Home | History | Annotate | Line # | Download | only in lint1
scan.l revision 1.56
      1 %{
      2 /* $NetBSD: scan.l,v 1.56 2014/04/18 01:39:57 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 1.56 2014/04/18 01:39:57 christos Exp $");
     39 #endif
     40 
     41 #include <stdlib.h>
     42 #include <string.h>
     43 #include <limits.h>
     44 #include <float.h>
     45 #include <ctype.h>
     46 #include <errno.h>
     47 #include <math.h>
     48 
     49 #include "lint1.h"
     50 #include "cgram.h"
     51 
     52 #define CHAR_MASK	(~(~0 << CHAR_BIT))
     53 
     54 /* Current position (its also updated when an included file is parsed) */
     55 pos_t	curr_pos = { 1, "", 0 };
     56 
     57 /*
     58  * Current position in C source (not updated when an included file is
     59  * parsed).
     60  */
     61 pos_t	csrc_pos = { 1, "", 0 };
     62 
     63 /* Are we parsing a gcc attribute? */
     64 int attron;
     65 
     66 static	void	incline(void);
     67 static	void	badchar(int);
     68 static	sbuf_t	*allocsb(void);
     69 static	void	freesb(sbuf_t *);
     70 static	int	inpc(void);
     71 static	int	hash(const char *);
     72 static	sym_t	*search(sbuf_t *);
     73 static	int	name(void);
     74 static	int	keyw(sym_t *);
     75 static	int	icon(int);
     76 static	int	fcon(void);
     77 static	int	operator(int, op_t);
     78 static	int	ccon(void);
     79 static	int	wccon(void);
     80 static	int	getescc(int);
     81 static	void	directive(void);
     82 static	void	comment(void);
     83 static	void	slashslashcomment(void);
     84 static	int	string(void);
     85 static	int	wcstrg(void);
     86 
     87 %}
     88 
     89 L	[_A-Za-z]
     90 D	[0-9]
     91 NZD	[1-9]
     92 OD	[0-7]
     93 HD	[0-9A-Fa-f]
     94 EX	([eE][+-]?[0-9]+)
     95 HX	(p[+-]?[0-9A-Fa-f]+)
     96 TL	([fFlL]?[i]?)
     97 
     98 %option nounput
     99 
    100 %%
    101 
    102 {L}({L}|{D})*		 	return (name());
    103 0{OD}*[lLuU]*			return (icon(8));
    104 {NZD}{D}*[lLuU]*		return (icon(10));
    105 0[xX]{HD}+[lLuU]*		return (icon(16));
    106 {D}+\.{D}*{EX}?{TL}	|
    107 {D}+{EX}{TL}		|
    108 0[xX]{HD}+\.{HD}*{HX}{TL} |
    109 0[xX]{HD}+{HX}{TL} 	|
    110 \.{D}+{EX}?{TL}		return (fcon());
    111 "="				return (operator(T_ASSIGN, ASSIGN));
    112 "*="				return (operator(T_OPASS, MULASS));
    113 "/="				return (operator(T_OPASS, DIVASS));
    114 "%="				return (operator(T_OPASS, MODASS));
    115 "+="				return (operator(T_OPASS, ADDASS));
    116 "-="				return (operator(T_OPASS, SUBASS));
    117 "<<="				return (operator(T_OPASS, SHLASS));
    118 ">>="				return (operator(T_OPASS, SHRASS));
    119 "&="				return (operator(T_OPASS, ANDASS));
    120 "^="				return (operator(T_OPASS, XORASS));
    121 "|="				return (operator(T_OPASS, ORASS));
    122 "||"				return (operator(T_LOGOR, LOGOR));
    123 "&&"				return (operator(T_LOGAND, LOGAND));
    124 "|"				return (operator(T_OR, OR));
    125 "&"				return (operator(T_AND, AND));
    126 "^"				return (operator(T_XOR, XOR));
    127 "=="				return (operator(T_EQOP, EQ));
    128 "!="				return (operator(T_EQOP, NE));
    129 "<"				return (operator(T_RELOP, LT));
    130 ">"				return (operator(T_RELOP, GT));
    131 "<="				return (operator(T_RELOP, LE));
    132 ">="				return (operator(T_RELOP, GE));
    133 "<<"				return (operator(T_SHFTOP, SHL));
    134 ">>"				return (operator(T_SHFTOP, SHR));
    135 "++"				return (operator(T_INCDEC, INC));
    136 "--"				return (operator(T_INCDEC, DEC));
    137 "->"				return (operator(T_STROP, ARROW));
    138 "."				return (operator(T_STROP, POINT));
    139 "+"				return (operator(T_ADDOP, PLUS));
    140 "-"				return (operator(T_ADDOP, MINUS));
    141 "*"				return (operator(T_MULT, MULT));
    142 "/"				return (operator(T_DIVOP, DIV));
    143 "%"				return (operator(T_DIVOP, MOD));
    144 "!"				return (operator(T_UNOP, NOT));
    145 "~"				return (operator(T_UNOP, COMPL));
    146 "\""				return (string());
    147 "L\""				return (wcstrg());
    148 ";"				return (T_SEMI);
    149 "{"				return (T_LBRACE);
    150 "}"				return (T_RBRACE);
    151 ","				return (T_COMMA);
    152 ":"				return (T_COLON);
    153 "?"				return (T_QUEST);
    154 "["				return (T_LBRACK);
    155 "]"				return (T_RBRACK);
    156 "("				return (T_LPARN);
    157 ")"				return (T_RPARN);
    158 "..."				return (T_ELLIPSE);
    159 "'"				return (ccon());
    160 "L'"				return (wccon());
    161 ^#.*$				directive();
    162 \n				incline();
    163 \t|" "|\f|\v			;
    164 "/*"				comment();
    165 "//"				slashslashcomment();
    166 .				badchar(yytext[0]);
    167 
    168 %%
    169 
    170 static void
    171 incline(void)
    172 {
    173 	curr_pos.p_line++;
    174 	curr_pos.p_uniq = 0;
    175 	if (curr_pos.p_file == csrc_pos.p_file) {
    176 		csrc_pos.p_line++;
    177 		csrc_pos.p_uniq = 0;
    178 	}
    179 }
    180 
    181 static void
    182 badchar(int c)
    183 {
    184 
    185 	/* unknown character \%o */
    186 	error(250, c);
    187 }
    188 
    189 /*
    190  * Keywords.
    191  * During initialisation they are written to the symbol table.
    192  */
    193 static	struct	kwtab {
    194 	const	char *kw_name;	/* keyword */
    195 	int	kw_token;	/* token returned by yylex() */
    196 	scl_t	kw_scl;		/* storage class if kw_token T_SCLASS */
    197 	tspec_t	kw_tspec;	/* type spec. if kw_token T_TYPE or T_SOU */
    198 	tqual_t	kw_tqual;	/* type qual. fi kw_token T_QUAL */
    199 	u_int	kw_c89 : 1;	/* c89 keyword */
    200 	u_int	kw_c99 : 1;	/* c99 keyword */
    201 	u_int	kw_gcc : 1;	/* GCC keyword */
    202 	u_int	kw_attr : 1;	/* GCC attribute, keyword */
    203 } kwtab[] = {
    204 	{ "__alignof__", T_ALIGNOF,	0,	0,	0,	  0, 0, 0, 0 },
    205 	{ "__attribute__",T_ATTRIBUTE,	0,	0,	0,	  0, 0, 1, 0 },
    206 	{ "attribute",	T_ATTRIBUTE,	0,	0,	0,	  0, 0, 1, 0 },
    207 	{ "__packed__",	T_AT_PACKED,	0,	0,	0,	  0, 0, 1, 1 },
    208 	{ "packed",	T_AT_PACKED,	0,	0,	0,	  0, 0, 1, 1 },
    209 	{ "__aligned__",T_AT_ALIGNED,	0,	0,	0,	  0, 0, 1, 1 },
    210 	{ "aligned",	T_AT_ALIGNED,	0,	0,	0,	  0, 0, 1, 1 },
    211 	{ "__transparent_union__",T_AT_TUNION,0,0,	0,	  0, 0, 1, 1 },
    212 	{ "transparent_union",T_AT_TUNION,0,	0,	0,	  0, 0, 1, 1 },
    213 	{ "__unused__",	T_AT_UNUSED,	0,	0,	0,	  0, 0, 1, 1 },
    214 	{ "unused",	T_AT_UNUSED,	0,	0,	0,	  0, 0, 1, 1 },
    215 	{ "__deprecated__",T_AT_DEPRECATED,0,	0,	0,	  0, 0, 1, 1 },
    216 	{ "deprecated",	T_AT_DEPRECATED,0,	0,	0,	  0, 0, 1, 1 },
    217 	{ "__may_alias__",T_AT_MAY_ALIAS,0,	0,	0,	  0, 0, 1, 1 },
    218 	{ "may_alias",	T_AT_MAY_ALIAS,	0,	0,	0,	  0, 0, 1, 1 },
    219 	{ "format",	T_AT_FORMAT,	0,	0,	0,	  0, 0, 1, 1 },
    220 	{ "__format__",	T_AT_FORMAT,	0,	0,	0,	  0, 0, 1, 1 },
    221 	{ "printf",	T_AT_FORMAT_PRINTF,0,	0,	0,	  0, 0, 1, 1 },
    222 	{ "__printf__",	T_AT_FORMAT_PRINTF,0,	0,	0,	  0, 0, 1, 1 },
    223 	{ "scanf",	T_AT_FORMAT_SCANF,0,	0,	0,	  0, 0, 1, 1 },
    224 	{ "__scanf__",	T_AT_FORMAT_SCANF,0,	0,	0,	  0, 0, 1, 1 },
    225 	{ "strftime",	T_AT_FORMAT_SCANF,0,	0,	0,	  0, 0, 1, 1 },
    226 	{ "__strftime__",T_AT_FORMAT_STRFTIME,0,0,	0,	  0, 0, 1, 1 },
    227 	{ "pure",	T_AT_PURE,	0,	0,	0,	  0, 0, 1, 1 },
    228 	{ "__pure__",	T_AT_PURE,	0,	0,	0,	  0, 0, 1, 1 },
    229 	{ "noreturn",	T_AT_NORETURN,	0,	0,	0,	  0, 0, 1, 1 },
    230 	{ "__noreturn__",T_AT_NORETURN,	0,	0,	0,	  0, 0, 1, 1 },
    231 	{ "asm",	T_ASM,		0,	0,	0,	  0, 0, 1, 1 },
    232 	{ "__asm",	T_ASM,		0,	0,	0,	  0, 0, 0, 0 },
    233 	{ "__asm__",	T_ASM,		0,	0,	0,	  0, 0, 0, 0 },
    234 	{ "auto",	T_SCLASS,	AUTO,	0,	0,	  0, 0, 0, 0 },
    235 	{ "break",	T_BREAK,	0,	0,	0,	  0, 0, 0, 0 },
    236 	{ "_Bool",	T_TYPE,		0,	BOOL,	0,	  0, 1, 0, 0 },
    237 	{ "case",	T_CASE,		0,	0,	0,	  0, 0, 0, 0 },
    238 	{ "char",	T_TYPE,		0,	CHAR,	0,	  0, 0, 0, 0 },
    239 	{ "const",	T_QUAL,		0,	0,	CONST,	  1, 0, 0, 0 },
    240 	{ "_Complex",	T_TYPE,		0,	COMPLEX,0,	  0, 1, 0, 0 },
    241 	{ "__const__",	T_QUAL,		0,	0,	CONST,	  0, 0, 0, 0 },
    242 	{ "__const",	T_QUAL,		0,	0,	CONST,	  0, 0, 0, 0 },
    243 	{ "continue",	T_CONTINUE,	0,	0,	0,	  0, 0, 0, 0 },
    244 	{ "default",	T_DEFAULT,	0,	0,	0,	  0, 0, 0, 0 },
    245 	{ "do",		T_DO,		0,	0,	0,	  0, 0, 0, 0 },
    246 	{ "double",	T_TYPE,		0,	DOUBLE,	0,	  0, 0, 0, 0 },
    247 	{ "else",	T_ELSE,		0,	0,	0,	  0, 0, 0, 0 },
    248 	{ "enum",	T_ENUM,		0,	0,	0,	  0, 0, 0, 0 },
    249 	{ "__extension__", T_EXTENSION,	0,	0,	0,	  0, 0, 1, 0 },
    250 	{ "extern",	T_SCLASS,	EXTERN,	0,	0,	  0, 0, 0, 0 },
    251 	{ "float",	T_TYPE,		0,	FLOAT,	0,	  0, 0, 0, 0 },
    252 	{ "for",	T_FOR,		0,	0,	0,	  0, 0, 0, 0 },
    253 	{ "goto",	T_GOTO,		0,	0,	0,	  0, 0, 0, 0 },
    254 	{ "if",		T_IF,		0,	0,	0,	  0, 0, 0, 0 },
    255 	{ "__imag__",	T_IMAG,		0,	0,	0,	  0, 1, 0, 0 },
    256 	{ "inline",	T_SCLASS,	INLINE,	0,	0,	  0, 1, 0, 0 },
    257 	{ "__inline__",	T_SCLASS,	INLINE,	0,	0,	  0, 0, 0, 0 },
    258 	{ "__inline",	T_SCLASS,	INLINE,	0,	0,	  0, 0, 0, 0 },
    259 	{ "int",	T_TYPE,		0,	INT,	0,	  0, 0, 0, 0 },
    260 	{ "__symbolrename", T_SYMBOLRENAME, 0,	0,	0,	  0, 0, 0, 0 },
    261 	{ "long",	T_TYPE,		0,	LONG,	0,	  0, 0, 0, 0 },
    262 	{ "__real__",	T_REAL,		0,	0,	0,	  0, 1, 0, 0 },
    263 	{ "register",	T_SCLASS,	REG,	0,	0,	  0, 0, 0, 0 },
    264 	{ "__restrict__",T_QUAL,	0,	0,	RESTRICT, 0, 1, 0, 0 },
    265 	{ "restrict",	T_QUAL,		0,	0,	RESTRICT, 0, 1, 0, 0 },
    266 	{ "return",	T_RETURN,	0,	0,	0,	  0, 0, 0, 0 },
    267 	{ "__packed",	T_PACKED,	0,	0,	0,	  0, 0, 0, 0 },
    268 	{ "short",	T_TYPE,		0,	SHORT,	0,	  0, 0, 0, 0 },
    269 	{ "signed",	T_TYPE,		0,	SIGNED,	0,	  1, 0, 0, 0 },
    270 	{ "__signed__",	T_TYPE,		0,	SIGNED,	0,	  0, 0, 0, 0 },
    271 	{ "__signed",	T_TYPE,		0,	SIGNED,	0,	  0, 0, 0, 0 },
    272 	{ "sizeof",	T_SIZEOF,	0,	0,	0,	  0, 0, 0, 0 },
    273 	{ "static",	T_SCLASS,	STATIC,	0,	0,	  0, 0, 0, 0 },
    274 	{ "struct",	T_SOU,		0,	STRUCT,	0,	  0, 0, 0, 0 },
    275 	{ "switch",	T_SWITCH,	0,	0,	0,	  0, 0, 0, 0 },
    276 	{ "typedef",	T_SCLASS,	TYPEDEF, 0,	0,	  0, 0, 0, 0 },
    277 	{ "typeof",	T_TYPEOF,	0,	0,	0,	  0, 0, 1, 0 },
    278 	{ "__typeof",	T_TYPEOF,	0,	0,	0,	  0, 0, 1, 0 },
    279 	{ "__typeof__",	T_TYPEOF,	0,	0,	0,	  0, 0, 1, 0 },
    280 	{ "union",	T_SOU,		0,	UNION,	0,	  0, 0, 0, 0 },
    281 	{ "unsigned",	T_TYPE,		0,	UNSIGN,	0,	  0, 0, 0, 0 },
    282 	{ "void",	T_TYPE,		0,	VOID,	0,	  0, 0, 0, 0 },
    283 	{ "volatile",	T_QUAL,		0,	0,	VOLATILE, 1, 0, 0, 0 },
    284 	{ "__volatile__", T_QUAL,	0,	0,	VOLATILE, 0, 0, 0, 0 },
    285 	{ "__volatile",	T_QUAL,		0,	0,	VOLATILE, 0, 0, 0, 0 },
    286 	{ "while",	T_WHILE,	0,	0,	0,	  0, 0, 0, 0 },
    287 	{ NULL,		0,		0,	0,	0,	  0, 0, 0, 0 }
    288 };
    289 
    290 /* Symbol table */
    291 static	sym_t	*symtab[HSHSIZ1];
    292 
    293 /* bit i of the entry with index i is set */
    294 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
    295 
    296 /* least significant i bits are set in the entry with index i */
    297 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
    298 
    299 /* least significant i bits are not set in the entry with index i */
    300 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
    301 
    302 /* free list for sbuf structures */
    303 static	sbuf_t	 *sbfrlst;
    304 
    305 /* Typ of next expected symbol */
    306 symt_t	symtyp;
    307 
    308 
    309 /*
    310  * All keywords are written to the symbol table. This saves us looking
    311  * in a extra table for each name we found.
    312  */
    313 void
    314 initscan(void)
    315 {
    316 	struct	kwtab *kw;
    317 	sym_t	*sym;
    318 	size_t	h, i;
    319 	uint64_t uq;
    320 
    321 	for (kw = kwtab; kw->kw_name != NULL; kw++) {
    322 		if ((kw->kw_c89 || kw->kw_c99) && tflag)
    323 			continue;
    324 		if (kw->kw_c99 && !(Sflag || gflag))
    325 			continue;
    326 		if (kw->kw_gcc && !gflag)
    327 			continue;
    328 		sym = getblk(sizeof (sym_t));
    329 		sym->s_name = kw->kw_name;
    330 		sym->s_keyw = kw;
    331 		sym->s_value.v_quad = kw->kw_token;
    332 		if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
    333 			sym->s_tspec = kw->kw_tspec;
    334 		} else if (kw->kw_token == T_SCLASS) {
    335 			sym->s_scl = kw->kw_scl;
    336 		} else if (kw->kw_token == T_QUAL) {
    337 			sym->s_tqual = kw->kw_tqual;
    338 		}
    339 		h = hash(sym->s_name);
    340 		if ((sym->s_link = symtab[h]) != NULL)
    341 			symtab[h]->s_rlink = &sym->s_link;
    342 		sym->s_rlink = &symtab[h];
    343 		symtab[h] = sym;
    344 	}
    345 
    346 	/* initialize bit-masks for quads */
    347 	for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
    348 		qbmasks[i] = (uint64_t)1 << i;
    349 		uq = ~(uint64_t)0 << i;
    350 		qumasks[i] = uq;
    351 		qlmasks[i] = ~uq;
    352 	}
    353 	qumasks[i] = 0;
    354 	qlmasks[i] = ~(uint64_t)0;
    355 }
    356 
    357 /*
    358  * Get a free sbuf structure, if possible from the free list
    359  */
    360 static sbuf_t *
    361 allocsb(void)
    362 {
    363 	sbuf_t	*sb;
    364 
    365 	if ((sb = sbfrlst) != NULL) {
    366 		sbfrlst = sb->sb_nxt;
    367 #ifdef BLKDEBUG
    368 		(void)memset(sb, 0, sizeof (*sb));
    369 #else
    370 		sb->sb_nxt = NULL;
    371 #endif
    372 	} else {
    373 		sb = xmalloc(sizeof (sbuf_t));
    374 		(void)memset(sb, 0, sizeof (*sb));
    375 	}
    376 	return (sb);
    377 }
    378 
    379 /*
    380  * Put a sbuf structure to the free list
    381  */
    382 static void
    383 freesb(sbuf_t *sb)
    384 {
    385 
    386 	(void)memset(sb, ZERO, sizeof (*sb));
    387 	sb->sb_nxt = sbfrlst;
    388 	sbfrlst = sb;
    389 }
    390 
    391 /*
    392  * Read a character and ensure that it is positive (except EOF).
    393  * Increment line count(s) if necessary.
    394  */
    395 static int
    396 inpc(void)
    397 {
    398 	int	c;
    399 
    400 	if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
    401 		incline();
    402 	return (c);
    403 }
    404 
    405 static int
    406 hash(const char *s)
    407 {
    408 	u_int	v;
    409 	const	u_char *us;
    410 
    411 	v = 0;
    412 	for (us = (const u_char *)s; *us != '\0'; us++) {
    413 		v = (v << sizeof (v)) + *us;
    414 		v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
    415 	}
    416 	return (v % HSHSIZ1);
    417 }
    418 
    419 /*
    420  * Lex has found a letter followed by zero or more letters or digits.
    421  * It looks for a symbol in the symbol table with the same name. This
    422  * symbol must either be a keyword or a symbol of the type required by
    423  * symtyp (label, member, tag, ...).
    424  *
    425  * If it is a keyword, the token is returned. In some cases it is described
    426  * more deeply by data written to yylval.
    427  *
    428  * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
    429  * is stored in yylval. This struct contains the name of the symbol, it's
    430  * length and hash value. If there is already a symbol of the same name
    431  * and type in the symbol table, the sbuf struct also contains a pointer
    432  * to the symbol table entry.
    433  */
    434 static int
    435 name(void)
    436 {
    437 	char	*s;
    438 	sbuf_t	*sb;
    439 	sym_t	*sym;
    440 	int	tok;
    441 
    442 	sb = allocsb();
    443 	sb->sb_name = yytext;
    444 	sb->sb_len = yyleng;
    445 	sb->sb_hash = hash(yytext);
    446 	if ((sym = search(sb)) != NULL && sym->s_keyw) {
    447 		freesb(sb);
    448 		return (keyw(sym));
    449 	}
    450 
    451 	sb->sb_sym = sym;
    452 
    453 	if (sym != NULL) {
    454 		if (blklev < sym->s_blklev)
    455 			LERROR("name()");
    456 		sb->sb_name = sym->s_name;
    457 		sb->sb_len = strlen(sym->s_name);
    458 		tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
    459 	} else {
    460 		s = getblk(yyleng + 1);
    461 		(void)memcpy(s, yytext, yyleng + 1);
    462 		sb->sb_name = s;
    463 		sb->sb_len = yyleng;
    464 		tok = T_NAME;
    465 	}
    466 
    467 	yylval.y_sb = sb;
    468 	return (tok);
    469 }
    470 
    471 static sym_t *
    472 search(sbuf_t *sb)
    473 {
    474 	sym_t	*sym;
    475 
    476 	for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
    477 		if (strcmp(sym->s_name, sb->sb_name) == 0) {
    478 			if (sym->s_keyw) {
    479 				struct kwtab *kw = sym->s_keyw;
    480 				if (!kw->kw_attr || attron)
    481 					return (sym);
    482 			} else if (sym->s_kind == symtyp)
    483 				return sym;
    484 		}
    485 	}
    486 
    487 	return (NULL);
    488 }
    489 
    490 static int
    491 keyw(sym_t *sym)
    492 {
    493 	int	t;
    494 
    495 	if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
    496 		yylval.y_scl = sym->s_scl;
    497 	} else if (t == T_TYPE || t == T_SOU) {
    498 		yylval.y_tspec = sym->s_tspec;
    499 	} else if (t == T_QUAL) {
    500 		yylval.y_tqual = sym->s_tqual;
    501 	}
    502 	return (t);
    503 }
    504 
    505 /*
    506  * Convert a string representing an integer into internal representation.
    507  * The value is returned in yylval. icon() (and yylex()) returns T_CON.
    508  */
    509 static int
    510 icon(int base)
    511 {
    512 	int	l_suffix, u_suffix;
    513 	int	len;
    514 	const	char *cp;
    515 	char	c, *eptr;
    516 	tspec_t	typ;
    517 	uint64_t uq = 0;
    518 	int	ansiu;
    519 	static	tspec_t contypes[2][3] = {
    520 		{ INT,  LONG,  QUAD },
    521 		{ UINT, ULONG, UQUAD }
    522 	};
    523 
    524 	cp = yytext;
    525 	len = yyleng;
    526 
    527 	/* skip 0x */
    528 	if (base == 16) {
    529 		cp += 2;
    530 		len -= 2;
    531 	}
    532 
    533 	/* read suffixes */
    534 	l_suffix = u_suffix = 0;
    535 	for ( ; ; ) {
    536 		if ((c = cp[len - 1]) == 'l' || c == 'L') {
    537 			l_suffix++;
    538 		} else if (c == 'u' || c == 'U') {
    539 			u_suffix++;
    540 		} else {
    541 			break;
    542 		}
    543 		len--;
    544 	}
    545 	if (l_suffix > 2 || u_suffix > 1) {
    546 		/* malformed integer constant */
    547 		warning(251);
    548 		if (l_suffix > 2)
    549 			l_suffix = 2;
    550 		if (u_suffix > 1)
    551 			u_suffix = 1;
    552 	}
    553 	if (tflag && u_suffix != 0) {
    554 		/* suffix U is illegal in traditional C */
    555 		warning(97);
    556 	}
    557 	typ = contypes[u_suffix][l_suffix];
    558 
    559 	errno = 0;
    560 
    561 	uq = strtouq(cp, &eptr, base);
    562 	if (eptr != cp + len)
    563 		LERROR("icon()");
    564 	if (errno != 0)
    565 		/* integer constant out of range */
    566 		warning(252);
    567 
    568 	/*
    569 	 * If the value is too big for the current type, we must choose
    570 	 * another type.
    571 	 */
    572 	ansiu = 0;
    573 	switch (typ) {
    574 	case INT:
    575 		if (uq <= TARG_INT_MAX) {
    576 			/* ok */
    577 		} else if (uq <= TARG_UINT_MAX && base != 10) {
    578 			typ = UINT;
    579 		} else if (uq <= TARG_LONG_MAX) {
    580 			typ = LONG;
    581 		} else {
    582 			typ = ULONG;
    583 			if (uq > TARG_ULONG_MAX) {
    584 				/* integer constant out of range */
    585 				warning(252);
    586 			}
    587 		}
    588 		if (typ == UINT || typ == ULONG) {
    589 			if (tflag) {
    590 				typ = LONG;
    591 			} else if (!sflag) {
    592 				/*
    593 				 * Remember that the constant is unsigned
    594 				 * only in ANSI C
    595 				 */
    596 				ansiu = 1;
    597 			}
    598 		}
    599 		break;
    600 	case UINT:
    601 		if (uq > TARG_UINT_MAX) {
    602 			typ = ULONG;
    603 			if (uq > TARG_ULONG_MAX) {
    604 				/* integer constant out of range */
    605 				warning(252);
    606 			}
    607 		}
    608 		break;
    609 	case LONG:
    610 		if (uq > TARG_LONG_MAX && !tflag) {
    611 			typ = ULONG;
    612 			if (!sflag)
    613 				ansiu = 1;
    614 			if (uq > TARG_ULONG_MAX) {
    615 				/* integer constant out of range */
    616 				warning(252);
    617 			}
    618 		}
    619 		break;
    620 	case ULONG:
    621 		if (uq > TARG_ULONG_MAX) {
    622 			/* integer constant out of range */
    623 			warning(252);
    624 		}
    625 		break;
    626 	case QUAD:
    627 		if (uq > TARG_QUAD_MAX && !tflag) {
    628 			typ = UQUAD;
    629 			if (!sflag)
    630 				ansiu = 1;
    631 		}
    632 		break;
    633 	case UQUAD:
    634 		if (uq > TARG_UQUAD_MAX) {
    635 			/* integer constant out of range */
    636 			warning(252);
    637 		}
    638 		break;
    639 		/* LINTED206: (enumeration values not handled in switch) */
    640 	case STRUCT:
    641 	case VOID:
    642 	case LDOUBLE:
    643 	case FUNC:
    644 	case ARRAY:
    645 	case PTR:
    646 	case ENUM:
    647 	case UNION:
    648 	case SIGNED:
    649 	case NOTSPEC:
    650 	case DOUBLE:
    651 	case FLOAT:
    652 	case USHORT:
    653 	case SHORT:
    654 	case UCHAR:
    655 	case SCHAR:
    656 	case CHAR:
    657 	case BOOL:
    658 	case UNSIGN:
    659 	case FCOMPLEX:
    660 	case DCOMPLEX:
    661 	case LCOMPLEX:
    662 	case COMPLEX:
    663 		break;
    664 
    665 	case NTSPEC:	/* this value unused */
    666 		break;
    667 	}
    668 
    669 	uq = (uint64_t)xsign((int64_t)uq, typ, -1);
    670 
    671 	(yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
    672 	yylval.y_val->v_ansiu = ansiu;
    673 	yylval.y_val->v_quad = (int64_t)uq;
    674 
    675 	return (T_CON);
    676 }
    677 
    678 /*
    679  * Returns 1 if t is a signed type and the value is negative.
    680  *
    681  * len is the number of significant bits. If len is -1, len is set
    682  * to the width of type t.
    683  */
    684 int
    685 sign(int64_t q, tspec_t t, int len)
    686 {
    687 
    688 	if (t == PTR || isutyp(t))
    689 		return (0);
    690 	return (msb(q, t, len));
    691 }
    692 
    693 int
    694 msb(int64_t q, tspec_t t, int len)
    695 {
    696 
    697 	if (len <= 0)
    698 		len = size(t);
    699 	return ((q & qbmasks[len - 1]) != 0);
    700 }
    701 
    702 /*
    703  * Extends the sign of q.
    704  */
    705 int64_t
    706 xsign(int64_t q, tspec_t t, int len)
    707 {
    708 
    709 	if (len <= 0)
    710 		len = size(t);
    711 
    712 	if (t == PTR || isutyp(t) || !sign(q, t, len)) {
    713 		q &= qlmasks[len];
    714 	} else {
    715 		q |= qumasks[len];
    716 	}
    717 	return (q);
    718 }
    719 
    720 /*
    721  * Convert a string representing a floating point value into its interal
    722  * representation. Type and value are returned in yylval. fcon()
    723  * (and yylex()) returns T_CON.
    724  * XXX Currently it is not possible to convert constants of type
    725  * long double which are greater than DBL_MAX.
    726  */
    727 static int
    728 fcon(void)
    729 {
    730 	const	char *cp;
    731 	int	len;
    732 	tspec_t typ;
    733 	char	c, *eptr;
    734 	double	d;
    735 	float	f = 0;
    736 
    737 	cp = yytext;
    738 	len = yyleng;
    739 
    740 	if (cp[len - 1] == 'i') {
    741 		/* imaginary, do nothing for now */
    742 		len--;
    743 	}
    744 	if ((c = cp[len - 1]) == 'f' || c == 'F') {
    745 		typ = FLOAT;
    746 		len--;
    747 	} else if (c == 'l' || c == 'L') {
    748 		typ = LDOUBLE;
    749 		len--;
    750 	} else {
    751 		if (c == 'd' || c == 'D')
    752 			len--;
    753 		typ = DOUBLE;
    754 	}
    755 
    756 	if (tflag && typ != DOUBLE) {
    757 		/* suffixes F and L are illegal in traditional C */
    758 		warning(98);
    759 	}
    760 
    761 	errno = 0;
    762 	d = strtod(cp, &eptr);
    763 	if (eptr != cp + len) {
    764 		switch (*eptr) {
    765 		/*
    766 		 * XXX: non-native non-current strtod() may not handle hex
    767 		 * floats, ignore the rest if we find traces of hex float
    768 		 * syntax...
    769 		 */
    770 		case 'p':
    771 		case 'P':
    772 		case 'x':
    773 		case 'X':
    774 			d = 0;
    775 			errno = 0;
    776 			break;
    777 		default:
    778 			LERROR("fcon(%s->%s)", cp, eptr);
    779 		}
    780 	}
    781 	if (errno != 0)
    782 		/* floating-point constant out of range */
    783 		warning(248);
    784 
    785 	if (typ == FLOAT) {
    786 		f = (float)d;
    787 		if (!finite(f)) {
    788 			/* floating-point constant out of range */
    789 			warning(248);
    790 			f = f > 0 ? FLT_MAX : -FLT_MAX;
    791 		}
    792 	}
    793 
    794 	(yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
    795 	if (typ == FLOAT) {
    796 		yylval.y_val->v_ldbl = f;
    797 	} else {
    798 		yylval.y_val->v_ldbl = d;
    799 	}
    800 
    801 	return (T_CON);
    802 }
    803 
    804 static int
    805 operator(int t, op_t o)
    806 {
    807 
    808 	yylval.y_op = o;
    809 	return (t);
    810 }
    811 
    812 /*
    813  * Called if lex found a leading \'.
    814  */
    815 static int
    816 ccon(void)
    817 {
    818 	size_t	n;
    819 	int val, c;
    820 	char	cv;
    821 
    822 	n = 0;
    823 	val = 0;
    824 	while ((c = getescc('\'')) >= 0) {
    825 		val = (val << CHAR_BIT) + c;
    826 		n++;
    827 	}
    828 	if (c == -2) {
    829 		/* unterminated character constant */
    830 		error(253);
    831 	} else {
    832 		if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
    833 			/* too many characters in character constant */
    834 			error(71);
    835 		} else if (n > 1) {
    836 			/* multi-character character constant */
    837 			warning(294);
    838 		} else if (n == 0) {
    839 			/* empty character constant */
    840 			error(73);
    841 		}
    842 	}
    843 	if (n == 1) {
    844 		cv = (char)val;
    845 		val = cv;
    846 	}
    847 
    848 	yylval.y_val = xcalloc(1, sizeof (val_t));
    849 	yylval.y_val->v_tspec = INT;
    850 	yylval.y_val->v_quad = val;
    851 
    852 	return (T_CON);
    853 }
    854 
    855 /*
    856  * Called if lex found a leading L\'
    857  */
    858 static int
    859 wccon(void)
    860 {
    861 	static	char buf[MB_LEN_MAX + 1];
    862 	size_t	i;
    863 	int c;
    864 	wchar_t	wc;
    865 
    866 	i = 0;
    867 	while ((c = getescc('\'')) >= 0) {
    868 		if (i < MB_CUR_MAX)
    869 			buf[i] = (char)c;
    870 		i++;
    871 	}
    872 
    873 	wc = 0;
    874 
    875 	if (c == -2) {
    876 		/* unterminated character constant */
    877 		error(253);
    878 	} else if (c == 0) {
    879 		/* empty character constant */
    880 		error(73);
    881 	} else {
    882 		if (i > MB_CUR_MAX) {
    883 			i = MB_CUR_MAX;
    884 			/* too many characters in character constant */
    885 			error(71);
    886 		} else {
    887 			buf[i] = '\0';
    888 			(void)mbtowc(NULL, NULL, 0);
    889 			if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
    890 				/* invalid multibyte character */
    891 				error(291);
    892 		}
    893 	}
    894 
    895 	yylval.y_val = xcalloc(1, sizeof (val_t));
    896 	yylval.y_val->v_tspec = WCHAR;
    897 	yylval.y_val->v_quad = wc;
    898 
    899 	return (T_CON);
    900 }
    901 
    902 /*
    903  * Read a character which is part of a character constant or of a string
    904  * and handle escapes.
    905  *
    906  * The Argument is the character which delimits the character constant or
    907  * string.
    908  *
    909  * Returns -1 if the end of the character constant or string is reached,
    910  * -2 if the EOF is reached, and the character otherwise.
    911  */
    912 static int
    913 getescc(int d)
    914 {
    915 	static	int pbc = -1;
    916 	int	n, c, v;
    917 
    918 	if (pbc == -1) {
    919 		c = inpc();
    920 	} else {
    921 		c = pbc;
    922 		pbc = -1;
    923 	}
    924 	if (c == d)
    925 		return (-1);
    926 	switch (c) {
    927 	case '\n':
    928 		if (tflag) {
    929 			/* newline in string or char constant */
    930 			error(254);
    931 			return (-2);
    932 		}
    933 		return (c);
    934 	case EOF:
    935 		return (-2);
    936 	case '\\':
    937 		switch (c = inpc()) {
    938 		case '"':
    939 			if (tflag && d == '\'')
    940 				/* \" inside character constant undef. ... */
    941 				warning(262);
    942 			return ('"');
    943 		case '\'':
    944 			return ('\'');
    945 		case '?':
    946 			if (tflag)
    947 				/* \? undefined in traditional C */
    948 				warning(263);
    949 			return ('?');
    950 		case '\\':
    951 			return ('\\');
    952 		case 'a':
    953 			if (tflag)
    954 				/* \a undefined in traditional C */
    955 				warning(81);
    956 			return ('\a');
    957 		case 'b':
    958 			return ('\b');
    959 		case 'f':
    960 			return ('\f');
    961 		case 'n':
    962 			return ('\n');
    963 		case 'r':
    964 			return ('\r');
    965 		case 't':
    966 			return ('\t');
    967 		case 'v':
    968 			if (tflag)
    969 				/* \v undefined in traditional C */
    970 				warning(264);
    971 			return ('\v');
    972 		case '8': case '9':
    973 			/* bad octal digit %c */
    974 			warning(77, c);
    975 			/* FALLTHROUGH */
    976 		case '0': case '1': case '2': case '3':
    977 		case '4': case '5': case '6': case '7':
    978 			n = 3;
    979 			v = 0;
    980 			do {
    981 				v = (v << 3) + (c - '0');
    982 				c = inpc();
    983 			} while (--n && isdigit(c) && (tflag || c <= '7'));
    984 			if (tflag && n > 0 && isdigit(c))
    985 				/* bad octal digit %c */
    986 				warning(77, c);
    987 			pbc = c;
    988 			if (v > TARG_UCHAR_MAX) {
    989 				/* character escape does not fit in char. */
    990 				warning(76);
    991 				v &= CHAR_MASK;
    992 			}
    993 			return (v);
    994 		case 'x':
    995 			if (tflag)
    996 				/* \x undefined in traditional C */
    997 				warning(82);
    998 			v = 0;
    999 			n = 0;
   1000 			while ((c = inpc()) >= 0 && isxdigit(c)) {
   1001 				c = isdigit(c) ?
   1002 					c - '0' : toupper(c) - 'A' + 10;
   1003 				v = (v << 4) + c;
   1004 				if (n >= 0) {
   1005 					if ((v & ~CHAR_MASK) != 0) {
   1006 						/* overflow in hex escape */
   1007 						warning(75);
   1008 						n = -1;
   1009 					} else {
   1010 						n++;
   1011 					}
   1012 				}
   1013 			}
   1014 			pbc = c;
   1015 			if (n == 0) {
   1016 				/* no hex digits follow \x */
   1017 				error(74);
   1018 			} if (n == -1) {
   1019 				v &= CHAR_MASK;
   1020 			}
   1021 			return (v);
   1022 		case '\n':
   1023 			return (getescc(d));
   1024 		case EOF:
   1025 			return (-2);
   1026 		default:
   1027 			if (isprint(c)) {
   1028 				/* dubious escape \%c */
   1029 				warning(79, c);
   1030 			} else {
   1031 				/* dubious escape \%o */
   1032 				warning(80, c);
   1033 			}
   1034 		}
   1035 	}
   1036 	return (c);
   1037 }
   1038 
   1039 /*
   1040  * Called for preprocessor directives. Currently implemented are:
   1041  *	# lineno
   1042  *	# lineno "filename"
   1043  */
   1044 static void
   1045 directive(void)
   1046 {
   1047 	const	char *cp, *fn;
   1048 	char	c, *eptr;
   1049 	size_t	fnl;
   1050 	long	ln;
   1051 	static	int first = 1;
   1052 
   1053 	/* Go to first non-whitespace after # */
   1054 	for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
   1055 		continue;
   1056 
   1057 	if (!isdigit((unsigned char)c)) {
   1058 	error:
   1059 		/* undefined or invalid # directive */
   1060 		warning(255);
   1061 		return;
   1062 	}
   1063 	ln = strtol(--cp, &eptr, 10);
   1064 	if (cp == eptr)
   1065 		goto error;
   1066 	if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
   1067 		goto error;
   1068 	while ((c = *cp++) == ' ' || c == '\t')
   1069 		continue;
   1070 	if (c != '\0') {
   1071 		if (c != '"')
   1072 			goto error;
   1073 		fn = cp;
   1074 		while ((c = *cp) != '"' && c != '\0')
   1075 			cp++;
   1076 		if (c != '"')
   1077 			goto error;
   1078 		if ((fnl = cp++ - fn) > PATH_MAX)
   1079 			goto error;
   1080 		while ((c = *cp++) == ' ' || c == '\t')
   1081 			continue;
   1082 #if 0
   1083 		if (c != '\0')
   1084 			warning("extra character(s) after directive");
   1085 #endif
   1086 
   1087 		/* empty string means stdin */
   1088 		if (fnl == 0) {
   1089 			fn = "{standard input}";
   1090 			fnl = 16;			/* strlen (fn) */
   1091 		}
   1092 		curr_pos.p_file = fnnalloc(fn, fnl);
   1093 		/*
   1094 		 * If this is the first directive, the name is the name
   1095 		 * of the C source file as specified at the command line.
   1096 		 * It is written to the output file.
   1097 		 */
   1098 		if (first) {
   1099 			csrc_pos.p_file = curr_pos.p_file;
   1100 			outsrc(curr_pos.p_file);
   1101 			first = 0;
   1102 		}
   1103 	}
   1104 	curr_pos.p_line = (int)ln - 1;
   1105 	curr_pos.p_uniq = 0;
   1106 	if (curr_pos.p_file == csrc_pos.p_file) {
   1107 		csrc_pos.p_line = (int)ln - 1;
   1108 		csrc_pos.p_uniq = 0;
   1109 	}
   1110 }
   1111 
   1112 /*
   1113  * Handle lint comments. Following comments are currently understood:
   1114  *	ARGSUSEDn
   1115  *	BITFIELDTYPE
   1116  *	CONSTCOND CONSTANTCOND CONSTANTCONDITION
   1117  *	FALLTHRU FALLTHROUGH
   1118  *	LINTLIBRARY
   1119  *	LINTEDn NOSTRICTn
   1120  *	LONGLONG
   1121  *	NOTREACHED
   1122  *	PRINTFLIKEn
   1123  *	PROTOLIB
   1124  *	SCANFLIKEn
   1125  *	VARARGSn
   1126  * If one of this comments is recognized, the arguments, if any, are
   1127  * parsed and a function which handles this comment is called.
   1128  */
   1129 static void
   1130 comment(void)
   1131 {
   1132 	int	c, lc;
   1133 	static struct {
   1134 		const	char *keywd;
   1135 		int	arg;
   1136 		void	(*func)(int);
   1137 	} keywtab[] = {
   1138 		{ "ARGSUSED",		1,	argsused	},
   1139 		{ "BITFIELDTYPE",	0,	bitfieldtype	},
   1140 		{ "CONSTCOND",		0,	constcond	},
   1141 		{ "CONSTANTCOND",	0,	constcond	},
   1142 		{ "CONSTANTCONDITION",	0,	constcond	},
   1143 		{ "FALLTHRU",		0,	fallthru	},
   1144 		{ "FALLTHROUGH",	0,	fallthru	},
   1145 		{ "LINTLIBRARY",	0,	lintlib		},
   1146 		{ "LINTED",		1,	linted		},
   1147 		{ "LONGLONG",		0,	longlong	},
   1148 		{ "NOSTRICT",		1,	linted		},
   1149 		{ "NOTREACHED",		0,	notreach	},
   1150 		{ "PRINTFLIKE",		1,	printflike	},
   1151 		{ "PROTOLIB",		1,	protolib	},
   1152 		{ "SCANFLIKE",		1,	scanflike	},
   1153 		{ "VARARGS",		1,	varargs		},
   1154 	};
   1155 	char	keywd[32];
   1156 	char	arg[32];
   1157 	size_t	l, i;
   1158 	int	a;
   1159 	int	eoc;
   1160 
   1161 	eoc = 0;
   1162 
   1163 	/* Skip white spaces after the start of the comment */
   1164 	while ((c = inpc()) != EOF && isspace(c))
   1165 		continue;
   1166 
   1167 	/* Read the potential keyword to keywd */
   1168 	l = 0;
   1169 	while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
   1170 		keywd[l++] = (char)c;
   1171 		c = inpc();
   1172 	}
   1173 	keywd[l] = '\0';
   1174 
   1175 	/* look for the keyword */
   1176 	for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
   1177 		if (strcmp(keywtab[i].keywd, keywd) == 0)
   1178 			break;
   1179 	}
   1180 	if (i == sizeof (keywtab) / sizeof (keywtab[0]))
   1181 		goto skip_rest;
   1182 
   1183 	/* skip white spaces after the keyword */
   1184 	while (c != EOF && isspace(c))
   1185 		c = inpc();
   1186 
   1187 	/* read the argument, if the keyword accepts one and there is one */
   1188 	l = 0;
   1189 	if (keywtab[i].arg) {
   1190 		while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
   1191 			arg[l++] = (char)c;
   1192 			c = inpc();
   1193 		}
   1194 	}
   1195 	arg[l] = '\0';
   1196 	a = l != 0 ? atoi(arg) : -1;
   1197 
   1198 	/* skip white spaces after the argument */
   1199 	while (c != EOF && isspace(c))
   1200 		c = inpc();
   1201 
   1202 	if (c != '*' || (c = inpc()) != '/') {
   1203 		if (keywtab[i].func != linted)
   1204 			/* extra characters in lint comment */
   1205 			warning(257);
   1206 	} else {
   1207 		/*
   1208 		 * remember that we have already found the end of the
   1209 		 * comment
   1210 		 */
   1211 		eoc = 1;
   1212 	}
   1213 
   1214 	if (keywtab[i].func != NULL)
   1215 		(*keywtab[i].func)(a);
   1216 
   1217  skip_rest:
   1218 	while (!eoc) {
   1219 		lc = c;
   1220 		if ((c = inpc()) == EOF) {
   1221 			/* unterminated comment */
   1222 			error(256);
   1223 			break;
   1224 		}
   1225 		if (lc == '*' && c == '/')
   1226 			eoc = 1;
   1227 	}
   1228 }
   1229 
   1230 /*
   1231  * Handle // style comments
   1232  */
   1233 static void
   1234 slashslashcomment(void)
   1235 {
   1236 	int c;
   1237 
   1238 	if (!Sflag && !gflag)
   1239 		/* // comments only supported in C99 */
   1240 		(void)gnuism(312, tflag ? "traditional" : "ANSI");
   1241 
   1242 	while ((c = inpc()) != EOF && c != '\n')
   1243 		continue;
   1244 }
   1245 
   1246 /*
   1247  * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
   1248  * clrwflgs() is called after function definitions and global and
   1249  * local declarations and definitions. It is also called between
   1250  * the controlling expression and the body of control statements
   1251  * (if, switch, for, while).
   1252  */
   1253 void
   1254 clrwflgs(void)
   1255 {
   1256 
   1257 	lwarn = LWARN_ALL;
   1258 	quadflg = 0;
   1259 	ccflg = 0;
   1260 }
   1261 
   1262 /*
   1263  * Strings are stored in a dynamically alloceted buffer and passed
   1264  * in yylval.y_xstrg to the parser. The parser or the routines called
   1265  * by the parser are responsible for freeing this buffer.
   1266  */
   1267 static int
   1268 string(void)
   1269 {
   1270 	u_char	*s;
   1271 	int	c;
   1272 	size_t	len, max;
   1273 	strg_t	*strg;
   1274 
   1275 	s = xmalloc(max = 64);
   1276 
   1277 	len = 0;
   1278 	while ((c = getescc('"')) >= 0) {
   1279 		/* +1 to reserve space for a trailing NUL character */
   1280 		if (len + 1 == max)
   1281 			s = xrealloc(s, max *= 2);
   1282 		s[len++] = (char)c;
   1283 	}
   1284 	s[len] = '\0';
   1285 	if (c == -2)
   1286 		/* unterminated string constant */
   1287 		error(258);
   1288 
   1289 	strg = xcalloc(1, sizeof (strg_t));
   1290 	strg->st_tspec = CHAR;
   1291 	strg->st_len = len;
   1292 	strg->st_cp = s;
   1293 
   1294 	yylval.y_strg = strg;
   1295 	return (T_STRING);
   1296 }
   1297 
   1298 static int
   1299 wcstrg(void)
   1300 {
   1301 	char	*s;
   1302 	int	c, n;
   1303 	size_t	i, wi;
   1304 	size_t	len, max, wlen;
   1305 	wchar_t	*ws;
   1306 	strg_t	*strg;
   1307 
   1308 	s = xmalloc(max = 64);
   1309 	len = 0;
   1310 	while ((c = getescc('"')) >= 0) {
   1311 		/* +1 to save space for a trailing NUL character */
   1312 		if (len + 1 >= max)
   1313 			s = xrealloc(s, max *= 2);
   1314 		s[len++] = (char)c;
   1315 	}
   1316 	s[len] = '\0';
   1317 	if (c == -2)
   1318 		/* unterminated string constant */
   1319 		error(258);
   1320 
   1321 	/* get length of wide-character string */
   1322 	(void)mblen(NULL, 0);
   1323 	for (i = 0, wlen = 0; i < len; i += n, wlen++) {
   1324 		if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
   1325 			/* invalid multibyte character */
   1326 			error(291);
   1327 			break;
   1328 		}
   1329 		if (n == 0)
   1330 			n = 1;
   1331 	}
   1332 
   1333 	ws = xmalloc((wlen + 1) * sizeof (wchar_t));
   1334 
   1335 	/* convert from multibyte to wide char */
   1336 	(void)mbtowc(NULL, NULL, 0);
   1337 	for (i = 0, wi = 0; i < len; i += n, wi++) {
   1338 		if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
   1339 			break;
   1340 		if (n == 0)
   1341 			n = 1;
   1342 	}
   1343 	ws[wi] = 0;
   1344 	free(s);
   1345 
   1346 	strg = xcalloc(1, sizeof (strg_t));
   1347 	strg->st_tspec = WCHAR;
   1348 	strg->st_len = wlen;
   1349 	strg->st_wcp = ws;
   1350 
   1351 	yylval.y_strg = strg;
   1352 	return (T_STRING);
   1353 }
   1354 
   1355 /*
   1356  * As noted above the scanner does not create new symbol table entries
   1357  * for symbols it cannot find in the symbol table. This is to avoid
   1358  * putting undeclared symbols into the symbol table if a syntax error
   1359  * occurs.
   1360  *
   1361  * getsym() is called as soon as it is probably ok to put the symbol to
   1362  * the symbol table. This does not mean that it is not possible that
   1363  * symbols are put to the symbol table which are than not completely
   1364  * declared due to syntax errors. To avoid too many problems in this
   1365  * case symbols get type int in getsym().
   1366  *
   1367  * XXX calls to getsym() should be delayed until decl1*() is called
   1368  */
   1369 sym_t *
   1370 getsym(sbuf_t *sb)
   1371 {
   1372 	dinfo_t	*di;
   1373 	char	*s;
   1374 	sym_t	*sym;
   1375 
   1376 	sym = sb->sb_sym;
   1377 
   1378 	/*
   1379 	 * During member declaration it is possible that name() looked
   1380 	 * for symbols of type FVFT, although it should have looked for
   1381 	 * symbols of type FTAG. Same can happen for labels. Both cases
   1382 	 * are compensated here.
   1383 	 */
   1384 	if (symtyp == FMOS || symtyp == FLAB) {
   1385 		if (sym == NULL || sym->s_kind == FVFT)
   1386 			sym = search(sb);
   1387 	}
   1388 
   1389 	if (sym != NULL) {
   1390 		if (sym->s_kind != symtyp)
   1391 			LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
   1392 		symtyp = FVFT;
   1393 		freesb(sb);
   1394 		return (sym);
   1395 	}
   1396 
   1397 	/* create a new symbol table entry */
   1398 
   1399 	/* labels must always be allocated at level 1 (outhermost block) */
   1400 	if (symtyp == FLAB) {
   1401 		sym = getlblk(1, sizeof (sym_t));
   1402 		s = getlblk(1, sb->sb_len + 1);
   1403 		(void)memcpy(s, sb->sb_name, sb->sb_len + 1);
   1404 		sym->s_name = s;
   1405 		sym->s_blklev = 1;
   1406 		di = dcs;
   1407 		while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
   1408 			di = di->d_nxt;
   1409 		if (di->d_ctx != AUTO)
   1410 			LERROR("storesym()");
   1411 	} else {
   1412 		sym = getblk(sizeof (sym_t));
   1413 		sym->s_name = sb->sb_name;
   1414 		sym->s_blklev = blklev;
   1415 		di = dcs;
   1416 	}
   1417 
   1418 	UNIQUE_CURR_POS(sym->s_dpos);
   1419 	if ((sym->s_kind = symtyp) != FLAB)
   1420 		sym->s_type = gettyp(INT);
   1421 
   1422 	symtyp = FVFT;
   1423 
   1424 	if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
   1425 		symtab[sb->sb_hash]->s_rlink = &sym->s_link;
   1426 	sym->s_rlink = &symtab[sb->sb_hash];
   1427 	symtab[sb->sb_hash] = sym;
   1428 
   1429 	*di->d_ldlsym = sym;
   1430 	di->d_ldlsym = &sym->s_dlnxt;
   1431 
   1432 	freesb(sb);
   1433 	return (sym);
   1434 }
   1435 
   1436 /*
   1437  * Construct a temporary symbol. The symbol starts with a digit, so that
   1438  * it is illegal.
   1439  */
   1440 sym_t *
   1441 mktempsym(type_t *t)
   1442 {
   1443 	static int n = 0;
   1444 	int h;
   1445 	char *s = getlblk(blklev, 64);
   1446 	sym_t *sym = getblk(sizeof (sym_t));
   1447 
   1448 	(void)snprintf(s, 64, "%.8d_tmp", n++);
   1449 	h = hash(s);
   1450 
   1451 	sym->s_name = s;
   1452 	sym->s_type = t;
   1453 	sym->s_blklev = blklev;
   1454 	sym->s_scl = AUTO;
   1455 	sym->s_kind = FVFT;
   1456 	sym->s_used = 1;
   1457 	sym->s_set = 1;
   1458 
   1459 	if ((sym->s_link = symtab[h]) != NULL)
   1460 		symtab[h]->s_rlink = &sym->s_link;
   1461 	sym->s_rlink = &symtab[h];
   1462 	symtab[h] = sym;
   1463 
   1464 	*dcs->d_ldlsym = sym;
   1465 	dcs->d_ldlsym = &sym->s_dlnxt;
   1466 
   1467 	return sym;
   1468 }
   1469 
   1470 /*
   1471  * Remove a symbol forever from the symbol table. s_blklev
   1472  * is set to -1 to avoid that the symbol will later be put
   1473  * back to the symbol table.
   1474  */
   1475 void
   1476 rmsym(sym_t *sym)
   1477 {
   1478 
   1479 	if ((*sym->s_rlink = sym->s_link) != NULL)
   1480 		sym->s_link->s_rlink = sym->s_rlink;
   1481 	sym->s_blklev = -1;
   1482 	sym->s_link = NULL;
   1483 }
   1484 
   1485 /*
   1486  * Remove a list of symbols declared at one level from the symbol
   1487  * table.
   1488  */
   1489 void
   1490 rmsyms(sym_t *syms)
   1491 {
   1492 	sym_t	*sym;
   1493 
   1494 	for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
   1495 		if (sym->s_blklev != -1) {
   1496 			if ((*sym->s_rlink = sym->s_link) != NULL)
   1497 				sym->s_link->s_rlink = sym->s_rlink;
   1498 			sym->s_link = NULL;
   1499 			sym->s_rlink = NULL;
   1500 		}
   1501 	}
   1502 }
   1503 
   1504 /*
   1505  * Put a symbol into the symbol table
   1506  */
   1507 void
   1508 inssym(int bl, sym_t *sym)
   1509 {
   1510 	int	h;
   1511 
   1512 	h = hash(sym->s_name);
   1513 	if ((sym->s_link = symtab[h]) != NULL)
   1514 		symtab[h]->s_rlink = &sym->s_link;
   1515 	sym->s_rlink = &symtab[h];
   1516 	symtab[h] = sym;
   1517 	sym->s_blklev = bl;
   1518 	if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
   1519 		LERROR("inssym()");
   1520 }
   1521 
   1522 /*
   1523  * Called at level 0 after syntax errors
   1524  * Removes all symbols which are not declared at level 0 from the
   1525  * symbol table. Also frees all memory which is not associated with
   1526  * level 0.
   1527  */
   1528 void
   1529 cleanup(void)
   1530 {
   1531 	sym_t	*sym, *nsym;
   1532 	int	i;
   1533 
   1534 	for (i = 0; i < HSHSIZ1; i++) {
   1535 		for (sym = symtab[i]; sym != NULL; sym = nsym) {
   1536 			nsym = sym->s_link;
   1537 			if (sym->s_blklev >= 1) {
   1538 				if ((*sym->s_rlink = nsym) != NULL)
   1539 					nsym->s_rlink = sym->s_rlink;
   1540 			}
   1541 		}
   1542 	}
   1543 
   1544 	for (i = mblklev; i > 0; i--)
   1545 		freelblk(i);
   1546 }
   1547 
   1548 /*
   1549  * Create a new symbol with the name of an existing symbol.
   1550  */
   1551 sym_t *
   1552 pushdown(sym_t *sym)
   1553 {
   1554 	int	h;
   1555 	sym_t	*nsym;
   1556 
   1557 	h = hash(sym->s_name);
   1558 	nsym = getblk(sizeof (sym_t));
   1559 	if (sym->s_blklev > blklev)
   1560 		LERROR("pushdown()");
   1561 	nsym->s_name = sym->s_name;
   1562 	UNIQUE_CURR_POS(nsym->s_dpos);
   1563 	nsym->s_kind = sym->s_kind;
   1564 	nsym->s_blklev = blklev;
   1565 
   1566 	if ((nsym->s_link = symtab[h]) != NULL)
   1567 		symtab[h]->s_rlink = &nsym->s_link;
   1568 	nsym->s_rlink = &symtab[h];
   1569 	symtab[h] = nsym;
   1570 
   1571 	*dcs->d_ldlsym = nsym;
   1572 	dcs->d_ldlsym = &nsym->s_dlnxt;
   1573 
   1574 	return (nsym);
   1575 }
   1576 
   1577 /*
   1578  * Free any dynamically allocated memory referenced by
   1579  * the value stack or yylval.
   1580  * The type of information in yylval is described by tok.
   1581  */
   1582 void
   1583 freeyyv(void *sp, int tok)
   1584 {
   1585 	if (tok == T_NAME || tok == T_TYPENAME) {
   1586 		sbuf_t *sb = *(sbuf_t **)sp;
   1587 		freesb(sb);
   1588 	} else if (tok == T_CON) {
   1589 		val_t *val = *(val_t **)sp;
   1590 		free(val);
   1591 	} else if (tok == T_STRING) {
   1592 		strg_t *strg = *(strg_t **)sp;
   1593 		if (strg->st_tspec == CHAR) {
   1594 			free(strg->st_cp);
   1595 		} else if (strg->st_tspec == WCHAR) {
   1596 			free(strg->st_wcp);
   1597 		} else {
   1598 			LERROR("fryylv()");
   1599 		}
   1600 		free(strg);
   1601 	}
   1602 }
   1603