scan.l revision 1.66
1 %{ 2 /* $NetBSD: scan.l,v 1.66 2016/07/20 17:42:14 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 1.66 2016/07/20 17:42:14 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 { "__gnu_inline__",T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 }, 219 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 }, 220 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 221 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 222 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 }, 223 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 224 { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 225 { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 226 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 227 { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 228 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 229 { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 230 { "strftime", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 231 { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 }, 232 { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 233 { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 234 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 235 { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 236 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 237 { "__sentinel__",T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 238 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 239 { "__format_arg__", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 240 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 241 { "__returns_twice__", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 242 { "cold", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 243 { "__cold__", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 244 { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 0 }, 245 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 246 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 247 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 }, 248 { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 }, 249 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 }, 250 { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 }, 251 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 }, 252 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 }, 253 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 }, 254 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 255 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 256 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 }, 257 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 }, 258 { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 }, 259 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 }, 260 { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 }, 261 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 }, 262 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 }, 263 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 }, 264 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 }, 265 { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 }, 266 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 }, 267 { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 }, 268 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 }, 269 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 }, 270 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 271 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 272 { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 }, 273 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 }, 274 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 }, 275 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 }, 276 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 }, 277 { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 278 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 279 { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 }, 280 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 }, 281 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 }, 282 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 }, 283 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 284 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 285 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 }, 286 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 }, 287 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 }, 288 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 }, 289 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 }, 290 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 291 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 292 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 293 { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 }, 294 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 }, 295 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 }, 296 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 }, 297 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 298 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 299 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 300 { "__visibility",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 301 { "__visibility__",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 302 { "weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 303 { "__weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 304 { "__weak__", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 305 { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 }, 306 { NULL, 0, 0, 0, 0, 0, 0, 0, 0 } 307 }; 308 309 /* Symbol table */ 310 static sym_t *symtab[HSHSIZ1]; 311 312 /* bit i of the entry with index i is set */ 313 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 314 315 /* least significant i bits are set in the entry with index i */ 316 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 317 318 /* least significant i bits are not set in the entry with index i */ 319 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 320 321 /* free list for sbuf structures */ 322 static sbuf_t *sbfrlst; 323 324 /* Typ of next expected symbol */ 325 symt_t symtyp; 326 327 328 /* 329 * All keywords are written to the symbol table. This saves us looking 330 * in a extra table for each name we found. 331 */ 332 void 333 initscan(void) 334 { 335 struct kwtab *kw; 336 sym_t *sym; 337 size_t h, i; 338 uint64_t uq; 339 340 for (kw = kwtab; kw->kw_name != NULL; kw++) { 341 if ((kw->kw_c89 || kw->kw_c99) && tflag) 342 continue; 343 if (kw->kw_c99 && !(Sflag || gflag)) 344 continue; 345 if (kw->kw_gcc && !gflag) 346 continue; 347 sym = getblk(sizeof (sym_t)); 348 sym->s_name = kw->kw_name; 349 sym->s_keyw = kw; 350 sym->s_value.v_quad = kw->kw_token; 351 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 352 sym->s_tspec = kw->kw_tspec; 353 } else if (kw->kw_token == T_SCLASS) { 354 sym->s_scl = kw->kw_scl; 355 } else if (kw->kw_token == T_QUAL) { 356 sym->s_tqual = kw->kw_tqual; 357 } 358 h = hash(sym->s_name); 359 if ((sym->s_link = symtab[h]) != NULL) 360 symtab[h]->s_rlink = &sym->s_link; 361 sym->s_rlink = &symtab[h]; 362 symtab[h] = sym; 363 } 364 365 /* initialize bit-masks for quads */ 366 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 367 qbmasks[i] = (uint64_t)1 << i; 368 uq = ~(uint64_t)0 << i; 369 qumasks[i] = uq; 370 qlmasks[i] = ~uq; 371 } 372 qumasks[i] = 0; 373 qlmasks[i] = ~(uint64_t)0; 374 } 375 376 /* 377 * Get a free sbuf structure, if possible from the free list 378 */ 379 static sbuf_t * 380 allocsb(void) 381 { 382 sbuf_t *sb; 383 384 if ((sb = sbfrlst) != NULL) { 385 sbfrlst = sb->sb_nxt; 386 #ifdef BLKDEBUG 387 (void)memset(sb, 0, sizeof (*sb)); 388 #else 389 sb->sb_nxt = NULL; 390 #endif 391 } else { 392 sb = xmalloc(sizeof (sbuf_t)); 393 (void)memset(sb, 0, sizeof (*sb)); 394 } 395 return (sb); 396 } 397 398 /* 399 * Put a sbuf structure to the free list 400 */ 401 static void 402 freesb(sbuf_t *sb) 403 { 404 405 (void)memset(sb, ZERO, sizeof (*sb)); 406 sb->sb_nxt = sbfrlst; 407 sbfrlst = sb; 408 } 409 410 /* 411 * Read a character and ensure that it is positive (except EOF). 412 * Increment line count(s) if necessary. 413 */ 414 static int 415 inpc(void) 416 { 417 int c; 418 419 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 420 incline(); 421 return (c); 422 } 423 424 static int 425 hash(const char *s) 426 { 427 u_int v; 428 const u_char *us; 429 430 v = 0; 431 for (us = (const u_char *)s; *us != '\0'; us++) { 432 v = (v << sizeof (v)) + *us; 433 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 434 } 435 return (v % HSHSIZ1); 436 } 437 438 /* 439 * Lex has found a letter followed by zero or more letters or digits. 440 * It looks for a symbol in the symbol table with the same name. This 441 * symbol must either be a keyword or a symbol of the type required by 442 * symtyp (label, member, tag, ...). 443 * 444 * If it is a keyword, the token is returned. In some cases it is described 445 * more deeply by data written to yylval. 446 * 447 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 448 * is stored in yylval. This struct contains the name of the symbol, its 449 * length and hash value. If there is already a symbol of the same name 450 * and type in the symbol table, the sbuf struct also contains a pointer 451 * to the symbol table entry. 452 */ 453 static int 454 name(void) 455 { 456 char *s; 457 sbuf_t *sb; 458 sym_t *sym; 459 int tok; 460 461 sb = allocsb(); 462 sb->sb_name = yytext; 463 sb->sb_len = yyleng; 464 sb->sb_hash = hash(yytext); 465 if ((sym = search(sb)) != NULL && sym->s_keyw) { 466 freesb(sb); 467 return (keyw(sym)); 468 } 469 470 sb->sb_sym = sym; 471 472 if (sym != NULL) { 473 if (blklev < sym->s_blklev) 474 LERROR("name()"); 475 sb->sb_name = sym->s_name; 476 sb->sb_len = strlen(sym->s_name); 477 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 478 } else { 479 s = getblk(yyleng + 1); 480 (void)memcpy(s, yytext, yyleng + 1); 481 sb->sb_name = s; 482 sb->sb_len = yyleng; 483 tok = T_NAME; 484 } 485 486 yylval.y_sb = sb; 487 return (tok); 488 } 489 490 static sym_t * 491 search(sbuf_t *sb) 492 { 493 sym_t *sym; 494 495 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 496 if (strcmp(sym->s_name, sb->sb_name) == 0) { 497 if (sym->s_keyw) { 498 struct kwtab *kw = sym->s_keyw; 499 if (!kw->kw_attr || attron) 500 return (sym); 501 } else if (sym->s_kind == symtyp) 502 return sym; 503 } 504 } 505 506 return (NULL); 507 } 508 509 static int 510 keyw(sym_t *sym) 511 { 512 int t; 513 514 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 515 yylval.y_scl = sym->s_scl; 516 } else if (t == T_TYPE || t == T_SOU) { 517 yylval.y_tspec = sym->s_tspec; 518 } else if (t == T_QUAL) { 519 yylval.y_tqual = sym->s_tqual; 520 } 521 return (t); 522 } 523 524 /* 525 * Convert a string representing an integer into internal representation. 526 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 527 */ 528 static int 529 icon(int base) 530 { 531 int l_suffix, u_suffix; 532 int len; 533 const char *cp; 534 char c, *eptr; 535 tspec_t typ; 536 uint64_t uq = 0; 537 int ansiu; 538 static tspec_t contypes[2][3] = { 539 { INT, LONG, QUAD }, 540 { UINT, ULONG, UQUAD } 541 }; 542 543 cp = yytext; 544 len = yyleng; 545 546 /* skip 0[xX] or 0[bB] */ 547 if (base == 16 || base == 2) { 548 cp += 2; 549 len -= 2; 550 } 551 552 /* read suffixes */ 553 l_suffix = u_suffix = 0; 554 for ( ; ; ) { 555 if ((c = cp[len - 1]) == 'l' || c == 'L') { 556 l_suffix++; 557 } else if (c == 'u' || c == 'U') { 558 u_suffix++; 559 } else { 560 break; 561 } 562 len--; 563 } 564 if (l_suffix > 2 || u_suffix > 1) { 565 /* malformed integer constant */ 566 warning(251); 567 if (l_suffix > 2) 568 l_suffix = 2; 569 if (u_suffix > 1) 570 u_suffix = 1; 571 } 572 if (tflag && u_suffix != 0) { 573 /* suffix U is illegal in traditional C */ 574 warning(97); 575 } 576 typ = contypes[u_suffix][l_suffix]; 577 578 errno = 0; 579 580 uq = strtouq(cp, &eptr, base); 581 if (eptr != cp + len) 582 LERROR("icon()"); 583 if (errno != 0) 584 /* integer constant out of range */ 585 warning(252); 586 587 /* 588 * If the value is too big for the current type, we must choose 589 * another type. 590 */ 591 ansiu = 0; 592 switch (typ) { 593 case INT: 594 if (uq <= TARG_INT_MAX) { 595 /* ok */ 596 } else if (uq <= TARG_UINT_MAX && base != 10) { 597 typ = UINT; 598 } else if (uq <= TARG_LONG_MAX) { 599 typ = LONG; 600 } else { 601 typ = ULONG; 602 if (uq > TARG_ULONG_MAX) { 603 /* integer constant out of range */ 604 warning(252); 605 } 606 } 607 if (typ == UINT || typ == ULONG) { 608 if (tflag) { 609 typ = LONG; 610 } else if (!sflag) { 611 /* 612 * Remember that the constant is unsigned 613 * only in ANSI C 614 */ 615 ansiu = 1; 616 } 617 } 618 break; 619 case UINT: 620 if (uq > TARG_UINT_MAX) { 621 typ = ULONG; 622 if (uq > TARG_ULONG_MAX) { 623 /* integer constant out of range */ 624 warning(252); 625 } 626 } 627 break; 628 case LONG: 629 if (uq > TARG_LONG_MAX && !tflag) { 630 typ = ULONG; 631 if (!sflag) 632 ansiu = 1; 633 if (uq > TARG_ULONG_MAX) { 634 /* integer constant out of range */ 635 warning(252); 636 } 637 } 638 break; 639 case ULONG: 640 if (uq > TARG_ULONG_MAX) { 641 /* integer constant out of range */ 642 warning(252); 643 } 644 break; 645 case QUAD: 646 if (uq > TARG_QUAD_MAX && !tflag) { 647 typ = UQUAD; 648 if (!sflag) 649 ansiu = 1; 650 } 651 break; 652 case UQUAD: 653 if (uq > TARG_UQUAD_MAX) { 654 /* integer constant out of range */ 655 warning(252); 656 } 657 break; 658 /* LINTED206: (enumeration values not handled in switch) */ 659 case STRUCT: 660 case VOID: 661 case LDOUBLE: 662 case FUNC: 663 case ARRAY: 664 case PTR: 665 case ENUM: 666 case UNION: 667 case SIGNED: 668 case NOTSPEC: 669 case DOUBLE: 670 case FLOAT: 671 case USHORT: 672 case SHORT: 673 case UCHAR: 674 case SCHAR: 675 case CHAR: 676 case BOOL: 677 case UNSIGN: 678 case FCOMPLEX: 679 case DCOMPLEX: 680 case LCOMPLEX: 681 case COMPLEX: 682 break; 683 684 case NTSPEC: /* this value unused */ 685 break; 686 } 687 688 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 689 690 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 691 yylval.y_val->v_ansiu = ansiu; 692 yylval.y_val->v_quad = (int64_t)uq; 693 694 return (T_CON); 695 } 696 697 /* 698 * Returns 1 if t is a signed type and the value is negative. 699 * 700 * len is the number of significant bits. If len is -1, len is set 701 * to the width of type t. 702 */ 703 int 704 sign(int64_t q, tspec_t t, int len) 705 { 706 707 if (t == PTR || isutyp(t)) 708 return (0); 709 return (msb(q, t, len)); 710 } 711 712 int 713 msb(int64_t q, tspec_t t, int len) 714 { 715 716 if (len <= 0) 717 len = size(t); 718 return ((q & qbmasks[len - 1]) != 0); 719 } 720 721 /* 722 * Extends the sign of q. 723 */ 724 int64_t 725 xsign(int64_t q, tspec_t t, int len) 726 { 727 728 if (len <= 0) 729 len = size(t); 730 731 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 732 q &= qlmasks[len]; 733 } else { 734 q |= qumasks[len]; 735 } 736 return (q); 737 } 738 739 /* 740 * Convert a string representing a floating point value into its interal 741 * representation. Type and value are returned in yylval. fcon() 742 * (and yylex()) returns T_CON. 743 * XXX Currently it is not possible to convert constants of type 744 * long double which are greater than DBL_MAX. 745 */ 746 static int 747 fcon(void) 748 { 749 const char *cp; 750 int len; 751 tspec_t typ; 752 char c, *eptr; 753 double d; 754 float f = 0; 755 756 cp = yytext; 757 len = yyleng; 758 759 if (cp[len - 1] == 'i') { 760 /* imaginary, do nothing for now */ 761 len--; 762 } 763 if ((c = cp[len - 1]) == 'f' || c == 'F') { 764 typ = FLOAT; 765 len--; 766 } else if (c == 'l' || c == 'L') { 767 typ = LDOUBLE; 768 len--; 769 } else { 770 if (c == 'd' || c == 'D') 771 len--; 772 typ = DOUBLE; 773 } 774 775 if (tflag && typ != DOUBLE) { 776 /* suffixes F and L are illegal in traditional C */ 777 warning(98); 778 } 779 780 errno = 0; 781 d = strtod(cp, &eptr); 782 if (eptr != cp + len) { 783 switch (*eptr) { 784 /* 785 * XXX: non-native non-current strtod() may not handle hex 786 * floats, ignore the rest if we find traces of hex float 787 * syntax... 788 */ 789 case 'p': 790 case 'P': 791 case 'x': 792 case 'X': 793 d = 0; 794 errno = 0; 795 break; 796 default: 797 LERROR("fcon(%s->%s)", cp, eptr); 798 } 799 } 800 if (errno != 0) 801 /* floating-point constant out of range */ 802 warning(248); 803 804 if (typ == FLOAT) { 805 f = (float)d; 806 if (!finite(f)) { 807 /* floating-point constant out of range */ 808 warning(248); 809 f = f > 0 ? FLT_MAX : -FLT_MAX; 810 } 811 } 812 813 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 814 if (typ == FLOAT) { 815 yylval.y_val->v_ldbl = f; 816 } else { 817 yylval.y_val->v_ldbl = d; 818 } 819 820 return (T_CON); 821 } 822 823 static int 824 operator(int t, op_t o) 825 { 826 827 yylval.y_op = o; 828 return (t); 829 } 830 831 /* 832 * Called if lex found a leading \'. 833 */ 834 static int 835 ccon(void) 836 { 837 size_t n; 838 int val, c; 839 char cv; 840 841 n = 0; 842 val = 0; 843 while ((c = getescc('\'')) >= 0) { 844 val = (val << CHAR_BIT) + c; 845 n++; 846 } 847 if (c == -2) { 848 /* unterminated character constant */ 849 error(253); 850 } else { 851 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 852 /* too many characters in character constant */ 853 error(71); 854 } else if (n > 1) { 855 /* multi-character character constant */ 856 warning(294); 857 } else if (n == 0) { 858 /* empty character constant */ 859 error(73); 860 } 861 } 862 if (n == 1) { 863 cv = (char)val; 864 val = cv; 865 } 866 867 yylval.y_val = xcalloc(1, sizeof (val_t)); 868 yylval.y_val->v_tspec = INT; 869 yylval.y_val->v_quad = val; 870 871 return (T_CON); 872 } 873 874 /* 875 * Called if lex found a leading L\' 876 */ 877 static int 878 wccon(void) 879 { 880 static char buf[MB_LEN_MAX + 1]; 881 size_t i; 882 int c; 883 wchar_t wc; 884 885 i = 0; 886 while ((c = getescc('\'')) >= 0) { 887 if (i < MB_CUR_MAX) 888 buf[i] = (char)c; 889 i++; 890 } 891 892 wc = 0; 893 894 if (c == -2) { 895 /* unterminated character constant */ 896 error(253); 897 } else if (c == 0) { 898 /* empty character constant */ 899 error(73); 900 } else { 901 if (i > MB_CUR_MAX) { 902 i = MB_CUR_MAX; 903 /* too many characters in character constant */ 904 error(71); 905 } else { 906 buf[i] = '\0'; 907 (void)mbtowc(NULL, NULL, 0); 908 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 909 /* invalid multibyte character */ 910 error(291); 911 } 912 } 913 914 yylval.y_val = xcalloc(1, sizeof (val_t)); 915 yylval.y_val->v_tspec = WCHAR; 916 yylval.y_val->v_quad = wc; 917 918 return (T_CON); 919 } 920 921 /* 922 * Read a character which is part of a character constant or of a string 923 * and handle escapes. 924 * 925 * The Argument is the character which delimits the character constant or 926 * string. 927 * 928 * Returns -1 if the end of the character constant or string is reached, 929 * -2 if the EOF is reached, and the character otherwise. 930 */ 931 static int 932 getescc(int d) 933 { 934 static int pbc = -1; 935 int n, c, v; 936 937 if (pbc == -1) { 938 c = inpc(); 939 } else { 940 c = pbc; 941 pbc = -1; 942 } 943 if (c == d) 944 return (-1); 945 switch (c) { 946 case '\n': 947 if (tflag) { 948 /* newline in string or char constant */ 949 error(254); 950 return (-2); 951 } 952 return (c); 953 case EOF: 954 return (-2); 955 case '\\': 956 switch (c = inpc()) { 957 case '"': 958 if (tflag && d == '\'') 959 /* \" inside character constant undef. ... */ 960 warning(262); 961 return ('"'); 962 case '\'': 963 return ('\''); 964 case '?': 965 if (tflag) 966 /* \? undefined in traditional C */ 967 warning(263); 968 return ('?'); 969 case '\\': 970 return ('\\'); 971 case 'a': 972 if (tflag) 973 /* \a undefined in traditional C */ 974 warning(81); 975 return ('\a'); 976 case 'b': 977 return ('\b'); 978 case 'f': 979 return ('\f'); 980 case 'n': 981 return ('\n'); 982 case 'r': 983 return ('\r'); 984 case 't': 985 return ('\t'); 986 case 'v': 987 if (tflag) 988 /* \v undefined in traditional C */ 989 warning(264); 990 return ('\v'); 991 case '8': case '9': 992 /* bad octal digit %c */ 993 warning(77, c); 994 /* FALLTHROUGH */ 995 case '0': case '1': case '2': case '3': 996 case '4': case '5': case '6': case '7': 997 n = 3; 998 v = 0; 999 do { 1000 v = (v << 3) + (c - '0'); 1001 c = inpc(); 1002 } while (--n && isdigit(c) && (tflag || c <= '7')); 1003 if (tflag && n > 0 && isdigit(c)) 1004 /* bad octal digit %c */ 1005 warning(77, c); 1006 pbc = c; 1007 if (v > TARG_UCHAR_MAX) { 1008 /* character escape does not fit in char. */ 1009 warning(76); 1010 v &= CHAR_MASK; 1011 } 1012 return (v); 1013 case 'x': 1014 if (tflag) 1015 /* \x undefined in traditional C */ 1016 warning(82); 1017 v = 0; 1018 n = 0; 1019 while ((c = inpc()) >= 0 && isxdigit(c)) { 1020 c = isdigit(c) ? 1021 c - '0' : toupper(c) - 'A' + 10; 1022 v = (v << 4) + c; 1023 if (n >= 0) { 1024 if ((v & ~CHAR_MASK) != 0) { 1025 /* overflow in hex escape */ 1026 warning(75); 1027 n = -1; 1028 } else { 1029 n++; 1030 } 1031 } 1032 } 1033 pbc = c; 1034 if (n == 0) { 1035 /* no hex digits follow \x */ 1036 error(74); 1037 } if (n == -1) { 1038 v &= CHAR_MASK; 1039 } 1040 return (v); 1041 case '\n': 1042 return (getescc(d)); 1043 case EOF: 1044 return (-2); 1045 default: 1046 if (isprint(c)) { 1047 /* dubious escape \%c */ 1048 warning(79, c); 1049 } else { 1050 /* dubious escape \%o */ 1051 warning(80, c); 1052 } 1053 } 1054 } 1055 return (c); 1056 } 1057 1058 /* 1059 * Called for preprocessor directives. Currently implemented are: 1060 * # lineno 1061 * # lineno "filename" 1062 */ 1063 static void 1064 directive(void) 1065 { 1066 const char *cp, *fn; 1067 char c, *eptr; 1068 size_t fnl; 1069 long ln; 1070 static int first = 1; 1071 1072 /* Go to first non-whitespace after # */ 1073 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1074 continue; 1075 1076 if (!isdigit((unsigned char)c)) { 1077 error: 1078 /* undefined or invalid # directive */ 1079 warning(255); 1080 return; 1081 } 1082 ln = strtol(--cp, &eptr, 10); 1083 if (cp == eptr) 1084 goto error; 1085 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1086 goto error; 1087 while ((c = *cp++) == ' ' || c == '\t') 1088 continue; 1089 if (c != '\0') { 1090 if (c != '"') 1091 goto error; 1092 fn = cp; 1093 while ((c = *cp) != '"' && c != '\0') 1094 cp++; 1095 if (c != '"') 1096 goto error; 1097 if ((fnl = cp++ - fn) > PATH_MAX) 1098 goto error; 1099 while ((c = *cp++) == ' ' || c == '\t') 1100 continue; 1101 #if 0 1102 if (c != '\0') 1103 warning("extra character(s) after directive"); 1104 #endif 1105 1106 /* empty string means stdin */ 1107 if (fnl == 0) { 1108 fn = "{standard input}"; 1109 fnl = 16; /* strlen (fn) */ 1110 } 1111 curr_pos.p_file = fnnalloc(fn, fnl); 1112 /* 1113 * If this is the first directive, the name is the name 1114 * of the C source file as specified at the command line. 1115 * It is written to the output file. 1116 */ 1117 if (first) { 1118 csrc_pos.p_file = curr_pos.p_file; 1119 outsrc(curr_pos.p_file); 1120 first = 0; 1121 } 1122 } 1123 curr_pos.p_line = (int)ln - 1; 1124 curr_pos.p_uniq = 0; 1125 if (curr_pos.p_file == csrc_pos.p_file) { 1126 csrc_pos.p_line = (int)ln - 1; 1127 csrc_pos.p_uniq = 0; 1128 } 1129 } 1130 1131 /* 1132 * Handle lint comments. Following comments are currently understood: 1133 * ARGSUSEDn 1134 * BITFIELDTYPE 1135 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1136 * FALLTHRU FALLTHROUGH 1137 * LINTLIBRARY 1138 * LINTEDn NOSTRICTn 1139 * LONGLONG 1140 * NOTREACHED 1141 * PRINTFLIKEn 1142 * PROTOLIB 1143 * SCANFLIKEn 1144 * VARARGSn 1145 * If one of this comments is recognized, the arguments, if any, are 1146 * parsed and a function which handles this comment is called. 1147 */ 1148 static void 1149 comment(void) 1150 { 1151 int c, lc; 1152 static struct { 1153 const char *keywd; 1154 int arg; 1155 void (*func)(int); 1156 } keywtab[] = { 1157 { "ARGSUSED", 1, argsused }, 1158 { "BITFIELDTYPE", 0, bitfieldtype }, 1159 { "CONSTCOND", 0, constcond }, 1160 { "CONSTANTCOND", 0, constcond }, 1161 { "CONSTANTCONDITION", 0, constcond }, 1162 { "FALLTHRU", 0, fallthru }, 1163 { "FALLTHROUGH", 0, fallthru }, 1164 { "LINTLIBRARY", 0, lintlib }, 1165 { "LINTED", 1, linted }, 1166 { "LONGLONG", 0, longlong }, 1167 { "NOSTRICT", 1, linted }, 1168 { "NOTREACHED", 0, notreach }, 1169 { "PRINTFLIKE", 1, printflike }, 1170 { "PROTOLIB", 1, protolib }, 1171 { "SCANFLIKE", 1, scanflike }, 1172 { "VARARGS", 1, varargs }, 1173 }; 1174 char keywd[32]; 1175 char arg[32]; 1176 size_t l, i; 1177 int a; 1178 int eoc; 1179 1180 eoc = 0; 1181 1182 /* Skip white spaces after the start of the comment */ 1183 while ((c = inpc()) != EOF && isspace(c)) 1184 continue; 1185 1186 /* Read the potential keyword to keywd */ 1187 l = 0; 1188 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1189 keywd[l++] = (char)c; 1190 c = inpc(); 1191 } 1192 keywd[l] = '\0'; 1193 1194 /* look for the keyword */ 1195 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1196 if (strcmp(keywtab[i].keywd, keywd) == 0) 1197 break; 1198 } 1199 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1200 goto skip_rest; 1201 1202 /* skip white spaces after the keyword */ 1203 while (c != EOF && isspace(c)) 1204 c = inpc(); 1205 1206 /* read the argument, if the keyword accepts one and there is one */ 1207 l = 0; 1208 if (keywtab[i].arg) { 1209 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1210 arg[l++] = (char)c; 1211 c = inpc(); 1212 } 1213 } 1214 arg[l] = '\0'; 1215 a = l != 0 ? atoi(arg) : -1; 1216 1217 /* skip white spaces after the argument */ 1218 while (c != EOF && isspace(c)) 1219 c = inpc(); 1220 1221 if (c != '*' || (c = inpc()) != '/') { 1222 if (keywtab[i].func != linted) 1223 /* extra characters in lint comment */ 1224 warning(257); 1225 } else { 1226 /* 1227 * remember that we have already found the end of the 1228 * comment 1229 */ 1230 eoc = 1; 1231 } 1232 1233 if (keywtab[i].func != NULL) 1234 (*keywtab[i].func)(a); 1235 1236 skip_rest: 1237 while (!eoc) { 1238 lc = c; 1239 if ((c = inpc()) == EOF) { 1240 /* unterminated comment */ 1241 error(256); 1242 break; 1243 } 1244 if (lc == '*' && c == '/') 1245 eoc = 1; 1246 } 1247 } 1248 1249 /* 1250 * Handle // style comments 1251 */ 1252 static void 1253 slashslashcomment(void) 1254 { 1255 int c; 1256 1257 if (!Sflag && !gflag) 1258 /* // comments only supported in C99 */ 1259 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1260 1261 while ((c = inpc()) != EOF && c != '\n') 1262 continue; 1263 } 1264 1265 /* 1266 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1267 * clrwflgs() is called after function definitions and global and 1268 * local declarations and definitions. It is also called between 1269 * the controlling expression and the body of control statements 1270 * (if, switch, for, while). 1271 */ 1272 void 1273 clrwflgs(void) 1274 { 1275 1276 lwarn = LWARN_ALL; 1277 quadflg = 0; 1278 ccflg = 0; 1279 } 1280 1281 /* 1282 * Strings are stored in a dynamically alloceted buffer and passed 1283 * in yylval.y_xstrg to the parser. The parser or the routines called 1284 * by the parser are responsible for freeing this buffer. 1285 */ 1286 static int 1287 string(void) 1288 { 1289 u_char *s; 1290 int c; 1291 size_t len, max; 1292 strg_t *strg; 1293 1294 s = xmalloc(max = 64); 1295 1296 len = 0; 1297 while ((c = getescc('"')) >= 0) { 1298 /* +1 to reserve space for a trailing NUL character */ 1299 if (len + 1 == max) 1300 s = xrealloc(s, max *= 2); 1301 s[len++] = (char)c; 1302 } 1303 s[len] = '\0'; 1304 if (c == -2) 1305 /* unterminated string constant */ 1306 error(258); 1307 1308 strg = xcalloc(1, sizeof (strg_t)); 1309 strg->st_tspec = CHAR; 1310 strg->st_len = len; 1311 strg->st_cp = s; 1312 1313 yylval.y_strg = strg; 1314 return (T_STRING); 1315 } 1316 1317 static int 1318 wcstrg(void) 1319 { 1320 char *s; 1321 int c, n; 1322 size_t i, wi; 1323 size_t len, max, wlen; 1324 wchar_t *ws; 1325 strg_t *strg; 1326 1327 s = xmalloc(max = 64); 1328 len = 0; 1329 while ((c = getescc('"')) >= 0) { 1330 /* +1 to save space for a trailing NUL character */ 1331 if (len + 1 >= max) 1332 s = xrealloc(s, max *= 2); 1333 s[len++] = (char)c; 1334 } 1335 s[len] = '\0'; 1336 if (c == -2) 1337 /* unterminated string constant */ 1338 error(258); 1339 1340 /* get length of wide-character string */ 1341 (void)mblen(NULL, 0); 1342 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1343 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1344 /* invalid multibyte character */ 1345 error(291); 1346 break; 1347 } 1348 if (n == 0) 1349 n = 1; 1350 } 1351 1352 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1353 1354 /* convert from multibyte to wide char */ 1355 (void)mbtowc(NULL, NULL, 0); 1356 for (i = 0, wi = 0; i < len; i += n, wi++) { 1357 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1358 break; 1359 if (n == 0) 1360 n = 1; 1361 } 1362 ws[wi] = 0; 1363 free(s); 1364 1365 strg = xcalloc(1, sizeof (strg_t)); 1366 strg->st_tspec = WCHAR; 1367 strg->st_len = wlen; 1368 strg->st_wcp = ws; 1369 1370 yylval.y_strg = strg; 1371 return (T_STRING); 1372 } 1373 1374 /* 1375 * As noted above the scanner does not create new symbol table entries 1376 * for symbols it cannot find in the symbol table. This is to avoid 1377 * putting undeclared symbols into the symbol table if a syntax error 1378 * occurs. 1379 * 1380 * getsym() is called as soon as it is probably ok to put the symbol to 1381 * the symbol table. This does not mean that it is not possible that 1382 * symbols are put to the symbol table which are than not completely 1383 * declared due to syntax errors. To avoid too many problems in this 1384 * case symbols get type int in getsym(). 1385 * 1386 * XXX calls to getsym() should be delayed until decl1*() is called 1387 */ 1388 sym_t * 1389 getsym(sbuf_t *sb) 1390 { 1391 dinfo_t *di; 1392 char *s; 1393 sym_t *sym; 1394 1395 sym = sb->sb_sym; 1396 1397 /* 1398 * During member declaration it is possible that name() looked 1399 * for symbols of type FVFT, although it should have looked for 1400 * symbols of type FTAG. Same can happen for labels. Both cases 1401 * are compensated here. 1402 */ 1403 if (symtyp == FMOS || symtyp == FLAB) { 1404 if (sym == NULL || sym->s_kind == FVFT) 1405 sym = search(sb); 1406 } 1407 1408 if (sym != NULL) { 1409 if (sym->s_kind != symtyp) 1410 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1411 symtyp = FVFT; 1412 freesb(sb); 1413 return (sym); 1414 } 1415 1416 /* create a new symbol table entry */ 1417 1418 /* labels must always be allocated at level 1 (outhermost block) */ 1419 if (symtyp == FLAB) { 1420 sym = getlblk(1, sizeof (sym_t)); 1421 s = getlblk(1, sb->sb_len + 1); 1422 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1423 sym->s_name = s; 1424 sym->s_blklev = 1; 1425 di = dcs; 1426 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1427 di = di->d_nxt; 1428 if (di->d_ctx != AUTO) 1429 LERROR("storesym()"); 1430 } else { 1431 sym = getblk(sizeof (sym_t)); 1432 sym->s_name = sb->sb_name; 1433 sym->s_blklev = blklev; 1434 di = dcs; 1435 } 1436 1437 UNIQUE_CURR_POS(sym->s_dpos); 1438 if ((sym->s_kind = symtyp) != FLAB) 1439 sym->s_type = gettyp(INT); 1440 1441 symtyp = FVFT; 1442 1443 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1444 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1445 sym->s_rlink = &symtab[sb->sb_hash]; 1446 symtab[sb->sb_hash] = sym; 1447 1448 *di->d_ldlsym = sym; 1449 di->d_ldlsym = &sym->s_dlnxt; 1450 1451 freesb(sb); 1452 return (sym); 1453 } 1454 1455 /* 1456 * Construct a temporary symbol. The symbol starts with a digit, so that 1457 * it is illegal. 1458 */ 1459 sym_t * 1460 mktempsym(type_t *t) 1461 { 1462 static int n = 0; 1463 int h; 1464 char *s = getlblk(blklev, 64); 1465 sym_t *sym = getblk(sizeof (sym_t)); 1466 1467 (void)snprintf(s, 64, "%.8d_tmp", n++); 1468 h = hash(s); 1469 1470 sym->s_name = s; 1471 sym->s_type = t; 1472 sym->s_blklev = blklev; 1473 sym->s_scl = AUTO; 1474 sym->s_kind = FVFT; 1475 sym->s_used = 1; 1476 sym->s_set = 1; 1477 1478 if ((sym->s_link = symtab[h]) != NULL) 1479 symtab[h]->s_rlink = &sym->s_link; 1480 sym->s_rlink = &symtab[h]; 1481 symtab[h] = sym; 1482 1483 *dcs->d_ldlsym = sym; 1484 dcs->d_ldlsym = &sym->s_dlnxt; 1485 1486 return sym; 1487 } 1488 1489 /* 1490 * Remove a symbol forever from the symbol table. s_blklev 1491 * is set to -1 to avoid that the symbol will later be put 1492 * back to the symbol table. 1493 */ 1494 void 1495 rmsym(sym_t *sym) 1496 { 1497 1498 if ((*sym->s_rlink = sym->s_link) != NULL) 1499 sym->s_link->s_rlink = sym->s_rlink; 1500 sym->s_blklev = -1; 1501 sym->s_link = NULL; 1502 } 1503 1504 /* 1505 * Remove a list of symbols declared at one level from the symbol 1506 * table. 1507 */ 1508 void 1509 rmsyms(sym_t *syms) 1510 { 1511 sym_t *sym; 1512 1513 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1514 if (sym->s_blklev != -1) { 1515 if ((*sym->s_rlink = sym->s_link) != NULL) 1516 sym->s_link->s_rlink = sym->s_rlink; 1517 sym->s_link = NULL; 1518 sym->s_rlink = NULL; 1519 } 1520 } 1521 } 1522 1523 /* 1524 * Put a symbol into the symbol table 1525 */ 1526 void 1527 inssym(int bl, sym_t *sym) 1528 { 1529 int h; 1530 1531 h = hash(sym->s_name); 1532 if ((sym->s_link = symtab[h]) != NULL) 1533 symtab[h]->s_rlink = &sym->s_link; 1534 sym->s_rlink = &symtab[h]; 1535 symtab[h] = sym; 1536 sym->s_blklev = bl; 1537 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1538 LERROR("inssym()"); 1539 } 1540 1541 /* 1542 * Called at level 0 after syntax errors 1543 * Removes all symbols which are not declared at level 0 from the 1544 * symbol table. Also frees all memory which is not associated with 1545 * level 0. 1546 */ 1547 void 1548 cleanup(void) 1549 { 1550 sym_t *sym, *nsym; 1551 int i; 1552 1553 for (i = 0; i < HSHSIZ1; i++) { 1554 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1555 nsym = sym->s_link; 1556 if (sym->s_blklev >= 1) { 1557 if ((*sym->s_rlink = nsym) != NULL) 1558 nsym->s_rlink = sym->s_rlink; 1559 } 1560 } 1561 } 1562 1563 for (i = mblklev; i > 0; i--) 1564 freelblk(i); 1565 } 1566 1567 /* 1568 * Create a new symbol with the name of an existing symbol. 1569 */ 1570 sym_t * 1571 pushdown(sym_t *sym) 1572 { 1573 int h; 1574 sym_t *nsym; 1575 1576 h = hash(sym->s_name); 1577 nsym = getblk(sizeof (sym_t)); 1578 if (sym->s_blklev > blklev) 1579 LERROR("pushdown()"); 1580 nsym->s_name = sym->s_name; 1581 UNIQUE_CURR_POS(nsym->s_dpos); 1582 nsym->s_kind = sym->s_kind; 1583 nsym->s_blklev = blklev; 1584 1585 if ((nsym->s_link = symtab[h]) != NULL) 1586 symtab[h]->s_rlink = &nsym->s_link; 1587 nsym->s_rlink = &symtab[h]; 1588 symtab[h] = nsym; 1589 1590 *dcs->d_ldlsym = nsym; 1591 dcs->d_ldlsym = &nsym->s_dlnxt; 1592 1593 return (nsym); 1594 } 1595 1596 /* 1597 * Free any dynamically allocated memory referenced by 1598 * the value stack or yylval. 1599 * The type of information in yylval is described by tok. 1600 */ 1601 void 1602 freeyyv(void *sp, int tok) 1603 { 1604 if (tok == T_NAME || tok == T_TYPENAME) { 1605 sbuf_t *sb = *(sbuf_t **)sp; 1606 freesb(sb); 1607 } else if (tok == T_CON) { 1608 val_t *val = *(val_t **)sp; 1609 free(val); 1610 } else if (tok == T_STRING) { 1611 strg_t *strg = *(strg_t **)sp; 1612 if (strg->st_tspec == CHAR) { 1613 free(strg->st_cp); 1614 } else if (strg->st_tspec == WCHAR) { 1615 free(strg->st_wcp); 1616 } else { 1617 LERROR("fryylv()"); 1618 } 1619 free(strg); 1620 } 1621 } 1622
Indexes created Thu Oct 02 14:10:14 GMT 2025