scan.l revision 1.76
1 %{ 2 /* $NetBSD: scan.l,v 1.76 2016/12/30 19:53:08 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 1.76 2016/12/30 19:53:08 christos Exp $"); 39 #endif 40 41 #include <stdlib.h> 42 #include <string.h> 43 #include <limits.h> 44 #include <float.h> 45 #include <ctype.h> 46 #include <errno.h> 47 #include <math.h> 48 49 #include "lint1.h" 50 #include "cgram.h" 51 52 #define CHAR_MASK ((int)(~(~0U << CHAR_BIT))) 53 54 /* Current position (its also updated when an included file is parsed) */ 55 pos_t curr_pos = { 1, "", 0 }; 56 57 /* 58 * Current position in C source (not updated when an included file is 59 * parsed). 60 */ 61 pos_t csrc_pos = { 1, "", 0 }; 62 63 /* Are we parsing a gcc attribute? */ 64 int attron; 65 66 static void incline(void); 67 static void badchar(int); 68 static sbuf_t *allocsb(void); 69 static void freesb(sbuf_t *); 70 static int inpc(void); 71 static int hash(const char *); 72 static sym_t *search(sbuf_t *); 73 static int name(void); 74 static int keyw(sym_t *); 75 static int icon(int); 76 static int fcon(void); 77 static int operator(int, op_t); 78 static int ccon(void); 79 static int wccon(void); 80 static int getescc(int); 81 static void directive(void); 82 static void comment(void); 83 static void slashslashcomment(void); 84 static int string(void); 85 static int wcstrg(void); 86 87 %} 88 89 90 L [_A-Za-z] 91 D [0-9] 92 NZD [1-9] 93 BD [0-1] 94 OD [0-7] 95 HD [0-9A-Fa-f] 96 EX ([eE][+-]?[0-9]+) 97 HX (p[+-]?[0-9A-Fa-f]+) 98 TL ([fFlL]?[i]?) 99 100 %option nounput 101 102 %% 103 104 {L}({L}|{D})* return (name()); 105 0[bB]{BD}+[lLuU]* return (icon(2)); 106 0{OD}*[lLuU]* return (icon(8)); 107 {NZD}{D}*[lLuU]* return (icon(10)); 108 0[xX]{HD}+[lLuU]* return (icon(16)); 109 {D}+\.{D}*{EX}?{TL} | 110 {D}+{EX}{TL} | 111 0[xX]{HD}+\.{HD}*{HX}{TL} | 112 0[xX]{HD}+{HX}{TL} | 113 \.{D}+{EX}?{TL} return (fcon()); 114 "=" return (operator(T_ASSIGN, ASSIGN)); 115 "*=" return (operator(T_OPASS, MULASS)); 116 "/=" return (operator(T_OPASS, DIVASS)); 117 "%=" return (operator(T_OPASS, MODASS)); 118 "+=" return (operator(T_OPASS, ADDASS)); 119 "-=" return (operator(T_OPASS, SUBASS)); 120 "<<=" return (operator(T_OPASS, SHLASS)); 121 ">>=" return (operator(T_OPASS, SHRASS)); 122 "&=" return (operator(T_OPASS, ANDASS)); 123 "^=" return (operator(T_OPASS, XORASS)); 124 "|=" return (operator(T_OPASS, ORASS)); 125 "||" return (operator(T_LOGOR, LOGOR)); 126 "&&" return (operator(T_LOGAND, LOGAND)); 127 "|" return (operator(T_OR, OR)); 128 "&" return (operator(T_AND, AND)); 129 "^" return (operator(T_XOR, XOR)); 130 "==" return (operator(T_EQOP, EQ)); 131 "!=" return (operator(T_EQOP, NE)); 132 "<" return (operator(T_RELOP, LT)); 133 ">" return (operator(T_RELOP, GT)); 134 "<=" return (operator(T_RELOP, LE)); 135 ">=" return (operator(T_RELOP, GE)); 136 "<<" return (operator(T_SHFTOP, SHL)); 137 ">>" return (operator(T_SHFTOP, SHR)); 138 "++" return (operator(T_INCDEC, INC)); 139 "--" return (operator(T_INCDEC, DEC)); 140 "->" return (operator(T_STROP, ARROW)); 141 "." return (operator(T_STROP, POINT)); 142 "+" return (operator(T_ADDOP, PLUS)); 143 "-" return (operator(T_ADDOP, MINUS)); 144 "*" return (operator(T_MULT, MULT)); 145 "/" return (operator(T_DIVOP, DIV)); 146 "%" return (operator(T_DIVOP, MOD)); 147 "!" return (operator(T_UNOP, NOT)); 148 "~" return (operator(T_UNOP, COMPL)); 149 "\"" return (string()); 150 "L\"" return (wcstrg()); 151 ";" return (T_SEMI); 152 "{" return (T_LBRACE); 153 "}" return (T_RBRACE); 154 "," return (T_COMMA); 155 ":" return (T_COLON); 156 "?" return (T_QUEST); 157 "[" return (T_LBRACK); 158 "]" return (T_RBRACK); 159 "(" return (T_LPARN); 160 ")" return (T_RPARN); 161 "..." return (T_ELLIPSE); 162 "'" return (ccon()); 163 "L'" return (wccon()); 164 ^#.*$ directive(); 165 \n incline(); 166 \t|" "|\f|\v ; 167 "/*" comment(); 168 "//" slashslashcomment(); 169 . badchar(yytext[0]); 170 171 %% 172 173 static void 174 incline(void) 175 { 176 curr_pos.p_line++; 177 curr_pos.p_uniq = 0; 178 if (curr_pos.p_file == csrc_pos.p_file) { 179 csrc_pos.p_line++; 180 csrc_pos.p_uniq = 0; 181 } 182 } 183 184 static void 185 badchar(int c) 186 { 187 188 /* unknown character \%o */ 189 error(250, c); 190 } 191 192 /* 193 * Keywords. 194 * During initialisation they are written to the symbol table. 195 */ 196 static struct kwtab { 197 const char *kw_name; /* keyword */ 198 int kw_token; /* token returned by yylex() */ 199 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 200 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 201 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 202 u_int kw_c89 : 1; /* c89 keyword */ 203 u_int kw_c99 : 1; /* c99 keyword */ 204 u_int kw_gcc : 1; /* GCC keyword */ 205 u_int kw_attr : 1; /* GCC attribute, keyword */ 206 u_int kw_deco : 3; /* name[1] __name[2] __name__[4] */ 207 } kwtab[] = { 208 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 }, 209 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 }, 210 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 }, 211 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 212 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 }, 213 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 }, 214 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 }, 215 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 }, 216 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 }, 217 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 }, 218 { "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 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 }, 222 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 }, 223 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 }, 224 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 }, 225 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 }, 226 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 }, 227 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 }, 228 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 }, 229 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 }, 230 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 }, 231 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 }, 232 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 }, 233 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 }, 234 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 }, 235 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 }, 236 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 }, 237 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 }, 238 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 }, 239 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 }, 240 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 }, 241 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 }, 242 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 }, 243 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 }, 244 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 }, 245 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 }, 246 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 }, 247 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 }, 248 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 }, 249 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 250 0, 0, 0, 0,0,1,1,5 }, 251 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 }, 252 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 }, 253 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 }, 254 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 }, 255 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 256 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 257 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 258 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 259 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 260 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 261 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 262 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 263 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 264 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 265 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 266 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 267 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 268 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 269 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 270 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 271 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 }, 272 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 273 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 274 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 275 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 276 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 277 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 }, 278 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 279 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 280 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 281 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 282 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 283 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 284 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 285 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 286 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 287 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 288 }; 289 290 /* Symbol table */ 291 static sym_t *symtab[HSHSIZ1]; 292 293 /* bit i of the entry with index i is set */ 294 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 295 296 /* least significant i bits are set in the entry with index i */ 297 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 298 299 /* least significant i bits are not set in the entry with index i */ 300 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 301 302 /* free list for sbuf structures */ 303 static sbuf_t *sbfrlst; 304 305 /* Typ of next expected symbol */ 306 symt_t symtyp; 307 308 309 static void 310 addkw(struct kwtab *kw, int deco) 311 { 312 sym_t *sym; 313 size_t h; 314 char buf[256]; 315 const char *name; 316 317 if (!(kw->kw_deco & deco)) 318 return; 319 320 switch (deco) { 321 case 1: 322 name = kw->kw_name; 323 break; 324 case 2: 325 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 326 name = strdup(buf); 327 break; 328 case 4: 329 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 330 name = strdup(buf); 331 break; 332 default: 333 abort(); 334 } 335 336 if (name == NULL) 337 err(1, "Can't init symbol table"); 338 339 sym = getblk(sizeof (sym_t)); 340 sym->s_name = name; 341 sym->s_keyw = kw; 342 sym->s_value.v_quad = kw->kw_token; 343 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 344 sym->s_tspec = kw->kw_tspec; 345 } else if (kw->kw_token == T_SCLASS) { 346 sym->s_scl = kw->kw_scl; 347 } else if (kw->kw_token == T_QUAL) { 348 sym->s_tqual = kw->kw_tqual; 349 } 350 h = hash(sym->s_name); 351 if ((sym->s_link = symtab[h]) != NULL) 352 symtab[h]->s_rlink = &sym->s_link; 353 sym->s_rlink = &symtab[h]; 354 symtab[h] = sym; 355 } 356 357 /* 358 * All keywords are written to the symbol table. This saves us looking 359 * in a extra table for each name we found. 360 */ 361 void 362 initscan(void) 363 { 364 struct kwtab *kw; 365 size_t i; 366 uint64_t uq; 367 368 for (kw = kwtab; kw->kw_name != NULL; kw++) { 369 if ((kw->kw_c89 || kw->kw_c99) && tflag) 370 continue; 371 if (kw->kw_c99 && !(Sflag || gflag)) 372 continue; 373 if (kw->kw_gcc && !gflag) 374 continue; 375 addkw(kw, 1); 376 addkw(kw, 2); 377 addkw(kw, 4); 378 } 379 380 /* initialize bit-masks for quads */ 381 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 382 qbmasks[i] = (uint64_t)1 << i; 383 uq = ~(uint64_t)0 << i; 384 qumasks[i] = uq; 385 qlmasks[i] = ~uq; 386 } 387 qumasks[i] = 0; 388 qlmasks[i] = ~(uint64_t)0; 389 } 390 391 /* 392 * Get a free sbuf structure, if possible from the free list 393 */ 394 static sbuf_t * 395 allocsb(void) 396 { 397 sbuf_t *sb; 398 399 if ((sb = sbfrlst) != NULL) { 400 sbfrlst = sb->sb_nxt; 401 #ifdef BLKDEBUG 402 (void)memset(sb, 0, sizeof (*sb)); 403 #else 404 sb->sb_nxt = NULL; 405 #endif 406 } else { 407 sb = xmalloc(sizeof (sbuf_t)); 408 (void)memset(sb, 0, sizeof (*sb)); 409 } 410 return (sb); 411 } 412 413 /* 414 * Put a sbuf structure to the free list 415 */ 416 static void 417 freesb(sbuf_t *sb) 418 { 419 420 (void)memset(sb, ZERO, sizeof (*sb)); 421 sb->sb_nxt = sbfrlst; 422 sbfrlst = sb; 423 } 424 425 /* 426 * Read a character and ensure that it is positive (except EOF). 427 * Increment line count(s) if necessary. 428 */ 429 static int 430 inpc(void) 431 { 432 int c; 433 434 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 435 incline(); 436 return (c); 437 } 438 439 static int 440 hash(const char *s) 441 { 442 u_int v; 443 const u_char *us; 444 445 v = 0; 446 for (us = (const u_char *)s; *us != '\0'; us++) { 447 v = (v << sizeof (v)) + *us; 448 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 449 } 450 return (v % HSHSIZ1); 451 } 452 453 /* 454 * Lex has found a letter followed by zero or more letters or digits. 455 * It looks for a symbol in the symbol table with the same name. This 456 * symbol must either be a keyword or a symbol of the type required by 457 * symtyp (label, member, tag, ...). 458 * 459 * If it is a keyword, the token is returned. In some cases it is described 460 * more deeply by data written to yylval. 461 * 462 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 463 * is stored in yylval. This struct contains the name of the symbol, its 464 * length and hash value. If there is already a symbol of the same name 465 * and type in the symbol table, the sbuf struct also contains a pointer 466 * to the symbol table entry. 467 */ 468 static int 469 name(void) 470 { 471 char *s; 472 sbuf_t *sb; 473 sym_t *sym; 474 int tok; 475 476 sb = allocsb(); 477 sb->sb_name = yytext; 478 sb->sb_len = yyleng; 479 sb->sb_hash = hash(yytext); 480 if ((sym = search(sb)) != NULL && sym->s_keyw) { 481 freesb(sb); 482 return (keyw(sym)); 483 } 484 485 sb->sb_sym = sym; 486 487 if (sym != NULL) { 488 if (blklev < sym->s_blklev) 489 LERROR("name()"); 490 sb->sb_name = sym->s_name; 491 sb->sb_len = strlen(sym->s_name); 492 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 493 } else { 494 s = getblk(yyleng + 1); 495 (void)memcpy(s, yytext, yyleng + 1); 496 sb->sb_name = s; 497 sb->sb_len = yyleng; 498 tok = T_NAME; 499 } 500 501 yylval.y_sb = sb; 502 return (tok); 503 } 504 505 static sym_t * 506 search(sbuf_t *sb) 507 { 508 sym_t *sym; 509 510 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 511 if (strcmp(sym->s_name, sb->sb_name) == 0) { 512 if (sym->s_keyw) { 513 struct kwtab *kw = sym->s_keyw; 514 if (!kw->kw_attr || attron) 515 return (sym); 516 } else if (!attron && sym->s_kind == symtyp) 517 return sym; 518 } 519 } 520 521 return (NULL); 522 } 523 524 static int 525 keyw(sym_t *sym) 526 { 527 int t; 528 529 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 530 yylval.y_scl = sym->s_scl; 531 } else if (t == T_TYPE || t == T_SOU) { 532 yylval.y_tspec = sym->s_tspec; 533 } else if (t == T_QUAL) { 534 yylval.y_tqual = sym->s_tqual; 535 } 536 return (t); 537 } 538 539 /* 540 * Convert a string representing an integer into internal representation. 541 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 542 */ 543 static int 544 icon(int base) 545 { 546 int l_suffix, u_suffix; 547 int len; 548 const char *cp; 549 char c, *eptr; 550 tspec_t typ; 551 uint64_t uq = 0; 552 int ansiu; 553 static tspec_t contypes[2][3] = { 554 { INT, LONG, QUAD }, 555 { UINT, ULONG, UQUAD } 556 }; 557 558 cp = yytext; 559 len = yyleng; 560 561 /* skip 0[xX] or 0[bB] */ 562 if (base == 16 || base == 2) { 563 cp += 2; 564 len -= 2; 565 } 566 567 /* read suffixes */ 568 l_suffix = u_suffix = 0; 569 for ( ; ; ) { 570 if ((c = cp[len - 1]) == 'l' || c == 'L') { 571 l_suffix++; 572 } else if (c == 'u' || c == 'U') { 573 u_suffix++; 574 } else { 575 break; 576 } 577 len--; 578 } 579 if (l_suffix > 2 || u_suffix > 1) { 580 /* malformed integer constant */ 581 warning(251); 582 if (l_suffix > 2) 583 l_suffix = 2; 584 if (u_suffix > 1) 585 u_suffix = 1; 586 } 587 if (tflag && u_suffix != 0) { 588 /* suffix U is illegal in traditional C */ 589 warning(97); 590 } 591 typ = contypes[u_suffix][l_suffix]; 592 593 errno = 0; 594 595 uq = strtouq(cp, &eptr, base); 596 if (eptr != cp + len) 597 LERROR("icon()"); 598 if (errno != 0) 599 /* integer constant out of range */ 600 warning(252); 601 602 /* 603 * If the value is too big for the current type, we must choose 604 * another type. 605 */ 606 ansiu = 0; 607 switch (typ) { 608 case INT: 609 if (uq <= TARG_INT_MAX) { 610 /* ok */ 611 } else if (uq <= TARG_UINT_MAX && base != 10) { 612 typ = UINT; 613 } else if (uq <= TARG_LONG_MAX) { 614 typ = LONG; 615 } else { 616 typ = ULONG; 617 if (uq > TARG_ULONG_MAX) { 618 /* integer constant out of range */ 619 warning(252); 620 } 621 } 622 if (typ == UINT || typ == ULONG) { 623 if (tflag) { 624 typ = LONG; 625 } else if (!sflag) { 626 /* 627 * Remember that the constant is unsigned 628 * only in ANSI C 629 */ 630 ansiu = 1; 631 } 632 } 633 break; 634 case UINT: 635 if (uq > TARG_UINT_MAX) { 636 typ = ULONG; 637 if (uq > TARG_ULONG_MAX) { 638 /* integer constant out of range */ 639 warning(252); 640 } 641 } 642 break; 643 case LONG: 644 if (uq > TARG_LONG_MAX && !tflag) { 645 typ = ULONG; 646 if (!sflag) 647 ansiu = 1; 648 if (uq > TARG_ULONG_MAX) { 649 /* integer constant out of range */ 650 warning(252); 651 } 652 } 653 break; 654 case ULONG: 655 if (uq > TARG_ULONG_MAX) { 656 /* integer constant out of range */ 657 warning(252); 658 } 659 break; 660 case QUAD: 661 if (uq > TARG_QUAD_MAX && !tflag) { 662 typ = UQUAD; 663 if (!sflag) 664 ansiu = 1; 665 } 666 break; 667 case UQUAD: 668 if (uq > TARG_UQUAD_MAX) { 669 /* integer constant out of range */ 670 warning(252); 671 } 672 break; 673 /* LINTED206: (enumeration values not handled in switch) */ 674 case STRUCT: 675 case VOID: 676 case LDOUBLE: 677 case FUNC: 678 case ARRAY: 679 case PTR: 680 case ENUM: 681 case UNION: 682 case SIGNED: 683 case NOTSPEC: 684 case DOUBLE: 685 case FLOAT: 686 case USHORT: 687 case SHORT: 688 case UCHAR: 689 case SCHAR: 690 case CHAR: 691 case BOOL: 692 case UNSIGN: 693 case FCOMPLEX: 694 case DCOMPLEX: 695 case LCOMPLEX: 696 case COMPLEX: 697 break; 698 699 case NTSPEC: /* this value unused */ 700 break; 701 } 702 703 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 704 705 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 706 yylval.y_val->v_ansiu = ansiu; 707 yylval.y_val->v_quad = (int64_t)uq; 708 709 return (T_CON); 710 } 711 712 /* 713 * Returns 1 if t is a signed type and the value is negative. 714 * 715 * len is the number of significant bits. If len is -1, len is set 716 * to the width of type t. 717 */ 718 int 719 sign(int64_t q, tspec_t t, int len) 720 { 721 722 if (t == PTR || isutyp(t)) 723 return (0); 724 return (msb(q, t, len)); 725 } 726 727 int 728 msb(int64_t q, tspec_t t, int len) 729 { 730 731 if (len <= 0) 732 len = size(t); 733 return ((q & qbmasks[len - 1]) != 0); 734 } 735 736 /* 737 * Extends the sign of q. 738 */ 739 int64_t 740 xsign(int64_t q, tspec_t t, int len) 741 { 742 743 if (len <= 0) 744 len = size(t); 745 746 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 747 q &= qlmasks[len]; 748 } else { 749 q |= qumasks[len]; 750 } 751 return (q); 752 } 753 754 /* 755 * Convert a string representing a floating point value into its interal 756 * representation. Type and value are returned in yylval. fcon() 757 * (and yylex()) returns T_CON. 758 * XXX Currently it is not possible to convert constants of type 759 * long double which are greater than DBL_MAX. 760 */ 761 static int 762 fcon(void) 763 { 764 const char *cp; 765 int len; 766 tspec_t typ; 767 char c, *eptr; 768 double d; 769 float f = 0; 770 771 cp = yytext; 772 len = yyleng; 773 774 if (cp[len - 1] == 'i') { 775 /* imaginary, do nothing for now */ 776 len--; 777 } 778 if ((c = cp[len - 1]) == 'f' || c == 'F') { 779 typ = FLOAT; 780 len--; 781 } else if (c == 'l' || c == 'L') { 782 typ = LDOUBLE; 783 len--; 784 } else { 785 if (c == 'd' || c == 'D') 786 len--; 787 typ = DOUBLE; 788 } 789 790 if (tflag && typ != DOUBLE) { 791 /* suffixes F and L are illegal in traditional C */ 792 warning(98); 793 } 794 795 errno = 0; 796 d = strtod(cp, &eptr); 797 if (eptr != cp + len) { 798 switch (*eptr) { 799 /* 800 * XXX: non-native non-current strtod() may not handle hex 801 * floats, ignore the rest if we find traces of hex float 802 * syntax... 803 */ 804 case 'p': 805 case 'P': 806 case 'x': 807 case 'X': 808 d = 0; 809 errno = 0; 810 break; 811 default: 812 LERROR("fcon(%s->%s)", cp, eptr); 813 } 814 } 815 if (errno != 0) 816 /* floating-point constant out of range */ 817 warning(248); 818 819 if (typ == FLOAT) { 820 f = (float)d; 821 if (!finite(f)) { 822 /* floating-point constant out of range */ 823 warning(248); 824 f = f > 0 ? FLT_MAX : -FLT_MAX; 825 } 826 } 827 828 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 829 if (typ == FLOAT) { 830 yylval.y_val->v_ldbl = f; 831 } else { 832 yylval.y_val->v_ldbl = d; 833 } 834 835 return (T_CON); 836 } 837 838 static int 839 operator(int t, op_t o) 840 { 841 842 yylval.y_op = o; 843 return (t); 844 } 845 846 /* 847 * Called if lex found a leading \'. 848 */ 849 static int 850 ccon(void) 851 { 852 size_t n; 853 int val, c; 854 char cv; 855 856 n = 0; 857 val = 0; 858 while ((c = getescc('\'')) >= 0) { 859 val = (val << CHAR_BIT) + c; 860 n++; 861 } 862 if (c == -2) { 863 /* unterminated character constant */ 864 error(253); 865 } else { 866 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 867 /* too many characters in character constant */ 868 error(71); 869 } else if (n > 1) { 870 /* multi-character character constant */ 871 warning(294); 872 } else if (n == 0) { 873 /* empty character constant */ 874 error(73); 875 } 876 } 877 if (n == 1) { 878 cv = (char)val; 879 val = cv; 880 } 881 882 yylval.y_val = xcalloc(1, sizeof (val_t)); 883 yylval.y_val->v_tspec = INT; 884 yylval.y_val->v_quad = val; 885 886 return (T_CON); 887 } 888 889 /* 890 * Called if lex found a leading L\' 891 */ 892 static int 893 wccon(void) 894 { 895 static char buf[MB_LEN_MAX + 1]; 896 size_t i; 897 int c; 898 wchar_t wc; 899 900 i = 0; 901 while ((c = getescc('\'')) >= 0) { 902 if (i < MB_CUR_MAX) 903 buf[i] = (char)c; 904 i++; 905 } 906 907 wc = 0; 908 909 if (c == -2) { 910 /* unterminated character constant */ 911 error(253); 912 } else if (c == 0) { 913 /* empty character constant */ 914 error(73); 915 } else { 916 if (i > MB_CUR_MAX) { 917 i = MB_CUR_MAX; 918 /* too many characters in character constant */ 919 error(71); 920 } else { 921 buf[i] = '\0'; 922 (void)mbtowc(NULL, NULL, 0); 923 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 924 /* invalid multibyte character */ 925 error(291); 926 } 927 } 928 929 yylval.y_val = xcalloc(1, sizeof (val_t)); 930 yylval.y_val->v_tspec = WCHAR; 931 yylval.y_val->v_quad = wc; 932 933 return (T_CON); 934 } 935 936 /* 937 * Read a character which is part of a character constant or of a string 938 * and handle escapes. 939 * 940 * The Argument is the character which delimits the character constant or 941 * string. 942 * 943 * Returns -1 if the end of the character constant or string is reached, 944 * -2 if the EOF is reached, and the character otherwise. 945 */ 946 static int 947 getescc(int d) 948 { 949 static int pbc = -1; 950 int n, c, v; 951 952 if (pbc == -1) { 953 c = inpc(); 954 } else { 955 c = pbc; 956 pbc = -1; 957 } 958 if (c == d) 959 return (-1); 960 switch (c) { 961 case '\n': 962 if (tflag) { 963 /* newline in string or char constant */ 964 error(254); 965 return (-2); 966 } 967 return (c); 968 case EOF: 969 return (-2); 970 case '\\': 971 switch (c = inpc()) { 972 case '"': 973 if (tflag && d == '\'') 974 /* \" inside character constant undef. ... */ 975 warning(262); 976 return ('"'); 977 case '\'': 978 return ('\''); 979 case '?': 980 if (tflag) 981 /* \? undefined in traditional C */ 982 warning(263); 983 return ('?'); 984 case '\\': 985 return ('\\'); 986 case 'a': 987 if (tflag) 988 /* \a undefined in traditional C */ 989 warning(81); 990 return ('\a'); 991 case 'b': 992 return ('\b'); 993 case 'f': 994 return ('\f'); 995 case 'n': 996 return ('\n'); 997 case 'r': 998 return ('\r'); 999 case 't': 1000 return ('\t'); 1001 case 'v': 1002 if (tflag) 1003 /* \v undefined in traditional C */ 1004 warning(264); 1005 return ('\v'); 1006 case '8': case '9': 1007 /* bad octal digit %c */ 1008 warning(77, c); 1009 /* FALLTHROUGH */ 1010 case '0': case '1': case '2': case '3': 1011 case '4': case '5': case '6': case '7': 1012 n = 3; 1013 v = 0; 1014 do { 1015 v = (v << 3) + (c - '0'); 1016 c = inpc(); 1017 } while (--n && isdigit(c) && (tflag || c <= '7')); 1018 if (tflag && n > 0 && isdigit(c)) 1019 /* bad octal digit %c */ 1020 warning(77, c); 1021 pbc = c; 1022 if (v > TARG_UCHAR_MAX) { 1023 /* character escape does not fit in char. */ 1024 warning(76); 1025 v &= CHAR_MASK; 1026 } 1027 return (v); 1028 case 'x': 1029 if (tflag) 1030 /* \x undefined in traditional C */ 1031 warning(82); 1032 v = 0; 1033 n = 0; 1034 while ((c = inpc()) >= 0 && isxdigit(c)) { 1035 c = isdigit(c) ? 1036 c - '0' : toupper(c) - 'A' + 10; 1037 v = (v << 4) + c; 1038 if (n >= 0) { 1039 if ((v & ~CHAR_MASK) != 0) { 1040 /* overflow in hex escape */ 1041 warning(75); 1042 n = -1; 1043 } else { 1044 n++; 1045 } 1046 } 1047 } 1048 pbc = c; 1049 if (n == 0) { 1050 /* no hex digits follow \x */ 1051 error(74); 1052 } if (n == -1) { 1053 v &= CHAR_MASK; 1054 } 1055 return (v); 1056 case '\n': 1057 return (getescc(d)); 1058 case EOF: 1059 return (-2); 1060 default: 1061 if (isprint(c)) { 1062 /* dubious escape \%c */ 1063 warning(79, c); 1064 } else { 1065 /* dubious escape \%o */ 1066 warning(80, c); 1067 } 1068 } 1069 } 1070 return (c); 1071 } 1072 1073 /* 1074 * Called for preprocessor directives. Currently implemented are: 1075 * # lineno 1076 * # lineno "filename" 1077 */ 1078 static void 1079 directive(void) 1080 { 1081 const char *cp, *fn; 1082 char c, *eptr; 1083 size_t fnl; 1084 long ln; 1085 static int first = 1; 1086 1087 /* Go to first non-whitespace after # */ 1088 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1089 continue; 1090 1091 if (!isdigit((unsigned char)c)) { 1092 error: 1093 /* undefined or invalid # directive */ 1094 warning(255); 1095 return; 1096 } 1097 ln = strtol(--cp, &eptr, 10); 1098 if (cp == eptr) 1099 goto error; 1100 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1101 goto error; 1102 while ((c = *cp++) == ' ' || c == '\t') 1103 continue; 1104 if (c != '\0') { 1105 if (c != '"') 1106 goto error; 1107 fn = cp; 1108 while ((c = *cp) != '"' && c != '\0') 1109 cp++; 1110 if (c != '"') 1111 goto error; 1112 if ((fnl = cp++ - fn) > PATH_MAX) 1113 goto error; 1114 while ((c = *cp++) == ' ' || c == '\t') 1115 continue; 1116 #if 0 1117 if (c != '\0') 1118 warning("extra character(s) after directive"); 1119 #endif 1120 1121 /* empty string means stdin */ 1122 if (fnl == 0) { 1123 fn = "{standard input}"; 1124 fnl = 16; /* strlen (fn) */ 1125 } 1126 curr_pos.p_file = fnnalloc(fn, fnl); 1127 /* 1128 * If this is the first directive, the name is the name 1129 * of the C source file as specified at the command line. 1130 * It is written to the output file. 1131 */ 1132 if (first) { 1133 csrc_pos.p_file = curr_pos.p_file; 1134 outsrc(fnxform(curr_pos.p_file, 1135 strlen(curr_pos.p_file))); 1136 first = 0; 1137 } 1138 } 1139 curr_pos.p_line = (int)ln - 1; 1140 curr_pos.p_uniq = 0; 1141 if (curr_pos.p_file == csrc_pos.p_file) { 1142 csrc_pos.p_line = (int)ln - 1; 1143 csrc_pos.p_uniq = 0; 1144 } 1145 } 1146 1147 /* 1148 * Handle lint comments. Following comments are currently understood: 1149 * ARGSUSEDn 1150 * BITFIELDTYPE 1151 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1152 * FALLTHRU FALLTHROUGH 1153 * LINTLIBRARY 1154 * LINTEDn NOSTRICTn 1155 * LONGLONG 1156 * NOTREACHED 1157 * PRINTFLIKEn 1158 * PROTOLIB 1159 * SCANFLIKEn 1160 * VARARGSn 1161 * If one of this comments is recognized, the arguments, if any, are 1162 * parsed and a function which handles this comment is called. 1163 */ 1164 static void 1165 comment(void) 1166 { 1167 int c, lc; 1168 static struct { 1169 const char *keywd; 1170 int arg; 1171 void (*func)(int); 1172 } keywtab[] = { 1173 { "ARGSUSED", 1, argsused }, 1174 { "BITFIELDTYPE", 0, bitfieldtype }, 1175 { "CONSTCOND", 0, constcond }, 1176 { "CONSTANTCOND", 0, constcond }, 1177 { "CONSTANTCONDITION", 0, constcond }, 1178 { "FALLTHRU", 0, fallthru }, 1179 { "FALLTHROUGH", 0, fallthru }, 1180 { "LINTLIBRARY", 0, lintlib }, 1181 { "LINTED", 1, linted }, 1182 { "LONGLONG", 0, longlong }, 1183 { "NOSTRICT", 1, linted }, 1184 { "NOTREACHED", 0, notreach }, 1185 { "PRINTFLIKE", 1, printflike }, 1186 { "PROTOLIB", 1, protolib }, 1187 { "SCANFLIKE", 1, scanflike }, 1188 { "VARARGS", 1, varargs }, 1189 }; 1190 char keywd[32]; 1191 char arg[32]; 1192 size_t l, i; 1193 int a; 1194 int eoc; 1195 1196 eoc = 0; 1197 1198 /* Skip white spaces after the start of the comment */ 1199 while ((c = inpc()) != EOF && isspace(c)) 1200 continue; 1201 1202 /* Read the potential keyword to keywd */ 1203 l = 0; 1204 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1205 keywd[l++] = (char)c; 1206 c = inpc(); 1207 } 1208 keywd[l] = '\0'; 1209 1210 /* look for the keyword */ 1211 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1212 if (strcmp(keywtab[i].keywd, keywd) == 0) 1213 break; 1214 } 1215 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1216 goto skip_rest; 1217 1218 /* skip white spaces after the keyword */ 1219 while (c != EOF && isspace(c)) 1220 c = inpc(); 1221 1222 /* read the argument, if the keyword accepts one and there is one */ 1223 l = 0; 1224 if (keywtab[i].arg) { 1225 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1226 arg[l++] = (char)c; 1227 c = inpc(); 1228 } 1229 } 1230 arg[l] = '\0'; 1231 a = l != 0 ? atoi(arg) : -1; 1232 1233 /* skip white spaces after the argument */ 1234 while (c != EOF && isspace(c)) 1235 c = inpc(); 1236 1237 if (c != '*' || (c = inpc()) != '/') { 1238 if (keywtab[i].func != linted) 1239 /* extra characters in lint comment */ 1240 warning(257); 1241 } else { 1242 /* 1243 * remember that we have already found the end of the 1244 * comment 1245 */ 1246 eoc = 1; 1247 } 1248 1249 if (keywtab[i].func != NULL) 1250 (*keywtab[i].func)(a); 1251 1252 skip_rest: 1253 while (!eoc) { 1254 lc = c; 1255 if ((c = inpc()) == EOF) { 1256 /* unterminated comment */ 1257 error(256); 1258 break; 1259 } 1260 if (lc == '*' && c == '/') 1261 eoc = 1; 1262 } 1263 } 1264 1265 /* 1266 * Handle // style comments 1267 */ 1268 static void 1269 slashslashcomment(void) 1270 { 1271 int c; 1272 1273 if (!Sflag && !gflag) 1274 /* // comments only supported in C99 */ 1275 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1276 1277 while ((c = inpc()) != EOF && c != '\n') 1278 continue; 1279 } 1280 1281 /* 1282 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1283 * clrwflgs() is called after function definitions and global and 1284 * local declarations and definitions. It is also called between 1285 * the controlling expression and the body of control statements 1286 * (if, switch, for, while). 1287 */ 1288 void 1289 clrwflgs(void) 1290 { 1291 1292 lwarn = LWARN_ALL; 1293 quadflg = 0; 1294 ccflg = 0; 1295 } 1296 1297 /* 1298 * Strings are stored in a dynamically alloceted buffer and passed 1299 * in yylval.y_xstrg to the parser. The parser or the routines called 1300 * by the parser are responsible for freeing this buffer. 1301 */ 1302 static int 1303 string(void) 1304 { 1305 u_char *s; 1306 int c; 1307 size_t len, max; 1308 strg_t *strg; 1309 1310 s = xmalloc(max = 64); 1311 1312 len = 0; 1313 while ((c = getescc('"')) >= 0) { 1314 /* +1 to reserve space for a trailing NUL character */ 1315 if (len + 1 == max) 1316 s = xrealloc(s, max *= 2); 1317 s[len++] = (char)c; 1318 } 1319 s[len] = '\0'; 1320 if (c == -2) 1321 /* unterminated string constant */ 1322 error(258); 1323 1324 strg = xcalloc(1, sizeof (strg_t)); 1325 strg->st_tspec = CHAR; 1326 strg->st_len = len; 1327 strg->st_cp = s; 1328 1329 yylval.y_strg = strg; 1330 return (T_STRING); 1331 } 1332 1333 static int 1334 wcstrg(void) 1335 { 1336 char *s; 1337 int c, n; 1338 size_t i, wi; 1339 size_t len, max, wlen; 1340 wchar_t *ws; 1341 strg_t *strg; 1342 1343 s = xmalloc(max = 64); 1344 len = 0; 1345 while ((c = getescc('"')) >= 0) { 1346 /* +1 to save space for a trailing NUL character */ 1347 if (len + 1 >= max) 1348 s = xrealloc(s, max *= 2); 1349 s[len++] = (char)c; 1350 } 1351 s[len] = '\0'; 1352 if (c == -2) 1353 /* unterminated string constant */ 1354 error(258); 1355 1356 /* get length of wide-character string */ 1357 (void)mblen(NULL, 0); 1358 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1359 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1360 /* invalid multibyte character */ 1361 error(291); 1362 break; 1363 } 1364 if (n == 0) 1365 n = 1; 1366 } 1367 1368 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1369 1370 /* convert from multibyte to wide char */ 1371 (void)mbtowc(NULL, NULL, 0); 1372 for (i = 0, wi = 0; i < len; i += n, wi++) { 1373 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1374 break; 1375 if (n == 0) 1376 n = 1; 1377 } 1378 ws[wi] = 0; 1379 free(s); 1380 1381 strg = xcalloc(1, sizeof (strg_t)); 1382 strg->st_tspec = WCHAR; 1383 strg->st_len = wlen; 1384 strg->st_wcp = ws; 1385 1386 yylval.y_strg = strg; 1387 return (T_STRING); 1388 } 1389 1390 /* 1391 * As noted above the scanner does not create new symbol table entries 1392 * for symbols it cannot find in the symbol table. This is to avoid 1393 * putting undeclared symbols into the symbol table if a syntax error 1394 * occurs. 1395 * 1396 * getsym() is called as soon as it is probably ok to put the symbol to 1397 * the symbol table. This does not mean that it is not possible that 1398 * symbols are put to the symbol table which are than not completely 1399 * declared due to syntax errors. To avoid too many problems in this 1400 * case symbols get type int in getsym(). 1401 * 1402 * XXX calls to getsym() should be delayed until decl1*() is called 1403 */ 1404 sym_t * 1405 getsym(sbuf_t *sb) 1406 { 1407 dinfo_t *di; 1408 char *s; 1409 sym_t *sym; 1410 1411 sym = sb->sb_sym; 1412 1413 /* 1414 * During member declaration it is possible that name() looked 1415 * for symbols of type FVFT, although it should have looked for 1416 * symbols of type FTAG. Same can happen for labels. Both cases 1417 * are compensated here. 1418 */ 1419 if (symtyp == FMOS || symtyp == FLAB) { 1420 if (sym == NULL || sym->s_kind == FVFT) 1421 sym = search(sb); 1422 } 1423 1424 if (sym != NULL) { 1425 if (sym->s_kind != symtyp) 1426 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1427 symtyp = FVFT; 1428 freesb(sb); 1429 return (sym); 1430 } 1431 1432 /* create a new symbol table entry */ 1433 1434 /* labels must always be allocated at level 1 (outhermost block) */ 1435 if (symtyp == FLAB) { 1436 sym = getlblk(1, sizeof (sym_t)); 1437 s = getlblk(1, sb->sb_len + 1); 1438 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1439 sym->s_name = s; 1440 sym->s_blklev = 1; 1441 di = dcs; 1442 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1443 di = di->d_nxt; 1444 if (di->d_ctx != AUTO) 1445 LERROR("storesym()"); 1446 } else { 1447 sym = getblk(sizeof (sym_t)); 1448 sym->s_name = sb->sb_name; 1449 sym->s_blklev = blklev; 1450 di = dcs; 1451 } 1452 1453 UNIQUE_CURR_POS(sym->s_dpos); 1454 if ((sym->s_kind = symtyp) != FLAB) 1455 sym->s_type = gettyp(INT); 1456 1457 symtyp = FVFT; 1458 1459 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1460 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1461 sym->s_rlink = &symtab[sb->sb_hash]; 1462 symtab[sb->sb_hash] = sym; 1463 1464 *di->d_ldlsym = sym; 1465 di->d_ldlsym = &sym->s_dlnxt; 1466 1467 freesb(sb); 1468 return (sym); 1469 } 1470 1471 /* 1472 * Construct a temporary symbol. The symbol starts with a digit, so that 1473 * it is illegal. 1474 */ 1475 sym_t * 1476 mktempsym(type_t *t) 1477 { 1478 static int n = 0; 1479 int h; 1480 char *s = getlblk(blklev, 64); 1481 sym_t *sym = getblk(sizeof (sym_t)); 1482 1483 (void)snprintf(s, 64, "%.8d_tmp", n++); 1484 h = hash(s); 1485 1486 sym->s_name = s; 1487 sym->s_type = t; 1488 sym->s_blklev = blklev; 1489 sym->s_scl = AUTO; 1490 sym->s_kind = FVFT; 1491 sym->s_used = 1; 1492 sym->s_set = 1; 1493 1494 if ((sym->s_link = symtab[h]) != NULL) 1495 symtab[h]->s_rlink = &sym->s_link; 1496 sym->s_rlink = &symtab[h]; 1497 symtab[h] = sym; 1498 1499 *dcs->d_ldlsym = sym; 1500 dcs->d_ldlsym = &sym->s_dlnxt; 1501 1502 return sym; 1503 } 1504 1505 /* 1506 * Remove a symbol forever from the symbol table. s_blklev 1507 * is set to -1 to avoid that the symbol will later be put 1508 * back to the symbol table. 1509 */ 1510 void 1511 rmsym(sym_t *sym) 1512 { 1513 1514 if ((*sym->s_rlink = sym->s_link) != NULL) 1515 sym->s_link->s_rlink = sym->s_rlink; 1516 sym->s_blklev = -1; 1517 sym->s_link = NULL; 1518 } 1519 1520 /* 1521 * Remove a list of symbols declared at one level from the symbol 1522 * table. 1523 */ 1524 void 1525 rmsyms(sym_t *syms) 1526 { 1527 sym_t *sym; 1528 1529 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1530 if (sym->s_blklev != -1) { 1531 if ((*sym->s_rlink = sym->s_link) != NULL) 1532 sym->s_link->s_rlink = sym->s_rlink; 1533 sym->s_link = NULL; 1534 sym->s_rlink = NULL; 1535 } 1536 } 1537 } 1538 1539 /* 1540 * Put a symbol into the symbol table 1541 */ 1542 void 1543 inssym(int bl, sym_t *sym) 1544 { 1545 int h; 1546 1547 h = hash(sym->s_name); 1548 if ((sym->s_link = symtab[h]) != NULL) 1549 symtab[h]->s_rlink = &sym->s_link; 1550 sym->s_rlink = &symtab[h]; 1551 symtab[h] = sym; 1552 sym->s_blklev = bl; 1553 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1554 LERROR("inssym()"); 1555 } 1556 1557 /* 1558 * Called at level 0 after syntax errors 1559 * Removes all symbols which are not declared at level 0 from the 1560 * symbol table. Also frees all memory which is not associated with 1561 * level 0. 1562 */ 1563 void 1564 cleanup(void) 1565 { 1566 sym_t *sym, *nsym; 1567 int i; 1568 1569 for (i = 0; i < HSHSIZ1; i++) { 1570 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1571 nsym = sym->s_link; 1572 if (sym->s_blklev >= 1) { 1573 if ((*sym->s_rlink = nsym) != NULL) 1574 nsym->s_rlink = sym->s_rlink; 1575 } 1576 } 1577 } 1578 1579 for (i = mblklev; i > 0; i--) 1580 freelblk(i); 1581 } 1582 1583 /* 1584 * Create a new symbol with the name of an existing symbol. 1585 */ 1586 sym_t * 1587 pushdown(sym_t *sym) 1588 { 1589 int h; 1590 sym_t *nsym; 1591 1592 h = hash(sym->s_name); 1593 nsym = getblk(sizeof (sym_t)); 1594 if (sym->s_blklev > blklev) 1595 LERROR("pushdown()"); 1596 nsym->s_name = sym->s_name; 1597 UNIQUE_CURR_POS(nsym->s_dpos); 1598 nsym->s_kind = sym->s_kind; 1599 nsym->s_blklev = blklev; 1600 1601 if ((nsym->s_link = symtab[h]) != NULL) 1602 symtab[h]->s_rlink = &nsym->s_link; 1603 nsym->s_rlink = &symtab[h]; 1604 symtab[h] = nsym; 1605 1606 *dcs->d_ldlsym = nsym; 1607 dcs->d_ldlsym = &nsym->s_dlnxt; 1608 1609 return (nsym); 1610 } 1611 1612 /* 1613 * Free any dynamically allocated memory referenced by 1614 * the value stack or yylval. 1615 * The type of information in yylval is described by tok. 1616 */ 1617 void 1618 freeyyv(void *sp, int tok) 1619 { 1620 if (tok == T_NAME || tok == T_TYPENAME) { 1621 sbuf_t *sb = *(sbuf_t **)sp; 1622 freesb(sb); 1623 } else if (tok == T_CON) { 1624 val_t *val = *(val_t **)sp; 1625 free(val); 1626 } else if (tok == T_STRING) { 1627 strg_t *strg = *(strg_t **)sp; 1628 if (strg->st_tspec == CHAR) { 1629 free(strg->st_cp); 1630 } else if (strg->st_tspec == WCHAR) { 1631 free(strg->st_wcp); 1632 } else { 1633 LERROR("fryylv()"); 1634 } 1635 free(strg); 1636 } 1637 } 1638
Indexes created Thu Oct 02 01:09:59 GMT 2025