Home | History | Annotate | Line # | Download | only in lint1
lex.c revision 1.127
      1 /* $NetBSD: lex.c,v 1.127 2022/04/30 21:38:03 rillig Exp $ */
      2 
      3 /*
      4  * Copyright (c) 1996 Christopher G. Demetriou.  All Rights Reserved.
      5  * Copyright (c) 1994, 1995 Jochen Pohl
      6  * All Rights Reserved.
      7  *
      8  * Redistribution and use in source and binary forms, with or without
      9  * modification, are permitted provided that the following conditions
     10  * are met:
     11  * 1. Redistributions of source code must retain the above copyright
     12  *    notice, this list of conditions and the following disclaimer.
     13  * 2. Redistributions in binary form must reproduce the above copyright
     14  *    notice, this list of conditions and the following disclaimer in the
     15  *    documentation and/or other materials provided with the distribution.
     16  * 3. All advertising materials mentioning features or use of this software
     17  *    must display the following acknowledgement:
     18  *      This product includes software developed by Jochen Pohl for
     19  *      The NetBSD Project.
     20  * 4. The name of the author may not be used to endorse or promote products
     21  *    derived from this software without specific prior written permission.
     22  *
     23  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
     24  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
     25  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
     26  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
     27  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
     28  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
     29  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
     30  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
     31  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
     32  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     33  */
     34 
     35 #if HAVE_NBTOOL_CONFIG_H
     36 #include "nbtool_config.h"
     37 #endif
     38 
     39 #include <sys/cdefs.h>
     40 #if defined(__RCSID) && !defined(lint)
     41 __RCSID("$NetBSD: lex.c,v 1.127 2022/04/30 21:38:03 rillig Exp $");
     42 #endif
     43 
     44 #include <ctype.h>
     45 #include <errno.h>
     46 #include <float.h>
     47 #include <limits.h>
     48 #include <math.h>
     49 #include <stdlib.h>
     50 #include <string.h>
     51 
     52 #include "lint1.h"
     53 #include "cgram.h"
     54 
     55 #define CHAR_MASK	((1U << CHAR_SIZE) - 1)
     56 
     57 
     58 /* Current position (it's also updated when an included file is parsed) */
     59 pos_t	curr_pos = { "", 1, 0 };
     60 
     61 /*
     62  * Current position in C source (not updated when an included file is
     63  * parsed).
     64  */
     65 pos_t	csrc_pos = { "", 1, 0 };
     66 
     67 bool in_gcc_attribute;
     68 bool in_system_header;
     69 
     70 /*
     71  * Valid values for 'since' are 78, 90, 99, 11.
     72  *
     73  * As of 2022-04-30, lint treats 11 like 99, in order to provide good error
     74  * messages instead of a simple parse error.  If the keyword '_Generic' were
     75  * not defined, it would be interpreted as an implicit function call, leading
     76  * to a parse error.
     77  */
     78 #define kwdef(name, token, scl, tspec, tqual,	since, gcc, attr, deco) \
     79 	{ \
     80 		name, token, scl, tspec, tqual, \
     81 		(since) == 90, \
     82 		/* CONSTCOND */ (since) == 99 || (since) == 11, \
     83 		(gcc) > 0, (attr) > 0, \
     84 		((deco) & 1) != 0, ((deco) & 2) != 0, ((deco) & 4) != 0, \
     85 	}
     86 #define kwdef_token(name, token,		since, gcc, deco) \
     87 	kwdef(name, token, 0, 0, 0,		since, gcc, 0, deco)
     88 #define kwdef_sclass(name, sclass,		since, gcc, deco) \
     89 	kwdef(name, T_SCLASS, sclass, 0, 0,	since, gcc, 0, deco)
     90 #define kwdef_type(name, tspec,			since, gcc, deco) \
     91 	kwdef(name, T_TYPE, 0, tspec, 0,	since, gcc, 0, deco)
     92 #define kwdef_tqual(name, tqual,		since, gcc, deco) \
     93 	kwdef(name, T_QUAL, 0, 0, tqual,	since, gcc, 0, deco)
     94 #define kwdef_keyword(name, token) \
     95 	kwdef(name, token, 0, 0, 0,		78, 0, 0, 1)
     96 #define kwdef_gcc_attr(name, token) \
     97 	kwdef(name, token, 0, 0, 0,		78, 1, 1, 5)
     98 
     99 /* During initialization, these keywords are written to the symbol table. */
    100 static const struct keyword {
    101 	const	char *kw_name;	/* keyword */
    102 	int	kw_token;	/* token returned by yylex() */
    103 	scl_t	kw_scl;		/* storage class if kw_token T_SCLASS */
    104 	tspec_t	kw_tspec;	/* type spec. if kw_token
    105 				 * T_TYPE or T_STRUCT_OR_UNION */
    106 	tqual_t	kw_tqual;	/* type qual. if kw_token T_QUAL */
    107 	bool	kw_c90:1;	/* C90 keyword */
    108 	bool	kw_c99_or_c11:1; /* C99 or C11 keyword */
    109 	bool	kw_gcc:1;	/* GCC keyword */
    110 	bool	kw_attr:1;	/* GCC attribute */
    111 	bool	kw_plain:1;	/* 'name' */
    112 	bool	kw_leading:1;	/* '__name' */
    113 	bool	kw_both:1;	/* '__name__' */
    114 } keywords[] = {
    115 	kwdef_keyword(	"_Alignas",	T_ALIGNAS),
    116 	kwdef_keyword(	"_Alignof",	T_ALIGNOF),
    117 	kwdef_token(	"__alignof__",	T_ALIGNOF,		78,0,1),
    118 	kwdef_token(	"asm",		T_ASM,			78,1,7),
    119 	kwdef_token(	"attribute",	T_ATTRIBUTE,		78,1,6),
    120 	kwdef_sclass(	"auto",		AUTO,			78,0,1),
    121 	kwdef_type(	"_Bool",	BOOL,			99,0,1),
    122 	kwdef_keyword(	"break",	T_BREAK),
    123 	kwdef_token(	"__builtin_offsetof", T_BUILTIN_OFFSETOF, 78,1,1),
    124 	kwdef_keyword(	"case",		T_CASE),
    125 	kwdef_type(	"char",		CHAR,			78,0,1),
    126 	kwdef_type(	"_Complex",	COMPLEX,		99,0,1),
    127 	kwdef_tqual(	"const",	CONST,			90,0,7),
    128 	kwdef_keyword(	"continue",	T_CONTINUE),
    129 	kwdef_keyword(	"default",	T_DEFAULT),
    130 	kwdef_keyword(	"do",		T_DO),
    131 	kwdef_type(	"double",	DOUBLE,			78,0,1),
    132 	kwdef_keyword(	"else",		T_ELSE),
    133 	kwdef_keyword(	"enum",		T_ENUM),
    134 	kwdef_token(	"__extension__",T_EXTENSION,		78,1,1),
    135 	kwdef_sclass(	"extern",	EXTERN,			78,0,1),
    136 	kwdef_type(	"float",	FLOAT,			78,0,1),
    137 	kwdef_keyword(	"for",		T_FOR),
    138 	kwdef_token(	"_Generic",	T_GENERIC,		11,0,1),
    139 	kwdef_keyword(	"goto",		T_GOTO),
    140 	kwdef_keyword(	"if",		T_IF),
    141 	kwdef_token(	"__imag__",	T_IMAG,			78,1,1),
    142 	kwdef_sclass(	"inline",	INLINE,			99,0,7),
    143 	kwdef_type(	"int",		INT,			78,0,1),
    144 #ifdef INT128_SIZE
    145 	kwdef_type(	"__int128_t",	INT128,			99,0,1),
    146 #endif
    147 	kwdef_type(	"long",		LONG,			78,0,1),
    148 	kwdef_token(	"_Noreturn",	T_NORETURN,		11,0,1),
    149 	kwdef_token(	"__packed",	T_PACKED,		78,0,1),
    150 	kwdef_token(	"__real__",	T_REAL,			78,1,1),
    151 	kwdef_sclass(	"register",	REG,			78,0,1),
    152 	kwdef_tqual(	"restrict",	RESTRICT,		99,0,7),
    153 	kwdef_keyword(	"return",	T_RETURN),
    154 	kwdef(		"section",	T_AT_SECTION,	0,0,0,	78,1,1,7),
    155 	kwdef_type(	"short",	SHORT,			78,0,1),
    156 	kwdef_type(	"signed",	SIGNED,			90,0,3),
    157 	kwdef_keyword(	"sizeof",	T_SIZEOF),
    158 	kwdef_sclass(	"static",	STATIC,			78,0,1),
    159 	kwdef_keyword(	"_Static_assert",	T_STATIC_ASSERT),
    160 	kwdef("struct",	T_STRUCT_OR_UNION, 0,	STRUCT,	0,	78,0,0,1),
    161 	kwdef_keyword(	"switch",	T_SWITCH),
    162 	kwdef_token(	"__symbolrename",	T_SYMBOLRENAME,	78,0,1),
    163 	kwdef_tqual(	"__thread",	THREAD,			78,1,1),
    164 	kwdef_tqual(	"_Thread_local", THREAD,		11,0,1),
    165 	kwdef_sclass(	"typedef",	TYPEDEF,		78,0,1),
    166 	kwdef_token(	"typeof",	T_TYPEOF,		78,1,7),
    167 #ifdef INT128_SIZE
    168 	kwdef_type(	"__uint128_t",	UINT128,		99,0,1),
    169 #endif
    170 	kwdef("union",	T_STRUCT_OR_UNION, 0,	UNION,	0,	78,0,0,1),
    171 	kwdef_type(	"unsigned",	UNSIGN,			78,0,1),
    172 	kwdef_type(	"void",		VOID,			78,0,1),
    173 	kwdef_tqual(	"volatile",	VOLATILE,		90,0,7),
    174 	kwdef_keyword(	"while",	T_WHILE),
    175 
    176 	kwdef_gcc_attr(	"alias",	T_AT_ALIAS),
    177 	kwdef_gcc_attr(	"aligned",	T_AT_ALIGNED),
    178 	kwdef_gcc_attr(	"alloc_size",	T_AT_ALLOC_SIZE),
    179 	kwdef_gcc_attr(	"always_inline",T_AT_ALWAYS_INLINE),
    180 	kwdef_gcc_attr(	"bounded",	T_AT_BOUNDED),
    181 	kwdef_gcc_attr(	"buffer",	T_AT_BUFFER),
    182 	kwdef_gcc_attr(	"cold",		T_AT_COLD),
    183 	kwdef_gcc_attr(	"common",	T_AT_COMMON),
    184 	kwdef_gcc_attr(	"constructor",	T_AT_CONSTRUCTOR),
    185 	kwdef_gcc_attr(	"deprecated",	T_AT_DEPRECATED),
    186 	kwdef_gcc_attr(	"destructor",	T_AT_DESTRUCTOR),
    187 	kwdef_gcc_attr(	"disable_sanitizer_instrumentation",
    188 	    T_AT_DISABLE_SANITIZER_INSTRUMENTATION),
    189 	kwdef_gcc_attr(	"fallthrough",	T_AT_FALLTHROUGH),
    190 	kwdef_gcc_attr(	"format",	T_AT_FORMAT),
    191 	kwdef_gcc_attr(	"format_arg",	T_AT_FORMAT_ARG),
    192 	kwdef_gcc_attr(	"gnu_inline",	T_AT_GNU_INLINE),
    193 	kwdef_gcc_attr(	"gnu_printf",	T_AT_FORMAT_GNU_PRINTF),
    194 	kwdef_gcc_attr(	"hot",		T_AT_HOT),
    195 	kwdef_gcc_attr(	"malloc",	T_AT_MALLOC),
    196 	kwdef_gcc_attr(	"may_alias",	T_AT_MAY_ALIAS),
    197 	kwdef_gcc_attr(	"minbytes",	T_AT_MINBYTES),
    198 	kwdef_gcc_attr(	"mode",		T_AT_MODE),
    199 	kwdef_gcc_attr("no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION),
    200 	kwdef_gcc_attr(	"no_sanitize",	T_AT_NO_SANITIZE),
    201 	kwdef_gcc_attr(	"no_sanitize_thread", T_AT_NO_SANITIZE_THREAD),
    202 	kwdef_gcc_attr(	"noinline",	T_AT_NOINLINE),
    203 	kwdef_gcc_attr(	"nonnull",	T_AT_NONNULL),
    204 	kwdef_gcc_attr(	"nonstring",	T_AT_NONSTRING),
    205 	kwdef_gcc_attr(	"noreturn",	T_AT_NORETURN),
    206 	kwdef_gcc_attr(	"nothrow",	T_AT_NOTHROW),
    207 	kwdef_gcc_attr(	"optimize",	T_AT_OPTIMIZE),
    208 	kwdef_gcc_attr(	"optnone",	T_AT_OPTNONE),
    209 	kwdef_gcc_attr(	"packed",	T_AT_PACKED),
    210 	kwdef_gcc_attr(	"pcs",		T_AT_PCS),
    211 	kwdef_gcc_attr(	"printf",	T_AT_FORMAT_PRINTF),
    212 	kwdef_gcc_attr(	"pure",		T_AT_PURE),
    213 	kwdef_gcc_attr(	"regparm",	T_AT_REGPARM),
    214 	kwdef_gcc_attr(	"returns_nonnull", T_AT_RETURNS_NONNULL),
    215 	kwdef_gcc_attr(	"returns_twice", T_AT_RETURNS_TWICE),
    216 	kwdef_gcc_attr(	"scanf",	T_AT_FORMAT_SCANF),
    217 	kwdef_gcc_attr(	"sentinel",	T_AT_SENTINEL),
    218 	kwdef_gcc_attr(	"strfmon",	T_AT_FORMAT_STRFMON),
    219 	kwdef_gcc_attr(	"strftime",	T_AT_FORMAT_STRFTIME),
    220 	kwdef_gcc_attr(	"string",	T_AT_STRING),
    221 	kwdef_gcc_attr(	"syslog",	T_AT_FORMAT_SYSLOG),
    222 	kwdef_gcc_attr(	"target",	T_AT_TARGET),
    223 	kwdef_gcc_attr(	"tls_model",	T_AT_TLS_MODEL),
    224 	kwdef_gcc_attr(	"transparent_union", T_AT_TUNION),
    225 	kwdef_gcc_attr(	"unused",	T_AT_UNUSED),
    226 	kwdef_gcc_attr(	"used",		T_AT_USED),
    227 	kwdef_gcc_attr(	"visibility",	T_AT_VISIBILITY),
    228 	kwdef_gcc_attr(	"warn_unused_result", T_AT_WARN_UNUSED_RESULT),
    229 	kwdef_gcc_attr(	"weak",		T_AT_WEAK),
    230 #undef kwdef
    231 #undef kwdef_token
    232 #undef kwdef_sclass
    233 #undef kwdef_type
    234 #undef kwdef_tqual
    235 #undef kwdef_keyword
    236 #undef kwdef_gcc_attr
    237 };
    238 
    239 /* Symbol table */
    240 static	sym_t	*symtab[HSHSIZ1];
    241 
    242 /* type of next expected symbol */
    243 symt_t	symtyp;
    244 
    245 
    246 static	int	get_escaped_char(int);
    247 
    248 
    249 static unsigned int
    250 hash(const char *s)
    251 {
    252 	unsigned int v;
    253 	const char *p;
    254 
    255 	v = 0;
    256 	for (p = s; *p != '\0'; p++) {
    257 		v = (v << 4) + (unsigned char)*p;
    258 		v ^= v >> 28;
    259 	}
    260 	return v % HSHSIZ1;
    261 }
    262 
    263 static void
    264 symtab_add(sym_t *sym)
    265 {
    266 	unsigned int h;
    267 
    268 	h = hash(sym->s_name);
    269 	if ((sym->s_symtab_next = symtab[h]) != NULL)
    270 		symtab[h]->s_symtab_ref = &sym->s_symtab_next;
    271 	sym->s_symtab_ref = &symtab[h];
    272 	symtab[h] = sym;
    273 }
    274 
    275 static sym_t *
    276 symtab_search(sbuf_t *sb)
    277 {
    278 
    279 	unsigned int h = hash(sb->sb_name);
    280 	for (sym_t *sym = symtab[h]; sym != NULL; sym = sym->s_symtab_next) {
    281 		if (strcmp(sym->s_name, sb->sb_name) != 0)
    282 			continue;
    283 
    284 		const struct keyword *kw = sym->s_keyword;
    285 		if (kw != NULL && !kw->kw_attr)
    286 			return sym;
    287 		if (kw != NULL && in_gcc_attribute)
    288 			return sym;
    289 		if (kw == NULL && !in_gcc_attribute && sym->s_kind == symtyp)
    290 			return sym;
    291 	}
    292 
    293 	return NULL;
    294 }
    295 
    296 static void
    297 symtab_remove(sym_t *sym)
    298 {
    299 
    300 	if ((*sym->s_symtab_ref = sym->s_symtab_next) != NULL)
    301 		sym->s_symtab_next->s_symtab_ref = sym->s_symtab_ref;
    302 	sym->s_symtab_next = NULL;
    303 }
    304 
    305 static void
    306 symtab_remove_locals(void)
    307 {
    308 
    309 	for (size_t i = 0; i < HSHSIZ1; i++) {
    310 		for (sym_t *sym = symtab[i]; sym != NULL; ) {
    311 			sym_t *next = sym->s_symtab_next;
    312 			if (sym->s_block_level >= 1)
    313 				symtab_remove(sym);
    314 			sym = next;
    315 		}
    316 	}
    317 }
    318 
    319 #ifdef DEBUG
    320 static int
    321 sym_by_name(const void *va, const void *vb)
    322 {
    323 	const sym_t *a = *(const sym_t *const *)va;
    324 	const sym_t *b = *(const sym_t *const *)vb;
    325 
    326 	return strcmp(a->s_name, b->s_name);
    327 }
    328 
    329 struct syms {
    330 	const sym_t **items;
    331 	size_t len;
    332 	size_t cap;
    333 };
    334 
    335 static void
    336 syms_add(struct syms *syms, const sym_t *sym)
    337 {
    338 	while (syms->len >= syms->cap) {
    339 		syms->cap *= 2;
    340 		syms->items = xrealloc(syms->items,
    341 		    syms->cap * sizeof(syms->items[0]));
    342 	}
    343 	syms->items[syms->len++] = sym;
    344 }
    345 
    346 void
    347 debug_symtab(void)
    348 {
    349 	struct syms syms = { xcalloc(64, sizeof(syms.items[0])), 0, 64 };
    350 
    351 	for (int level = -1;; level++) {
    352 		bool more = false;
    353 		size_t n = sizeof(symtab) / sizeof(symtab[0]);
    354 
    355 		syms.len = 0;
    356 		for (size_t i = 0; i < n; i++) {
    357 			for (sym_t *sym = symtab[i]; sym != NULL;) {
    358 				if (sym->s_block_level == level &&
    359 				    sym->s_keyword == NULL)
    360 					syms_add(&syms, sym);
    361 				if (sym->s_block_level > level)
    362 					more = true;
    363 				sym = sym->s_symtab_next;
    364 			}
    365 		}
    366 
    367 		if (syms.len > 0) {
    368 			debug_printf("symbol table level %d\n", level);
    369 			debug_indent_inc();
    370 			qsort(syms.items, syms.len, sizeof(syms.items[0]),
    371 			    sym_by_name);
    372 			for (size_t i = 0; i < syms.len; i++)
    373 				debug_sym("", syms.items[i], "\n");
    374 			debug_indent_dec();
    375 
    376 			lint_assert(level != -1);
    377 		}
    378 
    379 		if (!more)
    380 			break;
    381 	}
    382 
    383 	free(syms.items);
    384 }
    385 #endif
    386 
    387 static void
    388 add_keyword(const struct keyword *kw, bool leading, bool trailing)
    389 {
    390 	sym_t *sym;
    391 	char buf[256];
    392 	const char *name;
    393 
    394 	if (!leading && !trailing) {
    395 		name = kw->kw_name;
    396 	} else {
    397 		(void)snprintf(buf, sizeof(buf), "%s%s%s",
    398 		    leading ? "__" : "", kw->kw_name, trailing ? "__" : "");
    399 		name = xstrdup(buf);
    400 	}
    401 
    402 	sym = block_zero_alloc(sizeof(*sym));
    403 	sym->s_name = name;
    404 	sym->s_keyword = kw;
    405 	sym->u.s_keyword.sk_token = kw->kw_token;
    406 	if (kw->kw_token == T_TYPE || kw->kw_token == T_STRUCT_OR_UNION) {
    407 		sym->u.s_keyword.sk_tspec = kw->kw_tspec;
    408 	} else if (kw->kw_token == T_SCLASS) {
    409 		sym->s_scl = kw->kw_scl;
    410 	} else if (kw->kw_token == T_QUAL) {
    411 		sym->u.s_keyword.sk_qualifier = kw->kw_tqual;
    412 	}
    413 
    414 	symtab_add(sym);
    415 }
    416 
    417 static bool
    418 is_keyword_known(const struct keyword *kw)
    419 {
    420 
    421 	if ((kw->kw_c90 || kw->kw_c99_or_c11) && !allow_c90)
    422 		return false;
    423 
    424 	/*
    425 	 * In the 1990s, GCC defined several keywords that were later
    426 	 * incorporated into C99, therefore in GCC mode, all C99 keywords are
    427 	 * made available.  The C11 keywords are made available as well, but
    428 	 * there are so few that they don't matter practically.
    429 	 */
    430 	if (allow_gcc)
    431 		return true;
    432 	if (kw->kw_gcc)
    433 		return false;
    434 
    435 	if (kw->kw_c99_or_c11 && !allow_c99)
    436 		return false;
    437 	return true;
    438 }
    439 
    440 /*
    441  * All keywords are written to the symbol table. This saves us looking
    442  * in an extra table for each name we found.
    443  */
    444 void
    445 initscan(void)
    446 {
    447 	const struct keyword *kw, *end;
    448 
    449 	end = keywords + sizeof(keywords) / sizeof(keywords[0]);
    450 	for (kw = keywords; kw != end; kw++) {
    451 		if (!is_keyword_known(kw))
    452 			continue;
    453 		if (kw->kw_plain)
    454 			add_keyword(kw, false, false);
    455 		if (kw->kw_leading)
    456 			add_keyword(kw, true, false);
    457 		if (kw->kw_both)
    458 			add_keyword(kw, true, true);
    459 	}
    460 }
    461 
    462 /*
    463  * Read a character and ensure that it is positive (except EOF).
    464  * Increment line count(s) if necessary.
    465  */
    466 static int
    467 inpc(void)
    468 {
    469 	int	c;
    470 
    471 	if ((c = lex_input()) == EOF)
    472 		return c;
    473 	c &= CHAR_MASK;
    474 	if (c == '\0')
    475 		return EOF;	/* lex returns 0 on EOF. */
    476 	if (c == '\n')
    477 		lex_next_line();
    478 	return c;
    479 }
    480 
    481 static int
    482 lex_keyword(sym_t *sym)
    483 {
    484 	int	t;
    485 
    486 	if ((t = sym->u.s_keyword.sk_token) == T_SCLASS) {
    487 		yylval.y_scl = sym->s_scl;
    488 	} else if (t == T_TYPE || t == T_STRUCT_OR_UNION) {
    489 		yylval.y_tspec = sym->u.s_keyword.sk_tspec;
    490 	} else if (t == T_QUAL) {
    491 		yylval.y_tqual = sym->u.s_keyword.sk_qualifier;
    492 	}
    493 	return t;
    494 }
    495 
    496 /*
    497  * Lex has found a letter followed by zero or more letters or digits.
    498  * It looks for a symbol in the symbol table with the same name. This
    499  * symbol must either be a keyword or a symbol of the type required by
    500  * symtyp (label, member, tag, ...).
    501  *
    502  * If it is a keyword, the token is returned. In some cases it is described
    503  * more deeply by data written to yylval.
    504  *
    505  * If it is a symbol, T_NAME is returned and the name is stored in yylval.
    506  * If there is already a symbol of the same name and type in the symbol
    507  * table, yylval.y_name->sb_sym points there.
    508  */
    509 extern int
    510 lex_name(const char *yytext, size_t yyleng)
    511 {
    512 	char	*s;
    513 	sbuf_t	*sb;
    514 	sym_t	*sym;
    515 	int	tok;
    516 
    517 	sb = xmalloc(sizeof(*sb));
    518 	sb->sb_name = yytext;
    519 	sb->sb_len = yyleng;
    520 	if ((sym = symtab_search(sb)) != NULL && sym->s_keyword != NULL) {
    521 		free(sb);
    522 		return lex_keyword(sym);
    523 	}
    524 
    525 	sb->sb_sym = sym;
    526 
    527 	if (sym != NULL) {
    528 		lint_assert(block_level >= sym->s_block_level);
    529 		sb->sb_name = sym->s_name;
    530 		tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
    531 	} else {
    532 		s = block_zero_alloc(yyleng + 1);
    533 		(void)memcpy(s, yytext, yyleng + 1);
    534 		sb->sb_name = s;
    535 		tok = T_NAME;
    536 	}
    537 
    538 	yylval.y_name = sb;
    539 	return tok;
    540 }
    541 
    542 /*
    543  * Convert a string representing an integer into internal representation.
    544  * Return T_CON, storing the numeric value in yylval, for yylex.
    545  */
    546 int
    547 lex_integer_constant(const char *yytext, size_t yyleng, int base)
    548 {
    549 	int	l_suffix, u_suffix;
    550 	size_t	len;
    551 	const	char *cp;
    552 	char	c, *eptr;
    553 	tspec_t	typ;
    554 	bool	ansiu;
    555 	bool	warned = false;
    556 	uint64_t uq = 0;
    557 
    558 	/* C11 6.4.4.1p5 */
    559 	static const tspec_t suffix_type[2][3] = {
    560 		{ INT,  LONG,  QUAD, },
    561 		{ UINT, ULONG, UQUAD, }
    562 	};
    563 
    564 	cp = yytext;
    565 	len = yyleng;
    566 
    567 	/* skip 0[xX] or 0[bB] */
    568 	if (base == 16 || base == 2) {
    569 		cp += 2;
    570 		len -= 2;
    571 	}
    572 
    573 	/* read suffixes */
    574 	l_suffix = u_suffix = 0;
    575 	for (;;) {
    576 		if ((c = cp[len - 1]) == 'l' || c == 'L') {
    577 			l_suffix++;
    578 		} else if (c == 'u' || c == 'U') {
    579 			u_suffix++;
    580 		} else {
    581 			break;
    582 		}
    583 		len--;
    584 	}
    585 	if (l_suffix > 2 || u_suffix > 1) {
    586 		/* malformed integer constant */
    587 		warning(251);
    588 		if (l_suffix > 2)
    589 			l_suffix = 2;
    590 		if (u_suffix > 1)
    591 			u_suffix = 1;
    592 	}
    593 	if (!allow_c90 && u_suffix != 0) {
    594 		/* suffix U is illegal in traditional C */
    595 		warning(97);
    596 	}
    597 	typ = suffix_type[u_suffix][l_suffix];
    598 
    599 	errno = 0;
    600 
    601 	uq = (uint64_t)strtoull(cp, &eptr, base);
    602 	lint_assert(eptr == cp + len);
    603 	if (errno != 0) {
    604 		/* integer constant out of range */
    605 		warning(252);
    606 		warned = true;
    607 	}
    608 
    609 	/*
    610 	 * If the value is too big for the current type, we must choose
    611 	 * another type.
    612 	 */
    613 	ansiu = false;
    614 	switch (typ) {
    615 	case INT:
    616 		if (uq <= TARG_INT_MAX) {
    617 			/* ok */
    618 		} else if (uq <= TARG_UINT_MAX && base != 10) {
    619 			typ = UINT;
    620 		} else if (uq <= TARG_LONG_MAX) {
    621 			typ = LONG;
    622 		} else {
    623 			typ = ULONG;
    624 			if (uq > TARG_ULONG_MAX && !warned) {
    625 				/* integer constant out of range */
    626 				warning(252);
    627 			}
    628 		}
    629 		if (typ == UINT || typ == ULONG) {
    630 			if (!allow_c90) {
    631 				typ = LONG;
    632 			} else if (!sflag) {
    633 				/*
    634 				 * Remember that the constant is unsigned
    635 				 * only in ANSI C
    636 				 */
    637 				ansiu = true;
    638 			}
    639 		}
    640 		break;
    641 	case UINT:
    642 		if (uq > TARG_UINT_MAX) {
    643 			typ = ULONG;
    644 			if (uq > TARG_ULONG_MAX && !warned) {
    645 				/* integer constant out of range */
    646 				warning(252);
    647 			}
    648 		}
    649 		break;
    650 	case LONG:
    651 		if (uq > TARG_LONG_MAX && allow_c90) {
    652 			typ = ULONG;
    653 			if (!sflag)
    654 				ansiu = true;
    655 			if (uq > TARG_ULONG_MAX && !warned) {
    656 				/* integer constant out of range */
    657 				warning(252);
    658 			}
    659 		}
    660 		break;
    661 	case ULONG:
    662 		if (uq > TARG_ULONG_MAX && !warned) {
    663 			/* integer constant out of range */
    664 			warning(252);
    665 		}
    666 		break;
    667 	case QUAD:
    668 		if (uq > TARG_QUAD_MAX && allow_c90) {
    669 			typ = UQUAD;
    670 			if (!sflag)
    671 				ansiu = true;
    672 		}
    673 		break;
    674 	case UQUAD:
    675 		if (uq > TARG_UQUAD_MAX && !warned) {
    676 			/* integer constant out of range */
    677 			warning(252);
    678 		}
    679 		break;
    680 	default:
    681 		break;
    682 	}
    683 
    684 	uq = (uint64_t)convert_integer((int64_t)uq, typ, 0);
    685 
    686 	yylval.y_val = xcalloc(1, sizeof(*yylval.y_val));
    687 	yylval.y_val->v_tspec = typ;
    688 	yylval.y_val->v_unsigned_since_c90 = ansiu;
    689 	yylval.y_val->v_quad = (int64_t)uq;
    690 
    691 	return T_CON;
    692 }
    693 
    694 /*
    695  * Extend or truncate q to match t.  If t is signed, sign-extend.
    696  *
    697  * len is the number of significant bits. If len is -1, len is set
    698  * to the width of type t.
    699  */
    700 int64_t
    701 convert_integer(int64_t q, tspec_t t, unsigned int len)
    702 {
    703 	uint64_t vbits;
    704 
    705 	if (len == 0)
    706 		len = size_in_bits(t);
    707 
    708 	vbits = value_bits(len);
    709 	return t == PTR || is_uinteger(t) || ((q & bit(len - 1)) == 0)
    710 	    ? (int64_t)(q & vbits)
    711 	    : (int64_t)(q | ~vbits);
    712 }
    713 
    714 /*
    715  * Convert a string representing a floating point value into its numerical
    716  * representation. Type and value are returned in yylval.
    717  *
    718  * XXX Currently it is not possible to convert constants of type
    719  * long double which are greater than DBL_MAX.
    720  */
    721 int
    722 lex_floating_constant(const char *yytext, size_t yyleng)
    723 {
    724 	const	char *cp;
    725 	size_t	len;
    726 	tspec_t typ;
    727 	char	c, *eptr;
    728 	double	d;
    729 	float	f = 0;
    730 
    731 	cp = yytext;
    732 	len = yyleng;
    733 
    734 	if (cp[len - 1] == 'i')
    735 		len--;		/* imaginary, do nothing for now */
    736 
    737 	if ((c = cp[len - 1]) == 'f' || c == 'F') {
    738 		typ = FLOAT;
    739 		len--;
    740 	} else if (c == 'l' || c == 'L') {
    741 		typ = LDOUBLE;
    742 		len--;
    743 	} else {
    744 		if (c == 'd' || c == 'D')
    745 			len--;
    746 		typ = DOUBLE;
    747 	}
    748 
    749 	if (!allow_c90 && typ != DOUBLE) {
    750 		/* suffixes F and L are illegal in traditional C */
    751 		warning(98);
    752 	}
    753 
    754 	errno = 0;
    755 	d = strtod(cp, &eptr);
    756 	if (eptr != cp + len) {
    757 		switch (*eptr) {
    758 			/*
    759 			 * XXX: non-native non-current strtod() may not handle hex
    760 			 * floats, ignore the rest if we find traces of hex float
    761 			 * syntax...
    762 			 */
    763 		case 'p':
    764 		case 'P':
    765 		case 'x':
    766 		case 'X':
    767 			d = 0;
    768 			errno = 0;
    769 			break;
    770 		default:
    771 			INTERNAL_ERROR("lex_floating_constant(%s->%s)",
    772 			    cp, eptr);
    773 		}
    774 	}
    775 	if (errno != 0)
    776 		/* floating-point constant out of range */
    777 		warning(248);
    778 
    779 	if (typ == FLOAT) {
    780 		f = (float)d;
    781 		if (isfinite(f) == 0) {
    782 			/* floating-point constant out of range */
    783 			warning(248);
    784 			f = f > 0 ? FLT_MAX : -FLT_MAX;
    785 		}
    786 	}
    787 
    788 	yylval.y_val = xcalloc(1, sizeof(*yylval.y_val));
    789 	yylval.y_val->v_tspec = typ;
    790 	if (typ == FLOAT) {
    791 		yylval.y_val->v_ldbl = f;
    792 	} else {
    793 		yylval.y_val->v_ldbl = d;
    794 	}
    795 
    796 	return T_CON;
    797 }
    798 
    799 int
    800 lex_operator(int t, op_t o)
    801 {
    802 
    803 	yylval.y_op = o;
    804 	return t;
    805 }
    806 
    807 /* Called if lex found a leading "'". */
    808 int
    809 lex_character_constant(void)
    810 {
    811 	size_t	n;
    812 	int val, c;
    813 
    814 	n = 0;
    815 	val = 0;
    816 	while ((c = get_escaped_char('\'')) >= 0) {
    817 		val = (val << CHAR_SIZE) + c;
    818 		n++;
    819 	}
    820 	if (c == -2) {
    821 		/* unterminated character constant */
    822 		error(253);
    823 	} else if (n > sizeof(int) || (n > 1 && (pflag || hflag))) {
    824 		/* XXX: should rather be sizeof(TARG_INT) */
    825 
    826 		/* too many characters in character constant */
    827 		error(71);
    828 	} else if (n > 1) {
    829 		/* multi-character character constant */
    830 		warning(294);
    831 	} else if (n == 0) {
    832 		/* empty character constant */
    833 		error(73);
    834 	}
    835 	if (n == 1)
    836 		val = (int)convert_integer(val, CHAR, CHAR_SIZE);
    837 
    838 	yylval.y_val = xcalloc(1, sizeof(*yylval.y_val));
    839 	yylval.y_val->v_tspec = INT;
    840 	yylval.y_val->v_quad = val;
    841 
    842 	return T_CON;
    843 }
    844 
    845 /*
    846  * Called if lex found a leading L\'
    847  */
    848 int
    849 lex_wide_character_constant(void)
    850 {
    851 	static	char buf[MB_LEN_MAX + 1];
    852 	size_t	n, nmax;
    853 	int c;
    854 	wchar_t	wc;
    855 
    856 	nmax = MB_CUR_MAX;
    857 
    858 	n = 0;
    859 	while ((c = get_escaped_char('\'')) >= 0) {
    860 		if (n < nmax)
    861 			buf[n] = (char)c;
    862 		n++;
    863 	}
    864 
    865 	wc = 0;
    866 
    867 	if (c == -2) {
    868 		/* unterminated character constant */
    869 		error(253);
    870 	} else if (n == 0) {
    871 		/* empty character constant */
    872 		error(73);
    873 	} else if (n > nmax) {
    874 		n = nmax;
    875 		/* too many characters in character constant */
    876 		error(71);
    877 	} else {
    878 		buf[n] = '\0';
    879 		(void)mbtowc(NULL, NULL, 0);
    880 		if (mbtowc(&wc, buf, nmax) < 0)
    881 			/* invalid multibyte character */
    882 			error(291);
    883 	}
    884 
    885 	yylval.y_val = xcalloc(1, sizeof(*yylval.y_val));
    886 	yylval.y_val->v_tspec = WCHAR;
    887 	yylval.y_val->v_quad = wc;
    888 
    889 	return T_CON;
    890 }
    891 
    892 /*
    893  * Read a character which is part of a character constant or of a string
    894  * and handle escapes.
    895  *
    896  * The argument is the character which delimits the character constant or
    897  * string.
    898  *
    899  * Returns -1 if the end of the character constant or string is reached,
    900  * -2 if the EOF is reached, and the character otherwise.
    901  */
    902 static int
    903 get_escaped_char(int delim)
    904 {
    905 	static	int pbc = -1;
    906 	int	n, c, v;
    907 
    908 	if (pbc == -1) {
    909 		c = inpc();
    910 	} else {
    911 		c = pbc;
    912 		pbc = -1;
    913 	}
    914 	if (c == delim)
    915 		return -1;
    916 	switch (c) {
    917 	case '\n':
    918 		if (!allow_c90) {
    919 			/* newline in string or char constant */
    920 			error(254);
    921 			return -2;
    922 		}
    923 		return c;
    924 	case 0:
    925 		/* syntax error '%s' */
    926 		error(249, "EOF or null byte in literal");
    927 		return -2;
    928 	case EOF:
    929 		return -2;
    930 	case '\\':
    931 		switch (c = inpc()) {
    932 		case '"':
    933 			if (!allow_c90 && delim == '\'')
    934 				/* \" inside character constants undef... */
    935 				warning(262);
    936 			return '"';
    937 		case '\'':
    938 			return '\'';
    939 		case '?':
    940 			if (!allow_c90)
    941 				/* \? undefined in traditional C */
    942 				warning(263);
    943 			return '?';
    944 		case '\\':
    945 			return '\\';
    946 		case 'a':
    947 			if (!allow_c90)
    948 				/* \a undefined in traditional C */
    949 				warning(81);
    950 			return '\a';
    951 		case 'b':
    952 			return '\b';
    953 		case 'f':
    954 			return '\f';
    955 		case 'n':
    956 			return '\n';
    957 		case 'r':
    958 			return '\r';
    959 		case 't':
    960 			return '\t';
    961 		case 'v':
    962 			if (!allow_c90)
    963 				/* \v undefined in traditional C */
    964 				warning(264);
    965 			return '\v';
    966 		case '8': case '9':
    967 			/* bad octal digit %c */
    968 			warning(77, c);
    969 			/* FALLTHROUGH */
    970 		case '0': case '1': case '2': case '3':
    971 		case '4': case '5': case '6': case '7':
    972 			n = 3;
    973 			v = 0;
    974 			do {
    975 				v = (v << 3) + (c - '0');
    976 				c = inpc();
    977 			} while (--n > 0 && '0' <= c && c <= '7');
    978 			pbc = c;
    979 			if (v > TARG_UCHAR_MAX) {
    980 				/* character escape does not fit in character */
    981 				warning(76);
    982 				v &= CHAR_MASK;
    983 			}
    984 			return v;
    985 		case 'x':
    986 			if (!allow_c90)
    987 				/* \x undefined in traditional C */
    988 				warning(82);
    989 			v = 0;
    990 			n = 0;
    991 			while (c = inpc(), isxdigit(c)) {
    992 				c = isdigit(c) ?
    993 				    c - '0' : toupper(c) - 'A' + 10;
    994 				v = (v << 4) + c;
    995 				if (n >= 0) {
    996 					if ((v & ~CHAR_MASK) != 0) {
    997 						/* overflow in hex escape */
    998 						warning(75);
    999 						n = -1;
   1000 					} else {
   1001 						n++;
   1002 					}
   1003 				}
   1004 			}
   1005 			pbc = c;
   1006 			if (n == 0) {
   1007 				/* no hex digits follow \x */
   1008 				error(74);
   1009 			} if (n == -1) {
   1010 				v &= CHAR_MASK;
   1011 			}
   1012 			return v;
   1013 		case '\n':
   1014 			return get_escaped_char(delim);
   1015 		case EOF:
   1016 			return -2;
   1017 		default:
   1018 			if (isprint(c)) {
   1019 				/* dubious escape \%c */
   1020 				warning(79, c);
   1021 			} else {
   1022 				/* dubious escape \%o */
   1023 				warning(80, c);
   1024 			}
   1025 		}
   1026 	}
   1027 	return c;
   1028 }
   1029 
   1030 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */
   1031 static void
   1032 parse_line_directive_flags(const char *p,
   1033 			   bool *is_begin, bool *is_end, bool *is_system)
   1034 {
   1035 
   1036 	*is_begin = false;
   1037 	*is_end = false;
   1038 	*is_system = false;
   1039 
   1040 	while (*p != '\0') {
   1041 		const char *word_start, *word_end;
   1042 
   1043 		while (ch_isspace(*p))
   1044 			p++;
   1045 
   1046 		word_start = p;
   1047 		while (*p != '\0' && !ch_isspace(*p))
   1048 			p++;
   1049 		word_end = p;
   1050 
   1051 		if (word_end - word_start == 1 && word_start[0] == '1')
   1052 			*is_begin = true;
   1053 		if (word_end - word_start == 1 && word_start[0] == '2')
   1054 			*is_end = true;
   1055 		if (word_end - word_start == 1 && word_start[0] == '3')
   1056 			*is_system = true;
   1057 		/* Flag '4' is only interesting for C++. */
   1058 	}
   1059 }
   1060 
   1061 /*
   1062  * Called for preprocessor directives. Currently implemented are:
   1063  *	# pragma [argument...]
   1064  *	# lineno
   1065  *	# lineno "filename"
   1066  *	# lineno "filename" GCC-flag...
   1067  */
   1068 void
   1069 lex_directive(const char *yytext)
   1070 {
   1071 	const	char *cp, *fn;
   1072 	char	c, *eptr;
   1073 	size_t	fnl;
   1074 	long	ln;
   1075 	bool	is_begin, is_end, is_system;
   1076 
   1077 	static	bool first = true;
   1078 
   1079 	/* Go to first non-whitespace after # */
   1080 	for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
   1081 		continue;
   1082 
   1083 	if (!ch_isdigit(c)) {
   1084 		if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6]))
   1085 			return;
   1086 	error:
   1087 		/* undefined or invalid # directive */
   1088 		warning(255);
   1089 		return;
   1090 	}
   1091 	ln = strtol(--cp, &eptr, 10);
   1092 	if (eptr == cp)
   1093 		goto error;
   1094 	if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
   1095 		goto error;
   1096 	while ((c = *cp++) == ' ' || c == '\t')
   1097 		continue;
   1098 	if (c != '\0') {
   1099 		if (c != '"')
   1100 			goto error;
   1101 		fn = cp;
   1102 		while ((c = *cp) != '"' && c != '\0')
   1103 			cp++;
   1104 		if (c != '"')
   1105 			goto error;
   1106 		if ((fnl = cp++ - fn) > PATH_MAX)
   1107 			goto error;
   1108 		/* empty string means stdin */
   1109 		if (fnl == 0) {
   1110 			fn = "{standard input}";
   1111 			fnl = 16;			/* strlen (fn) */
   1112 		}
   1113 		curr_pos.p_file = record_filename(fn, fnl);
   1114 		/*
   1115 		 * If this is the first directive, the name is the name
   1116 		 * of the C source file as specified at the command line.
   1117 		 * It is written to the output file.
   1118 		 */
   1119 		if (first) {
   1120 			csrc_pos.p_file = curr_pos.p_file;
   1121 			outsrc(transform_filename(curr_pos.p_file,
   1122 			    strlen(curr_pos.p_file)));
   1123 			first = false;
   1124 		}
   1125 
   1126 		parse_line_directive_flags(cp, &is_begin, &is_end, &is_system);
   1127 		update_location(curr_pos.p_file, (int)ln, is_begin, is_end);
   1128 		in_system_header = is_system;
   1129 	}
   1130 	curr_pos.p_line = (int)ln - 1;
   1131 	curr_pos.p_uniq = 0;
   1132 	if (curr_pos.p_file == csrc_pos.p_file) {
   1133 		csrc_pos.p_line = (int)ln - 1;
   1134 		csrc_pos.p_uniq = 0;
   1135 	}
   1136 }
   1137 
   1138 /*
   1139  * Handle lint comments such as ARGSUSED.
   1140  *
   1141  * If one of these comments is recognized, the argument, if any, is
   1142  * parsed and a function which handles this comment is called.
   1143  */
   1144 void
   1145 lex_comment(void)
   1146 {
   1147 	int	c, lc;
   1148 	static const struct {
   1149 		const	char *keywd;
   1150 		bool	arg;
   1151 		void	(*func)(int);
   1152 	} keywtab[] = {
   1153 		{ "ARGSUSED",		true,	argsused	},
   1154 		{ "BITFIELDTYPE",	false,	bitfieldtype	},
   1155 		{ "CONSTCOND",		false,	constcond	},
   1156 		{ "CONSTANTCOND",	false,	constcond	},
   1157 		{ "CONSTANTCONDITION",	false,	constcond	},
   1158 		{ "FALLTHRU",		false,	fallthru	},
   1159 		{ "FALLTHROUGH",	false,	fallthru	},
   1160 		{ "FALL THROUGH",	false,	fallthru	},
   1161 		{ "fallthrough",	false,	fallthru	},
   1162 		{ "LINTLIBRARY",	false,	lintlib		},
   1163 		{ "LINTED",		true,	linted		},
   1164 		{ "LONGLONG",		false,	longlong	},
   1165 		{ "NOSTRICT",		true,	linted		},
   1166 		{ "NOTREACHED",		false,	not_reached	},
   1167 		{ "PRINTFLIKE",		true,	printflike	},
   1168 		{ "PROTOLIB",		true,	protolib	},
   1169 		{ "SCANFLIKE",		true,	scanflike	},
   1170 		{ "VARARGS",		true,	varargs		},
   1171 	};
   1172 	char	keywd[32];
   1173 	char	arg[32];
   1174 	size_t	l, i;
   1175 	int	a;
   1176 	bool	eoc;
   1177 
   1178 	eoc = false;
   1179 
   1180 	/* Skip whitespace after the start of the comment */
   1181 	while (c = inpc(), isspace(c))
   1182 		continue;
   1183 
   1184 	/* Read the potential keyword to keywd */
   1185 	l = 0;
   1186 	while (c != EOF && l < sizeof(keywd) - 1 &&
   1187 	    (isalpha(c) || isspace(c))) {
   1188 		if (islower(c) && l > 0 && ch_isupper(keywd[0]))
   1189 			break;
   1190 		keywd[l++] = (char)c;
   1191 		c = inpc();
   1192 	}
   1193 	while (l > 0 && ch_isspace(keywd[l - 1]))
   1194 		l--;
   1195 	keywd[l] = '\0';
   1196 
   1197 	/* look for the keyword */
   1198 	for (i = 0; i < sizeof(keywtab) / sizeof(keywtab[0]); i++) {
   1199 		if (strcmp(keywtab[i].keywd, keywd) == 0)
   1200 			break;
   1201 	}
   1202 	if (i == sizeof(keywtab) / sizeof(keywtab[0]))
   1203 		goto skip_rest;
   1204 
   1205 	/* skip whitespace after the keyword */
   1206 	while (isspace(c))
   1207 		c = inpc();
   1208 
   1209 	/* read the argument, if the keyword accepts one and there is one */
   1210 	l = 0;
   1211 	if (keywtab[i].arg) {
   1212 		while (isdigit(c) && l < sizeof(arg) - 1) {
   1213 			arg[l++] = (char)c;
   1214 			c = inpc();
   1215 		}
   1216 	}
   1217 	arg[l] = '\0';
   1218 	a = l != 0 ? atoi(arg) : -1;
   1219 
   1220 	/* skip whitespace after the argument */
   1221 	while (isspace(c))
   1222 		c = inpc();
   1223 
   1224 	if (c != '*' || (c = inpc()) != '/') {
   1225 		if (keywtab[i].func != linted)
   1226 			/* extra characters in lint comment */
   1227 			warning(257);
   1228 	} else {
   1229 		/*
   1230 		 * remember that we have already found the end of the
   1231 		 * comment
   1232 		 */
   1233 		eoc = true;
   1234 	}
   1235 
   1236 	if (keywtab[i].func != NULL)
   1237 		(*keywtab[i].func)(a);
   1238 
   1239 skip_rest:
   1240 	while (!eoc) {
   1241 		lc = c;
   1242 		if ((c = inpc()) == EOF) {
   1243 			/* unterminated comment */
   1244 			error(256);
   1245 			break;
   1246 		}
   1247 		if (lc == '*' && c == '/')
   1248 			eoc = true;
   1249 	}
   1250 }
   1251 
   1252 /*
   1253  * Handle // style comments
   1254  */
   1255 void
   1256 lex_slash_slash_comment(void)
   1257 {
   1258 	int c;
   1259 
   1260 	if (!allow_c99 && !allow_gcc)
   1261 		/* %s does not support // comments */
   1262 		gnuism(312, allow_c90 ? "C90" : "traditional C");
   1263 
   1264 	while ((c = inpc()) != EOF && c != '\n')
   1265 		continue;
   1266 }
   1267 
   1268 /*
   1269  * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
   1270  * clear_warn_flags is called after function definitions and global and
   1271  * local declarations and definitions. It is also called between
   1272  * the controlling expression and the body of control statements
   1273  * (if, switch, for, while).
   1274  */
   1275 void
   1276 clear_warn_flags(void)
   1277 {
   1278 
   1279 	lwarn = LWARN_ALL;
   1280 	quadflg = false;
   1281 	constcond_flag = false;
   1282 }
   1283 
   1284 /*
   1285  * Strings are stored in a dynamically allocated buffer and passed
   1286  * in yylval.y_string to the parser. The parser or the routines called
   1287  * by the parser are responsible for freeing this buffer.
   1288  */
   1289 int
   1290 lex_string(void)
   1291 {
   1292 	unsigned char *s;
   1293 	int	c;
   1294 	size_t	len, max;
   1295 	strg_t	*strg;
   1296 
   1297 	s = xmalloc(max = 64);
   1298 
   1299 	len = 0;
   1300 	while ((c = get_escaped_char('"')) >= 0) {
   1301 		/* +1 to reserve space for a trailing NUL character */
   1302 		if (len + 1 == max)
   1303 			s = xrealloc(s, max *= 2);
   1304 		s[len++] = (char)c;
   1305 	}
   1306 	s[len] = '\0';
   1307 	if (c == -2)
   1308 		/* unterminated string constant */
   1309 		error(258);
   1310 
   1311 	strg = xcalloc(1, sizeof(*strg));
   1312 	strg->st_char = true;
   1313 	strg->st_len = len;
   1314 	strg->st_mem = s;
   1315 
   1316 	yylval.y_string = strg;
   1317 	return T_STRING;
   1318 }
   1319 
   1320 int
   1321 lex_wide_string(void)
   1322 {
   1323 	char	*s;
   1324 	int	c, n;
   1325 	size_t	i, wi;
   1326 	size_t	len, max, wlen;
   1327 	wchar_t	*ws;
   1328 	strg_t	*strg;
   1329 
   1330 	s = xmalloc(max = 64);
   1331 	len = 0;
   1332 	while ((c = get_escaped_char('"')) >= 0) {
   1333 		/* +1 to save space for a trailing NUL character */
   1334 		if (len + 1 >= max)
   1335 			s = xrealloc(s, max *= 2);
   1336 		s[len++] = (char)c;
   1337 	}
   1338 	s[len] = '\0';
   1339 	if (c == -2)
   1340 		/* unterminated string constant */
   1341 		error(258);
   1342 
   1343 	/* get length of wide-character string */
   1344 	(void)mblen(NULL, 0);
   1345 	for (i = 0, wlen = 0; i < len; i += n, wlen++) {
   1346 		if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
   1347 			/* invalid multibyte character */
   1348 			error(291);
   1349 			break;
   1350 		}
   1351 		if (n == 0)
   1352 			n = 1;
   1353 	}
   1354 
   1355 	ws = xmalloc((wlen + 1) * sizeof(*ws));
   1356 
   1357 	/* convert from multibyte to wide char */
   1358 	(void)mbtowc(NULL, NULL, 0);
   1359 	for (i = 0, wi = 0; i < len; i += n, wi++) {
   1360 		if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
   1361 			break;
   1362 		if (n == 0)
   1363 			n = 1;
   1364 	}
   1365 	ws[wi] = 0;
   1366 	free(s);
   1367 
   1368 	strg = xcalloc(1, sizeof(*strg));
   1369 	strg->st_char = false;
   1370 	strg->st_len = wlen;
   1371 	strg->st_mem = ws;
   1372 
   1373 	yylval.y_string = strg;
   1374 	return T_STRING;
   1375 }
   1376 
   1377 void
   1378 lex_next_line(void)
   1379 {
   1380 	curr_pos.p_line++;
   1381 	curr_pos.p_uniq = 0;
   1382 	debug_step("parsing %s:%d", curr_pos.p_file, curr_pos.p_line);
   1383 	if (curr_pos.p_file == csrc_pos.p_file) {
   1384 		csrc_pos.p_line++;
   1385 		csrc_pos.p_uniq = 0;
   1386 	}
   1387 }
   1388 
   1389 void
   1390 lex_unknown_character(int c)
   1391 {
   1392 
   1393 	/* unknown character \%o */
   1394 	error(250, c);
   1395 }
   1396 
   1397 /*
   1398  * The scanner does not create new symbol table entries for symbols it cannot
   1399  * find in the symbol table. This is to avoid putting undeclared symbols into
   1400  * the symbol table if a syntax error occurs.
   1401  *
   1402  * getsym is called as soon as it is probably ok to put the symbol in the
   1403  * symbol table. It is still possible that symbols are put in the symbol
   1404  * table that are not completely declared due to syntax errors. To avoid too
   1405  * many problems in this case, symbols get type 'int' in getsym.
   1406  *
   1407  * XXX calls to getsym should be delayed until declare_1_* is called.
   1408  */
   1409 sym_t *
   1410 getsym(sbuf_t *sb)
   1411 {
   1412 	dinfo_t	*di;
   1413 	char	*s;
   1414 	sym_t	*sym;
   1415 
   1416 	sym = sb->sb_sym;
   1417 
   1418 	/*
   1419 	 * During member declaration it is possible that name() looked
   1420 	 * for symbols of type FVFT, although it should have looked for
   1421 	 * symbols of type FTAG. Same can happen for labels. Both cases
   1422 	 * are compensated here.
   1423 	 */
   1424 	if (symtyp == FMEMBER || symtyp == FLABEL) {
   1425 		if (sym == NULL || sym->s_kind == FVFT)
   1426 			sym = symtab_search(sb);
   1427 	}
   1428 
   1429 	if (sym != NULL) {
   1430 		lint_assert(sym->s_kind == symtyp);
   1431 		symtyp = FVFT;
   1432 		free(sb);
   1433 		return sym;
   1434 	}
   1435 
   1436 	/* create a new symbol table entry */
   1437 
   1438 	/* labels must always be allocated at level 1 (outermost block) */
   1439 	if (symtyp == FLABEL) {
   1440 		sym = level_zero_alloc(1, sizeof(*sym));
   1441 		s = level_zero_alloc(1, sb->sb_len + 1);
   1442 		(void)memcpy(s, sb->sb_name, sb->sb_len + 1);
   1443 		sym->s_name = s;
   1444 		sym->s_block_level = 1;
   1445 		di = dcs;
   1446 		while (di->d_enclosing != NULL &&
   1447 		    di->d_enclosing->d_enclosing != NULL)
   1448 			di = di->d_enclosing;
   1449 		lint_assert(di->d_kind == DK_AUTO);
   1450 	} else {
   1451 		sym = block_zero_alloc(sizeof(*sym));
   1452 		sym->s_name = sb->sb_name;
   1453 		sym->s_block_level = block_level;
   1454 		di = dcs;
   1455 	}
   1456 
   1457 	UNIQUE_CURR_POS(sym->s_def_pos);
   1458 	if ((sym->s_kind = symtyp) != FLABEL)
   1459 		sym->s_type = gettyp(INT);
   1460 
   1461 	symtyp = FVFT;
   1462 
   1463 	symtab_add(sym);
   1464 
   1465 	*di->d_ldlsym = sym;
   1466 	di->d_ldlsym = &sym->s_level_next;
   1467 
   1468 	free(sb);
   1469 	return sym;
   1470 }
   1471 
   1472 /*
   1473  * Construct a temporary symbol. The symbol name starts with a digit, making
   1474  * the name illegal.
   1475  */
   1476 sym_t *
   1477 mktempsym(type_t *tp)
   1478 {
   1479 	static unsigned n = 0;
   1480 	char *s = level_zero_alloc((size_t)block_level, 64);
   1481 	sym_t *sym = block_zero_alloc(sizeof(*sym));
   1482 	scl_t scl;
   1483 
   1484 	(void)snprintf(s, 64, "%.8u_tmp", n++);
   1485 
   1486 	scl = dcs->d_scl;
   1487 	if (scl == NOSCL)
   1488 		scl = block_level > 0 ? AUTO : EXTERN;
   1489 
   1490 	sym->s_name = s;
   1491 	sym->s_type = tp;
   1492 	sym->s_block_level = block_level;
   1493 	sym->s_scl = scl;
   1494 	sym->s_kind = FVFT;
   1495 	sym->s_used = true;
   1496 	sym->s_set = true;
   1497 
   1498 	symtab_add(sym);
   1499 
   1500 	*dcs->d_ldlsym = sym;
   1501 	dcs->d_ldlsym = &sym->s_level_next;
   1502 
   1503 	return sym;
   1504 }
   1505 
   1506 /* Remove a symbol forever from the symbol table. */
   1507 void
   1508 rmsym(sym_t *sym)
   1509 {
   1510 
   1511 	debug_step("rmsym '%s' %s '%s'",
   1512 	    sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type));
   1513 	symtab_remove(sym);
   1514 
   1515 	/* avoid that the symbol will later be put back to the symbol table */
   1516 	sym->s_block_level = -1;
   1517 }
   1518 
   1519 /*
   1520  * Remove all symbols from the symbol table that have the same level as the
   1521  * given symbol.
   1522  */
   1523 void
   1524 rmsyms(sym_t *syms)
   1525 {
   1526 	sym_t	*sym;
   1527 
   1528 	/* Note the use of s_level_next instead of s_symtab_next. */
   1529 	for (sym = syms; sym != NULL; sym = sym->s_level_next) {
   1530 		if (sym->s_block_level != -1) {
   1531 			debug_step("rmsyms '%s' %s '%s'",
   1532 			    sym->s_name, symt_name(sym->s_kind),
   1533 			    type_name(sym->s_type));
   1534 			symtab_remove(sym);
   1535 			sym->s_symtab_ref = NULL;
   1536 		}
   1537 	}
   1538 }
   1539 
   1540 /*
   1541  * Put a symbol into the symbol table.
   1542  */
   1543 void
   1544 inssym(int level, sym_t *sym)
   1545 {
   1546 
   1547 	debug_step("inssym '%s' %s '%s'",
   1548 	    sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type));
   1549 	symtab_add(sym);
   1550 	sym->s_block_level = level;
   1551 
   1552 	/*
   1553 	 * Placing the inner symbols to the beginning of the list ensures
   1554 	 * that these symbols are preferred over symbols from the outer
   1555 	 * blocks that happen to have the same name.
   1556 	 */
   1557 	lint_assert(sym->s_symtab_next != NULL
   1558 	    ? sym->s_block_level >= sym->s_symtab_next->s_block_level
   1559 	    : true);
   1560 }
   1561 
   1562 /*
   1563  * Called at level 0 after syntax errors.
   1564  *
   1565  * Removes all symbols which are not declared at level 0 from the
   1566  * symbol table. Also frees all memory which is not associated with
   1567  * level 0.
   1568  */
   1569 void
   1570 clean_up_after_error(void)
   1571 {
   1572 
   1573 	symtab_remove_locals();
   1574 
   1575 	for (size_t i = mem_block_level; i > 0; i--)
   1576 		level_free_all(i);
   1577 }
   1578 
   1579 /* Create a new symbol with the same name as an existing symbol. */
   1580 sym_t *
   1581 pushdown(const sym_t *sym)
   1582 {
   1583 	sym_t	*nsym;
   1584 
   1585 	debug_step("pushdown '%s' %s '%s'",
   1586 	    sym->s_name, symt_name(sym->s_kind), type_name(sym->s_type));
   1587 	nsym = block_zero_alloc(sizeof(*nsym));
   1588 	lint_assert(sym->s_block_level <= block_level);
   1589 	nsym->s_name = sym->s_name;
   1590 	UNIQUE_CURR_POS(nsym->s_def_pos);
   1591 	nsym->s_kind = sym->s_kind;
   1592 	nsym->s_block_level = block_level;
   1593 
   1594 	symtab_add(nsym);
   1595 
   1596 	*dcs->d_ldlsym = nsym;
   1597 	dcs->d_ldlsym = &nsym->s_level_next;
   1598 
   1599 	return nsym;
   1600 }
   1601 
   1602 /*
   1603  * Free any dynamically allocated memory referenced by
   1604  * the value stack or yylval.
   1605  * The type of information in yylval is described by tok.
   1606  */
   1607 void
   1608 freeyyv(void *sp, int tok)
   1609 {
   1610 	if (tok == T_NAME || tok == T_TYPENAME) {
   1611 		sbuf_t *sb = *(sbuf_t **)sp;
   1612 		free(sb);
   1613 	} else if (tok == T_CON) {
   1614 		val_t *val = *(val_t **)sp;
   1615 		free(val);
   1616 	} else if (tok == T_STRING) {
   1617 		strg_t *strg = *(strg_t **)sp;
   1618 		free(strg->st_mem);
   1619 		free(strg);
   1620 	}
   1621 }
   1622