xref: /NextBSD/usr.bin/xlint/lint1/scan.l (revision c21ffb8d6aca32c9584cfa072f309a5890a21aea)
1 %{
2 /* $NetBSD: scan.l,v 1.37 2007/02/06 00:08:31 he 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 1.37 2007/02/06 00:08:31 he Exp $");
39 #endif
40 __FBSDID("$FreeBSD$");
41 
42 #include <stdlib.h>
43 #include <string.h>
44 #include <limits.h>
45 #include <float.h>
46 #include <ctype.h>
47 #include <errno.h>
48 #include <err.h>
49 #include <math.h>
50 
51 #include "lint1.h"
52 #include "cgram.h"
53 
54 #define CHAR_MASK	(~(~0 << CHAR_BIT))
55 
56 /* Current position (its also updated when an included file is parsed) */
57 pos_t	curr_pos = { 1, "", 0 };
58 
59 /*
60  * Current position in C source (not updated when an included file is
61  * parsed).
62  */
63 pos_t	csrc_pos = { 1, "", 0 };
64 
65 static	void	incline(void);
66 static	void	badchar(int);
67 static	sbuf_t	*allocsb(void);
68 static	void	freesb(sbuf_t *);
69 static	int	inpc(void);
70 static	int	hash(const char *);
71 static	sym_t	*search(sbuf_t *);
72 static	int	name(void);
73 static	int	keyw(sym_t *);
74 static	int	icon(int);
75 static	int	fcon(void);
76 static	int	operator(int, op_t);
77 static	int	ccon(void);
78 static	int	wccon(void);
79 static	int	getescc(int);
80 static	void	directive(void);
81 static	void	comment(void);
82 static	void	slashslashcomment(void);
83 static	int	string(void);
84 static	int	wcstrg(void);
85 
86 %}
87 
88 %option nounput
89 
90 L	[_A-Za-z]
91 D	[0-9]
92 NZD	[1-9]
93 OD	[0-7]
94 HD	[0-9A-Fa-f]
95 EX	([eE][+-]?[0-9]+)
96 
97 %%
98 
99 {L}({L}|{D})*			return (name());
100 0{OD}*[lLuU]*			return (icon(8));
101 {NZD}{D}*[lLuU]*		return (icon(10));
102 0[xX]{HD}+[lLuU]*		return (icon(16));
103 {D}+\.{D}*{EX}?[fFlL]?		|
104 {D}+{EX}[fFlL]?			|
105 0[xX]{HD}+p{HD}+[fFlL]?		|
106 \.{D}+{EX}?[fFlL]?		return (fcon());
107 "="				return (operator(T_ASSIGN, ASSIGN));
108 "*="				return (operator(T_OPASS, MULASS));
109 "/="				return (operator(T_OPASS, DIVASS));
110 "%="				return (operator(T_OPASS, MODASS));
111 "+="				return (operator(T_OPASS, ADDASS));
112 "-="				return (operator(T_OPASS, SUBASS));
113 "<<="				return (operator(T_OPASS, SHLASS));
114 ">>="				return (operator(T_OPASS, SHRASS));
115 "&="				return (operator(T_OPASS, ANDASS));
116 "^="				return (operator(T_OPASS, XORASS));
117 "|="				return (operator(T_OPASS, ORASS));
118 "||"				return (operator(T_LOGOR, LOGOR));
119 "&&"				return (operator(T_LOGAND, LOGAND));
120 "|"				return (operator(T_OR, OR));
121 "&"				return (operator(T_AND, AND));
122 "^"				return (operator(T_XOR, XOR));
123 "=="				return (operator(T_EQOP, EQ));
124 "!="				return (operator(T_EQOP, NE));
125 "<"				return (operator(T_RELOP, LT));
126 ">"				return (operator(T_RELOP, GT));
127 "<="				return (operator(T_RELOP, LE));
128 ">="				return (operator(T_RELOP, GE));
129 "<<"				return (operator(T_SHFTOP, SHL));
130 ">>"				return (operator(T_SHFTOP, SHR));
131 "++"				return (operator(T_INCDEC, INC));
132 "--"				return (operator(T_INCDEC, DEC));
133 "->"				return (operator(T_STROP, ARROW));
134 "."				return (operator(T_STROP, POINT));
135 "+"				return (operator(T_ADDOP, PLUS));
136 "-"				return (operator(T_ADDOP, MINUS));
137 "*"				return (operator(T_MULT, MULT));
138 "/"				return (operator(T_DIVOP, DIV));
139 "%"				return (operator(T_DIVOP, MOD));
140 "!"				return (operator(T_UNOP, NOT));
141 "~"				return (operator(T_UNOP, COMPL));
142 "\""				return (string());
143 "L\""				return (wcstrg());
144 ";"				return (T_SEMI);
145 "{"				return (T_LBRACE);
146 "}"				return (T_RBRACE);
147 ","				return (T_COMMA);
148 ":"				return (T_COLON);
149 "?"				return (T_QUEST);
150 "["				return (T_LBRACK);
151 "]"				return (T_RBRACK);
152 "("				return (T_LPARN);
153 ")"				return (T_RPARN);
154 "..."				return (T_ELLIPSE);
155 "'"				return (ccon());
156 "L'"				return (wccon());
157 ^#.*$				directive();
158 \n				incline();
159 \t|" "|\f|\v			;
160 "/*"				comment();
161 "//"				slashslashcomment();
162 .				badchar(yytext[0]);
163 
164 %%
165 
166 static void
167 incline(void)
168 {
169 	curr_pos.p_line++;
170 	curr_pos.p_uniq = 0;
171 	if (curr_pos.p_file == csrc_pos.p_file) {
172 		csrc_pos.p_line++;
173 		csrc_pos.p_uniq = 0;
174 	}
175 }
176 
177 static void
badchar(int c)178 badchar(int c)
179 {
180 
181 	/* unknown character \%o */
182 	error(250, c);
183 }
184 
185 /*
186  * Keywords.
187  * During initialisation they are written to the symbol table.
188  */
189 static	struct	kwtab {
190 	const	char *kw_name;	/* keyword */
191 	int	kw_token;	/* token returned by yylex() */
192 	scl_t	kw_scl;		/* storage class if kw_token T_SCLASS */
193 	tspec_t	kw_tspec;	/* type spec. if kw_token T_TYPE or T_SOU */
194 	tqual_t	kw_tqual;	/* type qual. fi kw_token T_QUAL */
195 	u_int	kw_c89;		/* c89 keyword */
196 	u_int	kw_c99;		/* c99 keyword */
197 	u_int	kw_gcc;		/* GCC keyword */
198 } kwtab[] = {
199 	{ "asm",	T_ASM,		0,	0,	0,	  0, 0, 1 },
200 	{ "__asm",	T_ASM,		0,	0,	0,	  0, 0, 0 },
201 	{ "__asm__",	T_ASM,		0,	0,	0,	  0, 0, 0 },
202 	{ "auto",	T_SCLASS,	AUTO,	0,	0,	  0, 0, 0 },
203 	{ "break",	T_BREAK,	0,	0,	0,	  0, 0, 0 },
204 	{ "case",	T_CASE,		0,	0,	0,	  0, 0, 0 },
205 	{ "char",	T_TYPE,		0,	CHAR,	0,	  0, 0, 0 },
206 	{ "const",	T_QUAL,		0,	0,	CONST,	  1, 0, 0 },
207 	{ "__const__",	T_QUAL,		0,	0,	CONST,	  0, 0, 0 },
208 	{ "__const",	T_QUAL,		0,	0,	CONST,	  0, 0, 0 },
209 	{ "continue",	T_CONTINUE,	0,	0,	0,	  0, 0, 0 },
210 	{ "default",	T_DEFAULT,	0,	0,	0,	  0, 0, 0 },
211 	{ "do",		T_DO,		0,	0,	0,	  0, 0, 0 },
212 	{ "double",	T_TYPE,		0,	DOUBLE,	0,	  0, 0, 0 },
213 	{ "else",	T_ELSE,		0,	0,	0,	  0, 0, 0 },
214 	{ "enum",	T_ENUM,		0,	0,	0,	  0, 0, 0 },
215 	{ "extern",	T_SCLASS,	EXTERN,	0,	0,	  0, 0, 0 },
216 	{ "float",	T_TYPE,		0,	FLOAT,	0,	  0, 0, 0 },
217 	{ "for",	T_FOR,		0,	0,	0,	  0, 0, 0 },
218 	{ "goto",	T_GOTO,		0,	0,	0,	  0, 0, 0 },
219 	{ "if",		T_IF,		0,	0,	0,	  0, 0, 0 },
220 	{ "inline",	T_SCLASS,	INLINE,	0,	0,	  0, 1, 0 },
221 	{ "__inline__",	T_SCLASS,	INLINE,	0,	0,	  0, 0, 0 },
222 	{ "__inline",	T_SCLASS,	INLINE,	0,	0,	  0, 0, 0 },
223 	{ "int",	T_TYPE,		0,	INT,	0,	  0, 0, 0 },
224 	{ "__symbolrename", T_SYMBOLRENAME, 0,	0,	0,	  0, 0, 0 },
225 	{ "long",	T_TYPE,		0,	LONG,	0,	  0, 0, 0 },
226 	{ "register",	T_SCLASS,	REG,	0,	0,	  0, 0, 0 },
227 	{ "return",	T_RETURN,	0,	0,	0,	  0, 0, 0 },
228 	{ "short",	T_TYPE,		0,	SHORT,	0,	  0, 0, 0 },
229 	{ "signed",	T_TYPE,		0,	SIGNED,	0,	  1, 0, 0 },
230 	{ "__signed__",	T_TYPE,		0,	SIGNED,	0,	  0, 0, 0 },
231 	{ "__signed",	T_TYPE,		0,	SIGNED,	0,	  0, 0, 0 },
232 	{ "sizeof",	T_SIZEOF,	0,	0,	0,	  0, 0, 0 },
233 	{ "static",	T_SCLASS,	STATIC,	0,	0,	  0, 0, 0 },
234 	{ "struct",	T_SOU,		0,	STRUCT,	0,	  0, 0, 0 },
235 	{ "switch",	T_SWITCH,	0,	0,	0,	  0, 0, 0 },
236 	{ "typedef",	T_SCLASS,	TYPEDEF, 0,	0,	  0, 0, 0 },
237 	{ "union",	T_SOU,		0,	UNION,	0,	  0, 0, 0 },
238 	{ "unsigned",	T_TYPE,		0,	UNSIGN,	0,	  0, 0, 0 },
239 	{ "void",	T_TYPE,		0,	VOID,	0,	  0, 0, 0 },
240 	{ "volatile",	T_QUAL,		0,	0,	VOLATILE, 1, 0, 0 },
241 	{ "__volatile__", T_QUAL,	0,	0,	VOLATILE, 0, 0, 0 },
242 	{ "__volatile",	T_QUAL,		0,	0,	VOLATILE, 0, 0, 0 },
243 	{ "while",	T_WHILE,	0,	0,	0,	  0, 0, 0 },
244 	{ NULL,		0,		0,	0,	0,	  0, 0, 0 }
245 };
246 
247 /* Symbol table */
248 static	sym_t	*symtab[HSHSIZ1];
249 
250 /* bit i of the entry with index i is set */
251 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
252 
253 /* least significant i bits are set in the entry with index i */
254 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
255 
256 /* least significant i bits are not set in the entry with index i */
257 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
258 
259 /* free list for sbuf structures */
260 static	sbuf_t	 *sbfrlst;
261 
262 /* type of next expected symbol */
263 symt_t	symtyp;
264 
265 
266 /*
267  * All keywords are written to the symbol table. This saves us looking
268  * in an extra table for each name we found.
269  */
270 void
initscan(void)271 initscan(void)
272 {
273 	struct	kwtab *kw;
274 	sym_t	*sym;
275 	int	h, i;
276 	uint64_t uq;
277 
278 	for (kw = kwtab; kw->kw_name != NULL; kw++) {
279 		if ((kw->kw_c89 || kw->kw_c99) && tflag)
280 			continue;
281 		if (kw->kw_c99 && !(Sflag || gflag))
282 			continue;
283 		if (kw->kw_gcc && !gflag)
284 			continue;
285 		sym = getblk(sizeof (sym_t));
286 		sym->s_name = kw->kw_name;
287 		sym->s_keyw = 1;
288 		sym->s_value.v_quad = kw->kw_token;
289 		if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
290 			sym->s_tspec = kw->kw_tspec;
291 		} else if (kw->kw_token == T_SCLASS) {
292 			sym->s_scl = kw->kw_scl;
293 		} else if (kw->kw_token == T_QUAL) {
294 			sym->s_tqual = kw->kw_tqual;
295 		}
296 		h = hash(sym->s_name);
297 		if ((sym->s_link = symtab[h]) != NULL)
298 			symtab[h]->s_rlink = &sym->s_link;
299 		(symtab[h] = sym)->s_rlink = &symtab[h];
300 	}
301 
302 	/* initialize bit-masks for quads */
303 	for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
304 		qbmasks[i] = (uint64_t)1 << i;
305 		uq = ~(uint64_t)0 << i;
306 		qumasks[i] = uq;
307 		qlmasks[i] = ~uq;
308 	}
309 	qumasks[i] = 0;
310 	qlmasks[i] = ~(uint64_t)0;
311 }
312 
313 /*
314  * Get a free sbuf structure, if possible from the free list
315  */
316 static sbuf_t *
allocsb(void)317 allocsb(void)
318 {
319 	sbuf_t	*sb;
320 
321 	if ((sb = sbfrlst) != NULL) {
322 		sbfrlst = sb->sb_nxt;
323 	} else {
324 		if ((sb = malloc(sizeof (sbuf_t))) == NULL)
325 			nomem();
326 	}
327 	(void)memset(sb, 0, sizeof (*sb));
328 	return (sb);
329 }
330 
331 /*
332  * Put a sbuf structure to the free list
333  */
334 static void
freesb(sbuf_t * sb)335 freesb(sbuf_t *sb)
336 {
337 
338 	sb->sb_nxt = sbfrlst;
339 	sbfrlst = sb;
340 }
341 
342 /*
343  * Read a character and ensure that it is positive (except EOF).
344  * Increment line count(s) if necessary.
345  */
346 static int
inpc(void)347 inpc(void)
348 {
349 	int	c;
350 
351 	if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
352 		incline();
353 	return (c);
354 }
355 
356 static int
hash(const char * s)357 hash(const char *s)
358 {
359 	u_int	v;
360 	const	u_char *us;
361 
362 	v = 0;
363 	for (us = (const u_char *)s; *us != '\0'; us++) {
364 		v = (v << sizeof (v)) + *us;
365 		v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
366 	}
367 	return (v % HSHSIZ1);
368 }
369 
370 /*
371  * Lex has found a letter followed by zero or more letters or digits.
372  * It looks for a symbol in the symbol table with the same name. This
373  * symbol must either be a keyword or a symbol of the type required by
374  * symtyp (label, member, tag, ...).
375  *
376  * If it is a keyword, the token is returned. In some cases it is described
377  * more deeply by data written to yylval.
378  *
379  * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
380  * is stored in yylval. This struct contains the name of the symbol, it's
381  * length and hash value. If there is already a symbol of the same name
382  * and type in the symbol table, the sbuf struct also contains a pointer
383  * to the symbol table entry.
384  */
385 static int
name(void)386 name(void)
387 {
388 	char	*s;
389 	sbuf_t	*sb;
390 	sym_t	*sym;
391 	int	tok;
392 
393 	sb = allocsb();
394 	sb->sb_name = yytext;
395 	sb->sb_len = yyleng;
396 	sb->sb_hash = hash(yytext);
397 
398 	if ((sym = search(sb)) != NULL && sym->s_keyw) {
399 		freesb(sb);
400 		return (keyw(sym));
401 	}
402 
403 	sb->sb_sym = sym;
404 
405 	if (sym != NULL) {
406 		if (blklev < sym->s_blklev)
407 			LERROR("name()");
408 		sb->sb_name = sym->s_name;
409 		sb->sb_len = strlen(sym->s_name);
410 		tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
411 	} else {
412 		s = getblk(yyleng + 1);
413 		(void)memcpy(s, yytext, yyleng + 1);
414 		sb->sb_name = s;
415 		sb->sb_len = yyleng;
416 		tok = T_NAME;
417 	}
418 
419 	yylval.y_sb = sb;
420 	return (tok);
421 }
422 
423 static sym_t *
search(sbuf_t * sb)424 search(sbuf_t *sb)
425 {
426 	sym_t	*sym;
427 
428 	for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
429 		if (strcmp(sym->s_name, sb->sb_name) == 0) {
430 			if (sym->s_keyw || sym->s_kind == symtyp)
431 				return (sym);
432 		}
433 	}
434 
435 	return (NULL);
436 }
437 
438 static int
keyw(sym_t * sym)439 keyw(sym_t *sym)
440 {
441 	int	t;
442 
443 	if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
444 		yylval.y_scl = sym->s_scl;
445 	} else if (t == T_TYPE || t == T_SOU) {
446 		yylval.y_tspec = sym->s_tspec;
447 	} else if (t == T_QUAL) {
448 		yylval.y_tqual = sym->s_tqual;
449 	}
450 	return (t);
451 }
452 
453 /*
454  * Convert a string representing an integer into internal representation.
455  * The value is returned in yylval. icon() (and yylex()) returns T_CON.
456  */
457 static int
icon(int base)458 icon(int base)
459 {
460 	int	l_suffix, u_suffix;
461 	int	len;
462 	const	char *cp;
463 	char	c, *eptr;
464 	tspec_t	typ;
465 	u_long	ul = 0;
466 	uint64_t uq = 0;
467 	int	ansiu;
468 	static	tspec_t contypes[2][3] = {
469 		{ INT,  LONG,  QUAD },
470 		{ UINT, ULONG, UQUAD }
471 	};
472 
473 	cp = yytext;
474 	len = yyleng;
475 
476 	/* skip 0x */
477 	if (base == 16) {
478 		cp += 2;
479 		len -= 2;
480 	}
481 
482 	/* read suffixes */
483 	l_suffix = u_suffix = 0;
484 	for ( ; ; ) {
485 		if ((c = cp[len - 1]) == 'l' || c == 'L') {
486 			l_suffix++;
487 		} else if (c == 'u' || c == 'U') {
488 			u_suffix++;
489 		} else {
490 			break;
491 		}
492 		len--;
493 	}
494 	if (l_suffix > 2 || u_suffix > 1) {
495 		/* malformed integer constant */
496 		warning(251);
497 		if (l_suffix > 2)
498 			l_suffix = 2;
499 		if (u_suffix > 1)
500 			u_suffix = 1;
501 	}
502 	if (tflag && u_suffix != 0) {
503 		/* suffix U is illegal in traditional C */
504 		warning(97);
505 	}
506 	typ = contypes[u_suffix][l_suffix];
507 
508 	errno = 0;
509 	if (l_suffix < 2) {
510 		ul = strtoul(cp, &eptr, base);
511 	} else {
512 		uq = strtouq(cp, &eptr, base);
513 	}
514 	if (eptr != cp + len)
515 		LERROR("icon()");
516 	if (errno != 0)
517 		/* integer constant out of range */
518 		warning(252);
519 
520 	/*
521 	 * If the value is too big for the current type, we must choose
522 	 * another type.
523 	 */
524 	ansiu = 0;
525 	switch (typ) {
526 	case INT:
527 		if (ul <= INT_MAX) {
528 			/* ok */
529 		} else if (ul <= (unsigned)UINT_MAX && base != 10) {
530 			typ = UINT;
531 #if INT_MAX != LONG_MAX
532 		} else if (ul <= LONG_MAX) {
533 			typ = LONG;
534 #endif
535 		} else {
536 			typ = ULONG;
537 		}
538 		if (typ == UINT || typ == ULONG) {
539 			if (tflag) {
540 				typ = LONG;
541 			} else if (!sflag) {
542 				/*
543 				 * Remember that the constant is unsigned
544 				 * only in ANSI C
545 				 */
546 				ansiu = 1;
547 			}
548 		}
549 		break;
550 	case UINT:
551 		if (ul > (u_int)UINT_MAX)
552 			typ = ULONG;
553 		break;
554 	case LONG:
555 		if (ul > LONG_MAX && !tflag) {
556 			typ = ULONG;
557 			if (!sflag)
558 				ansiu = 1;
559 		}
560 		break;
561 	case QUAD:
562 		if (uq > QUAD_MAX && !tflag) {
563 			typ = UQUAD;
564 			if (!sflag)
565 				ansiu = 1;
566 		}
567 		break;
568 		/* LINTED (enumeration values not handled in switch) */
569 	case STRUCT:
570 	case VOID:
571 	case LDOUBLE:
572 	case FUNC:
573 	case ARRAY:
574 	case PTR:
575 	case ENUM:
576 	case UNION:
577 	case SIGNED:
578 	case NOTSPEC:
579 	case DOUBLE:
580 	case FLOAT:
581 	case UQUAD:
582 	case ULONG:
583 	case USHORT:
584 	case SHORT:
585 	case UCHAR:
586 	case SCHAR:
587 	case CHAR:
588 	case UNSIGN:
589 		break;
590 	}
591 
592 	if (typ != QUAD && typ != UQUAD) {
593 		if (isutyp(typ)) {
594 			uq = ul;
595 		} else {
596 			uq = (int64_t)(long)ul;
597 		}
598 	}
599 
600 	uq = (uint64_t)xsign((int64_t)uq, typ, -1);
601 
602 	if ((yylval.y_val = calloc(1, sizeof(val_t))) == NULL)
603 		nomem();
604 	yylval.y_val->v_tspec = typ;
605 	yylval.y_val->v_ansiu = ansiu;
606 	yylval.y_val->v_quad = (int64_t)uq;
607 
608 	return (T_CON);
609 }
610 
611 /*
612  * Returns 1 if t is a signed type and the value is negative.
613  *
614  * len is the number of significant bits. If len is -1, len is set
615  * to the width of type t.
616  */
617 int
sign(int64_t q,tspec_t t,int len)618 sign(int64_t q, tspec_t t, int len)
619 {
620 
621 	if (t == PTR || isutyp(t))
622 		return (0);
623 	return (msb(q, t, len));
624 }
625 
626 int
msb(int64_t q,tspec_t t,int len)627 msb(int64_t q, tspec_t t, int len)
628 {
629 
630 	if (len <= 0)
631 		len = size(t);
632 	return ((q & qbmasks[len - 1]) != 0);
633 }
634 
635 /*
636  * Extends the sign of q.
637  */
638 int64_t
xsign(int64_t q,tspec_t t,int len)639 xsign(int64_t q, tspec_t t, int len)
640 {
641 
642 	if (len <= 0)
643 		len = size(t);
644 
645 	if (t == PTR || isutyp(t) || !sign(q, t, len)) {
646 		q &= qlmasks[len];
647 	} else {
648 		q |= qumasks[len];
649 	}
650 	return (q);
651 }
652 
653 /*
654  * Convert a string representing a floating point value into its interal
655  * representation. Type and value are returned in yylval. fcon()
656  * (and yylex()) returns T_CON.
657  * XXX Currently it is not possible to convert constants of type
658  * long double which are greater than DBL_MAX.
659  */
660 static int
fcon(void)661 fcon(void)
662 {
663 	const	char *cp;
664 	int	len;
665 	tspec_t typ;
666 	char	c, *eptr;
667 	double	d;
668 	float	f = 0;
669 
670 	cp = yytext;
671 	len = yyleng;
672 
673 	if ((c = cp[len - 1]) == 'f' || c == 'F') {
674 		typ = FLOAT;
675 		len--;
676 	} else if (c == 'l' || c == 'L') {
677 		typ = LDOUBLE;
678 		len--;
679 	} else {
680 		typ = DOUBLE;
681 	}
682 
683 	if (tflag && typ != DOUBLE) {
684 		/* suffixes F and L are illegal in traditional C */
685 		warning(98);
686 	}
687 
688 	errno = 0;
689 	d = strtod(cp, &eptr);
690 	if (eptr != cp + len) {
691 		switch (*eptr) {
692 		/*
693 		 * XXX: non-native non-current strtod() may not handle hex
694 		 * floats, ignore the rest if we find traces of hex float
695 		 * syntax...
696 		 */
697 		case 'p':
698 		case 'P':
699 		case 'x':
700 		case 'X':
701 			d = 0;
702 			errno = 0;
703 			break;
704 		default:
705 			LERROR("fcon()");
706 		}
707 	}
708 	if (errno != 0)
709 		/* floating-point constant out of range */
710 		warning(248);
711 
712 	if (typ == FLOAT) {
713 		f = (float)d;
714 		if (!finite(f)) {
715 			/* floating-point constant out of range */
716 			warning(248);
717 			f = f > 0 ? FLT_MAX : -FLT_MAX;
718 		}
719 	}
720 
721 	if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL)
722 		nomem();
723 	yylval.y_val->v_tspec = typ;
724 	if (typ == FLOAT) {
725 		yylval.y_val->v_ldbl = f;
726 	} else {
727 		yylval.y_val->v_ldbl = d;
728 	}
729 
730 	return (T_CON);
731 }
732 
733 static int
operator(int t,op_t o)734 operator(int t, op_t o)
735 {
736 
737 	yylval.y_op = o;
738 	return (t);
739 }
740 
741 /*
742  * Called if lex found a leading \'.
743  */
744 static int
ccon(void)745 ccon(void)
746 {
747 	int	n, val, c;
748 	char	cv;
749 
750 	n = 0;
751 	val = 0;
752 	while ((c = getescc('\'')) >= 0) {
753 		val = (val << CHAR_BIT) + c;
754 		n++;
755 	}
756 	if (c == -2) {
757 		/* unterminated character constant */
758 		error(253);
759 	} else {
760 		if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
761 			/* too many characters in character constant */
762 			error(71);
763 		} else if (n > 1) {
764 			/* multi-character character constant */
765 			warning(294);
766 		} else if (n == 0) {
767 			/* empty character constant */
768 			error(73);
769 		}
770 	}
771 	if (n == 1) {
772 		cv = (char)val;
773 		val = cv;
774 	}
775 
776 	yylval.y_val = xcalloc(1, sizeof (val_t));
777 	yylval.y_val->v_tspec = INT;
778 	yylval.y_val->v_quad = val;
779 
780 	return (T_CON);
781 }
782 
783 /*
784  * Called if lex found a leading L\'
785  */
786 static int
wccon(void)787 wccon(void)
788 {
789 	static	char buf[MB_LEN_MAX + 1];
790 	int	i, c;
791 	wchar_t	wc;
792 
793 	i = 0;
794 	while ((c = getescc('\'')) >= 0) {
795 		if (i < MB_CUR_MAX)
796 			buf[i] = (char)c;
797 		i++;
798 	}
799 
800 	wc = 0;
801 
802 	if (c == -2) {
803 		/* unterminated character constant */
804 		error(253);
805 	} else if (c == 0) {
806 		/* empty character constant */
807 		error(73);
808 	} else {
809 		if (i > MB_CUR_MAX) {
810 			i = MB_CUR_MAX;
811 			/* too many characters in character constant */
812 			error(71);
813 		} else {
814 			buf[i] = '\0';
815 			(void)mbtowc(NULL, NULL, 0);
816 			if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
817 				/* invalid multibyte character */
818 				error(291);
819 		}
820 	}
821 
822 	if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL)
823 		nomem();
824 	yylval.y_val->v_tspec = WCHAR;
825 	yylval.y_val->v_quad = wc;
826 
827 	return (T_CON);
828 }
829 
830 /*
831  * Read a character which is part of a character constant or of a string
832  * and handle escapes.
833  *
834  * The Argument is the character which delimits the character constant or
835  * string.
836  *
837  * Returns -1 if the end of the character constant or string is reached,
838  * -2 if the EOF is reached, and the character otherwise.
839  */
840 static int
getescc(int d)841 getescc(int d)
842 {
843 	static	int pbc = -1;
844 	int	n, c, v;
845 
846 	if (pbc == -1) {
847 		c = inpc();
848 	} else {
849 		c = pbc;
850 		pbc = -1;
851 	}
852 	if (c == d)
853 		return (-1);
854 	switch (c) {
855 	case '\n':
856 		if (tflag) {
857 			/* newline in string or char constant */
858 			error(254);
859 			return (-2);
860 		}
861 		return (c);
862 	case EOF:
863 		return (-2);
864 	case '\\':
865 		switch (c = inpc()) {
866 		case '"':
867 			if (tflag && d == '\'')
868 				/* \" inside character constant undef. ... */
869 				warning(262);
870 			return ('"');
871 		case '\'':
872 			return ('\'');
873 		case '?':
874 			if (tflag)
875 				/* \? undefined in traditional C */
876 				warning(263);
877 			return ('?');
878 		case '\\':
879 			return ('\\');
880 		case 'a':
881 			if (tflag)
882 				/* \a undefined in traditional C */
883 				warning(81);
884 			return ('\a');
885 		case 'b':
886 			return ('\b');
887 		case 'f':
888 			return ('\f');
889 		case 'n':
890 			return ('\n');
891 		case 'r':
892 			return ('\r');
893 		case 't':
894 			return ('\t');
895 		case 'v':
896 			if (tflag)
897 				/* \v undefined in traditional C */
898 				warning(264);
899 			return ('\v');
900 		case '8': case '9':
901 			/* bad octal digit %c */
902 			warning(77, c);
903 			/* FALLTHROUGH */
904 		case '0': case '1': case '2': case '3':
905 		case '4': case '5': case '6': case '7':
906 			n = 3;
907 			v = 0;
908 			do {
909 				v = (v << 3) + (c - '0');
910 				c = inpc();
911 			} while (--n && isdigit(c) && (tflag || c <= '7'));
912 			if (tflag && n > 0 && isdigit(c))
913 				/* bad octal digit %c */
914 				warning(77, c);
915 			pbc = c;
916 			if (v > UCHAR_MAX) {
917 				/* character escape does not fit in char. */
918 				warning(76);
919 				v &= CHAR_MASK;
920 			}
921 			return (v);
922 		case 'x':
923 			if (tflag)
924 				/* \x undefined in traditional C */
925 				warning(82);
926 			v = 0;
927 			n = 0;
928 			while ((c = inpc()) >= 0 && isxdigit(c)) {
929 				c = isdigit(c) ?
930 					c - '0' : toupper(c) - 'A' + 10;
931 				v = (v << 4) + c;
932 				if (n >= 0) {
933 					if ((v & ~CHAR_MASK) != 0) {
934 						/* overflow in hex escape */
935 						warning(75);
936 						n = -1;
937 					} else {
938 						n++;
939 					}
940 				}
941 			}
942 			pbc = c;
943 			if (n == 0) {
944 				/* no hex digits follow \x */
945 				error(74);
946 			} if (n == -1) {
947 				v &= CHAR_MASK;
948 			}
949 			return (v);
950 		case '\n':
951 			return (getescc(d));
952 		case EOF:
953 			return (-2);
954 		default:
955 			if (isprint(c)) {
956 				/* dubious escape \%c */
957 				warning(79, c);
958 			} else {
959 				/* dubious escape \%o */
960 				warning(80, c);
961 			}
962 		}
963 	}
964 	return (c);
965 }
966 
967 /*
968  * Called for preprocessor directives. Currently implemented are:
969  *	# lineno
970  *	# lineno "filename"
971  */
972 static void
directive(void)973 directive(void)
974 {
975 	const	char *cp, *fn;
976 	char	c, *eptr;
977 	size_t	fnl;
978 	long	ln;
979 	static	int first = 1;
980 
981 	/* Go to first non-whitespace after # */
982 	for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
983 		continue;
984 
985 	if (!isdigit((unsigned char)c)) {
986 	error:
987 		/* undefined or invalid # directive */
988 		warning(255);
989 		return;
990 	}
991 	ln = strtol(--cp, &eptr, 10);
992 	if (cp == eptr)
993 		goto error;
994 	if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
995 		goto error;
996 	while ((c = *cp++) == ' ' || c == '\t')
997 		continue;
998 	if (c != '\0') {
999 		if (c != '"')
1000 			goto error;
1001 		fn = cp;
1002 		while ((c = *cp) != '"' && c != '\0')
1003 			cp++;
1004 		if (c != '"')
1005 			goto error;
1006 		if ((fnl = cp++ - fn) > PATH_MAX)
1007 			goto error;
1008 		while ((c = *cp++) == ' ' || c == '\t')
1009 			continue;
1010 #if 0
1011 		if (c != '\0')
1012 			warning("extra character(s) after directive");
1013 #endif
1014 
1015 		/* empty string means stdin */
1016 		if (fnl == 0) {
1017 			fn = "{standard input}";
1018 			fnl = 16;			/* strlen (fn) */
1019 		}
1020 		curr_pos.p_file = fnnalloc(fn, fnl);
1021 		/*
1022 		 * If this is the first directive, the name is the name
1023 		 * of the C source file as specified at the command line.
1024 		 * It is written to the output file.
1025 		 */
1026 		if (first) {
1027 			csrc_pos.p_file = curr_pos.p_file;
1028 			outsrc(curr_pos.p_file);
1029 			first = 0;
1030 		}
1031 	}
1032 	curr_pos.p_line = (int)ln - 1;
1033 	curr_pos.p_uniq = 0;
1034 	if (curr_pos.p_file == csrc_pos.p_file) {
1035 		csrc_pos.p_line = (int)ln - 1;
1036 		csrc_pos.p_uniq = 0;
1037 	}
1038 }
1039 
1040 /*
1041  * Handle lint comments. Following comments are currently understood:
1042  *	ARGSUSEDn
1043  *	BITFIELDTYPE
1044  *	CONSTCOND CONSTANTCOND CONSTANTCONDITION
1045  *	FALLTHRU FALLTHROUGH
1046  *	LINTLIBRARY
1047  *	LINTED NOSTRICT
1048  *	LONGLONG
1049  *	NOTREACHED
1050  *	PRINTFLIKEn
1051  *	PROTOLIB
1052  *	SCANFLIKEn
1053  *	VARARGSn
1054  * If one of this comments is recognized, the arguments, if any, are
1055  * parsed and a function which handles this comment is called.
1056  */
1057 static void
comment(void)1058 comment(void)
1059 {
1060 	int	c, lc;
1061 	static struct {
1062 		const	char *keywd;
1063 		int	arg;
1064 		void	(*func)(int);
1065 	} keywtab[] = {
1066 		{ "ARGSUSED",		1,	argsused	},
1067 		{ "BITFIELDTYPE",	0,	bitfieldtype	},
1068 		{ "CONSTCOND",		0,	constcond	},
1069 		{ "CONSTANTCOND",	0,	constcond	},
1070 		{ "CONSTANTCONDITION",	0,	constcond	},
1071 		{ "FALLTHRU",		0,	fallthru	},
1072 		{ "FALLTHROUGH",	0,	fallthru	},
1073 		{ "LINTLIBRARY",	0,	lintlib		},
1074 		{ "LINTED",		0,	linted		},
1075 		{ "LONGLONG",		0,	longlong	},
1076 		{ "NOSTRICT",		0,	linted		},
1077 		{ "NOTREACHED",		0,	notreach	},
1078 		{ "PRINTFLIKE",		1,	printflike	},
1079 		{ "PROTOLIB",		1,	protolib	},
1080 		{ "SCANFLIKE",		1,	scanflike	},
1081 		{ "VARARGS",		1,	varargs		},
1082 	};
1083 	char	keywd[32];
1084 	char	arg[32];
1085 	int	l, i, a;
1086 	int	eoc;
1087 
1088 	eoc = 0;
1089 
1090 	/* Skip white spaces after the start of the comment */
1091 	while ((c = inpc()) != EOF && isspace(c))
1092 		continue;
1093 
1094 	/* Read the potential keyword to keywd */
1095 	l = 0;
1096 	while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1097 		keywd[l++] = (char)c;
1098 		c = inpc();
1099 	}
1100 	keywd[l] = '\0';
1101 
1102 	/* look for the keyword */
1103 	for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1104 		if (strcmp(keywtab[i].keywd, keywd) == 0)
1105 			break;
1106 	}
1107 	if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1108 		goto skip_rest;
1109 
1110 	/* skip white spaces after the keyword */
1111 	while (c != EOF && isspace(c))
1112 		c = inpc();
1113 
1114 	/* read the argument, if the keyword accepts one and there is one */
1115 	l = 0;
1116 	if (keywtab[i].arg) {
1117 		while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1118 			arg[l++] = (char)c;
1119 			c = inpc();
1120 		}
1121 	}
1122 	arg[l] = '\0';
1123 	a = l != 0 ? atoi(arg) : -1;
1124 
1125 	/* skip white spaces after the argument */
1126 	while (c != EOF && isspace(c))
1127 		c = inpc();
1128 
1129 	if (c != '*' || (c = inpc()) != '/') {
1130 		if (keywtab[i].func != linted)
1131 			/* extra characters in lint comment */
1132 			warning(257);
1133 	} else {
1134 		/*
1135 		 * remember that we have already found the end of the
1136 		 * comment
1137 		 */
1138 		eoc = 1;
1139 	}
1140 
1141 	if (keywtab[i].func != NULL)
1142 		(*keywtab[i].func)(a);
1143 
1144  skip_rest:
1145 	while (!eoc) {
1146 		lc = c;
1147 		if ((c = inpc()) == EOF) {
1148 			/* unterminated comment */
1149 			error(256);
1150 			break;
1151 		}
1152 		if (lc == '*' && c == '/')
1153 			eoc = 1;
1154 	}
1155 }
1156 
1157 /*
1158  * Handle // style comments
1159  */
1160 static void
slashslashcomment(void)1161 slashslashcomment(void)
1162 {
1163 	int c;
1164 
1165 	if (!Sflag && !gflag)
1166 		/* // comments only supported in C99 */
1167 		(void)gnuism(312, tflag ? "traditional" : "ANSI");
1168 
1169 	while ((c = inpc()) != EOF && c != '\n')
1170 		continue;
1171 }
1172 
1173 /*
1174  * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1175  * clrwflgs() is called after function definitions and global and
1176  * local declarations and definitions. It is also called between
1177  * the controlling expression and the body of control statements
1178  * (if, switch, for, while).
1179  */
1180 void
clrwflgs(void)1181 clrwflgs(void)
1182 {
1183 
1184 	nowarn = 0;
1185 	quadflg = 0;
1186 	ccflg = 0;
1187 }
1188 
1189 /*
1190  * Strings are stored in a dynamically alloceted buffer and passed
1191  * in yylval.y_xstrg to the parser. The parser or the routines called
1192  * by the parser are responsible for freeing this buffer.
1193  */
1194 static int
string(void)1195 string(void)
1196 {
1197 	u_char	*s;
1198 	int	c;
1199 	size_t	len, max;
1200 	strg_t	*strg;
1201 
1202 	if ((s = malloc(max = 64)) == NULL)
1203 		nomem();
1204 
1205 	len = 0;
1206 	while ((c = getescc('"')) >= 0) {
1207 		/* +1 to reserve space for a trailing NUL character */
1208 		if (len + 1 == max)
1209 			if ((s = realloc(s, max *= 2)) == NULL)
1210 				nomem();
1211 		s[len++] = (char)c;
1212 	}
1213 	s[len] = '\0';
1214 	if (c == -2)
1215 		/* unterminated string constant */
1216 		error(258);
1217 
1218 	if ((strg = calloc(1, sizeof (strg_t))) == NULL)
1219 		nomem();
1220 	strg->st_tspec = CHAR;
1221 	strg->st_len = len;
1222 	strg->st_cp = s;
1223 
1224 	yylval.y_strg = strg;
1225 	return (T_STRING);
1226 }
1227 
1228 static int
wcstrg(void)1229 wcstrg(void)
1230 {
1231 	char	*s;
1232 	int	c, i, n, wi;
1233 	size_t	len, max, wlen;
1234 	wchar_t	*ws;
1235 	strg_t	*strg;
1236 
1237 	if ((s = malloc(max = 64)) == NULL)
1238 		nomem();
1239 	len = 0;
1240 	while ((c = getescc('"')) >= 0) {
1241 		/* +1 to save space for a trailing NUL character */
1242 		if (len + 1 >= max)
1243 			if ((s = realloc(s, max *= 2)) == NULL)
1244 				nomem();
1245 		s[len++] = (char)c;
1246 	}
1247 	s[len] = '\0';
1248 	if (c == -2)
1249 		/* unterminated string constant */
1250 		error(258);
1251 
1252 	/* get length of wide character string */
1253 	(void)mblen(NULL, 0);
1254 	for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1255 		if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1256 			/* invalid multibyte character */
1257 			error(291);
1258 			break;
1259 		}
1260 		if (n == 0)
1261 			n = 1;
1262 	}
1263 
1264 	if ((ws = malloc((wlen + 1) * sizeof (wchar_t))) == NULL)
1265 		nomem();
1266 
1267 	/* convert from multibyte to wide char */
1268 	(void)mbtowc(NULL, NULL, 0);
1269 	for (i = 0, wi = 0; i < len; i += n, wi++) {
1270 		if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1271 			break;
1272 		if (n == 0)
1273 			n = 1;
1274 	}
1275 	ws[wi] = 0;
1276 	free(s);
1277 
1278 	if ((strg = calloc(1, sizeof (strg_t))) == NULL)
1279 		nomem();
1280 	strg->st_tspec = WCHAR;
1281 	strg->st_len = wlen;
1282 	strg->st_wcp = ws;
1283 
1284 	yylval.y_strg = strg;
1285 	return (T_STRING);
1286 }
1287 
1288 /*
1289  * As noted above the scanner does not create new symbol table entries
1290  * for symbols it cannot find in the symbol table. This is to avoid
1291  * putting undeclared symbols into the symbol table if a syntax error
1292  * occurs.
1293  *
1294  * getsym() is called as soon as it is probably ok to put the symbol to
1295  * the symbol table. This does not mean that it is not possible that
1296  * symbols are put to the symbol table which are than not completely
1297  * declared due to syntax errors. To avoid too many problems in this
1298  * case symbols get type int in getsym().
1299  *
1300  * XXX calls to getsym() should be delayed until decl1*() is called
1301  */
1302 sym_t *
getsym(sbuf_t * sb)1303 getsym(sbuf_t *sb)
1304 {
1305 	dinfo_t	*di;
1306 	char	*s;
1307 	sym_t	*sym;
1308 
1309 	sym = sb->sb_sym;
1310 
1311 	/*
1312 	 * During member declaration it is possible that name() looked
1313 	 * for symbols of type FVFT, although it should have looked for
1314 	 * symbols of type FTAG. Same can happen for labels. Both cases
1315 	 * are compensated here.
1316 	 */
1317 	if (symtyp == FMOS || symtyp == FLAB) {
1318 		if (sym == NULL || sym->s_kind == FVFT)
1319 			sym = search(sb);
1320 	}
1321 
1322 	if (sym != NULL) {
1323 		if (sym->s_kind != symtyp)
1324 			LERROR("storesym()");
1325 		symtyp = FVFT;
1326 		freesb(sb);
1327 		return (sym);
1328 	}
1329 
1330 	/* create a new symbol table entry */
1331 
1332 	/* labels must always be allocated at level 1 (outhermost block) */
1333 	if (symtyp == FLAB) {
1334 		sym = getlblk(1, sizeof (sym_t));
1335 		s = getlblk(1, sb->sb_len + 1);
1336 		(void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1337 		sym->s_name = s;
1338 		sym->s_blklev = 1;
1339 		di = dcs;
1340 		while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1341 			di = di->d_nxt;
1342 		if (di->d_ctx != AUTO)
1343 			LERROR("storesym()");
1344 	} else {
1345 		sym = getblk(sizeof (sym_t));
1346 		sym->s_name = sb->sb_name;
1347 		sym->s_blklev = blklev;
1348 		di = dcs;
1349 	}
1350 
1351 	UNIQUE_CURR_POS(sym->s_dpos);
1352 	if ((sym->s_kind = symtyp) != FLAB)
1353 		sym->s_type = gettyp(INT);
1354 
1355 	symtyp = FVFT;
1356 
1357 	if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1358 		symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1359 	(symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash];
1360 
1361 	*di->d_ldlsym = sym;
1362 	di->d_ldlsym = &sym->s_dlnxt;
1363 
1364 	freesb(sb);
1365 	return (sym);
1366 }
1367 
1368 /*
1369  * Construct a temporary symbol. The symbol starts with a digit, so that
1370  * it is illegal.
1371  */
1372 sym_t *
mktempsym(type_t * t)1373 mktempsym(type_t *t)
1374 {
1375 	static int n = 0;
1376 	int h;
1377 	char *s = getlblk(blklev, 64);
1378 	sym_t *sym = getblk(sizeof (sym_t));
1379 
1380 	(void)snprintf(s, 64, "%.8d_tmp", n++);
1381 	h = hash(s);
1382 
1383 	sym->s_name = s;
1384 	sym->s_type = t;
1385 	sym->s_blklev = blklev;
1386 	sym->s_scl = AUTO;
1387 	sym->s_kind = FVFT;
1388 	sym->s_used = 1;
1389 	sym->s_set = 1;
1390 
1391 	if ((sym->s_link = symtab[h]) != NULL)
1392 		symtab[h]->s_rlink = &sym->s_link;
1393 	(symtab[h] = sym)->s_rlink = &symtab[h];
1394 
1395 	*dcs->d_ldlsym = sym;
1396 	dcs->d_ldlsym = &sym->s_dlnxt;
1397 
1398 	return sym;
1399 }
1400 
1401 /*
1402  * Remove a symbol forever from the symbol table. s_blklev
1403  * is set to -1 to avoid that the symbol will later be put
1404  * back to the symbol table.
1405  */
1406 void
rmsym(sym_t * sym)1407 rmsym(sym_t *sym)
1408 {
1409 
1410 	if ((*sym->s_rlink = sym->s_link) != NULL)
1411 		sym->s_link->s_rlink = sym->s_rlink;
1412 	sym->s_blklev = -1;
1413 	sym->s_link = NULL;
1414 }
1415 
1416 /*
1417  * Remove a list of symbols declared at one level from the symbol
1418  * table.
1419  */
1420 void
rmsyms(sym_t * syms)1421 rmsyms(sym_t *syms)
1422 {
1423 	sym_t	*sym;
1424 
1425 	for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1426 		if (sym->s_blklev != -1) {
1427 			if ((*sym->s_rlink = sym->s_link) != NULL)
1428 				sym->s_link->s_rlink = sym->s_rlink;
1429 			sym->s_link = NULL;
1430 			sym->s_rlink = NULL;
1431 		}
1432 	}
1433 }
1434 
1435 /*
1436  * Put a symbol into the symbol table
1437  */
1438 void
inssym(int bl,sym_t * sym)1439 inssym(int bl, sym_t *sym)
1440 {
1441 	int	h;
1442 
1443 	h = hash(sym->s_name);
1444 	if ((sym->s_link = symtab[h]) != NULL)
1445 		symtab[h]->s_rlink = &sym->s_link;
1446 	(symtab[h] = sym)->s_rlink = &symtab[h];
1447 	sym->s_blklev = bl;
1448 	if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1449 		LERROR("inssym()");
1450 }
1451 
1452 /*
1453  * Called at level 0 after syntax errors
1454  * Removes all symbols which are not declared at level 0 from the
1455  * symbol table. Also frees all memory which is not associated with
1456  * level 0.
1457  */
1458 void
cleanup(void)1459 cleanup(void)
1460 {
1461 	sym_t	*sym, *nsym;
1462 	int	i;
1463 
1464 	for (i = 0; i < HSHSIZ1; i++) {
1465 		for (sym = symtab[i]; sym != NULL; sym = nsym) {
1466 			nsym = sym->s_link;
1467 			if (sym->s_blklev >= 1) {
1468 				if ((*sym->s_rlink = nsym) != NULL)
1469 					nsym->s_rlink = sym->s_rlink;
1470 			}
1471 		}
1472 	}
1473 
1474 	for (i = mblklev; i > 0; i--)
1475 		freelblk(i);
1476 }
1477 
1478 /*
1479  * Create a new symbol with the name of an existing symbol.
1480  */
1481 sym_t *
pushdown(sym_t * sym)1482 pushdown(sym_t *sym)
1483 {
1484 	int	h;
1485 	sym_t	*nsym;
1486 
1487 	h = hash(sym->s_name);
1488 	nsym = getblk(sizeof (sym_t));
1489 	if (sym->s_blklev > blklev)
1490 		LERROR("pushdown()");
1491 	nsym->s_name = sym->s_name;
1492 	UNIQUE_CURR_POS(nsym->s_dpos);
1493 	nsym->s_kind = sym->s_kind;
1494 	nsym->s_blklev = blklev;
1495 
1496 	if ((nsym->s_link = symtab[h]) != NULL)
1497 		symtab[h]->s_rlink = &nsym->s_link;
1498 	(symtab[h] = nsym)->s_rlink = &symtab[h];
1499 
1500 	*dcs->d_ldlsym = nsym;
1501 	dcs->d_ldlsym = &nsym->s_dlnxt;
1502 
1503 	return (nsym);
1504 }
1505 
1506 /*
1507  * Free any dynamically allocated memory referenced by
1508  * the value stack or yylval.
1509  * The type of information in yylval is described by tok.
1510  */
1511 void
freeyyv(void * sp,int tok)1512 freeyyv(void *sp, int tok)
1513 {
1514 	if (tok == T_NAME || tok == T_TYPENAME) {
1515 		sbuf_t *sb = *(sbuf_t **)sp;
1516 		freesb(sb);
1517 	} else if (tok == T_CON) {
1518 		val_t *val = *(val_t **)sp;
1519 		free(val);
1520 	} else if (tok == T_STRING) {
1521 		strg_t *strg = *(strg_t **)sp;
1522 		if (strg->st_tspec == CHAR) {
1523 			free(strg->st_cp);
1524 		} else if (strg->st_tspec == WCHAR) {
1525 			free(strg->st_wcp);
1526 		} else {
1527 			LERROR("fryylv()");
1528 		}
1529 		free(strg);
1530 	}
1531 }
1532