scan.l revision 1.55
1 %{ 2 /* $NetBSD: scan.l,v 1.55 2014/04/18 00:23:46 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 1.55 2014/04/18 00:23:46 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 (~(~0 << 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 L [_A-Za-z] 90 D [0-9] 91 NZD [1-9] 92 OD [0-7] 93 HD [0-9A-Fa-f] 94 EX ([eE][+-]?[0-9]+) 95 HX (p[+-]?[0-9A-Fa-f]+) 96 TL ([fFlL]?[i]?) 97 98 %option nounput 99 100 %% 101 102 {L}({L}|{D})* return (name()); 103 0{OD}*[lLuU]* return (icon(8)); 104 {NZD}{D}*[lLuU]* return (icon(10)); 105 0[xX]{HD}+[lLuU]* return (icon(16)); 106 {D}+\.{D}*{EX}?{TL} | 107 {D}+{EX}{TL} | 108 0[xX]{HD}+\.{HD}*{HX}{TL} | 109 0[xX]{HD}+{HX}{TL} | 110 \.{D}+{EX}?{TL} return (fcon()); 111 "=" return (operator(T_ASSIGN, ASSIGN)); 112 "*=" return (operator(T_OPASS, MULASS)); 113 "/=" return (operator(T_OPASS, DIVASS)); 114 "%=" return (operator(T_OPASS, MODASS)); 115 "+=" return (operator(T_OPASS, ADDASS)); 116 "-=" return (operator(T_OPASS, SUBASS)); 117 "<<=" return (operator(T_OPASS, SHLASS)); 118 ">>=" return (operator(T_OPASS, SHRASS)); 119 "&=" return (operator(T_OPASS, ANDASS)); 120 "^=" return (operator(T_OPASS, XORASS)); 121 "|=" return (operator(T_OPASS, ORASS)); 122 "||" return (operator(T_LOGOR, LOGOR)); 123 "&&" return (operator(T_LOGAND, LOGAND)); 124 "|" return (operator(T_OR, OR)); 125 "&" return (operator(T_AND, AND)); 126 "^" return (operator(T_XOR, XOR)); 127 "==" return (operator(T_EQOP, EQ)); 128 "!=" return (operator(T_EQOP, NE)); 129 "<" return (operator(T_RELOP, LT)); 130 ">" return (operator(T_RELOP, GT)); 131 "<=" return (operator(T_RELOP, LE)); 132 ">=" return (operator(T_RELOP, GE)); 133 "<<" return (operator(T_SHFTOP, SHL)); 134 ">>" return (operator(T_SHFTOP, SHR)); 135 "++" return (operator(T_INCDEC, INC)); 136 "--" return (operator(T_INCDEC, DEC)); 137 "->" return (operator(T_STROP, ARROW)); 138 "." return (operator(T_STROP, POINT)); 139 "+" return (operator(T_ADDOP, PLUS)); 140 "-" return (operator(T_ADDOP, MINUS)); 141 "*" return (operator(T_MULT, MULT)); 142 "/" return (operator(T_DIVOP, DIV)); 143 "%" return (operator(T_DIVOP, MOD)); 144 "!" return (operator(T_UNOP, NOT)); 145 "~" return (operator(T_UNOP, COMPL)); 146 "\"" return (string()); 147 "L\"" return (wcstrg()); 148 ";" return (T_SEMI); 149 "{" return (T_LBRACE); 150 "}" return (T_RBRACE); 151 "," return (T_COMMA); 152 ":" return (T_COLON); 153 "?" return (T_QUEST); 154 "[" return (T_LBRACK); 155 "]" return (T_RBRACK); 156 "(" return (T_LPARN); 157 ")" return (T_RPARN); 158 "..." return (T_ELLIPSE); 159 "'" return (ccon()); 160 "L'" return (wccon()); 161 ^#.*$ directive(); 162 \n incline(); 163 \t|" "|\f|\v ; 164 "/*" comment(); 165 "//" slashslashcomment(); 166 . badchar(yytext[0]); 167 168 %% 169 170 static void 171 incline(void) 172 { 173 curr_pos.p_line++; 174 curr_pos.p_uniq = 0; 175 if (curr_pos.p_file == csrc_pos.p_file) { 176 csrc_pos.p_line++; 177 csrc_pos.p_uniq = 0; 178 } 179 } 180 181 static void 182 badchar(int c) 183 { 184 185 /* unknown character \%o */ 186 error(250, c); 187 } 188 189 /* 190 * Keywords. 191 * During initialisation they are written to the symbol table. 192 */ 193 static struct kwtab { 194 const char *kw_name; /* keyword */ 195 int kw_token; /* token returned by yylex() */ 196 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 197 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 198 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 199 u_int kw_c89 : 1; /* c89 keyword */ 200 u_int kw_c99 : 1; /* c99 keyword */ 201 u_int kw_gcc : 1; /* GCC keyword */ 202 u_int kw_attr : 1; /* GCC attribute, keyword */ 203 } kwtab[] = { 204 { "__alignof__", T_ALIGNOF, 0, 0, 0, 0, 0, 0, 0 }, 205 { "__attribute__",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 }, 206 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 }, 207 { "__packed__", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 }, 208 { "packed", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 }, 209 { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 210 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 211 { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1, 1 }, 212 { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1, 1 }, 213 { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 214 { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 215 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 216 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 217 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 }, 218 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 219 { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 220 { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 221 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 222 { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 223 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 224 { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 225 { "strftime", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 226 { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 }, 227 { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 228 { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 229 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 230 { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 231 { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 1 }, 232 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 233 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 234 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 }, 235 { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 }, 236 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 }, 237 { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 }, 238 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 }, 239 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 }, 240 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 }, 241 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 242 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 243 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 }, 244 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 }, 245 { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 }, 246 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 }, 247 { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 }, 248 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 }, 249 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 }, 250 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 }, 251 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 }, 252 { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 }, 253 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 }, 254 { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 }, 255 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 }, 256 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 }, 257 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 258 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 259 { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 }, 260 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 }, 261 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 }, 262 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 }, 263 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 }, 264 { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 265 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 266 { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 }, 267 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 }, 268 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 }, 269 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 }, 270 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 271 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 272 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 }, 273 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 }, 274 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 }, 275 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 }, 276 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 }, 277 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 278 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 279 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 280 { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 }, 281 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 }, 282 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 }, 283 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 }, 284 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 285 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 286 { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 }, 287 { NULL, 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 /* 310 * All keywords are written to the symbol table. This saves us looking 311 * in a extra table for each name we found. 312 */ 313 void 314 initscan(void) 315 { 316 struct kwtab *kw; 317 sym_t *sym; 318 size_t h, i; 319 uint64_t uq; 320 321 for (kw = kwtab; kw->kw_name != NULL; kw++) { 322 if ((kw->kw_c89 || kw->kw_c99) && tflag) 323 continue; 324 if (kw->kw_c99 && !(Sflag || gflag)) 325 continue; 326 if (kw->kw_gcc && !gflag) 327 continue; 328 sym = getblk(sizeof (sym_t)); 329 sym->s_name = kw->kw_name; 330 sym->s_keyw = kw; 331 sym->s_value.v_quad = kw->kw_token; 332 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 333 sym->s_tspec = kw->kw_tspec; 334 } else if (kw->kw_token == T_SCLASS) { 335 sym->s_scl = kw->kw_scl; 336 } else if (kw->kw_token == T_QUAL) { 337 sym->s_tqual = kw->kw_tqual; 338 } 339 h = hash(sym->s_name); 340 if ((sym->s_link = symtab[h]) != NULL) 341 symtab[h]->s_rlink = &sym->s_link; 342 sym->s_rlink = &symtab[h]; 343 symtab[h] = sym; 344 } 345 346 /* initialize bit-masks for quads */ 347 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 348 qbmasks[i] = (uint64_t)1 << i; 349 uq = ~(uint64_t)0 << i; 350 qumasks[i] = uq; 351 qlmasks[i] = ~uq; 352 } 353 qumasks[i] = 0; 354 qlmasks[i] = ~(uint64_t)0; 355 } 356 357 /* 358 * Get a free sbuf structure, if possible from the free list 359 */ 360 static sbuf_t * 361 allocsb(void) 362 { 363 sbuf_t *sb; 364 365 if ((sb = sbfrlst) != NULL) { 366 sbfrlst = sb->sb_nxt; 367 #ifdef BLKDEBUG 368 (void)memset(sb, 0, sizeof (*sb)); 369 #else 370 sb->sb_nxt = NULL; 371 #endif 372 } else { 373 sb = xmalloc(sizeof (sbuf_t)); 374 (void)memset(sb, 0, sizeof (*sb)); 375 } 376 return (sb); 377 } 378 379 /* 380 * Put a sbuf structure to the free list 381 */ 382 static void 383 freesb(sbuf_t *sb) 384 { 385 386 (void)memset(sb, ZERO, sizeof (*sb)); 387 sb->sb_nxt = sbfrlst; 388 sbfrlst = sb; 389 } 390 391 /* 392 * Read a character and ensure that it is positive (except EOF). 393 * Increment line count(s) if necessary. 394 */ 395 static int 396 inpc(void) 397 { 398 int c; 399 400 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 401 incline(); 402 return (c); 403 } 404 405 static int 406 hash(const char *s) 407 { 408 u_int v; 409 const u_char *us; 410 411 v = 0; 412 for (us = (const u_char *)s; *us != '\0'; us++) { 413 v = (v << sizeof (v)) + *us; 414 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 415 } 416 return (v % HSHSIZ1); 417 } 418 419 /* 420 * Lex has found a letter followed by zero or more letters or digits. 421 * It looks for a symbol in the symbol table with the same name. This 422 * symbol must either be a keyword or a symbol of the type required by 423 * symtyp (label, member, tag, ...). 424 * 425 * If it is a keyword, the token is returned. In some cases it is described 426 * more deeply by data written to yylval. 427 * 428 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 429 * is stored in yylval. This struct contains the name of the symbol, it's 430 * length and hash value. If there is already a symbol of the same name 431 * and type in the symbol table, the sbuf struct also contains a pointer 432 * to the symbol table entry. 433 */ 434 static int 435 name(void) 436 { 437 char *s; 438 sbuf_t *sb; 439 sym_t *sym; 440 int tok; 441 442 sb = allocsb(); 443 sb->sb_name = yytext; 444 sb->sb_len = yyleng; 445 sb->sb_hash = hash(yytext); 446 if ((sym = search(sb)) != NULL && sym->s_keyw) { 447 struct kwtab *kw = sym->s_keyw; 448 if (!kw->kw_attr || attron) { 449 freesb(sb); 450 return (keyw(sym)); 451 } 452 } 453 454 sb->sb_sym = sym; 455 456 if (sym != NULL) { 457 if (blklev < sym->s_blklev) 458 LERROR("name()"); 459 sb->sb_name = sym->s_name; 460 sb->sb_len = strlen(sym->s_name); 461 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 462 } else { 463 s = getblk(yyleng + 1); 464 (void)memcpy(s, yytext, yyleng + 1); 465 sb->sb_name = s; 466 sb->sb_len = yyleng; 467 tok = T_NAME; 468 } 469 470 yylval.y_sb = sb; 471 return (tok); 472 } 473 474 static sym_t * 475 search(sbuf_t *sb) 476 { 477 sym_t *sym; 478 479 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 480 if (strcmp(sym->s_name, sb->sb_name) == 0) { 481 if (sym->s_keyw || sym->s_kind == symtyp) 482 return (sym); 483 } 484 } 485 486 return (NULL); 487 } 488 489 static int 490 keyw(sym_t *sym) 491 { 492 int t; 493 494 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 495 yylval.y_scl = sym->s_scl; 496 } else if (t == T_TYPE || t == T_SOU) { 497 yylval.y_tspec = sym->s_tspec; 498 } else if (t == T_QUAL) { 499 yylval.y_tqual = sym->s_tqual; 500 } 501 return (t); 502 } 503 504 /* 505 * Convert a string representing an integer into internal representation. 506 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 507 */ 508 static int 509 icon(int base) 510 { 511 int l_suffix, u_suffix; 512 int len; 513 const char *cp; 514 char c, *eptr; 515 tspec_t typ; 516 uint64_t uq = 0; 517 int ansiu; 518 static tspec_t contypes[2][3] = { 519 { INT, LONG, QUAD }, 520 { UINT, ULONG, UQUAD } 521 }; 522 523 cp = yytext; 524 len = yyleng; 525 526 /* skip 0x */ 527 if (base == 16) { 528 cp += 2; 529 len -= 2; 530 } 531 532 /* read suffixes */ 533 l_suffix = u_suffix = 0; 534 for ( ; ; ) { 535 if ((c = cp[len - 1]) == 'l' || c == 'L') { 536 l_suffix++; 537 } else if (c == 'u' || c == 'U') { 538 u_suffix++; 539 } else { 540 break; 541 } 542 len--; 543 } 544 if (l_suffix > 2 || u_suffix > 1) { 545 /* malformed integer constant */ 546 warning(251); 547 if (l_suffix > 2) 548 l_suffix = 2; 549 if (u_suffix > 1) 550 u_suffix = 1; 551 } 552 if (tflag && u_suffix != 0) { 553 /* suffix U is illegal in traditional C */ 554 warning(97); 555 } 556 typ = contypes[u_suffix][l_suffix]; 557 558 errno = 0; 559 560 uq = strtouq(cp, &eptr, base); 561 if (eptr != cp + len) 562 LERROR("icon()"); 563 if (errno != 0) 564 /* integer constant out of range */ 565 warning(252); 566 567 /* 568 * If the value is too big for the current type, we must choose 569 * another type. 570 */ 571 ansiu = 0; 572 switch (typ) { 573 case INT: 574 if (uq <= TARG_INT_MAX) { 575 /* ok */ 576 } else if (uq <= TARG_UINT_MAX && base != 10) { 577 typ = UINT; 578 } else if (uq <= TARG_LONG_MAX) { 579 typ = LONG; 580 } else { 581 typ = ULONG; 582 if (uq > TARG_ULONG_MAX) { 583 /* integer constant out of range */ 584 warning(252); 585 } 586 } 587 if (typ == UINT || typ == ULONG) { 588 if (tflag) { 589 typ = LONG; 590 } else if (!sflag) { 591 /* 592 * Remember that the constant is unsigned 593 * only in ANSI C 594 */ 595 ansiu = 1; 596 } 597 } 598 break; 599 case UINT: 600 if (uq > TARG_UINT_MAX) { 601 typ = ULONG; 602 if (uq > TARG_ULONG_MAX) { 603 /* integer constant out of range */ 604 warning(252); 605 } 606 } 607 break; 608 case LONG: 609 if (uq > TARG_LONG_MAX && !tflag) { 610 typ = ULONG; 611 if (!sflag) 612 ansiu = 1; 613 if (uq > TARG_ULONG_MAX) { 614 /* integer constant out of range */ 615 warning(252); 616 } 617 } 618 break; 619 case ULONG: 620 if (uq > TARG_ULONG_MAX) { 621 /* integer constant out of range */ 622 warning(252); 623 } 624 break; 625 case QUAD: 626 if (uq > TARG_QUAD_MAX && !tflag) { 627 typ = UQUAD; 628 if (!sflag) 629 ansiu = 1; 630 } 631 break; 632 case UQUAD: 633 if (uq > TARG_UQUAD_MAX) { 634 /* integer constant out of range */ 635 warning(252); 636 } 637 break; 638 /* LINTED206: (enumeration values not handled in switch) */ 639 case STRUCT: 640 case VOID: 641 case LDOUBLE: 642 case FUNC: 643 case ARRAY: 644 case PTR: 645 case ENUM: 646 case UNION: 647 case SIGNED: 648 case NOTSPEC: 649 case DOUBLE: 650 case FLOAT: 651 case USHORT: 652 case SHORT: 653 case UCHAR: 654 case SCHAR: 655 case CHAR: 656 case BOOL: 657 case UNSIGN: 658 case FCOMPLEX: 659 case DCOMPLEX: 660 case LCOMPLEX: 661 case COMPLEX: 662 break; 663 664 case NTSPEC: /* this value unused */ 665 break; 666 } 667 668 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 669 670 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 671 yylval.y_val->v_ansiu = ansiu; 672 yylval.y_val->v_quad = (int64_t)uq; 673 674 return (T_CON); 675 } 676 677 /* 678 * Returns 1 if t is a signed type and the value is negative. 679 * 680 * len is the number of significant bits. If len is -1, len is set 681 * to the width of type t. 682 */ 683 int 684 sign(int64_t q, tspec_t t, int len) 685 { 686 687 if (t == PTR || isutyp(t)) 688 return (0); 689 return (msb(q, t, len)); 690 } 691 692 int 693 msb(int64_t q, tspec_t t, int len) 694 { 695 696 if (len <= 0) 697 len = size(t); 698 return ((q & qbmasks[len - 1]) != 0); 699 } 700 701 /* 702 * Extends the sign of q. 703 */ 704 int64_t 705 xsign(int64_t q, tspec_t t, int len) 706 { 707 708 if (len <= 0) 709 len = size(t); 710 711 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 712 q &= qlmasks[len]; 713 } else { 714 q |= qumasks[len]; 715 } 716 return (q); 717 } 718 719 /* 720 * Convert a string representing a floating point value into its interal 721 * representation. Type and value are returned in yylval. fcon() 722 * (and yylex()) returns T_CON. 723 * XXX Currently it is not possible to convert constants of type 724 * long double which are greater than DBL_MAX. 725 */ 726 static int 727 fcon(void) 728 { 729 const char *cp; 730 int len; 731 tspec_t typ; 732 char c, *eptr; 733 double d; 734 float f = 0; 735 736 cp = yytext; 737 len = yyleng; 738 739 if (cp[len - 1] == 'i') { 740 /* imaginary, do nothing for now */ 741 len--; 742 } 743 if ((c = cp[len - 1]) == 'f' || c == 'F') { 744 typ = FLOAT; 745 len--; 746 } else if (c == 'l' || c == 'L') { 747 typ = LDOUBLE; 748 len--; 749 } else { 750 if (c == 'd' || c == 'D') 751 len--; 752 typ = DOUBLE; 753 } 754 755 if (tflag && typ != DOUBLE) { 756 /* suffixes F and L are illegal in traditional C */ 757 warning(98); 758 } 759 760 errno = 0; 761 d = strtod(cp, &eptr); 762 if (eptr != cp + len) { 763 switch (*eptr) { 764 /* 765 * XXX: non-native non-current strtod() may not handle hex 766 * floats, ignore the rest if we find traces of hex float 767 * syntax... 768 */ 769 case 'p': 770 case 'P': 771 case 'x': 772 case 'X': 773 d = 0; 774 errno = 0; 775 break; 776 default: 777 LERROR("fcon(%s->%s)", cp, eptr); 778 } 779 } 780 if (errno != 0) 781 /* floating-point constant out of range */ 782 warning(248); 783 784 if (typ == FLOAT) { 785 f = (float)d; 786 if (!finite(f)) { 787 /* floating-point constant out of range */ 788 warning(248); 789 f = f > 0 ? FLT_MAX : -FLT_MAX; 790 } 791 } 792 793 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 794 if (typ == FLOAT) { 795 yylval.y_val->v_ldbl = f; 796 } else { 797 yylval.y_val->v_ldbl = d; 798 } 799 800 return (T_CON); 801 } 802 803 static int 804 operator(int t, op_t o) 805 { 806 807 yylval.y_op = o; 808 return (t); 809 } 810 811 /* 812 * Called if lex found a leading \'. 813 */ 814 static int 815 ccon(void) 816 { 817 size_t n; 818 int val, c; 819 char cv; 820 821 n = 0; 822 val = 0; 823 while ((c = getescc('\'')) >= 0) { 824 val = (val << CHAR_BIT) + c; 825 n++; 826 } 827 if (c == -2) { 828 /* unterminated character constant */ 829 error(253); 830 } else { 831 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 832 /* too many characters in character constant */ 833 error(71); 834 } else if (n > 1) { 835 /* multi-character character constant */ 836 warning(294); 837 } else if (n == 0) { 838 /* empty character constant */ 839 error(73); 840 } 841 } 842 if (n == 1) { 843 cv = (char)val; 844 val = cv; 845 } 846 847 yylval.y_val = xcalloc(1, sizeof (val_t)); 848 yylval.y_val->v_tspec = INT; 849 yylval.y_val->v_quad = val; 850 851 return (T_CON); 852 } 853 854 /* 855 * Called if lex found a leading L\' 856 */ 857 static int 858 wccon(void) 859 { 860 static char buf[MB_LEN_MAX + 1]; 861 size_t i; 862 int c; 863 wchar_t wc; 864 865 i = 0; 866 while ((c = getescc('\'')) >= 0) { 867 if (i < MB_CUR_MAX) 868 buf[i] = (char)c; 869 i++; 870 } 871 872 wc = 0; 873 874 if (c == -2) { 875 /* unterminated character constant */ 876 error(253); 877 } else if (c == 0) { 878 /* empty character constant */ 879 error(73); 880 } else { 881 if (i > MB_CUR_MAX) { 882 i = MB_CUR_MAX; 883 /* too many characters in character constant */ 884 error(71); 885 } else { 886 buf[i] = '\0'; 887 (void)mbtowc(NULL, NULL, 0); 888 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 889 /* invalid multibyte character */ 890 error(291); 891 } 892 } 893 894 yylval.y_val = xcalloc(1, sizeof (val_t)); 895 yylval.y_val->v_tspec = WCHAR; 896 yylval.y_val->v_quad = wc; 897 898 return (T_CON); 899 } 900 901 /* 902 * Read a character which is part of a character constant or of a string 903 * and handle escapes. 904 * 905 * The Argument is the character which delimits the character constant or 906 * string. 907 * 908 * Returns -1 if the end of the character constant or string is reached, 909 * -2 if the EOF is reached, and the character otherwise. 910 */ 911 static int 912 getescc(int d) 913 { 914 static int pbc = -1; 915 int n, c, v; 916 917 if (pbc == -1) { 918 c = inpc(); 919 } else { 920 c = pbc; 921 pbc = -1; 922 } 923 if (c == d) 924 return (-1); 925 switch (c) { 926 case '\n': 927 if (tflag) { 928 /* newline in string or char constant */ 929 error(254); 930 return (-2); 931 } 932 return (c); 933 case EOF: 934 return (-2); 935 case '\\': 936 switch (c = inpc()) { 937 case '"': 938 if (tflag && d == '\'') 939 /* \" inside character constant undef. ... */ 940 warning(262); 941 return ('"'); 942 case '\'': 943 return ('\''); 944 case '?': 945 if (tflag) 946 /* \? undefined in traditional C */ 947 warning(263); 948 return ('?'); 949 case '\\': 950 return ('\\'); 951 case 'a': 952 if (tflag) 953 /* \a undefined in traditional C */ 954 warning(81); 955 return ('\a'); 956 case 'b': 957 return ('\b'); 958 case 'f': 959 return ('\f'); 960 case 'n': 961 return ('\n'); 962 case 'r': 963 return ('\r'); 964 case 't': 965 return ('\t'); 966 case 'v': 967 if (tflag) 968 /* \v undefined in traditional C */ 969 warning(264); 970 return ('\v'); 971 case '8': case '9': 972 /* bad octal digit %c */ 973 warning(77, c); 974 /* FALLTHROUGH */ 975 case '0': case '1': case '2': case '3': 976 case '4': case '5': case '6': case '7': 977 n = 3; 978 v = 0; 979 do { 980 v = (v << 3) + (c - '0'); 981 c = inpc(); 982 } while (--n && isdigit(c) && (tflag || c <= '7')); 983 if (tflag && n > 0 && isdigit(c)) 984 /* bad octal digit %c */ 985 warning(77, c); 986 pbc = c; 987 if (v > TARG_UCHAR_MAX) { 988 /* character escape does not fit in char. */ 989 warning(76); 990 v &= CHAR_MASK; 991 } 992 return (v); 993 case 'x': 994 if (tflag) 995 /* \x undefined in traditional C */ 996 warning(82); 997 v = 0; 998 n = 0; 999 while ((c = inpc()) >= 0 && isxdigit(c)) { 1000 c = isdigit(c) ? 1001 c - '0' : toupper(c) - 'A' + 10; 1002 v = (v << 4) + c; 1003 if (n >= 0) { 1004 if ((v & ~CHAR_MASK) != 0) { 1005 /* overflow in hex escape */ 1006 warning(75); 1007 n = -1; 1008 } else { 1009 n++; 1010 } 1011 } 1012 } 1013 pbc = c; 1014 if (n == 0) { 1015 /* no hex digits follow \x */ 1016 error(74); 1017 } if (n == -1) { 1018 v &= CHAR_MASK; 1019 } 1020 return (v); 1021 case '\n': 1022 return (getescc(d)); 1023 case EOF: 1024 return (-2); 1025 default: 1026 if (isprint(c)) { 1027 /* dubious escape \%c */ 1028 warning(79, c); 1029 } else { 1030 /* dubious escape \%o */ 1031 warning(80, c); 1032 } 1033 } 1034 } 1035 return (c); 1036 } 1037 1038 /* 1039 * Called for preprocessor directives. Currently implemented are: 1040 * # lineno 1041 * # lineno "filename" 1042 */ 1043 static void 1044 directive(void) 1045 { 1046 const char *cp, *fn; 1047 char c, *eptr; 1048 size_t fnl; 1049 long ln; 1050 static int first = 1; 1051 1052 /* Go to first non-whitespace after # */ 1053 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1054 continue; 1055 1056 if (!isdigit((unsigned char)c)) { 1057 error: 1058 /* undefined or invalid # directive */ 1059 warning(255); 1060 return; 1061 } 1062 ln = strtol(--cp, &eptr, 10); 1063 if (cp == eptr) 1064 goto error; 1065 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1066 goto error; 1067 while ((c = *cp++) == ' ' || c == '\t') 1068 continue; 1069 if (c != '\0') { 1070 if (c != '"') 1071 goto error; 1072 fn = cp; 1073 while ((c = *cp) != '"' && c != '\0') 1074 cp++; 1075 if (c != '"') 1076 goto error; 1077 if ((fnl = cp++ - fn) > PATH_MAX) 1078 goto error; 1079 while ((c = *cp++) == ' ' || c == '\t') 1080 continue; 1081 #if 0 1082 if (c != '\0') 1083 warning("extra character(s) after directive"); 1084 #endif 1085 1086 /* empty string means stdin */ 1087 if (fnl == 0) { 1088 fn = "{standard input}"; 1089 fnl = 16; /* strlen (fn) */ 1090 } 1091 curr_pos.p_file = fnnalloc(fn, fnl); 1092 /* 1093 * If this is the first directive, the name is the name 1094 * of the C source file as specified at the command line. 1095 * It is written to the output file. 1096 */ 1097 if (first) { 1098 csrc_pos.p_file = curr_pos.p_file; 1099 outsrc(curr_pos.p_file); 1100 first = 0; 1101 } 1102 } 1103 curr_pos.p_line = (int)ln - 1; 1104 curr_pos.p_uniq = 0; 1105 if (curr_pos.p_file == csrc_pos.p_file) { 1106 csrc_pos.p_line = (int)ln - 1; 1107 csrc_pos.p_uniq = 0; 1108 } 1109 } 1110 1111 /* 1112 * Handle lint comments. Following comments are currently understood: 1113 * ARGSUSEDn 1114 * BITFIELDTYPE 1115 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1116 * FALLTHRU FALLTHROUGH 1117 * LINTLIBRARY 1118 * LINTEDn NOSTRICTn 1119 * LONGLONG 1120 * NOTREACHED 1121 * PRINTFLIKEn 1122 * PROTOLIB 1123 * SCANFLIKEn 1124 * VARARGSn 1125 * If one of this comments is recognized, the arguments, if any, are 1126 * parsed and a function which handles this comment is called. 1127 */ 1128 static void 1129 comment(void) 1130 { 1131 int c, lc; 1132 static struct { 1133 const char *keywd; 1134 int arg; 1135 void (*func)(int); 1136 } keywtab[] = { 1137 { "ARGSUSED", 1, argsused }, 1138 { "BITFIELDTYPE", 0, bitfieldtype }, 1139 { "CONSTCOND", 0, constcond }, 1140 { "CONSTANTCOND", 0, constcond }, 1141 { "CONSTANTCONDITION", 0, constcond }, 1142 { "FALLTHRU", 0, fallthru }, 1143 { "FALLTHROUGH", 0, fallthru }, 1144 { "LINTLIBRARY", 0, lintlib }, 1145 { "LINTED", 1, linted }, 1146 { "LONGLONG", 0, longlong }, 1147 { "NOSTRICT", 1, linted }, 1148 { "NOTREACHED", 0, notreach }, 1149 { "PRINTFLIKE", 1, printflike }, 1150 { "PROTOLIB", 1, protolib }, 1151 { "SCANFLIKE", 1, scanflike }, 1152 { "VARARGS", 1, varargs }, 1153 }; 1154 char keywd[32]; 1155 char arg[32]; 1156 size_t l, i; 1157 int a; 1158 int eoc; 1159 1160 eoc = 0; 1161 1162 /* Skip white spaces after the start of the comment */ 1163 while ((c = inpc()) != EOF && isspace(c)) 1164 continue; 1165 1166 /* Read the potential keyword to keywd */ 1167 l = 0; 1168 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1169 keywd[l++] = (char)c; 1170 c = inpc(); 1171 } 1172 keywd[l] = '\0'; 1173 1174 /* look for the keyword */ 1175 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1176 if (strcmp(keywtab[i].keywd, keywd) == 0) 1177 break; 1178 } 1179 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1180 goto skip_rest; 1181 1182 /* skip white spaces after the keyword */ 1183 while (c != EOF && isspace(c)) 1184 c = inpc(); 1185 1186 /* read the argument, if the keyword accepts one and there is one */ 1187 l = 0; 1188 if (keywtab[i].arg) { 1189 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1190 arg[l++] = (char)c; 1191 c = inpc(); 1192 } 1193 } 1194 arg[l] = '\0'; 1195 a = l != 0 ? atoi(arg) : -1; 1196 1197 /* skip white spaces after the argument */ 1198 while (c != EOF && isspace(c)) 1199 c = inpc(); 1200 1201 if (c != '*' || (c = inpc()) != '/') { 1202 if (keywtab[i].func != linted) 1203 /* extra characters in lint comment */ 1204 warning(257); 1205 } else { 1206 /* 1207 * remember that we have already found the end of the 1208 * comment 1209 */ 1210 eoc = 1; 1211 } 1212 1213 if (keywtab[i].func != NULL) 1214 (*keywtab[i].func)(a); 1215 1216 skip_rest: 1217 while (!eoc) { 1218 lc = c; 1219 if ((c = inpc()) == EOF) { 1220 /* unterminated comment */ 1221 error(256); 1222 break; 1223 } 1224 if (lc == '*' && c == '/') 1225 eoc = 1; 1226 } 1227 } 1228 1229 /* 1230 * Handle // style comments 1231 */ 1232 static void 1233 slashslashcomment(void) 1234 { 1235 int c; 1236 1237 if (!Sflag && !gflag) 1238 /* // comments only supported in C99 */ 1239 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1240 1241 while ((c = inpc()) != EOF && c != '\n') 1242 continue; 1243 } 1244 1245 /* 1246 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1247 * clrwflgs() is called after function definitions and global and 1248 * local declarations and definitions. It is also called between 1249 * the controlling expression and the body of control statements 1250 * (if, switch, for, while). 1251 */ 1252 void 1253 clrwflgs(void) 1254 { 1255 1256 lwarn = LWARN_ALL; 1257 quadflg = 0; 1258 ccflg = 0; 1259 } 1260 1261 /* 1262 * Strings are stored in a dynamically alloceted buffer and passed 1263 * in yylval.y_xstrg to the parser. The parser or the routines called 1264 * by the parser are responsible for freeing this buffer. 1265 */ 1266 static int 1267 string(void) 1268 { 1269 u_char *s; 1270 int c; 1271 size_t len, max; 1272 strg_t *strg; 1273 1274 s = xmalloc(max = 64); 1275 1276 len = 0; 1277 while ((c = getescc('"')) >= 0) { 1278 /* +1 to reserve space for a trailing NUL character */ 1279 if (len + 1 == max) 1280 s = xrealloc(s, max *= 2); 1281 s[len++] = (char)c; 1282 } 1283 s[len] = '\0'; 1284 if (c == -2) 1285 /* unterminated string constant */ 1286 error(258); 1287 1288 strg = xcalloc(1, sizeof (strg_t)); 1289 strg->st_tspec = CHAR; 1290 strg->st_len = len; 1291 strg->st_cp = s; 1292 1293 yylval.y_strg = strg; 1294 return (T_STRING); 1295 } 1296 1297 static int 1298 wcstrg(void) 1299 { 1300 char *s; 1301 int c, n; 1302 size_t i, wi; 1303 size_t len, max, wlen; 1304 wchar_t *ws; 1305 strg_t *strg; 1306 1307 s = xmalloc(max = 64); 1308 len = 0; 1309 while ((c = getescc('"')) >= 0) { 1310 /* +1 to save space for a trailing NUL character */ 1311 if (len + 1 >= max) 1312 s = xrealloc(s, max *= 2); 1313 s[len++] = (char)c; 1314 } 1315 s[len] = '\0'; 1316 if (c == -2) 1317 /* unterminated string constant */ 1318 error(258); 1319 1320 /* get length of wide-character string */ 1321 (void)mblen(NULL, 0); 1322 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1323 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1324 /* invalid multibyte character */ 1325 error(291); 1326 break; 1327 } 1328 if (n == 0) 1329 n = 1; 1330 } 1331 1332 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1333 1334 /* convert from multibyte to wide char */ 1335 (void)mbtowc(NULL, NULL, 0); 1336 for (i = 0, wi = 0; i < len; i += n, wi++) { 1337 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1338 break; 1339 if (n == 0) 1340 n = 1; 1341 } 1342 ws[wi] = 0; 1343 free(s); 1344 1345 strg = xcalloc(1, sizeof (strg_t)); 1346 strg->st_tspec = WCHAR; 1347 strg->st_len = wlen; 1348 strg->st_wcp = ws; 1349 1350 yylval.y_strg = strg; 1351 return (T_STRING); 1352 } 1353 1354 /* 1355 * As noted above the scanner does not create new symbol table entries 1356 * for symbols it cannot find in the symbol table. This is to avoid 1357 * putting undeclared symbols into the symbol table if a syntax error 1358 * occurs. 1359 * 1360 * getsym() is called as soon as it is probably ok to put the symbol to 1361 * the symbol table. This does not mean that it is not possible that 1362 * symbols are put to the symbol table which are than not completely 1363 * declared due to syntax errors. To avoid too many problems in this 1364 * case symbols get type int in getsym(). 1365 * 1366 * XXX calls to getsym() should be delayed until decl1*() is called 1367 */ 1368 sym_t * 1369 getsym(sbuf_t *sb) 1370 { 1371 dinfo_t *di; 1372 char *s; 1373 sym_t *sym; 1374 1375 sym = sb->sb_sym; 1376 1377 /* 1378 * During member declaration it is possible that name() looked 1379 * for symbols of type FVFT, although it should have looked for 1380 * symbols of type FTAG. Same can happen for labels. Both cases 1381 * are compensated here. 1382 */ 1383 if (symtyp == FMOS || symtyp == FLAB) { 1384 if (sym == NULL || sym->s_kind == FVFT) 1385 sym = search(sb); 1386 } 1387 1388 if (sym != NULL) { 1389 if (sym->s_kind != symtyp) 1390 LERROR("storesym()"); 1391 symtyp = FVFT; 1392 freesb(sb); 1393 return (sym); 1394 } 1395 1396 /* create a new symbol table entry */ 1397 1398 /* labels must always be allocated at level 1 (outhermost block) */ 1399 if (symtyp == FLAB) { 1400 sym = getlblk(1, sizeof (sym_t)); 1401 s = getlblk(1, sb->sb_len + 1); 1402 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1403 sym->s_name = s; 1404 sym->s_blklev = 1; 1405 di = dcs; 1406 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1407 di = di->d_nxt; 1408 if (di->d_ctx != AUTO) 1409 LERROR("storesym()"); 1410 } else { 1411 sym = getblk(sizeof (sym_t)); 1412 sym->s_name = sb->sb_name; 1413 sym->s_blklev = blklev; 1414 di = dcs; 1415 } 1416 1417 UNIQUE_CURR_POS(sym->s_dpos); 1418 if ((sym->s_kind = symtyp) != FLAB) 1419 sym->s_type = gettyp(INT); 1420 1421 symtyp = FVFT; 1422 1423 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1424 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1425 sym->s_rlink = &symtab[sb->sb_hash]; 1426 symtab[sb->sb_hash] = sym; 1427 1428 *di->d_ldlsym = sym; 1429 di->d_ldlsym = &sym->s_dlnxt; 1430 1431 freesb(sb); 1432 return (sym); 1433 } 1434 1435 /* 1436 * Construct a temporary symbol. The symbol starts with a digit, so that 1437 * it is illegal. 1438 */ 1439 sym_t * 1440 mktempsym(type_t *t) 1441 { 1442 static int n = 0; 1443 int h; 1444 char *s = getlblk(blklev, 64); 1445 sym_t *sym = getblk(sizeof (sym_t)); 1446 1447 (void)snprintf(s, 64, "%.8d_tmp", n++); 1448 h = hash(s); 1449 1450 sym->s_name = s; 1451 sym->s_type = t; 1452 sym->s_blklev = blklev; 1453 sym->s_scl = AUTO; 1454 sym->s_kind = FVFT; 1455 sym->s_used = 1; 1456 sym->s_set = 1; 1457 1458 if ((sym->s_link = symtab[h]) != NULL) 1459 symtab[h]->s_rlink = &sym->s_link; 1460 sym->s_rlink = &symtab[h]; 1461 symtab[h] = sym; 1462 1463 *dcs->d_ldlsym = sym; 1464 dcs->d_ldlsym = &sym->s_dlnxt; 1465 1466 return sym; 1467 } 1468 1469 /* 1470 * Remove a symbol forever from the symbol table. s_blklev 1471 * is set to -1 to avoid that the symbol will later be put 1472 * back to the symbol table. 1473 */ 1474 void 1475 rmsym(sym_t *sym) 1476 { 1477 1478 if ((*sym->s_rlink = sym->s_link) != NULL) 1479 sym->s_link->s_rlink = sym->s_rlink; 1480 sym->s_blklev = -1; 1481 sym->s_link = NULL; 1482 } 1483 1484 /* 1485 * Remove a list of symbols declared at one level from the symbol 1486 * table. 1487 */ 1488 void 1489 rmsyms(sym_t *syms) 1490 { 1491 sym_t *sym; 1492 1493 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1494 if (sym->s_blklev != -1) { 1495 if ((*sym->s_rlink = sym->s_link) != NULL) 1496 sym->s_link->s_rlink = sym->s_rlink; 1497 sym->s_link = NULL; 1498 sym->s_rlink = NULL; 1499 } 1500 } 1501 } 1502 1503 /* 1504 * Put a symbol into the symbol table 1505 */ 1506 void 1507 inssym(int bl, sym_t *sym) 1508 { 1509 int h; 1510 1511 h = hash(sym->s_name); 1512 if ((sym->s_link = symtab[h]) != NULL) 1513 symtab[h]->s_rlink = &sym->s_link; 1514 sym->s_rlink = &symtab[h]; 1515 symtab[h] = sym; 1516 sym->s_blklev = bl; 1517 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1518 LERROR("inssym()"); 1519 } 1520 1521 /* 1522 * Called at level 0 after syntax errors 1523 * Removes all symbols which are not declared at level 0 from the 1524 * symbol table. Also frees all memory which is not associated with 1525 * level 0. 1526 */ 1527 void 1528 cleanup(void) 1529 { 1530 sym_t *sym, *nsym; 1531 int i; 1532 1533 for (i = 0; i < HSHSIZ1; i++) { 1534 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1535 nsym = sym->s_link; 1536 if (sym->s_blklev >= 1) { 1537 if ((*sym->s_rlink = nsym) != NULL) 1538 nsym->s_rlink = sym->s_rlink; 1539 } 1540 } 1541 } 1542 1543 for (i = mblklev; i > 0; i--) 1544 freelblk(i); 1545 } 1546 1547 /* 1548 * Create a new symbol with the name of an existing symbol. 1549 */ 1550 sym_t * 1551 pushdown(sym_t *sym) 1552 { 1553 int h; 1554 sym_t *nsym; 1555 1556 h = hash(sym->s_name); 1557 nsym = getblk(sizeof (sym_t)); 1558 if (sym->s_blklev > blklev) 1559 LERROR("pushdown()"); 1560 nsym->s_name = sym->s_name; 1561 UNIQUE_CURR_POS(nsym->s_dpos); 1562 nsym->s_kind = sym->s_kind; 1563 nsym->s_blklev = blklev; 1564 1565 if ((nsym->s_link = symtab[h]) != NULL) 1566 symtab[h]->s_rlink = &nsym->s_link; 1567 nsym->s_rlink = &symtab[h]; 1568 symtab[h] = nsym; 1569 1570 *dcs->d_ldlsym = nsym; 1571 dcs->d_ldlsym = &nsym->s_dlnxt; 1572 1573 return (nsym); 1574 } 1575 1576 /* 1577 * Free any dynamically allocated memory referenced by 1578 * the value stack or yylval. 1579 * The type of information in yylval is described by tok. 1580 */ 1581 void 1582 freeyyv(void *sp, int tok) 1583 { 1584 if (tok == T_NAME || tok == T_TYPENAME) { 1585 sbuf_t *sb = *(sbuf_t **)sp; 1586 freesb(sb); 1587 } else if (tok == T_CON) { 1588 val_t *val = *(val_t **)sp; 1589 free(val); 1590 } else if (tok == T_STRING) { 1591 strg_t *strg = *(strg_t **)sp; 1592 if (strg->st_tspec == CHAR) { 1593 free(strg->st_cp); 1594 } else if (strg->st_tspec == WCHAR) { 1595 free(strg->st_wcp); 1596 } else { 1597 LERROR("fryylv()"); 1598 } 1599 free(strg); 1600 } 1601 } 1602
Indexes created Thu Oct 02 01:09:59 GMT 2025