xref: /netbsd/usr.bin/xlint/lint2/read.c (revision bf9ec67e)
1 /* $NetBSD: read.c,v 1.13 2002/01/31 19:36:56 tv 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 #include <sys/cdefs.h>
36 #if defined(__RCSID) && !defined(lint)
37 __RCSID("$NetBSD: read.c,v 1.13 2002/01/31 19:36:56 tv Exp $");
38 #endif
39 
40 #include <ctype.h>
41 #include <limits.h>
42 #include <stdio.h>
43 #include <stdlib.h>
44 #include <string.h>
45 
46 #include "lint2.h"
47 
48 
49 /* index of current (included) source file */
50 static	int	srcfile;
51 
52 /*
53  * The array pointed to by inpfns maps the file name indices of input files
54  * to the file name indices used in lint2
55  */
56 static	short	*inpfns;
57 static	size_t	ninpfns;
58 
59 /*
60  * The array pointed to by *fnames maps file name indizes to file names.
61  * Indices of type short are used instead of pointers to save memory.
62  */
63 const	char **fnames;
64 static	size_t	nfnames;
65 
66 /*
67  * Types are shared (to save memory for the types itself) and accessed
68  * via indices (to save memory for references to types (indices are short)).
69  * To share types, a equal type must be located fast. This is done by a
70  * hash table. Access by indices is done via an array of pointers to the
71  * types.
72  */
73 typedef struct thtab {
74 	const	char *th_name;
75 	u_short	th_idx;
76 	struct	thtab *th_nxt;
77 } thtab_t;
78 static	thtab_t	**thtab;		/* hash table */
79 type_t	**tlst;				/* array for indexed access */
80 static	size_t	tlstlen;		/* length of tlst */
81 
82 static	hte_t **renametab;
83 
84 /* index of current C source file (as spezified at the command line) */
85 static	int	csrcfile;
86 
87 
88 static	void	inperr(void);
89 static	void	setsrc(const char *);
90 static	void	setfnid(int, const char *);
91 static	void	funccall(pos_t *, const char *);
92 static	void	decldef(pos_t *, const char *);
93 static	void	usedsym(pos_t *, const char *);
94 static	u_short	inptype(const char *, const char **);
95 static	int	gettlen(const char *, const char **);
96 static	u_short	findtype(const char *, size_t, int);
97 static	u_short	storetyp(type_t *, const char *, size_t, int);
98 static	int	thash(const char *, size_t);
99 static	char	*inpqstrg(const char *, const char **);
100 static	const	char *inpname(const char *, const char **);
101 static	int	getfnidx(const char *);
102 
103 void
104 readfile(const char *name)
105 {
106 	FILE	*inp;
107 	size_t	len;
108 	const	char *cp;
109 	char	*line, *eptr, rt = '\0';
110 	int	cline, isrc, iline;
111 	pos_t	pos;
112 
113 	if (inpfns == NULL)
114 		inpfns = xcalloc(ninpfns = 128, sizeof (short));
115 	if (fnames == NULL)
116 		fnames = xcalloc(nfnames = 256, sizeof (char *));
117 	if (tlstlen == 0)
118 		tlst = xcalloc(tlstlen = 256, sizeof (type_t *));
119 	if (thtab == NULL)
120 		thtab = xcalloc(THSHSIZ2, sizeof (thtab_t));
121 
122 	_inithash(&renametab);
123 
124 	srcfile = getfnidx(name);
125 
126 	if ((inp = fopen(name, "r")) == NULL)
127 		err(1, "cannot open %s", name);
128 
129 	while ((line = fgetln(inp, &len)) != NULL) {
130 
131 		if (len == 0 || line[len - 1] != '\n')
132 			inperr();
133 		line[len - 1] = '\0';
134 		cp = line;
135 
136 		/* line number in csrcfile */
137 		cline = (int)strtol(cp, &eptr, 10);
138 		if (cp == eptr) {
139 		        cline = -1;
140 		} else {
141 			cp = eptr;
142 		}
143 
144 		/* record type */
145 		if (*cp != '\0') {
146 			rt = *cp++;
147 		} else {
148 			inperr();
149 		}
150 
151 		if (rt == 'S') {
152 			setsrc(cp);
153 			continue;
154 		} else if (rt == 's') {
155 			setfnid(cline, cp);
156 			continue;
157 		}
158 
159 		/*
160 		 * Index of (included) source file. If this index is
161 		 * different from csrcfile, it refers to an included
162 		 * file.
163 		 */
164 		isrc = (int)strtol(cp, &eptr, 10);
165 		if (cp == eptr)
166 			inperr();
167 		cp = eptr;
168 		isrc = inpfns[isrc];
169 
170 		/* line number in isrc */
171 		if (*cp++ != '.')
172 			inperr();
173 		iline = (int)strtol(cp, &eptr, 10);
174 		if (cp == eptr)
175 			inperr();
176 		cp = eptr;
177 
178 		pos.p_src = (u_short)csrcfile;
179 		pos.p_line = (u_short)cline;
180 		pos.p_isrc = (u_short)isrc;
181 		pos.p_iline = (u_short)iline;
182 
183 		/* process rest of this record */
184 		switch (rt) {
185 		case 'c':
186 			funccall(&pos, cp);
187 			break;
188 		case 'd':
189 			decldef(&pos, cp);
190 			break;
191 		case 'u':
192 			usedsym(&pos, cp);
193 			break;
194 		default:
195 			inperr();
196 		}
197 
198 	}
199 
200 	_destroyhash(renametab);
201 
202 	if (ferror(inp))
203 		err(1, "read error on %s", name);
204 
205 	(void)fclose(inp);
206 }
207 
208 
209 static void
210 inperr(void)
211 {
212 
213 	errx(1, "input file error: %s", fnames[srcfile]);
214 }
215 
216 /*
217  * Set the name of the C source file of the .ln file which is
218  * currently read.
219  */
220 static void
221 setsrc(const char *cp)
222 {
223 
224 	csrcfile = getfnidx(cp);
225 }
226 
227 /*
228  * setfnid() gets as input an index as used in an input file and the
229  * associated file name. If necessary, it creates a new lint2 file
230  * name index for this file name and creates the mapping of the index
231  * as used in the input file to the index used in lint2.
232  */
233 static void
234 setfnid(int fid, const char *cp)
235 {
236 
237 	if (fid == -1)
238 		inperr();
239 
240 	if (fid >= ninpfns) {
241 		inpfns = xrealloc(inpfns, (ninpfns * 2) * sizeof (short));
242 		(void)memset(inpfns + ninpfns, 0, ninpfns * sizeof (short));
243 		ninpfns *= 2;
244 	}
245 	/*
246 	 * Should always be true because indices written in the output
247 	 * file by lint1 are always the previous index + 1.
248 	 */
249 	if (fid >= ninpfns)
250 		errx(1, "internal error: setfnid()");
251 	inpfns[fid] = (u_short)getfnidx(cp);
252 }
253 
254 /*
255  * Process a function call record (c-record).
256  */
257 static void
258 funccall(pos_t *posp, const char *cp)
259 {
260 	arginf_t *ai, **lai;
261 	char	c, *eptr;
262 	int	rused, rdisc;
263 	hte_t	*hte;
264 	fcall_t	*fcall;
265 	const char *name;
266 
267 	fcall = xalloc(sizeof (fcall_t));
268 	STRUCT_ASSIGN(fcall->f_pos, *posp);
269 
270 	/* read flags */
271 	rused = rdisc = 0;
272 	lai = &fcall->f_args;
273 	while ((c = *cp) == 'u' || c == 'i' || c == 'd' ||
274 	       c == 'z' || c == 'p' || c == 'n' || c == 's') {
275 		cp++;
276 		switch (c) {
277 		case 'u':
278 			if (rused || rdisc)
279 				inperr();
280 			rused = 1;
281 			break;
282 		case 'i':
283 			if (rused || rdisc)
284 				inperr();
285 			break;
286 		case 'd':
287 			if (rused || rdisc)
288 				inperr();
289 			rdisc = 1;
290 			break;
291 		case 'z':
292 		case 'p':
293 		case 'n':
294 		case 's':
295 			ai = xalloc(sizeof (arginf_t));
296 			ai->a_num = (int)strtol(cp, &eptr, 10);
297 			if (cp == eptr)
298 				inperr();
299 			cp = eptr;
300 			if (c == 'z') {
301 				ai->a_pcon = ai->a_zero = 1;
302 			} else if (c == 'p') {
303 				ai->a_pcon = 1;
304 			} else if (c == 'n') {
305 				ai->a_ncon = 1;
306 			} else {
307 				ai->a_fmt = 1;
308 				ai->a_fstrg = inpqstrg(cp, &cp);
309 			}
310 			*lai = ai;
311 			lai = &ai->a_nxt;
312 			break;
313 		}
314 	}
315 	fcall->f_rused = rused;
316 	fcall->f_rdisc = rdisc;
317 
318 	/* read name of function */
319 	name = inpname(cp, &cp);
320 
321 	/* first look it up in the renaming table, then in the normal table */
322 	hte = _hsearch(renametab, name, 0);
323 	if (hte != NULL)
324 		hte = hte->h_hte;
325 	else
326 		hte = hsearch(name, 1);
327 	hte->h_used = 1;
328 
329 	fcall->f_type = inptype(cp, &cp);
330 
331 	*hte->h_lcall = fcall;
332 	hte->h_lcall = &fcall->f_nxt;
333 
334 	if (*cp != '\0')
335 		inperr();
336 }
337 
338 /*
339  * Process a declaration or definition (d-record).
340  */
341 static void
342 decldef(pos_t *posp, const char *cp)
343 {
344 	sym_t	*symp, sym;
345 	char	c, *ep, *pos1;
346 	int	used, renamed;
347 	hte_t	*hte, *renamehte = NULL;
348 	const char *name, *rename;
349 
350 	(void)memset(&sym, 0, sizeof (sym));
351 	STRUCT_ASSIGN(sym.s_pos, *posp);
352 	sym.s_def = NODECL;
353 
354 	used = 0;
355 
356 	while ((c = *cp) == 't' || c == 'd' || c == 'e' || c == 'u' ||
357 	       c == 'r' || c == 'o' || c == 's' || c == 'v' ||
358 	       c == 'P' || c == 'S') {
359 		cp++;
360 		switch (c) {
361 		case 't':
362 			if (sym.s_def != NODECL)
363 				inperr();
364 			sym.s_def = TDEF;
365 			break;
366 		case 'd':
367 			if (sym.s_def != NODECL)
368 				inperr();
369 			sym.s_def = DEF;
370 			break;
371 		case 'e':
372 			if (sym.s_def != NODECL)
373 				inperr();
374 			sym.s_def = DECL;
375 			break;
376 		case 'u':
377 			if (used)
378 				inperr();
379 			used = 1;
380 			break;
381 		case 'r':
382 			if (sym.s_rval)
383 				inperr();
384 			sym.s_rval = 1;
385 			break;
386 		case 'o':
387 			if (sym.s_osdef)
388 				inperr();
389 			sym.s_osdef = 1;
390 			break;
391 		case 's':
392 			if (sym.s_static)
393 				inperr();
394 			sym.s_static = 1;
395 			break;
396 		case 'v':
397 			if (sym.s_va)
398 				inperr();
399 			sym.s_va = 1;
400 			sym.s_nva = (short)strtol(cp, &ep, 10);
401 			if (cp == ep)
402 				inperr();
403 			cp = ep;
404 			break;
405 		case 'P':
406 			if (sym.s_prfl)
407 				inperr();
408 			sym.s_prfl = 1;
409 			sym.s_nprfl = (short)strtol(cp, &ep, 10);
410 			if (cp == ep)
411 				inperr();
412 			cp = ep;
413 			break;
414 		case 'S':
415 			if (sym.s_scfl)
416 				inperr();
417 			sym.s_scfl = 1;
418 			sym.s_nscfl = (short)strtol(cp, &ep, 10);
419 			if (cp == ep)
420 				inperr();
421 			cp = ep;
422 			break;
423 		}
424 	}
425 
426 	/* read symbol name, doing renaming if necessary */
427 	name = inpname(cp, &cp);
428 	renamed = 0;
429 	if (*cp == 'r') {
430 		cp++;
431 		name = xstrdup(name);
432 		rename = inpname(cp, &cp);
433 
434 		/* enter it and see if it's already been renamed */
435 		renamehte = _hsearch(renametab, name, 1);
436 		if (renamehte->h_hte == NULL) {
437 			hte = hsearch(rename, 1);
438 			renamehte->h_hte = hte;
439 			renamed = 1;
440 		} else if (strcmp((hte = renamehte->h_hte)->h_name, rename)) {
441 			pos1 = xstrdup(mkpos(&renamehte->h_syms->s_pos));
442 			/* %s renamed multiple times\t%s  ::  %s */
443 			msg(18, name, pos1, mkpos(&sym.s_pos));
444 			free(pos1);
445 		}
446 		free((char *)name);
447 	} else {
448 		/* it might be a previously-done rename */
449 		hte = _hsearch(renametab, name, 0);
450 		if (hte != NULL)
451 			hte = hte->h_hte;
452 		else
453 			hte = hsearch(name, 1);
454 	}
455 	hte->h_used |= used;
456 	if (sym.s_def == DEF || sym.s_def == TDEF)
457 		hte->h_def = 1;
458 
459 	sym.s_type = inptype(cp, &cp);
460 
461 	/*
462 	 * Allocate memory for this symbol only if it was not already
463 	 * declared or tentatively defined at the same location with
464 	 * the same type. Works only for symbols with external linkage,
465 	 * because static symbols, tentatively defined at the same location
466 	 * but in different translation units are really different symbols.
467 	 */
468 	for (symp = hte->h_syms; symp != NULL; symp = symp->s_nxt) {
469 		if (symp->s_pos.p_isrc == sym.s_pos.p_isrc &&
470 		    symp->s_pos.p_iline == sym.s_pos.p_iline &&
471 		    symp->s_type == sym.s_type &&
472 		    ((symp->s_def == DECL && sym.s_def == DECL) ||
473 		     (!sflag && symp->s_def == TDEF && sym.s_def == TDEF)) &&
474 		    !symp->s_static && !sym.s_static) {
475 			break;
476 		}
477 	}
478 
479 	if (symp == NULL) {
480 		/* allocsym reserviert keinen Platz fuer s_nva */
481 		if (sym.s_va || sym.s_prfl || sym.s_scfl) {
482 			symp = xalloc(sizeof (sym_t));
483 			STRUCT_ASSIGN(*symp, sym);
484 		} else {
485 			symp = xalloc(sizeof (symp->s_s));
486 			STRUCT_ASSIGN(symp->s_s, sym.s_s);
487 		}
488 		*hte->h_lsym = symp;
489 		hte->h_lsym = &symp->s_nxt;
490 
491 		/* XXX hack so we can remember where a symbol was renamed */
492 		if (renamed)
493 			renamehte->h_syms = symp;
494 	}
495 
496 	if (*cp != '\0')
497 		inperr();
498 }
499 
500 /*
501  * Read an u-record (emited by lint1 if a symbol was used).
502  */
503 static void
504 usedsym(pos_t *posp, const char *cp)
505 {
506 	usym_t	*usym;
507 	hte_t	*hte;
508 	const char *name;
509 
510 	usym = xalloc(sizeof (usym_t));
511 	STRUCT_ASSIGN(usym->u_pos, *posp);
512 
513 	/* needed as delimiter between two numbers */
514 	if (*cp++ != 'x')
515 		inperr();
516 
517 	name = inpname(cp, &cp);
518 	hte = _hsearch(renametab, name, 0);
519 	if (hte != NULL)
520 		hte = hte->h_hte;
521 	else
522 		hte = hsearch(name, 1);
523 	hte->h_used = 1;
524 
525 	*hte->h_lusym = usym;
526 	hte->h_lusym = &usym->u_nxt;
527 }
528 
529 /*
530  * Read a type and return the index of this type.
531  */
532 static u_short
533 inptype(const char *cp, const char **epp)
534 {
535 	char	c, s, *eptr;
536 	const	char *ep;
537 	type_t	*tp;
538 	int	narg, i, osdef = 0;
539 	size_t	tlen;
540 	u_short	tidx;
541 	int	h;
542 
543 	/* If we have this type already, return it's index. */
544 	tlen = gettlen(cp, &ep);
545 	h = thash(cp, tlen);
546 	if ((tidx = findtype(cp, tlen, h)) != 0) {
547 		*epp = ep;
548 		return (tidx);
549 	}
550 
551 	/* No, we must create a new type. */
552 	tp = xalloc(sizeof (type_t));
553 
554 	tidx = storetyp(tp, cp, tlen, h);
555 
556 	c = *cp++;
557 
558 	while (c == 'c' || c == 'v') {
559 		if (c == 'c') {
560 			tp->t_const = 1;
561 		} else {
562 			tp->t_volatile = 1;
563 		}
564 		c = *cp++;
565 	}
566 
567 	if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
568 		s = c;
569 		c = *cp++;
570 	} else {
571 		s = '\0';
572 	}
573 
574 	switch (c) {
575 	case 'C':
576 		tp->t_tspec = s == 's' ? SCHAR : (s == 'u' ? UCHAR : CHAR);
577 		break;
578 	case 'S':
579 		tp->t_tspec = s == 'u' ? USHORT : SHORT;
580 		break;
581 	case 'I':
582 		tp->t_tspec = s == 'u' ? UINT : INT;
583 		break;
584 	case 'L':
585 		tp->t_tspec = s == 'u' ? ULONG : LONG;
586 		break;
587 	case 'Q':
588 		tp->t_tspec = s == 'u' ? UQUAD : QUAD;
589 		break;
590 	case 'D':
591 		tp->t_tspec = s == 's' ? FLOAT : (s == 'l' ? LDOUBLE : DOUBLE);
592 		break;
593 	case 'V':
594 		tp->t_tspec = VOID;
595 		break;
596 	case 'P':
597 		tp->t_tspec = PTR;
598 		break;
599 	case 'A':
600 		tp->t_tspec = ARRAY;
601 		break;
602 	case 'F':
603 	case 'f':
604 		osdef = c == 'f';
605 		tp->t_tspec = FUNC;
606 		break;
607 	case 'T':
608 		tp->t_tspec = s == 'e' ? ENUM : (s == 's' ? STRUCT : UNION);
609 		break;
610 	}
611 
612 	switch (tp->t_tspec) {
613 	case ARRAY:
614 		tp->t_dim = (int)strtol(cp, &eptr, 10);
615 		cp = eptr;
616 		tp->t_subt = TP(inptype(cp, &cp));
617 		break;
618 	case PTR:
619 		tp->t_subt = TP(inptype(cp, &cp));
620 		break;
621 	case FUNC:
622 		c = *cp;
623 		if (isdigit((u_char)c)) {
624 			if (!osdef)
625 				tp->t_proto = 1;
626 			narg = (int)strtol(cp, &eptr, 10);
627 			cp = eptr;
628 			tp->t_args = xcalloc((size_t)(narg + 1),
629 					     sizeof (type_t *));
630 			for (i = 0; i < narg; i++) {
631 				if (i == narg - 1 && *cp == 'E') {
632 					tp->t_vararg = 1;
633 					cp++;
634 				} else {
635 					tp->t_args[i] = TP(inptype(cp, &cp));
636 				}
637 			}
638 		}
639 		tp->t_subt = TP(inptype(cp, &cp));
640 		break;
641 	case ENUM:
642 		tp->t_tspec = INT;
643 		tp->t_isenum = 1;
644 		/* FALLTHROUGH */
645 	case STRUCT:
646 	case UNION:
647 		switch (*cp++) {
648 		case '1':
649 			tp->t_istag = 1;
650 			tp->t_tag = hsearch(inpname(cp, &cp), 1);
651 			break;
652 		case '2':
653 			tp->t_istynam = 1;
654 			tp->t_tynam = hsearch(inpname(cp, &cp), 1);
655 			break;
656 		case '3':
657 			tp->t_isuniqpos = 1;
658 			tp->t_uniqpos.p_line = strtol(cp, &eptr, 10);
659 			cp = eptr;
660 			cp++;
661 			/* xlate to 'global' file name. */
662 			tp->t_uniqpos.p_file =
663 			    addoutfile(inpfns[strtol(cp, &eptr, 10)]);
664 			cp = eptr;
665 			cp++;
666 			tp->t_uniqpos.p_uniq = strtol(cp, &eptr, 10);
667 			cp = eptr;
668 			break;
669 		}
670 		break;
671 	case LONG:
672 	case VOID:
673 	case LDOUBLE:
674 	case DOUBLE:
675 	case FLOAT:
676 	case UQUAD:
677 	case QUAD:
678 	case ULONG:
679 	case UINT:
680 	case INT:
681 	case USHORT:
682 	case SHORT:
683 	case UCHAR:
684 	case SCHAR:
685 	case CHAR:
686 	case UNSIGN:
687 	case SIGNED:
688 	case NOTSPEC:
689 		break;
690 	case NTSPEC:
691 		abort();
692 	}
693 
694 	*epp = cp;
695 	return (tidx);
696 }
697 
698 /*
699  * Get the length of a type string.
700  */
701 static int
702 gettlen(const char *cp, const char **epp)
703 {
704 	const	char *cp1;
705 	char	c, s, *eptr;
706 	tspec_t	t;
707 	int	narg, i, cm, vm;
708 
709 	cp1 = cp;
710 
711 	c = *cp++;
712 
713 	cm = vm = 0;
714 
715 	while (c == 'c' || c == 'v') {
716 		if (c == 'c') {
717 			if (cm)
718 				inperr();
719 			cm = 1;
720 		} else {
721 			if (vm)
722 				inperr();
723 			vm = 1;
724 		}
725 		c = *cp++;
726 	}
727 
728 	if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
729 		s = c;
730 		c = *cp++;
731 	} else {
732 		s = '\0';
733 	}
734 
735 	t = NOTSPEC;
736 
737 	switch (c) {
738 	case 'C':
739 		if (s == 's') {
740 			t = SCHAR;
741 		} else if (s == 'u') {
742 			t = UCHAR;
743 		} else if (s == '\0') {
744 			t = CHAR;
745 		}
746 		break;
747 	case 'S':
748 		if (s == 'u') {
749 			t = USHORT;
750 		} else if (s == '\0') {
751 			t = SHORT;
752 		}
753 		break;
754 	case 'I':
755 		if (s == 'u') {
756 			t = UINT;
757 		} else if (s == '\0') {
758 			t = INT;
759 		}
760 		break;
761 	case 'L':
762 		if (s == 'u') {
763 			t = ULONG;
764 		} else if (s == '\0') {
765 			t = LONG;
766 		}
767 		break;
768 	case 'Q':
769 		if (s == 'u') {
770 			t = UQUAD;
771 		} else if (s == '\0') {
772 			t = QUAD;
773 		}
774 		break;
775 	case 'D':
776 		if (s == 's') {
777 			t = FLOAT;
778 		} else if (s == 'l') {
779 			t = LDOUBLE;
780 		} else if (s == '\0') {
781 			t = DOUBLE;
782 		}
783 		break;
784 	case 'V':
785 		if (s == '\0')
786 			t = VOID;
787 		break;
788 	case 'P':
789 		if (s == '\0')
790 			t = PTR;
791 		break;
792 	case 'A':
793 		if (s == '\0')
794 			t = ARRAY;
795 		break;
796 	case 'F':
797 	case 'f':
798 		if (s == '\0')
799 			t = FUNC;
800 		break;
801 	case 'T':
802 		if (s == 'e') {
803 			t = ENUM;
804 		} else if (s == 's') {
805 			t = STRUCT;
806 		} else if (s == 'u') {
807 			t = UNION;
808 		}
809 		break;
810 	default:
811 		inperr();
812 	}
813 
814 	if (t == NOTSPEC)
815 		inperr();
816 
817 	switch (t) {
818 	case ARRAY:
819 		(void)strtol(cp, &eptr, 10);
820 		if (cp == eptr)
821 			inperr();
822 		cp = eptr;
823 		(void)gettlen(cp, &cp);
824 		break;
825 	case PTR:
826 		(void)gettlen(cp, &cp);
827 		break;
828 	case FUNC:
829 		c = *cp;
830 		if (isdigit((u_char)c)) {
831 			narg = (int)strtol(cp, &eptr, 10);
832 			cp = eptr;
833 			for (i = 0; i < narg; i++) {
834 				if (i == narg - 1 && *cp == 'E') {
835 					cp++;
836 				} else {
837 					(void)gettlen(cp, &cp);
838 				}
839 			}
840 		}
841 		(void)gettlen(cp, &cp);
842 		break;
843 	case ENUM:
844 	case STRUCT:
845 	case UNION:
846 		switch (*cp++) {
847 		case '1':
848 			(void)inpname(cp, &cp);
849 			break;
850 		case '2':
851 			(void)inpname(cp, &cp);
852 			break;
853 		case '3':
854 			/* unique position: line.file.uniquifier */
855 			(void)strtol(cp, &eptr, 10);
856 			if (cp == eptr)
857 				inperr();
858 			cp = eptr;
859 			if (*cp++ != '.')
860 				inperr();
861 			(void)strtol(cp, &eptr, 10);
862 			if (cp == eptr)
863 				inperr();
864 			cp = eptr;
865 			if (*cp++ != '.')
866 				inperr();
867 			(void)strtol(cp, &eptr, 10);
868 			if (cp == eptr)
869 				inperr();
870 			cp = eptr;
871 			break;
872 		default:
873 			inperr();
874 		}
875 		break;
876 	case FLOAT:
877 	case USHORT:
878 	case SHORT:
879 	case UCHAR:
880 	case SCHAR:
881 	case CHAR:
882 	case UNSIGN:
883 	case SIGNED:
884 	case NOTSPEC:
885 	case INT:
886 	case UINT:
887 	case DOUBLE:
888 	case LDOUBLE:
889 	case VOID:
890 	case ULONG:
891 	case QUAD:
892 	case UQUAD:
893 	case LONG:
894 		break;
895 	case NTSPEC:
896 		abort();
897 	}
898 
899 	*epp = cp;
900 	return (cp - cp1);
901 }
902 
903 /*
904  * Search a type by it's type string.
905  */
906 static u_short
907 findtype(const char *cp, size_t len, int h)
908 {
909 	thtab_t	*thte;
910 
911 	for (thte = thtab[h]; thte != NULL; thte = thte->th_nxt) {
912 		if (strncmp(thte->th_name, cp, len) != 0)
913 			continue;
914 		if (thte->th_name[len] == '\0')
915 			return (thte->th_idx);
916 	}
917 
918 	return (0);
919 }
920 
921 /*
922  * Store a type and it's type string so we can later share this type
923  * if we read the same type string from the input file.
924  */
925 static u_short
926 storetyp(type_t *tp, const char *cp, size_t len, int h)
927 {
928 	static	u_int	tidx = 1;	/* 0 is reserved */
929 	thtab_t	*thte;
930 	char	*name;
931 
932 	if (tidx >= USHRT_MAX)
933 		errx(1, "sorry, too many types");
934 
935 	if (tidx == tlstlen - 1) {
936 		tlst = xrealloc(tlst, (tlstlen * 2) * sizeof (type_t *));
937 		(void)memset(tlst + tlstlen, 0, tlstlen * sizeof (type_t *));
938 		tlstlen *= 2;
939 	}
940 
941 	tlst[tidx] = tp;
942 
943 	/* create a hash table entry */
944 	name = xalloc(len + 1);
945 	(void)memcpy(name, cp, len);
946 	name[len] = '\0';
947 
948 	thte = xalloc(sizeof (thtab_t));
949 	thte->th_name = name;
950 	thte->th_idx = tidx;
951 	thte->th_nxt = thtab[h];
952 	thtab[h] = thte;
953 
954 	return ((u_short)tidx++);
955 }
956 
957 /*
958  * Hash function for types
959  */
960 static int
961 thash(const char *s, size_t len)
962 {
963 	u_int	v;
964 
965 	v = 0;
966 	while (len-- != 0) {
967 		v = (v << sizeof (v)) + (u_char)*s++;
968 		v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
969 	}
970 	return (v % THSHSIZ2);
971 }
972 
973 /*
974  * Read a string enclosed by "". This string may contain quoted chars.
975  */
976 static char *
977 inpqstrg(const char *src, const char **epp)
978 {
979 	char	*strg, *dst;
980 	size_t	slen;
981 	int	c;
982 	int	v;
983 
984 	dst = strg = xmalloc(slen = 32);
985 
986 	if ((c = *src++) != '"')
987 		inperr();
988 	if ((c = *src++) == '\0')
989 		inperr();
990 
991 	while (c != '"') {
992 		if (c == '\\') {
993 			if ((c = *src++) == '\0')
994 				inperr();
995 			switch (c) {
996 			case 'n':
997 				c = '\n';
998 				break;
999 			case 't':
1000 				c = '\t';
1001 				break;
1002 			case 'v':
1003 				c = '\v';
1004 				break;
1005 			case 'b':
1006 				c = '\b';
1007 				break;
1008 			case 'r':
1009 				c = '\r';
1010 				break;
1011 			case 'f':
1012 				c = '\f';
1013 				break;
1014 			case 'a':
1015 				c = '\a';
1016 				break;
1017 			case '\\':
1018 				c = '\\';
1019 				break;
1020 			case '"':
1021 				c = '"';
1022 				break;
1023 			case '\'':
1024 				c = '\'';
1025 				break;
1026 			case '0': case '1': case '2': case '3':
1027 				v = (c - '0') << 6;
1028 				if ((c = *src++) < '0' || c > '7')
1029 					inperr();
1030 				v |= (c - '0') << 3;
1031 				if ((c = *src++) < '0' || c > '7')
1032 					inperr();
1033 				v |= c - '0';
1034 				c = (u_char)v;
1035 				break;
1036 			default:
1037 				inperr();
1038 			}
1039 		}
1040 		/* keep space for trailing '\0' */
1041 		if (dst - strg == slen - 1) {
1042 			strg = xrealloc(strg, slen * 2);
1043 			dst = strg + (slen - 1);
1044 			slen *= 2;
1045 		}
1046 		*dst++ = (char)c;
1047 		if ((c = *src++) == '\0')
1048 			inperr();
1049 	}
1050 	*dst = '\0';
1051 
1052 	*epp = src;
1053 	return (strg);
1054 }
1055 
1056 /*
1057  * Read the name of a symbol in static memory.
1058  */
1059 static const char *
1060 inpname(const char *cp, const char **epp)
1061 {
1062 	static	char	*buf;
1063 	static	size_t	blen = 0;
1064 	size_t	len, i;
1065 	char	*eptr, c;
1066 
1067 	len = (int)strtol(cp, &eptr, 10);
1068 	if (cp == eptr)
1069 		inperr();
1070 	cp = eptr;
1071 	if (len + 1 > blen)
1072 		buf = xrealloc(buf, blen = len + 1);
1073 	for (i = 0; i < len; i++) {
1074 		c = *cp++;
1075 		if (!isalnum((unsigned char)c) && c != '_')
1076 			inperr();
1077 		buf[i] = c;
1078 	}
1079 	buf[i] = '\0';
1080 
1081 	*epp = cp;
1082 	return (buf);
1083 }
1084 
1085 /*
1086  * Return the index of a file name. If the name cannot be found, create
1087  * a new entry and return the index of the newly created entry.
1088  */
1089 static int
1090 getfnidx(const char *fn)
1091 {
1092 	int	i;
1093 
1094 	/* 0 ist reserved */
1095 	for (i = 1; fnames[i] != NULL; i++) {
1096 		if (strcmp(fnames[i], fn) == 0)
1097 			break;
1098 	}
1099 	if (fnames[i] != NULL)
1100 		return (i);
1101 
1102 	if (i == nfnames - 1) {
1103 		fnames = xrealloc(fnames, (nfnames * 2) * sizeof (char *));
1104 		(void)memset(fnames + nfnames, 0, nfnames * sizeof (char *));
1105 		nfnames *= 2;
1106 	}
1107 
1108 	fnames[i] = xstrdup(fn);
1109 	return (i);
1110 }
1111 
1112 /*
1113  * Separate symbols with static and external linkage.
1114  */
1115 void
1116 mkstatic(hte_t *hte)
1117 {
1118 	sym_t	*sym1, **symp, *sym;
1119 	fcall_t	**callp, *call;
1120 	usym_t	**usymp, *usym;
1121 	hte_t	*nhte;
1122 	int	ofnd;
1123 
1124 	/* Look for first static definition */
1125 	for (sym1 = hte->h_syms; sym1 != NULL; sym1 = sym1->s_nxt) {
1126 		if (sym1->s_static)
1127 			break;
1128 	}
1129 	if (sym1 == NULL)
1130 		return;
1131 
1132 	/* Do nothing if this name is used only in one translation unit. */
1133 	ofnd = 0;
1134 	for (sym = hte->h_syms; sym != NULL && !ofnd; sym = sym->s_nxt) {
1135 		if (sym->s_pos.p_src != sym1->s_pos.p_src)
1136 			ofnd = 1;
1137 	}
1138 	for (call = hte->h_calls; call != NULL && !ofnd; call = call->f_nxt) {
1139 		if (call->f_pos.p_src != sym1->s_pos.p_src)
1140 			ofnd = 1;
1141 	}
1142 	for (usym = hte->h_usyms; usym != NULL && !ofnd; usym = usym->u_nxt) {
1143 		if (usym->u_pos.p_src != sym1->s_pos.p_src)
1144 			ofnd = 1;
1145 	}
1146 	if (!ofnd) {
1147 		hte->h_used = 1;
1148 		/* errors about undef. static symbols are printed in lint1 */
1149 		hte->h_def = 1;
1150 		hte->h_static = 1;
1151 		return;
1152 	}
1153 
1154 	/*
1155 	 * Create a new hash table entry
1156 	 *
1157 	 * XXX this entry should be put at the beginning of the list to
1158 	 * avoid to process the same symbol twice.
1159 	 */
1160 	for (nhte = hte; nhte->h_link != NULL; nhte = nhte->h_link)
1161 		continue;
1162 	nhte->h_link = xmalloc(sizeof (hte_t));
1163 	nhte = nhte->h_link;
1164 	nhte->h_name = hte->h_name;
1165 	nhte->h_used = 1;
1166 	nhte->h_def = 1;	/* error in lint1 */
1167 	nhte->h_static = 1;
1168 	nhte->h_syms = NULL;
1169 	nhte->h_lsym = &nhte->h_syms;
1170 	nhte->h_calls = NULL;
1171 	nhte->h_lcall = &nhte->h_calls;
1172 	nhte->h_usyms = NULL;
1173 	nhte->h_lusym = &nhte->h_usyms;
1174 	nhte->h_link = NULL;
1175 	nhte->h_hte = NULL;
1176 
1177 	/*
1178 	 * move all symbols used in this translation unit into the new
1179 	 * hash table entry.
1180 	 */
1181 	for (symp = &hte->h_syms; (sym = *symp) != NULL; ) {
1182 		if (sym->s_pos.p_src == sym1->s_pos.p_src) {
1183 			sym->s_static = 1;
1184 			(*symp) = sym->s_nxt;
1185 			if (hte->h_lsym == &sym->s_nxt)
1186 				hte->h_lsym = symp;
1187 			sym->s_nxt = NULL;
1188 			*nhte->h_lsym = sym;
1189 			nhte->h_lsym = &sym->s_nxt;
1190 		} else {
1191 			symp = &sym->s_nxt;
1192 		}
1193 	}
1194 	for (callp = &hte->h_calls; (call = *callp) != NULL; ) {
1195 		if (call->f_pos.p_src == sym1->s_pos.p_src) {
1196 			(*callp) = call->f_nxt;
1197 			if (hte->h_lcall == &call->f_nxt)
1198 				hte->h_lcall = callp;
1199 			call->f_nxt = NULL;
1200 			*nhte->h_lcall = call;
1201 			nhte->h_lcall = &call->f_nxt;
1202 		} else {
1203 			callp = &call->f_nxt;
1204 		}
1205 	}
1206 	for (usymp = &hte->h_usyms; (usym = *usymp) != NULL; ) {
1207 		if (usym->u_pos.p_src == sym1->s_pos.p_src) {
1208 			(*usymp) = usym->u_nxt;
1209 			if (hte->h_lusym == &usym->u_nxt)
1210 				hte->h_lusym = usymp;
1211 			usym->u_nxt = NULL;
1212 			*nhte->h_lusym = usym;
1213 			nhte->h_lusym = &usym->u_nxt;
1214 		} else {
1215 			usymp = &usym->u_nxt;
1216 		}
1217 	}
1218 
1219 	/* h_def must be recalculated for old hte */
1220 	hte->h_def = nhte->h_def = 0;
1221 	for (sym = hte->h_syms; sym != NULL; sym = sym->s_nxt) {
1222 		if (sym->s_def == DEF || sym->s_def == TDEF) {
1223 			hte->h_def = 1;
1224 			break;
1225 		}
1226 	}
1227 
1228 	mkstatic(hte);
1229 }
1230