scan.l revision 1.72
1 %{ 2 /* $NetBSD: scan.l,v 1.72 2016/12/29 19:40:35 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 1.72 2016/12/29 19:40:35 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 { "__alias__", T_AT_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 213 { "alias", T_AT_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 214 { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 215 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 }, 216 { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1, 1 }, 217 { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1, 1 }, 218 { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 219 { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 }, 220 { "__used__", T_AT_USED, 0, 0, 0, 0, 0, 1, 1 }, 221 { "used", T_AT_USED, 0, 0, 0, 0, 0, 1, 1 }, 222 { "__always_inline__", T_AT_ALWAYS_INLINE, 223 0, 0, 0, 0, 0, 1, 1 }, 224 { "always_inline", T_AT_ALWAYS_INLINE, 225 0, 0, 0, 0, 0, 1, 1 }, 226 { "__gnu_inline__",T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 }, 227 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 }, 228 { "__constructor__",T_AT_CONSTRUCTOR,0, 0, 0, 0, 0, 1, 1 }, 229 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0, 0, 1, 1 }, 230 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 231 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 }, 232 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 }, 233 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 }, 234 { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 235 { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 }, 236 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 237 { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 }, 238 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 239 { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 }, 240 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0, 0, 1, 1 }, 241 { "__strfmon__",T_AT_FORMAT_STRFMON,0, 0, 0, 0, 0, 1, 1 }, 242 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0, 0, 1, 1 }, 243 { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 }, 244 { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 245 { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 }, 246 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 247 { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 }, 248 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 249 0, 0, 0, 0, 0, 1, 1 }, 250 { "__no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION, 251 0, 0, 0, 0, 0, 1, 1 }, 252 { "__no_instrument_function__", T_AT_NO_INSTRUMENT_FUNCTION, 253 0, 0, 0, 0, 0, 1, 1 }, 254 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 255 { "__sentinel__",T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 }, 256 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 257 { "__format_arg__", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 }, 258 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 259 { "__returns_twice__", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 }, 260 { "cold", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 261 { "__cold__", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 }, 262 { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 0 }, 263 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 264 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 }, 265 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 }, 266 { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 }, 267 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 }, 268 { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 }, 269 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 }, 270 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 }, 271 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 }, 272 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 273 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 }, 274 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 }, 275 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 }, 276 { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 }, 277 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 }, 278 { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 }, 279 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 }, 280 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 }, 281 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 }, 282 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 }, 283 { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 }, 284 { "_Generic", T_GENERIC, 0, 0, 0, 0, 1, 0, 0 }, 285 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 }, 286 { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 }, 287 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 }, 288 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 }, 289 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 290 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 }, 291 { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 }, 292 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 }, 293 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 }, 294 { "mode", T_AT_MODE, 0, 0, 0, 0, 0, 1, 1 }, 295 { "__mode", T_AT_MODE, 0, 0, 0, 0, 0, 1, 1 }, 296 { "__mode__", T_AT_MODE, 0, 0, 0, 0, 0, 1, 1 }, 297 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 }, 298 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 }, 299 { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 300 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 }, 301 { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 }, 302 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 }, 303 { "section", T_AT_SECTION, 0, 0, 0, 0, 0, 1, 1 }, 304 { "__section", T_AT_SECTION, 0, 0, 0, 0, 0, 1, 1 }, 305 { "__section__",T_AT_SECTION, 0, 0, 0, 0, 0, 1, 1 }, 306 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 }, 307 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 }, 308 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 309 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 }, 310 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 }, 311 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 }, 312 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 }, 313 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 }, 314 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 }, 315 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 316 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 317 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 }, 318 { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 }, 319 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 }, 320 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 }, 321 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 }, 322 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 323 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 }, 324 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 325 { "__visibility",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 326 { "__visibility__",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 }, 327 { "weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 328 { "__weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 329 { "__weak__", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 }, 330 { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 }, 331 { NULL, 0, 0, 0, 0, 0, 0, 0, 0 } 332 }; 333 334 /* Symbol table */ 335 static sym_t *symtab[HSHSIZ1]; 336 337 /* bit i of the entry with index i is set */ 338 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT]; 339 340 /* least significant i bits are set in the entry with index i */ 341 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1]; 342 343 /* least significant i bits are not set in the entry with index i */ 344 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1]; 345 346 /* free list for sbuf structures */ 347 static sbuf_t *sbfrlst; 348 349 /* Typ of next expected symbol */ 350 symt_t symtyp; 351 352 353 /* 354 * All keywords are written to the symbol table. This saves us looking 355 * in a extra table for each name we found. 356 */ 357 void 358 initscan(void) 359 { 360 struct kwtab *kw; 361 sym_t *sym; 362 size_t h, i; 363 uint64_t uq; 364 365 for (kw = kwtab; kw->kw_name != NULL; kw++) { 366 if ((kw->kw_c89 || kw->kw_c99) && tflag) 367 continue; 368 if (kw->kw_c99 && !(Sflag || gflag)) 369 continue; 370 if (kw->kw_gcc && !gflag) 371 continue; 372 sym = getblk(sizeof (sym_t)); 373 sym->s_name = kw->kw_name; 374 sym->s_keyw = kw; 375 sym->s_value.v_quad = kw->kw_token; 376 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 377 sym->s_tspec = kw->kw_tspec; 378 } else if (kw->kw_token == T_SCLASS) { 379 sym->s_scl = kw->kw_scl; 380 } else if (kw->kw_token == T_QUAL) { 381 sym->s_tqual = kw->kw_tqual; 382 } 383 h = hash(sym->s_name); 384 if ((sym->s_link = symtab[h]) != NULL) 385 symtab[h]->s_rlink = &sym->s_link; 386 sym->s_rlink = &symtab[h]; 387 symtab[h] = sym; 388 } 389 390 /* initialize bit-masks for quads */ 391 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) { 392 qbmasks[i] = (uint64_t)1 << i; 393 uq = ~(uint64_t)0 << i; 394 qumasks[i] = uq; 395 qlmasks[i] = ~uq; 396 } 397 qumasks[i] = 0; 398 qlmasks[i] = ~(uint64_t)0; 399 } 400 401 /* 402 * Get a free sbuf structure, if possible from the free list 403 */ 404 static sbuf_t * 405 allocsb(void) 406 { 407 sbuf_t *sb; 408 409 if ((sb = sbfrlst) != NULL) { 410 sbfrlst = sb->sb_nxt; 411 #ifdef BLKDEBUG 412 (void)memset(sb, 0, sizeof (*sb)); 413 #else 414 sb->sb_nxt = NULL; 415 #endif 416 } else { 417 sb = xmalloc(sizeof (sbuf_t)); 418 (void)memset(sb, 0, sizeof (*sb)); 419 } 420 return (sb); 421 } 422 423 /* 424 * Put a sbuf structure to the free list 425 */ 426 static void 427 freesb(sbuf_t *sb) 428 { 429 430 (void)memset(sb, ZERO, sizeof (*sb)); 431 sb->sb_nxt = sbfrlst; 432 sbfrlst = sb; 433 } 434 435 /* 436 * Read a character and ensure that it is positive (except EOF). 437 * Increment line count(s) if necessary. 438 */ 439 static int 440 inpc(void) 441 { 442 int c; 443 444 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 445 incline(); 446 return (c); 447 } 448 449 static int 450 hash(const char *s) 451 { 452 u_int v; 453 const u_char *us; 454 455 v = 0; 456 for (us = (const u_char *)s; *us != '\0'; us++) { 457 v = (v << sizeof (v)) + *us; 458 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 459 } 460 return (v % HSHSIZ1); 461 } 462 463 /* 464 * Lex has found a letter followed by zero or more letters or digits. 465 * It looks for a symbol in the symbol table with the same name. This 466 * symbol must either be a keyword or a symbol of the type required by 467 * symtyp (label, member, tag, ...). 468 * 469 * If it is a keyword, the token is returned. In some cases it is described 470 * more deeply by data written to yylval. 471 * 472 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 473 * is stored in yylval. This struct contains the name of the symbol, its 474 * length and hash value. If there is already a symbol of the same name 475 * and type in the symbol table, the sbuf struct also contains a pointer 476 * to the symbol table entry. 477 */ 478 static int 479 name(void) 480 { 481 char *s; 482 sbuf_t *sb; 483 sym_t *sym; 484 int tok; 485 486 sb = allocsb(); 487 sb->sb_name = yytext; 488 sb->sb_len = yyleng; 489 sb->sb_hash = hash(yytext); 490 if ((sym = search(sb)) != NULL && sym->s_keyw) { 491 freesb(sb); 492 return (keyw(sym)); 493 } 494 495 sb->sb_sym = sym; 496 497 if (sym != NULL) { 498 if (blklev < sym->s_blklev) 499 LERROR("name()"); 500 sb->sb_name = sym->s_name; 501 sb->sb_len = strlen(sym->s_name); 502 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 503 } else { 504 s = getblk(yyleng + 1); 505 (void)memcpy(s, yytext, yyleng + 1); 506 sb->sb_name = s; 507 sb->sb_len = yyleng; 508 tok = T_NAME; 509 } 510 511 yylval.y_sb = sb; 512 return (tok); 513 } 514 515 static sym_t * 516 search(sbuf_t *sb) 517 { 518 sym_t *sym; 519 520 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 521 if (strcmp(sym->s_name, sb->sb_name) == 0) { 522 if (sym->s_keyw) { 523 struct kwtab *kw = sym->s_keyw; 524 if (!kw->kw_attr || attron) 525 return (sym); 526 } else if (!attron && sym->s_kind == symtyp) 527 return sym; 528 } 529 } 530 531 return (NULL); 532 } 533 534 static int 535 keyw(sym_t *sym) 536 { 537 int t; 538 539 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 540 yylval.y_scl = sym->s_scl; 541 } else if (t == T_TYPE || t == T_SOU) { 542 yylval.y_tspec = sym->s_tspec; 543 } else if (t == T_QUAL) { 544 yylval.y_tqual = sym->s_tqual; 545 } 546 return (t); 547 } 548 549 /* 550 * Convert a string representing an integer into internal representation. 551 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 552 */ 553 static int 554 icon(int base) 555 { 556 int l_suffix, u_suffix; 557 int len; 558 const char *cp; 559 char c, *eptr; 560 tspec_t typ; 561 uint64_t uq = 0; 562 int ansiu; 563 static tspec_t contypes[2][3] = { 564 { INT, LONG, QUAD }, 565 { UINT, ULONG, UQUAD } 566 }; 567 568 cp = yytext; 569 len = yyleng; 570 571 /* skip 0[xX] or 0[bB] */ 572 if (base == 16 || base == 2) { 573 cp += 2; 574 len -= 2; 575 } 576 577 /* read suffixes */ 578 l_suffix = u_suffix = 0; 579 for ( ; ; ) { 580 if ((c = cp[len - 1]) == 'l' || c == 'L') { 581 l_suffix++; 582 } else if (c == 'u' || c == 'U') { 583 u_suffix++; 584 } else { 585 break; 586 } 587 len--; 588 } 589 if (l_suffix > 2 || u_suffix > 1) { 590 /* malformed integer constant */ 591 warning(251); 592 if (l_suffix > 2) 593 l_suffix = 2; 594 if (u_suffix > 1) 595 u_suffix = 1; 596 } 597 if (tflag && u_suffix != 0) { 598 /* suffix U is illegal in traditional C */ 599 warning(97); 600 } 601 typ = contypes[u_suffix][l_suffix]; 602 603 errno = 0; 604 605 uq = strtouq(cp, &eptr, base); 606 if (eptr != cp + len) 607 LERROR("icon()"); 608 if (errno != 0) 609 /* integer constant out of range */ 610 warning(252); 611 612 /* 613 * If the value is too big for the current type, we must choose 614 * another type. 615 */ 616 ansiu = 0; 617 switch (typ) { 618 case INT: 619 if (uq <= TARG_INT_MAX) { 620 /* ok */ 621 } else if (uq <= TARG_UINT_MAX && base != 10) { 622 typ = UINT; 623 } else if (uq <= TARG_LONG_MAX) { 624 typ = LONG; 625 } else { 626 typ = ULONG; 627 if (uq > TARG_ULONG_MAX) { 628 /* integer constant out of range */ 629 warning(252); 630 } 631 } 632 if (typ == UINT || typ == ULONG) { 633 if (tflag) { 634 typ = LONG; 635 } else if (!sflag) { 636 /* 637 * Remember that the constant is unsigned 638 * only in ANSI C 639 */ 640 ansiu = 1; 641 } 642 } 643 break; 644 case UINT: 645 if (uq > TARG_UINT_MAX) { 646 typ = ULONG; 647 if (uq > TARG_ULONG_MAX) { 648 /* integer constant out of range */ 649 warning(252); 650 } 651 } 652 break; 653 case LONG: 654 if (uq > TARG_LONG_MAX && !tflag) { 655 typ = ULONG; 656 if (!sflag) 657 ansiu = 1; 658 if (uq > TARG_ULONG_MAX) { 659 /* integer constant out of range */ 660 warning(252); 661 } 662 } 663 break; 664 case ULONG: 665 if (uq > TARG_ULONG_MAX) { 666 /* integer constant out of range */ 667 warning(252); 668 } 669 break; 670 case QUAD: 671 if (uq > TARG_QUAD_MAX && !tflag) { 672 typ = UQUAD; 673 if (!sflag) 674 ansiu = 1; 675 } 676 break; 677 case UQUAD: 678 if (uq > TARG_UQUAD_MAX) { 679 /* integer constant out of range */ 680 warning(252); 681 } 682 break; 683 /* LINTED206: (enumeration values not handled in switch) */ 684 case STRUCT: 685 case VOID: 686 case LDOUBLE: 687 case FUNC: 688 case ARRAY: 689 case PTR: 690 case ENUM: 691 case UNION: 692 case SIGNED: 693 case NOTSPEC: 694 case DOUBLE: 695 case FLOAT: 696 case USHORT: 697 case SHORT: 698 case UCHAR: 699 case SCHAR: 700 case CHAR: 701 case BOOL: 702 case UNSIGN: 703 case FCOMPLEX: 704 case DCOMPLEX: 705 case LCOMPLEX: 706 case COMPLEX: 707 break; 708 709 case NTSPEC: /* this value unused */ 710 break; 711 } 712 713 uq = (uint64_t)xsign((int64_t)uq, typ, -1); 714 715 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 716 yylval.y_val->v_ansiu = ansiu; 717 yylval.y_val->v_quad = (int64_t)uq; 718 719 return (T_CON); 720 } 721 722 /* 723 * Returns 1 if t is a signed type and the value is negative. 724 * 725 * len is the number of significant bits. If len is -1, len is set 726 * to the width of type t. 727 */ 728 int 729 sign(int64_t q, tspec_t t, int len) 730 { 731 732 if (t == PTR || isutyp(t)) 733 return (0); 734 return (msb(q, t, len)); 735 } 736 737 int 738 msb(int64_t q, tspec_t t, int len) 739 { 740 741 if (len <= 0) 742 len = size(t); 743 return ((q & qbmasks[len - 1]) != 0); 744 } 745 746 /* 747 * Extends the sign of q. 748 */ 749 int64_t 750 xsign(int64_t q, tspec_t t, int len) 751 { 752 753 if (len <= 0) 754 len = size(t); 755 756 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 757 q &= qlmasks[len]; 758 } else { 759 q |= qumasks[len]; 760 } 761 return (q); 762 } 763 764 /* 765 * Convert a string representing a floating point value into its interal 766 * representation. Type and value are returned in yylval. fcon() 767 * (and yylex()) returns T_CON. 768 * XXX Currently it is not possible to convert constants of type 769 * long double which are greater than DBL_MAX. 770 */ 771 static int 772 fcon(void) 773 { 774 const char *cp; 775 int len; 776 tspec_t typ; 777 char c, *eptr; 778 double d; 779 float f = 0; 780 781 cp = yytext; 782 len = yyleng; 783 784 if (cp[len - 1] == 'i') { 785 /* imaginary, do nothing for now */ 786 len--; 787 } 788 if ((c = cp[len - 1]) == 'f' || c == 'F') { 789 typ = FLOAT; 790 len--; 791 } else if (c == 'l' || c == 'L') { 792 typ = LDOUBLE; 793 len--; 794 } else { 795 if (c == 'd' || c == 'D') 796 len--; 797 typ = DOUBLE; 798 } 799 800 if (tflag && typ != DOUBLE) { 801 /* suffixes F and L are illegal in traditional C */ 802 warning(98); 803 } 804 805 errno = 0; 806 d = strtod(cp, &eptr); 807 if (eptr != cp + len) { 808 switch (*eptr) { 809 /* 810 * XXX: non-native non-current strtod() may not handle hex 811 * floats, ignore the rest if we find traces of hex float 812 * syntax... 813 */ 814 case 'p': 815 case 'P': 816 case 'x': 817 case 'X': 818 d = 0; 819 errno = 0; 820 break; 821 default: 822 LERROR("fcon(%s->%s)", cp, eptr); 823 } 824 } 825 if (errno != 0) 826 /* floating-point constant out of range */ 827 warning(248); 828 829 if (typ == FLOAT) { 830 f = (float)d; 831 if (!finite(f)) { 832 /* floating-point constant out of range */ 833 warning(248); 834 f = f > 0 ? FLT_MAX : -FLT_MAX; 835 } 836 } 837 838 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 839 if (typ == FLOAT) { 840 yylval.y_val->v_ldbl = f; 841 } else { 842 yylval.y_val->v_ldbl = d; 843 } 844 845 return (T_CON); 846 } 847 848 static int 849 operator(int t, op_t o) 850 { 851 852 yylval.y_op = o; 853 return (t); 854 } 855 856 /* 857 * Called if lex found a leading \'. 858 */ 859 static int 860 ccon(void) 861 { 862 size_t n; 863 int val, c; 864 char cv; 865 866 n = 0; 867 val = 0; 868 while ((c = getescc('\'')) >= 0) { 869 val = (val << CHAR_BIT) + c; 870 n++; 871 } 872 if (c == -2) { 873 /* unterminated character constant */ 874 error(253); 875 } else { 876 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 877 /* too many characters in character constant */ 878 error(71); 879 } else if (n > 1) { 880 /* multi-character character constant */ 881 warning(294); 882 } else if (n == 0) { 883 /* empty character constant */ 884 error(73); 885 } 886 } 887 if (n == 1) { 888 cv = (char)val; 889 val = cv; 890 } 891 892 yylval.y_val = xcalloc(1, sizeof (val_t)); 893 yylval.y_val->v_tspec = INT; 894 yylval.y_val->v_quad = val; 895 896 return (T_CON); 897 } 898 899 /* 900 * Called if lex found a leading L\' 901 */ 902 static int 903 wccon(void) 904 { 905 static char buf[MB_LEN_MAX + 1]; 906 size_t i; 907 int c; 908 wchar_t wc; 909 910 i = 0; 911 while ((c = getescc('\'')) >= 0) { 912 if (i < MB_CUR_MAX) 913 buf[i] = (char)c; 914 i++; 915 } 916 917 wc = 0; 918 919 if (c == -2) { 920 /* unterminated character constant */ 921 error(253); 922 } else if (c == 0) { 923 /* empty character constant */ 924 error(73); 925 } else { 926 if (i > MB_CUR_MAX) { 927 i = MB_CUR_MAX; 928 /* too many characters in character constant */ 929 error(71); 930 } else { 931 buf[i] = '\0'; 932 (void)mbtowc(NULL, NULL, 0); 933 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 934 /* invalid multibyte character */ 935 error(291); 936 } 937 } 938 939 yylval.y_val = xcalloc(1, sizeof (val_t)); 940 yylval.y_val->v_tspec = WCHAR; 941 yylval.y_val->v_quad = wc; 942 943 return (T_CON); 944 } 945 946 /* 947 * Read a character which is part of a character constant or of a string 948 * and handle escapes. 949 * 950 * The Argument is the character which delimits the character constant or 951 * string. 952 * 953 * Returns -1 if the end of the character constant or string is reached, 954 * -2 if the EOF is reached, and the character otherwise. 955 */ 956 static int 957 getescc(int d) 958 { 959 static int pbc = -1; 960 int n, c, v; 961 962 if (pbc == -1) { 963 c = inpc(); 964 } else { 965 c = pbc; 966 pbc = -1; 967 } 968 if (c == d) 969 return (-1); 970 switch (c) { 971 case '\n': 972 if (tflag) { 973 /* newline in string or char constant */ 974 error(254); 975 return (-2); 976 } 977 return (c); 978 case EOF: 979 return (-2); 980 case '\\': 981 switch (c = inpc()) { 982 case '"': 983 if (tflag && d == '\'') 984 /* \" inside character constant undef. ... */ 985 warning(262); 986 return ('"'); 987 case '\'': 988 return ('\''); 989 case '?': 990 if (tflag) 991 /* \? undefined in traditional C */ 992 warning(263); 993 return ('?'); 994 case '\\': 995 return ('\\'); 996 case 'a': 997 if (tflag) 998 /* \a undefined in traditional C */ 999 warning(81); 1000 return ('\a'); 1001 case 'b': 1002 return ('\b'); 1003 case 'f': 1004 return ('\f'); 1005 case 'n': 1006 return ('\n'); 1007 case 'r': 1008 return ('\r'); 1009 case 't': 1010 return ('\t'); 1011 case 'v': 1012 if (tflag) 1013 /* \v undefined in traditional C */ 1014 warning(264); 1015 return ('\v'); 1016 case '8': case '9': 1017 /* bad octal digit %c */ 1018 warning(77, c); 1019 /* FALLTHROUGH */ 1020 case '0': case '1': case '2': case '3': 1021 case '4': case '5': case '6': case '7': 1022 n = 3; 1023 v = 0; 1024 do { 1025 v = (v << 3) + (c - '0'); 1026 c = inpc(); 1027 } while (--n && isdigit(c) && (tflag || c <= '7')); 1028 if (tflag && n > 0 && isdigit(c)) 1029 /* bad octal digit %c */ 1030 warning(77, c); 1031 pbc = c; 1032 if (v > TARG_UCHAR_MAX) { 1033 /* character escape does not fit in char. */ 1034 warning(76); 1035 v &= CHAR_MASK; 1036 } 1037 return (v); 1038 case 'x': 1039 if (tflag) 1040 /* \x undefined in traditional C */ 1041 warning(82); 1042 v = 0; 1043 n = 0; 1044 while ((c = inpc()) >= 0 && isxdigit(c)) { 1045 c = isdigit(c) ? 1046 c - '0' : toupper(c) - 'A' + 10; 1047 v = (v << 4) + c; 1048 if (n >= 0) { 1049 if ((v & ~CHAR_MASK) != 0) { 1050 /* overflow in hex escape */ 1051 warning(75); 1052 n = -1; 1053 } else { 1054 n++; 1055 } 1056 } 1057 } 1058 pbc = c; 1059 if (n == 0) { 1060 /* no hex digits follow \x */ 1061 error(74); 1062 } if (n == -1) { 1063 v &= CHAR_MASK; 1064 } 1065 return (v); 1066 case '\n': 1067 return (getescc(d)); 1068 case EOF: 1069 return (-2); 1070 default: 1071 if (isprint(c)) { 1072 /* dubious escape \%c */ 1073 warning(79, c); 1074 } else { 1075 /* dubious escape \%o */ 1076 warning(80, c); 1077 } 1078 } 1079 } 1080 return (c); 1081 } 1082 1083 /* 1084 * Called for preprocessor directives. Currently implemented are: 1085 * # lineno 1086 * # lineno "filename" 1087 */ 1088 static void 1089 directive(void) 1090 { 1091 const char *cp, *fn; 1092 char c, *eptr; 1093 size_t fnl; 1094 long ln; 1095 static int first = 1; 1096 1097 /* Go to first non-whitespace after # */ 1098 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) 1099 continue; 1100 1101 if (!isdigit((unsigned char)c)) { 1102 error: 1103 /* undefined or invalid # directive */ 1104 warning(255); 1105 return; 1106 } 1107 ln = strtol(--cp, &eptr, 10); 1108 if (cp == eptr) 1109 goto error; 1110 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 1111 goto error; 1112 while ((c = *cp++) == ' ' || c == '\t') 1113 continue; 1114 if (c != '\0') { 1115 if (c != '"') 1116 goto error; 1117 fn = cp; 1118 while ((c = *cp) != '"' && c != '\0') 1119 cp++; 1120 if (c != '"') 1121 goto error; 1122 if ((fnl = cp++ - fn) > PATH_MAX) 1123 goto error; 1124 while ((c = *cp++) == ' ' || c == '\t') 1125 continue; 1126 #if 0 1127 if (c != '\0') 1128 warning("extra character(s) after directive"); 1129 #endif 1130 1131 /* empty string means stdin */ 1132 if (fnl == 0) { 1133 fn = "{standard input}"; 1134 fnl = 16; /* strlen (fn) */ 1135 } 1136 curr_pos.p_file = fnnalloc(fn, fnl); 1137 /* 1138 * If this is the first directive, the name is the name 1139 * of the C source file as specified at the command line. 1140 * It is written to the output file. 1141 */ 1142 if (first) { 1143 csrc_pos.p_file = curr_pos.p_file; 1144 outsrc(fnxform(curr_pos.p_file, 1145 strlen(curr_pos.p_file))); 1146 first = 0; 1147 } 1148 } 1149 curr_pos.p_line = (int)ln - 1; 1150 curr_pos.p_uniq = 0; 1151 if (curr_pos.p_file == csrc_pos.p_file) { 1152 csrc_pos.p_line = (int)ln - 1; 1153 csrc_pos.p_uniq = 0; 1154 } 1155 } 1156 1157 /* 1158 * Handle lint comments. Following comments are currently understood: 1159 * ARGSUSEDn 1160 * BITFIELDTYPE 1161 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1162 * FALLTHRU FALLTHROUGH 1163 * LINTLIBRARY 1164 * LINTEDn NOSTRICTn 1165 * LONGLONG 1166 * NOTREACHED 1167 * PRINTFLIKEn 1168 * PROTOLIB 1169 * SCANFLIKEn 1170 * VARARGSn 1171 * If one of this comments is recognized, the arguments, if any, are 1172 * parsed and a function which handles this comment is called. 1173 */ 1174 static void 1175 comment(void) 1176 { 1177 int c, lc; 1178 static struct { 1179 const char *keywd; 1180 int arg; 1181 void (*func)(int); 1182 } keywtab[] = { 1183 { "ARGSUSED", 1, argsused }, 1184 { "BITFIELDTYPE", 0, bitfieldtype }, 1185 { "CONSTCOND", 0, constcond }, 1186 { "CONSTANTCOND", 0, constcond }, 1187 { "CONSTANTCONDITION", 0, constcond }, 1188 { "FALLTHRU", 0, fallthru }, 1189 { "FALLTHROUGH", 0, fallthru }, 1190 { "LINTLIBRARY", 0, lintlib }, 1191 { "LINTED", 1, linted }, 1192 { "LONGLONG", 0, longlong }, 1193 { "NOSTRICT", 1, linted }, 1194 { "NOTREACHED", 0, notreach }, 1195 { "PRINTFLIKE", 1, printflike }, 1196 { "PROTOLIB", 1, protolib }, 1197 { "SCANFLIKE", 1, scanflike }, 1198 { "VARARGS", 1, varargs }, 1199 }; 1200 char keywd[32]; 1201 char arg[32]; 1202 size_t l, i; 1203 int a; 1204 int eoc; 1205 1206 eoc = 0; 1207 1208 /* Skip white spaces after the start of the comment */ 1209 while ((c = inpc()) != EOF && isspace(c)) 1210 continue; 1211 1212 /* Read the potential keyword to keywd */ 1213 l = 0; 1214 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1215 keywd[l++] = (char)c; 1216 c = inpc(); 1217 } 1218 keywd[l] = '\0'; 1219 1220 /* look for the keyword */ 1221 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1222 if (strcmp(keywtab[i].keywd, keywd) == 0) 1223 break; 1224 } 1225 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1226 goto skip_rest; 1227 1228 /* skip white spaces after the keyword */ 1229 while (c != EOF && isspace(c)) 1230 c = inpc(); 1231 1232 /* read the argument, if the keyword accepts one and there is one */ 1233 l = 0; 1234 if (keywtab[i].arg) { 1235 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1236 arg[l++] = (char)c; 1237 c = inpc(); 1238 } 1239 } 1240 arg[l] = '\0'; 1241 a = l != 0 ? atoi(arg) : -1; 1242 1243 /* skip white spaces after the argument */ 1244 while (c != EOF && isspace(c)) 1245 c = inpc(); 1246 1247 if (c != '*' || (c = inpc()) != '/') { 1248 if (keywtab[i].func != linted) 1249 /* extra characters in lint comment */ 1250 warning(257); 1251 } else { 1252 /* 1253 * remember that we have already found the end of the 1254 * comment 1255 */ 1256 eoc = 1; 1257 } 1258 1259 if (keywtab[i].func != NULL) 1260 (*keywtab[i].func)(a); 1261 1262 skip_rest: 1263 while (!eoc) { 1264 lc = c; 1265 if ((c = inpc()) == EOF) { 1266 /* unterminated comment */ 1267 error(256); 1268 break; 1269 } 1270 if (lc == '*' && c == '/') 1271 eoc = 1; 1272 } 1273 } 1274 1275 /* 1276 * Handle // style comments 1277 */ 1278 static void 1279 slashslashcomment(void) 1280 { 1281 int c; 1282 1283 if (!Sflag && !gflag) 1284 /* // comments only supported in C99 */ 1285 (void)gnuism(312, tflag ? "traditional" : "ANSI"); 1286 1287 while ((c = inpc()) != EOF && c != '\n') 1288 continue; 1289 } 1290 1291 /* 1292 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1293 * clrwflgs() is called after function definitions and global and 1294 * local declarations and definitions. It is also called between 1295 * the controlling expression and the body of control statements 1296 * (if, switch, for, while). 1297 */ 1298 void 1299 clrwflgs(void) 1300 { 1301 1302 lwarn = LWARN_ALL; 1303 quadflg = 0; 1304 ccflg = 0; 1305 } 1306 1307 /* 1308 * Strings are stored in a dynamically alloceted buffer and passed 1309 * in yylval.y_xstrg to the parser. The parser or the routines called 1310 * by the parser are responsible for freeing this buffer. 1311 */ 1312 static int 1313 string(void) 1314 { 1315 u_char *s; 1316 int c; 1317 size_t len, max; 1318 strg_t *strg; 1319 1320 s = xmalloc(max = 64); 1321 1322 len = 0; 1323 while ((c = getescc('"')) >= 0) { 1324 /* +1 to reserve space for a trailing NUL character */ 1325 if (len + 1 == max) 1326 s = xrealloc(s, max *= 2); 1327 s[len++] = (char)c; 1328 } 1329 s[len] = '\0'; 1330 if (c == -2) 1331 /* unterminated string constant */ 1332 error(258); 1333 1334 strg = xcalloc(1, sizeof (strg_t)); 1335 strg->st_tspec = CHAR; 1336 strg->st_len = len; 1337 strg->st_cp = s; 1338 1339 yylval.y_strg = strg; 1340 return (T_STRING); 1341 } 1342 1343 static int 1344 wcstrg(void) 1345 { 1346 char *s; 1347 int c, n; 1348 size_t i, wi; 1349 size_t len, max, wlen; 1350 wchar_t *ws; 1351 strg_t *strg; 1352 1353 s = xmalloc(max = 64); 1354 len = 0; 1355 while ((c = getescc('"')) >= 0) { 1356 /* +1 to save space for a trailing NUL character */ 1357 if (len + 1 >= max) 1358 s = xrealloc(s, max *= 2); 1359 s[len++] = (char)c; 1360 } 1361 s[len] = '\0'; 1362 if (c == -2) 1363 /* unterminated string constant */ 1364 error(258); 1365 1366 /* get length of wide-character string */ 1367 (void)mblen(NULL, 0); 1368 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1369 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1370 /* invalid multibyte character */ 1371 error(291); 1372 break; 1373 } 1374 if (n == 0) 1375 n = 1; 1376 } 1377 1378 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1379 1380 /* convert from multibyte to wide char */ 1381 (void)mbtowc(NULL, NULL, 0); 1382 for (i = 0, wi = 0; i < len; i += n, wi++) { 1383 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1384 break; 1385 if (n == 0) 1386 n = 1; 1387 } 1388 ws[wi] = 0; 1389 free(s); 1390 1391 strg = xcalloc(1, sizeof (strg_t)); 1392 strg->st_tspec = WCHAR; 1393 strg->st_len = wlen; 1394 strg->st_wcp = ws; 1395 1396 yylval.y_strg = strg; 1397 return (T_STRING); 1398 } 1399 1400 /* 1401 * As noted above the scanner does not create new symbol table entries 1402 * for symbols it cannot find in the symbol table. This is to avoid 1403 * putting undeclared symbols into the symbol table if a syntax error 1404 * occurs. 1405 * 1406 * getsym() is called as soon as it is probably ok to put the symbol to 1407 * the symbol table. This does not mean that it is not possible that 1408 * symbols are put to the symbol table which are than not completely 1409 * declared due to syntax errors. To avoid too many problems in this 1410 * case symbols get type int in getsym(). 1411 * 1412 * XXX calls to getsym() should be delayed until decl1*() is called 1413 */ 1414 sym_t * 1415 getsym(sbuf_t *sb) 1416 { 1417 dinfo_t *di; 1418 char *s; 1419 sym_t *sym; 1420 1421 sym = sb->sb_sym; 1422 1423 /* 1424 * During member declaration it is possible that name() looked 1425 * for symbols of type FVFT, although it should have looked for 1426 * symbols of type FTAG. Same can happen for labels. Both cases 1427 * are compensated here. 1428 */ 1429 if (symtyp == FMOS || symtyp == FLAB) { 1430 if (sym == NULL || sym->s_kind == FVFT) 1431 sym = search(sb); 1432 } 1433 1434 if (sym != NULL) { 1435 if (sym->s_kind != symtyp) 1436 LERROR("storesym(%d, %d)", sym->s_kind, symtyp); 1437 symtyp = FVFT; 1438 freesb(sb); 1439 return (sym); 1440 } 1441 1442 /* create a new symbol table entry */ 1443 1444 /* labels must always be allocated at level 1 (outhermost block) */ 1445 if (symtyp == FLAB) { 1446 sym = getlblk(1, sizeof (sym_t)); 1447 s = getlblk(1, sb->sb_len + 1); 1448 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1449 sym->s_name = s; 1450 sym->s_blklev = 1; 1451 di = dcs; 1452 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1453 di = di->d_nxt; 1454 if (di->d_ctx != AUTO) 1455 LERROR("storesym()"); 1456 } else { 1457 sym = getblk(sizeof (sym_t)); 1458 sym->s_name = sb->sb_name; 1459 sym->s_blklev = blklev; 1460 di = dcs; 1461 } 1462 1463 UNIQUE_CURR_POS(sym->s_dpos); 1464 if ((sym->s_kind = symtyp) != FLAB) 1465 sym->s_type = gettyp(INT); 1466 1467 symtyp = FVFT; 1468 1469 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1470 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1471 sym->s_rlink = &symtab[sb->sb_hash]; 1472 symtab[sb->sb_hash] = sym; 1473 1474 *di->d_ldlsym = sym; 1475 di->d_ldlsym = &sym->s_dlnxt; 1476 1477 freesb(sb); 1478 return (sym); 1479 } 1480 1481 /* 1482 * Construct a temporary symbol. The symbol starts with a digit, so that 1483 * it is illegal. 1484 */ 1485 sym_t * 1486 mktempsym(type_t *t) 1487 { 1488 static int n = 0; 1489 int h; 1490 char *s = getlblk(blklev, 64); 1491 sym_t *sym = getblk(sizeof (sym_t)); 1492 1493 (void)snprintf(s, 64, "%.8d_tmp", n++); 1494 h = hash(s); 1495 1496 sym->s_name = s; 1497 sym->s_type = t; 1498 sym->s_blklev = blklev; 1499 sym->s_scl = AUTO; 1500 sym->s_kind = FVFT; 1501 sym->s_used = 1; 1502 sym->s_set = 1; 1503 1504 if ((sym->s_link = symtab[h]) != NULL) 1505 symtab[h]->s_rlink = &sym->s_link; 1506 sym->s_rlink = &symtab[h]; 1507 symtab[h] = sym; 1508 1509 *dcs->d_ldlsym = sym; 1510 dcs->d_ldlsym = &sym->s_dlnxt; 1511 1512 return sym; 1513 } 1514 1515 /* 1516 * Remove a symbol forever from the symbol table. s_blklev 1517 * is set to -1 to avoid that the symbol will later be put 1518 * back to the symbol table. 1519 */ 1520 void 1521 rmsym(sym_t *sym) 1522 { 1523 1524 if ((*sym->s_rlink = sym->s_link) != NULL) 1525 sym->s_link->s_rlink = sym->s_rlink; 1526 sym->s_blklev = -1; 1527 sym->s_link = NULL; 1528 } 1529 1530 /* 1531 * Remove a list of symbols declared at one level from the symbol 1532 * table. 1533 */ 1534 void 1535 rmsyms(sym_t *syms) 1536 { 1537 sym_t *sym; 1538 1539 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1540 if (sym->s_blklev != -1) { 1541 if ((*sym->s_rlink = sym->s_link) != NULL) 1542 sym->s_link->s_rlink = sym->s_rlink; 1543 sym->s_link = NULL; 1544 sym->s_rlink = NULL; 1545 } 1546 } 1547 } 1548 1549 /* 1550 * Put a symbol into the symbol table 1551 */ 1552 void 1553 inssym(int bl, sym_t *sym) 1554 { 1555 int h; 1556 1557 h = hash(sym->s_name); 1558 if ((sym->s_link = symtab[h]) != NULL) 1559 symtab[h]->s_rlink = &sym->s_link; 1560 sym->s_rlink = &symtab[h]; 1561 symtab[h] = sym; 1562 sym->s_blklev = bl; 1563 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1564 LERROR("inssym()"); 1565 } 1566 1567 /* 1568 * Called at level 0 after syntax errors 1569 * Removes all symbols which are not declared at level 0 from the 1570 * symbol table. Also frees all memory which is not associated with 1571 * level 0. 1572 */ 1573 void 1574 cleanup(void) 1575 { 1576 sym_t *sym, *nsym; 1577 int i; 1578 1579 for (i = 0; i < HSHSIZ1; i++) { 1580 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1581 nsym = sym->s_link; 1582 if (sym->s_blklev >= 1) { 1583 if ((*sym->s_rlink = nsym) != NULL) 1584 nsym->s_rlink = sym->s_rlink; 1585 } 1586 } 1587 } 1588 1589 for (i = mblklev; i > 0; i--) 1590 freelblk(i); 1591 } 1592 1593 /* 1594 * Create a new symbol with the name of an existing symbol. 1595 */ 1596 sym_t * 1597 pushdown(sym_t *sym) 1598 { 1599 int h; 1600 sym_t *nsym; 1601 1602 h = hash(sym->s_name); 1603 nsym = getblk(sizeof (sym_t)); 1604 if (sym->s_blklev > blklev) 1605 LERROR("pushdown()"); 1606 nsym->s_name = sym->s_name; 1607 UNIQUE_CURR_POS(nsym->s_dpos); 1608 nsym->s_kind = sym->s_kind; 1609 nsym->s_blklev = blklev; 1610 1611 if ((nsym->s_link = symtab[h]) != NULL) 1612 symtab[h]->s_rlink = &nsym->s_link; 1613 nsym->s_rlink = &symtab[h]; 1614 symtab[h] = nsym; 1615 1616 *dcs->d_ldlsym = nsym; 1617 dcs->d_ldlsym = &nsym->s_dlnxt; 1618 1619 return (nsym); 1620 } 1621 1622 /* 1623 * Free any dynamically allocated memory referenced by 1624 * the value stack or yylval. 1625 * The type of information in yylval is described by tok. 1626 */ 1627 void 1628 freeyyv(void *sp, int tok) 1629 { 1630 if (tok == T_NAME || tok == T_TYPENAME) { 1631 sbuf_t *sb = *(sbuf_t **)sp; 1632 freesb(sb); 1633 } else if (tok == T_CON) { 1634 val_t *val = *(val_t **)sp; 1635 free(val); 1636 } else if (tok == T_STRING) { 1637 strg_t *strg = *(strg_t **)sp; 1638 if (strg->st_tspec == CHAR) { 1639 free(strg->st_cp); 1640 } else if (strg->st_tspec == WCHAR) { 1641 free(strg->st_wcp); 1642 } else { 1643 LERROR("fryylv()"); 1644 } 1645 free(strg); 1646 } 1647 } 1648
Indexes created Thu Oct 02 01:09:59 GMT 2025