NetBSD/sbin/fsck_ffs/setup.c

722 lines
20 KiB
C
Raw Normal View History

/* $NetBSD: setup.c,v 1.50 2001/09/18 08:38:28 lukem Exp $ */
1993-03-21 12:45:37 +03:00
/*
* Copyright (c) 1980, 1986, 1993
* The Regents of the University of California. All rights reserved.
1993-03-21 12:45:37 +03:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
1993-03-21 12:45:37 +03:00
#ifndef lint
#if 0
1997-09-16 20:44:43 +04:00
static char sccsid[] = "@(#)setup.c 8.10 (Berkeley) 5/9/95";
#else
__RCSID("$NetBSD: setup.c,v 1.50 2001/09/18 08:38:28 lukem Exp $");
#endif
1993-03-21 12:45:37 +03:00
#endif /* not lint */
#include <sys/param.h>
1994-04-25 22:28:42 +04:00
#include <sys/time.h>
1993-03-21 12:45:37 +03:00
#include <sys/stat.h>
#include <sys/ioctl.h>
1998-11-12 19:19:47 +03:00
#define FSTYPENAMES
1993-03-21 12:45:37 +03:00
#include <sys/disklabel.h>
#include <sys/file.h>
1997-09-16 20:44:43 +04:00
#include <ufs/ufs/dinode.h>
#include <ufs/ufs/ufs_bswap.h>
1997-09-16 20:44:43 +04:00
#include <ufs/ffs/fs.h>
#include <ufs/ffs/ffs_extern.h>
1997-09-16 20:44:43 +04:00
#include <ctype.h>
#include <err.h>
1993-03-21 12:45:37 +03:00
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
1993-03-21 12:45:37 +03:00
#include <string.h>
1993-03-21 12:45:37 +03:00
#include "fsck.h"
#include "extern.h"
1996-09-28 02:45:10 +04:00
#include "fsutil.h"
1993-03-21 12:45:37 +03:00
#define POWEROF2(num) (((num) & ((num) - 1)) == 0)
1997-09-16 20:44:43 +04:00
static void badsb __P((int, char *));
1998-07-27 00:32:42 +04:00
static int calcsb __P((const char *, int, struct fs *));
static struct disklabel *getdisklabel __P((const char *, int));
static int readsb __P((int));
1993-03-21 12:45:37 +03:00
1997-09-16 20:44:43 +04:00
/*
* Read in a superblock finding an alternate if necessary.
* Return 1 if successful, 0 if unsuccessful, -1 if filesystem
* is already clean (preen mode only).
*/
int
1993-03-21 12:45:37 +03:00
setup(dev)
1998-07-27 00:32:42 +04:00
const char *dev;
1993-03-21 12:45:37 +03:00
{
long cg, size, asked, i, j;
long bmapsize;
struct disklabel *lp;
off_t sizepb;
struct stat statb;
1993-03-21 12:45:37 +03:00
struct fs proto;
int doskipclean;
u_int64_t maxfilesize;
struct csum *ccsp;
1993-03-21 12:45:37 +03:00
havesb = 0;
fswritefd = -1;
doskipclean = skipclean;
1993-03-21 12:45:37 +03:00
if (stat(dev, &statb) < 0) {
printf("Can't stat %s: %s\n", dev, strerror(errno));
return (0);
}
if (forceimage) {
if (!S_ISREG(statb.st_mode)) {
pfatal("%s is not a regular file", dev);
if (reply("CONTINUE") == 0)
return (0);
}
} else if (!S_ISCHR(statb.st_mode)) {
1993-03-21 12:45:37 +03:00
pfatal("%s is not a character device", dev);
if (reply("CONTINUE") == 0)
return (0);
}
if ((fsreadfd = open(dev, O_RDONLY)) < 0) {
printf("Can't open %s: %s\n", dev, strerror(errno));
return (0);
}
if (preen == 0)
printf("** %s", dev);
if (nflag || (fswritefd = open(dev, O_WRONLY)) < 0) {
fswritefd = -1;
if (preen)
pfatal("NO WRITE ACCESS");
printf(" (NO WRITE)");
}
if (preen == 0)
printf("\n");
fsmodified = 0;
lfdir = 0;
initbarea(&sblk);
initbarea(&asblk);
sblk.b_un.b_buf = malloc(SBSIZE);
sblock = malloc(SBSIZE);
1993-03-21 12:45:37 +03:00
asblk.b_un.b_buf = malloc(SBSIZE);
altsblock = malloc(SBSIZE);
if (sblk.b_un.b_buf == NULL || asblk.b_un.b_buf == NULL ||
sblock == NULL || altsblock == NULL)
1997-09-16 20:44:43 +04:00
errx(EEXIT, "cannot allocate space for superblock");
if ((lp = getdisklabel(NULL, fsreadfd)) != NULL)
1993-03-21 12:45:37 +03:00
dev_bsize = secsize = lp->d_secsize;
else
dev_bsize = secsize = DEV_BSIZE;
/*
* Read in the superblock, looking for alternates if necessary
*/
if (readsb(1) == 0) {
if (bflag || preen || calcsb(dev, fsreadfd, &proto) == 0)
return(0);
if (reply("LOOK FOR ALTERNATE SUPERBLOCKS") == 0)
return (0);
for (cg = 0; cg < proto.fs_ncg; cg++) {
bflag = fsbtodb(&proto, cgsblock(&proto, cg));
if (readsb(0) != 0)
break;
}
if (cg >= proto.fs_ncg) {
printf("%s %s\n%s %s\n%s %s\n",
"SEARCH FOR ALTERNATE SUPER-BLOCK",
"FAILED. YOU MUST USE THE",
"-b OPTION TO fsck_ffs TO SPECIFY THE",
1993-03-21 12:45:37 +03:00
"LOCATION OF AN ALTERNATE",
"SUPER-BLOCK TO SUPPLY NEEDED",
"INFORMATION; SEE fsck_ffs(8).");
1993-03-21 12:45:37 +03:00
return(0);
}
doskipclean = 0;
1993-03-21 12:45:37 +03:00
pwarn("USING ALTERNATE SUPERBLOCK AT %d\n", bflag);
}
if (debug)
printf("clean = %d\n", sblock->fs_clean);
if (doswap)
doskipclean = 0;
if (sblock->fs_clean & FS_ISCLEAN) {
if (doskipclean) {
pwarn("%sile system is clean; not checking\n",
preen ? "f" : "** F");
return (-1);
}
if (!preen && !doswap)
pwarn("** File system is already clean\n");
}
maxfsblock = sblock->fs_size;
maxino = sblock->fs_ncg * sblock->fs_ipg;
sizepb = sblock->fs_bsize;
maxfilesize = sblock->fs_bsize * NDADDR - 1;
for (i = 0; i < NIADDR; i++) {
sizepb *= NINDIR(sblock);
maxfilesize += sizepb;
}
1993-03-21 12:45:37 +03:00
/*
* Check and potentially fix certain fields in the super block.
*/
if (sblock->fs_optim != FS_OPTTIME && sblock->fs_optim != FS_OPTSPACE) {
1993-03-21 12:45:37 +03:00
pfatal("UNDEFINED OPTIMIZATION IN SUPERBLOCK");
if (reply("SET TO DEFAULT") == 1) {
sblock->fs_optim = FS_OPTTIME;
1993-03-21 12:45:37 +03:00
sbdirty();
}
}
if ((sblock->fs_minfree < 0 || sblock->fs_minfree > 99)) {
1993-03-21 12:45:37 +03:00
pfatal("IMPOSSIBLE MINFREE=%d IN SUPERBLOCK",
sblock->fs_minfree);
1993-03-21 12:45:37 +03:00
if (reply("SET TO DEFAULT") == 1) {
sblock->fs_minfree = 10;
1993-03-21 12:45:37 +03:00
sbdirty();
}
}
if (sblock->fs_interleave < 1 ||
sblock->fs_interleave > sblock->fs_nsect) {
1993-03-21 12:45:37 +03:00
pwarn("IMPOSSIBLE INTERLEAVE=%d IN SUPERBLOCK",
sblock->fs_interleave);
sblock->fs_interleave = 1;
1993-03-21 12:45:37 +03:00
if (preen)
printf(" (FIXED)\n");
if (preen || reply("SET TO DEFAULT") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_npsect < sblock->fs_nsect ||
sblock->fs_npsect > sblock->fs_nsect*2) {
1993-03-21 12:45:37 +03:00
pwarn("IMPOSSIBLE NPSECT=%d IN SUPERBLOCK",
sblock->fs_npsect);
sblock->fs_npsect = sblock->fs_nsect;
1993-03-21 12:45:37 +03:00
if (preen)
printf(" (FIXED)\n");
if (preen || reply("SET TO DEFAULT") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_bmask != ~(sblock->fs_bsize - 1)) {
pwarn("INCORRECT BMASK=0x%x IN SUPERBLOCK",
sblock->fs_bmask);
sblock->fs_bmask = ~(sblock->fs_bsize - 1);
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_fmask != ~(sblock->fs_fsize - 1)) {
pwarn("INCORRECT FMASK=0x%x IN SUPERBLOCK",
sblock->fs_fmask);
sblock->fs_fmask = ~(sblock->fs_fsize - 1);
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_inodefmt >= FS_44INODEFMT) {
if (sblock->fs_maxfilesize != maxfilesize) {
pwarn("INCORRECT MAXFILESIZE=%lld IN SUPERBLOCK",
(unsigned long long)sblock->fs_maxfilesize);
sblock->fs_maxfilesize = maxfilesize;
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_maxsymlinklen != MAXSYMLINKLEN) {
pwarn("INCORRECT MAXSYMLINKLEN=%d IN SUPERBLOCK",
sblock->fs_maxsymlinklen);
sblock->fs_maxsymlinklen = MAXSYMLINKLEN;
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_qbmask != ~sblock->fs_bmask) {
pwarn("INCORRECT QBMASK=%llx IN SUPERBLOCK",
(unsigned long long)sblock->fs_qbmask);
sblock->fs_qbmask = ~sblock->fs_bmask;
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
if (sblock->fs_qfmask != ~sblock->fs_fmask) {
pwarn("INCORRECT QFMASK=%llx IN SUPERBLOCK",
(unsigned long long)sblock->fs_qfmask);
sblock->fs_qfmask = ~sblock->fs_fmask;
if (preen)
printf(" (FIXED)\n");
if (preen || reply("FIX") == 1) {
sbdirty();
dirty(&asblk);
}
}
newinofmt = 1;
} else {
sblock->fs_qbmask = ~sblock->fs_bmask;
sblock->fs_qfmask = ~sblock->fs_fmask;
newinofmt = 0;
}
/*
* Convert to new inode format.
*/
if (cvtlevel >= 2 && sblock->fs_inodefmt < FS_44INODEFMT) {
if (preen)
pwarn("CONVERTING TO NEW INODE FORMAT\n");
else if (!reply("CONVERT TO NEW INODE FORMAT"))
return(0);
doinglevel2++;
sblock->fs_inodefmt = FS_44INODEFMT;
sblock->fs_maxfilesize = maxfilesize;
sblock->fs_maxsymlinklen = MAXSYMLINKLEN;
sblock->fs_qbmask = ~sblock->fs_bmask;
sblock->fs_qfmask = ~sblock->fs_fmask;
sbdirty();
dirty(&asblk);
}
/*
* Convert to new cylinder group format.
*/
if (cvtlevel >= 1 && sblock->fs_postblformat == FS_42POSTBLFMT) {
if (preen)
pwarn("CONVERTING TO NEW CYLINDER GROUP FORMAT\n");
else if (!reply("CONVERT TO NEW CYLINDER GROUP FORMAT"))
return(0);
doinglevel1++;
sblock->fs_postblformat = FS_DYNAMICPOSTBLFMT;
sblock->fs_nrpos = 8;
sblock->fs_postbloff =
(char *)(&sblock->fs_opostbl[0][0]) -
(char *)(&sblock->fs_firstfield);
sblock->fs_rotbloff = &sblock->fs_space[0] -
(u_char *)(&sblock->fs_firstfield);
sblock->fs_cgsize =
fragroundup(sblock, CGSIZE(sblock));
sbdirty();
dirty(&asblk);
1993-03-21 12:45:37 +03:00
}
if (asblk.b_dirty && !bflag) {
memmove((struct fs*)sblk.b_un.b_fs, sblock, SBSIZE);
if (needswap)
ffs_sb_swap(sblock, (struct fs*)sblk.b_un.b_fs);
memmove(asblk.b_un.b_fs, sblk.b_un.b_fs, (size_t)sblock->fs_sbsize);
1993-03-21 12:45:37 +03:00
flush(fswritefd, &asblk);
}
/*
* read in the summary info.
*/
asked = 0;
sblock->fs_csp = (struct csum *)calloc(1, sblock->fs_cssize);
for (i = 0, j = 0; i < sblock->fs_cssize; i += sblock->fs_bsize, j++) {
size = sblock->fs_cssize - i < sblock->fs_bsize ?
sblock->fs_cssize - i : sblock->fs_bsize;
ccsp = (struct csum *)((char *)sblock->fs_csp + i);
if (bread(fsreadfd, (char *)ccsp,
fsbtodb(sblock, sblock->fs_csaddr + j * sblock->fs_frag),
1993-03-21 12:45:37 +03:00
size) != 0 && !asked) {
pfatal("BAD SUMMARY INFORMATION");
1999-11-15 22:18:24 +03:00
if (reply("CONTINUE") == 0) {
markclean = 0;
1997-09-16 20:44:43 +04:00
exit(EEXIT);
1999-11-15 22:18:24 +03:00
}
1993-03-21 12:45:37 +03:00
asked++;
}
if (doswap) {
ffs_csum_swap(ccsp, ccsp, size);
bwrite(fswritefd, (char *)ccsp,
1999-11-15 22:18:24 +03:00
fsbtodb(sblock,
sblock->fs_csaddr + j * sblock->fs_frag),
size);
}
if (needswap)
ffs_csum_swap(ccsp, ccsp, size);
1993-03-21 12:45:37 +03:00
}
/*
* allocate and initialize the necessary maps
*/
1995-03-21 04:30:09 +03:00
bmapsize = roundup(howmany(maxfsblock, NBBY), sizeof(int16_t));
1993-03-21 12:45:37 +03:00
blockmap = calloc((unsigned)bmapsize, sizeof (char));
if (blockmap == NULL) {
printf("cannot alloc %u bytes for blockmap\n",
(unsigned)bmapsize);
goto badsblabel;
1993-03-21 12:45:37 +03:00
}
statemap = calloc((unsigned)(maxino + 1), sizeof(char));
if (statemap == NULL) {
printf("cannot alloc %u bytes for statemap\n",
(unsigned)(maxino + 1));
goto badsblabel;
1993-03-21 12:45:37 +03:00
}
typemap = calloc((unsigned)(maxino + 1), sizeof(char));
if (typemap == NULL) {
printf("cannot alloc %u bytes for typemap\n",
(unsigned)(maxino + 1));
goto badsblabel;
}
1995-03-21 04:30:09 +03:00
lncntp = (int16_t *)calloc((unsigned)(maxino + 1), sizeof(int16_t));
1993-03-21 12:45:37 +03:00
if (lncntp == NULL) {
printf("cannot alloc %u bytes for lncntp\n",
(unsigned)((maxino + 1) * sizeof(int16_t)));
goto badsblabel;
1993-03-21 12:45:37 +03:00
}
/*
* cs_ndir may be inaccurate, particularly if we're using the -b
* option, so set a minimum to prevent bogus subdirectory reconnects
* and really inefficient directory scans.
* Also set a maximum in case the value is too large.
*/
numdirs = sblock->fs_cstotal.cs_ndir;
if (numdirs < 1024)
numdirs = 1024;
if (numdirs > maxino + 1)
numdirs = maxino + 1;
1993-03-21 12:45:37 +03:00
inplast = 0;
listmax = numdirs + 10;
inpsort = (struct inoinfo **)calloc((unsigned)listmax,
sizeof(struct inoinfo *));
inphead = (struct inoinfo **)calloc((unsigned)numdirs,
sizeof(struct inoinfo *));
if (inpsort == NULL || inphead == NULL) {
printf("cannot alloc %u bytes for inphead\n",
(unsigned)(numdirs * sizeof(struct inoinfo *)));
goto badsblabel;
1993-03-21 12:45:37 +03:00
}
1999-11-15 22:18:24 +03:00
cgrp = malloc(sblock->fs_cgsize);
if (cgrp == NULL) {
printf("cannot alloc %u bytes for cylinder group\n",
sblock->fs_cgsize);
goto badsblabel;
}
1993-03-21 12:45:37 +03:00
bufinit();
1999-11-15 22:18:24 +03:00
if (sblock->fs_flags & FS_DOSOFTDEP)
usedsoftdep = 1;
else
usedsoftdep = 0;
1993-03-21 12:45:37 +03:00
return (1);
badsblabel:
markclean=0;
ckfini();
1993-03-21 12:45:37 +03:00
return (0);
}
/*
* Read in the super block and its summary info.
*/
static int
1993-03-21 12:45:37 +03:00
readsb(listerr)
int listerr;
{
1997-09-16 20:44:43 +04:00
ufs_daddr_t super = bflag ? bflag : SBOFF / dev_bsize;
struct fs *fs;
1993-03-21 12:45:37 +03:00
if (bread(fsreadfd, (char *)sblk.b_un.b_fs, super, (long)SBSIZE) != 0)
1993-03-21 12:45:37 +03:00
return (0);
sblk.b_bno = super;
sblk.b_size = SBSIZE;
fs = sblk.b_un.b_fs;
/* auto detect byte order */
if( fs->fs_magic == FS_MAGIC) {
if (endian == 0 || BYTE_ORDER == endian) {
needswap = 0;
doswap = do_blkswap = do_dirswap = 0;
} else {
needswap = 1;
doswap = do_blkswap = do_dirswap = 1;
}
} else if (fs->fs_magic == bswap32(FS_MAGIC)) {
if (endian == 0 || BYTE_ORDER != endian) {
needswap = 1;
doswap = do_blkswap = do_dirswap = 0;
} else {
needswap = 0;
doswap = do_blkswap = do_dirswap = 1;
}
} else {
badsb(listerr, "MAGIC NUMBER WRONG");
return (0);
}
if (doswap) {
if (preen)
errx(EEXIT, "incompatible options -B and -p");
if (nflag)
errx(EEXIT, "incompatible options -B and -n");
if (endian == LITTLE_ENDIAN) {
1999-05-02 00:04:14 +04:00
if (!reply("CONVERT TO LITTLE ENDIAN"))
return 0;
} else if (endian == BIG_ENDIAN) {
1999-05-02 00:04:14 +04:00
if (!reply("CONVERT TO BIG ENDIAN"))
return 0;
} else
pfatal("INTERNAL ERROR: unknown endian");
}
if (needswap)
printf("** Swapped byte order\n");
/* swap SB byte order if asked */
if (doswap)
ffs_sb_swap(sblk.b_un.b_fs, sblk.b_un.b_fs);
memmove(sblock, sblk.b_un.b_fs, SBSIZE);
if (needswap)
ffs_sb_swap(sblk.b_un.b_fs, sblock);
1993-03-21 12:45:37 +03:00
/*
* run a few consistency checks of the super block
*/
if (sblock->fs_ncg < 1)
1993-03-21 12:45:37 +03:00
{ badsb(listerr, "NCG OUT OF RANGE"); return (0); }
if (sblock->fs_cpg < 1)
1993-03-21 12:45:37 +03:00
{ badsb(listerr, "CPG OUT OF RANGE"); return (0); }
if (sblock->fs_ncg * sblock->fs_cpg < sblock->fs_ncyl ||
(sblock->fs_ncg - 1) * sblock->fs_cpg >= sblock->fs_ncyl)
1993-03-21 12:45:37 +03:00
{ badsb(listerr, "NCYL LESS THAN NCG*CPG"); return (0); }
if (sblock->fs_sbsize > SBSIZE)
1993-03-21 12:45:37 +03:00
{ badsb(listerr, "SIZE PREPOSTEROUSLY LARGE"); return (0); }
/*
* Compute block size that the filesystem is based on,
* according to fsbtodb, and adjust superblock block number
* so we can tell if this is an alternate later.
*/
super *= dev_bsize;
dev_bsize = sblock->fs_fsize / fsbtodb(sblock, 1);
1993-03-21 12:45:37 +03:00
sblk.b_bno = super / dev_bsize;
if (bflag) {
havesb = 1;
return (1);
}
1993-03-21 12:45:37 +03:00
/*
* Set all possible fields that could differ, then do check
* of whole super block against an alternate super block.
* When an alternate super-block is specified this check is skipped.
*/
getblk(&asblk, cgsblock(sblock, sblock->fs_ncg - 1), sblock->fs_sbsize);
1993-03-21 12:45:37 +03:00
if (asblk.b_errs)
return (0);
/* swap SB byte order if asked */
if (doswap)
ffs_sb_swap(asblk.b_un.b_fs, asblk.b_un.b_fs);
memmove(altsblock, asblk.b_un.b_fs, sblock->fs_sbsize);
if (needswap)
ffs_sb_swap(asblk.b_un.b_fs, altsblock);
if (cmpsblks(sblock, altsblock)) {
1994-12-28 03:03:49 +03:00
if (debug) {
long *nlp, *olp, *endlp;
printf("superblock mismatches\n");
nlp = (long *)altsblock;
olp = (long *)sblock;
endlp = olp + (sblock->fs_sbsize / sizeof *olp);
1994-12-28 03:03:49 +03:00
for ( ; olp < endlp; olp++, nlp++) {
if (*olp == *nlp)
continue;
printf("offset %ld, original %lx, alternate %lx\n",
(long)(olp - (long *)sblock), *olp, *nlp);
1994-12-28 03:03:49 +03:00
}
}
1993-03-21 12:45:37 +03:00
badsb(listerr,
"VALUES IN SUPER BLOCK DISAGREE WITH THOSE IN FIRST ALTERNATE");
return (0);
}
/* Now we know the SB is valid, we can write it back if needed */
if (doswap) {
sbdirty();
dirty(&asblk);
}
1993-03-21 12:45:37 +03:00
havesb = 1;
return (1);
}
int
cmpsblks(const struct fs *sb, struct fs *asb)
{
/*
* Copy fields which we don't care if they're different in the
* alternate superblocks, as they're either likely to be
* different because they're per-cylinder-group specific, or
* because they're transient details which are only maintained
* in the primary superblock.
*/
asb->fs_firstfield = sb->fs_firstfield;
asb->fs_unused_1 = sb->fs_unused_1;
asb->fs_time = sb->fs_time;
asb->fs_cstotal = sb->fs_cstotal;
asb->fs_cgrotor = sb->fs_cgrotor;
asb->fs_fmod = sb->fs_fmod;
asb->fs_clean = sb->fs_clean;
asb->fs_ronly = sb->fs_ronly;
asb->fs_flags = sb->fs_flags;
asb->fs_maxcontig = sb->fs_maxcontig;
asb->fs_minfree = sb->fs_minfree;
asb->fs_optim = sb->fs_optim;
asb->fs_rotdelay = sb->fs_rotdelay;
asb->fs_maxbpg = sb->fs_maxbpg;
memmove(asb->fs_ocsp, sb->fs_ocsp, sizeof sb->fs_ocsp);
Incorporate the enhanced ffs_dirpref() by Grigoriy Orlov, as found in FreeBSD (three commits; the initial work, man page updates, and a fix to ffs_reload()), with the following differences: - Be consistent between newfs(8) and tunefs(8) as to the options which set and control the tuning parameters for this work (avgfilesize & avgfpdir) - Use u_int16_t instead of u_int8_t to keep track of the number of contiguous directories (suggested by Chuck Silvers) - Work within our FFS_EI framework - Ensure that fs->fs_maxclusters and fs->fs_contigdirs don't point to the same area of memory The new algorithm has a marked performance increase, especially when performing tasks such as untarring pkgsrc.tar.gz, etc. The original FreeBSD commit messages are attached: ===== mckusick 2001/04/10 01:39:00 PDT Directory layout preference improvements from Grigoriy Orlov <gluk@ptci.ru>. His description of the problem and solution follow. My own tests show speedups on typical filesystem intensive workloads of 5% to 12% which is very impressive considering the small amount of code change involved. ------ One day I noticed that some file operations run much faster on small file systems then on big ones. I've looked at the ffs algorithms, thought about them, and redesigned the dirpref algorithm. First I want to describe the results of my tests. These results are old and I have improved the algorithm after these tests were done. Nevertheless they show how big the perfomance speedup may be. I have done two file/directory intensive tests on a two OpenBSD systems with old and new dirpref algorithm. The first test is "tar -xzf ports.tar.gz", the second is "rm -rf ports". The ports.tar.gz file is the ports collection from the OpenBSD 2.8 release. It contains 6596 directories and 13868 files. The test systems are: 1. Celeron-450, 128Mb, two IDE drives, the system at wd0, file system for test is at wd1. Size of test file system is 8 Gb, number of cg=991, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=35 2. PIII-600, 128Mb, two IBM DTLA-307045 IDE drives at i815e, the system at wd0, file system for test is at wd1. Size of test file system is 40 Gb, number of cg=5324, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=50 You can get more info about the test systems and methods at: http://www.ptci.ru/gluk/dirpref/old/dirpref.html Test Results tar -xzf ports.tar.gz rm -rf ports mode old dirpref new dirpref speedup old dirprefnew dirpref speedup First system normal 667 472 1.41 477 331 1.44 async 285 144 1.98 130 14 9.29 sync 768 616 1.25 477 334 1.43 softdep 413 252 1.64 241 38 6.34 Second system normal 329 81 4.06 263.5 93.5 2.81 async 302 25.7 11.75 112 2.26 49.56 sync 281 57.0 4.93 263 90.5 2.9 softdep 341 40.6 8.4 284 4.76 59.66 "old dirpref" and "new dirpref" columns give a test time in seconds. speedup - speed increasement in times, ie. old dirpref / new dirpref. ------ Algorithm description The old dirpref algorithm is described in comments: /* * Find a cylinder to place a directory. * * The policy implemented by this algorithm is to select from * among those cylinder groups with above the average number of * free inodes, the one with the smallest number of directories. */ A new directory is allocated in a different cylinder groups than its parent directory resulting in a directory tree that is spreaded across all the cylinder groups. This spreading out results in a non-optimal access to the directories and files. When we have a small filesystem it is not a problem but when the filesystem is big then perfomance degradation becomes very apparent. What I mean by a big file system ? 1. A big filesystem is a filesystem which occupy 20-30 or more percent of total drive space, i.e. first and last cylinder are physically located relatively far from each other. 2. It has a relatively large number of cylinder groups, for example more cylinder groups than 50% of the buffers in the buffer cache. The first results in long access times, while the second results in many buffers being used by metadata operations. Such operations use cylinder group blocks and on-disk inode blocks. The cylinder group block (fs->fs_cblkno) contains struct cg, inode and block bit maps. It is 2k in size for the default filesystem parameters. If new and parent directories are located in different cylinder groups then the system performs more input/output operations and uses more buffers. On filesystems with many cylinder groups, lots of cache buffers are used for metadata operations. My solution for this problem is very simple. I allocate many directories in one cylinder group. I also do some things, so that the new allocation method does not cause excessive fragmentation and all directory inodes will not be located at a location far from its file's inodes and data. The algorithm is: /* * Find a cylinder group to place a directory. * * The policy implemented by this algorithm is to allocate a * directory inode in the same cylinder group as its parent * directory, but also to reserve space for its files inodes * and data. Restrict the number of directories which may be * allocated one after another in the same cylinder group * without intervening allocation of files. * * If we allocate a first level directory then force allocation * in another cylinder group. */ My early versions of dirpref give me a good results for a wide range of file operations and different filesystem capacities except one case: those applications that create their entire directory structure first and only later fill this structure with files. My solution for such and similar cases is to limit a number of directories which may be created one after another in the same cylinder group without intervening file creations. For this purpose, I allocate an array of counters at mount time. This array is linked to the superblock fs->fs_contigdirs[cg]. Each time a directory is created the counter increases and each time a file is created the counter decreases. A 60Gb filesystem with 8mb/cg requires 10kb of memory for the counters array. The maxcontigdirs is a maximum number of directories which may be created without an intervening file creation. I found in my tests that the best performance occurs when I restrict the number of directories in one cylinder group such that all its files may be located in the same cylinder group. There may be some deterioration in performance if all the file inodes are in the same cylinder group as its containing directory, but their data partially resides in a different cylinder group. The maxcontigdirs value is calculated to try to prevent this condition. Since there is no way to know how many files and directories will be allocated later I added two optimization parameters in superblock/tunefs. They are: int32_t fs_avgfilesize; /* expected average file size */ int32_t fs_avgfpdir; /* expected # of files per directory */ These parameters have reasonable defaults but may be tweeked for special uses of a filesystem. They are only necessary in rare cases like better tuning a filesystem being used to store a squid cache. I have been using this algorithm for about 3 months. I have done a lot of testing on filesystems with different capacities, average filesize, average number of files per directory, and so on. I think this algorithm has no negative impact on filesystem perfomance. It works better than the default one in all cases. The new dirpref will greatly improve untarring/removing/coping of big directories, decrease load on cvs servers and much more. The new dirpref doesn't speedup a compilation process, but also doesn't slow it down. Obtained from: Grigoriy Orlov <gluk@ptci.ru> ===== ===== iedowse 2001/04/23 17:37:17 PDT Pre-dirpref versions of fsck may zero out the new superblock fields fs_contigdirs, fs_avgfilesize and fs_avgfpdir. This could cause panics if these fields were zeroed while a filesystem was mounted read-only, and then remounted read-write. Add code to ffs_reload() which copies the fs_contigdirs pointer from the previous superblock, and reinitialises fs_avgf* if necessary. Reviewed by: mckusick ===== ===== nik 2001/04/10 03:36:44 PDT Add information about the new options to newfs and tunefs which set the expected average file size and number of files per directory. Could do with some fleshing out. =====
2001-09-06 06:16:00 +04:00
asb->fs_contigdirs = sb->fs_contigdirs;
asb->fs_csp = sb->fs_csp;
asb->fs_maxcluster = sb->fs_maxcluster;
memmove(asb->fs_fsmnt, sb->fs_fsmnt, sizeof sb->fs_fsmnt);
Incorporate the enhanced ffs_dirpref() by Grigoriy Orlov, as found in FreeBSD (three commits; the initial work, man page updates, and a fix to ffs_reload()), with the following differences: - Be consistent between newfs(8) and tunefs(8) as to the options which set and control the tuning parameters for this work (avgfilesize & avgfpdir) - Use u_int16_t instead of u_int8_t to keep track of the number of contiguous directories (suggested by Chuck Silvers) - Work within our FFS_EI framework - Ensure that fs->fs_maxclusters and fs->fs_contigdirs don't point to the same area of memory The new algorithm has a marked performance increase, especially when performing tasks such as untarring pkgsrc.tar.gz, etc. The original FreeBSD commit messages are attached: ===== mckusick 2001/04/10 01:39:00 PDT Directory layout preference improvements from Grigoriy Orlov <gluk@ptci.ru>. His description of the problem and solution follow. My own tests show speedups on typical filesystem intensive workloads of 5% to 12% which is very impressive considering the small amount of code change involved. ------ One day I noticed that some file operations run much faster on small file systems then on big ones. I've looked at the ffs algorithms, thought about them, and redesigned the dirpref algorithm. First I want to describe the results of my tests. These results are old and I have improved the algorithm after these tests were done. Nevertheless they show how big the perfomance speedup may be. I have done two file/directory intensive tests on a two OpenBSD systems with old and new dirpref algorithm. The first test is "tar -xzf ports.tar.gz", the second is "rm -rf ports". The ports.tar.gz file is the ports collection from the OpenBSD 2.8 release. It contains 6596 directories and 13868 files. The test systems are: 1. Celeron-450, 128Mb, two IDE drives, the system at wd0, file system for test is at wd1. Size of test file system is 8 Gb, number of cg=991, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=35 2. PIII-600, 128Mb, two IBM DTLA-307045 IDE drives at i815e, the system at wd0, file system for test is at wd1. Size of test file system is 40 Gb, number of cg=5324, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=50 You can get more info about the test systems and methods at: http://www.ptci.ru/gluk/dirpref/old/dirpref.html Test Results tar -xzf ports.tar.gz rm -rf ports mode old dirpref new dirpref speedup old dirprefnew dirpref speedup First system normal 667 472 1.41 477 331 1.44 async 285 144 1.98 130 14 9.29 sync 768 616 1.25 477 334 1.43 softdep 413 252 1.64 241 38 6.34 Second system normal 329 81 4.06 263.5 93.5 2.81 async 302 25.7 11.75 112 2.26 49.56 sync 281 57.0 4.93 263 90.5 2.9 softdep 341 40.6 8.4 284 4.76 59.66 "old dirpref" and "new dirpref" columns give a test time in seconds. speedup - speed increasement in times, ie. old dirpref / new dirpref. ------ Algorithm description The old dirpref algorithm is described in comments: /* * Find a cylinder to place a directory. * * The policy implemented by this algorithm is to select from * among those cylinder groups with above the average number of * free inodes, the one with the smallest number of directories. */ A new directory is allocated in a different cylinder groups than its parent directory resulting in a directory tree that is spreaded across all the cylinder groups. This spreading out results in a non-optimal access to the directories and files. When we have a small filesystem it is not a problem but when the filesystem is big then perfomance degradation becomes very apparent. What I mean by a big file system ? 1. A big filesystem is a filesystem which occupy 20-30 or more percent of total drive space, i.e. first and last cylinder are physically located relatively far from each other. 2. It has a relatively large number of cylinder groups, for example more cylinder groups than 50% of the buffers in the buffer cache. The first results in long access times, while the second results in many buffers being used by metadata operations. Such operations use cylinder group blocks and on-disk inode blocks. The cylinder group block (fs->fs_cblkno) contains struct cg, inode and block bit maps. It is 2k in size for the default filesystem parameters. If new and parent directories are located in different cylinder groups then the system performs more input/output operations and uses more buffers. On filesystems with many cylinder groups, lots of cache buffers are used for metadata operations. My solution for this problem is very simple. I allocate many directories in one cylinder group. I also do some things, so that the new allocation method does not cause excessive fragmentation and all directory inodes will not be located at a location far from its file's inodes and data. The algorithm is: /* * Find a cylinder group to place a directory. * * The policy implemented by this algorithm is to allocate a * directory inode in the same cylinder group as its parent * directory, but also to reserve space for its files inodes * and data. Restrict the number of directories which may be * allocated one after another in the same cylinder group * without intervening allocation of files. * * If we allocate a first level directory then force allocation * in another cylinder group. */ My early versions of dirpref give me a good results for a wide range of file operations and different filesystem capacities except one case: those applications that create their entire directory structure first and only later fill this structure with files. My solution for such and similar cases is to limit a number of directories which may be created one after another in the same cylinder group without intervening file creations. For this purpose, I allocate an array of counters at mount time. This array is linked to the superblock fs->fs_contigdirs[cg]. Each time a directory is created the counter increases and each time a file is created the counter decreases. A 60Gb filesystem with 8mb/cg requires 10kb of memory for the counters array. The maxcontigdirs is a maximum number of directories which may be created without an intervening file creation. I found in my tests that the best performance occurs when I restrict the number of directories in one cylinder group such that all its files may be located in the same cylinder group. There may be some deterioration in performance if all the file inodes are in the same cylinder group as its containing directory, but their data partially resides in a different cylinder group. The maxcontigdirs value is calculated to try to prevent this condition. Since there is no way to know how many files and directories will be allocated later I added two optimization parameters in superblock/tunefs. They are: int32_t fs_avgfilesize; /* expected average file size */ int32_t fs_avgfpdir; /* expected # of files per directory */ These parameters have reasonable defaults but may be tweeked for special uses of a filesystem. They are only necessary in rare cases like better tuning a filesystem being used to store a squid cache. I have been using this algorithm for about 3 months. I have done a lot of testing on filesystems with different capacities, average filesize, average number of files per directory, and so on. I think this algorithm has no negative impact on filesystem perfomance. It works better than the default one in all cases. The new dirpref will greatly improve untarring/removing/coping of big directories, decrease load on cvs servers and much more. The new dirpref doesn't speedup a compilation process, but also doesn't slow it down. Obtained from: Grigoriy Orlov <gluk@ptci.ru> ===== ===== iedowse 2001/04/23 17:37:17 PDT Pre-dirpref versions of fsck may zero out the new superblock fields fs_contigdirs, fs_avgfilesize and fs_avgfpdir. This could cause panics if these fields were zeroed while a filesystem was mounted read-only, and then remounted read-write. Add code to ffs_reload() which copies the fs_contigdirs pointer from the previous superblock, and reinitialises fs_avgf* if necessary. Reviewed by: mckusick ===== ===== nik 2001/04/10 03:36:44 PDT Add information about the new options to newfs and tunefs which set the expected average file size and number of files per directory. Could do with some fleshing out. =====
2001-09-06 06:16:00 +04:00
memmove(asb->fs_snapinum,
sb->fs_snapinum, sizeof sb->fs_snapinum);
asb->fs_avgfilesize = sb->fs_avgfilesize;
asb->fs_avgfpdir = sb->fs_avgfpdir;
memmove(asb->fs_sparecon,
sb->fs_sparecon, sizeof sb->fs_sparecon);
/*
* The following should not have to be copied, but need to be.
*/
asb->fs_fsbtodb = sb->fs_fsbtodb;
asb->fs_interleave = sb->fs_interleave;
asb->fs_npsect = sb->fs_npsect;
asb->fs_nrpos = sb->fs_nrpos;
asb->fs_qbmask = sb->fs_qbmask;
asb->fs_qfmask = sb->fs_qfmask;
asb->fs_state = sb->fs_state;
asb->fs_maxfilesize = sb->fs_maxfilesize;
/*
* Compare the superblocks, effectively checking every other
* field to see if they differ.
*/
return (memcmp(sb, asb, (int)sb->fs_sbsize));
}
1997-09-16 20:44:43 +04:00
static void
1993-03-21 12:45:37 +03:00
badsb(listerr, s)
int listerr;
char *s;
{
if (!listerr)
return;
if (preen)
printf("%s: ", cdevname());
1993-03-21 12:45:37 +03:00
pfatal("BAD SUPER BLOCK: %s\n", s);
}
/*
* Calculate a prototype superblock based on information in the disk label.
* When done the cgsblock macro can be calculated and the fs_ncg field
* can be used. Do NOT attempt to use other macros without verifying that
* their needed information is available!
*/
1997-09-16 20:44:43 +04:00
static int
1993-03-21 12:45:37 +03:00
calcsb(dev, devfd, fs)
1998-07-27 00:32:42 +04:00
const char *dev;
1993-03-21 12:45:37 +03:00
int devfd;
struct fs *fs;
1993-03-21 12:45:37 +03:00
{
struct disklabel *lp;
struct partition *pp;
char *cp;
1993-03-21 12:45:37 +03:00
int i;
cp = strchr(dev, '\0') - 1;
if ((cp == (char *)-1 || (*cp < 'a' || *cp > 'h')) && !isdigit(*cp)) {
1993-03-21 12:45:37 +03:00
pfatal("%s: CANNOT FIGURE OUT FILE SYSTEM PARTITION\n", dev);
return (0);
}
lp = getdisklabel(dev, devfd);
if (isdigit(*cp))
pp = &lp->d_partitions[0];
else
pp = &lp->d_partitions[*cp - 'a'];
if (pp->p_fstype != FS_BSDFFS) {
pfatal("%s: NOT LABELED AS A BSD FILE SYSTEM (%s)\n",
dev, pp->p_fstype < FSMAXTYPES ?
fstypenames[pp->p_fstype] : "unknown");
return (0);
}
/* avoid divide by 0 */
if (pp->p_fsize == 0 || pp->p_frag == 0)
return (0);
memset(fs, 0, sizeof(struct fs));
1993-03-21 12:45:37 +03:00
fs->fs_fsize = pp->p_fsize;
fs->fs_frag = pp->p_frag;
fs->fs_cpg = pp->p_cpg;
fs->fs_size = pp->p_size;
fs->fs_ntrak = lp->d_ntracks;
fs->fs_nsect = lp->d_nsectors;
fs->fs_spc = lp->d_secpercyl;
fs->fs_nspf = fs->fs_fsize / lp->d_secsize;
fs->fs_sblkno = roundup(
howmany(lp->d_bbsize + lp->d_sbsize, fs->fs_fsize),
fs->fs_frag);
fs->fs_cgmask = 0xffffffff;
for (i = fs->fs_ntrak; i > 1; i >>= 1)
fs->fs_cgmask <<= 1;
if (!POWEROF2(fs->fs_ntrak))
fs->fs_cgmask <<= 1;
fs->fs_cgoffset = roundup(
howmany(fs->fs_nsect, NSPF(fs)), fs->fs_frag);
fs->fs_fpg = (fs->fs_cpg * fs->fs_spc) / NSPF(fs);
fs->fs_ncg = howmany(fs->fs_size / fs->fs_spc, fs->fs_cpg);
for (fs->fs_fsbtodb = 0, i = NSPF(fs); i > 1; i >>= 1)
fs->fs_fsbtodb++;
dev_bsize = lp->d_secsize;
return (1);
}
static struct disklabel *
1993-03-21 12:45:37 +03:00
getdisklabel(s, fd)
1998-07-27 00:32:42 +04:00
const char *s;
1993-03-21 12:45:37 +03:00
int fd;
{
static struct disklabel lab;
if (ioctl(fd, DIOCGDINFO, (char *)&lab) < 0) {
if (s == NULL)
return ((struct disklabel *)NULL);
pwarn("ioctl (GCINFO): %s\n", strerror(errno));
1997-09-16 20:44:43 +04:00
errx(EEXIT, "%s: can't read disk label", s);
1993-03-21 12:45:37 +03:00
}
return (&lab);
}