scan.l revision 1.81
1 %{ 2 /* $NetBSD: scan.l,v 1.81 2018/06/29 20:18:36 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 1.81 2018/06/29 20:18:36 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 ((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 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 }, 212 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 213 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 }, 214 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 }, 215 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 }, 216 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 }, 217 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 }, 218 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 }, 219 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 }, 220 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 }, 221 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 }, 222 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 }, 223 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 }, 224 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 }, 225 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 }, 226 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 }, 227 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 }, 228 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 }, 229 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 }, 230 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 }, 231 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 }, 232 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 }, 233 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 }, 234 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 }, 235 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 }, 236 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 }, 237 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 }, 238 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 }, 239 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 }, 240 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 }, 241 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 }, 242 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 }, 243 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 }, 244 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 }, 245 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 }, 246 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 }, 247 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 }, 248 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 249 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 }, 250 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 }, 251 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 252 0, 0, 0, 0,0,1,1,5 }, 253 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 }, 254 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 }, 255 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 }, 256 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 }, 257 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 }, 258 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 259 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 260 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 261 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 262 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 263 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 264 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 265 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 266 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 267 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 268 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 269 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 270 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 271 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 272 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 273 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 274 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 }, 275 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 276 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 277 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 278 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 }, 279 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 280 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 281 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 }, 282 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 283 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 284 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 285 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 286 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 287 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 288 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 289 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 290 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 291 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 292 }; 293 294 /* Symbol table */ 295 static sym_t *symtab[HSHSIZ1]; 296 297 /* bit i of the entry with index i is set */ 298 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 299 300 /* least significant i bits are set in the entry with index i */ 301 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 302 303 /* least significant i bits are not set in the entry with index i */ 304 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 305 306 /* free list for sbuf structures */ 307 static sbuf_t *sbfrlst; 308 309 /* Typ of next expected symbol */ 310 symt_t symtyp; 311 312 313 static void 314 addkw(struct kwtab *kw, int deco) 315 { 316 sym_t *sym; 317 size_t h; 318 char buf[256]; 319 const char *name; 320 321 if (!(kw->kw_deco & deco)) 322 return; 323 324 switch (deco) { 325 case 1: 326 name = kw->kw_name; 327 break; 328 case 2: 329 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 330 name = strdup(buf); 331 break; 332 case 4: 333 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 334 name = strdup(buf); 335 break; 336 default: 337 abort(); 338 } 339 340 if (name == NULL) 341 err(1, "Can't init symbol table"); 342 343 sym = getblk(sizeof (sym_t)); 344 sym->s_name = name; 345 sym->s_keyw = kw; 346 sym->s_value.v_quad = kw->kw_token; 347 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 348 sym->s_tspec = kw->kw_tspec; 349 } else if (kw->kw_token == T_SCLASS) { 350 sym->s_scl = kw->kw_scl; 351 } else if (kw->kw_token == T_QUAL) { 352 sym->s_tqual = kw->kw_tqual; 353 } 354 h = hash(sym->s_name); 355 if ((sym->s_link = symtab[h]) != NULL) 356 symtab[h]->s_rlink = &sym->s_link; 357 sym->s_rlink = &symtab[h]; 358 symtab[h] = sym; 359 } 360 361 /* 362 * All keywords are written to the symbol table. This saves us looking 363 * in a extra table for each name we found. 364 */ 365 void 366 initscan(void) 367 { 368 struct kwtab *kw; 369 size_t i; 370 uint64_t uq; 371 372 for (kw = kwtab; kw->kw_name != NULL; kw++) { 373 if ((kw->kw_c89 || kw->kw_c99) && tflag) 374 continue; 375 if (kw->kw_c99 && !(Sflag || gflag)) 376 continue; 377 if (kw->kw_gcc && !gflag) 378 continue; 379 addkw(kw, 1); 380 addkw(kw, 2); 381 addkw(kw, 4); 382 } 383 384 /* initialize bit-masks for quads */ 385 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 386 qbmasks[i] = (uint64_t)1 << i; 387 uq = ~(uint64_t)0 << i; 388 qumasks[i] = uq; 389 qlmasks[i] = ~uq; 390 } 391 qumasks[i] = 0; 392 qlmasks[i] = ~(uint64_t)0; 393 } 394 395 /* 396 * Get a free sbuf structure, if possible from the free list 397 */ 398 static sbuf_t * 399 allocsb(void) 400 { 401 sbuf_t *sb; 402 403 if ((sb = sbfrlst) != NULL) { 404 sbfrlst = sb->sb_nxt; 405 #ifdef BLKDEBUG 406 (void)memset(sb, 0, sizeof (*sb)); 407 #else 408 sb->sb_nxt = NULL; 409 #endif 410 } else { 411 sb = xmalloc(sizeof (sbuf_t)); 412 (void)memset(sb, 0, sizeof (*sb)); 413 } 414 return (sb); 415 } 416 417 /* 418 * Put a sbuf structure to the free list 419 */ 420 static void 421 freesb(sbuf_t *sb) 422 { 423 424 (void)memset(sb, ZERO, sizeof (*sb)); 425 sb->sb_nxt = sbfrlst; 426 sbfrlst = sb; 427 } 428 429 /* 430 * Read a character and ensure that it is positive (except EOF). 431 * Increment line count(s) if necessary. 432 */ 433 static int 434 inpc(void) 435 { 436 int c; 437 438 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 439 incline(); 440 return (c); 441 } 442 443 static int 444 hash(const char *s) 445 { 446 u_int v; 447 const u_char *us; 448 449 v = 0; 450 for (us = (const u_char *)s; *us != '\0'; us++) { 451 v = (v << sizeof (v)) + *us; 452 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 453 } 454 return (v % HSHSIZ1); 455 } 456 457 /* 458 * Lex has found a letter followed by zero or more letters or digits. 459 * It looks for a symbol in the symbol table with the same name. This 460 * symbol must either be a keyword or a symbol of the type required by 461 * symtyp (label, member, tag, ...). 462 * 463 * If it is a keyword, the token is returned. In some cases it is described 464 * more deeply by data written to yylval. 465 * 466 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 467 * is stored in yylval. This struct contains the name of the symbol, its 468 * length and hash value. If there is already a symbol of the same name 469 * and type in the symbol table, the sbuf struct also contains a pointer 470 * to the symbol table entry. 471 */ 472 static int 473 name(void) 474 { 475 char *s; 476 sbuf_t *sb; 477 sym_t *sym; 478 int tok; 479 480 sb = allocsb(); 481 sb->sb_name = yytext; 482 sb->sb_len = yyleng; 483 sb->sb_hash = hash(yytext); 484 if ((sym = search(sb)) != NULL && sym->s_keyw) { 485 freesb(sb); 486 return (keyw(sym)); 487 } 488 489 sb->sb_sym = sym; 490 491 if (sym != NULL) { 492 if (blklev < sym->s_blklev) 493 LERROR("name()"); 494 sb->sb_name = sym->s_name; 495 sb->sb_len = strlen(sym->s_name); 496 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 497 } else { 498 s = getblk(yyleng + 1); 499 (void)memcpy(s, yytext, yyleng + 1); 500 sb->sb_name = s; 501 sb->sb_len = yyleng; 502 tok = T_NAME; 503 } 504 505 yylval.y_sb = sb; 506 return (tok); 507 } 508 509 static sym_t * 510 search(sbuf_t *sb) 511 { 512 sym_t *sym; 513 514 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 515 if (strcmp(sym->s_name, sb->sb_name) == 0) { 516 if (sym->s_keyw) { 517 struct kwtab *kw = sym->s_keyw; 518 if (!kw->kw_attr || attron) 519 return (sym); 520 } else if (!attron && sym->s_kind == symtyp) 521 return sym; 522 } 523 } 524 525 return (NULL); 526 } 527 528 static int 529 keyw(sym_t *sym) 530 { 531 int t; 532 533 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 534 yylval.y_scl = sym->s_scl; 535 } else if (t == T_TYPE || t == T_SOU) { 536 yylval.y_tspec = sym->s_tspec; 537 } else if (t == T_QUAL) { 538 yylval.y_tqual = sym->s_tqual; 539 } 540 return (t); 541 } 542 543 /* 544 * Convert a string representing an integer into internal representation. 545 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 546 */ 547 static int 548 icon(int base) 549 { 550 int l_suffix, u_suffix; 551 int len; 552 const char *cp; 553 char c, *eptr; 554 tspec_t typ; 555 uint64_t uq = 0; 556 int ansiu; 557 static tspec_t contypes[2][3] = { 558 { INT, LONG, QUAD }, 559 { UINT, ULONG, UQUAD } 560 }; 561 562 cp = yytext; 563 len = yyleng; 564 565 /* skip 0[xX] or 0[bB] */ 566 if (base == 16 || base == 2) { 567 cp += 2; 568 len -= 2; 569 } 570 571 /* read suffixes */ 572 l_suffix = u_suffix = 0; 573 for ( ; ; ) { 574 if ((c = cp[len - 1]) == 'l' || c == 'L') { 575 l_suffix++; 576 } else if (c == 'u' || c == 'U') { 577 u_suffix++; 578 } else { 579 break; 580 } 581 len--; 582 } 583 if (l_suffix > 2 || u_suffix > 1) { 584 /* malformed integer constant */ 585 warning(251); 586 if (l_suffix > 2) 587 l_suffix = 2; 588 if (u_suffix > 1) 589 u_suffix = 1; 590 } 591 if (tflag && u_suffix != 0) { 592 /* suffix U is illegal in traditional C */ 593 warning(97); 594 } 595 typ = contypes[u_suffix][l_suffix]; 596 597 errno = 0; 598 599 uq = strtouq(cp, &eptr, base); 600 if (eptr != cp + len) 601 LERROR("icon()"); 602 if (errno != 0) 603 /* integer constant out of range */ 604 warning(252); 605 606 /* 607 * If the value is too big for the current type, we must choose 608 * another type. 609 */ 610 ansiu = 0; 611 switch (typ) { 612 case INT: 613 if (uq <= TARG_INT_MAX) { 614 /* ok */ 615 } else if (uq <= TARG_UINT_MAX && base != 10) { 616 typ = UINT; 617 } else if (uq <= TARG_LONG_MAX) { 618 typ = LONG; 619 } else { 620 typ = ULONG; 621 if (uq > TARG_ULONG_MAX) { 622 /* integer constant out of range */ 623 warning(252); 624 } 625 } 626 if (typ == UINT || typ == ULONG) { 627 if (tflag) { 628 typ = LONG; 629 } else if (!sflag) { 630 /* 631 * Remember that the constant is unsigned 632 * only in ANSI C 633 */ 634 ansiu = 1; 635 } 636 } 637 break; 638 case UINT: 639 if (uq > TARG_UINT_MAX) { 640 typ = ULONG; 641 if (uq > TARG_ULONG_MAX) { 642 /* integer constant out of range */ 643 warning(252); 644 } 645 } 646 break; 647 case LONG: 648 if (uq > TARG_LONG_MAX && !tflag) { 649 typ = ULONG; 650 if (!sflag) 651 ansiu = 1; 652 if (uq > TARG_ULONG_MAX) { 653 /* integer constant out of range */ 654 warning(252); 655 } 656 } 657 break; 658 case ULONG: 659 if (uq > TARG_ULONG_MAX) { 660 /* integer constant out of range */ 661 warning(252); 662 } 663 break; 664 case QUAD: 665 if (uq > TARG_QUAD_MAX && !tflag) { 666 typ = UQUAD; 667 if (!sflag) 668 ansiu = 1; 669 } 670 break; 671 case UQUAD: 672 if (uq > TARG_UQUAD_MAX) { 673 /* integer constant out of range */ 674 warning(252); 675 } 676 break; 677 /* LINTED206: (enumeration values not handled in switch) */ 678 case STRUCT: 679 case VOID: 680 case LDOUBLE: 681 case FUNC: 682 case ARRAY: 683 case PTR: 684 case ENUM: 685 case UNION: 686 case SIGNED: 687 case NOTSPEC: 688 case DOUBLE: 689 case FLOAT: 690 case USHORT: 691 case SHORT: 692 case UCHAR: 693 case SCHAR: 694 case CHAR: 695 case BOOL: 696 case UNSIGN: 697 case FCOMPLEX: 698 case DCOMPLEX: 699 case LCOMPLEX: 700 case COMPLEX: 701 break; 702 703 case NTSPEC: /* this value unused */ 704 break; 705 } 706 707 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 708 709 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 710 yylval.y_val->v_ansiu = ansiu; 711 yylval.y_val->v_quad = (int64_t)uq; 712 713 return (T_CON); 714 } 715 716 /* 717 * Returns 1 if t is a signed type and the value is negative. 718 * 719 * len is the number of significant bits. If len is -1, len is set 720 * to the width of type t. 721 */ 722 int 723 sign(int64_t q, tspec_t t, int len) 724 { 725 726 if (t == PTR || isutyp(t)) 727 return (0); 728 return (msb(q, t, len)); 729 } 730 731 int 732 msb(int64_t q, tspec_t t, int len) 733 { 734 735 if (len <= 0) 736 len = size(t); 737 return ((q & qbmasks[len - 1]) != 0); 738 } 739 740 /* 741 * Extends the sign of q. 742 */ 743 int64_t 744 xsign(int64_t q, tspec_t t, int len) 745 { 746 747 if (len <= 0) 748 len = size(t); 749 750 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 751 q &= qlmasks[len]; 752 } else { 753 q |= qumasks[len]; 754 } 755 return (q); 756 } 757 758 /* 759 * Convert a string representing a floating point value into its interal 760 * representation. Type and value are returned in yylval. fcon() 761 * (and yylex()) returns T_CON. 762 * XXX Currently it is not possible to convert constants of type 763 * long double which are greater than DBL_MAX. 764 */ 765 static int 766 fcon(void) 767 { 768 const char *cp; 769 int len; 770 tspec_t typ; 771 char c, *eptr; 772 double d; 773 float f = 0; 774 775 cp = yytext; 776 len = yyleng; 777 778 if (cp[len - 1] == 'i') { 779 /* imaginary, do nothing for now */ 780 len--; 781 } 782 if ((c = cp[len - 1]) == 'f' || c == 'F') { 783 typ = FLOAT; 784 len--; 785 } else if (c == 'l' || c == 'L') { 786 typ = LDOUBLE; 787 len--; 788 } else { 789 if (c == 'd' || c == 'D') 790 len--; 791 typ = DOUBLE; 792 } 793 794 if (tflag && typ != DOUBLE) { 795 /* suffixes F and L are illegal in traditional C */ 796 warning(98); 797 } 798 799 errno = 0; 800 d = strtod(cp, &eptr); 801 if (eptr != cp + len) { 802 switch (*eptr) { 803 /* 804 * XXX: non-native non-current strtod() may not handle hex 805 * floats, ignore the rest if we find traces of hex float 806 * syntax... 807 */ 808 case 'p': 809 case 'P': 810 case 'x': 811 case 'X': 812 d = 0; 813 errno = 0; 814 break; 815 default: 816 LERROR("fcon(%s->%s)", cp, eptr); 817 } 818 } 819 if (errno != 0) 820 /* floating-point constant out of range */ 821 warning(248); 822 823 if (typ == FLOAT) { 824 f = (float)d; 825 if (!finite(f)) { 826 /* floating-point constant out of range */ 827 warning(248); 828 f = f > 0 ? FLT_MAX : -FLT_MAX; 829 } 830 } 831 832 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 833 if (typ == FLOAT) { 834 yylval.y_val->v_ldbl = f; 835 } else { 836 yylval.y_val->v_ldbl = d; 837 } 838 839 return (T_CON); 840 } 841 842 static int 843 operator(int t, op_t o) 844 { 845 846 yylval.y_op = o; 847 return (t); 848 } 849 850 /* 851 * Called if lex found a leading \'. 852 */ 853 static int 854 ccon(void) 855 { 856 size_t n; 857 int val, c; 858 char cv; 859 860 n = 0; 861 val = 0; 862 while ((c = getescc('\'')) >= 0) { 863 val = (val << CHAR_BIT) + c; 864 n++; 865 } 866 if (c == -2) { 867 /* unterminated character constant */ 868 error(253); 869 } else { 870 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 871 /* too many characters in character constant */ 872 error(71); 873 } else if (n > 1) { 874 /* multi-character character constant */ 875 warning(294); 876 } else if (n == 0) { 877 /* empty character constant */ 878 error(73); 879 } 880 } 881 if (n == 1) { 882 cv = (char)val; 883 val = cv; 884 } 885 886 yylval.y_val = xcalloc(1, sizeof (val_t)); 887 yylval.y_val->v_tspec = INT; 888 yylval.y_val->v_quad = val; 889 890 return (T_CON); 891 } 892 893 /* 894 * Called if lex found a leading L\' 895 */ 896 static int 897 wccon(void) 898 { 899 static char buf[MB_LEN_MAX + 1]; 900 size_t i; 901 int c; 902 wchar_t wc; 903 904 i = 0; 905 while ((c = getescc('\'')) >= 0) { 906 if (i < MB_CUR_MAX) 907 buf[i] = (char)c; 908 i++; 909 } 910 911 wc = 0; 912 913 if (c == -2) { 914 /* unterminated character constant */ 915 error(253); 916 } else if (c == 0) { 917 /* empty character constant */ 918 error(73); 919 } else { 920 if (i > MB_CUR_MAX) { 921 i = MB_CUR_MAX; 922 /* too many characters in character constant */ 923 error(71); 924 } else { 925 buf[i] = '\0'; 926 (void)mbtowc(NULL, NULL, 0); 927 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 928 /* invalid multibyte character */ 929 error(291); 930 } 931 } 932 933 yylval.y_val = xcalloc(1, sizeof (val_t)); 934 yylval.y_val->v_tspec = WCHAR; 935 yylval.y_val->v_quad = wc; 936 937 return (T_CON); 938 } 939 940 /* 941 * Read a character which is part of a character constant or of a string 942 * and handle escapes. 943 * 944 * The Argument is the character which delimits the character constant or 945 * string. 946 * 947 * Returns -1 if the end of the character constant or string is reached, 948 * -2 if the EOF is reached, and the character otherwise. 949 */ 950 static int 951 getescc(int d) 952 { 953 static int pbc = -1; 954 int n, c, v; 955 956 if (pbc == -1) { 957 c = inpc(); 958 } else { 959 c = pbc; 960 pbc = -1; 961 } 962 if (c == d) 963 return (-1); 964 switch (c) { 965 case '\n': 966 if (tflag) { 967 /* newline in string or char constant */ 968 error(254); 969 return (-2); 970 } 971 return (c); 972 case EOF: 973 return (-2); 974 case '\\': 975 switch (c = inpc()) { 976 case '"': 977 if (tflag && d == '\'') 978 /* \" inside character constant undef. ... */ 979 warning(262); 980 return ('"'); 981 case '\'': 982 return ('\''); 983 case '?': 984 if (tflag) 985 /* \? undefined in traditional C */ 986 warning(263); 987 return ('?'); 988 case '\\': 989 return ('\\'); 990 case 'a': 991 if (tflag) 992 /* \a undefined in traditional C */ 993 warning(81); 994 return ('\a'); 995 case 'b': 996 return ('\b'); 997 case 'f': 998 return ('\f'); 999 case 'n': 1000 return ('\n'); 1001 case 'r': 1002 return ('\r'); 1003 case 't': 1004 return ('\t'); 1005 case 'v': 1006 if (tflag) 1007 /* \v undefined in traditional C */ 1008 warning(264); 1009 return ('\v'); 1010 case '8': case '9': 1011 /* bad octal digit %c */ 1012 warning(77, c); 1013 /* FALLTHROUGH */ 1014 case '0': case '1': case '2': case '3': 1015 case '4': case '5': case '6': case '7': 1016 n = 3; 1017 v = 0; 1018 do { 1019 v = (v << 3) + (c - '0'); 1020 c = inpc(); 1021 } while (--n && isdigit(c) && (tflag || c <= '7')); 1022 if (tflag && n > 0 && isdigit(c)) 1023 /* bad octal digit %c */ 1024 warning(77, c); 1025 pbc = c; 1026 if (v > TARG_UCHAR_MAX) { 1027 /* character escape does not fit in char. */ 1028 warning(76); 1029 v &= CHAR_MASK; 1030 } 1031 return (v); 1032 case 'x': 1033 if (tflag) 1034 /* \x undefined in traditional C */ 1035 warning(82); 1036 v = 0; 1037 n = 0; 1038 while ((c = inpc()) >= 0 && isxdigit(c)) { 1039 c = isdigit(c) ? 1040 c - '0' : toupper(c) - 'A' + 10; 1041 v = (v << 4) + c; 1042 if (n >= 0) { 1043 if ((v & ~CHAR_MASK) != 0) { 1044 /* overflow in hex escape */ 1045 warning(75); 1046 n = -1; 1047 } else { 1048 n++; 1049 } 1050 } 1051 } 1052 pbc = c; 1053 if (n == 0) { 1054 /* no hex digits follow \x */ 1055 error(74); 1056 } if (n == -1) { 1057 v &= CHAR_MASK; 1058 } 1059 return (v); 1060 case '\n': 1061 return (getescc(d)); 1062 case EOF: 1063 return (-2); 1064 default: 1065 if (isprint(c)) { 1066 /* dubious escape \%c */ 1067 warning(79, c); 1068 } else { 1069 /* dubious escape \%o */ 1070 warning(80, c); 1071 } 1072 } 1073 } 1074 return (c); 1075 } 1076 1077 /* 1078 * Called for preprocessor directives. Currently implemented are: 1079 * # lineno 1080 * # lineno "filename" 1081 */ 1082 static void 1083 directive(void) 1084 { 1085 const char *cp, *fn; 1086 char c, *eptr; 1087 size_t fnl; 1088 long ln; 1089 static int first = 1; 1090 1091 /* Go to first non-whitespace after # */ 1092 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1093 continue; 1094 1095 if (!isdigit((unsigned char)c)) { 1096 if (strncmp(cp, "pragma", 6) == 0 1097 && isspace((unsigned char)cp[6])) 1098 return; 1099 error: 1100 /* undefined or invalid # directive */ 1101 warning(255); 1102 return; 1103 } 1104 ln = strtol(--cp, &eptr, 10); 1105 if (cp == eptr) 1106 goto error; 1107 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1108 goto error; 1109 while ((c = *cp++) == ' ' || c == '\t') 1110 continue; 1111 if (c != '\0') { 1112 if (c != '"') 1113 goto error; 1114 fn = cp; 1115 while ((c = *cp) != '"' && c != '\0') 1116 cp++; 1117 if (c != '"') 1118 goto error; 1119 if ((fnl = cp++ - fn) > PATH_MAX) 1120 goto error; 1121 while ((c = *cp++) == ' ' || c == '\t') 1122 continue; 1123 #if 0 1124 if (c != '\0') 1125 warning("extra character(s) after directive"); 1126 #endif 1127 1128 /* empty string means stdin */ 1129 if (fnl == 0) { 1130 fn = "{standard input}"; 1131 fnl = 16; /* strlen (fn) */ 1132 } 1133 curr_pos.p_file = fnnalloc(fn, fnl); 1134 /* 1135 * If this is the first directive, the name is the name 1136 * of the C source file as specified at the command line. 1137 * It is written to the output file. 1138 */ 1139 if (first) { 1140 csrc_pos.p_file = curr_pos.p_file; 1141 outsrc(fnxform(curr_pos.p_file, 1142 strlen(curr_pos.p_file))); 1143 first = 0; 1144 } 1145 } 1146 curr_pos.p_line = (int)ln - 1; 1147 curr_pos.p_uniq = 0; 1148 if (curr_pos.p_file == csrc_pos.p_file) { 1149 csrc_pos.p_line = (int)ln - 1; 1150 csrc_pos.p_uniq = 0; 1151 } 1152 } 1153 1154 /* 1155 * Handle lint comments. Following comments are currently understood: 1156 * ARGSUSEDn 1157 * BITFIELDTYPE 1158 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1159 * FALLTHRU FALLTHROUGH 1160 * LINTLIBRARY 1161 * LINTEDn NOSTRICTn 1162 * LONGLONG 1163 * NOTREACHED 1164 * PRINTFLIKEn 1165 * PROTOLIB 1166 * SCANFLIKEn 1167 * VARARGSn 1168 * If one of this comments is recognized, the arguments, if any, are 1169 * parsed and a function which handles this comment is called. 1170 */ 1171 static void 1172 comment(void) 1173 { 1174 int c, lc; 1175 static struct { 1176 const char *keywd; 1177 int arg; 1178 void (*func)(int); 1179 } keywtab[] = { 1180 { "ARGSUSED", 1, argsused }, 1181 { "BITFIELDTYPE", 0, bitfieldtype }, 1182 { "CONSTCOND", 0, constcond }, 1183 { "CONSTANTCOND", 0, constcond }, 1184 { "CONSTANTCONDITION", 0, constcond }, 1185 { "FALLTHRU", 0, fallthru }, 1186 { "FALLTHROUGH", 0, fallthru }, 1187 { "LINTLIBRARY", 0, lintlib }, 1188 { "LINTED", 1, linted }, 1189 { "LONGLONG", 0, longlong }, 1190 { "NOSTRICT", 1, linted }, 1191 { "NOTREACHED", 0, notreach }, 1192 { "PRINTFLIKE", 1, printflike }, 1193 { "PROTOLIB", 1, protolib }, 1194 { "SCANFLIKE", 1, scanflike }, 1195 { "VARARGS", 1, varargs }, 1196 }; 1197 char keywd[32]; 1198 char arg[32]; 1199 size_t l, i; 1200 int a; 1201 int eoc; 1202 1203 eoc = 0; 1204 1205 /* Skip white spaces after the start of the comment */ 1206 while ((c = inpc()) != EOF && isspace(c)) 1207 continue; 1208 1209 /* Read the potential keyword to keywd */ 1210 l = 0; 1211 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1212 keywd[l++] = (char)c; 1213 c = inpc(); 1214 } 1215 keywd[l] = '\0'; 1216 1217 /* look for the keyword */ 1218 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1219 if (strcmp(keywtab[i].keywd, keywd) == 0) 1220 break; 1221 } 1222 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1223 goto skip_rest; 1224 1225 /* skip white spaces after the keyword */ 1226 while (c != EOF && isspace(c)) 1227 c = inpc(); 1228 1229 /* read the argument, if the keyword accepts one and there is one */ 1230 l = 0; 1231 if (keywtab[i].arg) { 1232 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1233 arg[l++] = (char)c; 1234 c = inpc(); 1235 } 1236 } 1237 arg[l] = '\0'; 1238 a = l != 0 ? atoi(arg) : -1; 1239 1240 /* skip white spaces after the argument */ 1241 while (c != EOF && isspace(c)) 1242 c = inpc(); 1243 1244 if (c != '*' || (c = inpc()) != '/') { 1245 if (keywtab[i].func != linted) 1246 /* extra characters in lint comment */ 1247 warning(257); 1248 } else { 1249 /* 1250 * remember that we have already found the end of the 1251 * comment 1252 */ 1253 eoc = 1; 1254 } 1255 1256 if (keywtab[i].func != NULL) 1257 (*keywtab[i].func)(a); 1258 1259 skip_rest: 1260 while (!eoc) { 1261 lc = c; 1262 if ((c = inpc()) == EOF) { 1263 /* unterminated comment */ 1264 error(256); 1265 break; 1266 } 1267 if (lc == '*' && c == '/') 1268 eoc = 1; 1269 } 1270 } 1271 1272 /* 1273 * Handle // style comments 1274 */ 1275 static void 1276 slashslashcomment(void) 1277 { 1278 int c; 1279 1280 if (!Sflag && !gflag) 1281 /* // comments only supported in C99 */ 1282 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1283 1284 while ((c = inpc()) != EOF && c != '\n') 1285 continue; 1286 } 1287 1288 /* 1289 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1290 * clrwflgs() is called after function definitions and global and 1291 * local declarations and definitions. It is also called between 1292 * the controlling expression and the body of control statements 1293 * (if, switch, for, while). 1294 */ 1295 void 1296 clrwflgs(void) 1297 { 1298 1299 lwarn = LWARN_ALL; 1300 quadflg = 0; 1301 ccflg = 0; 1302 } 1303 1304 /* 1305 * Strings are stored in a dynamically alloceted buffer and passed 1306 * in yylval.y_xstrg to the parser. The parser or the routines called 1307 * by the parser are responsible for freeing this buffer. 1308 */ 1309 static int 1310 string(void) 1311 { 1312 u_char *s; 1313 int c; 1314 size_t len, max; 1315 strg_t *strg; 1316 1317 s = xmalloc(max = 64); 1318 1319 len = 0; 1320 while ((c = getescc('"')) >= 0) { 1321 /* +1 to reserve space for a trailing NUL character */ 1322 if (len + 1 == max) 1323 s = xrealloc(s, max *= 2); 1324 s[len++] = (char)c; 1325 } 1326 s[len] = '\0'; 1327 if (c == -2) 1328 /* unterminated string constant */ 1329 error(258); 1330 1331 strg = xcalloc(1, sizeof (strg_t)); 1332 strg->st_tspec = CHAR; 1333 strg->st_len = len; 1334 strg->st_cp = s; 1335 1336 yylval.y_strg = strg; 1337 return (T_STRING); 1338 } 1339 1340 static int 1341 wcstrg(void) 1342 { 1343 char *s; 1344 int c, n; 1345 size_t i, wi; 1346 size_t len, max, wlen; 1347 wchar_t *ws; 1348 strg_t *strg; 1349 1350 s = xmalloc(max = 64); 1351 len = 0; 1352 while ((c = getescc('"')) >= 0) { 1353 /* +1 to save space for a trailing NUL character */ 1354 if (len + 1 >= max) 1355 s = xrealloc(s, max *= 2); 1356 s[len++] = (char)c; 1357 } 1358 s[len] = '\0'; 1359 if (c == -2) 1360 /* unterminated string constant */ 1361 error(258); 1362 1363 /* get length of wide-character string */ 1364 (void)mblen(NULL, 0); 1365 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1366 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1367 /* invalid multibyte character */ 1368 error(291); 1369 break; 1370 } 1371 if (n == 0) 1372 n = 1; 1373 } 1374 1375 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1376 1377 /* convert from multibyte to wide char */ 1378 (void)mbtowc(NULL, NULL, 0); 1379 for (i = 0, wi = 0; i < len; i += n, wi++) { 1380 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1381 break; 1382 if (n == 0) 1383 n = 1; 1384 } 1385 ws[wi] = 0; 1386 free(s); 1387 1388 strg = xcalloc(1, sizeof (strg_t)); 1389 strg->st_tspec = WCHAR; 1390 strg->st_len = wlen; 1391 strg->st_wcp = ws; 1392 1393 yylval.y_strg = strg; 1394 return (T_STRING); 1395 } 1396 1397 /* 1398 * As noted above the scanner does not create new symbol table entries 1399 * for symbols it cannot find in the symbol table. This is to avoid 1400 * putting undeclared symbols into the symbol table if a syntax error 1401 * occurs. 1402 * 1403 * getsym() is called as soon as it is probably ok to put the symbol to 1404 * the symbol table. This does not mean that it is not possible that 1405 * symbols are put to the symbol table which are than not completely 1406 * declared due to syntax errors. To avoid too many problems in this 1407 * case symbols get type int in getsym(). 1408 * 1409 * XXX calls to getsym() should be delayed until decl1*() is called 1410 */ 1411 sym_t * 1412 getsym(sbuf_t *sb) 1413 { 1414 dinfo_t *di; 1415 char *s; 1416 sym_t *sym; 1417 1418 sym = sb->sb_sym; 1419 1420 /* 1421 * During member declaration it is possible that name() looked 1422 * for symbols of type FVFT, although it should have looked for 1423 * symbols of type FTAG. Same can happen for labels. Both cases 1424 * are compensated here. 1425 */ 1426 if (symtyp == FMOS || symtyp == FLAB) { 1427 if (sym == NULL || sym->s_kind == FVFT) 1428 sym = search(sb); 1429 } 1430 1431 if (sym != NULL) { 1432 if (sym->s_kind != symtyp) 1433 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1434 symtyp = FVFT; 1435 freesb(sb); 1436 return (sym); 1437 } 1438 1439 /* create a new symbol table entry */ 1440 1441 /* labels must always be allocated at level 1 (outhermost block) */ 1442 if (symtyp == FLAB) { 1443 sym = getlblk(1, sizeof (sym_t)); 1444 s = getlblk(1, sb->sb_len + 1); 1445 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1446 sym->s_name = s; 1447 sym->s_blklev = 1; 1448 di = dcs; 1449 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1450 di = di->d_nxt; 1451 if (di->d_ctx != AUTO) 1452 LERROR("storesym()"); 1453 } else { 1454 sym = getblk(sizeof (sym_t)); 1455 sym->s_name = sb->sb_name; 1456 sym->s_blklev = blklev; 1457 di = dcs; 1458 } 1459 1460 UNIQUE_CURR_POS(sym->s_dpos); 1461 if ((sym->s_kind = symtyp) != FLAB) 1462 sym->s_type = gettyp(INT); 1463 1464 symtyp = FVFT; 1465 1466 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1467 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1468 sym->s_rlink = &symtab[sb->sb_hash]; 1469 symtab[sb->sb_hash] = sym; 1470 1471 *di->d_ldlsym = sym; 1472 di->d_ldlsym = &sym->s_dlnxt; 1473 1474 freesb(sb); 1475 return (sym); 1476 } 1477 1478 /* 1479 * Construct a temporary symbol. The symbol starts with a digit, so that 1480 * it is illegal. 1481 */ 1482 sym_t * 1483 mktempsym(type_t *t) 1484 { 1485 static int n = 0; 1486 int h; 1487 char *s = getlblk(blklev, 64); 1488 sym_t *sym = getblk(sizeof (sym_t)); 1489 1490 (void)snprintf(s, 64, "%.8d_tmp", n++); 1491 h = hash(s); 1492 1493 sym->s_name = s; 1494 sym->s_type = t; 1495 sym->s_blklev = blklev; 1496 sym->s_scl = AUTO; 1497 sym->s_kind = FVFT; 1498 sym->s_used = 1; 1499 sym->s_set = 1; 1500 1501 if ((sym->s_link = symtab[h]) != NULL) 1502 symtab[h]->s_rlink = &sym->s_link; 1503 sym->s_rlink = &symtab[h]; 1504 symtab[h] = sym; 1505 1506 *dcs->d_ldlsym = sym; 1507 dcs->d_ldlsym = &sym->s_dlnxt; 1508 1509 return sym; 1510 } 1511 1512 /* 1513 * Remove a symbol forever from the symbol table. s_blklev 1514 * is set to -1 to avoid that the symbol will later be put 1515 * back to the symbol table. 1516 */ 1517 void 1518 rmsym(sym_t *sym) 1519 { 1520 1521 if ((*sym->s_rlink = sym->s_link) != NULL) 1522 sym->s_link->s_rlink = sym->s_rlink; 1523 sym->s_blklev = -1; 1524 sym->s_link = NULL; 1525 } 1526 1527 /* 1528 * Remove a list of symbols declared at one level from the symbol 1529 * table. 1530 */ 1531 void 1532 rmsyms(sym_t *syms) 1533 { 1534 sym_t *sym; 1535 1536 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1537 if (sym->s_blklev != -1) { 1538 if ((*sym->s_rlink = sym->s_link) != NULL) 1539 sym->s_link->s_rlink = sym->s_rlink; 1540 sym->s_link = NULL; 1541 sym->s_rlink = NULL; 1542 } 1543 } 1544 } 1545 1546 /* 1547 * Put a symbol into the symbol table 1548 */ 1549 void 1550 inssym(int bl, sym_t *sym) 1551 { 1552 int h; 1553 1554 h = hash(sym->s_name); 1555 if ((sym->s_link = symtab[h]) != NULL) 1556 symtab[h]->s_rlink = &sym->s_link; 1557 sym->s_rlink = &symtab[h]; 1558 symtab[h] = sym; 1559 sym->s_blklev = bl; 1560 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1561 LERROR("inssym()"); 1562 } 1563 1564 /* 1565 * Called at level 0 after syntax errors 1566 * Removes all symbols which are not declared at level 0 from the 1567 * symbol table. Also frees all memory which is not associated with 1568 * level 0. 1569 */ 1570 void 1571 cleanup(void) 1572 { 1573 sym_t *sym, *nsym; 1574 int i; 1575 1576 for (i = 0; i < HSHSIZ1; i++) { 1577 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1578 nsym = sym->s_link; 1579 if (sym->s_blklev >= 1) { 1580 if ((*sym->s_rlink = nsym) != NULL) 1581 nsym->s_rlink = sym->s_rlink; 1582 } 1583 } 1584 } 1585 1586 for (i = mblklev; i > 0; i--) 1587 freelblk(i); 1588 } 1589 1590 /* 1591 * Create a new symbol with the name of an existing symbol. 1592 */ 1593 sym_t * 1594 pushdown(sym_t *sym) 1595 { 1596 int h; 1597 sym_t *nsym; 1598 1599 h = hash(sym->s_name); 1600 nsym = getblk(sizeof (sym_t)); 1601 if (sym->s_blklev > blklev) 1602 LERROR("pushdown()"); 1603 nsym->s_name = sym->s_name; 1604 UNIQUE_CURR_POS(nsym->s_dpos); 1605 nsym->s_kind = sym->s_kind; 1606 nsym->s_blklev = blklev; 1607 1608 if ((nsym->s_link = symtab[h]) != NULL) 1609 symtab[h]->s_rlink = &nsym->s_link; 1610 nsym->s_rlink = &symtab[h]; 1611 symtab[h] = nsym; 1612 1613 *dcs->d_ldlsym = nsym; 1614 dcs->d_ldlsym = &nsym->s_dlnxt; 1615 1616 return (nsym); 1617 } 1618 1619 /* 1620 * Free any dynamically allocated memory referenced by 1621 * the value stack or yylval. 1622 * The type of information in yylval is described by tok. 1623 */ 1624 void 1625 freeyyv(void *sp, int tok) 1626 { 1627 if (tok == T_NAME || tok == T_TYPENAME) { 1628 sbuf_t *sb = *(sbuf_t **)sp; 1629 freesb(sb); 1630 } else if (tok == T_CON) { 1631 val_t *val = *(val_t **)sp; 1632 free(val); 1633 } else if (tok == T_STRING) { 1634 strg_t *strg = *(strg_t **)sp; 1635 if (strg->st_tspec == CHAR) { 1636 free(strg->st_cp); 1637 } else if (strg->st_tspec == WCHAR) { 1638 free(strg->st_wcp); 1639 } else { 1640 LERROR("fryylv()"); 1641 } 1642 free(strg); 1643 } 1644 } 1645
Indexes created Thu Oct 02 10:09:58 GMT 2025