1/* $NetBSD: read.c,v 1.19 2007/09/28 21:53:50 uwe 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 1.19 2007/09/28 21:53:50 uwe Exp $");
38#endif
39__FBSDID("$FreeBSD: stable/10/usr.bin/xlint/lint2/read.c 281168 2015-04-06 19:56:27Z pfg $");
40
41#include <ctype.h>
42#include <err.h>
43#include <limits.h>
44#include <stdio.h>
45#include <stdlib.h>
46#include <string.h>
47
48#include "lint2.h"
49
50
51/* index of current (included) source file */
52static int srcfile;
53
54/*
55 * The array pointed to by inpfns maps the file name indices of input files
56 * to the file name indices used in lint2
57 */
58static short *inpfns;
59static size_t ninpfns;
60
61/*
62 * The array pointed to by *fnames maps file name indizes to file names.
63 * Indices of type short are used instead of pointers to save memory.
64 */
65const char **fnames;
66static size_t nfnames;
67
68/*
69 * Types are shared (to save memory for the types itself) and accessed
70 * via indices (to save memory for references to types (indices are short)).
71 * To share types, an equal type must be located fast. This is done by a
72 * hash table. Access by indices is done via an array of pointers to the
73 * types.
74 */
75typedef struct thtab {
76 const char *th_name;
77 u_short th_idx;
78 struct thtab *th_nxt;
79} thtab_t;
80static thtab_t **thtab; /* hash table */
81type_t **tlst; /* array for indexed access */
82static size_t tlstlen; /* length of tlst */
83
84static hte_t **renametab;
85
86/* index of current C source file (as spezified at the command line) */
87static int csrcfile;
88
89
90#define inperr() inperror(__FILE__, __LINE__)
91static void inperror(const char *, size_t);
92static void setsrc(const char *);
93static void setfnid(int, const char *);
94static void funccall(pos_t *, const char *);
95static void decldef(pos_t *, const char *);
96static void usedsym(pos_t *, const char *);
97static u_short inptype(const char *, const char **);
98static int gettlen(const char *, const char **);
99static u_short findtype(const char *, size_t, int);
100static u_short storetyp(type_t *, const char *, size_t, int);
101static int thash(const char *, size_t);
102static char *inpqstrg(const char *, const char **);
103static const char *inpname(const char *, const char **);
104static int getfnidx(const char *);
105
106void
107readfile(const char *name)
108{
109 FILE *inp;
110 size_t len;
111 const char *cp;
112 char *line, *eptr, rt = '\0';
113 int cline, isrc, iline;
114 pos_t pos;
115
116 if (inpfns == NULL)
117 if ((inpfns = calloc(ninpfns = 128, sizeof (short))) == NULL)
118 nomem();
119 if (fnames == NULL)
120 if ((fnames = calloc(nfnames = 256, sizeof (char *))) == NULL)
121 nomem();
122 if (tlstlen == 0)
123 if ((tlst = calloc(tlstlen = 256, sizeof (type_t *))) == NULL)
124 nomem();
125 if (thtab == NULL)
126 if ((thtab = calloc(THSHSIZ2, sizeof (thtab_t))) == NULL)
127 nomem();
128
129 _inithash(&renametab);
130
131 srcfile = getfnidx(name);
132
133 if ((inp = fopen(name, "r")) == NULL)
134 err(1, "cannot open %s", name);
135
136 while ((line = fgetln(inp, &len)) != NULL) {
137
138 if (len == 0 || line[len - 1] != '\n')
139 inperr();
140 line[len - 1] = '\0';
141 cp = line;
142
143 /* line number in csrcfile */
144 cline = (int)strtol(cp, &eptr, 10);
145 if (cp == eptr) {
146 cline = -1;
147 } else {
148 cp = eptr;
149 }
150
151 /* record type */
152 if (*cp != '\0') {
153 rt = *cp++;
154 } else {
155 inperr();
156 }
157
158 if (rt == 'S') {
159 setsrc(cp);
160 continue;
161 } else if (rt == 's') {
162 setfnid(cline, cp);
163 continue;
164 }
165
166 /*
167 * Index of (included) source file. If this index is
168 * different from csrcfile, it refers to an included
169 * file.
170 */
171 isrc = (int)strtol(cp, &eptr, 10);
172 if (cp == eptr)
173 inperr();
174 cp = eptr;
175 isrc = inpfns[isrc];
176
177 /* line number in isrc */
178 if (*cp++ != '.')
179 inperr();
180 iline = (int)strtol(cp, &eptr, 10);
181 if (cp == eptr)
182 inperr();
183 cp = eptr;
184
185 pos.p_src = (u_short)csrcfile;
186 pos.p_line = (u_short)cline;
187 pos.p_isrc = (u_short)isrc;
188 pos.p_iline = (u_short)iline;
189
190 /* process rest of this record */
191 switch (rt) {
192 case 'c':
193 funccall(&pos, cp);
194 break;
195 case 'd':
196 decldef(&pos, cp);
197 break;
198 case 'u':
199 usedsym(&pos, cp);
200 break;
201 default:
202 inperr();
203 }
204
205 }
206
207 _destroyhash(renametab);
208
209 if (ferror(inp))
210 err(1, "read error on %s", name);
211
212 (void)fclose(inp);
213}
214
215
216static void
217inperror(const char *file, size_t line)
218{
219
220 errx(1, "%s,%zd: input file error: %s", file, line, fnames[srcfile]);
221}
222
223/*
224 * Set the name of the C source file of the .ln file which is
225 * currently read.
226 */
227static void
228setsrc(const char *cp)
229{
230
231 csrcfile = getfnidx(cp);
232}
233
234/*
235 * setfnid() gets as input an index as used in an input file and the
236 * associated file name. If necessary, it creates a new lint2 file
237 * name index for this file name and creates the mapping of the index
238 * as used in the input file to the index used in lint2.
239 */
240static void
241setfnid(int fid, const char *cp)
242{
243
244 if (fid == -1)
245 inperr();
246
247 if (fid >= ninpfns) {
248 if ((inpfns = realloc(inpfns, (ninpfns * 2) * sizeof (short)))
249 == NULL)
250 nomem();
251 (void)memset(inpfns + ninpfns, 0, ninpfns * sizeof (short));
252 ninpfns *= 2;
253 }
254 /*
255 * Should always be true because indices written in the output
256 * file by lint1 are always the previous index + 1.
257 */
258 if (fid >= ninpfns)
259 errx(1, "internal error: setfnid()");
260 inpfns[fid] = (u_short)getfnidx(cp);
261}
262
263/*
264 * Process a function call record (c-record).
265 */
266static void
267funccall(pos_t *posp, const char *cp)
268{
269 arginf_t *ai, **lai;
270 char c, *eptr;
271 int rused, rdisc;
272 hte_t *hte;
273 fcall_t *fcall;
274 const char *name;
275
276 fcall = xalloc(sizeof (fcall_t));
277 STRUCT_ASSIGN(fcall->f_pos, *posp);
278
279 /* read flags */
280 rused = rdisc = 0;
281 lai = &fcall->f_args;
282 while ((c = *cp) == 'u' || c == 'i' || c == 'd' ||
283 c == 'z' || c == 'p' || c == 'n' || c == 's') {
284 cp++;
285 switch (c) {
286 case 'u':
287 if (rused || rdisc)
288 inperr();
289 rused = 1;
290 break;
291 case 'i':
292 if (rused || rdisc)
293 inperr();
294 break;
295 case 'd':
296 if (rused || rdisc)
297 inperr();
298 rdisc = 1;
299 break;
300 case 'z':
301 case 'p':
302 case 'n':
303 case 's':
304 ai = xalloc(sizeof (arginf_t));
305 ai->a_num = (int)strtol(cp, &eptr, 10);
306 if (cp == eptr)
307 inperr();
308 cp = eptr;
309 if (c == 'z') {
310 ai->a_pcon = ai->a_zero = 1;
311 } else if (c == 'p') {
312 ai->a_pcon = 1;
313 } else if (c == 'n') {
314 ai->a_ncon = 1;
315 } else {
316 ai->a_fmt = 1;
317 ai->a_fstrg = inpqstrg(cp, &cp);
318 }
319 *lai = ai;
320 lai = &ai->a_nxt;
321 break;
322 }
323 }
324 fcall->f_rused = rused;
325 fcall->f_rdisc = rdisc;
326
327 /* read name of function */
328 name = inpname(cp, &cp);
329
330 /* first look it up in the renaming table, then in the normal table */
331 hte = _hsearch(renametab, name, 0);
332 if (hte != NULL)
333 hte = hte->h_hte;
334 else
335 hte = hsearch(name, 1);
336 hte->h_used = 1;
337
338 fcall->f_type = inptype(cp, &cp);
339
340 *hte->h_lcall = fcall;
341 hte->h_lcall = &fcall->f_nxt;
342
343 if (*cp != '\0')
344 inperr();
345}
346
347/*
348 * Process a declaration or definition (d-record).
349 */
350static void
351decldef(pos_t *posp, const char *cp)
352{
353 sym_t *symp, sym;
354 char c, *ep, *pos1;
355 int used, renamed;
356 hte_t *hte, *renamehte = NULL;
357 const char *name, *rename;
358
359 (void)memset(&sym, 0, sizeof (sym));
360 STRUCT_ASSIGN(sym.s_pos, *posp);
361 sym.s_def = NODECL;
362
363 used = 0;
364
365 while (strchr("tdeurosvPS", (c = *cp)) != NULL) {
366 cp++;
367 switch (c) {
368 case 't':
369 if (sym.s_def != NODECL)
370 inperr();
371 sym.s_def = TDEF;
372 break;
373 case 'd':
374 if (sym.s_def != NODECL)
375 inperr();
376 sym.s_def = DEF;
377 break;
378 case 'e':
379 if (sym.s_def != NODECL)
380 inperr();
381 sym.s_def = DECL;
382 break;
383 case 'u':
384 if (used)
385 inperr();
386 used = 1;
387 break;
388 case 'r':
389 if (sym.s_rval)
390 inperr();
391 sym.s_rval = 1;
392 break;
393 case 'o':
394 if (sym.s_osdef)
395 inperr();
396 sym.s_osdef = 1;
397 break;
398 case 's':
399 if (sym.s_static)
400 inperr();
401 sym.s_static = 1;
402 break;
403 case 'v':
404 if (sym.s_va)
405 inperr();
406 sym.s_va = 1;
407 sym.s_nva = (short)strtol(cp, &ep, 10);
408 if (cp == ep)
409 inperr();
410 cp = ep;
411 break;
412 case 'P':
413 if (sym.s_prfl)
414 inperr();
415 sym.s_prfl = 1;
416 sym.s_nprfl = (short)strtol(cp, &ep, 10);
417 if (cp == ep)
418 inperr();
419 cp = ep;
420 break;
421 case 'S':
422 if (sym.s_scfl)
423 inperr();
424 sym.s_scfl = 1;
425 sym.s_nscfl = (short)strtol(cp, &ep, 10);
426 if (cp == ep)
427 inperr();
428 cp = ep;
429 break;
430 }
431 }
432
433 /* read symbol name, doing renaming if necessary */
434 name = inpname(cp, &cp);
435 renamed = 0;
436 if (*cp == 'r') {
437 cp++;
438 name = xstrdup(name);
439 rename = inpname(cp, &cp);
440
441 /* enter it and see if it's already been renamed */
442 renamehte = _hsearch(renametab, name, 1);
443 if (renamehte->h_hte == NULL) {
444 hte = hsearch(rename, 1);
445 renamehte->h_hte = hte;
446 renamed = 1;
447 } else if (strcmp((hte = renamehte->h_hte)->h_name, rename)) {
448 pos1 = xstrdup(mkpos(&renamehte->h_syms->s_pos));
449 /* %s renamed multiple times\t%s :: %s */
450 msg(18, name, pos1, mkpos(&sym.s_pos));
451 free(pos1);
452 }
453 free((char *)name);
454 } else {
455 /* it might be a previously-done rename */
456 hte = _hsearch(renametab, name, 0);
457 if (hte != NULL)
458 hte = hte->h_hte;
459 else
460 hte = hsearch(name, 1);
461 }
462 hte->h_used |= used;
463 if (sym.s_def == DEF || sym.s_def == TDEF)
464 hte->h_def = 1;
465
466 sym.s_type = inptype(cp, &cp);
467
468 /*
469 * Allocate memory for this symbol only if it was not already
470 * declared or tentatively defined at the same location with
471 * the same type. Works only for symbols with external linkage,
472 * because static symbols, tentatively defined at the same location
473 * but in different translation units are really different symbols.
474 */
475 for (symp = hte->h_syms; symp != NULL; symp = symp->s_nxt) {
476 if (symp->s_pos.p_isrc == sym.s_pos.p_isrc &&
477 symp->s_pos.p_iline == sym.s_pos.p_iline &&
478 symp->s_type == sym.s_type &&
479 ((symp->s_def == DECL && sym.s_def == DECL) ||
480 (!sflag && symp->s_def == TDEF && sym.s_def == TDEF)) &&
481 !symp->s_static && !sym.s_static) {
482 break;
483 }
484 }
485
486 if (symp == NULL) {
487 /* allocsym reserviert keinen Platz fuer s_nva */
488 if (sym.s_va || sym.s_prfl || sym.s_scfl) {
489 symp = xalloc(sizeof (sym_t));
490 STRUCT_ASSIGN(*symp, sym);
491 } else {
492 symp = xalloc(sizeof (symp->s_s));
493 STRUCT_ASSIGN(symp->s_s, sym.s_s);
494 }
495 *hte->h_lsym = symp;
496 hte->h_lsym = &symp->s_nxt;
497
498 /* XXX hack so we can remember where a symbol was renamed */
499 if (renamed)
500 renamehte->h_syms = symp;
501 }
502
503 if (*cp != '\0')
504 inperr();
505}
506
507/*
508 * Read an u-record (emitted by lint1 if a symbol was used).
509 */
510static void
511usedsym(pos_t *posp, const char *cp)
512{
513 usym_t *usym;
514 hte_t *hte;
515 const char *name;
516
517 usym = xalloc(sizeof (usym_t));
518 STRUCT_ASSIGN(usym->u_pos, *posp);
519
520 /* needed as delimiter between two numbers */
521 if (*cp++ != 'x')
522 inperr();
523
524 name = inpname(cp, &cp);
525 hte = _hsearch(renametab, name, 0);
526 if (hte != NULL)
527 hte = hte->h_hte;
528 else
529 hte = hsearch(name, 1);
530 hte->h_used = 1;
531
532 *hte->h_lusym = usym;
533 hte->h_lusym = &usym->u_nxt;
534}
535
536/*
537 * Read a type and return the index of this type.
538 */
539static u_short
540inptype(const char *cp, const char **epp)
541{
542 char c, s, *eptr;
543 const char *ep;
544 type_t *tp;
545 int narg, i, osdef = 0;
546 size_t tlen;
547 u_short tidx, sidx;
548 int h;
549
550 /* If we have this type already, return it's index. */
551 tlen = gettlen(cp, &ep);
552 h = thash(cp, tlen);
553 if ((tidx = findtype(cp, tlen, h)) != 0) {
554 *epp = ep;
555 return (tidx);
556 }
557
558 /* No, we must create a new type. */
559 tp = xalloc(sizeof (type_t));
560
561 tidx = storetyp(tp, cp, tlen, h);
562
563 c = *cp++;
564
565 while (c == 'c' || c == 'v') {
566 if (c == 'c') {
567 tp->t_const = 1;
568 } else {
569 tp->t_volatile = 1;
570 }
571 c = *cp++;
572 }
573
574 if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
575 s = c;
576 c = *cp++;
577 } else {
578 s = '\0';
579 }
580
581 switch (c) {
582 case 'C':
583 tp->t_tspec = s == 's' ? SCHAR : (s == 'u' ? UCHAR : CHAR);
584 break;
585 case 'S':
586 tp->t_tspec = s == 'u' ? USHORT : SHORT;
587 break;
588 case 'I':
589 tp->t_tspec = s == 'u' ? UINT : INT;
590 break;
591 case 'L':
592 tp->t_tspec = s == 'u' ? ULONG : LONG;
593 break;
594 case 'Q':
595 tp->t_tspec = s == 'u' ? UQUAD : QUAD;
596 break;
597 case 'D':
598 tp->t_tspec = s == 's' ? FLOAT : (s == 'l' ? LDOUBLE : DOUBLE);
599 break;
600 case 'V':
601 tp->t_tspec = VOID;
602 break;
603 case 'P':
604 tp->t_tspec = PTR;
605 break;
606 case 'A':
607 tp->t_tspec = ARRAY;
608 break;
609 case 'F':
610 case 'f':
611 osdef = c == 'f';
612 tp->t_tspec = FUNC;
613 break;
614 case 'T':
615 tp->t_tspec = s == 'e' ? ENUM : (s == 's' ? STRUCT : UNION);
616 break;
617 }
618
619 switch (tp->t_tspec) {
620 case ARRAY:
621 tp->t_dim = (int)strtol(cp, &eptr, 10);
622 cp = eptr;
623 sidx = inptype(cp, &cp); /* force seq. point! (ditto below) */
624 tp->t_subt = TP(sidx);
625 break;
626 case PTR:
627 sidx = inptype(cp, &cp);
628 tp->t_subt = TP(sidx);
629 break;
630 case FUNC:
631 c = *cp;
632 if (isdigit((u_char)c)) {
633 if (!osdef)
634 tp->t_proto = 1;
635 narg = (int)strtol(cp, &eptr, 10);
636 cp = eptr;
637 if ((tp->t_args = calloc((size_t)(narg + 1),
638 sizeof (type_t *))) == NULL)
639 nomem();
640 for (i = 0; i < narg; i++) {
641 if (i == narg - 1 && *cp == 'E') {
642 tp->t_vararg = 1;
643 cp++;
644 } else {
645 sidx = inptype(cp, &cp);
646 tp->t_args[i] = TP(sidx);
647 }
648 }
649 }
650 sidx = inptype(cp, &cp);
651 tp->t_subt = TP(sidx);
652 break;
653 case ENUM:
654 tp->t_tspec = INT;
655 tp->t_isenum = 1;
656 /* FALLTHROUGH */
657 case STRUCT:
658 case UNION:
659 switch (*cp++) {
660 case '1':
661 tp->t_istag = 1;
662 tp->t_tag = hsearch(inpname(cp, &cp), 1);
663 break;
664 case '2':
665 tp->t_istynam = 1;
666 tp->t_tynam = hsearch(inpname(cp, &cp), 1);
667 break;
668 case '3':
669 tp->t_isuniqpos = 1;
670 tp->t_uniqpos.p_line = strtol(cp, &eptr, 10);
671 cp = eptr;
672 cp++;
673 /* xlate to 'global' file name. */
674 tp->t_uniqpos.p_file =
675 addoutfile(inpfns[strtol(cp, &eptr, 10)]);
676 cp = eptr;
677 cp++;
678 tp->t_uniqpos.p_uniq = strtol(cp, &eptr, 10);
679 cp = eptr;
680 break;
681 }
682 break;
683 case LONG:
684 case VOID:
685 case LDOUBLE:
686 case DOUBLE:
687 case FLOAT:
688 case UQUAD:
689 case QUAD:
690 case ULONG:
691 case UINT:
692 case INT:
693 case USHORT:
694 case SHORT:
695 case UCHAR:
696 case SCHAR:
697 case CHAR:
698 case UNSIGN:
699 case SIGNED:
700 case NOTSPEC:
701 break;
702 }
703
704 *epp = cp;
705 return (tidx);
706}
707
708/*
709 * Get the length of a type string.
710 */
711static int
712gettlen(const char *cp, const char **epp)
713{
714 const char *cp1;
715 char c, s, *eptr;
716 tspec_t t;
717 int narg, i, cm, vm;
718
719 cp1 = cp;
720
721 c = *cp++;
722
723 cm = vm = 0;
724
725 while (c == 'c' || c == 'v') {
726 if (c == 'c') {
727 if (cm)
728 inperr();
729 cm = 1;
730 } else {
731 if (vm)
732 inperr();
733 vm = 1;
734 }
735 c = *cp++;
736 }
737
738 if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
739 s = c;
740 c = *cp++;
741 } else {
742 s = '\0';
743 }
744
745 t = NOTSPEC;
746
747 switch (c) {
748 case 'C':
749 if (s == 's') {
750 t = SCHAR;
751 } else if (s == 'u') {
752 t = UCHAR;
753 } else if (s == '\0') {
754 t = CHAR;
755 }
756 break;
757 case 'S':
758 if (s == 'u') {
759 t = USHORT;
760 } else if (s == '\0') {
761 t = SHORT;
762 }
763 break;
764 case 'I':
765 if (s == 'u') {
766 t = UINT;
767 } else if (s == '\0') {
768 t = INT;
769 }
770 break;
771 case 'L':
772 if (s == 'u') {
773 t = ULONG;
774 } else if (s == '\0') {
775 t = LONG;
776 }
777 break;
778 case 'Q':
779 if (s == 'u') {
780 t = UQUAD;
781 } else if (s == '\0') {
782 t = QUAD;
783 }
784 break;
785 case 'D':
786 if (s == 's') {
787 t = FLOAT;
788 } else if (s == 'l') {
789 t = LDOUBLE;
790 } else if (s == '\0') {
791 t = DOUBLE;
792 }
793 break;
794 case 'V':
795 if (s == '\0')
796 t = VOID;
797 break;
798 case 'P':
799 if (s == '\0')
800 t = PTR;
801 break;
802 case 'A':
803 if (s == '\0')
804 t = ARRAY;
805 break;
806 case 'F':
807 case 'f':
808 if (s == '\0')
809 t = FUNC;
810 break;
811 case 'T':
812 if (s == 'e') {
813 t = ENUM;
814 } else if (s == 's') {
815 t = STRUCT;
816 } else if (s == 'u') {
817 t = UNION;
818 }
819 break;
820 default:
821 inperr();
822 }
823
824 if (t == NOTSPEC)
825 inperr();
826
827 switch (t) {
828 case ARRAY:
829 (void)strtol(cp, &eptr, 10);
830 if (cp == eptr)
831 inperr();
832 cp = eptr;
833 (void)gettlen(cp, &cp);
834 break;
835 case PTR:
836 (void)gettlen(cp, &cp);
837 break;
838 case FUNC:
839 c = *cp;
840 if (isdigit((u_char)c)) {
841 narg = (int)strtol(cp, &eptr, 10);
842 cp = eptr;
843 for (i = 0; i < narg; i++) {
844 if (i == narg - 1 && *cp == 'E') {
845 cp++;
846 } else {
847 (void)gettlen(cp, &cp);
848 }
849 }
850 }
851 (void)gettlen(cp, &cp);
852 break;
853 case ENUM:
854 case STRUCT:
855 case UNION:
856 switch (*cp++) {
857 case '1':
858 (void)inpname(cp, &cp);
859 break;
860 case '2':
861 (void)inpname(cp, &cp);
862 break;
863 case '3':
864 /* unique position: line.file.uniquifier */
865 (void)strtol(cp, &eptr, 10);
866 if (cp == eptr)
867 inperr();
868 cp = eptr;
869 if (*cp++ != '.')
870 inperr();
871 (void)strtol(cp, &eptr, 10);
872 if (cp == eptr)
873 inperr();
874 cp = eptr;
875 if (*cp++ != '.')
876 inperr();
877 (void)strtol(cp, &eptr, 10);
878 if (cp == eptr)
879 inperr();
880 cp = eptr;
881 break;
882 default:
883 inperr();
884 }
885 break;
886 case FLOAT:
887 case USHORT:
888 case SHORT:
889 case UCHAR:
890 case SCHAR:
891 case CHAR:
892 case UNSIGN:
893 case SIGNED:
894 case NOTSPEC:
895 case INT:
896 case UINT:
897 case DOUBLE:
898 case LDOUBLE:
899 case VOID:
900 case ULONG:
901 case QUAD:
902 case UQUAD:
903 case LONG:
904 break;
905 }
906
907 *epp = cp;
908 return (cp - cp1);
909}
910
911/*
912 * Search a type by it's type string.
913 */
914static u_short
915findtype(const char *cp, size_t len, int h)
916{
917 thtab_t *thte;
918
919 for (thte = thtab[h]; thte != NULL; thte = thte->th_nxt) {
920 if (strncmp(thte->th_name, cp, len) != 0)
921 continue;
922 if (thte->th_name[len] == '\0')
923 return (thte->th_idx);
924 }
925
926 return (0);
927}
928
929/*
930 * Store a type and it's type string so we can later share this type
931 * if we read the same type string from the input file.
932 */
933static u_short
934storetyp(type_t *tp, const char *cp, size_t len, int h)
935{
936 static u_int tidx = 1; /* 0 is reserved */
937 thtab_t *thte;
938 char *name;
939
940 if (tidx >= USHRT_MAX)
941 errx(1, "sorry, too many types");
942
943 if (tidx == tlstlen - 1) {
944 if ((tlst = realloc(tlst, (tlstlen * 2) * sizeof (type_t *)))
945 == NULL)
946 nomem();
947 (void)memset(tlst + tlstlen, 0, tlstlen * sizeof (type_t *));
948 tlstlen *= 2;
949 }
950
951 tlst[tidx] = tp;
952
953 /* create a hash table entry */
954 name = xalloc(len + 1);
955 (void)memcpy(name, cp, len);
956 name[len] = '\0';
957
958 thte = xalloc(sizeof (thtab_t));
959 thte->th_name = name;
960 thte->th_idx = tidx;
961 thte->th_nxt = thtab[h];
962 thtab[h] = thte;
963
964 return ((u_short)tidx++);
965}
966
967/*
968 * Hash function for types
969 */
970static int
971thash(const char *s, size_t len)
972{
973 u_int v;
974
975 v = 0;
976 while (len-- != 0) {
977 v = (v << sizeof (v)) + (u_char)*s++;
978 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
979 }
980 return (v % THSHSIZ2);
981}
982
983/*
984 * Read a string enclosed by "". This string may contain quoted chars.
985 */
986static char *
987inpqstrg(const char *src, const char **epp)
988{
989 char *strg, *dst;
990 size_t slen;
991 int c;
992 int v;
993
994 if ((dst = strg = malloc(slen = 32)) == NULL)
995 nomem();
996
997 if ((c = *src++) != '"')
998 inperr();
999 if ((c = *src++) == '\0')
1000 inperr();
1001
1002 while (c != '"') {
1003 if (c == '\\') {
1004 if ((c = *src++) == '\0')
1005 inperr();
1006 switch (c) {
1007 case 'n':
1008 c = '\n';
1009 break;
1010 case 't':
1011 c = '\t';
1012 break;
1013 case 'v':
1014 c = '\v';
1015 break;
1016 case 'b':
1017 c = '\b';
1018 break;
1019 case 'r':
1020 c = '\r';
1021 break;
1022 case 'f':
1023 c = '\f';
1024 break;
1025 case 'a':
1026 c = '\a';
1027 break;
1028 case '\\':
1029 c = '\\';
1030 break;
1031 case '"':
1032 c = '"';
1033 break;
1034 case '\'':
1035 c = '\'';
1036 break;
1037 case '0': case '1': case '2': case '3':
1038 v = (c - '0') << 6;
1039 if ((c = *src++) < '0' || c > '7')
1040 inperr();
1041 v |= (c - '0') << 3;
1042 if ((c = *src++) < '0' || c > '7')
1043 inperr();
1044 v |= c - '0';
1045 c = (u_char)v;
1046 break;
1047 default:
1048 inperr();
1049 }
1050 }
1051 /* keep space for trailing '\0' */
1052 if (dst - strg == slen - 1) {
1053 if ((strg = realloc(strg, slen * 2)) == NULL)
1054 nomem();
1055 dst = strg + (slen - 1);
1056 slen *= 2;
1057 }
1058 *dst++ = (char)c;
1059 if ((c = *src++) == '\0')
1060 inperr();
1061 }
1062 *dst = '\0';
1063
1064 *epp = src;
1065 return (strg);
1066}
1067
1068/*
1069 * Read the name of a symbol in static memory.
1070 */
1071static const char *
1072inpname(const char *cp, const char **epp)
1073{
1074 static char *buf;
1075 static size_t blen = 0;
1076 size_t len, i;
1077 char *eptr, c;
1078
1079 len = (int)strtol(cp, &eptr, 10);
1080 if (cp == eptr)
1081 inperr();
1082 cp = eptr;
1083 if (len + 1 > blen)
1084 if ((buf = realloc(buf, blen = len + 1)) == NULL)
1085 nomem();
1086 for (i = 0; i < len; i++) {
1087 c = *cp++;
1088 if (!isalnum((unsigned char)c) && c != '_')
1089 inperr();
1090 buf[i] = c;
1091 }
1092 buf[i] = '\0';
1093
1094 *epp = cp;
1095 return (buf);
1096}
1097
1098/*
1099 * Return the index of a file name. If the name cannot be found, create
1100 * a new entry and return the index of the newly created entry.
1101 */
1102static int
1103getfnidx(const char *fn)
1104{
1105 int i;
1106
1107 /* 0 ist reserved */
1108 for (i = 1; fnames[i] != NULL; i++) {
1109 if (strcmp(fnames[i], fn) == 0)
1110 break;
1111 }
1112 if (fnames[i] != NULL)
1113 return (i);
1114
1115 if (i == nfnames - 1) {
1116 if ((fnames = realloc(fnames, (nfnames * 2) * sizeof (char *)))
1117 == NULL)
1118 nomem();
1119 (void)memset(fnames + nfnames, 0, nfnames * sizeof (char *));
1120 nfnames *= 2;
1121 }
1122
1123 if ((fnames[i] = strdup(fn)) == NULL)
1124 nomem();
1125 return (i);
1126}
1127
1128/*
1129 * Separate symbols with static and external linkage.
1130 */
1131void
1132mkstatic(hte_t *hte)
1133{
1134 sym_t *sym1, **symp, *sym;
1135 fcall_t **callp, *call;
1136 usym_t **usymp, *usym;
1137 hte_t *nhte;
1138 int ofnd;
1139
1140 /* Look for first static definition */
1141 for (sym1 = hte->h_syms; sym1 != NULL; sym1 = sym1->s_nxt) {
1142 if (sym1->s_static)
1143 break;
1144 }
1145 if (sym1 == NULL)
1146 return;
1147
1148 /* Do nothing if this name is used only in one translation unit. */
1149 ofnd = 0;
1150 for (sym = hte->h_syms; sym != NULL && !ofnd; sym = sym->s_nxt) {
1151 if (sym->s_pos.p_src != sym1->s_pos.p_src)
1152 ofnd = 1;
1153 }
1154 for (call = hte->h_calls; call != NULL && !ofnd; call = call->f_nxt) {
1155 if (call->f_pos.p_src != sym1->s_pos.p_src)
1156 ofnd = 1;
1157 }
1158 for (usym = hte->h_usyms; usym != NULL && !ofnd; usym = usym->u_nxt) {
1159 if (usym->u_pos.p_src != sym1->s_pos.p_src)
1160 ofnd = 1;
1161 }
1162 if (!ofnd) {
1163 hte->h_used = 1;
1164 /* errors about undef. static symbols are printed in lint1 */
1165 hte->h_def = 1;
1166 hte->h_static = 1;
1167 return;
1168 }
1169
1170 /*
1171 * Create a new hash table entry
1172 *
1173 * XXX this entry should be put at the beginning of the list to
1174 * avoid to process the same symbol twice.
1175 */
1176 for (nhte = hte; nhte->h_link != NULL; nhte = nhte->h_link)
1177 continue;
1178 nhte->h_link = xmalloc(sizeof (hte_t));
1179 nhte = nhte->h_link;
1180 nhte->h_name = hte->h_name;
1181 nhte->h_used = 1;
1182 nhte->h_def = 1; /* error in lint1 */
1183 nhte->h_static = 1;
1184 nhte->h_syms = NULL;
1185 nhte->h_lsym = &nhte->h_syms;
1186 nhte->h_calls = NULL;
1187 nhte->h_lcall = &nhte->h_calls;
1188 nhte->h_usyms = NULL;
1189 nhte->h_lusym = &nhte->h_usyms;
1190 nhte->h_link = NULL;
1191 nhte->h_hte = NULL;
1192
1193 /*
1194 * move all symbols used in this translation unit into the new
1195 * hash table entry.
1196 */
1197 for (symp = &hte->h_syms; (sym = *symp) != NULL; ) {
1198 if (sym->s_pos.p_src == sym1->s_pos.p_src) {
1199 sym->s_static = 1;
1200 (*symp) = sym->s_nxt;
1201 if (hte->h_lsym == &sym->s_nxt)
1202 hte->h_lsym = symp;
1203 sym->s_nxt = NULL;
1204 *nhte->h_lsym = sym;
1205 nhte->h_lsym = &sym->s_nxt;
1206 } else {
1207 symp = &sym->s_nxt;
1208 }
1209 }
1210 for (callp = &hte->h_calls; (call = *callp) != NULL; ) {
1211 if (call->f_pos.p_src == sym1->s_pos.p_src) {
1212 (*callp) = call->f_nxt;
1213 if (hte->h_lcall == &call->f_nxt)
1214 hte->h_lcall = callp;
1215 call->f_nxt = NULL;
1216 *nhte->h_lcall = call;
1217 nhte->h_lcall = &call->f_nxt;
1218 } else {
1219 callp = &call->f_nxt;
1220 }
1221 }
1222 for (usymp = &hte->h_usyms; (usym = *usymp) != NULL; ) {
1223 if (usym->u_pos.p_src == sym1->s_pos.p_src) {
1224 (*usymp) = usym->u_nxt;
1225 if (hte->h_lusym == &usym->u_nxt)
1226 hte->h_lusym = usymp;
1227 usym->u_nxt = NULL;
1228 *nhte->h_lusym = usym;
1229 nhte->h_lusym = &usym->u_nxt;
1230 } else {
1231 usymp = &usym->u_nxt;
1232 }
1233 }
1234
1235 /* h_def must be recalculated for old hte */
1236 hte->h_def = nhte->h_def = 0;
1237 for (sym = hte->h_syms; sym != NULL; sym = sym->s_nxt) {
1238 if (sym->s_def == DEF || sym->s_def == TDEF) {
1239 hte->h_def = 1;
1240 break;
1241 }
1242 }
1243
1244 mkstatic(hte);
1245}
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 1.19 2007/09/28 21:53:50 uwe Exp $");
38#endif
39__FBSDID("$FreeBSD: stable/10/usr.bin/xlint/lint2/read.c 281168 2015-04-06 19:56:27Z pfg $");
40
41#include <ctype.h>
42#include <err.h>
43#include <limits.h>
44#include <stdio.h>
45#include <stdlib.h>
46#include <string.h>
47
48#include "lint2.h"
49
50
51/* index of current (included) source file */
52static int srcfile;
53
54/*
55 * The array pointed to by inpfns maps the file name indices of input files
56 * to the file name indices used in lint2
57 */
58static short *inpfns;
59static size_t ninpfns;
60
61/*
62 * The array pointed to by *fnames maps file name indizes to file names.
63 * Indices of type short are used instead of pointers to save memory.
64 */
65const char **fnames;
66static size_t nfnames;
67
68/*
69 * Types are shared (to save memory for the types itself) and accessed
70 * via indices (to save memory for references to types (indices are short)).
71 * To share types, an equal type must be located fast. This is done by a
72 * hash table. Access by indices is done via an array of pointers to the
73 * types.
74 */
75typedef struct thtab {
76 const char *th_name;
77 u_short th_idx;
78 struct thtab *th_nxt;
79} thtab_t;
80static thtab_t **thtab; /* hash table */
81type_t **tlst; /* array for indexed access */
82static size_t tlstlen; /* length of tlst */
83
84static hte_t **renametab;
85
86/* index of current C source file (as spezified at the command line) */
87static int csrcfile;
88
89
90#define inperr() inperror(__FILE__, __LINE__)
91static void inperror(const char *, size_t);
92static void setsrc(const char *);
93static void setfnid(int, const char *);
94static void funccall(pos_t *, const char *);
95static void decldef(pos_t *, const char *);
96static void usedsym(pos_t *, const char *);
97static u_short inptype(const char *, const char **);
98static int gettlen(const char *, const char **);
99static u_short findtype(const char *, size_t, int);
100static u_short storetyp(type_t *, const char *, size_t, int);
101static int thash(const char *, size_t);
102static char *inpqstrg(const char *, const char **);
103static const char *inpname(const char *, const char **);
104static int getfnidx(const char *);
105
106void
107readfile(const char *name)
108{
109 FILE *inp;
110 size_t len;
111 const char *cp;
112 char *line, *eptr, rt = '\0';
113 int cline, isrc, iline;
114 pos_t pos;
115
116 if (inpfns == NULL)
117 if ((inpfns = calloc(ninpfns = 128, sizeof (short))) == NULL)
118 nomem();
119 if (fnames == NULL)
120 if ((fnames = calloc(nfnames = 256, sizeof (char *))) == NULL)
121 nomem();
122 if (tlstlen == 0)
123 if ((tlst = calloc(tlstlen = 256, sizeof (type_t *))) == NULL)
124 nomem();
125 if (thtab == NULL)
126 if ((thtab = calloc(THSHSIZ2, sizeof (thtab_t))) == NULL)
127 nomem();
128
129 _inithash(&renametab);
130
131 srcfile = getfnidx(name);
132
133 if ((inp = fopen(name, "r")) == NULL)
134 err(1, "cannot open %s", name);
135
136 while ((line = fgetln(inp, &len)) != NULL) {
137
138 if (len == 0 || line[len - 1] != '\n')
139 inperr();
140 line[len - 1] = '\0';
141 cp = line;
142
143 /* line number in csrcfile */
144 cline = (int)strtol(cp, &eptr, 10);
145 if (cp == eptr) {
146 cline = -1;
147 } else {
148 cp = eptr;
149 }
150
151 /* record type */
152 if (*cp != '\0') {
153 rt = *cp++;
154 } else {
155 inperr();
156 }
157
158 if (rt == 'S') {
159 setsrc(cp);
160 continue;
161 } else if (rt == 's') {
162 setfnid(cline, cp);
163 continue;
164 }
165
166 /*
167 * Index of (included) source file. If this index is
168 * different from csrcfile, it refers to an included
169 * file.
170 */
171 isrc = (int)strtol(cp, &eptr, 10);
172 if (cp == eptr)
173 inperr();
174 cp = eptr;
175 isrc = inpfns[isrc];
176
177 /* line number in isrc */
178 if (*cp++ != '.')
179 inperr();
180 iline = (int)strtol(cp, &eptr, 10);
181 if (cp == eptr)
182 inperr();
183 cp = eptr;
184
185 pos.p_src = (u_short)csrcfile;
186 pos.p_line = (u_short)cline;
187 pos.p_isrc = (u_short)isrc;
188 pos.p_iline = (u_short)iline;
189
190 /* process rest of this record */
191 switch (rt) {
192 case 'c':
193 funccall(&pos, cp);
194 break;
195 case 'd':
196 decldef(&pos, cp);
197 break;
198 case 'u':
199 usedsym(&pos, cp);
200 break;
201 default:
202 inperr();
203 }
204
205 }
206
207 _destroyhash(renametab);
208
209 if (ferror(inp))
210 err(1, "read error on %s", name);
211
212 (void)fclose(inp);
213}
214
215
216static void
217inperror(const char *file, size_t line)
218{
219
220 errx(1, "%s,%zd: input file error: %s", file, line, fnames[srcfile]);
221}
222
223/*
224 * Set the name of the C source file of the .ln file which is
225 * currently read.
226 */
227static void
228setsrc(const char *cp)
229{
230
231 csrcfile = getfnidx(cp);
232}
233
234/*
235 * setfnid() gets as input an index as used in an input file and the
236 * associated file name. If necessary, it creates a new lint2 file
237 * name index for this file name and creates the mapping of the index
238 * as used in the input file to the index used in lint2.
239 */
240static void
241setfnid(int fid, const char *cp)
242{
243
244 if (fid == -1)
245 inperr();
246
247 if (fid >= ninpfns) {
248 if ((inpfns = realloc(inpfns, (ninpfns * 2) * sizeof (short)))
249 == NULL)
250 nomem();
251 (void)memset(inpfns + ninpfns, 0, ninpfns * sizeof (short));
252 ninpfns *= 2;
253 }
254 /*
255 * Should always be true because indices written in the output
256 * file by lint1 are always the previous index + 1.
257 */
258 if (fid >= ninpfns)
259 errx(1, "internal error: setfnid()");
260 inpfns[fid] = (u_short)getfnidx(cp);
261}
262
263/*
264 * Process a function call record (c-record).
265 */
266static void
267funccall(pos_t *posp, const char *cp)
268{
269 arginf_t *ai, **lai;
270 char c, *eptr;
271 int rused, rdisc;
272 hte_t *hte;
273 fcall_t *fcall;
274 const char *name;
275
276 fcall = xalloc(sizeof (fcall_t));
277 STRUCT_ASSIGN(fcall->f_pos, *posp);
278
279 /* read flags */
280 rused = rdisc = 0;
281 lai = &fcall->f_args;
282 while ((c = *cp) == 'u' || c == 'i' || c == 'd' ||
283 c == 'z' || c == 'p' || c == 'n' || c == 's') {
284 cp++;
285 switch (c) {
286 case 'u':
287 if (rused || rdisc)
288 inperr();
289 rused = 1;
290 break;
291 case 'i':
292 if (rused || rdisc)
293 inperr();
294 break;
295 case 'd':
296 if (rused || rdisc)
297 inperr();
298 rdisc = 1;
299 break;
300 case 'z':
301 case 'p':
302 case 'n':
303 case 's':
304 ai = xalloc(sizeof (arginf_t));
305 ai->a_num = (int)strtol(cp, &eptr, 10);
306 if (cp == eptr)
307 inperr();
308 cp = eptr;
309 if (c == 'z') {
310 ai->a_pcon = ai->a_zero = 1;
311 } else if (c == 'p') {
312 ai->a_pcon = 1;
313 } else if (c == 'n') {
314 ai->a_ncon = 1;
315 } else {
316 ai->a_fmt = 1;
317 ai->a_fstrg = inpqstrg(cp, &cp);
318 }
319 *lai = ai;
320 lai = &ai->a_nxt;
321 break;
322 }
323 }
324 fcall->f_rused = rused;
325 fcall->f_rdisc = rdisc;
326
327 /* read name of function */
328 name = inpname(cp, &cp);
329
330 /* first look it up in the renaming table, then in the normal table */
331 hte = _hsearch(renametab, name, 0);
332 if (hte != NULL)
333 hte = hte->h_hte;
334 else
335 hte = hsearch(name, 1);
336 hte->h_used = 1;
337
338 fcall->f_type = inptype(cp, &cp);
339
340 *hte->h_lcall = fcall;
341 hte->h_lcall = &fcall->f_nxt;
342
343 if (*cp != '\0')
344 inperr();
345}
346
347/*
348 * Process a declaration or definition (d-record).
349 */
350static void
351decldef(pos_t *posp, const char *cp)
352{
353 sym_t *symp, sym;
354 char c, *ep, *pos1;
355 int used, renamed;
356 hte_t *hte, *renamehte = NULL;
357 const char *name, *rename;
358
359 (void)memset(&sym, 0, sizeof (sym));
360 STRUCT_ASSIGN(sym.s_pos, *posp);
361 sym.s_def = NODECL;
362
363 used = 0;
364
365 while (strchr("tdeurosvPS", (c = *cp)) != NULL) {
366 cp++;
367 switch (c) {
368 case 't':
369 if (sym.s_def != NODECL)
370 inperr();
371 sym.s_def = TDEF;
372 break;
373 case 'd':
374 if (sym.s_def != NODECL)
375 inperr();
376 sym.s_def = DEF;
377 break;
378 case 'e':
379 if (sym.s_def != NODECL)
380 inperr();
381 sym.s_def = DECL;
382 break;
383 case 'u':
384 if (used)
385 inperr();
386 used = 1;
387 break;
388 case 'r':
389 if (sym.s_rval)
390 inperr();
391 sym.s_rval = 1;
392 break;
393 case 'o':
394 if (sym.s_osdef)
395 inperr();
396 sym.s_osdef = 1;
397 break;
398 case 's':
399 if (sym.s_static)
400 inperr();
401 sym.s_static = 1;
402 break;
403 case 'v':
404 if (sym.s_va)
405 inperr();
406 sym.s_va = 1;
407 sym.s_nva = (short)strtol(cp, &ep, 10);
408 if (cp == ep)
409 inperr();
410 cp = ep;
411 break;
412 case 'P':
413 if (sym.s_prfl)
414 inperr();
415 sym.s_prfl = 1;
416 sym.s_nprfl = (short)strtol(cp, &ep, 10);
417 if (cp == ep)
418 inperr();
419 cp = ep;
420 break;
421 case 'S':
422 if (sym.s_scfl)
423 inperr();
424 sym.s_scfl = 1;
425 sym.s_nscfl = (short)strtol(cp, &ep, 10);
426 if (cp == ep)
427 inperr();
428 cp = ep;
429 break;
430 }
431 }
432
433 /* read symbol name, doing renaming if necessary */
434 name = inpname(cp, &cp);
435 renamed = 0;
436 if (*cp == 'r') {
437 cp++;
438 name = xstrdup(name);
439 rename = inpname(cp, &cp);
440
441 /* enter it and see if it's already been renamed */
442 renamehte = _hsearch(renametab, name, 1);
443 if (renamehte->h_hte == NULL) {
444 hte = hsearch(rename, 1);
445 renamehte->h_hte = hte;
446 renamed = 1;
447 } else if (strcmp((hte = renamehte->h_hte)->h_name, rename)) {
448 pos1 = xstrdup(mkpos(&renamehte->h_syms->s_pos));
449 /* %s renamed multiple times\t%s :: %s */
450 msg(18, name, pos1, mkpos(&sym.s_pos));
451 free(pos1);
452 }
453 free((char *)name);
454 } else {
455 /* it might be a previously-done rename */
456 hte = _hsearch(renametab, name, 0);
457 if (hte != NULL)
458 hte = hte->h_hte;
459 else
460 hte = hsearch(name, 1);
461 }
462 hte->h_used |= used;
463 if (sym.s_def == DEF || sym.s_def == TDEF)
464 hte->h_def = 1;
465
466 sym.s_type = inptype(cp, &cp);
467
468 /*
469 * Allocate memory for this symbol only if it was not already
470 * declared or tentatively defined at the same location with
471 * the same type. Works only for symbols with external linkage,
472 * because static symbols, tentatively defined at the same location
473 * but in different translation units are really different symbols.
474 */
475 for (symp = hte->h_syms; symp != NULL; symp = symp->s_nxt) {
476 if (symp->s_pos.p_isrc == sym.s_pos.p_isrc &&
477 symp->s_pos.p_iline == sym.s_pos.p_iline &&
478 symp->s_type == sym.s_type &&
479 ((symp->s_def == DECL && sym.s_def == DECL) ||
480 (!sflag && symp->s_def == TDEF && sym.s_def == TDEF)) &&
481 !symp->s_static && !sym.s_static) {
482 break;
483 }
484 }
485
486 if (symp == NULL) {
487 /* allocsym reserviert keinen Platz fuer s_nva */
488 if (sym.s_va || sym.s_prfl || sym.s_scfl) {
489 symp = xalloc(sizeof (sym_t));
490 STRUCT_ASSIGN(*symp, sym);
491 } else {
492 symp = xalloc(sizeof (symp->s_s));
493 STRUCT_ASSIGN(symp->s_s, sym.s_s);
494 }
495 *hte->h_lsym = symp;
496 hte->h_lsym = &symp->s_nxt;
497
498 /* XXX hack so we can remember where a symbol was renamed */
499 if (renamed)
500 renamehte->h_syms = symp;
501 }
502
503 if (*cp != '\0')
504 inperr();
505}
506
507/*
508 * Read an u-record (emitted by lint1 if a symbol was used).
509 */
510static void
511usedsym(pos_t *posp, const char *cp)
512{
513 usym_t *usym;
514 hte_t *hte;
515 const char *name;
516
517 usym = xalloc(sizeof (usym_t));
518 STRUCT_ASSIGN(usym->u_pos, *posp);
519
520 /* needed as delimiter between two numbers */
521 if (*cp++ != 'x')
522 inperr();
523
524 name = inpname(cp, &cp);
525 hte = _hsearch(renametab, name, 0);
526 if (hte != NULL)
527 hte = hte->h_hte;
528 else
529 hte = hsearch(name, 1);
530 hte->h_used = 1;
531
532 *hte->h_lusym = usym;
533 hte->h_lusym = &usym->u_nxt;
534}
535
536/*
537 * Read a type and return the index of this type.
538 */
539static u_short
540inptype(const char *cp, const char **epp)
541{
542 char c, s, *eptr;
543 const char *ep;
544 type_t *tp;
545 int narg, i, osdef = 0;
546 size_t tlen;
547 u_short tidx, sidx;
548 int h;
549
550 /* If we have this type already, return it's index. */
551 tlen = gettlen(cp, &ep);
552 h = thash(cp, tlen);
553 if ((tidx = findtype(cp, tlen, h)) != 0) {
554 *epp = ep;
555 return (tidx);
556 }
557
558 /* No, we must create a new type. */
559 tp = xalloc(sizeof (type_t));
560
561 tidx = storetyp(tp, cp, tlen, h);
562
563 c = *cp++;
564
565 while (c == 'c' || c == 'v') {
566 if (c == 'c') {
567 tp->t_const = 1;
568 } else {
569 tp->t_volatile = 1;
570 }
571 c = *cp++;
572 }
573
574 if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
575 s = c;
576 c = *cp++;
577 } else {
578 s = '\0';
579 }
580
581 switch (c) {
582 case 'C':
583 tp->t_tspec = s == 's' ? SCHAR : (s == 'u' ? UCHAR : CHAR);
584 break;
585 case 'S':
586 tp->t_tspec = s == 'u' ? USHORT : SHORT;
587 break;
588 case 'I':
589 tp->t_tspec = s == 'u' ? UINT : INT;
590 break;
591 case 'L':
592 tp->t_tspec = s == 'u' ? ULONG : LONG;
593 break;
594 case 'Q':
595 tp->t_tspec = s == 'u' ? UQUAD : QUAD;
596 break;
597 case 'D':
598 tp->t_tspec = s == 's' ? FLOAT : (s == 'l' ? LDOUBLE : DOUBLE);
599 break;
600 case 'V':
601 tp->t_tspec = VOID;
602 break;
603 case 'P':
604 tp->t_tspec = PTR;
605 break;
606 case 'A':
607 tp->t_tspec = ARRAY;
608 break;
609 case 'F':
610 case 'f':
611 osdef = c == 'f';
612 tp->t_tspec = FUNC;
613 break;
614 case 'T':
615 tp->t_tspec = s == 'e' ? ENUM : (s == 's' ? STRUCT : UNION);
616 break;
617 }
618
619 switch (tp->t_tspec) {
620 case ARRAY:
621 tp->t_dim = (int)strtol(cp, &eptr, 10);
622 cp = eptr;
623 sidx = inptype(cp, &cp); /* force seq. point! (ditto below) */
624 tp->t_subt = TP(sidx);
625 break;
626 case PTR:
627 sidx = inptype(cp, &cp);
628 tp->t_subt = TP(sidx);
629 break;
630 case FUNC:
631 c = *cp;
632 if (isdigit((u_char)c)) {
633 if (!osdef)
634 tp->t_proto = 1;
635 narg = (int)strtol(cp, &eptr, 10);
636 cp = eptr;
637 if ((tp->t_args = calloc((size_t)(narg + 1),
638 sizeof (type_t *))) == NULL)
639 nomem();
640 for (i = 0; i < narg; i++) {
641 if (i == narg - 1 && *cp == 'E') {
642 tp->t_vararg = 1;
643 cp++;
644 } else {
645 sidx = inptype(cp, &cp);
646 tp->t_args[i] = TP(sidx);
647 }
648 }
649 }
650 sidx = inptype(cp, &cp);
651 tp->t_subt = TP(sidx);
652 break;
653 case ENUM:
654 tp->t_tspec = INT;
655 tp->t_isenum = 1;
656 /* FALLTHROUGH */
657 case STRUCT:
658 case UNION:
659 switch (*cp++) {
660 case '1':
661 tp->t_istag = 1;
662 tp->t_tag = hsearch(inpname(cp, &cp), 1);
663 break;
664 case '2':
665 tp->t_istynam = 1;
666 tp->t_tynam = hsearch(inpname(cp, &cp), 1);
667 break;
668 case '3':
669 tp->t_isuniqpos = 1;
670 tp->t_uniqpos.p_line = strtol(cp, &eptr, 10);
671 cp = eptr;
672 cp++;
673 /* xlate to 'global' file name. */
674 tp->t_uniqpos.p_file =
675 addoutfile(inpfns[strtol(cp, &eptr, 10)]);
676 cp = eptr;
677 cp++;
678 tp->t_uniqpos.p_uniq = strtol(cp, &eptr, 10);
679 cp = eptr;
680 break;
681 }
682 break;
683 case LONG:
684 case VOID:
685 case LDOUBLE:
686 case DOUBLE:
687 case FLOAT:
688 case UQUAD:
689 case QUAD:
690 case ULONG:
691 case UINT:
692 case INT:
693 case USHORT:
694 case SHORT:
695 case UCHAR:
696 case SCHAR:
697 case CHAR:
698 case UNSIGN:
699 case SIGNED:
700 case NOTSPEC:
701 break;
702 }
703
704 *epp = cp;
705 return (tidx);
706}
707
708/*
709 * Get the length of a type string.
710 */
711static int
712gettlen(const char *cp, const char **epp)
713{
714 const char *cp1;
715 char c, s, *eptr;
716 tspec_t t;
717 int narg, i, cm, vm;
718
719 cp1 = cp;
720
721 c = *cp++;
722
723 cm = vm = 0;
724
725 while (c == 'c' || c == 'v') {
726 if (c == 'c') {
727 if (cm)
728 inperr();
729 cm = 1;
730 } else {
731 if (vm)
732 inperr();
733 vm = 1;
734 }
735 c = *cp++;
736 }
737
738 if (c == 's' || c == 'u' || c == 'l' || c == 'e') {
739 s = c;
740 c = *cp++;
741 } else {
742 s = '\0';
743 }
744
745 t = NOTSPEC;
746
747 switch (c) {
748 case 'C':
749 if (s == 's') {
750 t = SCHAR;
751 } else if (s == 'u') {
752 t = UCHAR;
753 } else if (s == '\0') {
754 t = CHAR;
755 }
756 break;
757 case 'S':
758 if (s == 'u') {
759 t = USHORT;
760 } else if (s == '\0') {
761 t = SHORT;
762 }
763 break;
764 case 'I':
765 if (s == 'u') {
766 t = UINT;
767 } else if (s == '\0') {
768 t = INT;
769 }
770 break;
771 case 'L':
772 if (s == 'u') {
773 t = ULONG;
774 } else if (s == '\0') {
775 t = LONG;
776 }
777 break;
778 case 'Q':
779 if (s == 'u') {
780 t = UQUAD;
781 } else if (s == '\0') {
782 t = QUAD;
783 }
784 break;
785 case 'D':
786 if (s == 's') {
787 t = FLOAT;
788 } else if (s == 'l') {
789 t = LDOUBLE;
790 } else if (s == '\0') {
791 t = DOUBLE;
792 }
793 break;
794 case 'V':
795 if (s == '\0')
796 t = VOID;
797 break;
798 case 'P':
799 if (s == '\0')
800 t = PTR;
801 break;
802 case 'A':
803 if (s == '\0')
804 t = ARRAY;
805 break;
806 case 'F':
807 case 'f':
808 if (s == '\0')
809 t = FUNC;
810 break;
811 case 'T':
812 if (s == 'e') {
813 t = ENUM;
814 } else if (s == 's') {
815 t = STRUCT;
816 } else if (s == 'u') {
817 t = UNION;
818 }
819 break;
820 default:
821 inperr();
822 }
823
824 if (t == NOTSPEC)
825 inperr();
826
827 switch (t) {
828 case ARRAY:
829 (void)strtol(cp, &eptr, 10);
830 if (cp == eptr)
831 inperr();
832 cp = eptr;
833 (void)gettlen(cp, &cp);
834 break;
835 case PTR:
836 (void)gettlen(cp, &cp);
837 break;
838 case FUNC:
839 c = *cp;
840 if (isdigit((u_char)c)) {
841 narg = (int)strtol(cp, &eptr, 10);
842 cp = eptr;
843 for (i = 0; i < narg; i++) {
844 if (i == narg - 1 && *cp == 'E') {
845 cp++;
846 } else {
847 (void)gettlen(cp, &cp);
848 }
849 }
850 }
851 (void)gettlen(cp, &cp);
852 break;
853 case ENUM:
854 case STRUCT:
855 case UNION:
856 switch (*cp++) {
857 case '1':
858 (void)inpname(cp, &cp);
859 break;
860 case '2':
861 (void)inpname(cp, &cp);
862 break;
863 case '3':
864 /* unique position: line.file.uniquifier */
865 (void)strtol(cp, &eptr, 10);
866 if (cp == eptr)
867 inperr();
868 cp = eptr;
869 if (*cp++ != '.')
870 inperr();
871 (void)strtol(cp, &eptr, 10);
872 if (cp == eptr)
873 inperr();
874 cp = eptr;
875 if (*cp++ != '.')
876 inperr();
877 (void)strtol(cp, &eptr, 10);
878 if (cp == eptr)
879 inperr();
880 cp = eptr;
881 break;
882 default:
883 inperr();
884 }
885 break;
886 case FLOAT:
887 case USHORT:
888 case SHORT:
889 case UCHAR:
890 case SCHAR:
891 case CHAR:
892 case UNSIGN:
893 case SIGNED:
894 case NOTSPEC:
895 case INT:
896 case UINT:
897 case DOUBLE:
898 case LDOUBLE:
899 case VOID:
900 case ULONG:
901 case QUAD:
902 case UQUAD:
903 case LONG:
904 break;
905 }
906
907 *epp = cp;
908 return (cp - cp1);
909}
910
911/*
912 * Search a type by it's type string.
913 */
914static u_short
915findtype(const char *cp, size_t len, int h)
916{
917 thtab_t *thte;
918
919 for (thte = thtab[h]; thte != NULL; thte = thte->th_nxt) {
920 if (strncmp(thte->th_name, cp, len) != 0)
921 continue;
922 if (thte->th_name[len] == '\0')
923 return (thte->th_idx);
924 }
925
926 return (0);
927}
928
929/*
930 * Store a type and it's type string so we can later share this type
931 * if we read the same type string from the input file.
932 */
933static u_short
934storetyp(type_t *tp, const char *cp, size_t len, int h)
935{
936 static u_int tidx = 1; /* 0 is reserved */
937 thtab_t *thte;
938 char *name;
939
940 if (tidx >= USHRT_MAX)
941 errx(1, "sorry, too many types");
942
943 if (tidx == tlstlen - 1) {
944 if ((tlst = realloc(tlst, (tlstlen * 2) * sizeof (type_t *)))
945 == NULL)
946 nomem();
947 (void)memset(tlst + tlstlen, 0, tlstlen * sizeof (type_t *));
948 tlstlen *= 2;
949 }
950
951 tlst[tidx] = tp;
952
953 /* create a hash table entry */
954 name = xalloc(len + 1);
955 (void)memcpy(name, cp, len);
956 name[len] = '\0';
957
958 thte = xalloc(sizeof (thtab_t));
959 thte->th_name = name;
960 thte->th_idx = tidx;
961 thte->th_nxt = thtab[h];
962 thtab[h] = thte;
963
964 return ((u_short)tidx++);
965}
966
967/*
968 * Hash function for types
969 */
970static int
971thash(const char *s, size_t len)
972{
973 u_int v;
974
975 v = 0;
976 while (len-- != 0) {
977 v = (v << sizeof (v)) + (u_char)*s++;
978 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
979 }
980 return (v % THSHSIZ2);
981}
982
983/*
984 * Read a string enclosed by "". This string may contain quoted chars.
985 */
986static char *
987inpqstrg(const char *src, const char **epp)
988{
989 char *strg, *dst;
990 size_t slen;
991 int c;
992 int v;
993
994 if ((dst = strg = malloc(slen = 32)) == NULL)
995 nomem();
996
997 if ((c = *src++) != '"')
998 inperr();
999 if ((c = *src++) == '\0')
1000 inperr();
1001
1002 while (c != '"') {
1003 if (c == '\\') {
1004 if ((c = *src++) == '\0')
1005 inperr();
1006 switch (c) {
1007 case 'n':
1008 c = '\n';
1009 break;
1010 case 't':
1011 c = '\t';
1012 break;
1013 case 'v':
1014 c = '\v';
1015 break;
1016 case 'b':
1017 c = '\b';
1018 break;
1019 case 'r':
1020 c = '\r';
1021 break;
1022 case 'f':
1023 c = '\f';
1024 break;
1025 case 'a':
1026 c = '\a';
1027 break;
1028 case '\\':
1029 c = '\\';
1030 break;
1031 case '"':
1032 c = '"';
1033 break;
1034 case '\'':
1035 c = '\'';
1036 break;
1037 case '0': case '1': case '2': case '3':
1038 v = (c - '0') << 6;
1039 if ((c = *src++) < '0' || c > '7')
1040 inperr();
1041 v |= (c - '0') << 3;
1042 if ((c = *src++) < '0' || c > '7')
1043 inperr();
1044 v |= c - '0';
1045 c = (u_char)v;
1046 break;
1047 default:
1048 inperr();
1049 }
1050 }
1051 /* keep space for trailing '\0' */
1052 if (dst - strg == slen - 1) {
1053 if ((strg = realloc(strg, slen * 2)) == NULL)
1054 nomem();
1055 dst = strg + (slen - 1);
1056 slen *= 2;
1057 }
1058 *dst++ = (char)c;
1059 if ((c = *src++) == '\0')
1060 inperr();
1061 }
1062 *dst = '\0';
1063
1064 *epp = src;
1065 return (strg);
1066}
1067
1068/*
1069 * Read the name of a symbol in static memory.
1070 */
1071static const char *
1072inpname(const char *cp, const char **epp)
1073{
1074 static char *buf;
1075 static size_t blen = 0;
1076 size_t len, i;
1077 char *eptr, c;
1078
1079 len = (int)strtol(cp, &eptr, 10);
1080 if (cp == eptr)
1081 inperr();
1082 cp = eptr;
1083 if (len + 1 > blen)
1084 if ((buf = realloc(buf, blen = len + 1)) == NULL)
1085 nomem();
1086 for (i = 0; i < len; i++) {
1087 c = *cp++;
1088 if (!isalnum((unsigned char)c) && c != '_')
1089 inperr();
1090 buf[i] = c;
1091 }
1092 buf[i] = '\0';
1093
1094 *epp = cp;
1095 return (buf);
1096}
1097
1098/*
1099 * Return the index of a file name. If the name cannot be found, create
1100 * a new entry and return the index of the newly created entry.
1101 */
1102static int
1103getfnidx(const char *fn)
1104{
1105 int i;
1106
1107 /* 0 ist reserved */
1108 for (i = 1; fnames[i] != NULL; i++) {
1109 if (strcmp(fnames[i], fn) == 0)
1110 break;
1111 }
1112 if (fnames[i] != NULL)
1113 return (i);
1114
1115 if (i == nfnames - 1) {
1116 if ((fnames = realloc(fnames, (nfnames * 2) * sizeof (char *)))
1117 == NULL)
1118 nomem();
1119 (void)memset(fnames + nfnames, 0, nfnames * sizeof (char *));
1120 nfnames *= 2;
1121 }
1122
1123 if ((fnames[i] = strdup(fn)) == NULL)
1124 nomem();
1125 return (i);
1126}
1127
1128/*
1129 * Separate symbols with static and external linkage.
1130 */
1131void
1132mkstatic(hte_t *hte)
1133{
1134 sym_t *sym1, **symp, *sym;
1135 fcall_t **callp, *call;
1136 usym_t **usymp, *usym;
1137 hte_t *nhte;
1138 int ofnd;
1139
1140 /* Look for first static definition */
1141 for (sym1 = hte->h_syms; sym1 != NULL; sym1 = sym1->s_nxt) {
1142 if (sym1->s_static)
1143 break;
1144 }
1145 if (sym1 == NULL)
1146 return;
1147
1148 /* Do nothing if this name is used only in one translation unit. */
1149 ofnd = 0;
1150 for (sym = hte->h_syms; sym != NULL && !ofnd; sym = sym->s_nxt) {
1151 if (sym->s_pos.p_src != sym1->s_pos.p_src)
1152 ofnd = 1;
1153 }
1154 for (call = hte->h_calls; call != NULL && !ofnd; call = call->f_nxt) {
1155 if (call->f_pos.p_src != sym1->s_pos.p_src)
1156 ofnd = 1;
1157 }
1158 for (usym = hte->h_usyms; usym != NULL && !ofnd; usym = usym->u_nxt) {
1159 if (usym->u_pos.p_src != sym1->s_pos.p_src)
1160 ofnd = 1;
1161 }
1162 if (!ofnd) {
1163 hte->h_used = 1;
1164 /* errors about undef. static symbols are printed in lint1 */
1165 hte->h_def = 1;
1166 hte->h_static = 1;
1167 return;
1168 }
1169
1170 /*
1171 * Create a new hash table entry
1172 *
1173 * XXX this entry should be put at the beginning of the list to
1174 * avoid to process the same symbol twice.
1175 */
1176 for (nhte = hte; nhte->h_link != NULL; nhte = nhte->h_link)
1177 continue;
1178 nhte->h_link = xmalloc(sizeof (hte_t));
1179 nhte = nhte->h_link;
1180 nhte->h_name = hte->h_name;
1181 nhte->h_used = 1;
1182 nhte->h_def = 1; /* error in lint1 */
1183 nhte->h_static = 1;
1184 nhte->h_syms = NULL;
1185 nhte->h_lsym = &nhte->h_syms;
1186 nhte->h_calls = NULL;
1187 nhte->h_lcall = &nhte->h_calls;
1188 nhte->h_usyms = NULL;
1189 nhte->h_lusym = &nhte->h_usyms;
1190 nhte->h_link = NULL;
1191 nhte->h_hte = NULL;
1192
1193 /*
1194 * move all symbols used in this translation unit into the new
1195 * hash table entry.
1196 */
1197 for (symp = &hte->h_syms; (sym = *symp) != NULL; ) {
1198 if (sym->s_pos.p_src == sym1->s_pos.p_src) {
1199 sym->s_static = 1;
1200 (*symp) = sym->s_nxt;
1201 if (hte->h_lsym == &sym->s_nxt)
1202 hte->h_lsym = symp;
1203 sym->s_nxt = NULL;
1204 *nhte->h_lsym = sym;
1205 nhte->h_lsym = &sym->s_nxt;
1206 } else {
1207 symp = &sym->s_nxt;
1208 }
1209 }
1210 for (callp = &hte->h_calls; (call = *callp) != NULL; ) {
1211 if (call->f_pos.p_src == sym1->s_pos.p_src) {
1212 (*callp) = call->f_nxt;
1213 if (hte->h_lcall == &call->f_nxt)
1214 hte->h_lcall = callp;
1215 call->f_nxt = NULL;
1216 *nhte->h_lcall = call;
1217 nhte->h_lcall = &call->f_nxt;
1218 } else {
1219 callp = &call->f_nxt;
1220 }
1221 }
1222 for (usymp = &hte->h_usyms; (usym = *usymp) != NULL; ) {
1223 if (usym->u_pos.p_src == sym1->s_pos.p_src) {
1224 (*usymp) = usym->u_nxt;
1225 if (hte->h_lusym == &usym->u_nxt)
1226 hte->h_lusym = usymp;
1227 usym->u_nxt = NULL;
1228 *nhte->h_lusym = usym;
1229 nhte->h_lusym = &usym->u_nxt;
1230 } else {
1231 usymp = &usym->u_nxt;
1232 }
1233 }
1234
1235 /* h_def must be recalculated for old hte */
1236 hte->h_def = nhte->h_def = 0;
1237 for (sym = hte->h_syms; sym != NULL; sym = sym->s_nxt) {
1238 if (sym->s_def == DEF || sym->s_def == TDEF) {
1239 hte->h_def = 1;
1240 break;
1241 }
1242 }
1243
1244 mkstatic(hte);
1245}