk(sizeof (sym_t)); 1599 if (sym->s_blklev > blklev) 1600 LERROR("pushdown()"); 1601 nsym->s_name = sym->s_name; 1602 UNIQUE_CURR_POS(nsym->s_dpos); 1603 nsym->s_kind = sym->s_kind; 1604 nsym->s_blklev = blklev; 1605 1606 if ((nsym->s_link = symtab[h]) != NULL) 1607 symtab[h]->s_rlink = &nsym->s_link; 1608 nsym->s_rlink = &symtab[h]; 1609 symtab[h] = nsym; 1610 1611 *dcs->d_ldlsym = nsym; 1612 dcs->d_ldlsym = &nsym->s_dlnxt; 1613 1614 return (nsym); 1615 } 1616 1617 /* 1618 * Free any dynamically allocated memory referenced by 1619 * the value stack or yylval. 1620 * The type of information in yylval is described by tok. 1621 */ 1622 void 1623 freeyyv(void *sp, int tok) 1624 { 1625 if (tok == T_NAME || tok == T_TYPENAME) { 1626 sbuf_t *sb = *(sbuf_t **)sp; 1627 freesb(sb); 1628 } else if (tok == T_CON) { 1629 val_t *val = *(val_t **)sp; 1630 free(val); 1631 } else if (tok == T_STRING) { 1632 strg_t *strg = *(strg_t **)sp; 1633 if (strg->st_tspec == CHAR) { 1634 free(strg->st_cp); 1635 } else if (strg->st_tspec == WCHAR) { 1636 free(strg->st_wcp); 1637 } else { 1638 LERROR("fryylv()"); 1639 } 1640 free(strg); 1641 } 1642 } 1643
Indexes created Mon Oct 20 20:10:13 GMT 2025