scan.l revision 1.78
1 %{ 2 /* $NetBSD: scan.l,v 1.78 2017/01/15 21:10:24 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 1.78 2017/01/15 21:10:24 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 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 }, 256 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 }, 257 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 }, 258 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 }, 259 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 }, 260 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 }, 261 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 }, 262 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 }, 263 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 }, 264 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 }, 265 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 }, 266 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 }, 267 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 }, 268 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 }, 269 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 }, 270 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 }, 271 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 }, 272 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 }, 273 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 }, 274 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 }, 275 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 }, 276 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 }, 277 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 }, 278 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 }, 279 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 }, 280 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 }, 281 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 }, 282 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 }, 283 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 }, 284 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 }, 285 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 }, 286 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 }, 287 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 }, 288 { NULL, 0, 0, 0, 0, 0,0,0,0,0 } 289 }; 290 291 /* Symbol table */ 292 static sym_t *symtab[HSHSIZ1]; 293 294 /* bit i of the entry with index i is set */ 295 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 296 297 /* least significant i bits are set in the entry with index i */ 298 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 299 300 /* least significant i bits are not set in the entry with index i */ 301 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 302 303 /* free list for sbuf structures */ 304 static sbuf_t *sbfrlst; 305 306 /* Typ of next expected symbol */ 307 symt_t symtyp; 308 309 310 static void 311 addkw(struct kwtab *kw, int deco) 312 { 313 sym_t *sym; 314 size_t h; 315 char buf[256]; 316 const char *name; 317 318 if (!(kw->kw_deco & deco)) 319 return; 320 321 switch (deco) { 322 case 1: 323 name = kw->kw_name; 324 break; 325 case 2: 326 snprintf(buf, sizeof(buf), "__%s", kw->kw_name); 327 name = strdup(buf); 328 break; 329 case 4: 330 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name); 331 name = strdup(buf); 332 break; 333 default: 334 abort(); 335 } 336 337 if (name == NULL) 338 err(1, "Can't init symbol table"); 339 340 sym = getblk(sizeof (sym_t)); 341 sym->s_name = name; 342 sym->s_keyw = kw; 343 sym->s_value.v_quad = kw->kw_token; 344 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 345 sym->s_tspec = kw->kw_tspec; 346 } else if (kw->kw_token == T_SCLASS) { 347 sym->s_scl = kw->kw_scl; 348 } else if (kw->kw_token == T_QUAL) { 349 sym->s_tqual = kw->kw_tqual; 350 } 351 h = hash(sym->s_name); 352 if ((sym->s_link = symtab[h]) != NULL) 353 symtab[h]->s_rlink = &sym->s_link; 354 sym->s_rlink = &symtab[h]; 355 symtab[h] = sym; 356 } 357 358 /* 359 * All keywords are written to the symbol table. This saves us looking 360 * in a extra table for each name we found. 361 */ 362 void 363 initscan(void) 364 { 365 struct kwtab *kw; 366 size_t i; 367 uint64_t uq; 368 369 for (kw = kwtab; kw->kw_name != NULL; kw++) { 370 if ((kw->kw_c89 || kw->kw_c99) && tflag) 371 continue; 372 if (kw->kw_c99 && !(Sflag || gflag)) 373 continue; 374 if (kw->kw_gcc && !gflag) 375 continue; 376 addkw(kw, 1); 377 addkw(kw, 2); 378 addkw(kw, 4); 379 } 380 381 /* initialize bit-masks for quads */ 382 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 383 qbmasks[i] = (uint64_t)1 << i; 384 uq = ~(uint64_t)0 << i; 385 qumasks[i] = uq; 386 qlmasks[i] = ~uq; 387 } 388 qumasks[i] = 0; 389 qlmasks[i] = ~(uint64_t)0; 390 } 391 392 /* 393 * Get a free sbuf structure, if possible from the free list 394 */ 395 static sbuf_t * 396 allocsb(void) 397 { 398 sbuf_t *sb; 399 400 if ((sb = sbfrlst) != NULL) { 401 sbfrlst = sb->sb_nxt; 402 #ifdef BLKDEBUG 403 (void)memset(sb, 0, sizeof (*sb)); 404 #else 405 sb->sb_nxt = NULL; 406 #endif 407 } else { 408 sb = xmalloc(sizeof (sbuf_t)); 409 (void)memset(sb, 0, sizeof (*sb)); 410 } 411 return (sb); 412 } 413 414 /* 415 * Put a sbuf structure to the free list 416 */ 417 static void 418 freesb(sbuf_t *sb) 419 { 420 421 (void)memset(sb, ZERO, sizeof (*sb)); 422 sb->sb_nxt = sbfrlst; 423 sbfrlst = sb; 424 } 425 426 /* 427 * Read a character and ensure that it is positive (except EOF). 428 * Increment line count(s) if necessary. 429 */ 430 static int 431 inpc(void) 432 { 433 int c; 434 435 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 436 incline(); 437 return (c); 438 } 439 440 static int 441 hash(const char *s) 442 { 443 u_int v; 444 const u_char *us; 445 446 v = 0; 447 for (us = (const u_char *)s; *us != '\0'; us++) { 448 v = (v << sizeof (v)) + *us; 449 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 450 } 451 return (v % HSHSIZ1); 452 } 453 454 /* 455 * Lex has found a letter followed by zero or more letters or digits. 456 * It looks for a symbol in the symbol table with the same name. This 457 * symbol must either be a keyword or a symbol of the type required by 458 * symtyp (label, member, tag, ...). 459 * 460 * If it is a keyword, the token is returned. In some cases it is described 461 * more deeply by data written to yylval. 462 * 463 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 464 * is stored in yylval. This struct contains the name of the symbol, its 465 * length and hash value. If there is already a symbol of the same name 466 * and type in the symbol table, the sbuf struct also contains a pointer 467 * to the symbol table entry. 468 */ 469 static int 470 name(void) 471 { 472 char *s; 473 sbuf_t *sb; 474 sym_t *sym; 475 int tok; 476 477 sb = allocsb(); 478 sb->sb_name = yytext; 479 sb->sb_len = yyleng; 480 sb->sb_hash = hash(yytext); 481 if ((sym = search(sb)) != NULL && sym->s_keyw) { 482 freesb(sb); 483 return (keyw(sym)); 484 } 485 486 sb->sb_sym = sym; 487 488 if (sym != NULL) { 489 if (blklev < sym->s_blklev) 490 LERROR("name()"); 491 sb->sb_name = sym->s_name; 492 sb->sb_len = strlen(sym->s_name); 493 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 494 } else { 495 s = getblk(yyleng + 1); 496 (void)memcpy(s, yytext, yyleng + 1); 497 sb->sb_name = s; 498 sb->sb_len = yyleng; 499 tok = T_NAME; 500 } 501 502 yylval.y_sb = sb; 503 return (tok); 504 } 505 506 static sym_t * 507 search(sbuf_t *sb) 508 { 509 sym_t *sym; 510 511 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 512 if (strcmp(sym->s_name, sb->sb_name) == 0) { 513 if (sym->s_keyw) { 514 struct kwtab *kw = sym->s_keyw; 515 if (!kw->kw_attr || attron) 516 return (sym); 517 } else if (!attron && sym->s_kind == symtyp) 518 return sym; 519 } 520 } 521 522 return (NULL); 523 } 524 525 static int 526 keyw(sym_t *sym) 527 { 528 int t; 529 530 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 531 yylval.y_scl = sym->s_scl; 532 } else if (t == T_TYPE || t == T_SOU) { 533 yylval.y_tspec = sym->s_tspec; 534 } else if (t == T_QUAL) { 535 yylval.y_tqual = sym->s_tqual; 536 } 537 return (t); 538 } 539 540 /* 541 * Convert a string representing an integer into internal representation. 542 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 543 */ 544 static int 545 icon(int base) 546 { 547 int l_suffix, u_suffix; 548 int len; 549 const char *cp; 550 char c, *eptr; 551 tspec_t typ; 552 uint64_t uq = 0; 553 int ansiu; 554 static tspec_t contypes[2][3] = { 555 { INT, LONG, QUAD }, 556 { UINT, ULONG, UQUAD } 557 }; 558 559 cp = yytext; 560 len = yyleng; 561 562 /* skip 0[xX] or 0[bB] */ 563 if (base == 16 || base == 2) { 564 cp += 2; 565 len -= 2; 566 } 567 568 /* read suffixes */ 569 l_suffix = u_suffix = 0; 570 for ( ; ; ) { 571 if ((c = cp[len - 1]) == 'l' || c == 'L') { 572 l_suffix++; 573 } else if (c == 'u' || c == 'U') { 574 u_suffix++; 575 } else { 576 break; 577 } 578 len--; 579 } 580 if (l_suffix > 2 || u_suffix > 1) { 581 /* malformed integer constant */ 582 warning(251); 583 if (l_suffix > 2) 584 l_suffix = 2; 585 if (u_suffix > 1) 586 u_suffix = 1; 587 } 588 if (tflag && u_suffix != 0) { 589 /* suffix U is illegal in traditional C */ 590 warning(97); 591 } 592 typ = contypes[u_suffix][l_suffix]; 593 594 errno = 0; 595 596 uq = strtouq(cp, &eptr, base); 597 if (eptr != cp + len) 598 LERROR("icon()"); 599 if (errno != 0) 600 /* integer constant out of range */ 601 warning(252); 602 603 /* 604 * If the value is too big for the current type, we must choose 605 * another type. 606 */ 607 ansiu = 0; 608 switch (typ) { 609 case INT: 610 if (uq <= TARG_INT_MAX) { 611 /* ok */ 612 } else if (uq <= TARG_UINT_MAX && base != 10) { 613 typ = UINT; 614 } else if (uq <= TARG_LONG_MAX) { 615 typ = LONG; 616 } else { 617 typ = ULONG; 618 if (uq > TARG_ULONG_MAX) { 619 /* integer constant out of range */ 620 warning(252); 621 } 622 } 623 if (typ == UINT || typ == ULONG) { 624 if (tflag) { 625 typ = LONG; 626 } else if (!sflag) { 627 /* 628 * Remember that the constant is unsigned 629 * only in ANSI C 630 */ 631 ansiu = 1; 632 } 633 } 634 break; 635 case UINT: 636 if (uq > TARG_UINT_MAX) { 637 typ = ULONG; 638 if (uq > TARG_ULONG_MAX) { 639 /* integer constant out of range */ 640 warning(252); 641 } 642 } 643 break; 644 case LONG: 645 if (uq > TARG_LONG_MAX && !tflag) { 646 typ = ULONG; 647 if (!sflag) 648 ansiu = 1; 649 if (uq > TARG_ULONG_MAX) { 650 /* integer constant out of range */ 651 warning(252); 652 } 653 } 654 break; 655 case ULONG: 656 if (uq > TARG_ULONG_MAX) { 657 /* integer constant out of range */ 658 warning(252); 659 } 660 break; 661 case QUAD: 662 if (uq > TARG_QUAD_MAX && !tflag) { 663 typ = UQUAD; 664 if (!sflag) 665 ansiu = 1; 666 } 667 break; 668 case UQUAD: 669 if (uq > TARG_UQUAD_MAX) { 670 /* integer constant out of range */ 671 warning(252); 672 } 673 break; 674 /* LINTED206: (enumeration values not handled in switch) */ 675 case STRUCT: 676 case VOID: 677 case LDOUBLE: 678 case FUNC: 679 case ARRAY: 680 case PTR: 681 case ENUM: 682 case UNION: 683 case SIGNED: 684 case NOTSPEC: 685 case DOUBLE: 686 case FLOAT: 687 case USHORT: 688 case SHORT: 689 case UCHAR: 690 case SCHAR: 691 case CHAR: 692 case BOOL: 693 case UNSIGN: 694 case FCOMPLEX: 695 case DCOMPLEX: 696 case LCOMPLEX: 697 case COMPLEX: 698 break; 699 700 case NTSPEC: /* this value unused */ 701 break; 702 } 703 704 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 705 706 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 707 yylval.y_val->v_ansiu = ansiu; 708 yylval.y_val->v_quad = (int64_t)uq; 709 710 return (T_CON); 711 } 712 713 /* 714 * Returns 1 if t is a signed type and the value is negative. 715 * 716 * len is the number of significant bits. If len is -1, len is set 717 * to the width of type t. 718 */ 719 int 720 sign(int64_t q, tspec_t t, int len) 721 { 722 723 if (t == PTR || isutyp(t)) 724 return (0); 725 return (msb(q, t, len)); 726 } 727 728 int 729 msb(int64_t q, tspec_t t, int len) 730 { 731 732 if (len <= 0) 733 len = size(t); 734 return ((q & qbmasks[len - 1]) != 0); 735 } 736 737 /* 738 * Extends the sign of q. 739 */ 740 int64_t 741 xsign(int64_t q, tspec_t t, int len) 742 { 743 744 if (len <= 0) 745 len = size(t); 746 747 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 748 q &= qlmasks[len]; 749 } else { 750 q |= qumasks[len]; 751 } 752 return (q); 753 } 754 755 /* 756 * Convert a string representing a floating point value into its interal 757 * representation. Type and value are returned in yylval. fcon() 758 * (and yylex()) returns T_CON. 759 * XXX Currently it is not possible to convert constants of type 760 * long double which are greater than DBL_MAX. 761 */ 762 static int 763 fcon(void) 764 { 765 const char *cp; 766 int len; 767 tspec_t typ; 768 char c, *eptr; 769 double d; 770 float f = 0; 771 772 cp = yytext; 773 len = yyleng; 774 775 if (cp[len - 1] == 'i') { 776 /* imaginary, do nothing for now */ 777 len--; 778 } 779 if ((c = cp[len - 1]) == 'f' || c == 'F') { 780 typ = FLOAT; 781 len--; 782 } else if (c == 'l' || c == 'L') { 783 typ = LDOUBLE; 784 len--; 785 } else { 786 if (c == 'd' || c == 'D') 787 len--; 788 typ = DOUBLE; 789 } 790 791 if (tflag && typ != DOUBLE) { 792 /* suffixes F and L are illegal in traditional C */ 793 warning(98); 794 } 795 796 errno = 0; 797 d = strtod(cp, &eptr); 798 if (eptr != cp + len) { 799 switch (*eptr) { 800 /* 801 * XXX: non-native non-current strtod() may not handle hex 802 * floats, ignore the rest if we find traces of hex float 803 * syntax... 804 */ 805 case 'p': 806 case 'P': 807 case 'x': 808 case 'X': 809 d = 0; 810 errno = 0; 811 break; 812 default: 813 LERROR("fcon(%s->%s)", cp, eptr); 814 } 815 } 816 if (errno != 0) 817 /* floating-point constant out of range */ 818 warning(248); 819 820 if (typ == FLOAT) { 821 f = (float)d; 822 if (!finite(f)) { 823 /* floating-point constant out of range */ 824 warning(248); 825 f = f > 0 ? FLT_MAX : -FLT_MAX; 826 } 827 } 828 829 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 830 if (typ == FLOAT) { 831 yylval.y_val->v_ldbl = f; 832 } else { 833 yylval.y_val->v_ldbl = d; 834 } 835 836 return (T_CON); 837 } 838 839 static int 840 operator(int t, op_t o) 841 { 842 843 yylval.y_op = o; 844 return (t); 845 } 846 847 /* 848 * Called if lex found a leading \'. 849 */ 850 static int 851 ccon(void) 852 { 853 size_t n; 854 int val, c; 855 char cv; 856 857 n = 0; 858 val = 0; 859 while ((c = getescc('\'')) >= 0) { 860 val = (val << CHAR_BIT) + c; 861 n++; 862 } 863 if (c == -2) { 864 /* unterminated character constant */ 865 error(253); 866 } else { 867 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 868 /* too many characters in character constant */ 869 error(71); 870 } else if (n > 1) { 871 /* multi-character character constant */ 872 warning(294); 873 } else if (n == 0) { 874 /* empty character constant */ 875 error(73); 876 } 877 } 878 if (n == 1) { 879 cv = (char)val; 880 val = cv; 881 } 882 883 yylval.y_val = xcalloc(1, sizeof (val_t)); 884 yylval.y_val->v_tspec = INT; 885 yylval.y_val->v_quad = val; 886 887 return (T_CON); 888 } 889 890 /* 891 * Called if lex found a leading L\' 892 */ 893 static int 894 wccon(void) 895 { 896 static char buf[MB_LEN_MAX + 1]; 897 size_t i; 898 int c; 899 wchar_t wc; 900 901 i = 0; 902 while ((c = getescc('\'')) >= 0) { 903 if (i < MB_CUR_MAX) 904 buf[i] = (char)c; 905 i++; 906 } 907 908 wc = 0; 909 910 if (c == -2) { 911 /* unterminated character constant */ 912 error(253); 913 } else if (c == 0) { 914 /* empty character constant */ 915 error(73); 916 } else { 917 if (i > MB_CUR_MAX) { 918 i = MB_CUR_MAX; 919 /* too many characters in character constant */ 920 error(71); 921 } else { 922 buf[i] = '\0'; 923 (void)mbtowc(NULL, NULL, 0); 924 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 925 /* invalid multibyte character */ 926 error(291); 927 } 928 } 929 930 yylval.y_val = xcalloc(1, sizeof (val_t)); 931 yylval.y_val->v_tspec = WCHAR; 932 yylval.y_val->v_quad = wc; 933 934 return (T_CON); 935 } 936 937 /* 938 * Read a character which is part of a character constant or of a string 939 * and handle escapes. 940 * 941 * The Argument is the character which delimits the character constant or 942 * string. 943 * 944 * Returns -1 if the end of the character constant or string is reached, 945 * -2 if the EOF is reached, and the character otherwise. 946 */ 947 static int 948 getescc(int d) 949 { 950 static int pbc = -1; 951 int n, c, v; 952 953 if (pbc == -1) { 954 c = inpc(); 955 } else { 956 c = pbc; 957 pbc = -1; 958 } 959 if (c == d) 960 return (-1); 961 switch (c) { 962 case '\n': 963 if (tflag) { 964 /* newline in string or char constant */ 965 error(254); 966 return (-2); 967 } 968 return (c); 969 case EOF: 970 return (-2); 971 case '\\': 972 switch (c = inpc()) { 973 case '"': 974 if (tflag && d == '\'') 975 /* \" inside character constant undef. ... */ 976 warning(262); 977 return ('"'); 978 case '\'': 979 return ('\''); 980 case '?': 981 if (tflag) 982 /* \? undefined in traditional C */ 983 warning(263); 984 return ('?'); 985 case '\\': 986 return ('\\'); 987 case 'a': 988 if (tflag) 989 /* \a undefined in traditional C */ 990 warning(81); 991 return ('\a'); 992 case 'b': 993 return ('\b'); 994 case 'f': 995 return ('\f'); 996 case 'n': 997 return ('\n'); 998 case 'r': 999 return ('\r'); 1000 case 't': 1001 return ('\t'); 1002 case 'v': 1003 if (tflag) 1004 /* \v undefined in traditional C */ 1005 warning(264); 1006 return ('\v'); 1007 case '8': case '9': 1008 /* bad octal digit %c */ 1009 warning(77, c); 1010 /* FALLTHROUGH */ 1011 case '0': case '1': case '2': case '3': 1012 case '4': case '5': case '6': case '7': 1013 n = 3; 1014 v = 0; 1015 do { 1016 v = (v << 3) + (c - '0'); 1017 c = inpc(); 1018 } while (--n && isdigit(c) && (tflag || c <= '7')); 1019 if (tflag && n > 0 && isdigit(c)) 1020 /* bad octal digit %c */ 1021 warning(77, c); 1022 pbc = c; 1023 if (v > TARG_UCHAR_MAX) { 1024 /* character escape does not fit in char. */ 1025 warning(76); 1026 v &= CHAR_MASK; 1027 } 1028 return (v); 1029 case 'x': 1030 if (tflag) 1031 /* \x undefined in traditional C */ 1032 warning(82); 1033 v = 0; 1034 n = 0; 1035 while ((c = inpc()) >= 0 && isxdigit(c)) { 1036 c = isdigit(c) ? 1037 c - '0' : toupper(c) - 'A' + 10; 1038 v = (v << 4) + c; 1039 if (n >= 0) { 1040 if ((v & ~CHAR_MASK) != 0) { 1041 /* overflow in hex escape */ 1042 warning(75); 1043 n = -1; 1044 } else { 1045 n++; 1046 } 1047 } 1048 } 1049 pbc = c; 1050 if (n == 0) { 1051 /* no hex digits follow \x */ 1052 error(74); 1053 } if (n == -1) { 1054 v &= CHAR_MASK; 1055 } 1056 return (v); 1057 case '\n': 1058 return (getescc(d)); 1059 case EOF: 1060 return (-2); 1061 default: 1062 if (isprint(c)) { 1063 /* dubious escape \%c */ 1064 warning(79, c); 1065 } else { 1066 /* dubious escape \%o */ 1067 warning(80, c); 1068 } 1069 } 1070 } 1071 return (c); 1072 } 1073 1074 /* 1075 * Called for preprocessor directives. Currently implemented are: 1076 * # lineno 1077 * # lineno "filename" 1078 */ 1079 static void 1080 directive(void) 1081 { 1082 const char *cp, *fn; 1083 char c, *eptr; 1084 size_t fnl; 1085 long ln; 1086 static int first = 1; 1087 1088 /* Go to first non-whitespace after # */ 1089 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1090 continue; 1091 1092 if (!isdigit((unsigned char)c)) { 1093 if (strncmp(cp, "pragma", 6) == 0 1094 && isspace((unsigned char)cp[6])) 1095 return; 1096 error: 1097 /* undefined or invalid # directive */ 1098 warning(255); 1099 return; 1100 } 1101 ln = strtol(--cp, &eptr, 10); 1102 if (cp == eptr) 1103 goto error; 1104 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1105 goto error; 1106 while ((c = *cp++) == ' ' || c == '\t') 1107 continue; 1108 if (c != '\0') { 1109 if (c != '"') 1110 goto error; 1111 fn = cp; 1112 while ((c = *cp) != '"' && c != '\0') 1113 cp++; 1114 if (c != '"') 1115 goto error; 1116 if ((fnl = cp++ - fn) > PATH_MAX) 1117 goto error; 1118 while ((c = *cp++) == ' ' || c == '\t') 1119 continue; 1120 #if 0 1121 if (c != '\0') 1122 warning("extra character(s) after directive"); 1123 #endif 1124 1125 /* empty string means stdin */ 1126 if (fnl == 0) { 1127 fn = "{standard input}"; 1128 fnl = 16; /* strlen (fn) */ 1129 } 1130 curr_pos.p_file = fnnalloc(fn, fnl); 1131 /* 1132 * If this is the first directive, the name is the name 1133 * of the C source file as specified at the command line. 1134 * It is written to the output file. 1135 */ 1136 if (first) { 1137 csrc_pos.p_file = curr_pos.p_file; 1138 outsrc(fnxform(curr_pos.p_file, 1139 strlen(curr_pos.p_file))); 1140 first = 0; 1141 } 1142 } 1143 curr_pos.p_line = (int)ln - 1; 1144 curr_pos.p_uniq = 0; 1145 if (curr_pos.p_file == csrc_pos.p_file) { 1146 csrc_pos.p_line = (int)ln - 1; 1147 csrc_pos.p_uniq = 0; 1148 } 1149 } 1150 1151 /* 1152 * Handle lint comments. Following comments are currently understood: 1153 * ARGSUSEDn 1154 * BITFIELDTYPE 1155 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1156 * FALLTHRU FALLTHROUGH 1157 * LINTLIBRARY 1158 * LINTEDn NOSTRICTn 1159 * LONGLONG 1160 * NOTREACHED 1161 * PRINTFLIKEn 1162 * PROTOLIB 1163 * SCANFLIKEn 1164 * VARARGSn 1165 * If one of this comments is recognized, the arguments, if any, are 1166 * parsed and a function which handles this comment is called. 1167 */ 1168 static void 1169 comment(void) 1170 { 1171 int c, lc; 1172 static struct { 1173 const char *keywd; 1174 int arg; 1175 void (*func)(int); 1176 } keywtab[] = { 1177 { "ARGSUSED", 1, argsused }, 1178 { "BITFIELDTYPE", 0, bitfieldtype }, 1179 { "CONSTCOND", 0, constcond }, 1180 { "CONSTANTCOND", 0, constcond }, 1181 { "CONSTANTCONDITION", 0, constcond }, 1182 { "FALLTHRU", 0, fallthru }, 1183 { "FALLTHROUGH", 0, fallthru }, 1184 { "LINTLIBRARY", 0, lintlib }, 1185 { "LINTED", 1, linted }, 1186 { "LONGLONG", 0, longlong }, 1187 { "NOSTRICT", 1, linted }, 1188 { "NOTREACHED", 0, notreach }, 1189 { "PRINTFLIKE", 1, printflike }, 1190 { "PROTOLIB", 1, protolib }, 1191 { "SCANFLIKE", 1, scanflike }, 1192 { "VARARGS", 1, varargs }, 1193 }; 1194 char keywd[32]; 1195 char arg[32]; 1196 size_t l, i; 1197 int a; 1198 int eoc; 1199 1200 eoc = 0; 1201 1202 /* Skip white spaces after the start of the comment */ 1203 while ((c = inpc()) != EOF && isspace(c)) 1204 continue; 1205 1206 /* Read the potential keyword to keywd */ 1207 l = 0; 1208 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1209 keywd[l++] = (char)c; 1210 c = inpc(); 1211 } 1212 keywd[l] = '\0'; 1213 1214 /* look for the keyword */ 1215 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1216 if (strcmp(keywtab[i].keywd, keywd) == 0) 1217 break; 1218 } 1219 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1220 goto skip_rest; 1221 1222 /* skip white spaces after the keyword */ 1223 while (c != EOF && isspace(c)) 1224 c = inpc(); 1225 1226 /* read the argument, if the keyword accepts one and there is one */ 1227 l = 0; 1228 if (keywtab[i].arg) { 1229 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1230 arg[l++] = (char)c; 1231 c = inpc(); 1232 } 1233 } 1234 arg[l] = '\0'; 1235 a = l != 0 ? atoi(arg) : -1; 1236 1237 /* skip white spaces after the argument */ 1238 while (c != EOF && isspace(c)) 1239 c = inpc(); 1240 1241 if (c != '*' || (c = inpc()) != '/') { 1242 if (keywtab[i].func != linted) 1243 /* extra characters in lint comment */ 1244 warning(257); 1245 } else { 1246 /* 1247 * remember that we have already found the end of the 1248 * comment 1249 */ 1250 eoc = 1; 1251 } 1252 1253 if (keywtab[i].func != NULL) 1254 (*keywtab[i].func)(a); 1255 1256 skip_rest: 1257 while (!eoc) { 1258 lc = c; 1259 if ((c = inpc()) == EOF) { 1260 /* unterminated comment */ 1261 error(256); 1262 break; 1263 } 1264 if (lc == '*' && c == '/') 1265 eoc = 1; 1266 } 1267 } 1268 1269 /* 1270 * Handle // style comments 1271 */ 1272 static void 1273 slashslashcomment(void) 1274 { 1275 int c; 1276 1277 if (!Sflag && !gflag) 1278 /* // comments only supported in C99 */ 1279 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1280 1281 while ((c = inpc()) != EOF && c != '\n') 1282 continue; 1283 } 1284 1285 /* 1286 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1287 * clrwflgs() is called after function definitions and global and 1288 * local declarations and definitions. It is also called between 1289 * the controlling expression and the body of control statements 1290 * (if, switch, for, while). 1291 */ 1292 void 1293 clrwflgs(void) 1294 { 1295 1296 lwarn = LWARN_ALL; 1297 quadflg = 0; 1298 ccflg = 0; 1299 } 1300 1301 /* 1302 * Strings are stored in a dynamically alloceted buffer and passed 1303 * in yylval.y_xstrg to the parser. The parser or the routines called 1304 * by the parser are responsible for freeing this buffer. 1305 */ 1306 static int 1307 string(void) 1308 { 1309 u_char *s; 1310 int c; 1311 size_t len, max; 1312 strg_t *strg; 1313 1314 s = xmalloc(max = 64); 1315 1316 len = 0; 1317 while ((c = getescc('"')) >= 0) { 1318 /* +1 to reserve space for a trailing NUL character */ 1319 if (len + 1 == max) 1320 s = xrealloc(s, max *= 2); 1321 s[len++] = (char)c; 1322 } 1323 s[len] = '\0'; 1324 if (c == -2) 1325 /* unterminated string constant */ 1326 error(258); 1327 1328 strg = xcalloc(1, sizeof (strg_t)); 1329 strg->st_tspec = CHAR; 1330 strg->st_len = len; 1331 strg->st_cp = s; 1332 1333 yylval.y_strg = strg; 1334 return (T_STRING); 1335 } 1336 1337 static int 1338 wcstrg(void) 1339 { 1340 char *s; 1341 int c, n; 1342 size_t i, wi; 1343 size_t len, max, wlen; 1344 wchar_t *ws; 1345 strg_t *strg; 1346 1347 s = xmalloc(max = 64); 1348 len = 0; 1349 while ((c = getescc('"')) >= 0) { 1350 /* +1 to save space for a trailing NUL character */ 1351 if (len + 1 >= max) 1352 s = xrealloc(s, max *= 2); 1353 s[len++] = (char)c; 1354 } 1355 s[len] = '\0'; 1356 if (c == -2) 1357 /* unterminated string constant */ 1358 error(258); 1359 1360 /* get length of wide-character string */ 1361 (void)mblen(NULL, 0); 1362 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1363 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1364 /* invalid multibyte character */ 1365 error(291); 1366 break; 1367 } 1368 if (n == 0) 1369 n = 1; 1370 } 1371 1372 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1373 1374 /* convert from multibyte to wide char */ 1375 (void)mbtowc(NULL, NULL, 0); 1376 for (i = 0, wi = 0; i < len; i += n, wi++) { 1377 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1378 break; 1379 if (n == 0) 1380 n = 1; 1381 } 1382 ws[wi] = 0; 1383 free(s); 1384 1385 strg = xcalloc(1, sizeof (strg_t)); 1386 strg->st_tspec = WCHAR; 1387 strg->st_len = wlen; 1388 strg->st_wcp = ws; 1389 1390 yylval.y_strg = strg; 1391 return (T_STRING); 1392 } 1393 1394 /* 1395 * As noted above the scanner does not create new symbol table entries 1396 * for symbols it cannot find in the symbol table. This is to avoid 1397 * putting undeclared symbols into the symbol table if a syntax error 1398 * occurs. 1399 * 1400 * getsym() is called as soon as it is probably ok to put the symbol to 1401 * the symbol table. This does not mean that it is not possible that 1402 * symbols are put to the symbol table which are than not completely 1403 * declared due to syntax errors. To avoid too many problems in this 1404 * case symbols get type int in getsym(). 1405 * 1406 * XXX calls to getsym() should be delayed until decl1*() is called 1407 */ 1408 sym_t * 1409 getsym(sbuf_t *sb) 1410 { 1411 dinfo_t *di; 1412 char *s; 1413 sym_t *sym; 1414 1415 sym = sb->sb_sym; 1416 1417 /* 1418 * During member declaration it is possible that name() looked 1419 * for symbols of type FVFT, although it should have looked for 1420 * symbols of type FTAG. Same can happen for labels. Both cases 1421 * are compensated here. 1422 */ 1423 if (symtyp == FMOS || symtyp == FLAB) { 1424 if (sym == NULL || sym->s_kind == FVFT) 1425 sym = search(sb); 1426 } 1427 1428 if (sym != NULL) { 1429 if (sym->s_kind != symtyp) 1430 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1431 symtyp = FVFT; 1432 freesb(sb); 1433 return (sym); 1434 } 1435 1436 /* create a new symbol table entry */ 1437 1438 /* labels must always be allocated at level 1 (outhermost block) */ 1439 if (symtyp == FLAB) { 1440 sym = getlblk(1, sizeof (sym_t)); 1441 s = getlblk(1, sb->sb_len + 1); 1442 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1443 sym->s_name = s; 1444 sym->s_blklev = 1; 1445 di = dcs; 1446 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1447 di = di->d_nxt; 1448 if (di->d_ctx != AUTO) 1449 LERROR("storesym()"); 1450 } else { 1451 sym = getblk(sizeof (sym_t)); 1452 sym->s_name = sb->sb_name; 1453 sym->s_blklev = blklev; 1454 di = dcs; 1455 } 1456 1457 UNIQUE_CURR_POS(sym->s_dpos); 1458 if ((sym->s_kind = symtyp) != FLAB) 1459 sym->s_type = gettyp(INT); 1460 1461 symtyp = FVFT; 1462 1463 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1464 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1465 sym->s_rlink = &symtab[sb->sb_hash]; 1466 symtab[sb->sb_hash] = sym; 1467 1468 *di->d_ldlsym = sym; 1469 di->d_ldlsym = &sym->s_dlnxt; 1470 1471 freesb(sb); 1472 return (sym); 1473 } 1474 1475 /* 1476 * Construct a temporary symbol. The symbol starts with a digit, so that 1477 * it is illegal. 1478 */ 1479 sym_t * 1480 mktempsym(type_t *t) 1481 { 1482 static int n = 0; 1483 int h; 1484 char *s = getlblk(blklev, 64); 1485 sym_t *sym = getblk(sizeof (sym_t)); 1486 1487 (void)snprintf(s, 64, "%.8d_tmp", n++); 1488 h = hash(s); 1489 1490 sym->s_name = s; 1491 sym->s_type = t; 1492 sym->s_blklev = blklev; 1493 sym->s_scl = AUTO; 1494 sym->s_kind = FVFT; 1495 sym->s_used = 1; 1496 sym->s_set = 1; 1497 1498 if ((sym->s_link = symtab[h]) != NULL) 1499 symtab[h]->s_rlink = &sym->s_link; 1500 sym->s_rlink = &symtab[h]; 1501 symtab[h] = sym; 1502 1503 *dcs->d_ldlsym = sym; 1504 dcs->d_ldlsym = &sym->s_dlnxt; 1505 1506 return sym; 1507 } 1508 1509 /* 1510 * Remove a symbol forever from the symbol table. s_blklev 1511 * is set to -1 to avoid that the symbol will later be put 1512 * back to the symbol table. 1513 */ 1514 void 1515 rmsym(sym_t *sym) 1516 { 1517 1518 if ((*sym->s_rlink = sym->s_link) != NULL) 1519 sym->s_link->s_rlink = sym->s_rlink; 1520 sym->s_blklev = -1; 1521 sym->s_link = NULL; 1522 } 1523 1524 /* 1525 * Remove a list of symbols declared at one level from the symbol 1526 * table. 1527 */ 1528 void 1529 rmsyms(sym_t *syms) 1530 { 1531 sym_t *sym; 1532 1533 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1534 if (sym->s_blklev != -1) { 1535 if ((*sym->s_rlink = sym->s_link) != NULL) 1536 sym->s_link->s_rlink = sym->s_rlink; 1537 sym->s_link = NULL; 1538 sym->s_rlink = NULL; 1539 } 1540 } 1541 } 1542 1543 /* 1544 * Put a symbol into the symbol table 1545 */ 1546 void 1547 inssym(int bl, sym_t *sym) 1548 { 1549 int h; 1550 1551 h = hash(sym->s_name); 1552 if ((sym->s_link = symtab[h]) != NULL) 1553 symtab[h]->s_rlink = &sym->s_link; 1554 sym->s_rlink = &symtab[h]; 1555 symtab[h] = sym; 1556 sym->s_blklev = bl; 1557 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1558 LERROR("inssym()"); 1559 } 1560 1561 /* 1562 * Called at level 0 after syntax errors 1563 * Removes all symbols which are not declared at level 0 from the 1564 * symbol table. Also frees all memory which is not associated with 1565 * level 0. 1566 */ 1567 void 1568 cleanup(void) 1569 { 1570 sym_t *sym, *nsym; 1571 int i; 1572 1573 for (i = 0; i < HSHSIZ1; i++) { 1574 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1575 nsym = sym->s_link; 1576 if (sym->s_blklev >= 1) { 1577 if ((*sym->s_rlink = nsym) != NULL) 1578 nsym->s_rlink = sym->s_rlink; 1579 } 1580 } 1581 } 1582 1583 for (i = mblklev; i > 0; i--) 1584 freelblk(i); 1585 } 1586 1587 /* 1588 * Create a new symbol with the name of an existing symbol. 1589 */ 1590 sym_t * 1591 pushdown(sym_t *sym) 1592 { 1593 int h; 1594 sym_t *nsym; 1595 1596 h = hash(sym->s_name); 1597 nsym = getblk(sizeof (sym_t)); 1598 if (sym->s_blklev > blklev) 1599 LERROR("pushdown()"); 1600 nsym->s_name = sym->s_name; 1601 UNIQUE_CURR_POS(nsym->s_dpos); 1602 nsym->s_kind = sym->s_kind; 1603 nsym->s_blklev = blklev; 1604 1605 if ((nsym->s_link = symtab[h]) != NULL) 1606 symtab[h]->s_rlink = &nsym->s_link; 1607 nsym->s_rlink = &symtab[h]; 1608 symtab[h] = nsym; 1609 1610 *dcs->d_ldlsym = nsym; 1611 dcs->d_ldlsym = &nsym->s_dlnxt; 1612 1613 return (nsym); 1614 } 1615 1616 /* 1617 * Free any dynamically allocated memory referenced by 1618 * the value stack or yylval. 1619 * The type of information in yylval is described by tok. 1620 */ 1621 void 1622 freeyyv(void *sp, int tok) 1623 { 1624 if (tok == T_NAME || tok == T_TYPENAME) { 1625 sbuf_t *sb = *(sbuf_t **)sp; 1626 freesb(sb); 1627 } else if (tok == T_CON) { 1628 val_t *val = *(val_t **)sp; 1629 free(val); 1630 } else if (tok == T_STRING) { 1631 strg_t *strg = *(strg_t **)sp; 1632 if (strg->st_tspec == CHAR) { 1633 free(strg->st_cp); 1634 } else if (strg->st_tspec == WCHAR) { 1635 free(strg->st_wcp); 1636 } else { 1637 LERROR("fryylv()"); 1638 } 1639 free(strg); 1640 } 1641 } 1642
Indexes created Tue Sep 30 20:09:53 GMT 2025