scan.l revision 1.62
1 %{ 2 /* $NetBSD: scan.l,v 1.62 2015/11/12 17:34:01 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 1.62 2015/11/12 17:34:01 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 } kwtab[] = { 207 { "__alignof__", T_ALIGNOF, 0, 0, 0, 0, 0, 0, 0 }, 208 { "__attribute__",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 }, 209 { "__attribute",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 }, 210 { "__packed__", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 }, 211 { "packed", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 }, 212 { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 213 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 214 { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1, 1 }, 215 { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1, 1 }, 216 { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 217 { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 218 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 219 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 220 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 }, 221 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 222 { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 223 { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 224 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 225 { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 226 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 227 { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 228 { "strftime", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 229 { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 }, 230 { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 231 { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 232 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 233 { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 234 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 235 { "__sentinel__",T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 236 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 237 { "__format_arg__", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 238 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 239 { "__returns_twice__", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 240 { "cold", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 241 { "__cold__", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 242 { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 0 }, 243 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 244 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 245 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 }, 246 { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 }, 247 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 }, 248 { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 }, 249 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 }, 250 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 }, 251 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 }, 252 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 253 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 254 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 }, 255 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 }, 256 { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 }, 257 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 }, 258 { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 }, 259 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 }, 260 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 }, 261 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 }, 262 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 }, 263 { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 }, 264 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 }, 265 { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 }, 266 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 }, 267 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 }, 268 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 269 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 270 { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 }, 271 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 }, 272 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 }, 273 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 }, 274 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 }, 275 { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 276 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 277 { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 }, 278 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 }, 279 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 }, 280 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 }, 281 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 282 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 283 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 }, 284 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 }, 285 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 }, 286 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 }, 287 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 }, 288 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 289 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 290 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 291 { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 }, 292 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 }, 293 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 }, 294 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 }, 295 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 296 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 297 { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 }, 298 { NULL, 0, 0, 0, 0, 0, 0, 0, 0 } 299 }; 300 301 /* Symbol table */ 302 static sym_t *symtab[HSHSIZ1]; 303 304 /* bit i of the entry with index i is set */ 305 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 306 307 /* least significant i bits are set in the entry with index i */ 308 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 309 310 /* least significant i bits are not set in the entry with index i */ 311 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 312 313 /* free list for sbuf structures */ 314 static sbuf_t *sbfrlst; 315 316 /* Typ of next expected symbol */ 317 symt_t symtyp; 318 319 320 /* 321 * All keywords are written to the symbol table. This saves us looking 322 * in a extra table for each name we found. 323 */ 324 void 325 initscan(void) 326 { 327 struct kwtab *kw; 328 sym_t *sym; 329 size_t h, i; 330 uint64_t uq; 331 332 for (kw = kwtab; kw->kw_name != NULL; kw++) { 333 if ((kw->kw_c89 || kw->kw_c99) && tflag) 334 continue; 335 if (kw->kw_c99 && !(Sflag || gflag)) 336 continue; 337 if (kw->kw_gcc && !gflag) 338 continue; 339 sym = getblk(sizeof (sym_t)); 340 sym->s_name = kw->kw_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 /* initialize bit-masks for quads */ 358 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 359 qbmasks[i] = (uint64_t)1 << i; 360 uq = ~(uint64_t)0 << i; 361 qumasks[i] = uq; 362 qlmasks[i] = ~uq; 363 } 364 qumasks[i] = 0; 365 qlmasks[i] = ~(uint64_t)0; 366 } 367 368 /* 369 * Get a free sbuf structure, if possible from the free list 370 */ 371 static sbuf_t * 372 allocsb(void) 373 { 374 sbuf_t *sb; 375 376 if ((sb = sbfrlst) != NULL) { 377 sbfrlst = sb->sb_nxt; 378 #ifdef BLKDEBUG 379 (void)memset(sb, 0, sizeof (*sb)); 380 #else 381 sb->sb_nxt = NULL; 382 #endif 383 } else { 384 sb = xmalloc(sizeof (sbuf_t)); 385 (void)memset(sb, 0, sizeof (*sb)); 386 } 387 return (sb); 388 } 389 390 /* 391 * Put a sbuf structure to the free list 392 */ 393 static void 394 freesb(sbuf_t *sb) 395 { 396 397 (void)memset(sb, ZERO, sizeof (*sb)); 398 sb->sb_nxt = sbfrlst; 399 sbfrlst = sb; 400 } 401 402 /* 403 * Read a character and ensure that it is positive (except EOF). 404 * Increment line count(s) if necessary. 405 */ 406 static int 407 inpc(void) 408 { 409 int c; 410 411 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 412 incline(); 413 return (c); 414 } 415 416 static int 417 hash(const char *s) 418 { 419 u_int v; 420 const u_char *us; 421 422 v = 0; 423 for (us = (const u_char *)s; *us != '\0'; us++) { 424 v = (v << sizeof (v)) + *us; 425 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 426 } 427 return (v % HSHSIZ1); 428 } 429 430 /* 431 * Lex has found a letter followed by zero or more letters or digits. 432 * It looks for a symbol in the symbol table with the same name. This 433 * symbol must either be a keyword or a symbol of the type required by 434 * symtyp (label, member, tag, ...). 435 * 436 * If it is a keyword, the token is returned. In some cases it is described 437 * more deeply by data written to yylval. 438 * 439 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 440 * is stored in yylval. This struct contains the name of the symbol, its 441 * length and hash value. If there is already a symbol of the same name 442 * and type in the symbol table, the sbuf struct also contains a pointer 443 * to the symbol table entry. 444 */ 445 static int 446 name(void) 447 { 448 char *s; 449 sbuf_t *sb; 450 sym_t *sym; 451 int tok; 452 453 sb = allocsb(); 454 sb->sb_name = yytext; 455 sb->sb_len = yyleng; 456 sb->sb_hash = hash(yytext); 457 if ((sym = search(sb)) != NULL && sym->s_keyw) { 458 freesb(sb); 459 return (keyw(sym)); 460 } 461 462 sb->sb_sym = sym; 463 464 if (sym != NULL) { 465 if (blklev < sym->s_blklev) 466 LERROR("name()"); 467 sb->sb_name = sym->s_name; 468 sb->sb_len = strlen(sym->s_name); 469 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 470 } else { 471 s = getblk(yyleng + 1); 472 (void)memcpy(s, yytext, yyleng + 1); 473 sb->sb_name = s; 474 sb->sb_len = yyleng; 475 tok = T_NAME; 476 } 477 478 yylval.y_sb = sb; 479 return (tok); 480 } 481 482 static sym_t * 483 search(sbuf_t *sb) 484 { 485 sym_t *sym; 486 487 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 488 if (strcmp(sym->s_name, sb->sb_name) == 0) { 489 if (sym->s_keyw) { 490 struct kwtab *kw = sym->s_keyw; 491 if (!kw->kw_attr || attron) 492 return (sym); 493 } else if (sym->s_kind == symtyp) 494 return sym; 495 } 496 } 497 498 return (NULL); 499 } 500 501 static int 502 keyw(sym_t *sym) 503 { 504 int t; 505 506 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 507 yylval.y_scl = sym->s_scl; 508 } else if (t == T_TYPE || t == T_SOU) { 509 yylval.y_tspec = sym->s_tspec; 510 } else if (t == T_QUAL) { 511 yylval.y_tqual = sym->s_tqual; 512 } 513 return (t); 514 } 515 516 /* 517 * Convert a string representing an integer into internal representation. 518 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 519 */ 520 static int 521 icon(int base) 522 { 523 int l_suffix, u_suffix; 524 int len; 525 const char *cp; 526 char c, *eptr; 527 tspec_t typ; 528 uint64_t uq = 0; 529 int ansiu; 530 static tspec_t contypes[2][3] = { 531 { INT, LONG, QUAD }, 532 { UINT, ULONG, UQUAD } 533 }; 534 535 cp = yytext; 536 len = yyleng; 537 538 /* skip 0[xX] or 0[bB] */ 539 if (base == 16 || base == 2) { 540 cp += 2; 541 len -= 2; 542 } 543 544 /* read suffixes */ 545 l_suffix = u_suffix = 0; 546 for ( ; ; ) { 547 if ((c = cp[len - 1]) == 'l' || c == 'L') { 548 l_suffix++; 549 } else if (c == 'u' || c == 'U') { 550 u_suffix++; 551 } else { 552 break; 553 } 554 len--; 555 } 556 if (l_suffix > 2 || u_suffix > 1) { 557 /* malformed integer constant */ 558 warning(251); 559 if (l_suffix > 2) 560 l_suffix = 2; 561 if (u_suffix > 1) 562 u_suffix = 1; 563 } 564 if (tflag && u_suffix != 0) { 565 /* suffix U is illegal in traditional C */ 566 warning(97); 567 } 568 typ = contypes[u_suffix][l_suffix]; 569 570 errno = 0; 571 572 uq = strtouq(cp, &eptr, base); 573 if (eptr != cp + len) 574 LERROR("icon()"); 575 if (errno != 0) 576 /* integer constant out of range */ 577 warning(252); 578 579 /* 580 * If the value is too big for the current type, we must choose 581 * another type. 582 */ 583 ansiu = 0; 584 switch (typ) { 585 case INT: 586 if (uq <= TARG_INT_MAX) { 587 /* ok */ 588 } else if (uq <= TARG_UINT_MAX && base != 10) { 589 typ = UINT; 590 } else if (uq <= TARG_LONG_MAX) { 591 typ = LONG; 592 } else { 593 typ = ULONG; 594 if (uq > TARG_ULONG_MAX) { 595 /* integer constant out of range */ 596 warning(252); 597 } 598 } 599 if (typ == UINT || typ == ULONG) { 600 if (tflag) { 601 typ = LONG; 602 } else if (!sflag) { 603 /* 604 * Remember that the constant is unsigned 605 * only in ANSI C 606 */ 607 ansiu = 1; 608 } 609 } 610 break; 611 case UINT: 612 if (uq > TARG_UINT_MAX) { 613 typ = ULONG; 614 if (uq > TARG_ULONG_MAX) { 615 /* integer constant out of range */ 616 warning(252); 617 } 618 } 619 break; 620 case LONG: 621 if (uq > TARG_LONG_MAX && !tflag) { 622 typ = ULONG; 623 if (!sflag) 624 ansiu = 1; 625 if (uq > TARG_ULONG_MAX) { 626 /* integer constant out of range */ 627 warning(252); 628 } 629 } 630 break; 631 case ULONG: 632 if (uq > TARG_ULONG_MAX) { 633 /* integer constant out of range */ 634 warning(252); 635 } 636 break; 637 case QUAD: 638 if (uq > TARG_QUAD_MAX && !tflag) { 639 typ = UQUAD; 640 if (!sflag) 641 ansiu = 1; 642 } 643 break; 644 case UQUAD: 645 if (uq > TARG_UQUAD_MAX) { 646 /* integer constant out of range */ 647 warning(252); 648 } 649 break; 650 /* LINTED206: (enumeration values not handled in switch) */ 651 case STRUCT: 652 case VOID: 653 case LDOUBLE: 654 case FUNC: 655 case ARRAY: 656 case PTR: 657 case ENUM: 658 case UNION: 659 case SIGNED: 660 case NOTSPEC: 661 case DOUBLE: 662 case FLOAT: 663 case USHORT: 664 case SHORT: 665 case UCHAR: 666 case SCHAR: 667 case CHAR: 668 case BOOL: 669 case UNSIGN: 670 case FCOMPLEX: 671 case DCOMPLEX: 672 case LCOMPLEX: 673 case COMPLEX: 674 break; 675 676 case NTSPEC: /* this value unused */ 677 break; 678 } 679 680 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 681 682 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 683 yylval.y_val->v_ansiu = ansiu; 684 yylval.y_val->v_quad = (int64_t)uq; 685 686 return (T_CON); 687 } 688 689 /* 690 * Returns 1 if t is a signed type and the value is negative. 691 * 692 * len is the number of significant bits. If len is -1, len is set 693 * to the width of type t. 694 */ 695 int 696 sign(int64_t q, tspec_t t, int len) 697 { 698 699 if (t == PTR || isutyp(t)) 700 return (0); 701 return (msb(q, t, len)); 702 } 703 704 int 705 msb(int64_t q, tspec_t t, int len) 706 { 707 708 if (len <= 0) 709 len = size(t); 710 return ((q & qbmasks[len - 1]) != 0); 711 } 712 713 /* 714 * Extends the sign of q. 715 */ 716 int64_t 717 xsign(int64_t q, tspec_t t, int len) 718 { 719 720 if (len <= 0) 721 len = size(t); 722 723 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 724 q &= qlmasks[len]; 725 } else { 726 q |= qumasks[len]; 727 } 728 return (q); 729 } 730 731 /* 732 * Convert a string representing a floating point value into its interal 733 * representation. Type and value are returned in yylval. fcon() 734 * (and yylex()) returns T_CON. 735 * XXX Currently it is not possible to convert constants of type 736 * long double which are greater than DBL_MAX. 737 */ 738 static int 739 fcon(void) 740 { 741 const char *cp; 742 int len; 743 tspec_t typ; 744 char c, *eptr; 745 double d; 746 float f = 0; 747 748 cp = yytext; 749 len = yyleng; 750 751 if (cp[len - 1] == 'i') { 752 /* imaginary, do nothing for now */ 753 len--; 754 } 755 if ((c = cp[len - 1]) == 'f' || c == 'F') { 756 typ = FLOAT; 757 len--; 758 } else if (c == 'l' || c == 'L') { 759 typ = LDOUBLE; 760 len--; 761 } else { 762 if (c == 'd' || c == 'D') 763 len--; 764 typ = DOUBLE; 765 } 766 767 if (tflag && typ != DOUBLE) { 768 /* suffixes F and L are illegal in traditional C */ 769 warning(98); 770 } 771 772 errno = 0; 773 d = strtod(cp, &eptr); 774 if (eptr != cp + len) { 775 switch (*eptr) { 776 /* 777 * XXX: non-native non-current strtod() may not handle hex 778 * floats, ignore the rest if we find traces of hex float 779 * syntax... 780 */ 781 case 'p': 782 case 'P': 783 case 'x': 784 case 'X': 785 d = 0; 786 errno = 0; 787 break; 788 default: 789 LERROR("fcon(%s->%s)", cp, eptr); 790 } 791 } 792 if (errno != 0) 793 /* floating-point constant out of range */ 794 warning(248); 795 796 if (typ == FLOAT) { 797 f = (float)d; 798 if (!finite(f)) { 799 /* floating-point constant out of range */ 800 warning(248); 801 f = f > 0 ? FLT_MAX : -FLT_MAX; 802 } 803 } 804 805 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 806 if (typ == FLOAT) { 807 yylval.y_val->v_ldbl = f; 808 } else { 809 yylval.y_val->v_ldbl = d; 810 } 811 812 return (T_CON); 813 } 814 815 static int 816 operator(int t, op_t o) 817 { 818 819 yylval.y_op = o; 820 return (t); 821 } 822 823 /* 824 * Called if lex found a leading \'. 825 */ 826 static int 827 ccon(void) 828 { 829 size_t n; 830 int val, c; 831 char cv; 832 833 n = 0; 834 val = 0; 835 while ((c = getescc('\'')) >= 0) { 836 val = (val << CHAR_BIT) + c; 837 n++; 838 } 839 if (c == -2) { 840 /* unterminated character constant */ 841 error(253); 842 } else { 843 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 844 /* too many characters in character constant */ 845 error(71); 846 } else if (n > 1) { 847 /* multi-character character constant */ 848 warning(294); 849 } else if (n == 0) { 850 /* empty character constant */ 851 error(73); 852 } 853 } 854 if (n == 1) { 855 cv = (char)val; 856 val = cv; 857 } 858 859 yylval.y_val = xcalloc(1, sizeof (val_t)); 860 yylval.y_val->v_tspec = INT; 861 yylval.y_val->v_quad = val; 862 863 return (T_CON); 864 } 865 866 /* 867 * Called if lex found a leading L\' 868 */ 869 static int 870 wccon(void) 871 { 872 static char buf[MB_LEN_MAX + 1]; 873 size_t i; 874 int c; 875 wchar_t wc; 876 877 i = 0; 878 while ((c = getescc('\'')) >= 0) { 879 if (i < MB_CUR_MAX) 880 buf[i] = (char)c; 881 i++; 882 } 883 884 wc = 0; 885 886 if (c == -2) { 887 /* unterminated character constant */ 888 error(253); 889 } else if (c == 0) { 890 /* empty character constant */ 891 error(73); 892 } else { 893 if (i > MB_CUR_MAX) { 894 i = MB_CUR_MAX; 895 /* too many characters in character constant */ 896 error(71); 897 } else { 898 buf[i] = '\0'; 899 (void)mbtowc(NULL, NULL, 0); 900 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 901 /* invalid multibyte character */ 902 error(291); 903 } 904 } 905 906 yylval.y_val = xcalloc(1, sizeof (val_t)); 907 yylval.y_val->v_tspec = WCHAR; 908 yylval.y_val->v_quad = wc; 909 910 return (T_CON); 911 } 912 913 /* 914 * Read a character which is part of a character constant or of a string 915 * and handle escapes. 916 * 917 * The Argument is the character which delimits the character constant or 918 * string. 919 * 920 * Returns -1 if the end of the character constant or string is reached, 921 * -2 if the EOF is reached, and the character otherwise. 922 */ 923 static int 924 getescc(int d) 925 { 926 static int pbc = -1; 927 int n, c, v; 928 929 if (pbc == -1) { 930 c = inpc(); 931 } else { 932 c = pbc; 933 pbc = -1; 934 } 935 if (c == d) 936 return (-1); 937 switch (c) { 938 case '\n': 939 if (tflag) { 940 /* newline in string or char constant */ 941 error(254); 942 return (-2); 943 } 944 return (c); 945 case EOF: 946 return (-2); 947 case '\\': 948 switch (c = inpc()) { 949 case '"': 950 if (tflag && d == '\'') 951 /* \" inside character constant undef. ... */ 952 warning(262); 953 return ('"'); 954 case '\'': 955 return ('\''); 956 case '?': 957 if (tflag) 958 /* \? undefined in traditional C */ 959 warning(263); 960 return ('?'); 961 case '\\': 962 return ('\\'); 963 case 'a': 964 if (tflag) 965 /* \a undefined in traditional C */ 966 warning(81); 967 return ('\a'); 968 case 'b': 969 return ('\b'); 970 case 'f': 971 return ('\f'); 972 case 'n': 973 return ('\n'); 974 case 'r': 975 return ('\r'); 976 case 't': 977 return ('\t'); 978 case 'v': 979 if (tflag) 980 /* \v undefined in traditional C */ 981 warning(264); 982 return ('\v'); 983 case '8': case '9': 984 /* bad octal digit %c */ 985 warning(77, c); 986 /* FALLTHROUGH */ 987 case '0': case '1': case '2': case '3': 988 case '4': case '5': case '6': case '7': 989 n = 3; 990 v = 0; 991 do { 992 v = (v << 3) + (c - '0'); 993 c = inpc(); 994 } while (--n && isdigit(c) && (tflag || c <= '7')); 995 if (tflag && n > 0 && isdigit(c)) 996 /* bad octal digit %c */ 997 warning(77, c); 998 pbc = c; 999 if (v > TARG_UCHAR_MAX) { 1000 /* character escape does not fit in char. */ 1001 warning(76); 1002 v &= CHAR_MASK; 1003 } 1004 return (v); 1005 case 'x': 1006 if (tflag) 1007 /* \x undefined in traditional C */ 1008 warning(82); 1009 v = 0; 1010 n = 0; 1011 while ((c = inpc()) >= 0 && isxdigit(c)) { 1012 c = isdigit(c) ? 1013 c - '0' : toupper(c) - 'A' + 10; 1014 v = (v << 4) + c; 1015 if (n >= 0) { 1016 if ((v & ~CHAR_MASK) != 0) { 1017 /* overflow in hex escape */ 1018 warning(75); 1019 n = -1; 1020 } else { 1021 n++; 1022 } 1023 } 1024 } 1025 pbc = c; 1026 if (n == 0) { 1027 /* no hex digits follow \x */ 1028 error(74); 1029 } if (n == -1) { 1030 v &= CHAR_MASK; 1031 } 1032 return (v); 1033 case '\n': 1034 return (getescc(d)); 1035 case EOF: 1036 return (-2); 1037 default: 1038 if (isprint(c)) { 1039 /* dubious escape \%c */ 1040 warning(79, c); 1041 } else { 1042 /* dubious escape \%o */ 1043 warning(80, c); 1044 } 1045 } 1046 } 1047 return (c); 1048 } 1049 1050 /* 1051 * Called for preprocessor directives. Currently implemented are: 1052 * # lineno 1053 * # lineno "filename" 1054 */ 1055 static void 1056 directive(void) 1057 { 1058 const char *cp, *fn; 1059 char c, *eptr; 1060 size_t fnl; 1061 long ln; 1062 static int first = 1; 1063 1064 /* Go to first non-whitespace after # */ 1065 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1066 continue; 1067 1068 if (!isdigit((unsigned char)c)) { 1069 error: 1070 /* undefined or invalid # directive */ 1071 warning(255); 1072 return; 1073 } 1074 ln = strtol(--cp, &eptr, 10); 1075 if (cp == eptr) 1076 goto error; 1077 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1078 goto error; 1079 while ((c = *cp++) == ' ' || c == '\t') 1080 continue; 1081 if (c != '\0') { 1082 if (c != '"') 1083 goto error; 1084 fn = cp; 1085 while ((c = *cp) != '"' && c != '\0') 1086 cp++; 1087 if (c != '"') 1088 goto error; 1089 if ((fnl = cp++ - fn) > PATH_MAX) 1090 goto error; 1091 while ((c = *cp++) == ' ' || c == '\t') 1092 continue; 1093 #if 0 1094 if (c != '\0') 1095 warning("extra character(s) after directive"); 1096 #endif 1097 1098 /* empty string means stdin */ 1099 if (fnl == 0) { 1100 fn = "{standard input}"; 1101 fnl = 16; /* strlen (fn) */ 1102 } 1103 curr_pos.p_file = fnnalloc(fn, fnl); 1104 /* 1105 * If this is the first directive, the name is the name 1106 * of the C source file as specified at the command line. 1107 * It is written to the output file. 1108 */ 1109 if (first) { 1110 csrc_pos.p_file = curr_pos.p_file; 1111 outsrc(curr_pos.p_file); 1112 first = 0; 1113 } 1114 } 1115 curr_pos.p_line = (int)ln - 1; 1116 curr_pos.p_uniq = 0; 1117 if (curr_pos.p_file == csrc_pos.p_file) { 1118 csrc_pos.p_line = (int)ln - 1; 1119 csrc_pos.p_uniq = 0; 1120 } 1121 } 1122 1123 /* 1124 * Handle lint comments. Following comments are currently understood: 1125 * ARGSUSEDn 1126 * BITFIELDTYPE 1127 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1128 * FALLTHRU FALLTHROUGH 1129 * LINTLIBRARY 1130 * LINTEDn NOSTRICTn 1131 * LONGLONG 1132 * NOTREACHED 1133 * PRINTFLIKEn 1134 * PROTOLIB 1135 * SCANFLIKEn 1136 * VARARGSn 1137 * If one of this comments is recognized, the arguments, if any, are 1138 * parsed and a function which handles this comment is called. 1139 */ 1140 static void 1141 comment(void) 1142 { 1143 int c, lc; 1144 static struct { 1145 const char *keywd; 1146 int arg; 1147 void (*func)(int); 1148 } keywtab[] = { 1149 { "ARGSUSED", 1, argsused }, 1150 { "BITFIELDTYPE", 0, bitfieldtype }, 1151 { "CONSTCOND", 0, constcond }, 1152 { "CONSTANTCOND", 0, constcond }, 1153 { "CONSTANTCONDITION", 0, constcond }, 1154 { "FALLTHRU", 0, fallthru }, 1155 { "FALLTHROUGH", 0, fallthru }, 1156 { "LINTLIBRARY", 0, lintlib }, 1157 { "LINTED", 1, linted }, 1158 { "LONGLONG", 0, longlong }, 1159 { "NOSTRICT", 1, linted }, 1160 { "NOTREACHED", 0, notreach }, 1161 { "PRINTFLIKE", 1, printflike }, 1162 { "PROTOLIB", 1, protolib }, 1163 { "SCANFLIKE", 1, scanflike }, 1164 { "VARARGS", 1, varargs }, 1165 }; 1166 char keywd[32]; 1167 char arg[32]; 1168 size_t l, i; 1169 int a; 1170 int eoc; 1171 1172 eoc = 0; 1173 1174 /* Skip white spaces after the start of the comment */ 1175 while ((c = inpc()) != EOF && isspace(c)) 1176 continue; 1177 1178 /* Read the potential keyword to keywd */ 1179 l = 0; 1180 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1181 keywd[l++] = (char)c; 1182 c = inpc(); 1183 } 1184 keywd[l] = '\0'; 1185 1186 /* look for the keyword */ 1187 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1188 if (strcmp(keywtab[i].keywd, keywd) == 0) 1189 break; 1190 } 1191 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1192 goto skip_rest; 1193 1194 /* skip white spaces after the keyword */ 1195 while (c != EOF && isspace(c)) 1196 c = inpc(); 1197 1198 /* read the argument, if the keyword accepts one and there is one */ 1199 l = 0; 1200 if (keywtab[i].arg) { 1201 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1202 arg[l++] = (char)c; 1203 c = inpc(); 1204 } 1205 } 1206 arg[l] = '\0'; 1207 a = l != 0 ? atoi(arg) : -1; 1208 1209 /* skip white spaces after the argument */ 1210 while (c != EOF && isspace(c)) 1211 c = inpc(); 1212 1213 if (c != '*' || (c = inpc()) != '/') { 1214 if (keywtab[i].func != linted) 1215 /* extra characters in lint comment */ 1216 warning(257); 1217 } else { 1218 /* 1219 * remember that we have already found the end of the 1220 * comment 1221 */ 1222 eoc = 1; 1223 } 1224 1225 if (keywtab[i].func != NULL) 1226 (*keywtab[i].func)(a); 1227 1228 skip_rest: 1229 while (!eoc) { 1230 lc = c; 1231 if ((c = inpc()) == EOF) { 1232 /* unterminated comment */ 1233 error(256); 1234 break; 1235 } 1236 if (lc == '*' && c == '/') 1237 eoc = 1; 1238 } 1239 } 1240 1241 /* 1242 * Handle // style comments 1243 */ 1244 static void 1245 slashslashcomment(void) 1246 { 1247 int c; 1248 1249 if (!Sflag && !gflag) 1250 /* // comments only supported in C99 */ 1251 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1252 1253 while ((c = inpc()) != EOF && c != '\n') 1254 continue; 1255 } 1256 1257 /* 1258 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1259 * clrwflgs() is called after function definitions and global and 1260 * local declarations and definitions. It is also called between 1261 * the controlling expression and the body of control statements 1262 * (if, switch, for, while). 1263 */ 1264 void 1265 clrwflgs(void) 1266 { 1267 1268 lwarn = LWARN_ALL; 1269 quadflg = 0; 1270 ccflg = 0; 1271 } 1272 1273 /* 1274 * Strings are stored in a dynamically alloceted buffer and passed 1275 * in yylval.y_xstrg to the parser. The parser or the routines called 1276 * by the parser are responsible for freeing this buffer. 1277 */ 1278 static int 1279 string(void) 1280 { 1281 u_char *s; 1282 int c; 1283 size_t len, max; 1284 strg_t *strg; 1285 1286 s = xmalloc(max = 64); 1287 1288 len = 0; 1289 while ((c = getescc('"')) >= 0) { 1290 /* +1 to reserve space for a trailing NUL character */ 1291 if (len + 1 == max) 1292 s = xrealloc(s, max *= 2); 1293 s[len++] = (char)c; 1294 } 1295 s[len] = '\0'; 1296 if (c == -2) 1297 /* unterminated string constant */ 1298 error(258); 1299 1300 strg = xcalloc(1, sizeof (strg_t)); 1301 strg->st_tspec = CHAR; 1302 strg->st_len = len; 1303 strg->st_cp = s; 1304 1305 yylval.y_strg = strg; 1306 return (T_STRING); 1307 } 1308 1309 static int 1310 wcstrg(void) 1311 { 1312 char *s; 1313 int c, n; 1314 size_t i, wi; 1315 size_t len, max, wlen; 1316 wchar_t *ws; 1317 strg_t *strg; 1318 1319 s = xmalloc(max = 64); 1320 len = 0; 1321 while ((c = getescc('"')) >= 0) { 1322 /* +1 to save space for a trailing NUL character */ 1323 if (len + 1 >= max) 1324 s = xrealloc(s, max *= 2); 1325 s[len++] = (char)c; 1326 } 1327 s[len] = '\0'; 1328 if (c == -2) 1329 /* unterminated string constant */ 1330 error(258); 1331 1332 /* get length of wide-character string */ 1333 (void)mblen(NULL, 0); 1334 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1335 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1336 /* invalid multibyte character */ 1337 error(291); 1338 break; 1339 } 1340 if (n == 0) 1341 n = 1; 1342 } 1343 1344 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1345 1346 /* convert from multibyte to wide char */ 1347 (void)mbtowc(NULL, NULL, 0); 1348 for (i = 0, wi = 0; i < len; i += n, wi++) { 1349 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1350 break; 1351 if (n == 0) 1352 n = 1; 1353 } 1354 ws[wi] = 0; 1355 free(s); 1356 1357 strg = xcalloc(1, sizeof (strg_t)); 1358 strg->st_tspec = WCHAR; 1359 strg->st_len = wlen; 1360 strg->st_wcp = ws; 1361 1362 yylval.y_strg = strg; 1363 return (T_STRING); 1364 } 1365 1366 /* 1367 * As noted above the scanner does not create new symbol table entries 1368 * for symbols it cannot find in the symbol table. This is to avoid 1369 * putting undeclared symbols into the symbol table if a syntax error 1370 * occurs. 1371 * 1372 * getsym() is called as soon as it is probably ok to put the symbol to 1373 * the symbol table. This does not mean that it is not possible that 1374 * symbols are put to the symbol table which are than not completely 1375 * declared due to syntax errors. To avoid too many problems in this 1376 * case symbols get type int in getsym(). 1377 * 1378 * XXX calls to getsym() should be delayed until decl1*() is called 1379 */ 1380 sym_t * 1381 getsym(sbuf_t *sb) 1382 { 1383 dinfo_t *di; 1384 char *s; 1385 sym_t *sym; 1386 1387 sym = sb->sb_sym; 1388 1389 /* 1390 * During member declaration it is possible that name() looked 1391 * for symbols of type FVFT, although it should have looked for 1392 * symbols of type FTAG. Same can happen for labels. Both cases 1393 * are compensated here. 1394 */ 1395 if (symtyp == FMOS || symtyp == FLAB) { 1396 if (sym == NULL || sym->s_kind == FVFT) 1397 sym = search(sb); 1398 } 1399 1400 if (sym != NULL) { 1401 if (sym->s_kind != symtyp) 1402 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1403 symtyp = FVFT; 1404 freesb(sb); 1405 return (sym); 1406 } 1407 1408 /* create a new symbol table entry */ 1409 1410 /* labels must always be allocated at level 1 (outhermost block) */ 1411 if (symtyp == FLAB) { 1412 sym = getlblk(1, sizeof (sym_t)); 1413 s = getlblk(1, sb->sb_len + 1); 1414 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1415 sym->s_name = s; 1416 sym->s_blklev = 1; 1417 di = dcs; 1418 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1419 di = di->d_nxt; 1420 if (di->d_ctx != AUTO) 1421 LERROR("storesym()"); 1422 } else { 1423 sym = getblk(sizeof (sym_t)); 1424 sym->s_name = sb->sb_name; 1425 sym->s_blklev = blklev; 1426 di = dcs; 1427 } 1428 1429 UNIQUE_CURR_POS(sym->s_dpos); 1430 if ((sym->s_kind = symtyp) != FLAB) 1431 sym->s_type = gettyp(INT); 1432 1433 symtyp = FVFT; 1434 1435 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1436 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1437 sym->s_rlink = &symtab[sb->sb_hash]; 1438 symtab[sb->sb_hash] = sym; 1439 1440 *di->d_ldlsym = sym; 1441 di->d_ldlsym = &sym->s_dlnxt; 1442 1443 freesb(sb); 1444 return (sym); 1445 } 1446 1447 /* 1448 * Construct a temporary symbol. The symbol starts with a digit, so that 1449 * it is illegal. 1450 */ 1451 sym_t * 1452 mktempsym(type_t *t) 1453 { 1454 static int n = 0; 1455 int h; 1456 char *s = getlblk(blklev, 64); 1457 sym_t *sym = getblk(sizeof (sym_t)); 1458 1459 (void)snprintf(s, 64, "%.8d_tmp", n++); 1460 h = hash(s); 1461 1462 sym->s_name = s; 1463 sym->s_type = t; 1464 sym->s_blklev = blklev; 1465 sym->s_scl = AUTO; 1466 sym->s_kind = FVFT; 1467 sym->s_used = 1; 1468 sym->s_set = 1; 1469 1470 if ((sym->s_link = symtab[h]) != NULL) 1471 symtab[h]->s_rlink = &sym->s_link; 1472 sym->s_rlink = &symtab[h]; 1473 symtab[h] = sym; 1474 1475 *dcs->d_ldlsym = sym; 1476 dcs->d_ldlsym = &sym->s_dlnxt; 1477 1478 return sym; 1479 } 1480 1481 /* 1482 * Remove a symbol forever from the symbol table. s_blklev 1483 * is set to -1 to avoid that the symbol will later be put 1484 * back to the symbol table. 1485 */ 1486 void 1487 rmsym(sym_t *sym) 1488 { 1489 1490 if ((*sym->s_rlink = sym->s_link) != NULL) 1491 sym->s_link->s_rlink = sym->s_rlink; 1492 sym->s_blklev = -1; 1493 sym->s_link = NULL; 1494 } 1495 1496 /* 1497 * Remove a list of symbols declared at one level from the symbol 1498 * table. 1499 */ 1500 void 1501 rmsyms(sym_t *syms) 1502 { 1503 sym_t *sym; 1504 1505 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1506 if (sym->s_blklev != -1) { 1507 if ((*sym->s_rlink = sym->s_link) != NULL) 1508 sym->s_link->s_rlink = sym->s_rlink; 1509 sym->s_link = NULL; 1510 sym->s_rlink = NULL; 1511 } 1512 } 1513 } 1514 1515 /* 1516 * Put a symbol into the symbol table 1517 */ 1518 void 1519 inssym(int bl, sym_t *sym) 1520 { 1521 int h; 1522 1523 h = hash(sym->s_name); 1524 if ((sym->s_link = symtab[h]) != NULL) 1525 symtab[h]->s_rlink = &sym->s_link; 1526 sym->s_rlink = &symtab[h]; 1527 symtab[h] = sym; 1528 sym->s_blklev = bl; 1529 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1530 LERROR("inssym()"); 1531 } 1532 1533 /* 1534 * Called at level 0 after syntax errors 1535 * Removes all symbols which are not declared at level 0 from the 1536 * symbol table. Also frees all memory which is not associated with 1537 * level 0. 1538 */ 1539 void 1540 cleanup(void) 1541 { 1542 sym_t *sym, *nsym; 1543 int i; 1544 1545 for (i = 0; i < HSHSIZ1; i++) { 1546 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1547 nsym = sym->s_link; 1548 if (sym->s_blklev >= 1) { 1549 if ((*sym->s_rlink = nsym) != NULL) 1550 nsym->s_rlink = sym->s_rlink; 1551 } 1552 } 1553 } 1554 1555 for (i = mblklev; i > 0; i--) 1556 freelblk(i); 1557 } 1558 1559 /* 1560 * Create a new symbol with the name of an existing symbol. 1561 */ 1562 sym_t * 1563 pushdown(sym_t *sym) 1564 { 1565 int h; 1566 sym_t *nsym; 1567 1568 h = hash(sym->s_name); 1569 nsym = getblk(sizeof (sym_t)); 1570 if (sym->s_blklev > blklev) 1571 LERROR("pushdown()"); 1572 nsym->s_name = sym->s_name; 1573 UNIQUE_CURR_POS(nsym->s_dpos); 1574 nsym->s_kind = sym->s_kind; 1575 nsym->s_blklev = blklev; 1576 1577 if ((nsym->s_link = symtab[h]) != NULL) 1578 symtab[h]->s_rlink = &nsym->s_link; 1579 nsym->s_rlink = &symtab[h]; 1580 symtab[h] = nsym; 1581 1582 *dcs->d_ldlsym = nsym; 1583 dcs->d_ldlsym = &nsym->s_dlnxt; 1584 1585 return (nsym); 1586 } 1587 1588 /* 1589 * Free any dynamically allocated memory referenced by 1590 * the value stack or yylval. 1591 * The type of information in yylval is described by tok. 1592 */ 1593 void 1594 freeyyv(void *sp, int tok) 1595 { 1596 if (tok == T_NAME || tok == T_TYPENAME) { 1597 sbuf_t *sb = *(sbuf_t **)sp; 1598 freesb(sb); 1599 } else if (tok == T_CON) { 1600 val_t *val = *(val_t **)sp; 1601 free(val); 1602 } else if (tok == T_STRING) { 1603 strg_t *strg = *(strg_t **)sp; 1604 if (strg->st_tspec == CHAR) { 1605 free(strg->st_cp); 1606 } else if (strg->st_tspec == WCHAR) { 1607 free(strg->st_wcp); 1608 } else { 1609 LERROR("fryylv()"); 1610 } 1611 free(strg); 1612 } 1613 } 1614
Indexes created Tue Sep 30 20:09:53 GMT 2025