xref: /illumos-gate/usr/src/cmd/file/file.c (revision 03831d35)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
23 /*	  All Rights Reserved  	*/
24 
25 
26 /*	Copyright (c) 1987, 1988 Microsoft Corporation	*/
27 /*	  All Rights Reserved	*/
28 
29 /*
30  * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
31  * Use is subject to license terms.
32  */
33 
34 #pragma ident	"%Z%%M%	%I%	%E% SMI"
35 
36 #define	_LARGEFILE64_SOURCE
37 
38 #include <ctype.h>
39 #include <unistd.h>
40 #include <fcntl.h>
41 #include <signal.h>
42 #include <stdio.h>
43 #include <libelf.h>
44 #include <stdlib.h>
45 #include <limits.h>
46 #include <locale.h>
47 #include <wctype.h>
48 #include <string.h>
49 #include <errno.h>
50 #include <door.h>
51 #include <sys/param.h>
52 #include <sys/types.h>
53 #include <sys/mkdev.h>
54 #include <sys/stat.h>
55 #include <sys/elf.h>
56 #include <sys/elf_M32.h>
57 #include <sys/elf_SPARC.h>
58 #include <procfs.h>
59 #include <sys/core.h>
60 #include <sys/dumphdr.h>
61 #include <netinet/in.h>
62 #include <gelf.h>
63 #include <elfcap.h>
64 #include "file.h"
65 
66 typedef Elf64_Nhdr	GElf_Nhdr;
67 
68 /*
69  *	Misc
70  */
71 
72 #define	FBSZ		512
73 #define	MLIST_SZ	12
74 
75 /*
76  * The 0x8FCA0102 magic string was used in crash dumps generated by releases
77  * prior to Solaris 7.
78  */
79 #define	OLD_DUMP_MAGIC	0x8FCA0102
80 
81 #if defined(__sparc)
82 #define	NATIVE_ISA	"SPARC"
83 #define	OTHER_ISA	"Intel"
84 #else
85 #define	NATIVE_ISA	"Intel"
86 #define	OTHER_ISA	"SPARC"
87 #endif
88 
89 /* Assembly language comment char */
90 #ifdef pdp11
91 #define	ASCOMCHAR '/'
92 #else
93 #define	ASCOMCHAR '!'
94 #endif
95 
96 #pragma	align	16(fbuf)
97 static char	fbuf[FBSZ];
98 
99 /*
100  * Magic file variables
101  */
102 static intmax_t maxmagicoffset;
103 static intmax_t tmpmax;
104 static char	*magicbuf;
105 
106 static char	*dfile;
107 static char	*troff[] = {	/* new troff intermediate lang */
108 		"x", "T", "res", "init", "font", "202", "V0", "p1", 0};
109 
110 static char	*fort[] = {			/* FORTRAN */
111 		"function", "subroutine", "common", "dimension", "block",
112 		"integer", "real", "data", "double",
113 		"FUNCTION", "SUBROUTINE", "COMMON", "DIMENSION", "BLOCK",
114 		"INTEGER", "REAL", "DATA", "DOUBLE", 0};
115 
116 static char	*asc[] = {		/* Assembler Commands */
117 		"sys", "mov", "tst", "clr", "jmp", "cmp", "set", "inc",
118 		"dec", 0};
119 
120 static char	*c[] = {			/* C Language */
121 		"int", "char", "float", "double", "short", "long", "unsigned",
122 		"register", "static", "struct", "extern", 0};
123 
124 static char	*as[] = {	/* Assembler Pseudo Ops, prepended with '.' */
125 		"globl", "global", "ident", "file", "byte", "even",
126 		"text", "data", "bss", "comm", 0};
127 
128 /*
129  * The line and debug section names are used by the strip command.
130  * Any changes in the strip implementation need to be reflected here.
131  */
132 static char	*debug_sections[] = { /* Debug sections in a ELF file */
133 		".debug", ".stab", ".dwarf", ".line", NULL};
134 
135 
136 /* start for MB env */
137 static wchar_t wchar;
138 static int	length;
139 static int	IS_ascii;
140 static int	Max;
141 /* end for MB env */
142 static int	i;	/* global index into first 'fbsz' bytes of file */
143 static int	fbsz;
144 static int	ifd = -1;
145 static int	elffd = -1;
146 static int	tret;
147 static int	hflg;
148 static int	dflg;
149 static int	mflg;
150 static int	M_flg;
151 static int	iflg;
152 static struct stat64	mbuf;
153 
154 static char	**mlist1;	/* 1st ordered list of magic files */
155 static char	**mlist2;	/* 2nd ordered list of magic files */
156 static size_t	mlist1_sz;	/* number of ptrs allocated for mlist1 */
157 static size_t	mlist2_sz;	/* number of ptrs allocated for mlist2 */
158 static char	**mlist1p;	/* next entry in mlist1 */
159 static char	**mlist2p;	/* next entry in mlist2 */
160 
161 static ssize_t	mread;
162 
163 static void is_stripped(Elf *elf);
164 static Elf *is_elf_file(int elffd);
165 static void ar_coff_or_aout(int ifd);
166 static int type(char *file);
167 static int def_position_tests(void);
168 static void def_context_tests(void);
169 static int troffint(char *bp, int n);
170 static int lookup(char **tab);
171 static int ccom(void);
172 static int ascom(void);
173 static int sccs(void);
174 static int english(char *bp, int n);
175 static int old_core(Elf *elf, GElf_Ehdr *ehdr, int format);
176 static int core(Elf *elf, GElf_Ehdr *ehdr, int format);
177 static int shellscript(char buf[], struct stat64 *sb);
178 static int elf_check(Elf *elf);
179 static int get_door_target(char *, char *, size_t);
180 static int zipfile(char *, int);
181 static int is_crash_dump(const char *, int);
182 static void print_dumphdr(const int, const dumphdr_t *, uint32_t (*)(uint32_t),
183     const char *);
184 static uint32_t swap_uint32(uint32_t);
185 static uint32_t return_uint32(uint32_t);
186 static int is_in_list(char *[], char *);
187 static void usage(void);
188 static void default_magic(void);
189 static void add_to_mlist(char *, int);
190 static void fd_cleanup(void);
191 
192 #ifdef XPG4
193 	/* SUSv3 requires a single <space> after the colon */
194 #define	prf(x)	(void) printf("%s: ", x);
195 #else	/* !XPG4 */
196 #define	prf(x)	(void) printf("%s:%s", x, (int)strlen(x) > 6 ? "\t" : "\t\t");
197 #endif	/* XPG4 */
198 
199 int
200 main(int argc, char **argv)
201 {
202 	char	*p;
203 	int	ch;
204 	FILE	*fl;
205 	int	cflg = 0;
206 	int	eflg = 0;
207 	int	fflg = 0;
208 	char	*ap = NULL;
209 	int	pathlen;
210 	char	**filep;
211 
212 	(void) setlocale(LC_ALL, "");
213 #if !defined(TEXT_DOMAIN)	/* Should be defined by cc -D */
214 #define	TEXT_DOMAIN "SYS_TEST"	/* Use this only if it weren't */
215 #endif
216 	(void) textdomain(TEXT_DOMAIN);
217 
218 	while ((ch = getopt(argc, argv, "M:cdf:him:")) != EOF) {
219 		switch (ch) {
220 
221 		case 'M':
222 			add_to_mlist(optarg, !dflg);
223 			M_flg++;
224 			break;
225 
226 		case 'c':
227 			cflg++;
228 			break;
229 
230 		case 'd':
231 			if (!dflg) {
232 				default_magic();
233 				add_to_mlist(dfile, 0);
234 				dflg++;
235 			}
236 			break;
237 
238 		case 'f':
239 			fflg++;
240 			if ((fl = fopen(optarg, "r")) == NULL) {
241 				(void) fprintf(stderr,
242 					gettext("cannot open %s\n"), optarg);
243 				usage();
244 			}
245 			pathlen = pathconf("/", _PC_PATH_MAX);
246 			if (pathlen == -1) {
247 				(void) fprintf(stderr,
248 				    gettext("pathconf: cannot determine "
249 					"maximum path length\n"));
250 				exit(1);
251 			}
252 			pathlen += 2; /* for null and newline in fgets */
253 			ap = malloc(pathlen * sizeof (char));
254 			if (ap == NULL) {
255 				perror("malloc");
256 				exit(1);
257 			}
258 			break;
259 
260 		case 'h':
261 			hflg++;
262 			break;
263 
264 		case 'i':
265 			iflg++;
266 			break;
267 
268 		case 'm':
269 			add_to_mlist(optarg, !dflg);
270 			mflg++;
271 			break;
272 
273 		case '?':
274 			eflg++;
275 			break;
276 		}
277 	}
278 	if (!cflg && !fflg && (eflg || optind == argc))
279 		usage();
280 	if (iflg && (dflg || mflg || M_flg)) {
281 		usage();
282 	}
283 	if (iflg && cflg) {
284 		usage();
285 	}
286 
287 	if (!dflg && !mflg && !M_flg && !iflg) {
288 	/* no -d, -m, nor -M option; also -i option doesn't need magic  */
289 		default_magic();
290 		if (f_mkmtab(dfile, cflg, 0) == -1) {
291 			exit(2);
292 		}
293 	}
294 
295 	else if (mflg && !M_flg && !dflg) {
296 	/* -m specified without -d nor -M */
297 
298 #ifdef XPG4	/* For SUSv3 only */
299 
300 		/*
301 		 * The default position-dependent magic file tests
302 		 * in /etc/magic will follow all the -m magic tests.
303 		 */
304 
305 		for (filep = mlist1; filep < mlist1p; filep++) {
306 			if (f_mkmtab(*filep, cflg, 1) == -1) {
307 				exit(2);
308 			}
309 		}
310 		default_magic();
311 		if (f_mkmtab(dfile, cflg, 0) == -1) {
312 			exit(2);
313 		}
314 #else	/* !XPG4 */
315 		/*
316 		 * Retain Solaris file behavior for -m before SUSv3,
317 		 * when the new -d and -M options are not specified.
318 		 * Use the -m file specified in place of the default
319 		 * /etc/magic file.  Solaris file will
320 		 * now allow more than one magic file to be specified
321 		 * with multiple -m options, for consistency with
322 		 * other behavior.
323 		 *
324 		 * Put the magic table(s) specified by -m into
325 		 * the second magic table instead of the first
326 		 * (as indicated by the last argument to f_mkmtab()),
327 		 * since they replace the /etc/magic tests and
328 		 * must be executed alongside the default
329 		 * position-sensitive tests.
330 		 */
331 
332 		for (filep = mlist1; filep < mlist1p; filep++) {
333 			if (f_mkmtab(*filep, cflg, 0) == -1) {
334 				exit(2);
335 			}
336 		}
337 #endif /* XPG4 */
338 	} else {
339 		/*
340 		 * For any other combination of -d, -m, and -M,
341 		 * use the magic files in command-line order.
342 		 * Store the entries from the two separate lists of magic
343 		 * files, if any, into two separate magic file tables.
344 		 * mlist1: magic tests executed before default magic tests
345 		 * mlist2: default magic tests and after
346 		 */
347 		for (filep = mlist1; filep && (filep < mlist1p); filep++) {
348 			if (f_mkmtab(*filep, cflg, 1) == -1) {
349 				exit(2);
350 			}
351 		}
352 		for (filep = mlist2; filep && (filep < mlist2p); filep++) {
353 			if (f_mkmtab(*filep, cflg, 0) == -1) {
354 				exit(2);
355 			}
356 		}
357 	}
358 
359 	/* Initialize the magic file variables; check both magic tables */
360 	tmpmax = f_getmaxoffset(1);
361 	maxmagicoffset = f_getmaxoffset(0);
362 	if (maxmagicoffset < tmpmax) {
363 		maxmagicoffset = tmpmax;
364 	}
365 	if (maxmagicoffset < (intmax_t)FBSZ)
366 		maxmagicoffset = (intmax_t)FBSZ;
367 	if ((magicbuf = (char *)malloc(maxmagicoffset)) == NULL) {
368 		(void) fprintf(stderr, gettext("malloc failed\n"));
369 		exit(2);
370 	}
371 
372 	if (cflg) {
373 		f_prtmtab();
374 		if (ferror(stdout) != 0) {
375 			(void) fprintf(stderr, gettext("file: error writing to "
376 			    "stdout\n"));
377 			exit(1);
378 		}
379 		if (fclose(stdout) != 0) {
380 			perror(gettext("file: fclose failed"));
381 			exit(1);
382 		}
383 		exit(0);
384 	}
385 	for (; fflg || optind < argc; optind += !fflg) {
386 		register int	l;
387 
388 		if (fflg) {
389 			if ((p = fgets(ap, pathlen, fl)) == NULL) {
390 				fflg = 0;
391 				optind--;
392 				continue;
393 			}
394 			l = strlen(p);
395 			if (l > 0)
396 				p[l - 1] = '\0';
397 		} else
398 			p = argv[optind];
399 		prf(p);				/* print "file_name:<tab>" */
400 
401 		if (type(p))
402 			tret = 1;
403 	}
404 	if (ap != NULL)
405 		free(ap);
406 	if (tret != 0) {
407 		exit(tret);
408 	}
409 	if (ferror(stdout) != 0) {
410 		(void) fprintf(stderr, gettext("file: error writing to "
411 		    "stdout\n"));
412 		exit(1);
413 	}
414 	if (fclose(stdout) != 0) {
415 		perror(gettext("file: fclose failed"));
416 		exit(1);
417 	}
418 	return (0);
419 }
420 
421 static int
422 type(char *file)
423 {
424 	int	cc;
425 	char	buf[BUFSIZ];
426 	int	(*statf)() = hflg ? lstat64 : stat64;
427 
428 	i = 0;		/* reset index to beginning of file */
429 	ifd = -1;
430 	if ((*statf)(file, &mbuf) < 0) {
431 		if (statf == lstat64 || lstat64(file, &mbuf) < 0) {
432 			(void) printf(gettext("cannot open: %s\n"),
433 			    strerror(errno));
434 			return (0);		/* POSIX.2 */
435 		}
436 	}
437 	switch (mbuf.st_mode & S_IFMT) {
438 	case S_IFREG:
439 		if (iflg) {
440 			(void) printf(gettext("regular file\n"));
441 			return (0);
442 		}
443 		break;
444 	case S_IFCHR:
445 		(void) printf(gettext("character"));
446 		goto spcl;
447 
448 	case S_IFDIR:
449 		(void) printf(gettext("directory\n"));
450 		return (0);
451 
452 	case S_IFIFO:
453 		(void) printf(gettext("fifo\n"));
454 		return (0);
455 
456 	case S_IFLNK:
457 		if ((cc = readlink(file, buf, BUFSIZ)) < 0) {
458 			(void) printf(gettext("readlink error: %s\n"),
459 				strerror(errno));
460 			return (1);
461 		}
462 		buf[cc] = '\0';
463 		(void) printf(gettext("symbolic link to %s\n"), buf);
464 		return (0);
465 
466 	case S_IFBLK:
467 		(void) printf(gettext("block"));
468 					/* major and minor, see sys/mkdev.h */
469 spcl:
470 		(void) printf(gettext(" special (%d/%d)\n"),
471 		    major(mbuf.st_rdev), minor(mbuf.st_rdev));
472 		return (0);
473 
474 	case S_IFSOCK:
475 		(void) printf("socket\n");
476 		/* FIXME, should open and try to getsockname. */
477 		return (0);
478 
479 	case S_IFDOOR:
480 		if (get_door_target(file, buf, sizeof (buf)) == 0)
481 			(void) printf(gettext("door to %s\n"), buf);
482 		else
483 			(void) printf(gettext("door\n"));
484 		return (0);
485 
486 	}
487 
488 	if (elf_version(EV_CURRENT) == EV_NONE) {
489 		(void) printf(gettext("libelf is out of date\n"));
490 		return (1);
491 	}
492 
493 	ifd = open64(file, O_RDONLY);
494 	if (ifd < 0) {
495 		(void) printf(gettext("cannot open: %s\n"), strerror(errno));
496 		return (0);			/* POSIX.2 */
497 	}
498 
499 	/* need another fd for elf, since we might want to read the file too */
500 	elffd = open64(file, O_RDONLY);
501 	if (elffd < 0) {
502 		(void) printf(gettext("cannot open: %s\n"), strerror(errno));
503 		(void) close(ifd);
504 		ifd = -1;
505 		return (0);			/* POSIX.2 */
506 	}
507 	if ((fbsz = read(ifd, fbuf, FBSZ)) == -1) {
508 		(void) printf(gettext("cannot read: %s\n"), strerror(errno));
509 		(void) close(ifd);
510 		ifd = -1;
511 		return (0);			/* POSIX.2 */
512 	}
513 	if (fbsz == 0) {
514 		(void) printf(gettext("empty file\n"));
515 		fd_cleanup();
516 		return (0);
517 	}
518 
519 	/*
520 	 * First try user-specified position-dependent magic tests, if any,
521 	 * which need to execute before the default tests.
522 	 */
523 	if ((mread = pread(ifd, (void*)magicbuf, (size_t)maxmagicoffset,
524 			(off_t)0)) == -1) {
525 		(void) printf(gettext("cannot read: %s\n"), strerror(errno));
526 		fd_cleanup();
527 		return (0);
528 	}
529 
530 	/*
531 	 * ChecK against Magic Table entries.
532 	 * Check first magic table for magic tests to be applied
533 	 * before default tests.
534 	 * If no default tests are to be applied, all magic tests
535 	 * should occur in this magic table.
536 	 */
537 	switch (f_ckmtab(magicbuf, mread, 1)) {
538 		case -1:	/* Error */
539 			exit(2);
540 			break;
541 		case 0:		/* Not magic */
542 			break;
543 		default:	/* Switch is magic index */
544 			(void) putchar('\n');
545 			fd_cleanup();
546 			return (0);
547 			/* NOTREACHED */
548 			break;
549 	}
550 
551 	if (dflg || !M_flg) {
552 		/*
553 		 * default position-dependent tests,
554 		 * plus non-default magic tests, if any
555 		 */
556 		switch (def_position_tests()) {
557 			case -1:	/* error */
558 				fd_cleanup();
559 				return (1);
560 			case 1:	/* matching type found */
561 				fd_cleanup();
562 				return (0);
563 				/* NOTREACHED */
564 				break;
565 			case 0:		/* no matching type found */
566 				break;
567 		}
568 		/* default context-sensitive tests */
569 		def_context_tests();
570 	} else {
571 		/* no more tests to apply; no match was found */
572 		(void) printf(gettext("data\n"));
573 	}
574 	fd_cleanup();
575 	return (0);
576 }
577 
578 /*
579  * def_position_tests() - applies default position-sensitive tests,
580  *	looking for values in specific positions in the file.
581  *	These are followed by default (followed by possibly some
582  *	non-default) magic file tests.
583  *
584  *	All position-sensitive tests, default or otherwise, must
585  *	be applied before context-sensitive tests, to avoid
586  *	false context-sensitive matches.
587  *
588  * 	Returns -1 on error which should result in error (non-zero)
589  *	exit status for the file utility.
590  *	Returns 0 if no matching file type found.
591  *	Returns 1 if matching file type found.
592  */
593 
594 static int
595 def_position_tests(void)
596 {
597 	Elf	*elf;
598 
599 	if (sccs()) {	/* look for "1hddddd" where d is a digit */
600 		(void) printf("sccs \n");
601 		return (1);
602 	}
603 	if (fbuf[0] == '#' && fbuf[1] == '!' && shellscript(fbuf+2, &mbuf))
604 		return (1);
605 	if ((elf = is_elf_file(elffd)) != NULL) {
606 		(void) elf_check(elf);
607 		(void) elf_end(elf);
608 		(void) putchar('\n');
609 		return (1);
610 
611 	/* LINTED: pointer cast may result in improper alignment */
612 	} else if (*(int *)fbuf == CORE_MAGIC) {
613 		/* LINTED: pointer cast may result in improper alignment */
614 		struct core *corep = (struct core *)fbuf;
615 
616 		(void) printf("a.out core file");
617 
618 		if (*(corep->c_cmdname) != '\0')
619 			(void) printf(" from '%s'", corep->c_cmdname);
620 		(void) putchar('\n');
621 		return (1);
622 	}
623 
624 	/*
625 	 * ZIP files, JAR files, and Java executables
626 	 */
627 	if (zipfile(fbuf, ifd))
628 		return (1);
629 
630 	if (is_crash_dump(fbuf, ifd))
631 		return (1);
632 
633 	/*
634 	 * ChecK against Magic Table entries.
635 	 * The magic entries checked here always start with default
636 	 * magic tests and may be followed by other, non-default magic
637 	 * tests.  If no default tests are to be executed, all the
638 	 * magic tests should have been in the first magic table.
639 	 */
640 	switch (f_ckmtab(magicbuf, mread, 0)) {
641 		case -1:	/* Error */
642 			exit(2);
643 			break;
644 		case 0:		/* Not magic */
645 			return (0);
646 			/* NOTREACHED */
647 			break;
648 		default:	/* Switch is magic index */
649 
650 			/*
651 			 * f_ckmtab recognizes file type,
652 			 * check if it is PostScript.
653 			 * if not, check if elf or a.out
654 			 */
655 			if (magicbuf[0] == '%' && magicbuf[1] == '!') {
656 				(void) putchar('\n');
657 			} else {
658 
659 				/*
660 				 * Check that the file is executable (dynamic
661 				 * objects must be executable to be exec'ed,
662 				 * shared objects need not be, but by convention
663 				 * should be executable).
664 				 *
665 				 * Note that we should already have processed
666 				 * the file if it was an ELF file.
667 				 */
668 				ar_coff_or_aout(elffd);
669 				(void) putchar('\n');
670 			}
671 			return (1);
672 			/* NOTREACHED */
673 			break;
674 	}
675 
676 	return (0);	/* file was not identified */
677 }
678 
679 /*
680  * def_context_tests() - default context-sensitive tests.
681  *	These are the last tests to be applied.
682  *	If no match is found, prints out "data".
683  */
684 
685 static void
686 def_context_tests(void)
687 {
688 	int	j;
689 	int	nl;
690 	char	ch;
691 	int	len;
692 
693 	if (ccom() == 0)
694 		goto notc;
695 	while (fbuf[i] == '#') {
696 		j = i;
697 		while (fbuf[i++] != '\n') {
698 			if (i - j > 255) {
699 				(void) printf(gettext("data\n"));
700 				return;
701 			}
702 			if (i >= fbsz)
703 				goto notc;
704 		}
705 		if (ccom() == 0)
706 			goto notc;
707 	}
708 check:
709 	if (lookup(c) == 1) {
710 		while ((ch = fbuf[i]) != ';' && ch != '{') {
711 			if ((len = mblen(&fbuf[i], MB_CUR_MAX)) <= 0)
712 				len = 1;
713 			i += len;
714 			if (i >= fbsz)
715 				goto notc;
716 		}
717 		(void) printf(gettext("c program text"));
718 		goto outa;
719 	}
720 	nl = 0;
721 	while (fbuf[i] != '(') {
722 		if (fbuf[i] <= 0)
723 			goto notas;
724 		if (fbuf[i] == ';') {
725 			i++;
726 			goto check;
727 		}
728 		if (fbuf[i++] == '\n')
729 			if (nl++ > 6)
730 				goto notc;
731 		if (i >= fbsz)
732 			goto notc;
733 	}
734 	while (fbuf[i] != ')') {
735 		if (fbuf[i++] == '\n')
736 			if (nl++ > 6)
737 				goto notc;
738 		if (i >= fbsz)
739 			goto notc;
740 	}
741 	while (fbuf[i] != '{') {
742 		if ((len = mblen(&fbuf[i], MB_CUR_MAX)) <= 0)
743 			len = 1;
744 		if (fbuf[i] == '\n')
745 			if (nl++ > 6)
746 				goto notc;
747 		i += len;
748 		if (i >= fbsz)
749 			goto notc;
750 	}
751 	(void) printf(gettext("c program text"));
752 	goto outa;
753 notc:
754 	i = 0;			/* reset to begining of file again */
755 	while (fbuf[i] == 'c' || fbuf[i] == 'C'|| fbuf[i] == '!' ||
756 	    fbuf[i] == '*' || fbuf[i] == '\n') {
757 		while (fbuf[i++] != '\n')
758 			if (i >= fbsz)
759 				goto notfort;
760 	}
761 	if (lookup(fort) == 1) {
762 		(void) printf(gettext("fortran program text"));
763 		goto outa;
764 	}
765 notfort:			/* looking for assembler program */
766 	i = 0;			/* reset to beginning of file again */
767 	if (ccom() == 0)	/* assembler programs may contain */
768 				/* c-style comments */
769 		goto notas;
770 	if (ascom() == 0)
771 		goto notas;
772 	j = i - 1;
773 	if (fbuf[i] == '.') {
774 		i++;
775 		if (lookup(as) == 1) {
776 			(void) printf(gettext("assembler program text"));
777 			goto outa;
778 		} else if (j != -1 && fbuf[j] == '\n' && isalpha(fbuf[j + 2])) {
779 			(void) printf(
780 			    gettext("[nt]roff, tbl, or eqn input text"));
781 			goto outa;
782 		}
783 	}
784 	while (lookup(asc) == 0) {
785 		if (ccom() == 0)
786 			goto notas;
787 		if (ascom() == 0)
788 			goto notas;
789 		while (fbuf[i] != '\n' && fbuf[i++] != ':') {
790 			if (i >= fbsz)
791 				goto notas;
792 		}
793 		while (fbuf[i] == '\n' || fbuf[i] == ' ' || fbuf[i] == '\t')
794 			if (i++ >= fbsz)
795 				goto notas;
796 		j = i - 1;
797 		if (fbuf[i] == '.') {
798 			i++;
799 			if (lookup(as) == 1) {
800 				(void) printf(
801 				    gettext("assembler program text"));
802 				goto outa;
803 			} else if (fbuf[j] == '\n' && isalpha(fbuf[j+2])) {
804 				(void) printf(
805 				    gettext("[nt]roff, tbl, or eqn input "
806 				    "text"));
807 				goto outa;
808 			}
809 		}
810 	}
811 	(void) printf(gettext("assembler program text"));
812 	goto outa;
813 notas:
814 	/* start modification for multibyte env */
815 	IS_ascii = 1;
816 	if (fbsz < FBSZ)
817 		Max = fbsz;
818 	else
819 		Max = FBSZ - MB_LEN_MAX; /* prevent cut of wchar read */
820 	/* end modification for multibyte env */
821 
822 	for (i = 0; i < Max; /* null */)
823 		if (fbuf[i] & 0200) {
824 			IS_ascii = 0;
825 			if (fbuf[0] == '\100' && fbuf[1] == '\357') {
826 				(void) printf(gettext("troff output\n"));
827 				return;
828 			}
829 		/* start modification for multibyte env */
830 			if ((length = mbtowc(&wchar, &fbuf[i], MB_CUR_MAX))
831 			    <= 0 || !iswprint(wchar)) {
832 				(void) printf(gettext("data\n"));
833 				return;
834 			}
835 			i += length;
836 		}
837 		else
838 			i++;
839 	i = fbsz;
840 		/* end modification for multibyte env */
841 	if (mbuf.st_mode&(S_IXUSR|S_IXGRP|S_IXOTH))
842 		(void) printf(gettext("commands text"));
843 	else if (troffint(fbuf, fbsz))
844 		(void) printf(gettext("troff intermediate output text"));
845 	else if (english(fbuf, fbsz))
846 		(void) printf(gettext("English text"));
847 	else if (IS_ascii)
848 		(void) printf(gettext("ascii text"));
849 	else
850 		(void) printf(gettext("text")); /* for multibyte env */
851 outa:
852 	/*
853 	 * This code is to make sure that no MB char is cut in half
854 	 * while still being used.
855 	 */
856 	fbsz = (fbsz < FBSZ ? fbsz : fbsz - MB_CUR_MAX + 1);
857 	while (i < fbsz) {
858 		if (isascii(fbuf[i])) {
859 			i++;
860 			continue;
861 		} else {
862 			if ((length = mbtowc(&wchar, &fbuf[i], MB_CUR_MAX))
863 			    <= 0 || !iswprint(wchar)) {
864 				(void) printf(gettext(" with garbage\n"));
865 				return;
866 			}
867 			i = i + length;
868 		}
869 	}
870 	(void) printf("\n");
871 }
872 
873 static int
874 troffint(char *bp, int n)
875 {
876 	int k;
877 
878 	i = 0;
879 	for (k = 0; k < 6; k++) {
880 		if (lookup(troff) == 0)
881 			return (0);
882 		if (lookup(troff) == 0)
883 			return (0);
884 		while (i < n && bp[i] != '\n')
885 			i++;
886 		if (i++ >= n)
887 			return (0);
888 	}
889 	return (1);
890 }
891 
892 /*
893  * Determine if the passed descriptor describes an ELF file.
894  * If so, return the Elf handle.
895  */
896 static Elf *
897 is_elf_file(int elffd)
898 {
899 	Elf *elf;
900 
901 	elf = elf_begin(elffd, ELF_C_READ, (Elf *)0);
902 	switch (elf_kind(elf)) {
903 	case ELF_K_ELF:
904 		break;
905 	default:
906 		(void) elf_end(elf);
907 		elf = NULL;
908 		break;
909 	}
910 	return (elf);
911 }
912 
913 static void
914 ar_coff_or_aout(int elffd)
915 {
916 	Elf *elf;
917 
918 	/*
919 	 * Get the files elf descriptor and process it as an elf or
920 	 * a.out (4.x) file.
921 	 */
922 
923 	elf = elf_begin(elffd, ELF_C_READ, (Elf *)0);
924 	switch (elf_kind(elf)) {
925 		case ELF_K_AR :
926 			(void) printf(gettext(", not a dynamic executable "
927 			    "or shared object"));
928 			break;
929 		case ELF_K_COFF:
930 			(void) printf(gettext(", unsupported or unknown "
931 			    "file type"));
932 			break;
933 		default:
934 			/*
935 			 * This is either an unknown file or an aout format
936 			 * At this time, we don't print dynamic/stripped
937 			 * info. on a.out or non-Elf binaries.
938 			 */
939 			break;
940 	}
941 	(void) elf_end(elf);
942 }
943 
944 
945 static void
946 print_elf_type(Elf *elf, GElf_Ehdr *ehdr, int format)
947 {
948 	switch (ehdr->e_type) {
949 	case ET_NONE:
950 		(void) printf(" %s", gettext("unknown type"));
951 		break;
952 	case ET_REL:
953 		(void) printf(" %s", gettext("relocatable"));
954 		break;
955 	case ET_EXEC:
956 		(void) printf(" %s", gettext("executable"));
957 		break;
958 	case ET_DYN:
959 		(void) printf(" %s", gettext("dynamic lib"));
960 		break;
961 	case ET_CORE:
962 		if (old_core(elf, ehdr, format))
963 			(void) printf(" %s", gettext("pre-2.6 core file"));
964 		else
965 			(void) printf(" %s", gettext("core file"));
966 		break;
967 	default:
968 		break;
969 	}
970 }
971 
972 static void
973 print_elf_machine(int machine)
974 {
975 	switch (machine) {
976 	case EM_NONE:
977 		(void) printf(" %s", gettext("unknown machine"));
978 		break;
979 	case EM_M32:
980 		(void) printf(" %s", gettext("WE32100"));
981 		break;
982 	case EM_SPARC:
983 		(void) printf(" %s", gettext("SPARC"));
984 		break;
985 	case EM_386:
986 		(void) printf(" %s", gettext("80386"));
987 		break;
988 	case EM_68K:
989 		(void) printf(" %s", gettext("M68000"));
990 		break;
991 	case EM_88K:
992 		(void) printf(" %s", gettext("M88000"));
993 		break;
994 	case EM_486:
995 		(void) printf(" %s", gettext("80486"));
996 		break;
997 	case EM_860:
998 		(void) printf(" %s", gettext("i860"));
999 		break;
1000 	case EM_MIPS:
1001 		(void) printf(" %s", gettext("MIPS RS3000 Big-Endian"));
1002 		break;
1003 	case EM_MIPS_RS3_LE:
1004 		(void) printf(" %s", gettext("MIPS RS3000 Little-Endian"));
1005 		break;
1006 	case EM_RS6000:
1007 		(void) printf(" %s", gettext("MIPS RS6000"));
1008 		break;
1009 	case EM_PA_RISC:
1010 		(void) printf(" %s", gettext("PA-RISC"));
1011 		break;
1012 	case EM_nCUBE:
1013 		(void) printf(" %s", gettext("nCUBE"));
1014 		break;
1015 	case EM_VPP500:
1016 		(void) printf(" %s", gettext("VPP500"));
1017 		break;
1018 	case EM_SPARC32PLUS:
1019 		(void) printf(" %s", gettext("SPARC32PLUS"));
1020 		break;
1021 	case EM_PPC:
1022 		(void) printf(" %s", gettext("PowerPC"));
1023 		break;
1024 	case EM_SPARCV9:
1025 		(void) printf(" %s", gettext("SPARCV9"));
1026 		break;
1027 	case EM_IA_64:
1028 		(void) printf(" %s", gettext("IA64"));
1029 		break;
1030 	case EM_AMD64:
1031 		(void) printf(" %s", gettext("AMD64"));
1032 		break;
1033 	default:
1034 		break;
1035 	}
1036 }
1037 
1038 static void
1039 print_elf_datatype(int datatype)
1040 {
1041 	switch (datatype) {
1042 	case ELFDATA2LSB:
1043 		(void) printf(" %s", gettext("LSB"));
1044 		break;
1045 	case ELFDATA2MSB:
1046 		(void) printf(" %s", gettext("MSB"));
1047 		break;
1048 	default:
1049 		break;
1050 	}
1051 }
1052 
1053 static void
1054 print_elf_class(int class)
1055 {
1056 	switch (class) {
1057 	case ELFCLASS32:
1058 		(void) printf(" %s", gettext("32-bit"));
1059 		break;
1060 	case ELFCLASS64:
1061 		(void) printf(" %s", gettext("64-bit"));
1062 		break;
1063 	default:
1064 		break;
1065 	}
1066 }
1067 
1068 static void
1069 print_elf_flags(int machine, unsigned int flags)
1070 {
1071 	switch (machine) {
1072 	case EM_M32:
1073 		if (flags & EF_M32_MAU)
1074 			(void) printf("%s", gettext(", MAU Required"));
1075 		break;
1076 	case EM_SPARCV9:
1077 		if (flags & EF_SPARC_EXT_MASK) {
1078 			if (flags & EF_SPARC_SUN_US3) {
1079 				(void) printf("%s", gettext(
1080 				    ", UltraSPARC3 Extensions Required"));
1081 			} else if (flags & EF_SPARC_SUN_US1) {
1082 				(void) printf("%s", gettext(
1083 				    ", UltraSPARC1 Extensions Required"));
1084 			}
1085 			if (flags & EF_SPARC_HAL_R1)
1086 				(void) printf("%s", gettext(
1087 				    ", HaL R1 Extensions Required"));
1088 		}
1089 		break;
1090 	case EM_SPARC32PLUS:
1091 		if (flags & EF_SPARC_32PLUS)
1092 			(void) printf("%s", gettext(", V8+ Required"));
1093 		if (flags & EF_SPARC_SUN_US3) {
1094 			(void) printf("%s",
1095 				gettext(", UltraSPARC3 Extensions Required"));
1096 		} else if (flags & EF_SPARC_SUN_US1) {
1097 			(void) printf("%s",
1098 				gettext(", UltraSPARC1 Extensions Required"));
1099 		}
1100 		if (flags & EF_SPARC_HAL_R1)
1101 			(void) printf("%s",
1102 				gettext(", HaL R1 Extensions Required"));
1103 		break;
1104 	default:
1105 		break;
1106 	}
1107 }
1108 
1109 static int
1110 print_cap(Elf *elf, GElf_Ehdr *ehdr)
1111 {
1112 	Elf_Scn	*scn = 0;
1113 
1114 	/*
1115 	 * Traverse the files sections to see if any software/hardware
1116 	 * capabilities are available.
1117 	 */
1118 	while ((scn = elf_nextscn(elf, scn)) != 0) {
1119 		GElf_Word	ndx, capn;
1120 		GElf_Shdr	shdr;
1121 		Elf_Data	*data;
1122 
1123 		if (gelf_getshdr(scn, &shdr) == 0) {
1124 			(void) fprintf(stderr,
1125 			    gettext("can't read ELF section header\n"));
1126 			return (1);
1127 		}
1128 		if (shdr.sh_type != SHT_SUNW_cap)
1129 			continue;
1130 
1131 		/*
1132 		 * Get the data associated with the .cap section.
1133 		 */
1134 		if ((data = elf_getdata(scn, 0)) == 0) {
1135 			(void) fprintf(stderr,
1136 				gettext("can't read ELF section data\n"));
1137 			return (1);
1138 		}
1139 
1140 		capn = (GElf_Word)(shdr.sh_size / shdr.sh_entsize);
1141 		for (ndx = 0; ndx < capn; ndx++) {
1142 			char		str[100];
1143 			GElf_Cap	cap;
1144 
1145 			if (gelf_getcap(data, ndx, &cap) == NULL) {
1146 				(void) fprintf(stderr,
1147 				    gettext("can't read capabilities data\n"));
1148 				return (1);
1149 			}
1150 			if (cap.c_tag != CA_SUNW_NULL) {
1151 				(void) cap_val2str(cap.c_tag, cap.c_un.c_val,
1152 				    str, sizeof (str), 0, ehdr->e_machine);
1153 				(void) printf(" [%s]", str);
1154 			}
1155 		}
1156 	}
1157 	return (0);
1158 }
1159 
1160 static int
1161 elf_check(Elf *elf)
1162 {
1163 	GElf_Ehdr	ehdr;
1164 	GElf_Phdr	phdr;
1165 	int		dynamic, cnt;
1166 	char	*ident;
1167 	size_t	size;
1168 
1169 	/*
1170 	 * verify information in file header
1171 	 */
1172 	if (gelf_getehdr(elf, &ehdr) == (GElf_Ehdr *)0) {
1173 		(void) fprintf(stderr, gettext("can't read ELF header\n"));
1174 		return (1);
1175 	}
1176 	ident = elf_getident(elf, &size);
1177 	(void) printf("%s", gettext("ELF"));
1178 	print_elf_class(ident[EI_CLASS]);
1179 	print_elf_datatype(ident[EI_DATA]);
1180 	print_elf_type(elf, &ehdr, ident[EI_DATA]);
1181 	print_elf_machine(ehdr.e_machine);
1182 	if (ehdr.e_version == 1)
1183 		(void) printf(" %s %d",
1184 		    gettext("Version"), (int)ehdr.e_version);
1185 	print_elf_flags(ehdr.e_machine, ehdr.e_flags);
1186 
1187 	if (core(elf, &ehdr, ident[EI_DATA]))	/* check for core file */
1188 		return (0);
1189 
1190 	if (print_cap(elf, &ehdr))
1191 		return (1);
1192 
1193 	/*
1194 	 * check type
1195 	 */
1196 	if ((ehdr.e_type != ET_EXEC) && (ehdr.e_type != ET_DYN))
1197 		return (1);
1198 
1199 	/*
1200 	 * read program header and check for dynamic section
1201 	 */
1202 	if (ehdr.e_phnum == 0) {
1203 		(void) fprintf(stderr, gettext("can't read program header\n"));
1204 		return (1);
1205 	}
1206 
1207 	for (dynamic = 0, cnt = 0; cnt < (int)ehdr.e_phnum; cnt++) {
1208 		if (gelf_getphdr(elf, cnt, &phdr) == NULL) {
1209 			(void) fprintf(stderr,
1210 				gettext("can't read program header\n"));
1211 			return (1);
1212 		}
1213 		if (phdr.p_type == PT_DYNAMIC) {
1214 			dynamic = 1;
1215 			break;
1216 		}
1217 	}
1218 	if (dynamic)
1219 		(void) printf(gettext(", dynamically linked"));
1220 	else
1221 		(void) printf(gettext(", statically linked"));
1222 
1223 	is_stripped(elf);
1224 	return (0);
1225 }
1226 
1227 /*
1228  * is_stripped prints information on whether the executable has
1229  * been stripped.
1230  */
1231 static void
1232 is_stripped(Elf *elf)
1233 {
1234 	GElf_Shdr	shdr;
1235 	GElf_Ehdr	ehdr;
1236 	Elf_Scn		*scn, *nextscn;
1237 	char		*section_name;
1238 	int		symtab = 0;
1239 	int		debuginfo = 0;
1240 
1241 
1242 	if (gelf_getehdr(elf, &ehdr) == NULL) {
1243 		return;
1244 	}
1245 
1246 	/*
1247 	 * Definition time:
1248 	 *	- "not stripped" means that an executable file
1249 	 *	contains a Symbol Table (.symtab)
1250 	 *	- "stripped" means that an executable file
1251 	 *	does not contain a Symbol Table.
1252 	 * When strip -l or strip -x is run, it strips the
1253 	 * debugging information (.line section name (strip -l),
1254 	 * .line, .debug*, .stabs*, .dwarf* section names
1255 	 * and SHT_SUNW_DEBUGSTR and SHT_SUNW_DEBUG
1256 	 * section types (strip -x), however the Symbol
1257 	 * Table will still be present.
1258 	 * Therefore, if
1259 	 *	- No Symbol Table present, then report
1260 	 *		"stripped"
1261 	 *	- Symbol Table present with debugging
1262 	 *	information (line number or debug section names,
1263 	 *	or SHT_SUNW_DEBUGSTR or SHT_SUNW_DEBUG section
1264 	 *	types) then report:
1265 	 *		"not stripped"
1266 	 *	- Symbol Table present with no debugging
1267 	 *	information (line number or debug section names,
1268 	 *	or SHT_SUNW_DEBUGSTR or SHT_SUNW_DEBUG section
1269 	 *	types) then report:
1270 	 *		"not stripped, no debugging information
1271 	 *		available"
1272 	 */
1273 	scn = NULL;
1274 	while ((nextscn = elf_nextscn(elf, scn)) != NULL) {
1275 		if (symtab && debuginfo) {
1276 			break;
1277 		}
1278 
1279 		scn = nextscn;
1280 		if (gelf_getshdr(scn, &shdr) == NULL) {
1281 			continue;
1282 		}
1283 
1284 		if (!symtab && (shdr.sh_type == SHT_SYMTAB)) {
1285 			symtab++;
1286 			continue;
1287 		}
1288 
1289 		if (!debuginfo &&
1290 		    ((shdr.sh_type == SHT_SUNW_DEBUG) ||
1291 		    (shdr.sh_type == SHT_SUNW_DEBUGSTR) ||
1292 		    (((section_name = elf_strptr(elf, ehdr.e_shstrndx,
1293 		    (size_t)shdr.sh_name)) != NULL) &&
1294 		    (is_in_list(debug_sections, section_name))))) {
1295 			debuginfo++;
1296 		}
1297 	}
1298 
1299 	/*
1300 	 * Now that we've scanned all sections, print out the appropriate
1301 	 * diagnostic.
1302 	 */
1303 	if (symtab) {
1304 		(void) printf(gettext(", not stripped"));
1305 		if (!debuginfo) {
1306 			(void) printf(gettext(
1307 			    ", no debugging information available"));
1308 		}
1309 	} else {
1310 		(void) printf(gettext(", stripped"));
1311 	}
1312 }
1313 
1314 /*
1315  * lookup -
1316  * Attempts to match one of the strings from a list, 'tab',
1317  * with what is in the file, starting at the current index position 'i'.
1318  * Looks past any initial whitespace and expects whitespace or other
1319  * delimiting characters to follow the matched string.
1320  * A match identifies the file as being 'assembler', 'fortran', 'c', etc.
1321  * Returns 1 for a successful match, 0 otherwise.
1322  */
1323 static int
1324 lookup(char **tab)
1325 {
1326 	register char	r;
1327 	register int	k, j, l;
1328 
1329 	while (fbuf[i] == ' ' || fbuf[i] == '\t' || fbuf[i] == '\n')
1330 		i++;
1331 	for (j = 0; tab[j] != 0; j++) {
1332 		l = 0;
1333 		for (k = i; ((r = tab[j][l++]) == fbuf[k] && r != '\0'); k++);
1334 		if (r == '\0')
1335 			if (fbuf[k] == ' ' || fbuf[k] == '\n' ||
1336 			    fbuf[k] == '\t' || fbuf[k] == '{' ||
1337 			    fbuf[k] == '/') {
1338 				i = k;
1339 				return (1);
1340 			}
1341 	}
1342 	return (0);
1343 }
1344 
1345 /*
1346  * ccom -
1347  * Increments the current index 'i' into the file buffer 'fbuf' past any
1348  * whitespace lines and C-style comments found, starting at the current
1349  * position of 'i'.  Returns 1 as long as we don't increment i past the
1350  * size of fbuf (fbsz).  Otherwise, returns 0.
1351  */
1352 
1353 static int
1354 ccom(void)
1355 {
1356 	register char	cc;
1357 	int		len;
1358 
1359 	while ((cc = fbuf[i]) == ' ' || cc == '\t' || cc == '\n')
1360 		if (i++ >= fbsz)
1361 			return (0);
1362 	if (fbuf[i] == '/' && fbuf[i+1] == '*') {
1363 		i += 2;
1364 		while (fbuf[i] != '*' || fbuf[i+1] != '/') {
1365 			if (fbuf[i] == '\\')
1366 				i++;
1367 			if ((len = mblen(&fbuf[i], MB_CUR_MAX)) <= 0)
1368 				len = 1;
1369 			i += len;
1370 			if (i >= fbsz)
1371 				return (0);
1372 		}
1373 		if ((i += 2) >= fbsz)
1374 			return (0);
1375 	}
1376 	if (fbuf[i] == '\n')
1377 		if (ccom() == 0)
1378 			return (0);
1379 	return (1);
1380 }
1381 
1382 /*
1383  * ascom -
1384  * Increments the current index 'i' into the file buffer 'fbuf' past
1385  * consecutive assembler program comment lines starting with ASCOMCHAR,
1386  * starting at the current position of 'i'.
1387  * Returns 1 as long as we don't increment i past the
1388  * size of fbuf (fbsz).  Otherwise returns 0.
1389  */
1390 
1391 static int
1392 ascom(void)
1393 {
1394 	while (fbuf[i] == ASCOMCHAR) {
1395 		i++;
1396 		while (fbuf[i++] != '\n')
1397 			if (i >= fbsz)
1398 				return (0);
1399 		while (fbuf[i] == '\n')
1400 			if (i++ >= fbsz)
1401 				return (0);
1402 	}
1403 	return (1);
1404 }
1405 
1406 static int
1407 sccs(void)
1408 {				/* look for "1hddddd" where d is a digit */
1409 	register int j;
1410 
1411 	if (fbuf[0] == 1 && fbuf[1] == 'h') {
1412 		for (j = 2; j <= 6; j++) {
1413 			if (isdigit(fbuf[j]))
1414 				continue;
1415 			else
1416 				return (0);
1417 		}
1418 	} else {
1419 		return (0);
1420 	}
1421 	return (1);
1422 }
1423 
1424 static int
1425 english(char *bp, int n)
1426 {
1427 #define	NASC 128		/* number of ascii char ?? */
1428 	register int	j, vow, freq, rare, len;
1429 	register int	badpun = 0, punct = 0;
1430 	int	ct[NASC];
1431 
1432 	if (n < 50)
1433 		return (0); /* no point in statistics on squibs */
1434 	for (j = 0; j < NASC; j++)
1435 		ct[j] = 0;
1436 	for (j = 0; j < n; j += len) {
1437 		if ((unsigned char)bp[j] < NASC)
1438 			ct[bp[j]|040]++;
1439 		switch (bp[j]) {
1440 		case '.':
1441 		case ',':
1442 		case ')':
1443 		case '%':
1444 		case ';':
1445 		case ':':
1446 		case '?':
1447 			punct++;
1448 			if (j < n-1 && bp[j+1] != ' ' && bp[j+1] != '\n')
1449 				badpun++;
1450 		}
1451 		if ((len = mblen(&bp[j], MB_CUR_MAX)) <= 0)
1452 			len = 1;
1453 	}
1454 	if (badpun*5 > punct)
1455 		return (0);
1456 	vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
1457 	freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
1458 	rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
1459 	if (2*ct[';'] > ct['e'])
1460 		return (0);
1461 	if ((ct['>'] + ct['<'] + ct['/']) > ct['e'])
1462 		return (0);	/* shell file test */
1463 	return (vow * 5 >= n - ct[' '] && freq >= 10 * rare);
1464 }
1465 
1466 /*
1467  * Convert a word from an elf file to native format.
1468  * This is needed because there's no elf routine to
1469  * get and decode a Note section header.
1470  */
1471 static void
1472 convert_gelf_word(Elf *elf, GElf_Word *data, int version, int format)
1473 {
1474 	Elf_Data src, dst;
1475 
1476 	dst.d_buf = data;
1477 	dst.d_version = version;
1478 	dst.d_size = sizeof (GElf_Word);
1479 	dst.d_type = ELF_T_WORD;
1480 	src.d_buf = data;
1481 	src.d_version = version;
1482 	src.d_size = sizeof (GElf_Word);
1483 	src.d_type = ELF_T_WORD;
1484 	(void) gelf_xlatetom(elf, &dst, &src, format);
1485 }
1486 
1487 static void
1488 convert_gelf_nhdr(Elf *elf, GElf_Nhdr *nhdr, GElf_Word version, int format)
1489 {
1490 	convert_gelf_word(elf, &nhdr->n_namesz, version, format);
1491 	convert_gelf_word(elf, &nhdr->n_descsz, version, format);
1492 	convert_gelf_word(elf, &nhdr->n_type, version, format);
1493 }
1494 
1495 /*
1496  * Return true if it is an old (pre-restructured /proc) core file.
1497  */
1498 static int
1499 old_core(Elf *elf, GElf_Ehdr *ehdr, int format)
1500 {
1501 	register int inx;
1502 	GElf_Phdr phdr;
1503 	GElf_Phdr nphdr;
1504 	GElf_Nhdr nhdr;
1505 	off_t offset;
1506 
1507 	if (ehdr->e_type != ET_CORE)
1508 		return (0);
1509 	for (inx = 0; inx < (int)ehdr->e_phnum; inx++) {
1510 		if (gelf_getphdr(elf, inx, &phdr) == NULL) {
1511 			return (0);
1512 		}
1513 		if (phdr.p_type == PT_NOTE) {
1514 			/*
1515 			 * If the next segment is also a note, use it instead.
1516 			 */
1517 			if (gelf_getphdr(elf, inx+1, &nphdr) == NULL) {
1518 				return (0);
1519 			}
1520 			if (nphdr.p_type == PT_NOTE)
1521 				phdr = nphdr;
1522 			offset = (off_t)phdr.p_offset;
1523 			(void) pread(ifd, &nhdr, sizeof (GElf_Nhdr), offset);
1524 			convert_gelf_nhdr(elf, &nhdr, ehdr->e_version, format);
1525 			/*
1526 			 * Old core files have type NT_PRPSINFO.
1527 			 */
1528 			if (nhdr.n_type == NT_PRPSINFO)
1529 				return (1);
1530 			return (0);
1531 		}
1532 	}
1533 	return (0);
1534 }
1535 
1536 /*
1537  * If it's a core file, print out the name of the file that dumped core.
1538  */
1539 static int
1540 core(Elf *elf, GElf_Ehdr *ehdr, int format)
1541 {
1542 	register int inx;
1543 	char *psinfo;
1544 	GElf_Phdr phdr;
1545 	GElf_Phdr nphdr;
1546 	GElf_Nhdr nhdr;
1547 	off_t offset;
1548 
1549 	if (ehdr->e_type != ET_CORE)
1550 		return (0);
1551 	for (inx = 0; inx < (int)ehdr->e_phnum; inx++) {
1552 		if (gelf_getphdr(elf, inx, &phdr) == NULL) {
1553 			(void) fprintf(stderr,
1554 				gettext("can't read program header\n"));
1555 			return (0);
1556 		}
1557 		if (phdr.p_type == PT_NOTE) {
1558 			char *fname;
1559 			size_t size;
1560 			/*
1561 			 * If the next segment is also a note, use it instead.
1562 			 */
1563 			if (gelf_getphdr(elf, inx+1, &nphdr) == NULL) {
1564 				(void) fprintf(stderr,
1565 				    gettext("can't read program header\n"));
1566 				return (0);
1567 			}
1568 			if (nphdr.p_type == PT_NOTE)
1569 				phdr = nphdr;
1570 			offset = (off_t)phdr.p_offset;
1571 			(void) pread(ifd, &nhdr, sizeof (GElf_Nhdr), offset);
1572 			convert_gelf_nhdr(elf, &nhdr, ehdr->e_version, format);
1573 			/*
1574 			 * Note: the ABI states that n_namesz must
1575 			 * be rounded up to a 4 byte boundary.
1576 			 */
1577 			offset += sizeof (GElf_Nhdr) +
1578 			    ((nhdr.n_namesz + 0x03) & ~0x3);
1579 			size = nhdr.n_descsz;
1580 			psinfo = malloc(size);
1581 			(void) pread(ifd, psinfo, size, offset);
1582 			/*
1583 			 * We want to print the string contained
1584 			 * in psinfo->pr_fname[], where 'psinfo'
1585 			 * is either an old NT_PRPSINFO structure
1586 			 * or a new NT_PSINFO structure.
1587 			 *
1588 			 * Old core files have only type NT_PRPSINFO.
1589 			 * New core files have type NT_PSINFO.
1590 			 *
1591 			 * These structures are also different by
1592 			 * virtue of being contained in a core file
1593 			 * of either 32-bit or 64-bit type.
1594 			 *
1595 			 * To further complicate matters, we ourself
1596 			 * might be compiled either 32-bit or 64-bit.
1597 			 *
1598 			 * For these reason, we just *know* the offsets of
1599 			 * pr_fname[] into the four different structures
1600 			 * here, regardless of how we are compiled.
1601 			 */
1602 			if (gelf_getclass(elf) == ELFCLASS32) {
1603 				/* 32-bit core file, 32-bit structures */
1604 				if (nhdr.n_type == NT_PSINFO)
1605 					fname = psinfo + 88;
1606 				else	/* old: NT_PRPSINFO */
1607 					fname = psinfo + 84;
1608 			} else if (gelf_getclass(elf) == ELFCLASS64) {
1609 				/* 64-bit core file, 64-bit structures */
1610 				if (nhdr.n_type == NT_PSINFO)
1611 					fname = psinfo + 136;
1612 				else	/* old: NT_PRPSINFO */
1613 					fname = psinfo + 120;
1614 			} else {
1615 				free(psinfo);
1616 				break;
1617 			}
1618 			(void) printf(gettext(", from '%s'"), fname);
1619 			free(psinfo);
1620 			break;
1621 		}
1622 	}
1623 	return (1);
1624 }
1625 
1626 static int
1627 shellscript(char buf[], struct stat64 *sb)
1628 {
1629 	char *tp, *cp, *xp, *up, *gp;
1630 
1631 	cp = strchr(buf, '\n');
1632 	if (cp == NULL || cp - fbuf > fbsz)
1633 		return (0);
1634 	for (tp = buf; tp != cp && isspace((unsigned char)*tp); tp++)
1635 		if (!isascii(*tp))
1636 			return (0);
1637 	for (xp = tp; tp != cp && !isspace((unsigned char)*tp); tp++)
1638 		if (!isascii(*tp))
1639 			return (0);
1640 	if (tp == xp)
1641 		return (0);
1642 	if (sb->st_mode & S_ISUID)
1643 		up = gettext("set-uid ");
1644 	else
1645 		up = "";
1646 
1647 	if (sb->st_mode & S_ISGID)
1648 		gp = gettext("set-gid ");
1649 	else
1650 		gp = "";
1651 
1652 	if (strncmp(xp, "/bin/sh", tp - xp) == 0)
1653 		xp = gettext("shell");
1654 	else if (strncmp(xp, "/bin/csh", tp - xp) == 0)
1655 		xp = gettext("c-shell");
1656 	else if (strncmp(xp, "/usr/sbin/dtrace", tp - xp) == 0)
1657 		xp = gettext("DTrace");
1658 	else
1659 		*tp = '\0';
1660 	/*
1661 	 * TRANSLATION_NOTE
1662 	 * This message is printed by file command for shell scripts.
1663 	 * The first %s is for the translation for "set-uid " (if the script
1664 	 *   has the set-uid bit set), or is for an empty string (if the
1665 	 *   script does not have the set-uid bit set).
1666 	 * Similarly, the second %s is for the translation for "set-gid ",
1667 	 *   or is for an empty string.
1668 	 * The third %s is for the translation for either: "shell", "c-shell",
1669 	 *   or "DTrace", or is for the pathname of the program the script
1670 	 *   executes.
1671 	 */
1672 	(void) printf(gettext("%s%sexecutable %s script\n"), up, gp, xp);
1673 	return (1);
1674 }
1675 
1676 static int
1677 get_door_target(char *file, char *buf, size_t bufsize)
1678 {
1679 	int fd;
1680 	door_info_t di;
1681 	psinfo_t psinfo;
1682 
1683 	if ((fd = open64(file, O_RDONLY)) < 0 ||
1684 	    door_info(fd, &di) != 0) {
1685 		if (fd >= 0)
1686 			(void) close(fd);
1687 		return (-1);
1688 	}
1689 	(void) close(fd);
1690 
1691 	(void) sprintf(buf, "/proc/%ld/psinfo", di.di_target);
1692 	if ((fd = open64(buf, O_RDONLY)) < 0 ||
1693 	    read(fd, &psinfo, sizeof (psinfo)) != sizeof (psinfo)) {
1694 		if (fd >= 0)
1695 			(void) close(fd);
1696 		return (-1);
1697 	}
1698 	(void) close(fd);
1699 
1700 	(void) snprintf(buf, bufsize, "%s[%ld]", psinfo.pr_fname, di.di_target);
1701 	return (0);
1702 }
1703 
1704 /*
1705  * ZIP file header information
1706  */
1707 #define	SIGSIZ		4
1708 #define	LOCSIG		"PK\003\004"
1709 #define	LOCHDRSIZ	30
1710 
1711 #define	CH(b, n)	(((unsigned char *)(b))[n])
1712 #define	SH(b, n)	(CH(b, n) | (CH(b, n+1) << 8))
1713 #define	LG(b, n)	(SH(b, n) | (SH(b, n+2) << 16))
1714 
1715 #define	LOCNAM(b)	(SH(b, 26))	/* filename size */
1716 #define	LOCEXT(b)	(SH(b, 28))	/* extra field size */
1717 
1718 #define	XFHSIZ		4		/* header id, data size */
1719 #define	XFHID(b)	(SH(b, 0))	/* extract field header id */
1720 #define	XFDATASIZ(b)	(SH(b, 2))	/* extract field data size */
1721 #define	XFJAVASIG	0xcafe		/* java executables */
1722 
1723 static int
1724 zipfile(char *fbuf, int fd)
1725 {
1726 	off_t xoff, xoff_end;
1727 
1728 	if (strncmp(fbuf, LOCSIG, SIGSIZ) != 0)
1729 		return (0);
1730 
1731 	xoff = LOCHDRSIZ + LOCNAM(fbuf);
1732 	xoff_end = xoff + LOCEXT(fbuf);
1733 
1734 	while (xoff < xoff_end) {
1735 		char xfhdr[XFHSIZ];
1736 
1737 		if (pread(fd, xfhdr, XFHSIZ, xoff) != XFHSIZ)
1738 			break;
1739 
1740 		if (XFHID(xfhdr) == XFJAVASIG) {
1741 			(void) printf("%s\n", gettext("java program"));
1742 			return (1);
1743 		}
1744 		xoff += sizeof (xfhdr) + XFDATASIZ(xfhdr);
1745 	}
1746 
1747 	/*
1748 	 * We could just print "ZIP archive" here.
1749 	 *
1750 	 * However, customers may be using their own entries in
1751 	 * /etc/magic to distinguish one kind of ZIP file from another, so
1752 	 * let's defer the printing of "ZIP archive" to there.
1753 	 */
1754 	return (0);
1755 }
1756 
1757 static int
1758 is_crash_dump(const char *buf, int fd)
1759 {
1760 	/* LINTED: pointer cast may result in improper alignment */
1761 	const dumphdr_t *dhp = (const dumphdr_t *)buf;
1762 
1763 	/*
1764 	 * The current DUMP_MAGIC string covers Solaris 7 and later releases.
1765 	 * The utsname struct is only present in dumphdr_t's with dump_version
1766 	 * greater than or equal to 9.
1767 	 */
1768 	if (dhp->dump_magic == DUMP_MAGIC) {
1769 		print_dumphdr(fd, dhp, return_uint32, NATIVE_ISA);
1770 
1771 	} else if (dhp->dump_magic == swap_uint32(DUMP_MAGIC)) {
1772 		print_dumphdr(fd, dhp, swap_uint32, OTHER_ISA);
1773 
1774 	} else if (dhp->dump_magic == OLD_DUMP_MAGIC ||
1775 	    dhp->dump_magic == swap_uint32(OLD_DUMP_MAGIC)) {
1776 		char *isa = (dhp->dump_magic == OLD_DUMP_MAGIC ?
1777 		    NATIVE_ISA : OTHER_ISA);
1778 		(void) printf(gettext("SunOS 32-bit %s crash dump\n"), isa);
1779 
1780 	} else {
1781 		return (0);
1782 	}
1783 
1784 	return (1);
1785 }
1786 
1787 static void
1788 print_dumphdr(const int fd, const dumphdr_t *dhp, uint32_t (*swap)(uint32_t),
1789     const char *isa)
1790 {
1791 	dumphdr_t dh;
1792 
1793 	/*
1794 	 * A dumphdr_t is bigger than FBSZ, so we have to manually read the
1795 	 * rest of it.
1796 	 */
1797 	if (swap(dhp->dump_version) > 8 && pread(fd, &dh, sizeof (dumphdr_t),
1798 	    (off_t)0) == sizeof (dumphdr_t)) {
1799 		(void) printf(gettext(
1800 		    "%s %s %s %u-bit %s crash dump from '%s'\n"),
1801 		    dh.dump_utsname.sysname, dh.dump_utsname.release,
1802 		    dh.dump_utsname.version, swap(dh.dump_wordsize), isa,
1803 		    dh.dump_utsname.nodename);
1804 	} else {
1805 		(void) printf(gettext("SunOS %u-bit %s crash dump\n"),
1806 		    swap(dhp->dump_wordsize), isa);
1807 	}
1808 }
1809 
1810 static void
1811 usage(void)
1812 {
1813 	(void) fprintf(stderr, gettext(
1814 		"usage: file [-dh] [-M mfile] [-m mfile] [-f ffile] file ...\n"
1815 		"       file [-dh] [-M mfile] [-m mfile] -f ffile\n"
1816 		"       file -i [-h] [-f ffile] file ...\n"
1817 		"       file -i [-h] -f ffile\n"
1818 		"       file -c [-d] [-M mfile] [-m mfile]\n"));
1819 	exit(2);
1820 }
1821 
1822 static uint32_t
1823 swap_uint32(uint32_t in)
1824 {
1825 	uint32_t out;
1826 
1827 	out = (in & 0x000000ff) << 24;
1828 	out |= (in & 0x0000ff00) << 8; /* >> 8 << 16 */
1829 	out |= (in & 0x00ff0000) >> 8; /* >> 16 << 8 */
1830 	out |= (in & 0xff000000) >> 24;
1831 
1832 	return (out);
1833 }
1834 
1835 static uint32_t
1836 return_uint32(uint32_t in)
1837 {
1838 	return (in);
1839 }
1840 
1841 /*
1842  * Check if str is in the string list str_list.
1843  */
1844 static int
1845 is_in_list(char *str_list[], char *str)
1846 {
1847 	int i;
1848 
1849 	/*
1850 	 * Only need to compare the strlen(str_list[i]) bytes.
1851 	 * That way .stab will match on .stab* sections, and
1852 	 * .debug will match on .debug* sections.
1853 	 */
1854 	for (i = 0; str_list[i] != NULL; i++) {
1855 		if (strncmp(str_list[i], str, strlen(str_list[i])) == 0) {
1856 			return (1);
1857 		}
1858 	}
1859 	return (0);
1860 }
1861 
1862 /*
1863  * default_magic -
1864  *	allocate space for and create the default magic file
1865  *	name string.
1866  */
1867 
1868 static void
1869 default_magic(void)
1870 {
1871 	const char *msg_locale = setlocale(LC_MESSAGES, NULL);
1872 	struct stat	statbuf;
1873 
1874 	if ((dfile = (char *)malloc(strlen(msg_locale) + 35)) == NULL) {
1875 		perror("file");
1876 		exit(2);
1877 	}
1878 	(void) snprintf(dfile, strlen(msg_locale) + 35,
1879 	    "/usr/lib/locale/%s/LC_MESSAGES/magic", msg_locale);
1880 	if (stat(dfile, &statbuf) != 0) {
1881 		(void) strcpy(dfile, "/etc/magic");
1882 	}
1883 }
1884 
1885 /*
1886  * add_to_mlist -
1887  *	Add the given magic_file filename string to the list of magic
1888  *	files (mlist).  This list of files will later be examined, and
1889  *	each magic file's entries will be added in order to
1890  *	the mtab table.
1891  *
1892  *	The first flag is set to 1 to add to the first list, mlist1.
1893  *	The first flag is set to 0 to add to the second list, mlist2.
1894  */
1895 
1896 static void
1897 add_to_mlist(char *magic_file, int first)
1898 {
1899 	char	**mlist;	/* ordered list of magic files */
1900 	size_t	mlist_sz;	/* number of pointers allocated  for mlist */
1901 	char	**mlistp;	/* next entry in mlist */
1902 	size_t mlistp_off;
1903 
1904 	if (first) {
1905 		mlist = mlist1;
1906 		mlist_sz = mlist1_sz;
1907 		mlistp = mlist1p;
1908 	} else {
1909 		mlist = mlist2;
1910 		mlist_sz = mlist2_sz;
1911 		mlistp = mlist2p;
1912 	}
1913 
1914 	if (mlist == NULL) {	/* initial mlist allocation */
1915 		if ((mlist = (char **)calloc(MLIST_SZ, sizeof (char *)))
1916 		    == NULL) {
1917 			perror("file");
1918 			exit(2);
1919 		}
1920 		mlist_sz = MLIST_SZ;
1921 		mlistp = mlist;
1922 	}
1923 	if ((mlistp - mlist) >= mlist_sz) {
1924 		mlistp_off = mlistp - mlist;
1925 		mlist_sz *= 2;
1926 		if ((mlist = (char **)realloc(mlist,
1927 		    mlist_sz * sizeof (char *))) == NULL) {
1928 			perror("file");
1929 			exit(2);
1930 		}
1931 		mlistp = mlist + mlistp_off;
1932 	}
1933 	/*
1934 	 * now allocate memory for and copy the
1935 	 * magic file name string
1936 	 */
1937 	if ((*mlistp = malloc(strlen(magic_file) + 1)) == NULL) {
1938 		perror("file");
1939 		exit(2);
1940 	}
1941 	(void) strlcpy(*mlistp, magic_file, strlen(magic_file) + 1);
1942 	mlistp++;
1943 
1944 	if (first) {
1945 		mlist1 = mlist;
1946 		mlist1_sz = mlist_sz;
1947 		mlist1p = mlistp;
1948 	} else {
1949 		mlist2 = mlist;
1950 		mlist2_sz = mlist_sz;
1951 		mlist2p = mlistp;
1952 	}
1953 }
1954 
1955 static void
1956 fd_cleanup(void)
1957 {
1958 	if (ifd != -1) {
1959 		(void) close(ifd);
1960 		ifd = -1;
1961 	}
1962 	if (elffd != -1) {
1963 		(void) close(elffd);
1964 		elffd = -1;
1965 	}
1966 }
1967