main.c revision 247212
1/*
2 * Copyright (c) 1980, 1986, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29
30#if 0
31#ifndef lint
32static const char copyright[] =
33"@(#) Copyright (c) 1980, 1986, 1993\n\
34	The Regents of the University of California.  All rights reserved.\n";
35#endif /* not lint */
36
37#ifndef lint
38static char sccsid[] = "@(#)main.c	8.6 (Berkeley) 5/14/95";
39#endif /* not lint */
40#endif
41#include <sys/cdefs.h>
42__FBSDID("$FreeBSD: head/sbin/fsck_ffs/main.c 247212 2013-02-24 06:44:29Z mckusick $");
43
44#include <sys/param.h>
45#include <sys/file.h>
46#include <sys/mount.h>
47#include <sys/resource.h>
48#include <sys/stat.h>
49#include <sys/sysctl.h>
50#include <sys/uio.h>
51#include <sys/disklabel.h>
52
53#include <ufs/ufs/dinode.h>
54#include <ufs/ffs/fs.h>
55
56#include <err.h>
57#include <errno.h>
58#include <fstab.h>
59#include <grp.h>
60#include <mntopts.h>
61#include <paths.h>
62#include <stdint.h>
63#include <string.h>
64#include <time.h>
65
66#include "fsck.h"
67
68static void usage(void) __dead2;
69static int argtoi(int flag, const char *req, const char *str, int base);
70static int checkfilesys(char *filesys);
71static int chkdoreload(struct statfs *mntp);
72static struct statfs *getmntpt(const char *);
73
74int
75main(int argc, char *argv[])
76{
77	int ch;
78	struct rlimit rlimit;
79	struct itimerval itimerval;
80	int ret = 0;
81
82	sync();
83	skipclean = 1;
84	inoopt = 0;
85	while ((ch = getopt(argc, argv, "b:Bc:CdEfFm:npry")) != -1) {
86		switch (ch) {
87		case 'b':
88			skipclean = 0;
89			bflag = argtoi('b', "number", optarg, 10);
90			printf("Alternate super block location: %d\n", bflag);
91			break;
92
93		case 'B':
94			bkgrdflag = 1;
95			break;
96
97		case 'c':
98			skipclean = 0;
99			cvtlevel = argtoi('c', "conversion level", optarg, 10);
100			if (cvtlevel < 3)
101				errx(EEXIT, "cannot do level %d conversion",
102				    cvtlevel);
103			break;
104
105		case 'd':
106			debug++;
107			break;
108
109		case 'E':
110			Eflag++;
111			break;
112
113		case 'f':
114			skipclean = 0;
115			break;
116
117		case 'F':
118			bkgrdcheck = 1;
119			break;
120
121		case 'm':
122			lfmode = argtoi('m', "mode", optarg, 8);
123			if (lfmode &~ 07777)
124				errx(EEXIT, "bad mode to -m: %o", lfmode);
125			printf("** lost+found creation mode %o\n", lfmode);
126			break;
127
128		case 'n':
129			nflag++;
130			yflag = 0;
131			break;
132
133		case 'p':
134			preen++;
135			/*FALLTHROUGH*/
136
137		case 'C':
138			ckclean++;
139			break;
140
141		case 'r':
142			inoopt++;
143			break;
144
145		case 'y':
146			yflag++;
147			nflag = 0;
148			break;
149
150		default:
151			usage();
152		}
153	}
154	argc -= optind;
155	argv += optind;
156
157	if (!argc)
158		usage();
159
160	if (signal(SIGINT, SIG_IGN) != SIG_IGN)
161		(void)signal(SIGINT, catch);
162	if (ckclean)
163		(void)signal(SIGQUIT, catchquit);
164	signal(SIGINFO, infohandler);
165	if (bkgrdflag) {
166		signal(SIGALRM, alarmhandler);
167		itimerval.it_interval.tv_sec = 5;
168		itimerval.it_interval.tv_usec = 0;
169		itimerval.it_value.tv_sec = 5;
170		itimerval.it_value.tv_usec = 0;
171		setitimer(ITIMER_REAL, &itimerval, NULL);
172	}
173	/*
174	 * Push up our allowed memory limit so we can cope
175	 * with huge file systems.
176	 */
177	if (getrlimit(RLIMIT_DATA, &rlimit) == 0) {
178		rlimit.rlim_cur = rlimit.rlim_max;
179		(void)setrlimit(RLIMIT_DATA, &rlimit);
180	}
181	while (argc-- > 0)
182		(void)checkfilesys(*argv++);
183
184	if (returntosingle)
185		ret = 2;
186	exit(ret);
187}
188
189static int
190argtoi(int flag, const char *req, const char *str, int base)
191{
192	char *cp;
193	int ret;
194
195	ret = (int)strtol(str, &cp, base);
196	if (cp == str || *cp)
197		errx(EEXIT, "-%c flag requires a %s", flag, req);
198	return (ret);
199}
200
201/*
202 * Check the specified file system.
203 */
204/* ARGSUSED */
205static int
206checkfilesys(char *filesys)
207{
208	ufs2_daddr_t n_ffree, n_bfree;
209	struct dups *dp;
210	struct statfs *mntp;
211	struct stat snapdir;
212	struct group *grp;
213	struct iovec *iov;
214	char errmsg[255];
215	int iovlen;
216	int cylno;
217	intmax_t blks, files;
218	size_t size;
219
220	iov = NULL;
221	iovlen = 0;
222	errmsg[0] = '\0';
223
224	cdevname = filesys;
225	if (debug && ckclean)
226		pwarn("starting\n");
227	/*
228	 * Make best effort to get the disk name. Check first to see
229	 * if it is listed among the mounted file systems. Failing that
230	 * check to see if it is listed in /etc/fstab.
231	 */
232	mntp = getmntpt(filesys);
233	if (mntp != NULL)
234		filesys = mntp->f_mntfromname;
235	else
236		filesys = blockcheck(filesys);
237	/*
238	 * If -F flag specified, check to see whether a background check
239	 * is possible and needed. If possible and needed, exit with
240	 * status zero. Otherwise exit with status non-zero. A non-zero
241	 * exit status will cause a foreground check to be run.
242	 */
243	sblock_init();
244	if (bkgrdcheck) {
245		if ((fsreadfd = open(filesys, O_RDONLY)) < 0 || readsb(0) == 0)
246			exit(3);	/* Cannot read superblock */
247		close(fsreadfd);
248		/* Earlier background failed or journaled */
249		if (sblock.fs_flags & (FS_NEEDSFSCK | FS_SUJ))
250			exit(4);
251		if ((sblock.fs_flags & FS_DOSOFTDEP) == 0)
252			exit(5);	/* Not running soft updates */
253		size = MIBSIZE;
254		if (sysctlnametomib("vfs.ffs.adjrefcnt", adjrefcnt, &size) < 0)
255			exit(6);	/* Lacks kernel support */
256		if ((mntp == NULL && sblock.fs_clean == 1) ||
257		    (mntp != NULL && (sblock.fs_flags & FS_UNCLEAN) == 0))
258			exit(7);	/* Filesystem clean, report it now */
259		exit(0);
260	}
261	if (ckclean && skipclean) {
262		/*
263		 * If file system is gjournaled, check it here.
264		 */
265		if ((fsreadfd = open(filesys, O_RDONLY)) < 0 || readsb(0) == 0)
266			exit(3);	/* Cannot read superblock */
267		close(fsreadfd);
268		if ((sblock.fs_flags & FS_GJOURNAL) != 0) {
269			//printf("GJournaled file system detected on %s.\n",
270			//    filesys);
271			if (sblock.fs_clean == 1) {
272				pwarn("FILE SYSTEM CLEAN; SKIPPING CHECKS\n");
273				exit(0);
274			}
275			if ((sblock.fs_flags & (FS_UNCLEAN | FS_NEEDSFSCK)) == 0) {
276				gjournal_check(filesys);
277				if (chkdoreload(mntp) == 0)
278					exit(0);
279				exit(4);
280			} else {
281				pfatal(
282			    "UNEXPECTED INCONSISTENCY, CANNOT RUN FAST FSCK\n");
283			}
284		}
285	}
286	/*
287	 * If we are to do a background check:
288	 *	Get the mount point information of the file system
289	 *	create snapshot file
290	 *	return created snapshot file
291	 *	if not found, clear bkgrdflag and proceed with normal fsck
292	 */
293	if (bkgrdflag) {
294		if (mntp == NULL) {
295			bkgrdflag = 0;
296			pfatal("NOT MOUNTED, CANNOT RUN IN BACKGROUND\n");
297		} else if ((mntp->f_flags & MNT_SOFTDEP) == 0) {
298			bkgrdflag = 0;
299			pfatal(
300			  "NOT USING SOFT UPDATES, CANNOT RUN IN BACKGROUND\n");
301		} else if ((mntp->f_flags & MNT_RDONLY) != 0) {
302			bkgrdflag = 0;
303			pfatal("MOUNTED READ-ONLY, CANNOT RUN IN BACKGROUND\n");
304		} else if ((fsreadfd = open(filesys, O_RDONLY)) >= 0) {
305			if (readsb(0) != 0) {
306				if (sblock.fs_flags & (FS_NEEDSFSCK | FS_SUJ)) {
307					bkgrdflag = 0;
308					pfatal(
309			"UNEXPECTED INCONSISTENCY, CANNOT RUN IN BACKGROUND\n");
310				}
311				if ((sblock.fs_flags & FS_UNCLEAN) == 0 &&
312				    skipclean && ckclean) {
313					/*
314					 * file system is clean;
315					 * skip snapshot and report it clean
316					 */
317					pwarn(
318					"FILE SYSTEM CLEAN; SKIPPING CHECKS\n");
319					goto clean;
320				}
321			}
322			close(fsreadfd);
323		}
324		if (bkgrdflag) {
325			snprintf(snapname, sizeof snapname, "%s/.snap",
326			    mntp->f_mntonname);
327			if (stat(snapname, &snapdir) < 0) {
328				if (errno != ENOENT) {
329					bkgrdflag = 0;
330					pfatal(
331	"CANNOT FIND SNAPSHOT DIRECTORY %s: %s, CANNOT RUN IN BACKGROUND\n",
332					    snapname, strerror(errno));
333				} else if ((grp = getgrnam("operator")) == 0 ||
334				    mkdir(snapname, 0770) < 0 ||
335				    chown(snapname, -1, grp->gr_gid) < 0 ||
336				    chmod(snapname, 0770) < 0) {
337					bkgrdflag = 0;
338					pfatal(
339	"CANNOT CREATE SNAPSHOT DIRECTORY %s: %s, CANNOT RUN IN BACKGROUND\n",
340					    snapname, strerror(errno));
341				}
342			} else if (!S_ISDIR(snapdir.st_mode)) {
343				bkgrdflag = 0;
344				pfatal(
345			"%s IS NOT A DIRECTORY, CANNOT RUN IN BACKGROUND\n",
346				    snapname);
347			}
348		}
349		if (bkgrdflag) {
350			snprintf(snapname, sizeof snapname,
351			    "%s/.snap/fsck_snapshot", mntp->f_mntonname);
352			build_iovec(&iov, &iovlen, "fstype", "ffs", 4);
353			build_iovec(&iov, &iovlen, "from", snapname,
354			    (size_t)-1);
355			build_iovec(&iov, &iovlen, "fspath", mntp->f_mntonname,
356			    (size_t)-1);
357			build_iovec(&iov, &iovlen, "errmsg", errmsg,
358			    sizeof(errmsg));
359			build_iovec(&iov, &iovlen, "update", NULL, 0);
360			build_iovec(&iov, &iovlen, "snapshot", NULL, 0);
361
362			while (nmount(iov, iovlen, mntp->f_flags) < 0) {
363				if (errno == EEXIST && unlink(snapname) == 0)
364					continue;
365				bkgrdflag = 0;
366				pfatal("CANNOT CREATE SNAPSHOT %s: %s %s\n",
367				    snapname, strerror(errno), errmsg);
368				break;
369			}
370			if (bkgrdflag != 0)
371				filesys = snapname;
372		}
373	}
374
375	switch (setup(filesys)) {
376	case 0:
377		if (preen)
378			pfatal("CAN'T CHECK FILE SYSTEM.");
379		return (0);
380	case -1:
381	clean:
382		pwarn("clean, %ld free ", (long)(sblock.fs_cstotal.cs_nffree +
383		    sblock.fs_frag * sblock.fs_cstotal.cs_nbfree));
384		printf("(%jd frags, %jd blocks, %.1f%% fragmentation)\n",
385		    (intmax_t)sblock.fs_cstotal.cs_nffree,
386		    (intmax_t)sblock.fs_cstotal.cs_nbfree,
387		    sblock.fs_cstotal.cs_nffree * 100.0 / sblock.fs_dsize);
388		return (0);
389	}
390	/*
391	 * Determine if we can and should do journal recovery.
392	 */
393	if ((sblock.fs_flags & FS_SUJ) == FS_SUJ) {
394		if ((sblock.fs_flags & FS_NEEDSFSCK) != FS_NEEDSFSCK && skipclean) {
395			if (preen || reply("USE JOURNAL")) {
396				if (suj_check(filesys) == 0) {
397					printf("\n***** FILE SYSTEM MARKED CLEAN *****\n");
398					if (chkdoreload(mntp) == 0)
399						exit(0);
400					exit(4);
401				}
402			}
403			printf("** Skipping journal, falling through to full fsck\n\n");
404		}
405		/*
406		 * Write the superblock so we don't try to recover the
407		 * journal on another pass.
408		 */
409		sblock.fs_mtime = time(NULL);
410		sbdirty();
411	}
412
413	/*
414	 * Cleared if any questions answered no. Used to decide if
415	 * the superblock should be marked clean.
416	 */
417	resolved = 1;
418	/*
419	 * 1: scan inodes tallying blocks used
420	 */
421	if (preen == 0) {
422		printf("** Last Mounted on %s\n", sblock.fs_fsmnt);
423		if (mntp != NULL && mntp->f_flags & MNT_ROOTFS)
424			printf("** Root file system\n");
425		printf("** Phase 1 - Check Blocks and Sizes\n");
426	}
427	clock_gettime(CLOCK_REALTIME_PRECISE, &startprog);
428	pass1();
429	IOstats("Pass1");
430
431	/*
432	 * 1b: locate first references to duplicates, if any
433	 */
434	if (duplist) {
435		if (preen || usedsoftdep)
436			pfatal("INTERNAL ERROR: dups with %s%s%s",
437			    preen ? "-p" : "",
438			    (preen && usedsoftdep) ? " and " : "",
439			    usedsoftdep ? "softupdates" : "");
440		printf("** Phase 1b - Rescan For More DUPS\n");
441		pass1b();
442		IOstats("Pass1b");
443	}
444
445	/*
446	 * 2: traverse directories from root to mark all connected directories
447	 */
448	if (preen == 0)
449		printf("** Phase 2 - Check Pathnames\n");
450	pass2();
451	IOstats("Pass2");
452
453	/*
454	 * 3: scan inodes looking for disconnected directories
455	 */
456	if (preen == 0)
457		printf("** Phase 3 - Check Connectivity\n");
458	pass3();
459	IOstats("Pass3");
460
461	/*
462	 * 4: scan inodes looking for disconnected files; check reference counts
463	 */
464	if (preen == 0)
465		printf("** Phase 4 - Check Reference Counts\n");
466	pass4();
467	IOstats("Pass4");
468
469	/*
470	 * 5: check and repair resource counts in cylinder groups
471	 */
472	if (preen == 0)
473		printf("** Phase 5 - Check Cyl groups\n");
474	pass5();
475	IOstats("Pass5");
476
477	/*
478	 * print out summary statistics
479	 */
480	n_ffree = sblock.fs_cstotal.cs_nffree;
481	n_bfree = sblock.fs_cstotal.cs_nbfree;
482	files = maxino - ROOTINO - sblock.fs_cstotal.cs_nifree - n_files;
483	blks = n_blks +
484	    sblock.fs_ncg * (cgdmin(&sblock, 0) - cgsblock(&sblock, 0));
485	blks += cgsblock(&sblock, 0) - cgbase(&sblock, 0);
486	blks += howmany(sblock.fs_cssize, sblock.fs_fsize);
487	blks = maxfsblock - (n_ffree + sblock.fs_frag * n_bfree) - blks;
488	if (bkgrdflag && (files > 0 || blks > 0)) {
489		countdirs = sblock.fs_cstotal.cs_ndir - countdirs;
490		pwarn("Reclaimed: %ld directories, %jd files, %jd fragments\n",
491		    countdirs, files - countdirs, blks);
492	}
493	pwarn("%ld files, %jd used, %ju free ",
494	    (long)n_files, (intmax_t)n_blks,
495	    (uintmax_t)n_ffree + sblock.fs_frag * n_bfree);
496	printf("(%ju frags, %ju blocks, %.1f%% fragmentation)\n",
497	    (uintmax_t)n_ffree, (uintmax_t)n_bfree,
498	    n_ffree * 100.0 / sblock.fs_dsize);
499	if (debug) {
500		if (files < 0)
501			printf("%jd inodes missing\n", -files);
502		if (blks < 0)
503			printf("%jd blocks missing\n", -blks);
504		if (duplist != NULL) {
505			printf("The following duplicate blocks remain:");
506			for (dp = duplist; dp; dp = dp->next)
507				printf(" %jd,", (intmax_t)dp->dup);
508			printf("\n");
509		}
510	}
511	duplist = (struct dups *)0;
512	muldup = (struct dups *)0;
513	inocleanup();
514	if (fsmodified) {
515		sblock.fs_time = time(NULL);
516		sbdirty();
517	}
518	if (cvtlevel && sblk.b_dirty) {
519		/*
520		 * Write out the duplicate super blocks
521		 */
522		for (cylno = 0; cylno < sblock.fs_ncg; cylno++)
523			blwrite(fswritefd, (char *)&sblock,
524			    fsbtodb(&sblock, cgsblock(&sblock, cylno)),
525			    SBLOCKSIZE);
526	}
527	if (rerun)
528		resolved = 0;
529	finalIOstats();
530
531	/*
532	 * Check to see if the file system is mounted read-write.
533	 */
534	if (bkgrdflag == 0 && mntp != NULL && (mntp->f_flags & MNT_RDONLY) == 0)
535		resolved = 0;
536	ckfini(resolved);
537
538	for (cylno = 0; cylno < sblock.fs_ncg; cylno++)
539		if (inostathead[cylno].il_stat != NULL)
540			free((char *)inostathead[cylno].il_stat);
541	free((char *)inostathead);
542	inostathead = NULL;
543	if (fsmodified && !preen)
544		printf("\n***** FILE SYSTEM WAS MODIFIED *****\n");
545	if (rerun)
546		printf("\n***** PLEASE RERUN FSCK *****\n");
547	if (chkdoreload(mntp) != 0) {
548		if (!fsmodified)
549			return (0);
550		if (!preen)
551			printf("\n***** REBOOT NOW *****\n");
552		sync();
553		return (4);
554	}
555	return (0);
556}
557
558static int
559chkdoreload(struct statfs *mntp)
560{
561	struct iovec *iov;
562	int iovlen;
563	char errmsg[255];
564
565	if (mntp == NULL)
566		return (0);
567
568	iov = NULL;
569	iovlen = 0;
570	errmsg[0] = '\0';
571	/*
572	 * We modified a mounted file system.  Do a mount update on
573	 * it unless it is read-write, so we can continue using it
574	 * as safely as possible.
575	 */
576	if (mntp->f_flags & MNT_RDONLY) {
577		build_iovec(&iov, &iovlen, "fstype", "ffs", 4);
578		build_iovec(&iov, &iovlen, "from", mntp->f_mntfromname,
579		    (size_t)-1);
580		build_iovec(&iov, &iovlen, "fspath", mntp->f_mntonname,
581		    (size_t)-1);
582		build_iovec(&iov, &iovlen, "errmsg", errmsg,
583		    sizeof(errmsg));
584		build_iovec(&iov, &iovlen, "update", NULL, 0);
585		build_iovec(&iov, &iovlen, "reload", NULL, 0);
586		/*
587		 * XX: We need the following line until we clean up
588		 * nmount parsing of root mounts and NFS root mounts.
589		 */
590		build_iovec(&iov, &iovlen, "ro", NULL, 0);
591		if (nmount(iov, iovlen, mntp->f_flags) == 0) {
592			return (0);
593		}
594		pwarn("mount reload of '%s' failed: %s %s\n\n",
595		    mntp->f_mntonname, strerror(errno), errmsg);
596		return (1);
597	}
598	return (0);
599}
600
601/*
602 * Get the mount point information for name.
603 */
604static struct statfs *
605getmntpt(const char *name)
606{
607	struct stat devstat, mntdevstat;
608	char device[sizeof(_PATH_DEV) - 1 + MNAMELEN];
609	char *ddevname;
610	struct statfs *mntbuf, *statfsp;
611	int i, mntsize, isdev;
612
613	if (stat(name, &devstat) != 0)
614		return (NULL);
615	if (S_ISCHR(devstat.st_mode) || S_ISBLK(devstat.st_mode))
616		isdev = 1;
617	else
618		isdev = 0;
619	mntsize = getmntinfo(&mntbuf, MNT_NOWAIT);
620	for (i = 0; i < mntsize; i++) {
621		statfsp = &mntbuf[i];
622		ddevname = statfsp->f_mntfromname;
623		if (*ddevname != '/') {
624			strcpy(device, _PATH_DEV);
625			strcat(device, ddevname);
626			strcpy(statfsp->f_mntfromname, device);
627		}
628		if (isdev == 0) {
629			if (strcmp(name, statfsp->f_mntonname))
630				continue;
631			return (statfsp);
632		}
633		if (stat(ddevname, &mntdevstat) == 0 &&
634		    mntdevstat.st_rdev == devstat.st_rdev)
635			return (statfsp);
636	}
637	statfsp = NULL;
638	return (statfsp);
639}
640
641static void
642usage(void)
643{
644	(void) fprintf(stderr,
645"usage: %s [-BEFfnpry] [-b block] [-c level] [-m mode] filesystem ...\n",
646	    getprogname());
647	exit(1);
648}
649