scan.l revision 1.82
1 %{ 2 /* $NetBSD: scan.l,v 1.82 2018/07/08 17:48:42 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 1.82 2018/07/08 17:48:42 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 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 }, 255 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 }, 256 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 }, 257 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 }, 258 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 }, 259 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 260 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 261 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 262 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 263 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 264 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 265 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 266 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 267 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 268 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 269 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 270 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 271 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 272 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 273 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 274 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 275 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 }, 276 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 277 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 278 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 279 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 }, 280 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 281 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 282 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 }, 283 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 284 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 285 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 286 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 287 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 288 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 289 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 290 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 291 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 292 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 293 }; 294 295 /* Symbol table */ 296 static sym_t *symtab[HSHSIZ1]; 297 298 /* bit i of the entry with index i is set */ 299 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 300 301 /* least significant i bits are set in the entry with index i */ 302 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 303 304 /* least significant i bits are not set in the entry with index i */ 305 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 306 307 /* free list for sbuf structures */ 308 static sbuf_t *sbfrlst; 309 310 /* Typ of next expected symbol */ 311 symt_t symtyp; 312 313 314 static void 315 addkw(struct kwtab *kw, int deco) 316 { 317 sym_t *sym; 318 size_t h; 319 char buf[256]; 320 const char *name; 321 322 if (!(kw->kw_deco & deco)) 323 return; 324 325 switch (deco) { 326 case 1: 327 name = kw->kw_name; 328 break; 329 case 2: 330 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 331 name = strdup(buf); 332 break; 333 case 4: 334 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 335 name = strdup(buf); 336 break; 337 default: 338 abort(); 339 } 340 341 if (name == NULL) 342 err(1, "Can't init symbol table"); 343 344 sym = getblk(sizeof (sym_t)); 345 sym->s_name = name; 346 sym->s_keyw = kw; 347 sym->s_value.v_quad = kw->kw_token; 348 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 349 sym->s_tspec = kw->kw_tspec; 350 } else if (kw->kw_token == T_SCLASS) { 351 sym->s_scl = kw->kw_scl; 352 } else if (kw->kw_token == T_QUAL) { 353 sym->s_tqual = kw->kw_tqual; 354 } 355 h = hash(sym->s_name); 356 if ((sym->s_link = symtab[h]) != NULL) 357 symtab[h]->s_rlink = &sym->s_link; 358 sym->s_rlink = &symtab[h]; 359 symtab[h] = sym; 360 } 361 362 /* 363 * All keywords are written to the symbol table. This saves us looking 364 * in a extra table for each name we found. 365 */ 366 void 367 initscan(void) 368 { 369 struct kwtab *kw; 370 size_t i; 371 uint64_t uq; 372 373 for (kw = kwtab; kw->kw_name != NULL; kw++) { 374 if ((kw->kw_c89 || kw->kw_c99) && tflag) 375 continue; 376 if (kw->kw_c99 && !(Sflag || gflag)) 377 continue; 378 if (kw->kw_gcc && !gflag) 379 continue; 380 addkw(kw, 1); 381 addkw(kw, 2); 382 addkw(kw, 4); 383 } 384 385 /* initialize bit-masks for quads */ 386 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 387 qbmasks[i] = (uint64_t)1 << i; 388 uq = ~(uint64_t)0 << i; 389 qumasks[i] = uq; 390 qlmasks[i] = ~uq; 391 } 392 qumasks[i] = 0; 393 qlmasks[i] = ~(uint64_t)0; 394 } 395 396 /* 397 * Get a free sbuf structure, if possible from the free list 398 */ 399 static sbuf_t * 400 allocsb(void) 401 { 402 sbuf_t *sb; 403 404 if ((sb = sbfrlst) != NULL) { 405 sbfrlst = sb->sb_nxt; 406 #ifdef BLKDEBUG 407 (void)memset(sb, 0, sizeof (*sb)); 408 #else 409 sb->sb_nxt = NULL; 410 #endif 411 } else { 412 sb = xmalloc(sizeof (sbuf_t)); 413 (void)memset(sb, 0, sizeof (*sb)); 414 } 415 return (sb); 416 } 417 418 /* 419 * Put a sbuf structure to the free list 420 */ 421 static void 422 freesb(sbuf_t *sb) 423 { 424 425 (void)memset(sb, ZERO, sizeof (*sb)); 426 sb->sb_nxt = sbfrlst; 427 sbfrlst = sb; 428 } 429 430 /* 431 * Read a character and ensure that it is positive (except EOF). 432 * Increment line count(s) if necessary. 433 */ 434 static int 435 inpc(void) 436 { 437 int c; 438 439 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 440 incline(); 441 return (c); 442 } 443 444 static int 445 hash(const char *s) 446 { 447 u_int v; 448 const u_char *us; 449 450 v = 0; 451 for (us = (const u_char *)s; *us != '\0'; us++) { 452 v = (v << sizeof (v)) + *us; 453 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 454 } 455 return (v % HSHSIZ1); 456 } 457 458 /* 459 * Lex has found a letter followed by zero or more letters or digits. 460 * It looks for a symbol in the symbol table with the same name. This 461 * symbol must either be a keyword or a symbol of the type required by 462 * symtyp (label, member, tag, ...). 463 * 464 * If it is a keyword, the token is returned. In some cases it is described 465 * more deeply by data written to yylval. 466 * 467 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 468 * is stored in yylval. This struct contains the name of the symbol, its 469 * length and hash value. If there is already a symbol of the same name 470 * and type in the symbol table, the sbuf struct also contains a pointer 471 * to the symbol table entry. 472 */ 473 static int 474 name(void) 475 { 476 char *s; 477 sbuf_t *sb; 478 sym_t *sym; 479 int tok; 480 481 sb = allocsb(); 482 sb->sb_name = yytext; 483 sb->sb_len = yyleng; 484 sb->sb_hash = hash(yytext); 485 if ((sym = search(sb)) != NULL && sym->s_keyw) { 486 freesb(sb); 487 return (keyw(sym)); 488 } 489 490 sb->sb_sym = sym; 491 492 if (sym != NULL) { 493 if (blklev < sym->s_blklev) 494 LERROR("name()"); 495 sb->sb_name = sym->s_name; 496 sb->sb_len = strlen(sym->s_name); 497 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 498 } else { 499 s = getblk(yyleng + 1); 500 (void)memcpy(s, yytext, yyleng + 1); 501 sb->sb_name = s; 502 sb->sb_len = yyleng; 503 tok = T_NAME; 504 } 505 506 yylval.y_sb = sb; 507 return (tok); 508 } 509 510 static sym_t * 511 search(sbuf_t *sb) 512 { 513 sym_t *sym; 514 515 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 516 if (strcmp(sym->s_name, sb->sb_name) == 0) { 517 if (sym->s_keyw) { 518 struct kwtab *kw = sym->s_keyw; 519 if (!kw->kw_attr || attron) 520 return (sym); 521 } else if (!attron && sym->s_kind == symtyp) 522 return sym; 523 } 524 } 525 526 return (NULL); 527 } 528 529 static int 530 keyw(sym_t *sym) 531 { 532 int t; 533 534 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 535 yylval.y_scl = sym->s_scl; 536 } else if (t == T_TYPE || t == T_SOU) { 537 yylval.y_tspec = sym->s_tspec; 538 } else if (t == T_QUAL) { 539 yylval.y_tqual = sym->s_tqual; 540 } 541 return (t); 542 } 543 544 /* 545 * Convert a string representing an integer into internal representation. 546 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 547 */ 548 static int 549 icon(int base) 550 { 551 int l_suffix, u_suffix; 552 int len; 553 const char *cp; 554 char c, *eptr; 555 tspec_t typ; 556 uint64_t uq = 0; 557 int ansiu; 558 static tspec_t contypes[2][3] = { 559 { INT, LONG, QUAD }, 560 { UINT, ULONG, UQUAD } 561 }; 562 563 cp = yytext; 564 len = yyleng; 565 566 /* skip 0[xX] or 0[bB] */ 567 if (base == 16 || base == 2) { 568 cp += 2; 569 len -= 2; 570 } 571 572 /* read suffixes */ 573 l_suffix = u_suffix = 0; 574 for ( ; ; ) { 575 if ((c = cp[len - 1]) == 'l' || c == 'L') { 576 l_suffix++; 577 } else if (c == 'u' || c == 'U') { 578 u_suffix++; 579 } else { 580 break; 581 } 582 len--; 583 } 584 if (l_suffix > 2 || u_suffix > 1) { 585 /* malformed integer constant */ 586 warning(251); 587 if (l_suffix > 2) 588 l_suffix = 2; 589 if (u_suffix > 1) 590 u_suffix = 1; 591 } 592 if (tflag && u_suffix != 0) { 593 /* suffix U is illegal in traditional C */ 594 warning(97); 595 } 596 typ = contypes[u_suffix][l_suffix]; 597 598 errno = 0; 599 600 uq = strtouq(cp, &eptr, base); 601 if (eptr != cp + len) 602 LERROR("icon()"); 603 if (errno != 0) 604 /* integer constant out of range */ 605 warning(252); 606 607 /* 608 * If the value is too big for the current type, we must choose 609 * another type. 610 */ 611 ansiu = 0; 612 switch (typ) { 613 case INT: 614 if (uq <= TARG_INT_MAX) { 615 /* ok */ 616 } else if (uq <= TARG_UINT_MAX && base != 10) { 617 typ = UINT; 618 } else if (uq <= TARG_LONG_MAX) { 619 typ = LONG; 620 } else { 621 typ = ULONG; 622 if (uq > TARG_ULONG_MAX) { 623 /* integer constant out of range */ 624 warning(252); 625 } 626 } 627 if (typ == UINT || typ == ULONG) { 628 if (tflag) { 629 typ = LONG; 630 } else if (!sflag) { 631 /* 632 * Remember that the constant is unsigned 633 * only in ANSI C 634 */ 635 ansiu = 1; 636 } 637 } 638 break; 639 case UINT: 640 if (uq > TARG_UINT_MAX) { 641 typ = ULONG; 642 if (uq > TARG_ULONG_MAX) { 643 /* integer constant out of range */ 644 warning(252); 645 } 646 } 647 break; 648 case LONG: 649 if (uq > TARG_LONG_MAX && !tflag) { 650 typ = ULONG; 651 if (!sflag) 652 ansiu = 1; 653 if (uq > TARG_ULONG_MAX) { 654 /* integer constant out of range */ 655 warning(252); 656 } 657 } 658 break; 659 case ULONG: 660 if (uq > TARG_ULONG_MAX) { 661 /* integer constant out of range */ 662 warning(252); 663 } 664 break; 665 case QUAD: 666 if (uq > TARG_QUAD_MAX && !tflag) { 667 typ = UQUAD; 668 if (!sflag) 669 ansiu = 1; 670 } 671 break; 672 case UQUAD: 673 if (uq > TARG_UQUAD_MAX) { 674 /* integer constant out of range */ 675 warning(252); 676 } 677 break; 678 /* LINTED206: (enumeration values not handled in switch) */ 679 case STRUCT: 680 case VOID: 681 case LDOUBLE: 682 case FUNC: 683 case ARRAY: 684 case PTR: 685 case ENUM: 686 case UNION: 687 case SIGNED: 688 case NOTSPEC: 689 case DOUBLE: 690 case FLOAT: 691 case USHORT: 692 case SHORT: 693 case UCHAR: 694 case SCHAR: 695 case CHAR: 696 case BOOL: 697 case UNSIGN: 698 case FCOMPLEX: 699 case DCOMPLEX: 700 case LCOMPLEX: 701 case COMPLEX: 702 break; 703 704 case NTSPEC: /* this value unused */ 705 break; 706 } 707 708 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 709 710 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 711 yylval.y_val->v_ansiu = ansiu; 712 yylval.y_val->v_quad = (int64_t)uq; 713 714 return (T_CON); 715 } 716 717 /* 718 * Returns 1 if t is a signed type and the value is negative. 719 * 720 * len is the number of significant bits. If len is -1, len is set 721 * to the width of type t. 722 */ 723 int 724 sign(int64_t q, tspec_t t, int len) 725 { 726 727 if (t == PTR || isutyp(t)) 728 return (0); 729 return (msb(q, t, len)); 730 } 731 732 int 733 msb(int64_t q, tspec_t t, int len) 734 { 735 736 if (len <= 0) 737 len = size(t); 738 return ((q & qbmasks[len - 1]) != 0); 739 } 740 741 /* 742 * Extends the sign of q. 743 */ 744 int64_t 745 xsign(int64_t q, tspec_t t, int len) 746 { 747 748 if (len <= 0) 749 len = size(t); 750 751 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 752 q &= qlmasks[len]; 753 } else { 754 q |= qumasks[len]; 755 } 756 return (q); 757 } 758 759 /* 760 * Convert a string representing a floating point value into its interal 761 * representation. Type and value are returned in yylval. fcon() 762 * (and yylex()) returns T_CON. 763 * XXX Currently it is not possible to convert constants of type 764 * long double which are greater than DBL_MAX. 765 */ 766 static int 767 fcon(void) 768 { 769 const char *cp; 770 int len; 771 tspec_t typ; 772 char c, *eptr; 773 double d; 774 float f = 0; 775 776 cp = yytext; 777 len = yyleng; 778 779 if (cp[len - 1] == 'i') { 780 /* imaginary, do nothing for now */ 781 len--; 782 } 783 if ((c = cp[len - 1]) == 'f' || c == 'F') { 784 typ = FLOAT; 785 len--; 786 } else if (c == 'l' || c == 'L') { 787 typ = LDOUBLE; 788 len--; 789 } else { 790 if (c == 'd' || c == 'D') 791 len--; 792 typ = DOUBLE; 793 } 794 795 if (tflag && typ != DOUBLE) { 796 /* suffixes F and L are illegal in traditional C */ 797 warning(98); 798 } 799 800 errno = 0; 801 d = strtod(cp, &eptr); 802 if (eptr != cp + len) { 803 switch (*eptr) { 804 /* 805 * XXX: non-native non-current strtod() may not handle hex 806 * floats, ignore the rest if we find traces of hex float 807 * syntax... 808 */ 809 case 'p': 810 case 'P': 811 case 'x': 812 case 'X': 813 d = 0; 814 errno = 0; 815 break; 816 default: 817 LERROR("fcon(%s->%s)", cp, eptr); 818 } 819 } 820 if (errno != 0) 821 /* floating-point constant out of range */ 822 warning(248); 823 824 if (typ == FLOAT) { 825 f = (float)d; 826 if (!finite(f)) { 827 /* floating-point constant out of range */ 828 warning(248); 829 f = f > 0 ? FLT_MAX : -FLT_MAX; 830 } 831 } 832 833 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 834 if (typ == FLOAT) { 835 yylval.y_val->v_ldbl = f; 836 } else { 837 yylval.y_val->v_ldbl = d; 838 } 839 840 return (T_CON); 841 } 842 843 static int 844 operator(int t, op_t o) 845 { 846 847 yylval.y_op = o; 848 return (t); 849 } 850 851 /* 852 * Called if lex found a leading \'. 853 */ 854 static int 855 ccon(void) 856 { 857 size_t n; 858 int val, c; 859 char cv; 860 861 n = 0; 862 val = 0; 863 while ((c = getescc('\'')) >= 0) { 864 val = (val << CHAR_BIT) + c; 865 n++; 866 } 867 if (c == -2) { 868 /* unterminated character constant */ 869 error(253); 870 } else { 871 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 872 /* too many characters in character constant */ 873 error(71); 874 } else if (n > 1) { 875 /* multi-character character constant */ 876 warning(294); 877 } else if (n == 0) { 878 /* empty character constant */ 879 error(73); 880 } 881 } 882 if (n == 1) { 883 cv = (char)val; 884 val = cv; 885 } 886 887 yylval.y_val = xcalloc(1, sizeof (val_t)); 888 yylval.y_val->v_tspec = INT; 889 yylval.y_val->v_quad = val; 890 891 return (T_CON); 892 } 893 894 /* 895 * Called if lex found a leading L\' 896 */ 897 static int 898 wccon(void) 899 { 900 static char buf[MB_LEN_MAX + 1]; 901 size_t i; 902 int c; 903 wchar_t wc; 904 905 i = 0; 906 while ((c = getescc('\'')) >= 0) { 907 if (i < MB_CUR_MAX) 908 buf[i] = (char)c; 909 i++; 910 } 911 912 wc = 0; 913 914 if (c == -2) { 915 /* unterminated character constant */ 916 error(253); 917 } else if (c == 0) { 918 /* empty character constant */ 919 error(73); 920 } else { 921 if (i > MB_CUR_MAX) { 922 i = MB_CUR_MAX; 923 /* too many characters in character constant */ 924 error(71); 925 } else { 926 buf[i] = '\0'; 927 (void)mbtowc(NULL, NULL, 0); 928 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 929 /* invalid multibyte character */ 930 error(291); 931 } 932 } 933 934 yylval.y_val = xcalloc(1, sizeof (val_t)); 935 yylval.y_val->v_tspec = WCHAR; 936 yylval.y_val->v_quad = wc; 937 938 return (T_CON); 939 } 940 941 /* 942 * Read a character which is part of a character constant or of a string 943 * and handle escapes. 944 * 945 * The Argument is the character which delimits the character constant or 946 * string. 947 * 948 * Returns -1 if the end of the character constant or string is reached, 949 * -2 if the EOF is reached, and the character otherwise. 950 */ 951 static int 952 getescc(int d) 953 { 954 static int pbc = -1; 955 int n, c, v; 956 957 if (pbc == -1) { 958 c = inpc(); 959 } else { 960 c = pbc; 961 pbc = -1; 962 } 963 if (c == d) 964 return (-1); 965 switch (c) { 966 case '\n': 967 if (tflag) { 968 /* newline in string or char constant */ 969 error(254); 970 return (-2); 971 } 972 return (c); 973 case EOF: 974 return (-2); 975 case '\\': 976 switch (c = inpc()) { 977 case '"': 978 if (tflag && d == '\'') 979 /* \" inside character constant undef. ... */ 980 warning(262); 981 return ('"'); 982 case '\'': 983 return ('\''); 984 case '?': 985 if (tflag) 986 /* \? undefined in traditional C */ 987 warning(263); 988 return ('?'); 989 case '\\': 990 return ('\\'); 991 case 'a': 992 if (tflag) 993 /* \a undefined in traditional C */ 994 warning(81); 995 return ('\a'); 996 case 'b': 997 return ('\b'); 998 case 'f': 999 return ('\f'); 1000 case 'n': 1001 return ('\n'); 1002 case 'r': 1003 return ('\r'); 1004 case 't': 1005 return ('\t'); 1006 case 'v': 1007 if (tflag) 1008 /* \v undefined in traditional C */ 1009 warning(264); 1010 return ('\v'); 1011 case '8': case '9': 1012 /* bad octal digit %c */ 1013 warning(77, c); 1014 /* FALLTHROUGH */ 1015 case '0': case '1': case '2': case '3': 1016 case '4': case '5': case '6': case '7': 1017 n = 3; 1018 v = 0; 1019 do { 1020 v = (v << 3) + (c - '0'); 1021 c = inpc(); 1022 } while (--n && isdigit(c) && (tflag || c <= '7')); 1023 if (tflag && n > 0 && isdigit(c)) 1024 /* bad octal digit %c */ 1025 warning(77, c); 1026 pbc = c; 1027 if (v > TARG_UCHAR_MAX) { 1028 /* character escape does not fit in char. */ 1029 warning(76); 1030 v &= CHAR_MASK; 1031 } 1032 return (v); 1033 case 'x': 1034 if (tflag) 1035 /* \x undefined in traditional C */ 1036 warning(82); 1037 v = 0; 1038 n = 0; 1039 while ((c = inpc()) >= 0 && isxdigit(c)) { 1040 c = isdigit(c) ? 1041 c - '0' : toupper(c) - 'A' + 10; 1042 v = (v << 4) + c; 1043 if (n >= 0) { 1044 if ((v & ~CHAR_MASK) != 0) { 1045 /* overflow in hex escape */ 1046 warning(75); 1047 n = -1; 1048 } else { 1049 n++; 1050 } 1051 } 1052 } 1053 pbc = c; 1054 if (n == 0) { 1055 /* no hex digits follow \x */ 1056 error(74); 1057 } if (n == -1) { 1058 v &= CHAR_MASK; 1059 } 1060 return (v); 1061 case '\n': 1062 return (getescc(d)); 1063 case EOF: 1064 return (-2); 1065 default: 1066 if (isprint(c)) { 1067 /* dubious escape \%c */ 1068 warning(79, c); 1069 } else { 1070 /* dubious escape \%o */ 1071 warning(80, c); 1072 } 1073 } 1074 } 1075 return (c); 1076 } 1077 1078 /* 1079 * Called for preprocessor directives. Currently implemented are: 1080 * # lineno 1081 * # lineno "filename" 1082 */ 1083 static void 1084 directive(void) 1085 { 1086 const char *cp, *fn; 1087 char c, *eptr; 1088 size_t fnl; 1089 long ln; 1090 static int first = 1; 1091 1092 /* Go to first non-whitespace after # */ 1093 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1094 continue; 1095 1096 if (!isdigit((unsigned char)c)) { 1097 if (strncmp(cp, "pragma", 6) == 0 1098 && isspace((unsigned char)cp[6])) 1099 return; 1100 error: 1101 /* undefined or invalid # directive */ 1102 warning(255); 1103 return; 1104 } 1105 ln = strtol(--cp, &eptr, 10); 1106 if (cp == eptr) 1107 goto error; 1108 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1109 goto error; 1110 while ((c = *cp++) == ' ' || c == '\t') 1111 continue; 1112 if (c != '\0') { 1113 if (c != '"') 1114 goto error; 1115 fn = cp; 1116 while ((c = *cp) != '"' && c != '\0') 1117 cp++; 1118 if (c != '"') 1119 goto error; 1120 if ((fnl = cp++ - fn) > PATH_MAX) 1121 goto error; 1122 while ((c = *cp++) == ' ' || c == '\t') 1123 continue; 1124 #if 0 1125 if (c != '\0') 1126 warning("extra character(s) after directive"); 1127 #endif 1128 1129 /* empty string means stdin */ 1130 if (fnl == 0) { 1131 fn = "{standard input}"; 1132 fnl = 16; /* strlen (fn) */ 1133 } 1134 curr_pos.p_file = fnnalloc(fn, fnl); 1135 /* 1136 * If this is the first directive, the name is the name 1137 * of the C source file as specified at the command line. 1138 * It is written to the output file. 1139 */ 1140 if (first) { 1141 csrc_pos.p_file = curr_pos.p_file; 1142 outsrc(fnxform(curr_pos.p_file, 1143 strlen(curr_pos.p_file))); 1144 first = 0; 1145 } 1146 } 1147 curr_pos.p_line = (int)ln - 1; 1148 curr_pos.p_uniq = 0; 1149 if (curr_pos.p_file == csrc_pos.p_file) { 1150 csrc_pos.p_line = (int)ln - 1; 1151 csrc_pos.p_uniq = 0; 1152 } 1153 } 1154 1155 /* 1156 * Handle lint comments. Following comments are currently understood: 1157 * ARGSUSEDn 1158 * BITFIELDTYPE 1159 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1160 * FALLTHRU FALLTHROUGH 1161 * LINTLIBRARY 1162 * LINTEDn NOSTRICTn 1163 * LONGLONG 1164 * NOTREACHED 1165 * PRINTFLIKEn 1166 * PROTOLIB 1167 * SCANFLIKEn 1168 * VARARGSn 1169 * If one of this comments is recognized, the arguments, if any, are 1170 * parsed and a function which handles this comment is called. 1171 */ 1172 static void 1173 comment(void) 1174 { 1175 int c, lc; 1176 static struct { 1177 const char *keywd; 1178 int arg; 1179 void (*func)(int); 1180 } keywtab[] = { 1181 { "ARGSUSED", 1, argsused }, 1182 { "BITFIELDTYPE", 0, bitfieldtype }, 1183 { "CONSTCOND", 0, constcond }, 1184 { "CONSTANTCOND", 0, constcond }, 1185 { "CONSTANTCONDITION", 0, constcond }, 1186 { "FALLTHRU", 0, fallthru }, 1187 { "FALLTHROUGH", 0, fallthru }, 1188 { "LINTLIBRARY", 0, lintlib }, 1189 { "LINTED", 1, linted }, 1190 { "LONGLONG", 0, longlong }, 1191 { "NOSTRICT", 1, linted }, 1192 { "NOTREACHED", 0, notreach }, 1193 { "PRINTFLIKE", 1, printflike }, 1194 { "PROTOLIB", 1, protolib }, 1195 { "SCANFLIKE", 1, scanflike }, 1196 { "VARARGS", 1, varargs }, 1197 }; 1198 char keywd[32]; 1199 char arg[32]; 1200 size_t l, i; 1201 int a; 1202 int eoc; 1203 1204 eoc = 0; 1205 1206 /* Skip white spaces after the start of the comment */ 1207 while ((c = inpc()) != EOF && isspace(c)) 1208 continue; 1209 1210 /* Read the potential keyword to keywd */ 1211 l = 0; 1212 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1213 keywd[l++] = (char)c; 1214 c = inpc(); 1215 } 1216 keywd[l] = '\0'; 1217 1218 /* look for the keyword */ 1219 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1220 if (strcmp(keywtab[i].keywd, keywd) == 0) 1221 break; 1222 } 1223 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1224 goto skip_rest; 1225 1226 /* skip white spaces after the keyword */ 1227 while (c != EOF && isspace(c)) 1228 c = inpc(); 1229 1230 /* read the argument, if the keyword accepts one and there is one */ 1231 l = 0; 1232 if (keywtab[i].arg) { 1233 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1234 arg[l++] = (char)c; 1235 c = inpc(); 1236 } 1237 } 1238 arg[l] = '\0'; 1239 a = l != 0 ? atoi(arg) : -1; 1240 1241 /* skip white spaces after the argument */ 1242 while (c != EOF && isspace(c)) 1243 c = inpc(); 1244 1245 if (c != '*' || (c = inpc()) != '/') { 1246 if (keywtab[i].func != linted) 1247 /* extra characters in lint comment */ 1248 warning(257); 1249 } else { 1250 /* 1251 * remember that we have already found the end of the 1252 * comment 1253 */ 1254 eoc = 1; 1255 } 1256 1257 if (keywtab[i].func != NULL) 1258 (*keywtab[i].func)(a); 1259 1260 skip_rest: 1261 while (!eoc) { 1262 lc = c; 1263 if ((c = inpc()) == EOF) { 1264 /* unterminated comment */ 1265 error(256); 1266 break; 1267 } 1268 if (lc == '*' && c == '/') 1269 eoc = 1; 1270 } 1271 } 1272 1273 /* 1274 * Handle // style comments 1275 */ 1276 static void 1277 slashslashcomment(void) 1278 { 1279 int c; 1280 1281 if (!Sflag && !gflag) 1282 /* // comments only supported in C99 */ 1283 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1284 1285 while ((c = inpc()) != EOF && c != '\n') 1286 continue; 1287 } 1288 1289 /* 1290 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1291 * clrwflgs() is called after function definitions and global and 1292 * local declarations and definitions. It is also called between 1293 * the controlling expression and the body of control statements 1294 * (if, switch, for, while). 1295 */ 1296 void 1297 clrwflgs(void) 1298 { 1299 1300 lwarn = LWARN_ALL; 1301 quadflg = 0; 1302 ccflg = 0; 1303 } 1304 1305 /* 1306 * Strings are stored in a dynamically alloceted buffer and passed 1307 * in yylval.y_xstrg to the parser. The parser or the routines called 1308 * by the parser are responsible for freeing this buffer. 1309 */ 1310 static int 1311 string(void) 1312 { 1313 u_char *s; 1314 int c; 1315 size_t len, max; 1316 strg_t *strg; 1317 1318 s = xmalloc(max = 64); 1319 1320 len = 0; 1321 while ((c = getescc('"')) >= 0) { 1322 /* +1 to reserve space for a trailing NUL character */ 1323 if (len + 1 == max) 1324 s = xrealloc(s, max *= 2); 1325 s[len++] = (char)c; 1326 } 1327 s[len] = '\0'; 1328 if (c == -2) 1329 /* unterminated string constant */ 1330 error(258); 1331 1332 strg = xcalloc(1, sizeof (strg_t)); 1333 strg->st_tspec = CHAR; 1334 strg->st_len = len; 1335 strg->st_cp = s; 1336 1337 yylval.y_strg = strg; 1338 return (T_STRING); 1339 } 1340 1341 static int 1342 wcstrg(void) 1343 { 1344 char *s; 1345 int c, n; 1346 size_t i, wi; 1347 size_t len, max, wlen; 1348 wchar_t *ws; 1349 strg_t *strg; 1350 1351 s = xmalloc(max = 64); 1352 len = 0; 1353 while ((c = getescc('"')) >= 0) { 1354 /* +1 to save space for a trailing NUL character */ 1355 if (len + 1 >= max) 1356 s = xrealloc(s, max *= 2); 1357 s[len++] = (char)c; 1358 } 1359 s[len] = '\0'; 1360 if (c == -2) 1361 /* unterminated string constant */ 1362 error(258); 1363 1364 /* get length of wide-character string */ 1365 (void)mblen(NULL, 0); 1366 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1367 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1368 /* invalid multibyte character */ 1369 error(291); 1370 break; 1371 } 1372 if (n == 0) 1373 n = 1; 1374 } 1375 1376 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1377 1378 /* convert from multibyte to wide char */ 1379 (void)mbtowc(NULL, NULL, 0); 1380 for (i = 0, wi = 0; i < len; i += n, wi++) { 1381 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1382 break; 1383 if (n == 0) 1384 n = 1; 1385 } 1386 ws[wi] = 0; 1387 free(s); 1388 1389 strg = xcalloc(1, sizeof (strg_t)); 1390 strg->st_tspec = WCHAR; 1391 strg->st_len = wlen; 1392 strg->st_wcp = ws; 1393 1394 yylval.y_strg = strg; 1395 return (T_STRING); 1396 } 1397 1398 /* 1399 * As noted above the scanner does not create new symbol table entries 1400 * for symbols it cannot find in the symbol table. This is to avoid 1401 * putting undeclared symbols into the symbol table if a syntax error 1402 * occurs. 1403 * 1404 * getsym() is called as soon as it is probably ok to put the symbol to 1405 * the symbol table. This does not mean that it is not possible that 1406 * symbols are put to the symbol table which are than not completely 1407 * declared due to syntax errors. To avoid too many problems in this 1408 * case symbols get type int in getsym(). 1409 * 1410 * XXX calls to getsym() should be delayed until decl1*() is called 1411 */ 1412 sym_t * 1413 getsym(sbuf_t *sb) 1414 { 1415 dinfo_t *di; 1416 char *s; 1417 sym_t *sym; 1418 1419 sym = sb->sb_sym; 1420 1421 /* 1422 * During member declaration it is possible that name() looked 1423 * for symbols of type FVFT, although it should have looked for 1424 * symbols of type FTAG. Same can happen for labels. Both cases 1425 * are compensated here. 1426 */ 1427 if (symtyp == FMOS || symtyp == FLAB) { 1428 if (sym == NULL || sym->s_kind == FVFT) 1429 sym = search(sb); 1430 } 1431 1432 if (sym != NULL) { 1433 if (sym->s_kind != symtyp) 1434 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1435 symtyp = FVFT; 1436 freesb(sb); 1437 return (sym); 1438 } 1439 1440 /* create a new symbol table entry */ 1441 1442 /* labels must always be allocated at level 1 (outhermost block) */ 1443 if (symtyp == FLAB) { 1444 sym = getlblk(1, sizeof (sym_t)); 1445 s = getlblk(1, sb->sb_len + 1); 1446 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1447 sym->s_name = s; 1448 sym->s_blklev = 1; 1449 di = dcs; 1450 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1451 di = di->d_nxt; 1452 if (di->d_ctx != AUTO) 1453 LERROR("storesym()"); 1454 } else { 1455 sym = getblk(sizeof (sym_t)); 1456 sym->s_name = sb->sb_name; 1457 sym->s_blklev = blklev; 1458 di = dcs; 1459 } 1460 1461 UNIQUE_CURR_POS(sym->s_dpos); 1462 if ((sym->s_kind = symtyp) != FLAB) 1463 sym->s_type = gettyp(INT); 1464 1465 symtyp = FVFT; 1466 1467 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1468 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1469 sym->s_rlink = &symtab[sb->sb_hash]; 1470 symtab[sb->sb_hash] = sym; 1471 1472 *di->d_ldlsym = sym; 1473 di->d_ldlsym = &sym->s_dlnxt; 1474 1475 freesb(sb); 1476 return (sym); 1477 } 1478 1479 /* 1480 * Construct a temporary symbol. The symbol starts with a digit, so that 1481 * it is illegal. 1482 */ 1483 sym_t * 1484 mktempsym(type_t *t) 1485 { 1486 static int n = 0; 1487 int h; 1488 char *s = getlblk(blklev, 64); 1489 sym_t *sym = getblk(sizeof (sym_t)); 1490 1491 (void)snprintf(s, 64, "%.8d_tmp", n++); 1492 h = hash(s); 1493 1494 sym->s_name = s; 1495 sym->s_type = t; 1496 sym->s_blklev = blklev; 1497 sym->s_scl = AUTO; 1498 sym->s_kind = FVFT; 1499 sym->s_used = 1; 1500 sym->s_set = 1; 1501 1502 if ((sym->s_link = symtab[h]) != NULL) 1503 symtab[h]->s_rlink = &sym->s_link; 1504 sym->s_rlink = &symtab[h]; 1505 symtab[h] = sym; 1506 1507 *dcs->d_ldlsym = sym; 1508 dcs->d_ldlsym = &sym->s_dlnxt; 1509 1510 return sym; 1511 } 1512 1513 /* 1514 * Remove a symbol forever from the symbol table. s_blklev 1515 * is set to -1 to avoid that the symbol will later be put 1516 * back to the symbol table. 1517 */ 1518 void 1519 rmsym(sym_t *sym) 1520 { 1521 1522 if ((*sym->s_rlink = sym->s_link) != NULL) 1523 sym->s_link->s_rlink = sym->s_rlink; 1524 sym->s_blklev = -1; 1525 sym->s_link = NULL; 1526 } 1527 1528 /* 1529 * Remove a list of symbols declared at one level from the symbol 1530 * table. 1531 */ 1532 void 1533 rmsyms(sym_t *syms) 1534 { 1535 sym_t *sym; 1536 1537 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1538 if (sym->s_blklev != -1) { 1539 if ((*sym->s_rlink = sym->s_link) != NULL) 1540 sym->s_link->s_rlink = sym->s_rlink; 1541 sym->s_link = NULL; 1542 sym->s_rlink = NULL; 1543 } 1544 } 1545 } 1546 1547 /* 1548 * Put a symbol into the symbol table 1549 */ 1550 void 1551 inssym(int bl, sym_t *sym) 1552 { 1553 int h; 1554 1555 h = hash(sym->s_name); 1556 if ((sym->s_link = symtab[h]) != NULL) 1557 symtab[h]->s_rlink = &sym->s_link; 1558 sym->s_rlink = &symtab[h]; 1559 symtab[h] = sym; 1560 sym->s_blklev = bl; 1561 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1562 LERROR("inssym()"); 1563 } 1564 1565 /* 1566 * Called at level 0 after syntax errors 1567 * Removes all symbols which are not declared at level 0 from the 1568 * symbol table. Also frees all memory which is not associated with 1569 * level 0. 1570 */ 1571 void 1572 cleanup(void) 1573 { 1574 sym_t *sym, *nsym; 1575 int i; 1576 1577 for (i = 0; i < HSHSIZ1; i++) { 1578 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1579 nsym = sym->s_link; 1580 if (sym->s_blklev >= 1) { 1581 if ((*sym->s_rlink = nsym) != NULL) 1582 nsym->s_rlink = sym->s_rlink; 1583 } 1584 } 1585 } 1586 1587 for (i = mblklev; i > 0; i--) 1588 freelblk(i); 1589 } 1590 1591 /* 1592 * Create a new symbol with the name of an existing symbol. 1593 */ 1594 sym_t * 1595 pushdown(sym_t *sym) 1596 { 1597 int h; 1598 sym_t *nsym; 1599 1600 h = hash(sym->s_name); 1601 nsym = getblk(sizeof (sym_t)); 1602 if (sym->s_blklev > blklev) 1603 LERROR("pushdown()"); 1604 nsym->s_name = sym->s_name; 1605 UNIQUE_CURR_POS(nsym->s_dpos); 1606 nsym->s_kind = sym->s_kind; 1607 nsym->s_blklev = blklev; 1608 1609 if ((nsym->s_link = symtab[h]) != NULL) 1610 symtab[h]->s_rlink = &nsym->s_link; 1611 nsym->s_rlink = &symtab[h]; 1612 symtab[h] = nsym; 1613 1614 *dcs->d_ldlsym = nsym; 1615 dcs->d_ldlsym = &nsym->s_dlnxt; 1616 1617 return (nsym); 1618 } 1619 1620 /* 1621 * Free any dynamically allocated memory referenced by 1622 * the value stack or yylval. 1623 * The type of information in yylval is described by tok. 1624 */ 1625 void 1626 freeyyv(void *sp, int tok) 1627 { 1628 if (tok == T_NAME || tok == T_TYPENAME) { 1629 sbuf_t *sb = *(sbuf_t **)sp; 1630 freesb(sb); 1631 } else if (tok == T_CON) { 1632 val_t *val = *(val_t **)sp; 1633 free(val); 1634 } else if (tok == T_STRING) { 1635 strg_t *strg = *(strg_t **)sp; 1636 if (strg->st_tspec == CHAR) { 1637 free(strg->st_cp); 1638 } else if (strg->st_tspec == WCHAR) { 1639 free(strg->st_wcp); 1640 } else { 1641 LERROR("fryylv()"); 1642 } 1643 free(strg); 1644 } 1645 } 1646
Indexes created Thu Oct 02 14:10:14 GMT 2025